Instruments and Methods A high-precision method for measurement of paleoatmospheric CO 2 in small polar ice samples Jinho AHN, Edward J. BROOK, Kate HOWELL Department of Geosciences, Oregon State University, Corvallis, Oregon 97331-5506, USA E-mail: jinhoahn@gmail.com ABSTRACT. We describe a high-precision method, now in use in our laboratory, for measuring the CO 2 mixing ratio of ancient air trapped in polar ice cores. Occluded air in ice samples weighing 8–15 g is liberated by crushing with steel pins at –358C and trapped at –2638C in a cryogenic cold trap. CO 2 in the extracted air is analyzed using gas chromatography. Replicate measurements for several samples of high- quality ice from the Siple Dome and Taylor Dome Antarctic ice cores have pooled standard deviations of <0.9 ppm. This high-precision technique is directly applicable to high-temporal-resolution studies for detection of small CO 2 variations, for example CO 2 variations of a few parts per million on millennial to decadal scales. INTRODUCTION Carbon dioxide is the most important greenhouse gas directly impacted by human activities. The atmospheric CO 2 mixing ratio has increased 25% since the industrial revolution (Etheridge and others, 1996; MacFarling Meure and others, 2006), and its continuing increase will con- tribute a major fraction of future global warming (Solomon and others, 2007). The transfer of carbon between reservoirs in the carbon cycle ultimately controls the fate of fossil-fuel- derived CO 2 , but how climate and the carbon cycle are linked is only partly understood. Studies of how atmospheric CO 2 and climate are related in geologic history contribute to our understanding of Earth’s climate system. Ice cores are unique archives that allow direct measurement of atmos- pheric CO 2 content over 800 ka (Macfarling Meure and others, 2006; Lu ¨ thi and others, 2008), expanding instru- mental CO 2 records, which began only in 1958 (C.D. Keeling and T.P. Whorf, http://gcmd.nasa.gov/ records/GCMD_CDIAC_CO2_SIO.html). CO 2 records de- rived from Antarctic ice cores are widely demonstrated to be representative of atmospheric concentrations over several glacial–interglacial cycles (Fischer and others, 1999; Petit and others, 1999; Kawamura and others, 2003; Siegenthaler and others, 2005; Lu ¨thi and others, 2008). However, this is not the case for CO 2 records from Greenland ice cores where dust is present in high concentration. The dust includes carbonates (Anklin and others, 1995, 1997; Barnola and others, 1995; Smith and others, 1997a,b) and organic compounds (Tschumi and Stauffer, 2000) within the ice core, giving rise to in situ CO 2 production. There is general agreement of atmospheric CO 2 trends within deep Antarctic ice cores on millennial and longer timescales. However, disagreements in CO 2 concentrations of up to 5– 20 ppm between comparable cores and laboratories have been reported (Fischer and others, 1999; Petit and others, 1999; Stauffer, 2006). Given these discrepancies, there is an increasing demand for high-precision measurements combined with high- temporal-resolution CO 2 concentration records, to decipher the exact mechanisms controlling Earth’s climate system on millennial or sub-millennial timescales. For example, under- standing how atmospheric CO 2 varied with respect to abrupt climate change during the last ice age (e.g. Stauffer and others, 1998; Ahn and Brook, 2007, 2008) or during climate cycles of the late Holocene (MacFarling Meure and others, 2006) requires, in some cases, decadal data with a precision of at least 2 ppm, and preferably better. In addition, in deep ice-coring projects ice availability is limited, requiring analytical techniques suitable for small samples of <10 g. We note that the air age distribution also limits time resolution. Usually, low accumulations at coring sites give large age distributions and limit time resolution of gas records in ice cores (Spahni and others, 2003). Analysis of CO 2 in a small quantity of ice is challenging, primarily because of variable desorption and adsorption of CO 2 in the extraction line and chamber due to water-vapor content (Zumbrunn and others, 1982). Air should be extracted using a ‘dry’ mechanical method rather than simpler ‘wet’ extraction (melting under vacuum), as melting may induce a carbonate–acid reaction and increase CO 2 levels in the sample (Delmas and others, 1980). Conven- tional dry extraction methods include crushing ice with metal needles (Zumbrunn and others, 1982; Wahlen and others, 1991) or metal balls (Delmas and others, 1980), milling with blades (Moor and Stauffer, 1984; Nakazawa and others, 1993) or a ‘cheese grater’ (Etheridge and others, 1988). All methods require sustained cold conditions in the vacuum chamber. We chose to build a ‘needle crusher’ (Fig. 1) capable of crushing small (10 g) ice samples. The milling techniques and crushing with metal balls at present require relatively large samples (>500 and 40–50 g for blades and metal balls, respectively). The infrared laser spectroscopy (IRLS) method has been used successfully in several laboratories for analysis of air (<1 cm 3 STP) extracted from small ice samples (<10 g; Zumbrunn and others, 1982; Wahlen and others, 1991). However, the analytical precision with IRLS is generally 0.5% (1'), corresponding to 1.5 ppm CO 2 for measure- ments of bubble-free single-crystal ice (Monnin and others, 2001; Lu ¨thi and others, 2008). Instead, we used gas chromatography (GC) to analyze the small amount of air Journal of Glaciology, Vol. 55, No. 191, 2009 499