Recycling of Graphite During Himalayan Erosion: A Geological Stabilization of Carbon in the Crust

See allHide authors and affiliations

Science  07 Nov 2008:
Vol. 322, Issue 5903, pp. 943-945
DOI: 10.1126/science.1161408


At geological time scales, the role of continental erosion in the organic carbon (OC) cycle is determined by the balance between recent OC burial and petrogenic OC oxidation. Evaluating its net effect on the concentration of carbon dioxide and dioxygen in the atmosphere requires the fate of petrogenic OC to be assessed. Here, we report a multiscale (nanometer to micrometer) structural characterization of petrogenic OC in the Himalayan system. We show that graphitic carbon is preserved and buried in marine sediments, while the less graphitized forms are oxidized during fluvial transport. Radiocarbon dating indicates that 30 to 50% of the carbon initially present in the Himalayan rocks is conserved during the erosion cycle. Graphitization during metamorphism thus stabilizes carbon in the crust over geological time scales.

The burial of organic carbon (OC) in marine sediments represents the second largest sink of atmospheric CO2 after silicate weathering and subsequent carbonate precipitation (14). Rivers play a crucial role in this process by exporting a large flux of OC from the continents to the oceans (57), hence generating a long-term C sink. OC exported by rivers is, however, a mix of (i) “recent” OC (OCrecent) derived from plant detritus, associated soil organic matter, and autotrophic carbon production by aquatic plants, and (ii) petrogenic OC (OCpetro) derived from erosion of carbonaceous rocks [e.g., (811)]. Burial of the latter is a simple recycling of reduced C and has no effect on the long-term atmospheric CO2 and O2 levels. Conversely, its oxidation consumes O2 from and returns CO2 to the atmosphere, thereby counter-acting the effect of OCrecent burial. Addressing the role of continental erosion on the global C cycle thus requires assessing the fate of OCpetro during erosion and fluvial transport.

The Himalayan range undergoes an intense physical erosion producing an enormous flux of sediment (∼2 billion tons per year) carried by the Ganges-Brahmaputra (G-B) fluvial system to the Bengal Fan turbiditic system [e.g., (12)]. Recently, we showed that OC is very efficiently buried in sediments of the Bengal Fan (13). This contrasts with most large deltaic systems, including the Amazon (57). We estimated that Himalayan erosion is responsible for ∼10 to 20% of the global burial of OCrecent, implying a global impact on the OC cycle. The Himalayan example illustrates the importance of active orogens in the global C cycle and therefore offers an ideal case study to determine the fate of OCpetro during continental erosion.

Several methods have been developed to detect and quantify the most refractory pool of C in rocks, river sediments, and marine sediments, often referred to as black carbon [e.g., (1417)]. The selectivity of these methods is, however, based on the refractory character of the C particles, which is not unique to OCpetro, but also characterizes some soil OC (charcoals) or anthropic C (e.g., soots). Consequently, they isolate a much larger pool of C and thus overestimate the petrogenic component. Here, we combine Raman microspectroscopy (RM) and transmission electron microscopy (TEM) to (i) detect OCpetro in river and marine sediments and (ii) characterize its multiscale structural organization (nm to μm). RM allows in situ analysis, preserving the textural information of OC (particle size and shape) itself, but also its relations with minerals (aggregates and inclusions). Combining 002 lattice-fringes imaging and electron diffraction, TEM gives an insight into the nm-scale structure of the aromatic skeleton composing OCpetro (16, 17). Finally, we use radiocarbon dating of bulk acid-insoluble OC to quantify the proportion of OCpetro and OCrecent in river sediments (17).

We have analyzed a sample set covering the whole erosional and depositional system, including Himalayan source rocks, G-B river sediments, and Bengal Fan sediments (tables S1 and S2). To study the fate of OCpetro during fluvial transport, river sediments were collected from the Himalayan range to the Bangladesh delta (fig. S1). In the G-B system, suspended and bed sediments define a continuum from fine, clay-rich, surface-suspended sediments to coarse, quartz-rich, bed sediments. OC content increases with decreasing grain size and increasing proportions of phyllosilicates (18). To take this variability into account, we analyzed OC in suspended sediments collected along depth profiles within the river as well as from dredged riverbed sediments. To describe the Bengal Fan turbiditic system, we used sediments drilled by the research vessel (RV) Sonne cruise 93 (19) in the shelf, the middle fan channel levee system, and the deep fan (fig. S2).

