Experimental Characterization of Diamond ... - Springer Link

ISSN 0016-7029, Geochemistry International, 2008, Vol. 46, No. 6, pp. 531–553. © Pleiades Publishing, Ltd., 2008. Original Russian Text © Yu.A. Litvin, V.Yu. Litvin, A.A. Kadik, 2008, published in Geokhimiya, 2008, No. 6, pp. 579–602.

Experimental Characterization of Diamond Crystallization in Melts of Mantle Silicate–Carbonate–Carbon Systems at 7.0–8.5 GPa Yu. A. Litvina, V. Yu. Litvina, and A. A. Kadikb a

Institute of Experimental Mineralogy, Russian Academy of Sciences, Chernogolovka, Moscow oblast, 142432 Russia e-mail: [email protected] b Vernadsky Institute of Geochemistry and Analytical Chemistry, Russian Academy of Sciences, ul. Kosygina 19, Moscow, 119991 Russia Received March 3, 2007

Abstract—Diamond crystallization from carbon solutions in compositionally variable melts of model eclogite with dolomite [CaMg(CO3)2], potassium carbonate (K2CO3), and multicomponent K–Na–Ca–Mg–Fe carbonates was studied at 7.0–8.5 GPa. Concentration barriers for the nucleation of the diamond were determined at a standard pressure of 8.5 GPa for variable proportions of silicate and carbonate components in the growth solutions. They correspond to 35, 65, and 40 wt % of silicate components for systems with dolomite, K2CO3, and carbonatites, respectively. At higher contents of silicates in silicate–carbonate melts, the nucleation of diamond phase ceases, but diamond crystallization on seed crystals continues and is accompanied by the spontaneous crystallization of thermodynamically unstable graphite. In melts of the albite (NaAlSi3O8)–K2CO3–C compositions, the concentration barrier of diamond nucleation at 8.5 GPa is up to 90–92 wt % of the albite component, and diamond growth on seeds was observed in albite–carbon melts. Using mineralogical and experimental data, we developed a model of mantle carbonate–silicate (carbonatite) melts as the main parental media for natural diamonds; it was shown that the composition of the silicate constituent of such parental melts is variable and corresponds to the mantle ultrabasic–basic series. With respect to concentration contributions and dominant role in the genesis of diamond in the Earth’s mantle, major (carbonate and silicate) and minor or admixture components were distinguished. The latter include both soluble in carbonate–silicate melts (oxides, phosphates, chlorides, carbon dioxide, and water) and insoluble components (sulfides, metals, and carbides). Both major and minor components may affect the position of the concentration barriers of diamond nucleation in natural parent media. DOI: 10.1134/S0016702908060013

INTRODUCTION The determination of the chemical composition of parent media is crucial for the problem of diamond genesis under the conditions of the Earth’s mantle. The model of strongly compressed multicomponent carbonate–silicate melts with dissolved carbon [1] is in adequate agreement with experimental [1, 2] and mineralogical [3, 4] data. It also satisfies the experimental criterion of the syngenesis of diamond and growth inclusions in it, according to which a parent media must provide the formation of both diamonds and the mineral matter of primary inclusions in them [1, 5]. Variable proportions of carbonate and silicate components were observed in fragments of parent carbonatite media from inclusions in natural diamonds [3, 6]. Mineral inclusions also indicate [4] that the parent media contained oxide–silicate components of the main mantle assemblages, from olivine-bearing peridotite and pyroxenite rocks to eclogitic and grospyditic assemblages with quartz or coesite. Variable natural carbonate–silicate diamond-forming media may belong to

either carbonatite (more than 50 wt % carbonate components) or ultrabasic–basic compositions enriched in silicate components. It is impossible to determine the chemical nature and compositions of media responsible for diamond formation on the basis of mineralogical data only. They can be constrained by high-pressure experimental investigations, and the key criterion is the efficiency of the medium in diamond nucleation [1, 5]. The efficiency of carbonate–silicate melts with dissolved carbon (from a graphite source) in providing diamond nucleation was first demonstrated for kimberlitic rocks (which are known to be transporting rather than growth media for natural diamonds) in experiments at 7.0–7.7 GPa and 1800–2200°ë [7]. In the quench products of kimberlite melt, diamond coexisted with olivine, clinopyroxene, magnetite, perovskite, apatite, and Ca carbonate. Diamond was also synthesized in the carbonate–oxide–carbon system K2CO3– SiO2–Al2O3–MgO–ë (graphite) with high contents of carbonate (66.6 and 80 wt %) [8]; alkaline carbonate– silicate melts produced wadeite (K2Si4O9), forsterite,

531

532

LITVIN et al.

and magnesian spinel. At 6.3 GPa and 1650°C, diamond nucleation in melts of the K2CO3–Mg2SiO4 and K2CO3–SiO2 systems occurred at Mg2SiO4 and SiO2 contents of up to 50 and 25 wt %, respectively [9, 10]. An increase in the content of silicate or oxide led to the cessation of diamond nucleation, but diamond growth on seeds was observed up to 90 wt % Mg2SiO4 and 75 wt % SiO2. Forsterite, periclase, wadeite, or coesite were formed together with diamond. The model systems of eclogitic garnet and clinopyroxene with carbonates [CaMg(CO3)2 or K2CO3] were studied at 6.0–8.5 GPa and 1200–1800°ë [11]. Diamond nucleation ceased in these systems at certain contents of silicate components, after which diamond grew on seeds, which was accompanied by the crystallization of the thermodynamically unstable graphite phase. For carbonate–silicate melts of varying composition, Litvin et al. [11] developed a concept of the concentration barrier of diamond nucleation as a boundary characteristic of diamond formation. The concentration barrier is determined experimentally from the cessation of diamond nucleation and corresponds to a certain content of silicate components in the growth melt. The carbon supersaturation that is generated in the diamond-forming solution on the diamond nucleation barrier is critically low for diamond nucleation but sufficient for diamond growth on seeds (according to the accepted terminology, this is a boundary between the field of labile supersaturation of carbon solution with respect to diamond and the field of metastable supersaturation). A kinetic phenomenon is observed in metastably supersaturated solutions: the thermodynamically unstable graphite phase nucleates under the P–T conditions of diamond stability. Similar peculiarities of diamond formation in melt–solution systems were previously described in metal–carbon melts during the formation of metal-synthetic diamonds in them at high pressures and temperatures [12]. These features (diamond nucleation, diamond growth on seeds, and nucleation of unstable graphite) are well documented effects in experiments and convenient as a methodological basis for the investigation of diamond-forming systems. For instance, the fact of diamond nucleation indicates labile supersaturation in the diamond-forming melt–solution, and diamond growth on a seed is indicative of the state of its metastable supersaturation (this is also suggested by the nucleation of thermodynamically unstable graphite). Multicomponent K–Na–Ca–Mg–Fe-carbonate–silicate systems (chemical analogs of carbonatite inclusions in diamonds described in [3]) provide efficient diamond nucleation immediately after melting at 5.5−7.0 GPa and 1200–1570°ë. During the first minutes, extensive crystallization of diamonds of octahedral habit was observed in them [2]. At higher pressures of 7.0–8.5 GPa, avalanche-like crystallization of finegrained (0.01–100 µm) polycrystalline intergrowths, more than 1 mm in size, was observed during the first seconds after melting [13]. These aggregates are ana-

logs of monomineralic diamond rocks (diamondites) [14]. The melts of melanocratic carbonatites of the Chagatai complex (Uzbekistan), as model carbonate– silicate media, produced cogenetic diamond and silicate, oxide, phosphate, carbonate, and other minerals at 7.0 GPa and 1450–1550°ë [15, 16]. Diamond crystallization is characterized by high nucleation density and high growth rate, up to the avalanche-like formation of diamondites [13, 17]. The simultaneous formation of diamond and silicate minerals was observed at 5.5– 7.5 GPa and 1420–1700°C in the melts of the dolomite– garnet–clinopyroxene rocks of the Kokchetav massif, Kazakhstan [18, 19]. All these results are consistent with the criterion of syngenesis of diamond and growth inclusions in it [1, 5], because silicate, oxide, and carbonate minerals were formed in the same carbonate– silicate–carbon melts together with diamond. Together with the mineralogical data, they demonstrate the dominant role of carbonate–silicate (carbonatite) melts in diamond genesis. With respect to the syngenesis criterion, sulfide melts (one of the mineralogical versions of parent media [20–23]) could not be growth media for the majority of natural diamonds, although sulfide melts are efficient materials for diamond nucleation [5, 24–26]. It was experimentally shown that silicate and carbonate minerals are insoluble in sulfide melts and, therefore, cannot be formed in them [5, 27]. This means that the growth of diamond in sulfide media cannot be accompanied by the capture of silicate and carbonate minerals, which are absent in these media. Thus, high-pressure and high-temperature experiments showed that carbonate–silicate melts enriched in carbonate components are efficient diamond-forming media. Silicate-dominated silicate–carbonate melts are still poorly studied in this respect. These questions are of prime importance for the problem of diamond genesis, because a characteristic property of the chemical and phase composition of parent melts is their variability, which is distinctly reflected in silicate mineral inclusions [3, 4]. The material of inclusions records a complex history of evolution of parent carbonate–silicate magmas in the peridotite mantle. The experimental investigation of diamond formation conditions in variable carbonate–silicate magmas will eventually allow the determination of natural concentration boundaries for the multicomponent material, which corresponds to the concept of the parent diamond-forming medium, and reveal the physicochemical nature of these boundaries. This will lead to the clear understanding of scales, physicochemical mechanisms, and place of the processes of natural diamond formation in the complex system of mantle magmatism and solution of the key problem of geochemistry and genetic mineralogy. This paper reports the results of an experimental investigation at high pressures of up to 8.5 GPa and temperatures of up to 2000°ë of the diamond formation efficiency of carbonate–silicate–carbon melts depending on composition by the example of simple and multicomponent systems. Natural parent media are charac-

GEOCHEMISTRY INTERNATIONAL

Vol. 46

No. 6

2008

EXPERIMENTAL CHARACTERIZATION OF DIAMOND CRYSTALLIZATION

terized by significant chemical variations, whose limits are not obvious from mineralogical data but can be constrained by experiments. In this study, we begin from model compositions with various silicate and carbonate components of natural diamond-forming media. The experiments also aimed at searching for the silicate– carbonate melts enriched in silicate components relative to carbonate ones that are capable of diamond nucleation. The silicate constituent of parent media is represented by the eclogite assemblage of pyrope–grossular garnet and diopside–jadeite clinopyroxene. Their carbonate component was represented by ä2ëé3 (major component of carbonatite inclusions in natural diamond [3]), dolomite (contains the main components of mantle carbonatites, MgCO3 and CaCO3), and K−Na–Ca–Mg–Fe carbonatites as chemical analogs of inclusions in diamond [3]. The influence of the main silicate components of parent media on the concentration limits of diamond nucleation was investigated at 8.5 GPa in the model systems albite–K2CO3 (with components of diamondiferous coesite eclogites, jadeite and silica), forsterite–K2CO3, and silica–K2CO3. The obtained experimental data are applied to the development of the general physicochemical theory of diamond genesis. EXPERIMENTAL METHODS Finely ground mixtures of carbonate and silicates (Table 1) were blended with high-purity graphite in the weight proportion 60 : 40. The initial mixtures consisted of carbonates (CaCO3, MgCO3, K2CO3, Na2CO3, and FeCO3), synthetic quartz (SiO2), and amorphous materials with compositions of jadeite (NaAlSi2O6), albite (NaAlSi3O8), diopside (CaMgSi2O6), pyrope (Mg3Al2Si3O12), grossular (Ca3Al2Si3O12), and forsterite (Mg2SiO4) prepared from gels [28]. Glasses corresponding to the compositions of model eclogite (Di50Jd50)50(Gros60Prp40)50 (wt %) were prepared in a Pt capsule in a gas apparatus at an argon pressure of 0.1 GPa and 1350°ë for 60 min and quenched to a transparent glass.1 The glass was crushed to a grain size of about 0.5 mm and mixed with CaCO3 and MgCO3 (in the dolomite stoichiometry) to obtain the compositions Ecl50Dol50 and Ecl70Dol30 and with K2CO3 to obtain the compositions Ecl50(K2CO3)50 and Ecl70(K2CO3)30. The mixtures were melted under the same conditions and quenched to glasses (their compositions are given in Table 1). The use of either mixtures or glasses did not significantly affect the nucleation and 1 Symbols:

