Fast Gas Chromatography Combustion Isotope Ratio Mass Spectrometry

Authors:

Abstract
We report here the first coupling of fast GC to IRMS, in a system capable of 250 ms peak widths (fwhm) at 1 mL/min flow rates, one-fifth as narrow as any previously reported GCC-IRMS system. We developed an optimized postcolumn interface that results in minimal peak broadening, using a programmable temperature vaporization injector in place of a rotary valve or backflush system to divert solvent, a narrow capillary combustion reactor followed by a cryogenic water trap with narrow-bore (600 pmol C on column; precision for the narrow peaks (250 ms) was considerably better for injections 13C) < 1‰ was achievable for injections of 5−15 pmol on column for 250 ms wide peaks, 10-fold better precision than 2500 ms wide peaks, and within a factor of 3 of the counting statistics limit. For a mixture of 15 fatty acid methyl esters (FAME), 1.5 nmol C of each on column yielded typical SD(δ13Cpdb) = 0.4‰ for fast GC and 0.5‰ for conventional GC. For 14 of the 15 FAME, δ13C values between the two systems were within ± 1.5‰ and not significantly different. Fast GCC-IRMS required one-third the run time (450 s vs 1400 s) to achieve comparable resolution. Mean peak widths for fast GCC-IRMS of the FAME were 720 ms, compared to 650 ms by fast GC with flame ionization detection. At a 15-fold dilution (100 pmol C on column for each FAME), fast GCC-IRMS achieved ∼2-fold better precision and accuracy than similar injections on conventional GCC-IRMS. Finally, a mixture of 10 steroids (∼7 nmol C (100 ng) each on column) was analyzed with mean precision of SD(δ13C) = 0.2‰ in 620 s by fast GCC-IRMS, while conventional GCC-IRMS required 1200 s and achieved poorer resolution. δ13C values for the two system were similar (Δδ13C ≤ 2‰ for all steroids), indicating that accuracy is not compromised. In summary, fast GCC-IRMS can achieve similar precision to conventional GC with considerable time savings for standard sample sizes (>1 nmol C) and achieves modest precision (∼1‰) near the counting statistics limit on low level components.

This publication has 25 references indexed in Scilit: