Dear Paul and Others,
In reference to this topic, I would like to present our laboratory's
experience regarding (using Paul's terminology) scaling, linearity,
and drift, using a Finnigan Delta-Plus Advantage IRMS/Costech ECS
4010 EA system.
We have been using this EA/IRMS system for about five years. Our
usual routine lab standard for basic instrument calibration is
acetanilide (for about the last two years, from a bottle of
analytical reagent grade ("Baker Analyzed") material, and prior to
that, from a bottle of elemental analysis grade (Costech) material),
and our usual lab control standard is NBS 1572, Citrus Leaves. These
lab standards have been extensively analyzed against a number of
international certified reference materials, including NBS19, Sucrose
ANU, and PEF 1 (C only), USGS 25, 26, and 32 and IAEA N1 and N3 (N
only), and USGS 40 and 41 (dual element), at same-amount levels, as
well as at varying amounts. Here is a summary of our results.
Carbon
1. "Linearity"
Our standard calibration procedure entails the measurement of a range
of weights of acetanilide samples. There is always a significant
weight-to-delta bias, on the order of about one permil per 70ug of
carbon (200 - 700ug weight range); the exact slope and intercept of
the curve, of course, varies slightly from day to day. The weight
versus delta curves for four different materials, acetanilide,
sucrose, limestone (NBS19) and polyethylene film (PEF 1) vary only
slightly in their slopes (0.0067 to 0.0070 in one particular trial).
Based on this similarity with widely different sample types, we feel
that the weight-to-delta bias determined with the acetanilide
standards can be appropriately applied to the unknown samples
measured with those standards, as long as the amounts of carbon in
the unknowns fall within the weight range of the calibration standards.
2. "Scaling"
We determine the delta 13-C normalization curve for our instrument by
evaluating the measured values (after adjustment for weight-to-delta
bias, if needed) versus expected delta values for five certified
international standards, which cover a delta range of -32 to +38
permil. The slope of this curve for our instrument is very close to
unity (0.985), and the r-squared value for the curve is typically
better than 0.9999. The "known" values we use for our lab standards
(-29.50 for acetanilide and -27.20 for NBS 1572) are based on an
average of multiple analyses of these materials run in the same
batches as the certified standards. Our normalization procedure for
a typical sample batch run consists of simply adjusting the intercept
of the acetanilide weight vs delta calibration curve, to bring the
mean measured acetanilide and citrus (NBS 1572) samples to values
that are equally (but oppositely) offset (typically by no more than
+/- 0.1 permil) from their expected values (i.e., "splitting the
difference" between the two standards). For the vast majority of the
samples our lab analyzes, namely, biologically based samples, with
significant inherent variability even in replicate samples, we feel
that the roughly 0.15 permil potential inaccuracy per 10 permil
difference from our standards, is negligible, and additional
"scaling" is not required. For samples (and/or studies) that justify
it, we include additional control samples in the batch run, and add
the linearity slope to the adjustment calculation.
3. "Drift"
We always analyze additional acetanilide samples periodically
throughout a run (typically every 8 to 12 samples) as check
standards, as a means of monitoring of general instrument
performance. Our experience is that the drift in our instrument over
the course of a typical six hour run is rarely significant, that is,
the weight-adjusted values for the check standards virtually never
show any directional tendency outside the expected standard deviation
for measurements of this material (typically 0.20 permil), and drift
correction is not required.
Nitrogen
1. "Linearity"
Our standard calibration procedure for nitrogen is also based on a
range of weights of acetanilide. The slope of the weight versus
delta curve for nitrogen averages around 2 permil per 100ug of
nitrogen, although it can run anywhere from 1 to 3 permil per 100ug
on any given day. In addition to lower temporal consistency than for
carbon, the slope of the curve with different materials (and even
different sources of the same material) is also not as uniform. In a
comparison of 5 different sample types, acetanilide, potassium
nitrate (N3), and citrus leaves all had similar slopes (about 0.02),
while ammonium sulfate (N1) and glutamic acid (USGS 40) had slopes of
about 0.01. In another comparison, two different ammonium sulfate
samples (N1 and USGS 26) showed slopes of 0.019 and 0.031,
respectively. The lower precision for nitrogen delta measurements
at typically analyzed nitrogen levels (generally 0.3 permil at 20 -
100ug) may explain some of the variability, but the bias is
apparently also dependent on particular instrumental conditions
(probably mainly sample combustion), which can affect different
sample materials differently. For greatest accuracy in nitrogen
isotope ratio determinations, it is recommended that unknown samples
be analyzed within a narrow weight range, preferably corresponding to
an amount of nitrogen near the middle of the calibration weight range.
2. "Scaling"
The delta 15-N normalization curve for our instrument has been
determined by evaluating the measured values (after adjustment for
weight-to-delta bias) versus expected delta values for seven
certified international standards, which cover a delta range of -30
to +180 permil. The slope of this curve for our instrument is nearly
identical to that for carbon (0.987), and the r-squared value for the
curve is also typically better than 0.9999. The "known" values we
use for our lab standards (-0.40 for acetanilide and +4.90 for NBS
1572) are, as with the carbon values, based on an average of multiple
analyses of these materials in the same batches as the certified
standards. Our normalization procedure for delta 15-N is the same as
that for carbon, adjusting the calibration curve offset to split the
difference between the expected values for acetanilide and citrus.