Ab, Albite NaAlSi3O8; Cpx, clinopyroxene Di50Jd50; D, diamond; Di, diopside CaMgSi2O6; Dol, dolomite CaMg(CO3)2; Ecl, eclogite Cpx60Grt40; Fo, forsterite Mg2SiO4; G, graphite; Grt, garnet Gros60Prp40; Gros, grossular Ca3Al2Si3O12; Jd, jadeite NaAlSi2O6; Coes, coesite SiO2; KC, K carbonate K2ëO3; L, melt (as a quench phase); and Prp, pyrope Mg3Al2Si3O12. GEOCHEMISTRY INTERNATIONAL

Vol. 46

No. 6

533

crystallization of diamond; mixtures were used in the majority of experiments. The efficiency of the melts studied in providing diamond formation was estimated from the nucleation of the diamond phase. The extensive nucleation of diamond in experiments occurs immediately after the formation of carbon solutions in melts showing labile supersaturation to diamond (diamond formation is checked by an increase in the electric resistance of the sample owing to the transformation of conducting graphite into dielectric diamond in solution–melt of carbon). After the experiment, diamond crystals are readily identified under a microscope. This allowed us to compare media of various compositions in terms of their efficiency in diamond formation. Short-term extensive diamond nucleation under the influence of steady labile supersaturation of the melt with respect to diamond was considered efficient. The appearance of sparse spontaneous crystals is not an indicator of the high diamond-forming efficiency of the medium, because it could be caused by fluctuations in carbon concentration up to the level of labile supersaturation in the field of metastable supersaturation with respect to diamond. The probability of this process increases at long experimental durations, which can be misinterpreted as an indication for the long incubation period of diamond nucleation. Experiments were carried out in an anvil-with-hole apparatus with a toroidal gasket [29–31] manufactured from lithographic stone and a graphite heater (which simultaneously served as a container for sample material) with an inner diameter of 3 mm, an outer diameter of 5 mm, and a height of 7.5 mm (Fig. 1). Pressure in the central zone of the cell was determined with an accuracy of ±0.1 GPa using the room-temperature calibration curve of cell pressure (P, GPa)–oil pressure in the hydraulic ram (p, atm) (Fig. 2), which was constructed on the basis of electric resistance changes in reference materials: bismuth at 2.55 GPa (phase I−phase II), 2.70 GPa (II–III), and 7.70 GPa (III–IV) and barium at 5.5 GPa [32]. Temperature was determined with an accuracy of ±15–20°ë from a calibration curve constructed using Pt70Rh30/Pt94Rh06 thermocouples (wire thickness of 0.3 and 0.5 mm). The thermocouple junction was inserted into the central part of the cell (Fig. 1) and separated from the reaction mixture with MgO or ZrO2 coating. The dependence of temperature in the center of the cell on heating power is shown in Fig. 3. At high temperatures, pressure values were adjusted to fit the graphite–diamond equilibrium curve [33], and the correction was no higher than 0.03 GPa at temperatures of 1400–1800°ë [34]. At 7.0– 8.5 GPa and temperatures above 1500°ë, the lithographic stone (natural calcite limestone from Algeti, Georgia) of the assembly is melted near the contact with the graphite heater, and the carbonate melt reacts with the outer zone of the heater producing a crust of polycrystalline diamond, up to 100–150 µm thick. In high-temperature experiments lasting more than 2008

534

LITVIN et al.

Table 1. Compositions of starting materials and experimental silicate and silicate–carbonate mixtures Oxide composition, wt % Component composition K2O

Na2O

MgO

CaO

FeO

Al2O3

SiO2

CO2

–

–

21.86

30.41

–

–

–

47.73

K2CO3

68.16

–

–

–

–

–

–

31.84

Mixture K3

21.40

2.34

11.09

20.28

–

–

–

44.89

Mixture K5

18.55

1.69

8.30

15.08

15.89

–

–

40.49

Ecl = (Di50Jd50)50(Prp40Gros60)50

–

7.67

9.31

12.95

–

12.61

57.47

–

Ecl30Dol70

–

2.30

18.09

25.17

–

3.78

17.24

33.42

Ecl40Dol60

–

3.07

16.84

23.43

–

5.04

22.98

28.64

Ecl50Dol50

–

3.83

15.59

21.69

–

6.31

28.72

23.86

Ecl70Dol30

–

5.37

13.08

18.19

–

8.83

40.21

14.32

Ecl50(K2CO3)50

34.08

3.84

4.66

6.48

–

6.31

28.70

15.92

Ecl60(K2CO3)40

27.26

4.60

5.59

7.77

–

7.57

34.47

12.74

Ecl70(K2CO3)30

20.45

5.37

6.52

9.07

–

8.83

40.21

9.55

Ecl30(mixture K3)70

14.98

3.94

10.55

18.09

–

3.78

17.24

31.42

Ecl40(mixture K5)60

11.13

4.08

8.70

14.23

9.53

5.04

22.99

24.30

Ecl50(mixture K5)50

9.28

4.69

8.81

14.01

7.95

6.31

28.72

20.23

Ecl60(mixture K5)40

7.42

5.28

8.91

13.80

6.36

7.57

34.46

16.20

Ecl70(mixture K5)30

5.57

5.88

9.01

13.58

4.77

8.83

40.21

12.15

Ecl80(mixture K5)20

3.71

6.48

9.11

13.37

3.18

10.09

45.96

8.10

Ab = NaAlSi3O8

–

11.82

–

–

–

19.44

68.74

–

Ab40Dol60

–

4.73

13.12

18.25

–

7.78

27.49

28.63

Ab50Dol50

–

5.91

10.93

15.22

–

9.72

34.36

23.86

Ab50(K2CO3)50

34.08

5.91

–

–

–

9.72

34.36

15.93

Ab60(K2CO3)40

27.26

7.09

–

–

–

11.66

41.25

12.74

Ab70(K2CO3)30

20.45

8.27

–

–

–

13.61

48.12

9.55

Ab80(K2CO3)20

13.63

9.46

–

–

–

15.55

54.99

6.37

Ab90(K2CO3)10

6.82

10.64

–

–

–

17.50

61.86

3.18

Ab40(mixture K3)60

12.84

6.13

6.65

12.17

–

7.78

27.50

26.93

Ab50(mixture K3)50

10.70

7.08

5.55

10.14

–

9.72

34.37

22.45

Fo = Mg2SiO4

–

–

57.30

–

–

–

42.70

–

Fo50(K2CO3)50

34.08

–

28.65

–

–

–

21.35

15.92

Fo60(K2CO3)40

27.26

–

34.38

–

–

–

25.62

12.74

Fo65(K2CO3)35

23.86

–

37.24

–

–

–

27.76

11.14

Fo70(K2CO3)30

20.45

–

40.11

–

–

–

29.89

9.55

Fo80(K2CO3)20

13.63

–

45.84

–

–

–

34.16

6.37

–

–

–

–

–

–

100.00

–

(SiO2)30(K2CO3)70

47.71

–

–

–

–

–

30.00

22.29

(SiO2)40(K2CO3)60

40.90

–

–

–

–

–

40.00

19.10

(SiO2)50(K2CO3)50

34.08

–

–

–

–

–

50.00

15.92

Dol = CaMg(CO3)2

SiO2


Vol. 46

No. 6

2008


535

4 9 1

2 3

8

2

7

6 5 1 mm Fig. 1. Cell assembly of the anvil-with-hole apparatus with a toroidal gasket: 1, high-pressure gasket manufactured from Algeti limestone, Georgia (lithographic stone); 2, protective sleeve of pressed ZrO2 or MgO powder with up to 30 vol % of hexagonal BN; 3, electric furnace manufactured from high-purity graphite; 4, graphite plugs (1 mm thick); 5, sample; 6, thermocouple junction; 7, thermocouple wire; 8. Al2O3-based ceramic tube; 9, position of the “point” pressure sensor based on reference substances (a piece of wire, 0.3 mm in diameter and 0.5–0.7 mm long).

T, °C 2000

P, GPa 10

8

6

7.70 (Bi III–IV)

1500

5.50 (Ba I–II)

1000

4

500 2.70 (Bi II–III)

2

2.55(Bi I–II)

0

0

20 40 60 80 100 120 140 160 180 200 220 240 p, atm

Some features of methodical importance are related to the diamond-forming processes themselves. The initial polycrystalline graphite retained its characteristic fine-grained structure in the carbonate–silicate–graphite samples that were not affected by melting (Fig. 4a). GEOCHEMISTRY INTERNATIONAL

Vol. 46

No. 6

1000

1500

2000

2500 W, W

Fig. 3. Calibration curve of temperature in the center of the cell (T, °C) versus heating power (W, W) according to measurements with Pt70Rh30/Pt94Rh06thermocouples with 0.3 mm thick (unfilled circles) and 0.5 mm thick (filled circles) wires at a pressure of 7.0 GPa. The standard accuracy of temperature measurement by a thermocouple is ±3°ë.

Fig. 2. Curve of pressure calibration at room temperature: cell pressure (P, GPa) versus oil pressure in the hydraulic ram (p, atm).

20−40 min, the Ca carbonate melt may penetrate into the sample within the heater and contaminate the growth melt (from a few tenths to several weight percent ë‡ëé3). Contamination was not observed at durations of less than 10−15 min. The contamination was eliminated by using MgO or ZrO2 protective sleeves (Fig. 1).

500

After melting and crystallization of diamond and graphite, a characteristic zoning develops in the samples. Figure 4b presents a distinctly zoned sample with monocrystalline graphite in the center and diamond in the end zones. The boundaries between the zones are sharp and reflect the distribution of temperature influence on the sample and characteristics of the nonuniform temperature field. The lines in the image show schematic distribution of temperature along the sample axis, whereas the uniformity of temperature in lateral sections results in the uniform distribution of diamond grains (Fig. 4c). The highest temperatures are characteristic of the central section, and the vectors of temper2008

536

LITVIN et al. (‡)

(b)

(c)

[D] [G] [D] [G] [D] G SEM MAG: 1.00 kx HV: 20.0 kV VAC: HiVac

DET: SE Det + BSE Det DATE: 03/13/03 Device: MV2300

50 µm

SEM MAG: 67 x HV: 20.0 kV VAC: HiVac

(d)

DET: BSE De + SE Dete DATE: 01/15/03 Device: MV2300

1 mm


(e)

DET: BSE Detector DATE: 03/13/03 Device: MV2300

1 mm

(f)

L

(111) D

L

G (100) SEM MAG: 1.00 kx DET: BSE De + SE Dete HV: 20.0 kV DATE: 03/13/03 VAC: HiVac Device: MV2300

200 µm

SEM MAG: 3.00 kx HV: 20.0 kV VAC: HiVac


20 µm

SEM MAG: 2.00 kx DET: SE Detector HV: 20.0 kV DATE: 02/19/04 VAC: HiVac Device: MV2300

50 µm

Fig. 4. Experimental samples after quenching of diamond-forming carbonate–silicate–carbon and silicate–carbon melts. (a) Polycrystalline graphite (black particles) and a carbonate–silicate mixture (light gray) under subsolidus conditions, sample 1017, composition [Ecl50(K2CO3)50]60C40, 7.0 GPa, 1000°ë. (b) Zoning of the temperature field. The vectors of temperature gradient are shown by arrows and directed away from the central section of the sample (solid line) toward the end surfaces. They intersect the boundary sections (dashed lines) between the central zones, where graphite crystallizes (denoted as [G]), and marginal zones, where diamond crystallizes (denoted as [D]). Dark crystals of graphite and diamond occur within the carbonate–silicate melt, which is quenched to a light-colored material; sample 965, composition [Ecl50(K2CO3)50]60C40, 7.7 GPa, 1800°ë. (c) Lateral cross-section of a sample at a level of 30% (see text) showing uniform distribution of diamond grains (dark) in a quenched carbonate–silicate melt (light); sample 1009, composition [Ecl50(K2CO3)50]60C40, 8.5 GPa, 1720°ë. (d) Sample with efficient nucleation and extensive crystallization of diamond in a carbonate–silicate–carbon melt; sample 965, composition [Ecl50(K2CO3)50]60C40, 7.7 GPa, 1800°ë. (e) Sample with extensive crystallization of thermodynamically unstable monocrystalline graphite (black plates and spherules) in a carbonate–silicate–carbon melt; sample 965, composition [Ecl50(K2CO3)50]60C40, 7.7 GPa, 1800°ë. (f) Surface of a seed cuboctahedral diamond crystal after growth of newly formed diamond layers in a silicate–carbonate–carbon melt: (1) stepped layer-by-layer growth on the (111) octahedral face; the steps are decorated by white-colored remnants of the quenched growth medium; (2) rough micropyramidal growth on the (100) cubic face, represented by a dark grained material also with remnants of the growth medium in the right lower part of the photo; sample 1142, composition [Ab70(K2CO3)30]60C40, 8.5 GPa, 1740°ë.

