Its been a long time (over 20 years) but I recall that the major 
problem was the production of CO.  I used a Tesla coil to convert the 
CO to CO2.  CO is produced when the furnace is to hot.  I recall that 
there was an optimum temperature for best yield.




You wrote: 
>
>
>Dear Isogeochemists,
>
>About two years ago we built a vacuum extraction line to produce CO2
>from sulphates for oxygen isotopic analysis. The line is built 
following Rafter's (1967) principles, following the design at SURRC 
(Glasgow), whose
>line has been in succesful operation for some years now. We use an 
>induction heater to combust the sample, and a capacitance manometer to
>measure the yield of CO2 generated.
>
>The first few runs gave erratic results, with flashing being observed 
in
>the quartz reaction vessel. These problems were solved by changing to 
a
>reliable spectrographic-grade graphite supplyer, and by replacing the
>induction heater. Our original induction heater was a fixed-frequency,
>2 MHz., 0.8 kW unit that also caused plasma formation, together with
>strong isotopic fractionation, when used for D/H work. We have 
replaced
>it with a variable frequency, 50 to 225 kHz, 1 kW unit, that works 
well,
>at least as far as D/H work is concerned. **BEWARE any of you that 
intend
>to buy an induction heater for D/H analyses: Do not make my own 
mistake!
>It was a 8,000 Pounds Sterling error!**
>
>After receiving the new induction heater, and sucessfully testing it 
for
>Hydrogen, we have gone back to sulphate-O extractions. Now we get
>reproducible results, but still have problems: We can not get a delta
>18-O value of 9.3/9.4 permil (depending on reference value accepted) 
for
>NBS-127, and our yields are low.
>
>Disregarding all extractions with less than 75% yield we get, for 
NBS-127,
>a value of 8.59 +/- 0.25 permil (n=25). However, there seems to exist 
a
>relationship (although not a proper correlation!) between yield, 
>reproducibility and isotopic value, such that 20 analyses with yield
>between 75 and 92% give 8.57 +/- 0.27 permil, whereas 5 analyses with
>yield above 92% gave 8.69 +/- 0.11 permil.
>
>Some other samples (non-standards) of known isotopic composition are 
also
>consistently lighter than expected. However, within-sample 
reproducibility
>(including extraction) is again good.
>
>When trying to deal with these problems, we have followed the 
indications
>made in the literature available to us on the subject (our University
>only has journals from the mid-70's onwards).
>
>As said above, our method is based in that of Rafter's (1967). For 
NBS-127
>we subjected the sample to no chemical treatment. For the "unknown"
>(gypsum) samples, we used dissolution in 2.5M HCl, filtration and
>precipitation as BaSO4 with 5% BaCl2, washing and drying.
>The clean, dry sample was mixed with 1:1 spectrographic-grade graphite
>onto a pre-degassed ZGS Pt-crucible, that is loaded onto a quartz 
reaction
>tube, and the whole degassed overnight under vacuum at 200 deg. C. On 
>extracion, the sample is again degassed on line by using the induction
>heater at low power for 10 minutes (We do not know the exact 
temperature,
>although we are in the process of acquiring an infra-red t/c to this 
end).
>
>In succesive tests to try to improve results, on-line degassing was 
>increased to 1 hr at 400 deg. C, and lately, following Lang's et al. 
(1989)
>reccomendations, to 1 hr at 600 deg.C (in a tube furnace; always after
>overnight degassing at 200 deg.C).
>
>In order to discard Sakai and Krouse's (1971) concern that there might
>be isotopic exchange between product CO + CO2 and the hot quartz 
walls,
>we have replaced the original tulip-shaped, quartz crucible holder by
>a Pt-rod spiral, with NO apreciable difference in results. Exchange 
with
>the outside-tube quartz walls of the reaction vessel would be expected
>to be negligible at its near-LN2 temperature.
>
>Two additional modifications have been helpful in increasing yield:
>First, we have increased the graphite:sample ratio from 1:1 to 3:1, 
and
>second, we have added a lid to the crucible. Any insight on the 
>explanation of why this increased yield is wellcome, as well as 
indications
>on the reaction mechanism between graphite and BaSO4. How important is
>graphite grain-size? If this is indeed important, how to reduce its 
grain-
>size to a minimum? We have observed under the microscope that graphite
>and BaSO4 remain as discrete, isolated materials. What is meant by
>"intimate mixture" in the literature?
>
>Regarding possible memory effects, we assume these to be unimportant. 
We
>have been running succesive batches of the same sample (NBS-127) in 
clean,
>pre-degassed crucibles. The Pt-convertor plates are cleaned regularly
>every 2-3 runs, as well as the glass tube surrounding them. Reaction
>tubes are acid-cleaned after every run.
>
>Our feeling is that if we can increase yield above 95 %, we should get 

>closer to the 9.3/9.4 permil certified value for NBS-127. However, if 
this
>is not possible, and given the good reproducibility of results, we 
could
>always construct a calibration line, but to that end, we would require 
at
>least three -ideally more- well constrained, different enough points. 
Thus
>the question goes: In the event that this proved to be necessary, will
>somebody have a reliable internal standard that could be made 
available,
>with delta 18-O values different enough from NBS-127? A possible 
alternative
>might be reagent-grade BaSO4 from a reagents manufacturer. Would 
somebody 
>be willing to measure the oxygen isotopic composition of a given 
bottle
>of such material, and return it to us, such as to avoid inter-batch 
and
>inter-manufacturer variability?
>
>All tips, suggestions and help most appreciated!!!
>
>Cheers,
>
>
>Clemente
>
>Refs.:
>
>- Lang, J.R., Guan, Y. and Eastoe, Ch.J. (1989) "Stable Isotope 
Studies
>of Sulfates and Sulfides in the Mineral Park Porphyry Cu-Mo System,
>Arizona". Econ. Geol., 84, 650-662.
>
>- Rafter, T.A. (1967) "Oxygen isotopic composition of sulphates - Part 
I.
>A method for the extraction of oxygen and its quantitative conversion 
to
>carbon dioxide for isotope radiation measurements". New Zealand Jl. 
Sci.,
>10, 493-510.
>
>- Sakai, H. and Krouse, H.R. (1971) "Elimination of memory effects in
>18-O/16-O determinations in sulphates". Earth Planet. Sci. Letters, 
11,
>369-373.
>
>.... as said above, I have problems getting literature older than 1974 
...
>
>
>*/*/*/*/*/*/*/*/*/*/*/*/*/*/*/*
>Dr. Clemente Recio
>Laboratorio de Isotopos Estables
>Fac. de Ciencias
>Univ. de Salamanca
>Plaza de la Merced, S/N
>37008 - Salamanca
>Spain
>
>Phone (+34) 23 29 44 00, Ext. 1540
>Fax   (+34) 23 29 45 14
>E-mail: as in the heading
>*/*/*/*/*/*/*/*/*/*/*/*/*/*/*/*
>
>
>





Dr. Peter Kroopnick
Groundwater Technology Inc.
19033 W. Valley Hwy.
Suite D-104
Kent, WA 98032

Phone  206-251-5441
FAX    206-251-8452
email  [log in to unmask]