OCpetro has been detected by RM and TEM in all investigated sediments, including the most distal regions of the Bengal Fan (Fig. 1). Using RM, OCpetro was identified in various forms: (i) discrete “free” particles from a few to several tens of μm size; (ii) inclusions within quartz, calcite, and metamorphic minerals; and (iii) aggregates with minerals, mostly micas. The latter two have an obvious petrogenic origin, whereas in the former the most disordered C may be OCrecent (soot or charcoal) or OCpetro from very low-grade metamorphic rocks (17).

Fig. 1.

RM and TEM characterization of OCpetro in riverine and marine sediments. (A) Image of 002 lattice-fringes (002LF) of disordered and microporous OCpetro in Narayani bed load. (B) Image (002LF) of graphitic C in Narayani bed load. (C) Low-magnification (LM) image of a graphite particle in Narayani suspended load. (D) Image (002LF) of graphitic OCpetro in Lower Meghna bed load. (E) Image (002LF) of graphitic OCpetro in distal Bengal Fan sediment. (F) LM image of graphite particles in distal Bengal Fan sediment. In (A), (C), (D), and (E), the orange inset is a zoom (5-nm side) located on the main image. (G) Selection of representative Raman spectra from Himalayan rivers (red), Lower Meghna (green), and Bengal Fan sediments (blue). These spectra were obtained from both individual OCpetro particles (no mineral contribution in the spectrum) and OCpetro inclusions/aggregates within minerals (mineral contribution in the spectrum as depicted).

RM shows a large structural variety of OCpetro in bed and suspended loads from Himalayan rivers (Fig. 1). This is confirmed by TEM investigations that reveal the presence of turbostratic disordered OCpetro to perfectly crystalline graphite in these samples. The structural variety of OCpetro reflects the zoning of metamorphism in the source rocks of the Himalayan range [e.g., (20)], from low- to high-grade metamorphism (fig. S1). Similar structural variety is observed within the vertical depth profiles collected at the outflow of the range and in Bangladesh, although the less-ordered OCpetro is rarely observed. In all marine sediments, both RM and TEM investigations show that OCpetro detected by these methods is almost exclusively highly ordered polycrystalline or pristine graphite (Fig. 1) and that disordered OCpetro is rare (17). Because we do not observe any segregation in terms of crystallinity of OCpetro along vertical depth profiles in rivers, we posit that OCpetro is homogenously transported as a continuum through the river section. However, oxidation of the less-graphitic OCpetro occurs within the Himalayan range and during the floodplain transit, because these forms are virtually absent in the Bengal Fan sediments, which is consistent with OCrecent oxidation and replacement documented in the Gangetic floodplain (18). Such selective oxidation might be related to the higher chemical reactivity due to the presence of atoms other than C (primarily H and O) and enhanced by the high nano- to micro-porosity of disordered OCpetro compared with graphite.

Estimating the proportion of OC contained in the Himalayan rocks that is ultimately oxidized during the erosion cycle requires quantifying the amount of OCpetro in the sediments exported by G-B rivers. We measured the 14C composition of bulk acid-insoluble OC (expressed as the fraction of modern C, Fm) in sediments collected along depth profiles. OCpetro is radiocarbon free (Fm = 0), whereas OCrecent has variable amounts of radiocarbon (Fm > 0) depending on its residence time in the basin. In the Himalayan basin, a consequence of intense physical erosion is that none of the OCrecent particles contained in soils is radiocarbon free, which gives to OCpetro a unique geochemical signature (Fm = 0).