ature gradient are directed in opposite directions from this section (distance between the center and the rim of a sample is about 2 mm). Graphite crystallization and diamond growth on seeds in the central zone suggest metastable solution supersaturation with respect to diamond. This zone is 0.6–0.7 mm in size, i.e., it occupies about 30% of the distance between the center and the rim of the sample, which is comparable with the size of a thermocouple junction. Then, the zone of spontane-

ous diamond nucleation is observed. For sample no. 965, the temperature gradient vector intersects the boundary between the fields of labile (LSF) and metastable supersaturation (MSF) (Fig. 5), such that the temperature in the center corresponds to the MSF and that at the sample ends corresponds to the LSF. In various experiments, the LSF/MSF boundary moves along the sample axis with changes in temperature, and in the limiting cases, either diamond or graphite crystallizes


Vol. 46

No. 6

2008


throughout the volume of the sample. Taking into account these phenomena, the material for investigations was always collected from the aforementioned 30% level. The character of extensive diamond crystallization at this level is shown in Fig. 4c. The high intensity of diamond crystallization in the systems studied is illustrated by Figs. 4c and 4d, which also show the purely octahedral habit of diamond crystals (cubic faces did not develop). Figure 4e exhibits characteristic monocrystalline platy and spherulitic forms of crystallization of thermodynamically unstable graphite, which are significantly different from those of starting graphite shown in Fig. 4a. In methodical aspects, the use of seeds, cuboctahedral monocrystals of diamonds synthesized in melts of the Mn–Ni–C system, was informative for the characterization of melt capacity with respect to diamond growth. This stems from the fact that diamonds synthesized in metal systems have both octahedral (111) and cubic (100) faces, which are smooth owing to stepped layer-by-layer growth. In contrast, diamond synthesized in carbonate systems have only octahedral (111) faces, similar to natural diamonds, and stepped layerby-layer growth occurs on the (111) faces only. The investigation of new layers on the (111) and (100) faces of cuboctahedral seeds synthesized in metal systems revealed significant differences. The (111) faces showed smooth-faced stepped layer-by-layer growth (Fig. 4f), whereas rough micropyramidal growth patterns were observed on the (100) faces: rectangular growth micropyramids (half-octahedra) with axes perpendicular to the cubic face formed continuous layers. Such a characteristic transformation of initially smooth (100) faces to rough surfaces covered with growth micropyramids is an indicator of diamond growth on the seed [35]. Note that seeds do not trigger the nucleation and extensive crystallization of diamond; these processes occur spontaneously in carbon solution showing labile supersaturation with respect to diamond. Experimental samples were investigated using a CamScan M2300 (Vega TS 5230MM) electron microscope equipped with a Link INCA Energy analytical energy-dispersive system at the Institute of Experimental Mineralogy, Russian Academy of Sciences. Raman spectra were recorded on an RM1000 (Renishaw) spectrometer equipped with a Leica microscope at the same Institute. The spectra were excited by a 514.5 nm argon ion laser (Melles Griot). The excitation spot was 5 µm in size. EXPERIMENTAL RESULTS Experiments on the investigation of the efficiency of diamond-forming media with variable compositions were begun from two model multicomponent eclogite– carbonate systems (Table 1). The silicate component is represented by the synthetic omphacite–garnet eclogite Cpx60Grt40 (wt %), where Cpx = Di50Jd50 and GEOCHEMISTRY INTERNATIONAL

Vol. 46

No. 6

8.5

2

537

Diamond 5 4

3

7.5 1

6.5 LSF Graphite

5.5 MSF

4.5 1100

1300

1500

1700

1900

2100

2300

Fig. 5. P–T diagram of diamond crystallization with boundary conditions: 1, diamond–graphite equilibrium boundary [33]; 2, boundary of the eutectic melting of a K–Na–Mg– Ca–Fe-carbonatite sample with graphite [34]; 3, boundary curve between the regions of labile and metastable supersaturation (LSF and MSF, respectively) with respect to diamond for carbon solutions in K–Na–Mg–Ca–Fe carbonatite melt [34]; 4, LSF/MSF boundary for the composition Ecl50(K2CO3)50; and 5, LSF/MSF boundary for the composition Ab50(K2CO3)50.

Grt = Gros60Prp40 (mol %), and the carbonate part is either a magnesite–calcite mixture with the stoichiometry of dolomite, CaMg(CO3)2, or alkali carbonate K2ëO3. The choice of these carbonate compounds was dictated by the fact that Mg, Ca, and K carbonate components are predominant in primary carbonatite inclusions in diamond [3]. Dolomite is unstable at the depths of diamond formation [36], and the magnesite, MgCO3, and aragonite, CaCO3, end-members are probably the major components of mantle carbonatite melts, whose genetic relation with the parent diamond melts cannot be ruled out [1]. Of special interest is also the opportunity to compare the role of carbonate components in diamond-forming systems in two variants, alkaline and alkali-free. In this paper, experimental information on the conditions and mechanisms of crystallization of solid carbon phases, diamond and graphite, was analyzed employing the P–T diagram of the boundary conditions of diamond formation (Fig. 5). The P–T diagram of diamond crystallization reported in [34] for a multicomponent K–Na–Ca–Mg–Fe carbonate–carbon melt (mixture 5 in Table 1) was used as a reference diagram. The selected composition approaches most closely the composition of the carbonate component of carbonate–silicate inclusions in natural diamonds from kimberlites [3, 37]. This diagram also displays additional lines illustrating some of the results of our study. The questions of the boundary conditions of diamond formation in various systems are fundamentally important for the interpretation of experimental and mineralogical data; therefore, it is expedient to define 2008

538

LITVIN et al.

Table 2. Conditions and results of experiments on diamond crystallization in the model eclogite system Cpx60Grt40– CaMg(CO3)2 (dolomite) Run no. 1482 1488 992 1007 1018 964 967 1056 1053 1052 993 968 1020

Component composition Pressure, GPa of solvent, wt % Ecl30Dol70 Ecl40Dol60 Ecl50Dol50 Ecl50Dol50 Ecl50Dol50 Ecl50Dol50 Ecl50Dol50 Ecl60Dol40 Ecl60Dol40 Ecl60Dol40 Ecl70Dol30 Ecl70Dol30 Ecl70Dol30

Diamond crystallization

Temperature, °C

Duration, min

1750 1750 1800 1700 1710 1800 1820 1700 1720 1570 1800 1780 1570

10 10 40 42 40 40 45 4 14 40 40 40 40

8.5 8.5 8.5 8.5 7.8 7.8 7.6 8.5 8.5 8.5 8.5 7.8 7.8

preliminarily some relevant concepts and terms. The P−T boundary conditions (Fig. 5) include the parameters of (1) the graphite–diamond equilibrium curve [33] (independent of the composition of the diamond-forming system), (2) the curve of the pressure dependence of the solidus temperature of the diamond-forming system (sensitive to its composition), and (3) the boundary curve between the fields of metastable supersaturation (MSF) and labile solutions (LSF) [38], which is controlled by the activation energy of spontaneous diamond nucleation (depends on the composition of the system). The main physicochemical difference between these fields is that, under the conditions of MSF, supersaturation with respect to diamond in carbonate–carbon melt–solutions is sufficient to provide diamond growth on seeds (as well as nucleation and growth of the graphite phase, which is thermodynamically unstable under these conditions; this is a specific phenomenon of diamond-forming systems of the solution–melt type) but not sufficient for diamond nucleation. The degrees of supersaturation attained under LSF conditions are sufficient for the spontaneous formation of diamond nuclei. Physicochemical reasons for the appearance of the LSF and MSF in diamond-forming systems were discussed in detail elsewhere [12, 38, 39]. System Cpx60Grt40–CaMg(CO3)2–C The conditions and results of experiments with the model eclogite under reference conditions of 8.5 GPa and 1700–1800°ë are shown in Table 2. The comparison of results on diamond nucleation was feasible, because the carbonate–silicate system was completely molten under the experimental conditions (above the

spontaneous

on seeds

yes no no no no no no no no no no no no

yes yes yes yes yes yes yes yes yes yes yes yes yes

liquidus curve), and the compositions of melts were identical to the bulk composition of the system. Extensive diamond crystallization was observed at 8.5 GPa and 1750°ë (Figs. 6a, 6b) in the melts of the carbonate– silicate composition Ecl30Dol70, i.e., rather deep in the carbonatite part of the system, but was not detected at 8.5 GPa and 1750°ë in the composition Ecl40Dol60, as well as at 1700 and 1800°ë in the composition Ecl50Dol50. Flat hexagonal crystals of graphite and their intergrowth crystallized under these conditions (Figs. 6c, 6d). Newly grown diamond layers were observed on a seed crystal (Fig. 6e): intergrown rectangular pyramids on cubic faces (rough surface) and flat growth steps on octahedral faces (smooth-faced growth). The high mobility of low-viscosity carbonate– silicate melts is indicated by the characteristic vortex structures observed after quenching (Fig. 6f). At lower pressures, there was no diamond nucleation in the Ecl50Dol50 melts, and graphite crystallized as plates, corrugated intergrowths (Fig. 6g), and spherules. There was no diamond nucleation at high contents of the eclogite component (compositions Ecl60Dol40 and Ecl70Dol30) at 8.5 GPa, whereas extensive graphite crystallization was observed (Figs. 6h, 6i, 6k). This implies that carbon solubility in the silicate–carbonate melts is relatively high, but labile diamond supersaturation and, consequently, conditions necessary for diamond nucleation were not attained. The growth of diamond on seeds is indicated by the development of micropyramidal layers on the (100) face (Fig. 6j). Clinopyroxenes and garnets did not appear up to the composition Ecl70Dol30, because the system was in a completely molten state, but they were detected among the quench products of silicate–carbonate melts starting


Vol. 46

No. 6

2008


from the composition Ecl70Dol30 (Fig. 6l). This implies that the temperature of the liquidus curve increased with increasing content of silicate components, and the experimental conditions shifted from the field of homogeneous melt to the two-phase and three-phase fields of the liquidus crystallization of silicate minerals. In sample no. 1020, the average composition of grossular– pyrope solid solutions is close to Ca3.6Mg2.4Al2Si3O12, and that of clinopyroxene is Na0.23Ca0.77Mg0.73Al0.55Si1.72O6 (62% diopside endmember, 23% jadeite, and 15% Ca-Tschermak component). Under the conditions of total and partial melting, completely miscible carbonate–silicate melts are formed. During quenching, they produce complex heterogeneous and fragile dendritic carbonate–silicate aggregates, whose quantitative microprobe analysis is a difficult task. The semiquantitative investigation of the quench aggregates suggested that the compositions of diamond-forming carbonate–silicate melts are similar to those of the starting mixtures.

539

System Cpx60Grt40–K2CO3–ë

Thus, the concentration barrier of diamond nucleation in carbonate–eclogite melts at 8.5 GPa lies between 30 and 40 wt % of the eclogite component (35% was accepted), i.e., the spontaneous crystallization of diamond is possible only within the carbonatite segment of the dolomite–eclogite system. This is the case of the melt of the carbonate–silicate rock of the Kokchetav massif, which contains 70 wt % dolomite and 30 wt % silicate minerals (mainly, clinopyroxene and garnet) [19]. An increase in the content of silicate components in the melt inhibits diamond nucleation, gradually lowering its efficiency and eventually completely depressing it (mechanism of such an influence is related to a decrease in supersaturation levels from labile ones to the concentration barrier of diamond nucleation and further to the metastable region beyond the barrier). Diamond growth on seeds from metastable supersaturated solutions was observed within a considerable part of the silicate segment of compositions (up to 70 wt % silicate components).

