The nature and distribution of carbon in submarine basalts and peridotite nodules

Earth and Planetao' Science Letters, 56 ( 1981 ) 217-232

217

Elsevier Scientific Publishing Company. Amsterdam - Printed in The Netherlands

[6]

The nature and distribution of carbon in submarine basalts and peridotite nodules E.A. Mathez and J.R. Delaney 2 I Department of Geological Sciences, 2 Department of Oceanography, Universi O, of Washington, Seattle, WA 98195 (U.S.A.)

Received October 8, 1980 Revised version received August 12, 1981

Primary carbonaceous material has been identified in submarine basaltic glasses and mantle-derived peridotite nodules from alkali basalts using electron microprobe techniques. In the submarine rocks carbon occurs (I) in quench-produced microcracks in glasses and phenocrysts, (2) in vesicles, where it is preferentially concentrated on the sulfide spherules attached to vesicle walls, and (3) in microcracks and CO z-rich bubbles in inclusions of glass completely enclosed by phenocrysts. In peridotite nodules carbon exists in intergrain cracks, along grain boundaries, and on the walls of fluid inclusions disposed in two dimensional arrays. The carbonaceous material is believed to consist of a mixture of graphite, other forms of elemental carbon, and possibly small amounts of organic matter. It is suggested that carbon precipitates by disproportionation of CO according to the reaction 2 CO ~ C + CO 2 and that this reaction is catalyzed by sulfide-oxide surfaces in vesicles. Once deposition has begun, the reaction continues on carbon surfaces as well. Based on the large amounts of condensed carbon observed in some vapor inclusions and the apparent lack of oxidation features associated with them. it is proposed that carbon condensed from a magmatic vapor in which CO was a significant constituent. This implies that oxygen fugacities of undegassed basaltic melts under confining pressures of the shallow crust are typically lower than those of the Q F M buffer at equivalent temperatures. This is in agreement with some intrinsic oxygen fugacity measurements on similar undegassed materials. Regardless of the mechanism of its formation, the presence of carbon in CO 2-rich vesicles and inclusions in basaltic glasses and mantle nodules adds uncertainty to estimates of m i n i m u m pressures of entrapment based on measurements of fluid densities. Condensed carbon also accounts for some of the carbon isotopic characteristics of these rocks.

1. Introduction

Knowledge of the chemical equilibria among volatile elements in magmas is central to important petrologic processes such as mass transport and magma degassing. The equilibria depend on temperature, pressure, and the oxidation states and abundances of the elements involved. Carbon, in particular, is of interest because as CO 2 it is a major component of volcanic gases [1] and is observed as inclusions in mantle-derived peridotites [2,3]. Also, experiment has established that the presence of carbon profoundly affects the temperatures at which model mantle assemblages begin melting and the compositions of the partial melts over a wide range of upper mantle conditions [4,5].

Because most reactions that can be written among the major volatile species involve exchange of oxygen, the determination of oxygen fugacity becomes important in understanding volatile related processes. This point has been emphasized by Sato [6] for both lunar and terrestrial systems, and it can be further illustrated by consideration of phase relations in the C-O system (e.g. [7,8]). Oxygen fugacities under which graphite is stable are strongly temperature and pressure dependent. Of geologic interest is the fact that, except at low pressures ( < 100 bar), the vapor in equilibrium with graphite is dominated by CO 2 rather than CO through most of the subsolidus temperature range. At pressures greater than a few hundred bars, graphite is stable under conditions more oxidizing than the iron-wustite buffer at magmatic as well as

0012-821X/81/0000-0000/$02.75 '~ 1981 Elsevier Scientific Publishing Company

218 subsolidus temperatures. This means that graphite may be stable throughout much of the crust and upper mantle [6]. In fact, Hoefs [9] suggested that small amounts of condensed carbon exist in nearly all crustal rocks and minerals. Furthermore, if magmas are sufficiently reduced so that CO makes up a significant proportion of their vapor, then the CO2-rich gases observed in basalts and mantlederived peridotites at room temperature are not representative of magmatic ones. The observations described here suggest that this is in fact the case and that reactions involving the gas phase during cooling need to be considered. Primary condensed carbon exists in vesicles and microcracks in submarine basaltic glasses and in vapor inclusions in mantle-derived peridotite nodules recovered from alkalic basalts. Because they are rapidly quenched under moderate pressure, the submarine basaltic glasses remain largely undegassed. The mineral assemblages of the nodules are believed to have only partially re-equilibrated at low pressure. Using electron microprobe techniques, carbon has been identified in five submarine basalts and two nodules. In addition, it has been observed optically in many similar rocks. A catalytic mechanism involving deposition of carbon by disproportionation of CO from an original magmatic vapor is proposed to account for specific aspects of carbon distribution and texture, and the consequences of this mechanism are examined in terms of high-temperature oxidation.

Contamination becomes a problem only when the detection of low concentrations of carbon is desired, such as over small cracks. The primary concern was to avoid the introduction of carbon compounds to the specimen during preparation. For this reason, samples were mounted in a non-carbon-bearing medium (sodium metasilicate) and polished with alumina and water. In practice, even samples mounted in epoxy showed no severe contamination problems when aluminawater slurries were used in polishing. Aluminum was utilized as a conductive surface coat because it is non-toxic and relatively transparent to low-energy X-rays. However, it has the disadvantages of being highly reflective and susceptible to rapid oxidation. The ideal coat is thick enough for adequate conduction yet thin enough for optical and X-ray transparency. A practical means of producing such a coat is to monitor the resistance change across a glass slide during aluminum deposition and during subsequent introduction of atmosphere to the evaporation chamber, at which time a thin oxide surface layer forms almost immediately on the aluminum. The adequacy of the coat as a conductor is critical to low-level carbon analysis because it appears to affect the rate at which carbon from pump oils is deposited under the beam. Again, this is not a concern to the pres+nt study because relatively thick coats could be tolerated for studying the distribution of carbon-rich phases.

2. Analytical procedure

3. Samples

The carbon distribution was studied with electron microprobe procedures initially developed to detect low concentrations of carbon dissolved in glasses. Samples were run at 8 kV acceleration potential and 140 nA sample current (graphite target) on the ARL SEMQ microprobe at the Eidgen6ssische Technische Hochschule, Ziarich. This instrument is fitted with an LN 2 cold trap above the diffusion pump and a molecular sieve in the foreline between the diffusion and mechanical pumps. This configuration minimizes carbon buildup on the sample surface resulting from the cracking of pump oils under the electron beam.

The five submarine basalts used in this study are described in Table 1. They included a plagioclase-rich high-alumina basalt from the Kolbeinsey Ridge near Iceland (TR-139-25D), a magnesian lava from the east rift zone of Kilauea (9), a relatively primitive mid-oceanic ridge tholeiite from the FAMOUS area (526-1), and two more evolved lavas from near the Bouvet Islands in the South Atlantic (CHN and SLNT). The mineralogy and chemistry of the FAMOUS basalt are reported by Bryan and Moore [10]; Watson [11] has studied the glass inclusions and Graham [12] the gases of the Bouvet samples; the Hawaiian basalt has been

219 TABLE 1 Glass compositions, phenocryst assemblage and collection depths of submarine basalts studied for carbon TR- 139-25D [14]

526-1 [10]

CHN [1 I]

SiO 2 TiO 2 A1203 FeO * MnO MgO CaO NazO K 20 P205 S (ppm) Total

50.39 0.81 14.38 10.17 0.19 8.18 12.36 1.91 0.03 0.08 1120 98.22

51.36 0.78 15,03 7.86 0,10 9.01 12.38 2.04 0.10 890 * 98.75

Phenocryst assemblage **

pl, ol

ch, ol, pl

pl, ol, cpx

pl, ol, cpx

ol, pl, cpx

Collection depth (m)

1100-1200

2400

1800

1800

5000

5 I. 19 1.61 15.15 8.73 0.18 7.24 11.53 3.17 0.49 -

SLNT [l I]

9 [131

50.70 1.29 15.24 8.02 0.14 8.37 13.16 2.28 0.39 -

47.68 1.95 10.12 11.41 0.17 17.74 7.85 1.71 0.33 0.20

1040 * 99.39

99.59

* Mathez and Delaney, unpublished data. ** pl -- plagioclase; ol = olivine; cpx = Ca-rich pyroxene; ch = chromite.

described by Moore [13] and the Kolbeinsey Ridge sample is included in Mathez' [14] investigation of sulfur in submarine glasses. These rocks are all fresh, and only their glassy non-crystalline portions were examined for carbon. The glasses were collected in deep water, are sparsely vesicular, and have only partially degassed. Nodule UM-5 was collected from the 1801 Hualalai flow, Hawaii (B.E. Nordlie, personal communication). It is a coarse granular chromitebearing dunite (-Fo87 ) with minor Cr-diopside, accessory primary sulfides, and abundant vapor inclusions. The inclusions exist in crystals, along grain boundaries, and in the silicate glass observed in minor amounts in interstices and along grain boundaries of crystalline phases. Sample KH77-6 is a fresh protogranular spinel lherzolite from Kilbourne Hole, New Mexico. The trace element chemistry of this rock is reported by Irving [15]. It contains magnesian olivine (-Fog0), Cr-spinel, Cr -diopside, orthopyroxene, and accessory sulfides. Both nodules fall within the Group I suite of Frey and Prinz [ 16].

4. Carbon distribution

In the submarine basalt glasses carbon exists on vesicle walls, in quench-produced microcracks in matrix glass and crystals, and in the microcracks and vapor bubbles in glass inclusions in phenocrysts. In the nodules carbon is observed in vapor inclusions and cracks in crystals and along grain boundaries. 4.1. Microcracks

Although the numerous microcracks observed in fresh submarine basalt glasses may have originated through a variety of mechanisms, many appear to have formed during quenching. These include the ubiquitous curvilinear cracks in glass circumjacent to phenocrysts, particularly olivine (Fig. I), and the fractures in crystals themselves which terminate at crystal boundaries or within adjacent glass or which merge with the circumjacent cracks. Alteration minerals are absent in such cracks, at least in the rocks described here, and no evidence exists to suggest that they were once channels for hydrothermal fluids. Many of these cracks contain a carbon-rich

220

phase, as demonstrated by the scanning photomicrographs of carbon X-rays of Figs. 1 and 2. This phase appears as delicate semicontinuous dendritic masses and rosettes on crack walls (Fig. 2). The high carbon concentrations observed in scanning photomicrographs are correlative with locations along these cracks where the dendritic masses intersect with and are exposed at the sample surface; the carbon signal originates from these masses and not from other locations along cracks. Not all the opaque material in cracks is carbonaceous. Sulfides are also observed, for example. However, beam scans for sulfur X-rays similar to the carbon images demonstrate that sulfides typically exist as discrete individual grains and are much less abundant than carbonaceous matter. The carbon phase in Fig. 2 is easily observed because cracks in which it exists are relatively large. In smaller cracks, carbon appears as a "film" on crack walls and results in slight discoloration. In fact, the smallest cracks may be visible only for this reason. This is particularly true of cracks in glass inclusions, where the mass of carbon in them is so small that X-ray intensities are too low to

appear in scanning photomicrographs. However, unambiguous carbon X-ray signals can be generated from such cracks with a stationary electron beam. Nodules UM-5 and KH77-6 contain numerous inclusions filled with CO2-rich vapor. They are distributed in such a way as to define random surfaces through all constituent crystals. These inclusion arrays are common to nearly all mantlederived nodules [2,3,17]. Green and Radcliffe [17] suggest that the fluids in them were originally concentrated along crystal defects and grain boundaries during deformation of the nodules in the mantle. Evidently, cracks in which fluids were accumulated were subsequently annealed during transport of the nodule to the surface by the host magma. In addition to inclusion arrays, microcracks are also present in both nodules. Many microcracks and inclusion arrays are restricted to individual crystals, terminating at grain boundaries. Others traverse larger volumes of the nodules. Carbon exists in nearly all inclusion arrays and in many microcracks and grain boundaries as well.

200/zrn Fig. 1. Quench-produced microcracks in glass circumjacent to an olivine phenocryst. In the reflected light photomicrograph (right) the white phase is chromite containing glass inclusions. The beam scanning photomicrograph (left) is of carbon X-rays over the outlined area of the reflected light photomicrograph. The beam scan area appears slightly discolored on the reflected light photomicrograph from carbon deposition from the pump oils during beam loading. Spot density is proportional to carbon concentration. This sample was mounted in sodium metasilicate and polished with alumina-water slurries.

221

Iii IIIIII

IO0/ m Fig. 2, Dendrites and rosettes of carbonaceous matter on crack walls, transmitted light (right), where the focal plane is below the thin section surface, and beam scanning image of carbon X-rays (left) over the area outlined on the reflected light photomicrograph. In cracks such as these in which the opaque material is relatively coarse, carbon signals are produced from locations where it is exposed on the analytical surface. Sample CHN.

