C-HE SYSTEMATICS IN HOTSPOT XENOLITHS - IMPLICATIONS FOR MANTLE CARBON CONTENTS AND CARBON RECYCLING

We have measured the carbon and helium abundances and isotopic composi tions of high-pressure carbon dioxide fluid inclusions in ultramafic x enoliths from oceanic hotspot volcanos to examine the extent to which He and C are separated by igneous processes, and to determine whether or not the 'undegassed' isotopic character of hotspot helium extends t o carbon. These measurements place limits on upper mantle C contents a nd the on the fate of carbon recycled at subduction zones. Xenolith fl uid inclusions from the Loihi-Hawaii, Reunion and Kerguelen hotspots e xhibit carbon isotopic compositions (deltaC-13 = - 1.6 to - 10.8 parts per thousand) similar to, although somewhat more variable, than MORB. The greater larger range can be ascribed to physicochemical processes associated with the volcanic systems and there is no evidence that ho tspots and ridges tap isotopically distinct carbon sources. CO2 and He abundances vary by more than three orders of magnitude (1-260 ppm C a nd 10(-9)-10(-6) cc STP He/g) and are strongly correlated. The samples ' C/He-3 ratios (2-20 x 10(9)) largely overlap with mid-ocean ridge ba salt values (1-7 x 10(9)). The small overall scatter of C/He-3 ratios argues against residual carbon phases during melting and against large -scale diffusive He transport in the mantle. In contrast to these simi larities, most hotspot xenoliths and basalts with high He-3/He-4 ratio s have higher C/He-4 ratios (4-40 x 10(4), uncorrelated with C or He a bundance) than mid-ocean ridge basalts (0.5-7 x 10(4)). Given the simi larity in C/He-3 ratios, the lower C/He-4 ratios of MORB with respect to hotspot sources must be produced by radiogenic He-4 production. The measured C/He-4 ratios suggest that the upper mantle carbon content m ust be less than 500 ppm C and is probably in the range 50-250 ppm C. The upper limit relies only on the observed MORB C/He-4 ratios, a maxi mum mantle U concentration of 26 ppb, and the conclusion from helium i sotopic compositions that most upper mantle He-4 must be radiogenic ra ther than trapped primordial 4He. The lower C range relies on the dyna mics of the two-reservoir mantle model proposed by Kellog and Wasserbu rg 1! to explain ridge-hotspot He isotopic differences, and on the hy pothesis that the upper (ridge source) and lower (hotspot source) mant le reservoirs had similar initial C/He-3 and He-3/He-4 ratios. Within the framework of an initially homogeneous mantle which has differentia ted into two reservoirs, significant carbon recycling to the upper man tle is inconsistent with the similarity of hotspot and ridge C/He-3 ra tios. However, recycling of carbon to the lower mantle is consistent w ith C-He systematics and can account for both the similarity of ridge and hotspot carbon isotopic compositions and the low exospheric C inve ntory.