And the same reasoning as with carbon applies for not including a
scaling adjustment for most samples. But again, for samples (and/or
studies) that justify it, we include additional control samples in
the batch run, and add the linearity slope to the adjustment
calculation.
3. "Drift"
Periodic check standards are also analyzed in nitrogen isotope runs,
and as with carbon, a clear indication of instrument drift is rarely
observed.
In regard to obtaining accurate results for samples with disparate C/
N ratios in a single batch run, by calibrating our instrument with
acetanilide standards (C/N = 7) having a weight range from 100 to
900ug, and adjusting for weight-to-delta bias (linearity), we are
able to cover a C/N ratio range from less than 1 to over 50.
I hope some of you will find this helpful.
Best regards,
Robert Petty, Ph. D.
(Former) Manager, MSI Analytical Lab
Marine Science Institute
University of California
Santa Barbara, CA 93106
On Thursday, October 30, 2008, at 11:01AM, "stableisotopes"
<[log in to unmask]> wrote:
> To ever analyst analyzing 15N and 13C with an elemental analyzer (EA)
> and isotope ratio mass spectrometer (IRMS):
>
> I have been following the thread (bottom of this email) on
> standards for
> 13C 15N calibration.
>
> For the purposes of my discussion:
>
> 1) I refer to the use of two standards with an equal mass of N or
> C but
> different different isotope ratios, for example USGS 40 and USGS
> 41, as
> "scaling" standards, for calibrating the instruments scale.
>
> 2) I refer to the same standard at different masses as "linearity"
> calibration or checking. This is the change of isotope ratio with
> size.
>
> 3) I refer to the same standard of the same mass analyzed periodically
> throughout the analysis as "drift" correction. This is to correct for
> drift with time.
>
> I agree completely that we need a set of two standards that have high
> and low 15N and 13C delta values so that we can scale the instruments.
> The problem is that so far the urea, Acetanilide and glutamic acids
> have
> C/N ratios of 6/1 or less, which is much lower than 90% of the samples
> that we analyze. Most of our samples are plant or soil material,
> with a
> C/N ratio of 15 to 30/1. Our experience is that the change of
> linearity
> with size of the combined EA/IRMS system can be significant and
> must be
> corrected for on every analysis. This should not be confused with
> linearity checks of the reference gas alone in the IRMS. I have found
> several researchers who assume that since their IRMS reference gas was
> linear compared to changes in size that different sized samples
> from the
> EA should be as well. When they checked the combined EA/IRMS response
> to different sizes of the same standard they discovered that the
> response was not linear, and size dependant. This effect seems to
> change as the EA and IRMS age, and may change from day to day.
>
> In addition, we have found that the EA/IRMS combination can drift over
> time, even with instruments with reference gas injection, so drift
> correction standards at regular intervals are required.
>
> Therefore we find it necessary to do analyzes with drift correction
> standards every 12 samples, 10 variable weight standards in the
> analysis
> to check for linearity, and additional standards to check scaling. We
> usually drift correct and check linearity with standard sets of NIST
> 1547 peach leaves, which we have found to be uniformly ground out
> of the
> bottle down to 0.2 mg weight, have a C/N ratio of 16, and are
> available
> in over 100g quantities. We then check the scaling with another
> suitable
> quality control standard at a different isotope ratio but similar
> material and C/N ratio to the samples we are analyzing. Even this
> amount
> of standards means that we have 30 standards in a typical overnight
> analysis of 90 unknowns, resulting in 120 total tins for analysis.
>
> If we were to try to use glutamic acid we would have to weigh out two
> sets of standards to get the N and C into the same range as the
> samples. This means that we would now have to weigh out 40 drift and
> linearity standards for the analysis, plus additional standards for
> scaling. This results in almost as many standards as samples in the
> samples tray and is completely unrealistic for most work.
>
> I have had several researchers explain that they get around this
> problem
> by; a) analyzing the samples first on a normal EA for %N and %C; b)
> weighing out two aliquots of the sample at specific weight for
> either C
> or N analysis so that for an analysis each sample set has very similar
> mass of N or C; c) analyzing the samples in two separate analyzes, one
> for N and one for C. The problem with this approach, apart from the
> fact that it does not lend itself to analysis of samples sent in by
> researchers not present at our facility, is that this requires three
> analyzes, %, dN and dC, for one sample.
>
> All our experience in ecosystem science work has shown that inevitably
> the noise level of the experiment is in the field sampling, and
> that to
> get the noise level of an experiment down one should take and
> analyze as
> many field samples as possible. Therefore it is much more
> productive to
> analyze three field samples each for a combined %, dN and dC than to
> analyze one sample three separate times for %, dN and dC.
>
> This raises the question as to how other labs deal with the problems
> described above, and if they have found suitable standard material
> with
> isotope ratios of N and C at different ends of the scale but with
> suitable C/N ratios.