4.2. Vesicles

Vesicles in glassy basalts that have erupted in deep water typically possess a complex variety of features, including the ubiquitous sulfide spherules embedded in their walls [18,19]. These sulfides exhibit remarkable regularity in distribution and size within individual vesicles, show an increase in size and decrease in number in vesicles progressively further from pillow margins, have compositions similar to those of the sulfide globules in glasses, and are believed to form at magmatic temperatures. Typical spherule-bearing vesicles are shown in Figs. 3 and 4, along with beam scanning images of carbon X-rays. High concentrations of carbon in scanning photomicrographs can be seen to correlate with specific spherules observable optically (Fig. 3). The high spot densities in the. scanning

photomicrographs are not due to the higher Bremsstrahlung signal of the sulfide than surrounding glass, a possibility that was tested by scanning over sulfides and glass in juxtaposition on flat surfaces. Most, but not all, spherules yield either high carbon or high sulfur count rates, but not both (Fig. 4). In some vesicles, spherules from which high carbon signals originate, when observed in reflected light, lack the distinct metallic luster of other spherules (Fig. 4). However, no other differences, such as in size or spatial distribution, have been noted. These observations are interpreted to indicate that carbon exists as a surface layer on some of the sulfides, where it is preferentially concentrated compared to other surfaces in vesicles. At 1 atm and temperatures < 1200°C, it is known that small ( < 0.5 wt.%) amounts of carbon can dissolve in sulfide melts in equilibrium with

222

50~m Fig. 3. Typical vesicle in submarine basalt glass. The spots on the reflected light photomicrograph (right) are sulfide spherules. The beam scanning photomicrograph on the left is of carbon X-rays over this vesicle. Note that high carbon concentrations can be correlated with specific spherules. Sample CHN.

iron melts [20], and at higher pressures carbon solubilities may be greater. For this reason, a distinct carbon phase was sought in and around the margins of sulfide globules completely enclosed in glass. Although this search was not exhaustive, carbon has yet to be detected in sulfides unassociated with vapor. This indicates that carbon condensed from the vapor phase and was not exsolved from the sulfide. 4. 3. Glass inclusions

It is generally agreed that many glass inclusions in phenocrysts represent melt trapped prior to eruption and quenching [21], although their compositions usually require interpretation in terms of a multiplicity of processes [11,22]. Most inclusions contain a vapor bubble and are bisected by a microcrack (Fig. 5). Bubbles make up ~ 4% of the inclusion volume and appear to have exsolved after entrapment of melt by the host crystal [23]. The microcracks either do not penetrate the host crystal, terminate immediately within it, or form curved surfaces in the host crystal around the inclusion but confined to its immediate vicinity. Carbon occurs in these bubbles and micro-

cracks. Its presence has been confirmed in inclusions initially enclosed by the host crystal, and thus isolated from the outside environment, and then subsequently exposed on the sample surface by continued polishing. Because these inclusions were trapped under, magmatic conditions, the carbon in them must be primary; that is, it is not material introduced during or after the eruption and quenching of the lavas on the ocean floor. By analogy, carbon in microcracks and vesicles in the matrix glasses is also interpreted as primary. The distribution of carbon X-ray intensities generated from randomly chosen locations on walls of bubbles in glass inclusions are shown in Fig. 6. It can be seen that many bubbles contain large quantities of carbon and that the intensities exhibit wide variability. This variability extends to wide beam analyses of different bubbles (where beam diameter is adjusted to bubble diameter) and to multiple focused-beam analyses within a single bubble. Much of the variability is due to the uneven distribution of condensed carbon on bubble walls as well as to the fact that the signals were generated from highly irregular surfaces. Also, since melt inclusions are trapped over a range of pressures, the bubbles that exsolve in them should

Fig. 5. Glass inclusions in a plagioclase phenocryst, transmitted light. Many inclusions contain a vesicle and are bisected by a microcrack restricted to the inclusion or to the nearby vicinity in the phenocryst. Sample CHN.

Fig. 4. Spherule-bearing vesicle. The center beam scanning photomicrograph is of carbon and the bottom one of sulfur X-rays for the area outlined in the reflected light photomicrograph. Individual spherules usually yield either high sulfur or high carbon count rates but not both. Note that spherules which yield strong carbon signals are less reflective than other

contain varying amounts of carbon. It is of interest to estimate the relative amounts of condensed carbon and gas occupying bubbles. This can be accomplished from knowledge of bubble diameters, of their carbon contents, and of the relationships between the volumes from which characteristic carbon X-rays are produced and Xray excitation conditions and target properties (cf. [24]). For example, the average of all the data plotted in Fig. 6 is equivalent to 6.2 wt.% carbon. This would result from a carbon layer - 0.05 pm thick on a flat surface. For an average bubble of 10 pm diameter, an X-ray intensity equivalent to 6.2 wt.% carbon would correspond to an average carbon content of - 3 vol.%. It is estimated that most melt inclusion bubbles contain between 0.5 and 5 vol.% condensed carbon. It should be emphasized that large errors are associated with this estimate because of the uncertainties inherent in the above procedure and the variability of the carbon data. The main point is that bubbles in glass inclusions contain large and spherules but Sample CHN.

that

in all other

respects

they appear

similar.

224

30I~ x0

e-

r

o

Y

S

l

C

N

~

0

WT. % CARBON

50

Fig. 6. The distribution of carbon intensities generated with a focusedelectron beam randomly located in the interiors of bubbles in glass inclusions. 210 points.

variable amounts of condensed carbon. The few comparable data collected for vesicles in matrix glasses suggest that their carbon contents are much lower.

5. Nature of carbon

X-ray intensities corresponding to > 50 wt.% carbon, and in some instances > 70 wt.%, can be generated from exposed crack and vesicle wails. Inasmuch as these surfaces are irregular and marginally conductive, low X-ray intensities are anticipated from them relative to flat conductive surfaces of similar compositions. The high X-ray intensities limit the material to some form of elemental carbon or organic matter and specifically eliminate carbonate ( ~ 12 wt.% C) as a possible phase. SEM examination of fracture surfaces and vesicle walls revealed no recognizable crystalline morphologies. The carbonaceous matter in cracks sometimes appears to erode under the electron beam. However, carbon X-ray signals from cracks and bubbles do not decay, remaining constant with time, even under the high electron beam currents employed in this study. In addition to graphite, the possible forms in which carbon could exist include carbynes, hydrocarbons, and ill-defined "amorphous" matter.