The results of experiments with various compositions of the Cpx50Grt50–K2CO3–ë system are given in Table 3. In carbonate–silicate melts of the Ecl50(K2CO3)50 composition, intense spontaneous diamond crystallization was observed at 8.5 GPa and 1720°ë (Fig. 7a). An interesting example of diamond and graphite syngenesis can be seen in Fig. 7b, where an octahedral diamond crystal and a platy graphite grain were formed within a 20-µm area; in addition, there is a spinel twin of diamond. After diamond nucleation in the labile region and its growth, the degree of supersaturation decreased to the metastable level in the vicinity of the growing diamond crystal, where local nucleation and growth of thermodynamically unstable graphite occurred. It is interesting that thermodynamically unstable graphite was formed under the conditions of diamond stability in the same melt–solution of carbon. This scenario is appropriate for the explanation of the findings of inclusions of monocrystalline graphite in natural diamonds [43]. Extensive diamond crystallization could be observed at 8.0 GPa and 1800°ë in the Ecl50(K2CO3)50 composition (Fig. 7c). As pressure decreases to 7.0–7.5 GPa at 1500–1750°ë, diamond crystallization gives way to the spontaneous crystallization of graphite spherules (Fig. 7d), blocks, and plates, which is accompanied by diamond growth on seeds, stepped layer-by-layer on the (111) faces and rough pyramidal on the (100) faces. Thus, the LSF/MSF boundary for the Ecl50(K2CO3)50 composition lies within 7.5–8.0 GPa (Fig. 5), i.e., at much higher pressure than that for standard carbonatite. The nucleation and growth of diamond crystals are rather intense at 8.5 GPa and 1800°ë in the silicate–carbonate melt of the Ecl60(K2CO3)40 composition, which falls within the silicate-rich part of the system, i.e., outside the carbonatite segment (Fig. 7e). No spontaneous diamond nucleation was detected in the Ecl70(K2Cé3)30 and Ecl80(K2Cé3)20 compositions at 7.5–8.5 GPa, but graphite crystallized (Fig. 7f) and diamond grew on seeds under these conditions.

Carbonate–silicate melts are completely miscible under the P–T conditions of diamond stability [40], which is indicated, in particular, by their quench products (e.g., Figs. 6a–6c, 6h). A single liquid phase (carbonate–silicate melt) is formed. The proportions of carbonate and silicate components in it may vary, which affects the solubility of carbon in the melt and has serious consequences for diamond formation. The effect of complete carbonate–silicate liquid miscibility in parent diamond-forming melts provides meaning to the concept of the concentration barrier of nucleation. The effects of carbonate–silicate liquid immiscibility were detected both in experimental [41] and natural systems [42] at pressures below 2.56 Pa, but there is no ground for the assumption that these phenomena are significant for the processes of diamond formation [3].

Thus, the concentration barrier of diamond nucleation at 8.5 GPa is constrained between the Ecl60(K2CO3)40 and Ecl70(K2Cé3)30 compositions (we accepted a value of 65 wt % of silicate components), i.e., at higher contents of silicate components in silicate–carbonate melts. This is the first experimental example when the composition of diamond-forming liquid falls within the field of silicate–carbonate (basic) compositions with a considerable amount of silicate components. An increase in the content of silicate components is accompanied by a decrease in carbon solubility in silicate–carbonate melts and attainment of critically low labile supersaturation for diamond nucleation on the concentration barrier. Note that alkaline K carbonate is more favorable for diamond nucleation than Ca–Mg carbonate.


Vol. 46

No. 6

2008

540

LITVIN et al. (‡)

(b)

(c)

L

L L

G D D SEM MAG: 1.00 kx DET: SE Detector HV: 20.0 kV DATE: 03/22/06 VanKV Device: Vega TS5130MM

100 µm

SEM MAG: 3.20 kx DET: SE Detector HV: 20.0 kV DATE: 03/22/06 VanKV Device: Vega TS5130MM

(d)

20 µm

SEM MAG: 3.00 kx DET: BSE De + SE Det HV: 20.0 kV DATE: 03/15/06 Nekrasov Device: Vega TS5130MM

(e)

10 µm

(f)

(111)

L G

L SEM MAG: 3.00 kx DET: SE Det + BSE Dete HV: 20.0 kV DATE: 04/04/03 VAC: HiVac Device: MV2300

20 µm

L (100)


DET: SE Detector DATE: 04/04/03 Device: MV2300

200 µm


DET: BSE De + SE Det DATE: 01/15/03 Device: MV2300

100 µm

Fig. 6. Crystallization of diamond and graphite in the carbonate–silicate–carbon system Cpx60Grt40–CaMg(CO3)2–C. (a) and (b) extensive diamond crystallization; composition Ecl30Dol70 sample 1482, 8.5 GPa, 1750°C. (c) Unstable monocrystalline graphite occuring in situ in a quenched melt; composition Ecl40Dol60, sample 1488, 8.5 GPa, 1750°ë. (d) Growth of platy monocrystals of unstable graphite and their intergrowths in a carbonate–silicate melt; composition Ecl50Dol50, sample 1007, 8.5 GPa, 1700°ë. (e) Face growth of diamond on a cuboctahedral seed, micropyramidal on the (100) cube faces and stepped layer-by-layer on the (111) octahedral faces; composition Ecl50Dol50, sample 1007, 8.5 GPa, 1700°ë. (f) Vortex forms of convective mixing in a carbonate–silicate melt; composition Ecl50Dol50, sample 967, 7.6 GPa, 1820°ë. (g) Extensive crystallization of unstable graphite; composition Ecl50Dol50, sample 964, 7.8 GPa, 1800°ë. (h) Graphite monocrystals in a quenched silicate–carbonate melt; composition Ecl60Dol40, sample 1052, 8.5 GPa, 1570°ë. (i) Corrugated intergrowths of platy monocrystals of unstable graphite; composition Ecl60Dol40, sample 1052, 8.5 GPa, 1570°ë. (j) Micropyramidal growth of diamond on the (100) face in a silicate–carbonate melt with dissolved carbon; composition Ecl60Dol60, sample 1052, 8.5 GPa, 1570°ë. (k) Extensive crystallization of unstable monocrystalline platy graphite in a silicate–carbonate melt; composition Ecl60Dol40, sample 1053, 8.5 GPa, 1720°ë. (l) Simultaneous (cogenetic) crystallization of garnet, clinopyroxene, and unstable graphite in a silicate–carbonate melt; composition Ecl70Dol30, sample 1020, 7.8 GPa, 1570°ë.

The fact of spontaneous diamond nucleation as a criterion of diamond formation efficiency of a growth medium allows us to consider the concentration barrier of nucleation for systems with variable compositions under standardized physical parameters as a chemical boundary condition (Fig. 8). For carbonate–silicate systems, it is convenient to correlate the concentration barrier with the limiting content of silicate components, when the minimum labile supersaturation necessary for

diamond nucleation is reached. It is known [12, 19, 34, 38, 39] that this corresponds to the P–T parameters of the LSF/MSF boundary (Fig. 5), i.e., physicochemical conditions on the concentration barrier are identical to those of the LSF/MSF boundary. Experimental investigations on the basis of the criteria of diamond nucleation and syngenesis of diamond and inclusions coupled with the determination of the concentration barriers of nucleation provide a fundamental basis for the


Vol. 46

No. 6

2008

EXPERIMENTAL CHARACTERIZATION OF DIAMOND CRYSTALLIZATION (g)

(h)

541

(i) G

L L

G

L G



25 µm

SEM MAG: 4.00 kx DET: SE Det + BSE Det HV: 20.0 kV DATE: 04/04/03 VAC: HiVac Device: MV2300

(j)


(k)

20 µm

(l)

G

(111)

L

20 µm

G

(100)

L

Grt

G

L


50 µm


20 µm



10 µm

Fig. 6. Contd.

determination of chemical boundaries of natural parent media.

System Cpx60Grt40–model K–Na–Mg–Ca–Fe carbonatite–C

The speciation of carbon in carbonate–silicate melts during the formation of diamond can be indirectly characterized by the presence of microscopic graphite and diamond crystals detected by micro-Raman spectroscopy (Figs. 9a–9c) in homogeneous areas of quenched melts free of solid carbon phases (according to the data of electron microscopy). Such microscopic phases are formed under the conditions of rapid quenching (300°C/s) of carbonate–silicate melts nearly saturated in carbon, when the degree of supersaturation with respect to diamond increases sharply. This is in agreement with the concept of the atomic and cluster–atomic state of dissolved carbon. Quench microscopic phases of solid carbon were also detected in experiments on the investigation of carbon isotope fractionation during diamond formation [44].

This system is interesting as a proxy for the compositions of natural mafic carbonate–silicate parent media. The multicomponent composition of the carbonate constituent corresponds to the limiting carbonatite composition of carbonate–silicate inclusions in natural diamonds, which was calculated in [3]. This composition was experimentally studied and proved to be very efficient with respect to diamond formation [2, 34, 39]. Carbonatite composition with and without iron carbonate were used (mixtures K3 and K5 in Tables 1 and 4). At 8.5 GPa and 1700–1800°ë (Table 4), intense diamond nucleation was observed in the melts of the carbonatite segment of compositions up to Ecl40(K–Na–Mg–Ca–Fe carbonatite)60 (Figs. 10a, 10b). In contrast, diamond nucleation was inefficient in ironfree and iron-bearing melts with identical silicate–carbonate ratios, Ecl50(K–Na–Mg–Ca carbonatite)50 and Ecl50(K–Na–Mg–Ca–Fe carbonatite)50 (Figs. 10c, 10d).


Vol. 46

No. 6

2008

542

LITVIN et al.

Table 3. Conditions and results of experiments on diamond crystallization in the model eclogite system Cpx60Grt40–K2CO3 Run no. 1009 965 970 1013 1014 1017 1489 1010 1000 1019 966 969 1125

Component composition Pressure, GPa Temperature, °C Duration, min of solvent, wt % Ecl50(K2CO3)50 Ecl50(K2CO3)50 Ecl50(K2CO3)50 Ecl50(K2CO3)50 Ecl50(K2CO3)50 Ecl50(K2CO3)50 Ecl60(K2CO3)40 Ecl70(K2CO3)30 Ecl70(K2CO3)30 Ecl70(K2CO3)30 Ecl70(K2CO3)30 Ecl70(K2CO3)30 Ecl80(K2CO3)20

8.5 7.7 7.5 7.5 7.0 7.0 8.5 8.5 8.1 7.7 7.7 7.5 8.5

1720 1800 1800 1650 1500 1000 1800 1720 1800 1700 1800 1800 1800

Diamond did not nucleate, and only graphite crystallization was observed in the Ecl60(K–Na–Mg–Ca–Fe carbonatite)40 and Ecl70(K–Na–Mg–Ca–Fe carbonatite)30 compositions (Figs. 10e, 10f). Thus, all of the efficient diamond-forming compositions and, correspondingly, the point of the concentration limit of diamond nucleation are confined to the carbonatite segment. The concentration barrier of diamond nucleation corresponds to the Ecl40(K–Na–Mg–Ca–Fe carbonatite)60 composition (Fig. 8). Note that the multicomponent carbonatites are very favorable for diamond nucleation and may even provide avalanche-like formation of diamondites [1, 2, 14, 34]. This indicates that the inhibiting role of the silicate component is a powerful factor, and natural carbonatite melts may lose their high efficiency with respect to diamond formation in response to an increase in the content of silicate components. The investigation of systems of such a type, multicomponent and sufficiently representative with respect to the carbonatite and silicate constituents and with the compositions closely approaching the probable growth media of natural diamonds, is very important for the outlining of the compositions of variable natural parent diamond-forming media, which, as becomes obvious, are bounded by the concentration barriers of diamond nucleation. Model System Albite NaAlSi3O8–ä2ëé3–ë Our experimental approach allows us to combine the investigation of the inhibiting role of silicate components with the search for diamond-forming silicate– carbonate melts whose compositions are dominated by silicate components. The system albite NaAlSi3O8– ä2ëé3–ë was chosen as a model object, because its carbonate component is the most representative in carbonatite inclusions in natural diamonds [3, 6] and it was used in a number of experimental studies [9, 11]. Under

Diamond crystallization spontaneous

on seeds

yes yes no no no no yes no no no no no no

yes yes yes yes yes no yes yes yes yes yes yes yes

40 40 40 40 45 40 2 40 40 40 30 40 18

the P–T conditions of diamond formation, albite is unstable and is represented in melts by the jadeite (NaAlSi2O6) and silica (SiO2) components of the eclogite association. The conditions and results of experiments are shown in Table 5. At 8.5 GPa, all compositions of the carbonatite series are highly efficient in terms of diamond nucleation, up to the boundary composition (NaAlSi3O8)50(K2CO3)50 (Figs. 11a, 11b). Figure 10c shows an example of simultaneous graphite and diamond formation. In contrast to the case described above (Fig. 7b), intergrowth of diamond (left part of the sample showing flat faces) with a graphite hemisphere (right part of the sample showing rounded outlines) is observed. Figures 11d and 11e show a seed crystal before and after chemical cleaning, respectively. The character of stepped layer-by-layer growth on the (111) face suggests high rate of carbon transfer under the conditions of high supersaturation, when the concentration of carbon in the flow fluctuated up to labile supersaturation, which was sufficient for the nucleation and growth of small diamond crystals of octahedral shapes, as well as spinel and, occasionally, polysynthetic (fivepointed in shape) twins (Fig. 11d). These microscopic crystals adhered in an oriented way to the growing (111) face of the seed crystal (in Fig. 11d, this face is in part covered by a solid layer, which formed during quenching of carbonate–silicate growth melt). Rough micropyramidal layers were formed on the (100) face of the seed, which is characteristic of the whole range of silicate–carbonate melts (Fig. 11f). With decreasing pressure, the extensive spontaneous crystallization of diamond was observed in the melts of the (NaAlSi3O8)50(K2CO3)50 composition up to 7.8 GPa (Fig. 11g), but extensive spontaneous crystallization of unstable graphite was observed at 7.6 GPa


Vol. 46

No. 6

2008

EXPERIMENTAL CHARACTERIZATION OF DIAMOND CRYSTALLIZATION (‡)

(b)

543

(c) D

L

D G

L D

L

D SEM MAG: 1.00 kx HV: 20.0 kV VanKV


50 µm

SEM MAG: 4.00 kx DET: SE Det + BSE Det HV: 20.0 kV DATE: 03/13/03 VanKV Device: MV2300

(d)

20 µm

SEM MAG: 2.00 kx DET: BSE De + SE Det HV: 20.0 kV DATE: 01/15/03 Nekrasov Device: MV2300

(e)

(f)

G

G D L



25 µm

L

L

50 µm

SEM MAG: 2.00 x HV: 20.0 kV Nekrasov

DET: BSE De + SE Det DATE: 03/15/06 Device: Vega TS5130MM

25 µm



25 µm

Fig. 7. Crystallization of diamond and graphite in the silicate–carbonate–carbon system Cpx60Grt40–K2CO3–C. (a) Extensive diamond crystallization in a carbonate–silicate melt; composition Ecl50(K2CO3)50, sample 1009, 8.5 GPa, 1720°ë. (b) Simultaneous crystallization of diamond and thermodynamically unstable graphite; composition Ecl50(K2CO3)50, sample 1009, 8.5 GPa, 1720°ë. (c) Monocrystalline diamonds in a quenched carbonate–silicate melt; composition Ecl50(K2CO3)50, sample 965, 7.7 GPa, 1800°ë. (d) Spherules of unstable graphite in a quenched carbonate–silicate melt; composition Ecl50(K2CO3)50, sample 1013, 7.5 GPa, 1650°ë. (e) Extensive crystallization of diamond in a silicate–carbonate melt; composition Ecl60(K2CO3)40, sample 1489, 8.5 GPa, 1800°ë. (f) Extensive crystallization of unstable graphite in a silicate–carbonate melt; composition Ecl70(K2CO3)30, sample 1010, 8.5 GPa, 1720°ë.

(Fig. 11h). At 8.2 and 8.5 GPa, the nucleation and growth of diamond occurred in the (NaAlSi3O8)70(K2CO3)30 and (NaAlSi3O8)80(K2CO3)20 compositions, when cocrystallization of diamond with jadeite and coesite could be observed (Figs. 11i, 11j). The nucleation and growth of diamond crystals was also detected in the (NaAlSi3O8)90(K2CO3)10 composition (Fig. 11k). The simultaneous nucleation of diamond and graphite phases (Fig. 11l) is indicative of the minimum level of labile supersaturation coinciding with the maximum level of metastable supersaturation (consequence of spatial heterogeneity in the supersaturated state of the carbon solution). This effect suggests the proximity of the concentration barrier of diamond GEOCHEMISTRY INTERNATIONAL

Vol. 46

No. 6

nucleation; indeed, in melts of the (NaAlSi3O8)95(K2CO3)05 composition and in pure albite melt, only globules of unstable graphite were formed, and diamond grew on seeds. Thus, the concentration barrier of diamond nucleation in the NaAlSi3O8–K2CO3 system corresponds to 90–92 wt % of silicate components (Fig. 8). The obtained results show that alkaline aluminosilicate–carbonate growth media with high contents of alkaline aluminosilicate components may be highly efficient as diamond-forming media. Table 5 and Fig. 8 present the results of experiments on diamond crystallization at a standard for this study pressure of 8.5 GPa in the systems Mg2SiO4–K2CO3–graphite and SiO2–K2CO3– 2008

544

LITVIN et al.

Ab + KC

(92)

Ecl + KC

(65)

Ecl + KC

500

(65)

Fo + KC

400

(50)

Fo + KC SiO2 + KC

Intensity (‡) 600

300

(45)

200 100

(25)

Ecl + Carb Ecl + Dol

0

(40)

0

(35)

1000

2000

3000

4000

(b) Carb

20

40

60

80

Sil

Fig. 8. Position of the concentration barrier of diamond nucleation depending on the chemical composition of the growth carbonate–silicate–carbon melt at 8.5 GPa and 1700–1800°ë in the systems eclogite (Cpx60Grt40)–dolomite (CaMg(CO3)2)–carbon (C), eclogite Cpx60Grt40–ä carbonate (K2CO3)–carbon (C), eclogite Cpx60Grt40–KNa-Ca-Mg-(±Fe) carbonatite–carbon (C), albite NaAlSi3O8–ä carbonate (ä2ëé3)–carbon (C), forsterite– K carbonate (ä2ëé3)–carbon (C), and silica SiO2–ä carbonate (ä2ëé3)–carbon (C) (solid and dashed lines correspond to 8.5 and 6.3 GPa, respectively [9]). Numbers in parentheses show the content of silicate components in carbonate–silicate melts on respective concentration barriers of nucleation. Abbreviations: KC, ä2ëé3 and MCC, multicomponent K–Na–Ca–Mg–(±Fe) carbonatite.

2000 1500 1000 500 0 0

1000

2000

3000

(c)

graphite (which were previously investigated at lower pressures [9, 10]) compared with other carbonate–silicate compositions considered in this study. Intense nucleation and crystallization of diamond was observed in the silicate–carbonate segment for the (Mg2SiO4)60(K2CO3)40 composition, i.e., within the field of silicate–carbonate compositions (concentration barrier of nucleation is located at about 65 wt % of the forsterite component). Unstable graphite nucleated in the boundary melts of the (SiO2)50(K2CO3)50 composition, i.e., the composition of the concentration barrier of diamond nucleation lies within the carbonatite segment of carbonate–silicate compositions (about 45 wt % SiO2). It can be seen in Fig. 8 that the concentration barrier is shifted with increasing pressure to compositions enriched in silicate or oxide components. THE PROCESSES OF DIAMOND FORMATION UNDER MANTLE CONDITIONS Cogenetic (growth) inclusions in diamonds provide insight into the chemistry of the mantle diamond-forming medium. Carbonatite melts transformed during isochoric cooling to the association of carbonates, sili-

4000 3000 2000 1000 0 0

1000

2000

3000 4000 Wavenumber, cm–1

Fig. 9. Raman spectra of quenched carbonate–silicate–carbon growth media. (a) Characteristic bands of diamond at 1336 cm–1 and graphite at 1578 cm–1; composition Ecl70(K2CO3)30, sample 1010, 8.5 GPa, 1720°ë. (b) Characteristic bands of diamond at 1351 cm–1 and graphite at 1585 cm–1; composition Ecl70(K2CO3)30, sample 1010, 8.5 GPa, 1720°ë. (c) Characteristic bands of diamond at 1335 cm–1 and graphite at 1580 cm–1; composition Ecl50(K2CO3)50, sample 1013, 7.5 GPa, 1650°ë.


Vol. 46

No. 6

2008


545

Table 4. Conditions and results of experiments on diamond crystallization in the model eclogite system Cpx60Grt40–K-NaCa-Mg-Fe carbonatite (mixtures K3 and K5 in Table 1) Run no. 1496 1483 1127 1491 1068 1112 1069 1074

Component composition Pressure, GPa Temperature, °C Duration, min of solvent, wt % Ecl30(mixture K3)70 Ecl40(mixture K5)60 Ecl50(mixture K5)50 Ecl50(mixture K5)50 Ecl50(mixture K5)50 Ecl50(mixture K5)50 Ecl60(mixture K5)40 Ecl70(mixture K5)30

8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5

1770 1780 1800 1770 1700 1760 1780 1780

cates, phosphates, oxides, sulfides, chlorides, water, and carbon dioxide [3, 6, 45, 46] and various silicate minerals of ultramafic and mafic suites were documented as inclusions in diamond [4, 47, 48]. Carbonatite material was detected in a single inclusion with silicate minerals [49]. Sulfide inclusions are widespread [50–53]; carbonates [54–56] and aqueous K and Na chloride brines were reported [57]. The mineralogy of cogenetic inclusions unambiguously constrains the general chemical character of the parent medium of natural diamonds and growth inclusions in them. The chemical model of the parent medium can be approximated by the multicomponent system MgO–CaO–FeO (Fe2O3)–MnO–Na2O–K2O–Al2O3–Cr2O3–TiO2–ZrO2– SiO2–P2O5–CuS (Cu2S)–FeS (FeS2)–NiS–KCl–NaCl– CO2–H2O–C. The natural system is polyphase and shows considerable variations in bulk composition and the compositions of mineral phases, which include compounds of diverse chemical nature: Mg, Ca, Fe, Ti, Al, and Si oxides; Mg, Fe, and Ca silicates; Na, K, Mg, Ca, and Fe aluminosilicates; Fe, Ni, and Cu sulfides; Ca phosphates; Mg, Ca, Fe, K, and Na carbonates; K and Na chlorides; Si carbides; water; carbon dioxide; occasionally, native iron, methane, etc. The diversity of the chemical and phase composition of parent media does not allow us to unequivocally determine the composition of material responsible for the formation of the majority of natural diamonds on the basis of mineralogical data only. The difficulty arises because this information cannot be used to determine which of the chemically different substances in inclusions are efficient in diamond formation, and, correspondingly, the physicochemical mechanism of diamond formation under natural conditions cannot be fully understood. Experiments at high pressures and temperatures can be used to asses the efficiency of diamond formation in any medium, including those compositionally similar to primary inclusions in natural diamonds. The experimental criterion of diamond-forming efficiency is the fact of spontaneous diamond nucleation in the given medium during its testing as a melt with dissolved carGEOCHEMISTRY INTERNATIONAL

Vol. 46

No. 6

25 25 40 25 5 25 40 39

Diamond crystallization spontaneous

on seeds

yes yes no no no no no no

yes yes yes yes yes yes yes yes

bon under the P–T conditions of diamond stability. The experience of experimental investigations suggests that this criterion is necessary but not sufficient for the unambiguous solution of the problem of parent media for natural diamonds. The main uncertainty lies in the fact that the experiments identified several chemically different media associating with natural diamonds that are efficient in diamond formation. Among them are simple and double carbonates, multicomponent carbonate and carbonate–silicate melts with dissolved carbon [1, 2, 10, 34], sulfide–carbon melts [5, 24–27], chloride–carbon melts [58], C–O–H fluid phases (e.g., dense ç2é and ëé2 [10, 59–64]), silicate–carbon melts [65], and other substances. Nonetheless, although the experimental criterion of nucleation does not close the debate, it provides a basis for the more consistent consideration of the problem of the chemical and phase composition of the main diamond-forming media under mantle conditions [1, 10, 63, 64]. The experimental criterion of the syngenesis of diamonds and growth inclusions in them is more rigorous [1, 5]. According to this criterion, any parent medium must be similarly efficient in both diamond formation and coupled physicochemical processes producing numerous chemically different compounds observed in growth inclusions in diamonds. Using this criterion and experimental data on the immiscibility of sulfide melts with silicate, carbonate, and carbonate–silicate melts under the conditions of diamond formation, the idea of sulfide–carbon melts as the dominant diamond-forming media in the Earth’s mantle [21, 23, 53] was arguably rejected [1, 5, 27]. However, sulfide–carbon melts are efficient in diamond nucleation [24] and occur as immiscible admixture phases in the parent carbonate– silicate media, and they probably played a role, albeit minor, in diamond formation [5, 26, 27]. The criterion of syngenesis is at odds with the hypothesis of molten metals as dominant parent media for diamonds [66– 69]. The criteria of nucleation and syngenesis allow us to achieve consistency between experimental and mineralogical data and support the universal character of 2008

