Accelerator mass spectrometry (AMS) measurements of environmental C-14 have
been employed during the past decade at the several micromole level (tens
of mug carbon), but advanced research in the atmospheric and marine science
s demands still higher (mug) sensitivity, an extreme example being the dete
rmination of C-14 in elemental or "black" carbon (BC) at levels of 2-10 mug
per kg of Greenland snow and ice (Currie et al., 1998). A fundamental limi
tation for C-14 AMS is Poisson counting statistics, which sets in at about
1 mug modern-C. Using the small sample (25 mug) AMS target preparation faci
lity at NOSAMS (Pearson et al., 1998), and the microsample combustion-dilut
ion facility at NIST, we have demonstrated an intrinsic modern-C quantifica
tion limit (m(Q)) of ca. 0.9 mug, based on a 1-parameter fit to the empiric
al AMS variance function. (For environmental C-14, the modern carbon quanti
fication limit is defined as that mass (m(Q)) corresponding to 10% relative
standard deviation (rsd) for the fraction of modern carbon, sigma(*f(M))/(
f(M).) Stringent control, required for quantitative dilution factors (DL),
is achieved with the NIST on-line manometric/mass spectrometry facility tha
t compensates also for unsuspected trace impurities from vigorous chemical
processing (e.g., acid digestion). Our current combustion blank is trivial
(mean: 0.16+/-0.02 mug C, n = 13) but lognormally distributed (dispersion [
sigma]: 0.07 +/- 0.01 mug). An iterative numerical expression is introduced
to assess the quantitative impacts of fossil and modern carbon blank compo
nents on m(Q); and a new "clean chemistry" BC processing system is describe
d for the minimization of such blanks. For the assay of soot carbon in Gree
nland snow/ice, the overall processing blank has been reduced from nearly 7
mug total carbon to less than 1 mug, and is undetectable for BC. (C) 2000
Elsevier Science B.V. All rights reserved.