Hi Bill and others,

I cannot resist to react on this one... Your description clearly brought
forward one of the major caveats of single inlet-CF-mass spectrometry:
non-linearity! Now, there is nothing wrong in applying corrections. But
determination of a correction (e.g. linear regression) introduces an
uncertainty which has to be propagated to the final combined uncertainty of
a measurement, and therefore makes the result less precise. Actually I
cannot recall even one publication mentioning to include such an uncertainty
as mentioned here...!! Since we see very small precisions reported with the
single inlet-CF machines, you bring forward some clear arguments against
these (indirectly, yes).

Let me give an example - analysis of carbonates: If comparing phosphoric
acid digestion with double-inlet and single inlet-CF procedures, we see with
the last method generally smaller precisions are reported in literature
(also if comparing older work, based on double inlet, with newer single
inlet-CF type of work). If now considering the fact that single inlet MS is
a less precise way of measuring than double inlet MS and that for the single
inlet far smaller sample size is used compared with double inlet, where
introduction generally of more heterogeneity by the small sample size has to
be considered, how is it possible that single inlet-CF-MS results are given
with smaller precision than for double inlet results?

Moreover, with Bill's description, it also raises questions for solid
samples and the unavoidable inaccuracy of sample weight - in sense of amount
of element of interest for the isotopic signature and the variation in peak
sizes and shapes (same amount of gas but with clearly flatter and wider peak
will also give different results compared with a very narrow and higher
peak, following the description by Bill!) - and how far the delta values are
correct as produced in such machines (this counts for all single
inlet-CF-IRMS set-ups, not only TC/EA). If not aware of this 'amount'
problem, results may be quite wrong in some cases - how do we know to shift
the right from the wrong here? A nice precision just based on statistics
does not tell any of the story here for certain. And how can we correct for
an 'unknown' by regression in Bill's example - just be usuming that...???

One other comment I like to give here: the gas introduced at the open split
in front of the ion source often is named reference gas. This is a
completely incorrect name - this gas is and CANNOT be considered as
reference. It only serves to monitor the performance of our mass
spectrometer (to detect/register drift or eventual major MS problems, for
instance). Monitoring gas would be the correct term here. Using the wrong
term is seriously wrong, since it suggest something that is not... Reference
materials (preferably indentical in chemical, structural and isotopic
character to the sample material to be measured) have to pass the complete
system in exactly the same way as samples do. This monitoring gas does not
answer any of these requirements.

It is not that I am against these 'integrated' analytical systems - on the
contrary, I am happy they exist. But we should be more honest and realistic
in what they can do to my opinion. That is why I reacted...

Best wishes,
Pier.

****************************************************************
Delta Isotopes Consultancy

Dr. Pier A. de Groot
Pastoor Moorkensstraat 16
2400 Mol - Achterbos
Belgium
Tel. +32 (0)14 326 205
e-mail: [log in to unmask] or [log in to unmask]

Associate editor for stable isotopes of eEarth on-line magazine
http://www.electronic-earth.net

Visit my WEB-site about my ³Handbook of Stable Isotope Analytical
Techniques², with a link to the Elsevier web site on the handbook (marked:
ŒOrder Now¹):
http://users.pandora.be/handbook/index.html
last update: August 15, 2005
Volume I is now available. Volume II is expected to be available second half
of 2006.

****************************************************************
In a World Without Walls or Fences,
Who Needs Gates, or Windows?

****************************************************************



> From: Bill Showers <[log in to unmask]>
> Reply-To: Stable Isotope Geochemistry <[log in to unmask]>
> Date: Wed, 24 May 2006 15:21:11 -0400
> To: <[log in to unmask]>
> Subject: Re: [ISOGEOCHEM] calibration TC/EA for O2
> 
> Willi, Bruno, et al.,
> 
> The size non-linearity of CO is what caused us problems when we
> started using the TCEA several years ago.  We worked with Hairigh
> Avak at Bremen who helped us with this technical issue. Basically
> with liquids that are injected with an autosampler and a Hamilton
> micro-syringe you can inject about the same size sample every time
> into the TCEA and size linearity is not much of an issue between
> samples and standards. But with solids you can not weigh out exactly
> the same amount of oxygen in every sample, so your CO peaks from
> solid samples will encompass a range of sizes.
> 
> To get size linearity with CO, Hairigh had us basically detune the
> Delta to achieve maximum linearity.  This is best accomplished on our
> system by backing off on the extraction potential and then tuning
> everything else back up.  Maximum sensitivity for us is ~5.8
> extraction (out of 10), maximum linearity for CO is ~5.05 out of 10
> on the extraction setting.  We detune for for CO and O2, and have
> <0.1 per mil change between ~1-6 volts.
> 
> With a new filament and a cleaned source we can get good linearity
> for CO between 0.5 to 7 volts, but all solid samples need to be
> within this range to get good results. We then use tripicates of  3-4
> known isotopic standards that encompass the isotopic range of the
> samples we are running to calibrate each run for  the differences in
> slope for the isotopic response changes (with CF there is not a 1:1
> isotopic relationship between accepted isotopic values and measured
> values for standards due to instrument effects.  This linear
> correction then needs to be applied to the samples for each run).  I
> am not sure how you would correct for size and isotopic
> non-linearities in the same sample run.  I guess you would have to
> run low, medium and high isotopic standards at small, medium and
> large sizes and then have a multi-variate regression correction?
> 
> Detuning the extraction potential does decrease sensitivity, so there
> is a tradeoff.  Then there is the issue of what standards to use and
> what is the accepted isotopic value of these international
> standards.  Your corrected sample values can change depending upon
> what values you use to construct your linear regression (defined vs observed)
> 
> Since you can not run international standards every day in
> triplicate, you have to develop internal lab standards which are
> usually some sort of analytical grade reagent.  For CO we like sugars
> since they pyrolize so nicely, but they also have a limited isotopic
> range.  BaSO4s have low O-18 values and sucrose tends to be on the
> high side, and then there are plenty of compounds that are in the
> middle.  Soda nitre fertlizers from the Atacoma desert have very high
> O18 values because of their association with atmospheric deposition
> and you can order 5-20 lbs of that on line, enough for several
> careers.  Some folks have re-equilibrated reagent grade chemicals
> with waters of different O-18 compositions to get a wide range of
> isotopic compositions of the one particular compound they are
> running.  You still have to calibrate that with know standards.  But
> that is another discussion.  By detuning for size linearity and using
> a wide isotopic range of standards that bracket the sample isotopic
> values, we get consistent values run to run.  But we measure size
> linearity every day and throw out any sample values that fall out of
> that defined range.
> 
> Best,
> Bill
> 
> At 02:03 PM 5/24/2006, you wrote:
>>  Natalia, another thought...
>> 
>> 
>> Ohlsson and Wallmark published a paper that presents an algorithm to
>> mitigate the mass effect on isotope ratio analyses.  That is, the
>> ratio changes as a function of material combusted-- which is why we
>> are fairly careful about weighing standards precisely. In a previous
>> life, I modified their algorithm because we found that isotope ratio
>> dropped as a function of Total N or C combusted.  It dropped in a
>> manner that could be approximated by a first order decay curve.  We
>> never adopted this in the lab because it requires analysis of
>> standards with a range of masses bracketing the masses of the
>> samples within every run.  However, trials showed that this improved
>> precision in analysis of standards by as much as an order of
>> magnitude.  If it is important for you to run samples of varying
>> sizes you might give this method a try.
>> 
>> 
>> 
>> Reference    Ohlsson, K.E., and P.H. Wallmark. (1999) Novel
>> calibration with correction for drift and non-linear response for
>> continuous flow isotope ratio mass spectrometry applied to the
>> determination of (delta notation)15N, total nitrogen, (delta
>> notation)13C and total carbon in biological material.
>> The Analyst 124:571-577.
>> 
>> Good luck
>> 
>> John
>> 
>> John B. Cliff, Ph.D.
>> Research Scientist
>> Advanced Radioanalytical Chemistry Group
>> National Security Division
>> Pacific Northwest National Laboratory
>> Richland,  WA   U.S.A.
>> (509) 373-9003, fax (509) 376-5021
>> ([log in to unmask])
>> 
>> 
>> 
>> 
>> -----Original Message-----
>> From: Stable Isotope Geochemistry [mailto:[log in to unmask]]
>> On Behalf Of Dr. Bruno Glaser
>> Sent: Tuesday, May 23, 2006 11:00 AM
>> To: [log in to unmask]
>> Subject: Re: [ISOGEOCHEM] calibration TC/EA for O2
>> 
>> Dear Tiziano,
>> 
>> amount dependence might also be due to the split dilution,
>> especially from the Conlfo II such effects are known and can only be
>> avoided when using samples with the same amount of analyte element
>> or in other words using samples yielding e.g. around 2 V signal intensity.
>> 
>> Kind regards
>> 
>> Bruno
>> 
>> -----Original Message-----
>> From: Stable Isotope Geochemistry [mailto:[log in to unmask]]On
>> Behalf Of Tiziano Boschetti
>> Sent: Dienstag, 23. Mai 2006 17:31
>> To: [log in to unmask]
>> Subject: Re: [ISOGEOCHEM] calibration TC/EA for O2
>> 
>> 
>> Dear Willi and Natalia,
>> during the analysis of O-isotope I've noted an increase of the
>> background mass 30 (14N16O, 12C18O), usually the twice of 28.
>> Probably this could be due to the low purity of CO-reference gas
>> (4.7 grade), in fact I've also noted an increase of the 32 mass
>> signal (impurity of oxygen in CO?).
>> Low Purity of reference CO-gas could be resolved with a point-of-use
>> gas filtration or gas purifier (
>> http://www.alltechweb.com/productinfo/technical/datasheets/4001u.pdf
>> http://www.pall.com/microe_26234.asp?level1=0 ) What do you think?
>> Ciao!
>> *****
>> Tiziano Boschetti
>> University of Parma, Italy
>> 
>> *****
>> ----- Original Message -----
>> From: Willi A. Brand
>> To: [log in to unmask]
>> Sent: Tuesday, May 23, 2006 2:57 PM
>> Subject: Re: [ISOGEOCHEM] calibration TC/EA for O2
>> 
>> 
>> Dear Natalia,
>> I have two comments to your question.
>> 1. With a perfectly tuned ion source (perfect linearity for CO2 and N2) we
>> very consistently see a dependence of the 30/28 ratio upon signal size. It
>> is surprizingly constant at +0.3 per mill / Volt (1V = 3 nA 28). We
>> routinely correct for this, but we do not know the origin. One possibility
>> would be formation of an extra ion current on m/z 30 much like the H3+. If
>> small amounts of N2 are present in the reference gas, formation of NO can
>> occur on the filament, giving rise to NO+ (m/z 30) during reference
>> measurement. In this case, the correction would have to be applied to the
>> reference gas only. However, we see the same dependence upon signal size on
>> the sample peaks.
>> 2. We generally run our reduction at 1400$B!k(JC (tube-in-tube
>> arrangement with
>> the He flow reversal installed) and we adjust the sample size to about 1 mg
>> Oxygen to generate a larger ion current. We found this necessary in order to
>> be somewhat less dependent upon background and memory effects.
>> Regards   Willi
>> 
>> Sevastyanov wrote:
>> 
>> Hello!
>> I am a post-graduate student of V.I. Vernadsky Institute of Geochemistry and
>> Analytical Chemistry, Russian Academy of Science. I would like to ask some
>> questions about our DeltaptusXP mass spectrometer.
>> Two organic substances were analyzed for their $B&D(J18$B'0(J values using
>> our
>> pyrolysis EA-IRMS system: cellulose and benzoic acid (the reaction furnace
>> is heated to 13500C, silver capsules were used). A various samples amount
>> were examined, and our results revealed that there is a dependence of the
>> $B&D(J18$B'0(J values upon peak amplitude. These changes are not
>> great when peak
>> amplitudes exceed 2 V but increase rapidly when peak amplitudes become less
>> than 2 V. Thus, in the range from 2 to 1 V $B&D(J18$B'0(J values
>> become about 5 $B"s(J
>> smaller. We believe such phenomenon is not normal and we would like to know
>> how this situation may be improved and where this phenomenon comes from?
>> Where linearity area of our system ends? Is that true, that linearity bounds
>> by 2V? If this is so, what is the nature of such phenomenon?
>> 
>> 
>> Best regards,
>> 
>> 
>> Natalia Sokolova
>> post-graduate student
>> V.I. Vernadsky Institute of Geochemistry and Analytical Chemistry,
>> Russian Academy of Science
>> 19, Kosigin st., Moscow, 119991, Russia
>> 
>> 
>> 
>> 
>> 
>> --
>> .....................................................................
>> Willi A. Brand, Stable Isotope Laboratory      [log in to unmask]
>> Max-Planck-Institute for Biogeochemistry (Beutenberg Campus)
>> Hans-Knoell-Str. 10, 07745 Jena, Germany      Tel: +49-3641-576400
>> P.O.Box 100164,      07701 Jena, Germany      Fax: +49-3641-577400
>> http://www.bgc-jena.mpg.de/
>> http://www.bgc-jena.mpg.de/service/iso_gas_lab/
>> .....................................................................
>> GASIR 2005 in Jena:
>> http://www.bgc-jena.mpg.de/service/iso_gas_lab/gasir2005/index.shtml
>> .....................................................................
> 
> William J. Showers
> Dept of Marine, Earth & Atm Sciences
> Box 8208
> North Carolina State University
> Raleigh NC 27695
> 
> For express mail: 1125 Faucette Dr.
> 
> (919) 515 - 7143 Office
> (919) 515 - 7802 Fax
> (919) 515 - 3689 Lab
> (919) 515 - 7911 Field Lab
> [log in to unmask]
>