Carbynes consists of - C ~ C - units linked together to form a series of linear polymorphs (see Whittaker [25] and references therein). They have been identified in the Murchison (C2) and Allende (C3) carbonaceous chondrites [26,27] and have been found intermixed with some terrestrial graphites [28]. Although they are thermodynamically unstable at temperatures below - 2 3 0 0 ° C at all pressures [29], carbynes have been synthesized metastably at temperatures < 680°C by catalytic disproportionation of CO on chromites [26]. Hydrocarbons are well known from C1 carbonaceous chondrites [30,31]. Their presence in meteorites has been attributed to a Fisher-Tropsch type reaction [32] in which partial hydrogenation of CO occurs in the presence of a clay catalyst. Organic matter has also been reported in diamonds recovered from kimberlites [33] and in some alkaline igneous rocks [34,35]. In basalts, however, hydrocarbon concentrations are extremely low except where vegetation pyrolysis has occurred [36]. For example, Muenow [37] investigated the mass spectra in the range of 2-150 a.m.u, of gases liberated during progressive heating of tholeiitic glass from Hawaii. He reports small but detectable molecular ion intensities in the released gases characteristic of fragmentation spectra of various organic constituents. These may be indicative of the presence of "tarry" com-

225

pounds that might occupy cracks and vesicle walls, but their intensities were so small compared to those attributed to CO 2 and CO that their concentrations were considered to be "negligible". In a similar study, Graham [12] also observed no peaks that could be interpreted in terms of fragmentation patterns of organic matter in the hightemperature release spectra of sample9 from Hawaii and Bouvet samples C H N and SLNT, even though H 2 0 constituted a significant portion of the released gases in the latter two. Similar results have been obtained from investigations of typical MOR basalts [23], of relatively H20-rich tholeiites from near the Marianas arc [38], and of other basalts and nodules collected from Hawaii [39,40]. Large wavelength differences exist among the characteristic X-ray spectra of carbon combined in different ways. The spectra generated from the carbonaceous material in basalts and ultramafic nodules and from graphite are compared in Fig. 7. The two are remarkably similar on the high-

z

,

I

43.0

,

i

44.0

I

45.0

i

I

46.0

i 47,0

Fig. 7. Comparison of the wavelength distributions of characteristic carbon X-rays generated from graphite and carbonaceous matter in basalts and nodules. Vertical scale for graphite is 2 × that for natural carbon to facilitate comparison. The wavelength of the graphite centroid is 44.7000 A. Similar values measured for carbon combined in other material are as follows: calcite: 44.4974 A; SiC: 44.5825 A; vitreous carbon: 44.7900 A; carbon dissolved in synthetic diopside glass prepared at 20 kbar~ 1625°C by J. Holloway: 44.4074 A. The distribution of carbon intensities generated from random locations in the interiors of bubbles in glass inclusions in phenocrysts of submarine basalt glasses. 210 points.

wavelength shoulders of the peaks, but the former is distinctly broader and slightly more asymmetrical. These spectral features could result from a mixture of phases. In view of the observations summarized above, it is suggested that the condensed carbons in submarine basalts and mantle nodules consist of a mixture of graphite, other forms of elemental carbon (possible carbynes) and small amounts of organic matter.

6. Discussion 6.1. Vapors in submarine basalts For the present purposes, vapor equilibria in submarine basaltic melts can be approximated by those in the C-O system. This is justified for two reasons. First, at room temperature the vesicle gases in submarine tholeiitic glasses consist of nearly pure CO 2 [41]. The dominance of CO 2 is in part due to the fact that the solubilities of FeS and H 2 0 in mafic liquids are much higher than those of carbon species at moderate pressures ( > 2 0 0 bar). The glasses, for example, typically contain in solution 1000-1800 ppm sulfur [14], and the H 2 0 contents of these basalts are always much lower than saturation levels at their eruption depths [13,23]. Second, at oxygen fugacities near the graphite stability field, small amounts of hydrogen act mainly as dilutants in C-O vapors [7]. Thus, C H 4 is never an important hydrogen species in vapors consisting of more than 90% CO 2 + CO, even at high pressures [42]. The effect of addition of small amounts of H z to the C-O system is to depress the oxygen fugacity at which graphite is stable. 6.2. Equilibria in the C-O system The C-O system has been discussed in some detail by French and Eugster [7]. Equilibrium between graphite and a C-O vapor is governed by the relation: 2 CO ~ C + CO 2

(1)

The products of this reaction are favored with decreasing temperature and increasing pressure. This is emphasized by Fig. 8, which shows the

226

T (%) 200

300

4

400

500 ,

600 ,

,

800 , ,

,

in the metallurgical and surface science literature because they have important industrial applications. Among the materials that are active catalysts for CO are transition metals Fe, Ni, Mo, and Pt. It is believed that CO chemisorbs non-dissociatively on these metal substrates, resulting in modification of the orbital configuration of adsorbed CO from gaseous CO so that reaction of the two becomes possible (see Andersson et al. [48]). In such a case reaction (1) would be more properly expressed:

1200 ,i

p~~~lO0~P=500 P IO00BARS

o

o_~

3

a"° 2 2C -125.0

2

Oi

.0

= 15.0

i IO.O

5.0

I(~1T=K

Fig. 8. Composition of C-O vapor in equilibrium with graphite at various pressures as a function of temperature. Thermochemical data are from Robie et al. [43], CO 2 fugacity data are from B u r n h a m and Wall (unpublished data), and CO fugacities were calculated by the method of corresponding states (e.g. [44]). Ideal mixing as assumed; errors inherent in this assumption are small for P < I kbar.

composition of vapor in equilibrium with graphite as a function of temperature and pressure, and by Fig. 9, in which graphite stability curves are compared with some oxygen buffers. It is evident that graphite can exist under a wide range of geologic conditions. In Fig. 9 are also plotted contours of Pco/ (Pco + Pco2)" These isopleths are relatively insensitive to pressure changes. It can be seen that the vapor in equilibrium with graphite becomes progressively more CO 2-rich with decreasing temperature and that at moderate pressures and low temperatures graphite and nearly pure CO 2 can coexist.

