Rapid Sample Preparation14
Jan 2004 ___________________________________________________________________________________________________________ 14 Jan 2004 The conclusions from our analyses raise concerns about using the Brendel method for use with 18O and 14C isotopes in particular and as a method in general. Using wood, root and leaf samples from a diversty of plants we've found that the Brendel method appears to produce the least pure form of cellulose according to our stable isotope and 13C-NMR analysis and produces 18O values that are enriched relative to values one gets if they use the holo- and alpha-cellulose soxhlet methods. While it produces 13C results that are often similar to the alpha- and holo-cellulose soxhlet methods (especially for wood) the results are inconsistent. Modifiing the Brendel method helps with many of these problems, but our analyses of wood, leaf and root cellulose shows that the soxhlet method still produces the most reliable results. Several labs now have a "rapid" batch soxhlet method that makes it easier to process many samples in a relatively short time. We know that others have used the Brendel method and seem generally pleased with it, but our results are showing us that it is not the best method if you want the highest possible precision for stable isotope analyses and it produces erroneous results for 14C. Todd Dawson Center for Stable Isotope
Biogeochemistry ___________________________________________________________________________________________________________
refs, -- Mailing address: ___________________________________________________________________________________________________________ 16 Jan 2004 Additional results for both oxygen and carbon (stable and radio-) analyses have just been published in EPSL (please contact Pascale Poussart at Harvard, poussart@eps.harvard.edu, for a reprint). We have not made any purity tests of the product, such as are described in Brendel's paper for c13, and described by Todd in his email. Mike Lott made some intercomparisons between this modified Brendel method and the Jayme-Wise extraction in summer of 2002. He found no significant difference in results for o18 measurements using continuous flow analysis. We not not found any problems with reproducibility of measurements for either o18 or c13, within typical measurement precision of 0.2-0.3 permil SMOW on a continuous flow instrument. Todd, I would be interested to discuss a draft of your manuscript on method intercomparison. I am especially interested in why and how radiocarbon results (but not c13?) are affected by the Brendel protocol, but not by other techniques. I would also like to see how your purity results differ from Oliver's. Is this mainly due to the diverse and variable mixture of C compounds in soils and other heterogeneous samples compared to something relatively simple, like wood, or is it in the extraction chemistry? We have found both reasonable and strange radiocarbon results after processing wood via modified-Brendel. On wood from Costa Rica from the post-bomb period, we found prebomb (500-1000BP) ages which were replicated within c14 error by WHOI and CAMS radiocarbon labs. I chalked these results up to an unknown but large fossil radiocarbon contamination; maybe due to alpha-cellulose extraction? But in wood from Indonesia, Pascale Poussart has published radiocarbon results which are consistent within error with an age model developed from oxygen isotope data. (See our EPSL and GCA papers for the hypothesis behind this approach, based on work of Roden et al.) Then again, Pascale has also found radiocarbon ages which are too young, based on samples from crossdated (annual-ringed) teaks from Indonesia. Pascale can provide more details on those results and her thinking on their interpretation. Sincerely, ____________________________________________________________________________________________________________ 18 Jan 2004 I regularly use two soxhlet
setups at one time (a larger soxhlet would serve the same purpose).
The number of samples processed in each soxhlet is determined by the
mean sample size, but I usually extract 80-100 samples in less than
a week. If necessary, two extractions can be Regards, _________________________________________________________________________________________ |
|
|