Dear subscribers to the ring test and interested audience,
With this message I want to disseminate the results of the D/H ring test that I
proposed to conduct about a year ago.
Background of the test:
The proposal arose from a discussion regarding the ability of 'modern' mass
spectrometers to accurately measure the difference between SLAP and VSMOW which
should be -428 permil. Some of the highlights of the discussion in Isogeochem
were:
Max Coleman, 9/22/95:
The data we produced many years ago on a 602 when we published the zinc method
gave -426.7 per mil, uncorrected by any stretch factor. Now we can get values
as high as -350 per mil !!!
James O'Neil, 9/22/95:
"We too have noted that the difference between raw dD values on a "modern
machine" (the Delta-S) are A LOT smaller than they should be. That is, for some
newer machines, you can not just designate a value of 0.0 permil for SMOW in
your program, accept the questionable H3+ factors determined automatically for
you, and get anywhere near the correct dD value when it is large...
For dD values of around -700 or so, comparisons of analyses from a few
laboratories where different kinds of machines are in use are disturbingly bad
(something like 50-100 hundred permil different!)."
Torsten Vennemann, 9/25/95:
"...gases (large batches of which have been prepared from mixtures of
commercially available H2 and HD) that we have sent to four other laboratories
for D/H measurement gave values that were generally within 2-3 permil of our
values IF THOSE GASES WERE ANALYZED ON FINNIGAN MAT'S, but could differ by 10 to
70 permil from our values when analyzed on VG machines."
The discussion revealed a great deal of uncertainty regarding the D/H
performance of mass spectrometers in various labs. As a consequence the
comparability of D/H results reported in the literature was questioned.
The IAEA has recommended that laboratories normalize measured isotopic
differences to the set difference of -428 permil for SLAP/VSMOW, a proposal that
must be followed when data are reported relative to VSMOW.
This, of course, applies when the samples are clean water implying that all
samples must be converted to clean water prior to measurement.
For measurements where water cannot be measured or compared directly (eg direct
pyrolysis of organic compounds) it would be nice to separate effects of the mass
spectrometer from those of the sample preparation procedure. Since there are no
certified H2 gases available to independently determine the mass spec scaling
(and because I did not agree with Jim O'Neil's statement) I proposed this test.
Three batches of equilibrated hydrogen gas flasks were produced by Messer
Griesheim. The isotopic compositions I asked for were:
1st gas around (but not equal to) SMOW [GAS A],
2nd gas around -400 per mil [GAS B],
3rd gas around -700 per mil [GAS C].
The homogeneity of the flasks for each batch was checked by Tyler Coplen. Ty
measured 7 flasks of each of the batches. Within each batch the flasks were
statistically identical to better than 0.5 per mil (for more information please
consult Ty's report to Isogeochem, 6/7/96).
Forty laboratories had enlisted to participate in the test (and pay DM 600 for
the flasks). The bottles were shipped in June this year. To this date 36 labs
have sent their results back to me. One lab did not receive the flasks (sorry,
Carol), another participant is on a cruise and could not make the measurements
in time. Thus, out of 38 labs the response is 94.7%.
>From the 36 labs I obtained results from 40 instruments listed in tables 1 and
2:
Table 1: D/H ring test results, values vs Gas A
MS / year Instr # delA 1/delA delB delB/A delC delC/A
MAT252 / 95 1 0.29 -0.29 -408.89 -409.06 -701.69 -701.78
MAT252 / 94 6 0.05 -0.05 -406.50 -406.53 -697.70 -697.72
MAT252 / 90 3 0.00 0.00 -408.00 -408.00 -703.00 -703.00
MAT252 / 90 8 -0.08 0.08 -411.94 -411.89 -709.24 -709.22