6.3. Disproportionation of CO High activation energies are associated with the formation of graphite because reaction (1) is known to be "symmetry forbidden" [48,49]. That is, its progress to the right is kinetically unfavorable, and CO-bearing gases tend to persist metastably under conditions far removed from equilibrium. However, CO disproportionates in the presence of certain catalysts. This and related catalytic processes have been the subjects of numerous investigations

CO(aa, + CO(v) ~ C(ad) + CO:,v I

(la)

where the subscripts (ad) and (v) refer to adsorbed and vapor species, respectively. Reaction mechanisms involving other catalysts may be quite different, however, because experiments have shown that reaction rates are complexly dependent on the nature of the catalyst (e.g., its composition, surface structure, surface area), on impurities in the gas, and on temperature. For example, the addition of small amounts of H 2 to a CO gas greatly enhances the rate of CO disproportionation on iron [51], presumably because of the parallel carbon-producing reaction: H 2 + C O ---, C + H 2 0

(2)

In contrast, the addition of sulfur retards the reaction rate because of the formation of troilite, which is not as catalytically active as iron [52]. In addition, it has been claimed that various forms of iron oxides are capable of catalyzing reaction (1) [53], that CO can disproportionate in the presence of chromite to produce carbynes [26], and that carbon itself serves as an active catalyst [54]. If sulfide spherules on vesicle walls are similar to the sulfide globules in glasses [19], then they should consist of intimate intergrowths of pyrrhotire, Cu-Fe sulfides, pentlandite, magnetite, and silicate. Over much of the temperature range in which carbon may have begun precipitating, monosulfide solid solution should be present. Whether or not such an assemblage is a suitable catalyst for a complex CO-bearing gas at elevated temperatures and pressures has not, to our knowledge, been established through experiment. The preferential concentration of carbon on sulfides, the irregularity of its distribution on ves-

227

T(°C) 500

800

700

600

w

w

i

*

900

I000

.

,

I100 1200 1300 ,

w

e .

l'le

-6 • •

•

-8

f .,.(J,

-

,"

30% 40%

.

sp.~..

• " ,,

-I0

",

",

D "/e I0"/.

• I"/'.

-12

0 w,--

(.9 0 ._1

-14

SKAERGAARD LZ

-16

SUBMARINE BASALT OL

-18 -20 -22 -24 -26 • 14.0

~

!

13.0

i

12.0

!

I1.0

I

!

I

!

I

I0.0

9.0

8.0

7.0

6.0

I

104/T(°K) Fig. 9. Compositional and phase relations in the C-O system compared with some synthetic solid oxygen buffers and with intrinsic fo 2 measurements from some natural materials. QFM=quartz-fayalite-magnetite [45]; IW=iron-wustite. The Skaergaard LZ (Lower Zone) represents the range of intrinsic fo, measurements reported by Sato and Valenza [46], and the submarine basalt curve is from Sato [47]. Dotted lines are contours of "°co/( P('o + Pco2 ) in percent for pure C-O gas. These contours are not significantly pressure dependent. Other sources of data are the same as in Fig. 8.

icle walls, a n d its d e n d r i t i c a p p e a r a n c e in cracks are i n t e r p r e t e d to i n d i c a t e that c o n d e n s e d c a r b o n p r e c i p i t a t e s a c c o r d i n g to r e a c t i o n ( l a ) , that sulfide - o x i d e spherules act as catalysts, a n d that once the r e a c t i o n has been i n i t i a t e d it c o n t i n u e s b y autocatalysis. In view of the complexities of the catalytic process d e s c r i b e d above, c a r b o n d e p o s i t i o n m a y p r o c e e d b y a n u m b e r of r e a c t i o n paths, dep e n d i n g on the physical c o n d i t i o n s a n d phases present. W i t h o u t a t t e m p t i n g to specifically identify them, the process envisioned to o p e r a t e in the b a s a l t - v a p o r system is s c h e m a t i c a l l y p r e s e n t e d in Fig. 10.

6. 4. Implications for oxidation Because it is o b s e r v e d in cracks in glasses, c a r b o n must be stable at least at subsolidus temp e r a t u r e s below the liquid-glass transition. N e i t h e r its d i s t r i b u t i o n n o r the o p e r a t i o n of reaction (1) p r o v i d e direct i n f o r m a t i o n on its m a x i m u m temp e r a t u r e of stability a n d therefore on o x i d a t i o n states of basaltic m a g m a s or c o n d i t i o n s a t t e n d a n t on cooling. This is because there are at least two ways in which c a r b o n could form. O n the one hand, it could c o n d e n s e directly from CO-rich v a p o r at high t e m p e r a t u r e such that the a m o u n t of

228

GRAPHITE FORMING REACTIONS IN A VESICLE 2 CO --~ C + CO2

••

VA POR

~~,/COod

+ COg..-~ Cod+ COzg

~'~oxtoE ,~.,=.~COad+ COg,_~Cod+ CO2

~

~

' GRAPHITE

(AUTOCATALYSIS) g I

3 IJ.m

I

Fig. 10. Schematic representation of the catalytic disproportionation of CO in vesicles.

carbon formed is proportional to the CO content of the vapor. In this case, the formation of significant amounts of carbon would require that magmatic vapors be considerably more reduced than the QFM buffer. Alternatively, carbon could precipitate by breakdown of CO2, with CO as an intermediate product. In this case, it is not possible to deduce the vapor composition or its oxidation state from the amount of carbon present. The breakdown of CO z, however, would require an additional oxidation reaction involving the solid assemblage.

Carbon from CO-rich vapor. Consider a hypothetical case in which a basaltic melt quenches to glass and vesicles on the ocean floor. The system is at 1250°C under a confining pressure of 500 bar, and its oxidation state is such that the vapor is 30% CO. In such a case, f o 2 - 10--9.6, o r about two orders of magnitude below QFM (Fig. 9). For the sake of simplicity, it is assumed that the vesicle size is established at the initial temperature and pressure, that CO 2 and CO mix ideally, and that the vapor quenches isochemically. Then, if all CO were to react according to (1), a vesicle comprising 0.3 vol.% of condensed carbon and 99.7% COz would result. (For slowly cooled systems, the assumption that the vapor composition remains constant with cooling is not justified when the vapor and surrounding solid assemblage approach equi-

librium. Since the solid assemblage should follow a T-fo 2 path approximately parallel to a solid oxygen buffer such as QFM, the vapor phase should do likewise and thus become progressively more CO 2rich with falling temperature (Fig. 9). However, the arguments presented here are not changed qualitatively if the vapor actually quenches in this manner rather than isochemically.) In this same system consider a vapor bubble in a melt inclusion trapped in a phenocryst. Assume that conditions within the inclusion are similar to those beyond it except that the inclusion, because it was trapped in a subsurface magma reservoir, exists under a pressure of 3 kbar. Upon quenching, this high-pressure vapor bubble would consist of 2 vol.% C and 98% CO 2. Relations among carbon contents of a vapor inclusion, vapor composition and entrapment conditions are summarized in Fig. 11. There are several points to note here: First, for a single composition vapor, the higher the confining pressure the more condensed carbon that will be formed by the mechanism described above. This would explain, at least qualitatively, why large amounts of carbon are observed on walls of vapor bubbles in glass inclusions and at the same time why vesicles in the lower-pressure matrix glasses apparently contain relatively small amounts of carbon. Second, the large amounts of carbon ( ~ 3 vol.%) believed to be present in the glass inclusion bubbles should not be observed if melts are as oxidized as the QFM buffer. The carbon data indicate that oxidation states of basaltic melts in the highpressure inclusion environment are somewhat below QFM conditions. This oxidation argument is critically dependent on the amounts of condensed carbon present. As noted previously, these amounts are poorly constrained (Fig. 6). Nevertheless, the important point is that if carbon forms by the mechanism described above, its abundance in inclusions can be used to estimate magmatic oxidation states. The inference from the carbon data that oxidation states of basaltic magmas are below QFM is in agreement with some intrinsic fo2 measurements on rocks and minerals cooled under other than 1 atm confining pressures [46,47,55,56]. Of particu-