546

LITVIN et al. (‡)

(b)

(c) G

G L D D L L G SEM MAG: 1.00 kx HV: 20.0 kV VanKV

DET: SE Detector DATE: 03/22/06 Device: Vega TS5130MM

50 µm

SEM MAG: 2.50 kx DET: BSE De + SE Det HV: 20.0 kV DATE: 01/15/04 VanKV Device: MV2300

(d)

25 µm

SEM MAG: 4.00 kx DET: SE Det + BSE Det HV: 20.0 kV DATE: 05/16/03 Nekrasov Device: MV2300

(e)

(f)

G

G

20 µm

G L

L

L SEM MAG: 8.00 kx DET: SE Det + BSE Det HV: 20.0 kV DATE: 05/16/03 VAC: HiVac Device: MV2300

10 µm

SEM MAG: 12.00 kx HV: 20.0 kV Nekrasov


5 µm

SEM MAG: 4.00 kx DET: SE Det + BSE Det HV: 20.0 kV DATE: 05/16/03 VAC: HiVac Device: MV2300

20 µm

Fig. 10. Crystallization of diamond and graphite in the carbonate–silicate system Cpx60Grt40–K-Na-Ca-Mg-Fe carbonatite–C. (a) Extensive crystallization of unstable graphite; composition (Cpx60Grt40)70(K–Na–Ca–Mg–Fe carbonatite)30, sample 1069, 8.5 GPa, 1780°ë. (b) Spherule of unstable graphite; composition (Cpx60Grt40)70(K–Na–Ca–Mg–Fe carbonatite)30, sample 1069, 8.5 GPa, 1780°ë. (c) Extensive crystallization of unstable graphite; composition (Cpx60Grt40)50(K–Na–Ca–Mg–Fe carbonatite)50, sample 1068, 8.5 GPa, 1700°ë. (d) Elongated crystal of unstable graphite; composition (Cpx60Grt40)50(K–Na–Ca–Mg–Fe carbonatite)50, sample 1068, 8.5 GPa, 1700°ë. (e) Simultaneous crystallization of diamond and unstable graphite; composition (Cpx60Grt40)40(K–Na–Ca–Mg–Fe carbonatite)60, sample 1127, 8.5 GPa, 1800°ë. (f) Simultaneous extensive crystallization of diamond and unstable graphite; composition (Cpx60Grt40)40(K–Na–Ca–Mg–Fe carbonatite)60, sample 1483, 8.5 GPa, 1780°ë.

the carbonate–silicate model of parent media for the majority of diamonds [1, 58]. According to the carbonate–silicate model, the dominant parent media are multicomponent carbonate– silicate–carbon mantle melts, most of which are probably of carbonatite-like composition. It is usually believed that carbon for natural diamonds was supplied from possible carbon species in the host mantle [45]. Note that of direct significance for the genesis of diamond is elemental carbon dissolved in parent melt, which may have diverse sources. The parent media are chemically variable with respect to the main carbonate and silicate components of peridotite and eclogite

assemblages, which provide the major concentration contribution. Carbonate–silicate melts with dissolved carbon and variable composition are the dominant medium responsible for the formation of the majority of natural diamonds. Simultaneously, the carbonate– silicate melts are a kind of matrix for various admixture components, which are scavenged during melt formation and evolution from the multicomponent mantle material. With respect to concentration contributions and significance for diamond formation, the admixture components are subordinate, and their contents in the parent melts are strongly variable. Some admixture components, such as oxides, phosphates, chlorides,


Vol. 46

No. 6

2008


547

Table 5. Conditions and results of experiments on diamond crystallization in the model silicate (oxide)–carbonate systems albite (NaAlSi3O8)–K2CO3–C, forsterite (Mg2SiO4)–K2CO3–C, and silica (SiO2)–K2CO3–C Run no. 1113 1114 1115 1153 1116 1122 1123 1156 1156 1142 1157 1485 1145 1131 1143 1125 1486 1490 1494 1523 1524 1523 1527 1528 1495

Component composition Pressure, GPa Temperature, °C Duration, min of solvent, wt % Ab50(K2CO3)50 Ab50(K2CO3)50 Ab50(K2CO3)50 Ab50(K2CO3)50 Ab50(K2CO3)50 Ab60(K2CO3)40 Ab60(K2CO3)40 Ab60(K2CO3)40 Ab65(K2CO3)35 Ab70(K2CO3)30 Ab70(K2CO3)30 Ab70(K2CO3)30 Ab70(K2CO3)30 Ab70(K2CO3)30 Ab70(K2CO3)30 Ab80(K2CO3)20 Ab80(K2CO3)20 Ab90(K2CO3)10 Fo60K2CO3)40 Fo65(K2CO3)35 Fo70(K2CO3)30 Fo80(K2CO3)20 (SiO2)30(K2CO3)70 (SiO2)40(K2CO3)60 (SiO2)50(K2CO3)50

8.5 8.0 7.8 7.8 7.6 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.2 8.5 8.0 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5

1770 1780 1770 1800 1760 1710 1800 1770 1700 1740 1650 1760 1620 1800 1640 1800 1730 1740 1790 1800 1780 1720 1780 1780 1740

water, carbon dioxide, etc., are highly soluble in carbonate–silicate melts, whereas others, e.g., sulfides, metals, carbides, etc., can be classified as almost completely insoluble. The geochemical fate of admixture components depends on physicochemical transformations in the main carbonate–silicate parent medium. These transformations are related to the initial stages of their formation, which involve the material of the peridotite mantle, and the subsequent evolution, which is accompanied by contamination and fractional crystallization. Crystal fractionation is controlled by the main phase relations of the minerals and melts involved and can provide continuous transitions and genetic relations between garnet-peridotite and eclogite mineral assemblages in the parent media, similar to what may occur for the respective rocks of the Earth’s mantle [30]. Both major and admixture components are trapped by growing diamonds as minerals and melts. In terms of physical chemistry, the parent carbonate–silicate (carbonatite) and silicate–carbonate (ultraGEOCHEMISTRY INTERNATIONAL

Vol. 46

No. 6

30 25 25 25 25 25 30 35 7 25 40 6 20 4 40 18 6 10 10 10 10 10 10 10 10

Diamond crystallization spontaneous yes yes yes yes yes yes yes yes yes yes yes yes yes yes yes yes yes yes yes no no no yes yes no

on seeds yes yes yes yes yes yes yes yes yes yes yes yes yes yes yes yes yes yes yes yes yes yes yes yes yes

basic–basic) melts are characterized by complete liquid miscibility between their carbonate and silicate components. They are usually produced by the partial melting of mantle rocks, although their formation through the complete melting of parent media cannot be ruled out also. There is mineralogical evidence for the rather considerable variations in their compositions with respect to the carbonate–silicate ratio, silica content (from peridotite to eclogite assemblages), alkalinity (enriched in potassium or sodium, or alkali-free), and redox conditions (in addition to diamond, minor amounts of native metals, graphite, carbides, ëé2, and ëç4 were detected). Carbon is dissolved in carbonate and carbonate– silicate melts of any composition, probably in atomic and cluster–atomic forms, which is suggested by indirect observations and analyses of experimental products. This is also indicated by the data of micro-Raman spectroscopy and carbon isotope analyses of quenched carbonate phases [44]. 2008

548

LITVIN et al. (‡)

(b)

(c)

L G

L

D

D L

D SEM MAG: 1.00 kx HV: 20.0 kV VAC: HiVac

DET: SE Det + BSE Dete DATE: 13/08/03 Device: MV2300

100 µm



(d)

25 µm


(e)


25 µm

(f)

D L

(100) (111) D SEM MAG: 1.00 kx HV: 20.0 kV VAC: HiVac


50 µm


(111)


100 µm



100 µm

Fig. 11. Crystallization of diamond and graphite in the silicate–carbonate system albite–K2ëé3–C. (a) Extensive diamond crystallization in a carbonate–silicate melt; composition Ab50(K2CO3)50, sample 1113, 8.5 GPa, 1770°ë. (b) Spinel twins and intergrowths of octahedral diamond crystals; composition Ab50(K2CO3)50, sample 1113, 8.5 GPa, 1770°ë. (c) Intergrowth of diamond (left) and a semispherule of metastable graphite (right); composition Ab50(K2CO3)50, sample 1113, 8.5 GPa, 1770°ë. (d) The (111) face of a seed (before cleaning) with new stepped layers partly covered with remnants of quenched growth melt containing oriented spontaneously grown diamond crystals of octahedral habit, as well as spinel and polysynthetic twins (five-pointed star); composition Ab50(K2CO3)50, sample 1113, 8.5 GPa, 1770°ë. (e) The (111) face of a seed (after cleaning) with new stepped layers and adhered spontaneous octahedral diamond crystals; composition Ab50(K2CO3)50, sample 1113, 8.5 GPa, 1770°ë. (f) Rough micropyramidal growth on the (100) face; composition Ab60(K2CO3)40, sample 1122, 8.5 GPa, 1710°ë. (g) Extensive diamond crystallization in a carbonate–silicate melt; composition Ab50(K2CO3)50, sample 1153, 7.8 GPa, 1800°ë. (h) Extensive crystallization of metastable graphite in a carbonate–silicate melt; composition Ab50(K2CO3)50, sample 1116, 7.6 GPa, 1760°ë. (i) Extensive crystallization of diamond in a carbonate melt; composition Ab70(K2CO3)30, sample 1131, 8.5 GPa, 1800°ë. (j) Cogenetic crystallization of large coesite monocrystals, jadeite, and diamond in a silicate–carbonate melt; composition Ab80(K2CO3)20, sample 1125, 8.5 GPa, 1800°ë. (k) Extensive diamond crystallization in a silicate–carbonate melt; composition Ab90(K2CO3)10, sample 1490, 8.5 GPa, 1740°ë. (l) Cogenetic crystallization of diamond and unstable graphite in a silicate–carbonate melt; composition Ab90(K2CO3)10, sample 1490, 8.5 GPa, 1740°ë.

According to experimental data, carbonate–carbon and carbonate–silicate–carbon systems are stable buffer associations. At high pressures, carbonate melts congruently, and carbonate components are stable during the whole experiment and show no signs of decomposition with the liberation of carbon dioxide. The compositions of the carbonate–silicate–carbon systems include substances of very different redox activity, car-

bonates and solid carbon (graphite and diamond). The analysis of the material of quench melts showed that carbonate compounds and their melts occurring in a prolonged and continuous contact with elemental carbon (graphite or diamond) are not reduced by it, and carbon in the form of graphite, diamond, and carbon dissolved in the elemental form is not oxidized, i.e., there is no significant redox interaction between car-


Vol. 46

No. 6

2008

EXPERIMENTAL CHARACTERIZATION OF DIAMOND CRYSTALLIZATION (g)

(h)

549

(i)

L D

L G



50 µm



(j)

25 µm

L



(k)

Jd

D

50 µm

(l)

L G L D Coes

D

L SEM MAG: 6.00 kx HV: 20.0 kV VAC: HiVac


10 µm

SEM MAG: 4.00 kx DET: BSE De + SE Det HV: 20.0 kV DATE: 03/15/06 Nekrasov Device: Vega TS5130MM

20 µm

SEM MAG: 4.00 kx DET: BSE De + SE Det HV: 20.0 kV DATE: 03/15/06 Nekrasov Device: MV2300

20 µm

Fig. 11. Contd.

bonates and carbon. The reason is that buffer associations, for instance, carbonate–silicate–solid carbon, are formed and maintain the redox potential of diamond formation processes at the level of the Fe/FeO buffer. The parent media of diamonds are formed in experiments via the eutectic melting of carbonate–silicate material and generation of carbon solutions in carbonate–silicate melts, which show either labile or metastable supersaturation in diamond depending on P–T conditions. The supersaturated state of carbon solutions with respect to diamond is the reason for the nucleation of diamond and thermodynamically unstable graphite (as a kinetic phenomenon) and diamond growth on seeds. The formation of microscopic phases of graphite and diamond during quenching of experimental melts provides indirect evidence for the atomic (atomic–cluster) forms of dissolved carbon. The mechanisms of dissolution, solubility, and speciation of carbon in silicate–carbonate and carbonatite melts under the parameters of diamond formation were GEOCHEMISTRY INTERNATIONAL

Vol. 46

No. 6

not studied in detail. These problems were previously considered for silicate melts [70, 71]. It was shown that, at a pressure of 4.0 GPa, temperatures of 1500–1600°ë, and oxygen fugacity below the Fe–FeO buffer equilibrium, carbon is dissolved in a ferrobasalt melt both in 2– the elemental form and as the carbonate ion ( CO 3 ); however, the total solubility of all carbon species is not high and estimated as about 0.2 wt %. The formation of Si–C bonds was observed in silicate melts. The fraction 2– of CO 3 in silicate melts increases with increasing oxygen fugacity. Based on these observations, it is reasonable to suggest that in the case of silicate–carbonate melts, whose oxygen fugacity is higher than that of silicate melts, the possible existence of dissolved carbon in the elemental form is accompanied by an increase in 2– the fraction of CO 3 in such melts. The carbonate ion is formed at the expense of dissolved elemental carbon. The possible proportions of the concentrations of ele2008

550

LITVIN et al.

mental carbon and carbonate ions are not constrained and require special studies. The dissolved carbonate phase can also contribute to the formation of carbonate ions in the melt. If the concentration of hydrogen is sufficiently high in the sample surroundings (for instance, in the solid-phase cell of the high-pressure apparatus), part of carbon can be consumed for the formation of C−H bonds in the melt [70, 71], but these processes can be inhibited by relatively high oxygen fugacity values in the system. Thus, the possibility that diamond formation in the mantle is accompanied by the generation of C–O–H fluid components cannot be excluded; however, experimental results, including carbon isotope investigations, Raman spectroscopy, and electron microprobe investigations of quenched diamond-forming melts (which demonstrate the congruent character of carbonate melting), suggest that such fluid substances are subsidiary or admixture components in the composition of parent carbonate–silicate–carbon diamond-forming melts. The problems of the influence of minor components on the position of the concentration barriers of diamond nucleation acquired, thus, special significance, especially for multicomponent carbonate–silicate melts forming a basis for natural diamond parent media. Perhaps, the influence of oxides, phosphates, and chlorides, which may occur in minor concentrations, is not high. The most interesting are the soluble components of C–O–H fluids, whose contents in the parent carbonate–silicate melts may be as high as several weight percent. These problems were addressed in the experimental investigations of the simplified systems Mg2SiO4– H2O–C, SiO2–H2O–C, CaMgSi2O6–Na2C2O4–C, and NaAlSi2O6–Na2C2O4–C [10], which can be considered as limiting boundary compositions of natural multicomponent carbonate–silicate systems with dissolved minor fluid components. The ability to provide spontaneous diamond nucleation is a crucial property of natural parent melts, because diamond cannot be formed under natural conditions without spontaneous nucleation. The experimental criterion of nucleation is probably the only experimental way to determine the concentration barriers of diamond nucleation and, correspondingly, the natural boundaries of chemical composition for the concept of the parent diamond-forming medium. On the other hand, it is interesting that after the cessation of diamond nucleation beyond the concentration barrier, diamond may grow on seeds, and it cannot be ruled out that the seeded diamond growth can occur up to the boundary silicate–carbon melts. This allows us to suppose that, under the conditions of contamination or fractional crystallization, the point of the carbonate–silicate parent medium will inevitably move, and its composition may shift from the region of spontaneous diamond nucleation to the conditions of seeded growth, when diamond nucleation is impossible (transition from the LSF to MSF). The diamonds that were formed spontaneously may serve as seeds in the changed

medium showing a lower degree of supersaturation in carbon. The growth of diamonds continues but already beyond the concentration barrier in a parent silicate– carbonate–carbon melt relatively enriched in silicate components. Thus, it is possible that diamonds spontaneously formed in a carbonatite melt continue to grow as seeds in a basic (silicate–carbonate) melt under natural conditions. The diamond-forming carbonate–silicate melts and mantle carbonatites have probably a common origin. The genesis of mantle carbonatites is related to the partial melting of peridotite [72] or fluid–magma interaction of the plume–mantle type [73]. It is conceivable that future geochemical and experimental studies will result in that the concepts of the origin of mantle carbonatites and parent diamond-forming media will converge as concerned with closely related processes of mantle magmatism. CONCLUSIONS For the first time, diamond crystallization was experimentally studied at 8.5 GPa in carbon-bearing melts corresponding to mixtures of model eclogite with dolomite, K2CO3, and multicomponent K–Na–Ca–Mg–Fe carbonatites. The concentration barriers of diamond nucleation were determined for these melts as 35, 65, and 40 wt % of silicate components, respectively. At higher contents of silicate components, diamond may grow on seeds simultaneously with the nucleation of thermodynamically unstable graphite. In the melts of the albite–K2CO3 system, the concentration barrier of diamond nucleation at 8.5 GPa lies at 90–92 wt % of the albite component; and diamond grows on seeds in albite–carbon solution. Using experimental and mineralogical data, it was demonstrated that carbonate–silicate (carbonatite) melts are the dominant parent media of natural diamonds. The concept of the concentration barrier of diamond nucleation paves the way for the experimental quantification of the boundary compositions of natural parent diamond-forming media. ACKNOWLEDGMENTS This study was financially supported by Program no. P9 of the Presidium of the Russian Academy of Sciences “Investigations of Materials under Extreme Conditions,” the Russian Foundation for Basic Research (project nos. 05-05-64101, 05-02-17283, and 06-0564478), and international grant INTAS 05-10000087927 “Diamond and Graphite in Carbonatite Magmas.” REFERENCES 1. Yu. A. Litvin, “High Pressure Mineralogy of Diamond Genesis,” in Advances in High-Pressure Mineralogy, Ed. by E. Ohtani, Geol. Soc. Am. Spec. Pap. 421, 83–103 (2007).


Vol. 46

No. 6

2008

EXPERIMENTAL CHARACTERIZATION OF DIAMOND CRYSTALLIZATION 2. Yu. A. Litvin and V. A. Zharikov, “Experimental Modeling of Diamond Genesis: Diamond Crystallization in Multicomponent Carbonate–Silicate Melts at 5–7 GPa and 1200–1570°C,” Dokl. Akad. Nauk 372 (6), 808–811 (2000) [Dokl. Earth Sci. 372, 867–870 (2000)]. 3. M. Schrauder and O. Navon, “Hydrous and Carbonatite Mantle Fluids in Fibrous Diamonds from Jwaneng, Botswana,” Geochim. Cosmochim. Acta 58 (2), 761–771 (1994). 4. N. V. Sobolev, Deep-Seated Inclusions in Kimberlites and the Problem of the Composition of the Upper Mantle, (Nauka, Novosibirsk, 1974; Am. Geophys. Union, Washington, 1977). 5. Yu. A. Litvin, A. V. Shushkanova, and V. A. Zharikov, “Immiscibility of Sulfide–Silicate Melts in the Mantle: Role in the Syngenesis of Diamond and Inclusions (Based on Experiments at 7.0 GPa),” Dokl. Akad. Nauk 402 (5), 656–660 (2005) [Dokl. Earth Sci. 402, 719–722 (2005)]. 6. O. Navon, E. S. Izraeli, and O. Klein-BenDavid, “Fluid Inclusions in Diamonds—the Carbonatitic Connection,” in Proceedings of 8th Int. Kimberlite Conference, Victoria, Canada, 2003 (Victoria, 2003). 7. M. Arima, K. Nakayama, M. Akaishi, et al., “Crystallization of Diamonds from a Silicate Melt of Kimberlite Composition in High-Pressure and High-Temperature Experiments,” Geology 21, 968–970 (1993). 8. Yu. M. Borzdov, A. G. Sokol, Yu. N. Pal’yanov, et al., “The Study of Diamond Crystallization from Alkaline Silicate, Carbonate, and Carbonate–Silicate Melts,” Dokl. Akad. Nauk 366 (4), 530–533 (1999) [Dokl. Earth Sci. 366, 578–581 (1999)]. 9. A. F. Shatskii, Yu. M. Borzdov, A. G. Sokol, and Yu. N. Pal’yanov, “Features of Phase Formation and Crystallization in Ultrapotassic Carbonate–Silicate Systems with Carbon,” Geol. Geofiz. 43 (10), 940–950 (2002). 10. Yu. N. Pal’yanov, A. G. Sokol, and N. V. Sobolev, “Experimental Modeling of Mantle Diamond-Forming Processes,” Geol. Geofiz. 46 (12), 1290–1303 (2005). 11. V. Yu. Litvin, Yu. A. Litvin, and A. A. Kadik, “Diamond Syntheses from Silicate–Carbonate–Carbon Melts at 6– 8.5 GPa: Limits of Diamond Formation and Forms of Dissolved Carbon,” Exp. Geosci. 11 (1), 28–31 (2003). 12. Yu. A. Litvin, “On Formation Mechanism of Diamond in Metal–Carbon Systems,” Izv. Akad. Nauk SSSR, Ser. Neorgan. Mater. 4 (2), 175–182 (1968). 13. Yu. A. Litvin and A. V. Spivak, “Rapid Growth of Diamondite at the Contact between Graphite and Carbonate Melt: Experiments at 7.5–8.5 GPa,” Dokl. Akad. Nauk 391 (5), 673–677 (2003) [Dokl. Earth Sci. 391A, 888–891 (2003)]. 14. G. Kurat and G. Dobosi, “Garnet and Diopside-Bearing Diamondites (Framesites),” Mineral. Petrol. 69, 143–159 (2000). 15. Yu. A. Litvin, A. P. Jones, A. D. Beard, et al., “Crystallization of Diamond and Syngenetic Minerals in Melts of Diamondiferous Carbonatites of the Chagatai Massif, Uzbekistan: Experiment at 7.0 GPa,” Dokl. Akad. Nauk 381 (4), 528–531 (2001) [Dokl. Earth Sci. 381, 1066– 1069 (2001)]. GEOCHEMISTRY INTERNATIONAL

Vol. 46

No. 6

551

16. A. V. Bobrov, Yu. A. Litvin, and F. K. Divaev, “Phase Relations and Diamond Synthesis in the Carbonate–Silicate Rocks of the Chagatai Complex, Western Uzbekistan: Results of Experiments at P = 4–7 GPa and T = 1200–1700°C,” Geokhimiya, No. 1, 49–60 (2004) [Geochem. Int. 42, 39–48 (2004)]. 17. Yu. A. Litvin, G. Kurat, and G. Doboshi, “Experimental Investigation of Diamondite Formation in Carbonate– Silicate Melts: Model Approximation to Natural Processes,” Geol. Geofiz. 46 (12), 1304–1317 (2005). 18. Yu. N. Palyanov, V. S. Shatsky, A. G. Sokol, et al., “Crystallization of Metamorphic Diamond: An Experimental Modeling,” Dokl. Akad. Nauk 380 (5), 671–675 (2001) [Dokl. Earth Sci. 381, 935–938 (2001)]. 19. Yu. A. Litvin, A. V. Spivak, and Yu. A. Matveev, “Experimental Study of Diamond Formation in the Molten Carbonate–Silicate Rocks of the Kokchetav Metamorphic Complex at 5.5–7.5 GPa,” Geokhimiya, No. 11, 1191– 1200 (2003) [Geochem. Int. 41, 1090–1098 (2003)]. 20. P. C. Marx, “Pyrrhotine and the Origin of Terrestrial Diamonds,” Mineral. Mag. 38, 636–638 (1972). 21. S. E. Haggerty, “Diamond Genesis in Multiply Constrained Model,” Nature 320, 34–38 (1986). 22. G. P. Bulanova, Z. V. Spetsius, and N. V. Leskova, Sulfides in Diamonds and Xenoliths from Kimberlite Pipes of Yakutia (Nauka, Novosibirsk, 1990) [in Russian]. 23. Z. V. Spetsius, “Two Generations of Diamonds in the Eclogite Xenoliths,” in Proceedings of 7th Int. Kimberlite Conference, Cape Town, South Africa, 1999 (Red Roof Design, South Africa, 1999), Vol. 2, pp. 823–828. 24. Yu. A. Litvin, V. G. Butvina, A. V. Bobrov, and V. A. Zharikov, “The First Synthesis of Diamond in Sulfide–Carbon Systems: The Role of Sulfides in Diamond Genesis,” Dokl. Akad. Nauk 382 (1), 106–109 (2002) [Dokl. Earth Sci. 382, 40–43 (2002)]. 25. Yu. N. Pal’yanov, Yu. M. Borzdov, I. Yu. Ovchinnikov, and N. V. Sobolev, “Experimental Study of the Interaction between Pentlandite Melt and Carbon at Mantle PT Parameters: Condition of Diamond and Graphite Crystallization,” Dokl. Akad. Nauk 392 (3), 388–391 (2003) [Dokl. Earth Sci. 392, 1026–1029 (2003)]. 26. Yu. A. Litvin and V. G. Butvina, “Diamond-Forming Media in the System Eclogite–Carbonatite–Sulfide–Carbon: Experiments at 6.0–8.5 GPa,” Petrologiya 12 (4), 426–438 (2004) [Petrology 12, 377–387 (2004)]. 27. A. V. Shushkanova and Yu. A. Litvin, “Phase Relations during Melting of Diamond-Forming Carbonate–Silicate–Sulfide Systems,” Geol. Geofiz. 46 (12), 1304– 1317 (2005). 28. D. L. Hamilton and C. M. B. Henderson, “The Preparation of Silicate Composition by a Gelling Method,” Mineral. Mag. 38 (282), 832–838 (1968). 29. L. G. Khvostantsev, L. F. Vereshchagin, and A. P. Novikov, “Device of Toroid Type for High Pressure Generation,” High Temp.–High Press., No. 9, 637–639 (1977). 30. Yu. A. Litvin, Physicochemical Studies of the Melting of Deep-Seated Materials of the Earth, (Nauka, Moscow, 1991) [in Russian]. 31. M. Eremets, High Pressure Experimental Methods (Oxford Univ. Press, New York, 1996). 32. H. T. Hall, “Fixed Points near Room Temperature,” in Accurate Characterization of the High Pressure Envi2008

552

33. 34.

