Essential methodological improvements in the oxygen isotope ratio analysis of N-containing organic compounds

The quantitative conversion of organically bound oxygen into CO, a prerequisite for the 18O/16O analysis of organic compounds, is generally performed by high‐temperature conversion in the presence of carbon at ∼1450°C. Since this high‐temperature procedure demands complicated and expensive equipment, a lower temperature method that could be utilized on standard elemental analyzers was evaluated. By substituting glassy carbon with carbon black, the conversion temperature could be reduced to 1170°C. However, regardless of the temperature, N‐containing compounds yielded incorrect results, despite quantitative conversion of the bound oxygen into CO. We believe that the problems were partially caused by interfering gases produced by a secondary decomposition of N‐ and C‐containing polymers formed during the decomposition of the analyte. In order to overcome the interference, we replaced the gas chromatographic (GC) separation of CO and N2 by reversible CO adsorption, yielding the possibility of collecting and purifying the CO more efficiently. After CO collection, the interfering gases were vented by means of a specific stream diverter, thus preventing them from entering the trap and the mass spectrometer. Simultaneously, a make‐up He flow was used to purge the gas‐specific trap before the desorption of the CO and its subsequent mass spectrometric analysis. Furthermore, the formation of interfering gases was reduced by the use of polyethylene as an additive for analytes with a N:O ratio greater than 1. These methodological modifications to the thermal conversion of N‐containing analytes, depending on their structure or O:N ratio, led to satisfactory results and showed that it was possible to optimize the conditions for their individual oxygen isotope ratio analysis, even at 1170°C. With these methodological modifications, correct and precise δ18O results were obtained on N‐containing analytes even at 1170°C. Differences from the expected standard values were below ±1‰ with standard deviations of the analysis