229

LOG foz -8

-I0

-9

-II

which should proceed through the intermediate steps [7]:

-12

7.0

2 CO 2 -~ 2 C O + 0 2

6.0

Z

5.0

0 n~

lad

t) 4.0 ,m LU

U')

co ILl

hZ~3.0 ¢.~

cD

z

_z

o

z ~

2.0

z

o

_5 0

o I,C

0 o

.ao

.40

,so

.8o

l,oo

CO/(CO + CO z ) Fig. 11. The amounts of condensed carbon that form by reaction (1) in quenched vapor inclusions as a function of initial vapor composition and confining pressure. For example, a vapor inclusion consisting of 30% CO and 70% CO 2 at 1250°C and 1500 bars confining pressure would quench to an inclusion containing ~ I vol.% condensed carbon and 99% CO2.

lar interest are those of Sato and Valenza [46] on rocks from the Lower Zone of the Skaergaard Intrusion and of Sato [47] on an olivine separated from a submarine tholeiitic basalt recovered from abyssal depths near Hawaii. These data are plotted in Fig. 9. It can be seen that over the entire temperature range of the measurements, the oxidation states of these materials are such that their coexisting vapors would posses large proportions of CO. The sample studied by Sato [47] is from the same suite as the Hawaiian basalt included in this study. Breakdown of CO 2. An alternative mechanism by which carbon could form is by breakdown of CO 2 according to the reaction: CO2----~C-1"-O 2

(3)

2 CO -~ C + CO 2 Consequently, a catalytic process may control its progress, regardless of the CO 2 contents of the original vapor. This mechanism would operate if the graphite stability field is encountered under conditions where the vapor is nearly pure CO 2, in which case the amounts of condensed carbon to precipitate by disproportionation of CO in the original vapor would be negligible. This would be anticipated for a basaltic melt which exists under an oxidation state at or above the Q F M buffer, regardless of whether the vapor quenches along a T-fo,- path approximated by the Q F M buffer itself or along an isopleth (Fig. 9). If carbon forms in this way, no inferences regarding magmatic oxidation conditions can be made. It should be noted that the stoichiometry of reaction (3) requires that condensed carbon and oxygen gas form on a mole for mole basis. Therefore, the only way in which significant amounts of carbon could form would be if reaction (3) were accompanied by an additional one which consumes oxygen. Such reactions might involve oxidation of Fe z+ to Fe 3+ in the glass surrounding the vapor inclusions, oxidation of the sulfides to SO 2, or introduction of a component such as H 2 into the vapor and its subsequent oxidation t o H 2 0 (reaction (2)). No evidence has been found to suggest that such oxidation processes have occurred in the vicinity of vesicles, of vapor inclusions in the forsteritic olivines in nodules, or of the carbon-rich vapor bubbles in glass inclusions. But even if oxidation had occurred it is not clear that the petrographic evidence for it would be easily recognizable. The production of condensed carbon by breakdown of CO 2 must remain a possibility, although on the basis of current observations it is not favored. 6.5. CO: geobarornetry

If CO is a significant high-temperature vapor species, then the nearly pure CO 2 vapors observed

230

in nodules [2,3] and submarine basalts [41] must result in part from reactions that occur during cooling. It can be seen that reaction (1) has a negative AV. If CO breaks down in this manner, estimates of minimum pressures of entrapment based on measurements of fluid densities are in error by amounts proportional to the condensed carbon contents of the inclusions. If reaction (3) operates, then the same conclusion remains because of the necessity of removal of oxygen as condensed carbon forms. If hydrogen is present, the situation becomes more complex, not only because of the presence of hydrogen-bearing species in the vapor but also because of the possible formation of organic matter. Before CO 2 geobarometry can be applied with any confidence these complications need to be better understood.

6.6. Carbon isotopes It has been pointed out by Hoers [57,58] that 6~3C values of m a n y mantle nodules and igneous rocks are considerably more negative (typically - 2 0 to -30%0) than the mantle values estimated from carbonatites and kimberlites (typically - 2 to - 9%o [59,60]). Hoers [9] had proposed in an earlier work that igneous rocks typically contain small quantities of "elementary" carbon. Based on this knowledge, on the carbon isotopic similarity of some meteorites and terrestrial igneous rocks, and on isotopic fractionation known to result from the Fisher-Tropsch process [61], Hoefs attributed the isotopic character of the terrestrial rocks to the presence of elemental carbon. Pineau et al. [62] found 813C values of --7.5%° for the C O 2 gas fraction in some mid-ocean ridge basalts but 8~3C values of - 12 to - 13.7%0 for total carbon. Subsequently, it was shown through experiment that the CO 2 vapor was enriched in a 8~3C relative to the carbon dissolved in the silicate liquid [63]. The presence of carbonaceous matter in these rocks, however, also accounts for the different isotopic ratios of carbon in the gas fraction and in the whole rock.

glasses and peridotite nodules is due to the disproportionation of CO on suitable catalytic surfaces during cooling. In this respect, the process is similar to that invoked to account for the presence of elemental carbon and organic matter in meteorites. The carbonaceous material in the terrestrial rocks appears to consist of a mixture of graphite, other forms of elemental carbon, and possibly small amounts of organic matter. From the amounts of carbon inferred to exist in vapor bubbles of melt inclusions trapped in phenocrysts and the apparent absence of any oxidation features associated with these bubbles, it is hypothesized that CO is an important constituent of magmatic vapors. This in turn implies that basaltic liquids are somewhat more reduced than the Q F M buffer. The presence of carbon in vapor inclusions, no matter what the mechanism of its formation, casts doubt on the validity of pressure estimates of entrapment based on measurements of fluid densities. Carbon also accounts for the fact that 8~3C values of some mantle-derived rocks are more negative than the estimated mantle values.

