Download 277.02 Kb.

Clumped isotope measurements of small carbonate samples using a highefficiency dualreservoir technique Sierra V. Petersen^{*}, Daniel P. Schrag Harvard University, Department of Earth and Planetary Sciences, 20 Oxford Street, Cambridge, Massachusetts 02138, USA. * Author for Correspondence: speters@fas.harvard.edu Supplementary Material Figure S1: Diagram of 10mL reservoir attached to reference bellows. The MAT253 has a builtin ¼” Swagelok compression fitting as the output to the reference bellows. We attach the 10mL stainless steel reservoir (Swagelok piece # SS4CDTW10) to the fused silica capillaries (~1m length, 110m inner diameter, SGE # 0624459) using two converter pieces (1/4” male to 3/8” female, Swagelok piece # SS600R4, and 3/8” female to 1/16” female, Swagelok piece # SS60061). The capillaries are connected to the 1/16” Swagelok fitting using a graphiteVespel composite ferrule (SGE # 072663). On the sample side, the same 2 Swagelok connectors and 10mL reservoir were used to form identical volumes from which the gas bleeds down. However, the internal volume of the inlet valve adjacent to the sample reservoir is larger than the internal volume of the MAT 253 valve adjacent to the reference reservoir, so 87 glass beads (borosilicate, 3mm diameter, similar to VWR#26396630) were placed inside the sample reservoir to balance this volume difference.
Table S1: Slopes and 1 SE estimates from data in Figure 4 (_{48} vs _{47}), shown for each sample type regressed individually, compared with the slope found using the group fit. There is a slight dependence of the slope on the average _{47} composition of each sample type, with the least clumped samples (PUCO21000C) having the steeper slope, and the more clumped (Ref Gas, RTG) having a shallower slope. Figure S2: _{48} vs _{47raw} and vs _{47RFAC}. The fractionation relationship is preserved through the correction to the universal reference frame, with essentially no change in slope. The carbonate data shown here are a subset of data from measurement period #2 which were run at a starting voltage of m/z 47 = 33003800mV. These carbonates were corrected using only heated and equilibrated gases run within the same voltage range. If the difference in running voltage of the samples and gas standards were causing the observed fractionation, this reference frame correction done with similarsized carbonates and standard gases should remove the fractionation and flatten out the data in this plot. The fact that the slope is unchanged suggests that differences in running voltage between samples and standards do not cause the observed fractionation.
Table S2: Fitted slopes and intercepts for the four lines shown in Figure S2. The slope is essentially unchanged by the correction to the universal reference frame, and is statistically different from zero in all cases, suggesting that the reference frame correction does not remove the fractionation slope, even when the gases and carbonates are run at the same voltage. Figure S3: _{48} vs _{47raw} and _{48} vs _{47RFAC} for both CM2 and RTG for the 4 measurement periods (MP1 = 09/22/13 to 10/03/13; MP2 = 10/07/13 to 12/18/13; MP3 = 01/06/14 to 02/14/14; MP4 = 02/18/14 to 03/28/14). The PPQ trap material was changed during MP3 and did not have a large influence on the slope. Horizontal grey lines indicate the published value for each standard. RTG has few replicates in MP1, resulting in a more divergent slope.
Table S3: Slopes and intercepts of the lines shown in Figure S3 (_{48} vs _{47raw} and _{48} vs _{47RFAC}), with errors (1 SE). There is a slight noticeable offset between the slopes fitred to CM2 data and that fitted to RTG. Figure S4: ^{18}O (left) and ^{13}C (right) values vs _{48} for all CM2 (blue), RTG (green), and NBS19 (orange) points measured over 4 different measurement periods. No significant correlation is observed between _{48} and the stable isotope ratios, unlike between _{48} and _{47}. Plots of ^{13}C and ^{18}O values vs _{47RFAC}, _{47corr}, and ^{48} values look very similar because of the strong correlation between those quantities and _{48}. Figure S5: ^{18}O (left) and ^{13}C (right) values vs sample size for all CM2 (blue), RTG (green), and NBS19 (orange) points measured over 4 different measurement periods. There is no correlation between the stable isotopes and sample size. Plots of ^{13}C and ^{18}O values vs initial m/z 44 or m/z 47 look very similar because of the strong correlation between those quantities and sample size. Figure S6: _{48} vs _{47RFAC} (top) and Initial m/z 47 intensity vs _{48} (bottom), separated by sample size for CM2 runs over all four measurement periods. Smaller sample sizes tend to show higher _{48},_{ }and therefore _{47RFAC}, values, whereas larger samples tend to show lower _{48} and _{47RFAC} values, although a few points do not follow this. Grey dashed lines delineate “no fractionation”, or (_{48}) = 0, covering a range of values for the 4 measurement periods. A wider range of _{48} values would be considered uncontaminated using traditional tests (±1.5‰ around dashed lines). Figure S7: Residual yield (difference between observed yield, measured as increase in source vacuum gauge pressure, and fitted line shown in Figure 3a) vs _{48} for all carbonate data run over all four measurement periods. There is no correlation between residual yield and _{48}, indicating that partial yield is not causing the fractionation. If that were the case, we would expect to see the highest _{48} values occurring either at the highest residual yield (contaminant being added) or the lowest (fractionation occurring during loss of some gas). Instead we see nearzero residual values showing the highest _{48}. Figure S8: _{48} vs _{47raw} for runs of reference gas treated as a sample under three configurations: 1) reference gas passed through the full inlet; 2) reference gas frozen into the small Utrap (bypassing the PPQ trap); 3) reference gas expanded into the small Utrap (bypassing the PPQ trap). When run through the full inlet, the reference gas shows a similar slope to carbonate samples and heated gases (~0.05, see Figure 4, and Table S1). In configuration 2 (No PPQ, Freeze), a strong relationship exists between _{48} and _{47}, but with a slope that is twice as steep (0.111). Configuration 3 (No PPQ, Expand) shows no significant relationship and the data points are mostly clumped around the origin, the values that we would expect if there were no fractionation. 
search 