The results are presented in Fig. 2 as modern OC content (Fm × %OC) plotted as a function of the total OC content (see also fig. S4 and table S1). Sediments collected along depth profiles define linear trends in this diagram. Suspended and bed sediments from each depth profile thus have identical amount of OCpetro (17). This consistency derives from the generation of sediment continuums by mixing processes during fluvial transport. A statistical analysis of the trends drawn in Fig. 2 allows the determination of the amount of OCpetro (17). Large trans-Himalayan rivers sampled at the outflow of the range appear to carry variable amounts of OCpetro. We estimate an OCpetro content of 0.05% and 0.03% for the Narayani and Kosi, respectively. This variability likely derives from intrinsic characteristics of the drainage basins. Our data consistently indicate an OCpetro content of 0.02 to 0.03% in the sediments exported by the Brahmaputra and the Ganges sampled close to their mouths in the Bangladesh delta. We therefore estimate that sediments delivered to the Bay of Bengal contain ∼0.025% of OCpetro derived from the erosion of Himalayan rocks. Because the oxidation of OC in the marine system before deposition in the Bengal Fan is negligible (13), the OCpetro content of these sediments must be comparable to that of the sediments exported by G-B rivers. Moreover, for two sediments deposited in the channel-levee system of the Bengal Fan, we corrected the 14C composition of OC for their respective deposition age determined by 14C dating of foraminifera (21). Their position in Fig. 2 is highly consistent with sediments delivered by G-B rivers, indicating that they have similar content of OCpetro. Consequently, we estimate that Bengal Fan sediments contain ∼0.025% of OCpetro. To evaluate the fate of petrogenic OC during the Himalayan erosion cycle, this value must be compared to the mean OC content of Himalayan rocks. The latter has been estimated to be 0.05 to 0.08%, on the basis of individual rock samples and composite gravels extracted from the bed of Himalayan rivers (13, 22, 23). Based on these figures, at least 30% (40 ± 10) of the OC contained in the Himalayan rocks appears to be preserved and recycled during the erosion cycle, mostly through burial of highly graphitic carbon in the Bengal Fan sediments. In turn, the oxidative flux of OCpetro associated with Himalayan erosion is estimated to be 0.67 (±0.25) × 1011 mol/year, which is comparable to the long-term CO2 consumption by silicate weathering (0.6 × 1011 mol/year (24)) but one order of magnitude lower than the CO2 consumption by OCrecent burial in the Bengal Fan [3.1 (±0.3) × 1011 mol/year (13)].

Fig. 2.

(Top) Modern OC content (Fm × %OC) of sediments collected over depth profiles in rivers from the G-B system as a function of their total OC content. In this representation, sediments with similar amount of OCpetro define linear trends, whose intercept with the x axis gives %OCpetro (17). Each solid line represents the best linear fit defined by sediments from the same depth profile. Himalayan rivers sediments (green, Narayani; yellow, Kosi) have variable amounts of OCpetro, whereas Ganges (blue) and Brahmaputra (red) in Bangladesh have similar amounts of OCpetro around an average value of 0.025%. Radiocarbon composition of Bengal Fan sediments (hexagons) is comparable with that of river sediments, indicating they have similar content of OCpetro. (Bottom) Total OC content of Himalayan source rocks (black, High-Himalayan Crystalline; light gray, Lesser Himalaya; dark gray, composite gravel samples) from (13). The average OC content of source rocks ranges between 0.05 and 0.08%.

The Himalayan example highlights the role of mountain building in the long-term stabilization of Earth climate and atmospheric composition. Active orogenesis not only promotes CO2 consumption by both silicate weathering and terrestrial organic carbon burial but also stabilizes C in the crust over geological time scales through graphitization of OC (Fig. 3). Metamorphism during orogenesis actually transforms labile photosynthetic C into refractory petrogenic C through graphitization, the efficiency of this transformation being mainly controlled by the metamorphic temperature. The selective preservation of graphite during erosion and weathering indicates that graphitization occurring during metamorphism subtracts C from the external cycle (atmosphere-biosphere-ocean) and locks it into the geological cycle (Fig. 3). Mechanisms capable of returning graphitic C into the atmosphere involve subduction and volcanism, both having a very long characteristic time compared with surface processes. At geological time scales, the graphitization thus maintains an imbalance between photosynthesis and respiration by locking reduced C into the crust, hence promoting both consumption of CO2 and accumulation of O2 in the atmosphere.

Fig. 3.

A model of the long-term OC cycle highlighting the role of graphitization. The graphite formed during high-temperature (>500°C) metamorphism is preserved during continental erosion and returns to the atmosphere only throughout volcanism. Graphitization thus stabilizes reduced C, confining it to a geological subcycle and maintaining an imbalance between photosynthesis and respiration/oxidation over geological time scales.

Supporting Online Material

Material and Methods

SOM Text

Figs. S1 to S4

Tables S1 and S2


References and Notes

Stay Connected to Science

Navigate This Article