35.

36.

37.

38. 39.

40.

41.

42.

43. 44.

45.

46.

LITVIN et al. ronment, Ed. by E. C. Lloyd (Nat. Bur. Stand., Washington, 1971), U.S. Spec. Publ., No. 326 (1971). C. S. Kennedy and G. C. Kennedy, “The Equilibrium Boundary between Graphite and Diamond,” J. Geophys. Res. 81 (14), 2467–2470 (1976). A. V. Spivak and Yu. A. Litvin, “Diamond Syntheses in Multicomponent Carbonate–Carbon Melts of Natural Chemistry: Elementary Processes and Properties,” Diamond Relat. Mater., No. 13, 482–487 (2004). Yu. A. Litvin, L. T. Chudinovskikh, and V. A. Zharikov, “Crystallization of Diamond in the Na2Mg(CO3)2– K2Mg(CO3)2–C System at 8–10 GPa,” Dokl. Akad. Nauk 359 (6), 818–820 (1998) [Dokl. Earth Sci. 359, 433–435 (1998)]. I. Martinez, J. Zhang, and R. J. Reeder, “In Situ X-Ray Diffraction of Aragonite and Dolomite CaMg(CO3)2 at High Temperature: Evidence for Dolomite Breakdown to Aragonite and Magnesite,” Am. Mineral. 81, 611–624 (1996). Yu. A. Litvin and V. A. Zharikov, “Primary Fluid–Carbonatite Inclusions in Diamond: Experimental Modeling in the System K2O–Na2O–CaO–MgO–FeO–CO2 as a Diamond Formation Medium at 7–9 GPa,” Dokl. Akad. Nauk 367 (3), 397–401 (1999) [Dokl. Earth Sci. 367, 801–805 (1999)]. Yu. A. Litvin, “On the Problem of Diamond Origin,” Zap. Vses. Mineral. O-va 98 (1), 116–123 (1969). Yu. A. Litvin and A. V. Spivak, “Growth of Diamond Crystals at 5.5–8.5 GPa in the Carbonate–Carbon Melt– Solutions as Chemical Analogues of Natural DiamondForming Media,” Materialovedenie, No. 3, 27–34 (2004). Yu. A. Litvin, V. Yu. Litvin, A. A. Kadik, and V. A. Zharikov, “Formation of Reaction Garnets during Melting of the MgCO3–CaCO3–NaAlSiO4–SiO2 System at 7.0 GPa,” Dokl. Akad. Nauk 406 (1), 83–88 (2006) [Dokl. Earth Sci. 406, 41–45 (2006)]. M. B. Baker and P. J. Wyllie, “Liquid Immiscibility in a Nepheline–Carbonate System at 25 kbar and Implications for Carbonatite Origin,” Nature 346, 168–177 (1990). L. N. Kogarko, C. M. B. Hengerson, and H. Pacheko, “Primary Ca-Rich Carbonatite Magma and Carbonate– Silicate–Sulphide Liquid Immiscibility in the Upper Mantle,” Contrib. Mineral. Petrol. 121, 267–274 (1995). Glinnemann J., Kusaka K., Harris J.W. “Oriented Graphite Single-Crystal Inclusions in Diamond,” Z. Kristallogr. 218 (11), 733–739 (2003). Yu. A. Litvin, F. Pineau, and M. Javoy, “Carbon Isotope Fractionation on Diamond Synthesis in Carbonatite– Carbon Melts of Natural Chemistry (Experiments at 6.5– 7.5 GPa),” in Proceedings of 6th International Symposium on Applied Isotope Geochemistry, Prague, Czechia, 2005 (Prague, 2005), p. 143. O. Navon, “Diamond Formation in the Earth’s Mantle,” in Proceedings of 7th Int. Kimberlite Conference, Cape Town, South Africa, 1999, Ed. by J. J. Gurney (Red Roof Design, Cape Town, 1999), Vol. 2, pp. 584–604. D. A. Zedgenizov, H. Kagi, V. S. Shatsky, and N. V. Sobolev, “Carbonatitic Melts in Cuboid Diamonds from Udachnaya Kimberlite Pipe (Yakutia): Evidence from

47. 48. 49.

50. 51.

52. 53. 54.

55. 56.

57. 58.

59.

60.

61.

62.

Vibrational Spectroscopy,” Mineral. Mag. 68 (1), 61–73 (2004). H. O. A. Meyer, “Inclusions in Diamond in Mantle Xenoliths, " in Mantle Xenoliths, Ed. by P. H. Nixon, (Wiley, Chichester, 1987), pp. 501–523. L. Taylor and M. Anand, “Diamonds: Time Capsules from the Siberian Mantle,” Chem. Erde 64, 1–74 (2004). E. S. Izraeli, M. Schrauder, and O Navon, “On the Connection between Fluid- and Mineral-Inclusions in Diamonds,” in Extended Abstracts of 7th Int. Kimberlite Conference, Cape Town, South Africa, 1998 (Cape Town, 1998), pp. 352–354. W. E. Sharp, “Pyrrhotite, a Common Inclusion in South African Diamonds,” Nature 211, 402–403 (1966). E. S. Efimova, N. V. Sobolev, and L. N. Pospelova, “Inclusions of Sulfides in Diamonds and Characteristics of Their Paragenesis,” Zap. Vses. Mineral. O-va 112 (3), 300–310 (1983). G. P. Bulanova, W. L. Griffin, C. G. Ryan, et al., “Trace Elements in Sulfide Inclusions from Yakutian Diamonds,” Contrib. Mineral. Petrol. 124, 111–125 (1996). G. P. Bulanova, “Formation of Diamond,” J. Geochem. Explor. 53, 1–23 (1995). A. Wang, J. D. Pasteris, H. O. A. Meyer, and M. L. DeleDuboi, “Magnesite-Bearing Inclusion Assemblage in Natural Diamond,” Earth Planet. Sci. Lett. 141, 293–306 (1996). N. V. Sobolev, F. V. Kaminsky, W. L. Griffin, et al., “Mineral Inclusions in Diamonds from the Sputnik Kimberlite Pipe, Yakutia,” Lithos 39, 135–157 (1997). S. V. Titkov, A. I. Gorshkov, Yu. P. Solodova, et al., “Mineral Microinclusions in Cubic Diamonds from the Yakutian Deposits Based on Analytical Electron Microscopy Data,” Dokl. Akad. Nauk 410 (2), 255–258 (2006) [Dokl. Earth Sci. 410, 1101–1105 (2006)]. E. S. Izraeli, J. W. Harris, and O. Navon, “Brine Inclusions in Diamonds: A New Upper Mantle Fluid,” Earth Planet. Sci. Lett. 187, 323–332 (2001). Yu. A. Litvin, “Alkaline–Chloride Components in Processes of Diamond Growth in the Mantle and High-Pressure Experimental Conditions,” Dokl. Akad. Nauk 389 (3), 382–386 (2003) [Dokl. Earth Sci. 389A, 388–391 (2003)]. M. Akaishi and S. Yamaoka, “Crystallization of Diamond from C–O–H Fluid under High-Pressure and High-Temperature Conditions,” J. Cryst. Growth 209, 999–1003 (2000). S. Yamaoka, M. D. Shaji Kumar, M. Akaishi, and H. Kanda, “Reactions between Carbon and Water under Diamond-Stable High Pressure and High Temperature Conditions,” Diamond Relat. Mater. 9, 1480–1486 (2000). Yu. N. Pal’yanov, A. G. Sokol, A. F. Khokhryakov, et al., “Diamond and Graphite Crystallization in C–O–H Fluid at P–T Parameters of the Natural Diamond Formation,” Dokl. Akad. Nauk 375, 348–388 (2000) [Dokl. Earth Sci. 375, 1395–1398 (2000)]. A. G. Sokol, Yu. N. Pal’yanov, G. A. Pal’yanova, et al., “Diamond and Graphite Crystallization from C–O–H Fluids,” Diamond Relat. Mater. 10 (12), 2131–2136 (2001).


Vol. 46

No. 6

2008

EXPERIMENTAL CHARACTERIZATION OF DIAMOND CRYSTALLIZATION 63. Yu. N. Pal’yanov, A. G. Sokol, Yu. M. Borzdov, and A. F. Khokhryakov, “Alkaline Carbonate–Fluid Melts as the Medium for the Formation of Diamonds in the Earth’s Mantle: An Experimental Study,” Lithos 60, 145–159 (2002). 64. A. G. Sokol and Yu. N. Pal’yanov, “Diamond Crystallization in Fluid and Carbonate–Fluid Systems under Mantle P–T Conditions: 2. An Analytical Review of Experimental Data,” Geokhimiya, No. 11, 1157–1172 (2004) [Geochem. Int. 42, 1018–1032 (2004)]. 65. M. Akaishi, “Effect of Na2O or H2O Addition to SiO2 on the Synthesis of Diamond from Graphite,” in Proceedings of 3rd Int. Symposium ISAM'96, Tsukuba, Japan, 1996 (NIRIM, Tsukuba, 1996), pp. 75–80. 66. R. H. Wentorf and H. P. Bovenkerk, “On the Origin of Natural Diamonds,” Astrophys. J. 134, 995–1005 (1961). 67. A. I. Chepurov, I. I. Fedorov, and V. M. Sonin, Experimental Modeling of Diamond Formation (NITs OIGGM SO RAN, Novosibirsk, 1997) [in Russian]. 68. S. K. Simakov, “Redox State of Earth’s Upper Mantle Peridotites under the Ancient Cratons and Its Connection


Vol. 46

No. 6

69.

70. 71.

72.

73.

2008

553

with Diamond Genesis,” Geochim. Cosmochim. Acta 62 (10), 1811–1820 (1998). I. I. Fedorov, A. I. Chepurov, A. A. Chepurov, and A. V. Kuroedov, “Estimation of the Rate of Postcrystallization Self-Purification of Diamond from Metal Inclusions in the Earth’s Mantle,” Geokhimiya, No. 12, 1340– 1344 (2005) [Geochem. Int. 43, 1235–1240 (2005)]. A. A. Kadik, F. Pineau, Yu. A. Litvin, et al., “Formation of Carbon and Hydrogen Species in Magmas at Low Oxygen Fugacity,” J. Petrol. 45 (7), 1297–1310 (2004). A. A. Kadik, Yu. A. Litvin, V. V. Koltashev, et al., “Solubility of Hydrogen and Carbon in Reduced Magmas of the Early Earth’s Mantle,” Geokhimiya, No. 1, 38–53 (2006) [Geochem. Int. 44, 33–47 (2006)]. I. D. Ryabchikov, G. P. Brey, and V. K. Bulatov, “Carbonate Melts Equilibrated with Mantle Peridotites at 50 kbar,” Petrologiya 1, 189–194 (1993) [Petrology 1, 159–163 (1993)]. Yu. A. Litvin, “Mantle Hot Spots and Experiments up to 10 GPa: Alkaline Reactions, Carbonation of the Lithosphere, and New Diamond-Forming Systems,” Geol. Geofiz. 39, 1772–1779 (1998).