>
> I am looking forward to input on this.
>
> Paul Brooks
>
>
>
>
> Arndt Schimmelmann wrote:
>> Dear Hilmar and those with interest in acetanilides and ureas with
>> 13C
>> and/or 15N enrichment,
>>
>> I thank Ty Coplen for pointing out that USGS 40 and USGS 41 L-
>> glutamic
>> acids already provide suitable reference materials with two different
>> 13C/12C and 15N/14N ratios for two-point-calibrations. These
>> materials
>> should serve the purpose well for most labs. However, I would be
>> happy
>> to send test aliquots of Indiana University's acetanilides and ureas
>> with 3 levels of 13C and/or 15N abundances to qualified laboratories
>> who wish to participate in an informal multi-laboratory comparison.
>> Please contact me directly and do not reply to the entire list.
>> Thank you for your interest,
>> Arndt Schimmelmann
>>
>>
>> Prof. Dr. Hilmar Förstel wrote:
>>> Dear Arndt Schimmelmann,
>>> Thanks for your comment in a public forum. Does it make sense to
>>> make
>>> a laboratory comparison? A official round robin test may need too
>>> much effort.
>>>
>>> Yours Hilmar Foerstel
>>>
>>>
>>>> Hello Lora and others,
>>>> The use of a single batch of isotopically known acetanilide for
>>>> d13C
>>>> and/or d15N calibration is problematic because single-point
>>>> calibration does not account for differences in attenuation of
>>>> delta-scales among different mass-spectrometers (i.e. slope). We
>>>> should calibrate to the VPDB scale in a way that it correctly
>>>> reflects NBS-19 (+1.95) and L-SVEC (-46.6) values. This was
>>>> explained by Coplen et al. (2006) New guidelines for d13C
>>>> measurements. Analytical Chemistry 78 (7), 2439-2441. The rationale
>>>> is the same as for VSMOW and SLAP in hydrogen isotope systematics.
>>>> It follows that for internal calibration one should use pairs of
>>>> organic reference materials that express significant isotopic
>>>> differences. Our lab is currently developing acetanilides with 3
>>>> levels of 15N abundance (delta15N ca. +1.2, +19.6, and +40.6
>>>> permil;
>>>> for standards we use the IAEA- N-1 and IAEA-N-2 ammonium sulfates).
>>>> We could not obtain 13C-enriched acetanilide. Instead, we are
>>>> developing 3 batches of urea with different levels of 13C and 15N
>>>> abundances. A similar project is generating 4 nicotines with
>>>> different levels of 13C and 15N abundances for GC-IRMS. These
>>>> materials should be ready by the end of this year. Some compounds
>>>> are sufficiently characterized already now, but not yet listed
>>>> on my
>>>> website (http://mypage.iu.edu/~aschimme/hc.html). Please contact me
>>>> if you have interest.
>>>> Best regards,
>>>> Arndt Schimmelmann
>>>>
>>>> Lora L. Wingate wrote:
>>>>
>>>>> I am using Acetanilide for a C/N weight percentage check standard
>>>>> on our new Costech EA coupled to a Delta V Plus. I am interested
>>>>> as to whether there is a consensus as to its value for d15N and
>>>>> d13C. I have run it alongside and calibrated it relative to IAEA
>>>>> and USGS standards, but I am finding that my Acetanilide values
>>>>> are
>>>>> different from one lab's data report that I had the opportunity to
>>>>> look over. I'd be interested in knowing what values other labs
>>>>> are
>>>>> getting. Or, if it is a moot point and Acetanilide isn't deemed a
>>>>> reliable isotopic standard, despite the reproducible results we
>>>>> obtain. Thanks for any insight, and off list replies are welcome.
>>>>>
>>>>>
>>>>>
>>>>> Lora L. Wingate
>>>>> University of Michigan
>>>>> Department of Geological Sciences
>>>>> Stable Isotope Laboratory
>>>>> 1100 North University
>>>>> 1013 C.C. Little Building
>>>>> Ann Arbor, MI 48109-1005
>>>> --
>>>> Arndt Schimmelmann, Ph.D.
>>>> Senior Scientist
>>>> Indiana University
>>>> Department of Geological Sciences
>>>> Biogeochemical Laboratories
>>>> 1001 East 10th Street
>>>> Bloomington, IN 47405-1405
>>>> Ph (812) 855-7645
>>>> home (812) 339-3708
>>>> FAX (812) 855-7961
>>>> e-mail: [log in to unmask]
>>>> personal home page:
>>>> http://www.indiana.edu/~geosci/people/faculty2.php?n=schimmelmann
>>>> home page of Biogeochemical Laboratories:
>>>> http://www.indiana.edu/~geosci/research/biogeochem/
>>>>
>>
>>
>
>
> --
> Mr. Paul D. Brooks,
> Dept. Integrative Biology MC3140,
> 3060 Valley Life Sciences Building,
> UC Berkeley, 94720-3140.
>
> [log in to unmask]
>
> phone (510)643-1748
> FAX (510)643-1749
>
> http://ib.berkeley.edu/groups/biogeochemistry/
>
|