Acknowledgements Critical reviews of this manuscript by I.S. McCallum, B.W. Evans, J. Touret, M. Sato, and an anonymous reviewer are much appreciated. We are also indebted to G.C. Ulmer, who first directed our attention to the kinetic problems associated with graphite deposition. Most of the probe work was done at E.T.H. Ztirich. We extend particular thanks to J. Sommerauer and other friends in Ztirich for their help and support. E.B. Watson kindly provided the C H N and SLNT samples, J.G. Moore the Hawaiian basalt and A.J. Irving the Kilborne Hole nodule. This research is supported by N.S.F. grant EAR 7802714 to I.S. McCallum.

References 7. Conclusions The presence of primary carbonaceous matter in bubbles and microcracks of submarine basalt

I T.M. Gerlach and B.E. Nordlie, The C-O-H-S gaseous system, 1. Composition limits and trends in basaltic cases, Am. J. Sci. 275 (1974) 353.

231 2 E. Roedder, Liquid CO 2 inclusions in olivine-bearing nodules and phenocrysts from basalts, Am. Mineral. 50 (1965) 1746. 3 B.W. Murck, R.C. Burruss and L.S. Hollister, Phase equilibria in fluid inclusions in ultramafic xenoliths, Am. Mineral. 63 (1978) 40. 4 D.H. Eggler, The effect of CO 2 upon partial melting of peridotite in the system NazO-CaO-AI2Os-MgO-SiO2-CO 2 to 35 kb, with an analysis of melting in a peridotite-H20CO 2 system, Am. J. Sci. 287 (1978) 305. 5 P.J. Wyllie and W. Huang, Carbonation and melting reactions in the system CaO-MgO-SiO 2-CO 2 at mantle pressures with geophysical and petrological applications, Contrib. Mineral. Petrol. 54 (1976) 79. 6 M. Sato, Oxygen fugacity of basaltic magmas and the role of gas-forming elements, Geophys. Res. Lett. 5 (1978) 447. 7 B.M. French and H.P. Eugster, Experimental control of oxygen fugacities by graphite-gas equilibriums, J. Geophys. Res. 70 (1965) 1529. 8 E. Woermann, B. Knecht, M. Rosenhauer and G.C. Ulmer, The stability of graphite in the system C-O, Extended Abstr. contributed to the Second International Kimberlite Conference (1977). 9 J. Hoefs, Ein Beitrag zur Geochemie des Kohlenstoffs in magmatischen und metamorphen Gesteinen, Geochim. Cosmochim. Acta 29 (1965) 399. 10 W.B. Bryan and J.G. Moore, Compositional variations of young basalts in the Mid-Atlantic Ridge rift valley near lat. 36°49'N, Bull. Geol. Soc. Am. 88 (1977) 556. 11 E.B. Watson, Glass inclusions as samples of early magmatic liquid: determinative method and application to a South Atlantic basalt, J. Volcanol. Geotherm. Res. 1 (1976) 73. 12 D.G. Graham, A mass spectrometric investigation of the volatile content of deep submarine basalts, Ph.D. Dissertation, University of Hawaii (1978) 187 pp. 13 J.G. Moore, Petrology of deep'sea basalt near Hawaii, Am. J. Sci. 263 (1965) 40. 14 E.A. Mathez, Sulfur solubility and magmatic sulfides in submarine basalt glass, J. Geophys. Res. 81 (1976) 4269. 15 A.J. Irving, Petrology ahd geochemistry of ultramafic xenoliths in alkalic basalts and implications for magmatic processes within the mantle, Am. J. Sci. 280A (1980) 389. 16 F.A. Frey and M. Prinz, Ultramafic inclusions from San Carlos, Arizona: petrologic and geochemical data bearing on their pctrogenesis, Earth Planet. Left. 38 (1978) 129. 17 H.W. Green and S.V. Radcliffe, Fluid precipitates in rocks from the earth's mantle, Geol. Soc. Am. Bull. 86 (1975) 846. 18 J.G. Moore and L. Calk, Sulfide spherules in vesicles of dredged pillow basalt, Am. Mineral. 56 (1971) 476. 19 R.S. Yeats and E.A. Mathez, Decorated vesicles in deep-sea basalt glass, Eastern Pacific, J. Geophys. Res. 81 (1976) 4277. 20 E.T. Turkdogan and R.A. Hancock, Thermodynamics of carbon dissolved in icon alloys, III. Solubility of carbon in iron-sulphur melts, J. Iron Steel Inst. 179 (1955) 155. 21 A.T. Anderson, Chlorine, sulfur, and water in magmas and oceans, Geol. Soc. Am. Bull. 85 (1974) 1485.

22 M.S. Dungan and J.M. Rhodes, Residual glasses and melt inclusions in basalts from DSDP Legs 45 and 46: evidence for magma mixing, Contrib. Mineral. Petrol. 67 (1978) 417. 23 J.R. Delaney, D.W. Muenow, and D G . Graham, Abundance and distribution of water, carbon, and sulfur in the glassy rims of submarine pillow basalts, Geochim. Cosmochim. Acta 42 (1978) 581. 24 A.E. Andersen, Electron microprobe analysis of thin layers arid small particles with emphasis on light element determinations, in: The Electron microprobe, T.D. McKinley, K.F.J. Heinrich and D.B. Wittry eds. (Wiley, New York, N.Y., 1966) 58. 25 A.G. Whittaker, Carbon: a new view of its high-temperature behavior, Science 200 (1978) 793. 26 R. Hayatsu, R.G. Scott, M.H. Studier, R.S. Lewis and E. Anders, Carbynes in meteorites: detection, low temperature origin, and implications for interstellar molecules, Science 209 (1980) 1515. 27 A.G. Whittaker, E.J. Watts, R.S. Lewis and E. Anders, Carbynes: carriers of primordial noble gases in meteorites, Science 209 (1980) 1512. 28 A.G. Whittaker, Carbon: occurrence of carbyne forms of carbon in natural graphite, Carbon 17 (1979) 21. 29 F.P. Bundy, the P, T phase and reaction diagram for elemental carbon, 1979, J. Geophys. Res. 85 (1980) 6930. 30 J.M. Hayes, Organic constituents of meteorites, Geochim. Cosmochim. Acta 31 (1967) 1395. 31 G.P. Vd0vykin, Carbonaceous Matter of Meteorites (Organic Compounds, Diamonds, Graphite) (Nauka Press, Moscow, in Russian) English translation, NASA TT F-582 (1970) 32 E. Anders, R. Hayatsu and H.M. Studier, Organic compounds in meteorites, Science 182 (1973) 781. 33 C.E. Melton and A.A. Giardini, The composition and significance of gas released from natural diamonds from Africa and Brazil, Am. Mineral. 59 (1974) 775. 34 I.A. Petersilie, E.D. Andreeva and E.V. Sveshnikova, The organic substances in the rocks of certain alkaline massifs of Siberia, Izv. Acad. Sci. USSR, Geol. Ser. 6 (1965) 26 (English translation). 35 I.A. Petersilie and H. Sorensen, Hydrocarbon gases and bituminous substances in rocks from the Ilimaussaq alkaline intrusion, South Greenland, Lithos 3 (1970) 59. 36 E.J. Graeber, P.J. Modreski and T.M. Gerlach, Compositions of gases collected during the 1977 East Rift eruption, Kilauea, Hawaii, J. Volcanol. Geotherm Res. 5 (1979) 337. 37 D.W. Muenow, High temperature mass spectrometric gasrelease studies of Hawaiian volcanic glass: Pele's tears, Geochim. Cosmochim. Acta 37 (1973) 1551. 38 M.O. Garcia, N.W.K. Liu and D.W. Muenow, Volatiles in submarine volcanic rocks from the M;~riana Island arc and trough, Geochim. Cosmochim. Acta 43 (1979) 305. 39 J.S. Killingley and D.W. Muenow, A mass spectrometric method for the determination of the size distribution of CO 2 inclusions in olivines, Am. mineral. 59 (1974) 863. 40 J.S. Killingley and D.W. Muenow, Volatiles from Hawaiian submarine basalts determined by dynamic high temperature

232

41

42

43

44 45

46

47

48

49 50

51

52

53

mass spectrometry, Geochim. Cosmochim. Acta 39 (1975) 1467. J.G. Moore, J.N. Batchelder and C.G. Cunningham, CO 2filled vesicles in mid-oceanic basalt, J. Volcanol. Geotherm. Res. 2 (1977) 309. B.R. Frost, Mineral equilibria involving mixed-volatiles in a C-O-H fluid phase: the stabilities of graphite and siderite. Am. J. Sci. 279 (1979) 1033. R.A. Robie, B.S. Hemingway and J.R. Fisher, Thermodynamic properties of minerals and related substances at 298.15 K and I bar (105 pascals) pressure and at higher temperatures, U.S. Geol. Surv. Bull. 1452 (1978). R.H. Newton, Activity coefficients of gases, Ind. Eng. Chem. 37 (1935) 302. D.A. Hewitt, A redetermination of the fayalite-magnetitequartz equilibrium between 650 ° and 850°C, Am. J. Sci. 278 (1978) 715. M. Sato and M. Valenza, Oxygen fugacities of the layered series of the Skaergaard intrusion, East-Greenland, Am. J. Sci. 280A (1980) 134. M. Sato, Intrinsic oxygen fugacities of iron-bearing oxide and silicate minerals under low total pressure, Geol. Soc. Am. Mere. 135 (1972) 289. S. Andersson, B.I. Lundqvist and J.K. Ne~rskov, Possible mechanism for the catalytic action of Ni surfaces on the reaction 2 C O ~ C + C O 2, Proc. 7th Int. Vac. Congr., 3rd Int. Conf. Solid Surfaces, Vienna (1977) 815. R.B. Woodward and R. Hoffmann, The Conservation of Orbital Symmetry (Verlag Chemie, Weinheim, 1970) 178. L.A. Haas, S.E. Khalafalla and P.L. Weston, Jr., Kinetics of formation of carbon dioxide and carbon from carbon monoxide in presence of iron pellets, U.S. Bur. Mines, Dep. Invest. (I 968) 7064. E.T. Turkdogan and J.V. Vinters, Catalytic effect of iron on decomposition of carbon monoxide, I. Carbon deposition in H2-CO mixtures, MetaIl. Trans. 5 (1974) 11. R.G. Olsson and E.T. Turkdogan, Catalytic effect of iron on decomposition of carbon monoxide, II. Effect of additions of H E, H20, CO 2, SO 2 and H2S, Metall. Trans. 5 (1974) 21. G.D. Renshaw, C. Roscoe and P.L. Walker, Jr., Disproportionation of CO, I. Over iron and silicon-iron single crystals, J. Catal. 18 (1970) 164.

54 T.C. Tsao, K. Li and W.O. Philbrook, Kinetics of dissociation of carbon monoxide on a-iron, Proc. Conf. Metall. Soc. Can. Inst. Min. Metall., Annu. Vol. Featuring Molybdenum (1977). 55 R.J. Arculus and J.W. Delano, Intrinsic oxygen fugacities of thc earth's mantle: a new approach and results for spinel peridotites, in: Lunar and Planetary Science, XI (Lunar and Planetary Institute, Houston, Texas 1980) 28 (abstract). 56 D. Virgo, M. Rosenhauer and F.E. Huggins, Intrinsic oxygen fugacities of natural melanites and schorlomites and crystalchemical implications, Carnegie Inst. Washington Yearb. 75 (1976) 720. 57 J. Hoers, Ein Beitrag zur Isotopengeochemie des Kohlenstoffs in magmatischen Gesteinen, Contrib. Mineral. Petrol. 4l (1973) 177. 58 J. Hoefs, The carbon isotopic composition of CO 2 from fluid inclusions, Fortschr. Mineral. 52 (1975) 475. 59 P. Deines and D.P. Gold, The isotopic composition of carbonatite and kimberlite carbonates and their bearing on the isotopic composition of deep-seated carbon, Geochim. Cosmochim. Acta 37 (1973) 1709. 60 B.J. Kobelski, D.P. (}old and P. Deines, Variations in stable isotope compositions for carbon and oxygen in some South Africa Lesothan kimberlites, in: Kimberlites, Diatremes, and Diamonds: Their Geology, Petrology, and Geochemistry, F.R. Boyd and H.O.A. Meyer, eds., Proc. 2nd Int. Kimberlite Conf., Vol. 1 (American Geophysical Union, Washington, D.C., 1979) 252. 61 M.S. Lancet and E. Anders, Carbon isotopic fractionation in the Fisher-Tropsch synthesis and in meteorites, Science 170 (1970) 980. 62 F. Pineau, M. Javoy and Y. Bottinga, t3C/t2C ratios of rocks and inclusions in popping rocks of the Mid-Atlantic Ridge and their bearing on the problem of isotopic composition of deep-seated carbon, Earth Planet. Sci. Lett. 29 (1976) 413. 63 J. Javoy, F. Pincau and I. Iiyama, Experimental determination of the isotopic fractionation between gaseous CO 2 and carbon dissolved in tholeiitic magma, Contrib. Mineral. Petrol. 67 (1978) 35.