A modified extraction technique for liberating occluded gases from ice cores

We have developed a new dry extraction technique to extract air from large pieces of glacial ice. The primary reason for developing this technique was to be able to perform a single extraction and measure a single sample of a ir from an ice core sample for as many atmospheric constituents as possible . The procedure is modeled after the dry extraction -"cheese grater" design of Etheridge et al. [1988]. Extracted air samples are analyzed for the ele mental and isotopic composition of O-2 and N-2 as well as the CH4 concentra tion. Extensive experimental work to determine the integrity of the extract ion procedure yielded blank values and external precision which are compara ble with existing extraction procedures. Overall external precision for del ta (O2)/N-2, delta O-18 of O-2, and delta N-15 of NZ analyses is +/-2.1 par ts per thousand, +/-0.074 parts per thousand, and +/-0.045 parts per thousa nd, respectively. Variable delta N-O2/(2) results from ice which is either completely bubbly or clathrated are in good agreement with previous measure ments. Variable delta (O2)/N-2 results from Greenland Ice Sheet Project (GI SP) II ice samples spanning the clathrate formation region (1000-1500 m) ar e markedly different from previous results obtained with a "wet" extraction procedure, We attribute the differences to variable (O2)/N-2 ratios in bub bles and clathrates in the clathrate formation region combined with a 21% d ifference in our extraction efficiency for bubbly versus clathrated ice. Th e overall uncertainty and blank value for CH4 measurements are +/-19 ppb an d 16 ppb, respectively. CH4 concentrations for ice between 115 and 140 mete rs below the surface (mbs) from the GISP II ice core appear to be 4.3% high er than the average value measured by five other laboratories. We attribute our elevated values to uncertainties in the actual concentration of our wo rking standard and small differences in the CH4 concentration of the libera ted air relative to the total air trapped in ice. Our corrected CH4 data sp anning the last 25 kyr are indistinguishable from the Brook ef al, [1996] C H4 record from the same period.