MAT252 / 94 32 -0.31 0.31 -407.92 -407.74 -700.00 -699.91
MAT252 / 89 35 -0.75 0.75 -408.23 -407.79 -699.87 -699.64
MAT252 / 93 37 0.03 -0.02 -408.32 -408.33 -700.49 -700.50
DeltaS / 92 4 -0.50 0.50 -407.60 -407.30 -699.00 -698.85
DeltaS / 95 5 -1.30 1.30 -414.80 -414.04 -705.30 -704.92
DeltaS / 93 18 -11.40 11.53 -412.10 -405.32 -699.45 -695.98
DeltaS / 88 23 -0.68 0.68 -404.30 -403.89 -694.70 -694.49
DeltaS / 91 25 -0.20 0.20 -405.30 -405.18 -693.50 -693.44
DeltaS / 93 29 -0.20 0.20 -407.50 -407.38 -699.10 -699.04
Optima / 96 10 0.50 -0.50 -408.50 -408.80 -701.60 -701.75
Optima / 92 11 0.10 -0.10 -405.63 -405.69 -696.04 -696.07
Optima / 92 13 -0.20 0.20 -407.70 -407.58 -698.30 -698.24
Optima / 96 15 0.13 -0.13 -412.90 -412.98 -709.70 -709.74
ES 2020 30 0.00 0.00 -406.50 -406.50 -698.30 -698.30
Sira II / 91 7 0.10 -0.10 -396.30 -396.36 -680.30 -680.33
SIRA / 88 9 0.40 -0.40 -380.70 -380.95 -653.30 -653.44
Sira 12 / 86 20 0.03 -0.03 -401.65 -401.67 -689.06 -689.07
Sira 9 / 85 22 0.00 0.00 -404.60 -404.60 -695.90 -695.90
Sira II / 89 39 0.14 -0.13 -368.03 -368.12 -632.10 -632.15
Delta E / 86 12 -0.43 0.43 -405.70 -405.44 -693.30 -693.17
Delta E / 87 24 0.01 -0.01 -412.80 -412.81 -708.30 -708.30
Delta E / 86 27 -0.14 0.14 -406.50 -406.42 -698.50 -698.46
Delta E / 88 33 0.40 -0.40 -396.40 -396.64 -679.20 -679.33
MAT 251 / 86 16 0.00 0.00 -399.77 -399.77 -687.62 -687.62
MAT 251 / 82 26 -0.02 0.02 -407.19 -407.18 -699.19 -699.18
MAT 251 / 81 28 -0.50 0.50 -416.00 -415.71 -714.20 -714.06
MAT 250 / 78 31 0.00 0.00 -406.90 -406.90 -698.40 -698.40
MAT 250 / 80 34 0.00 0.00 -401.99 -401.99 -689.66 -689.66
MAT 251 / 82 36 0.10 -0.10 -397.00 -397.06 -680.70 -680.73
VG 602 E/84 14 -0.29 0.29 -393.56 -393.38 -674.22 -674.13
VG 602 D 19 -0.09 0.09 -413.40 -413.35 -707.50 -707.47
VG 602 C/7x 21 -0.51 0.51 -408.26 -407.96 -700.27 -700.12
VG 602 D / 78 38 0.05 -0.05 -403.66 -403.69 -691.06 -691.08
VG 602 / 75 40 0.00 0.00 -401.70 -401.70 -689.30 -689.30
MAT230/ 77 2 0.00 0.00 -392.63 -392.63 -673.99 -673.99
Prism II /89 17 -0.32 0.32 -461.97 -461.80 -791.97 -791.90
Mean values [6-52]* -404.06 -693.29
std.dev. 8.90 15.26
Mean values [6-26]* -408.03 -700.01
std.dev. 2.90 4.79
*These numbers refer to the lines in my spreadsheet. [6-52] are all instruments
except 17, which
was treated as an outlier. [6-26] are the 'modern' instruments from the top to
Instr.# 30
---------------------------------------------------------------------------------
-------------------------------
The columns del A, del B and del C are the values that were reported. In order
to make the numbers comparable I recalculated those values with the mean of gas
A set to zero using the usual del[x/st] = del[x/ws] + del[ws/st] + del[x/ws]
del[ws/st] / 1000. In our example, x is gas B or C, st is the average of gas A
set to zero, ws is the working gas in the reference bellows, in most cases gas
A. The column labelled [1/delA] reflects the del[ws/st] term.
The order of the listing is arbitrary (with a non-intentional bias for the
products of my employer), Instruments of the same kind have been grouped
together. The first 17 instruments represent the suite that is currently
available on the market ('modern' instruments). At the bottom I have given
average numbers for delB/A and delC/A for those and for all instruments
together (#17 taken out).
There is a notable difference in the average values given due to the fact that
some older instruments tend to underestimate the isotopic differences
systematically by a large amount. This is also reflected by the larger error
margins. The average values of the 'modern' instuments are probably more
representative. Here, the error margins are 3 and 5 per mil respectively which
is still high in comparison with the reproducibility that a single instrument is
capable of delivering. Instrument 17 was treated as an outlier for obvious
reasons. The results are in line with previous reports for this instrument (see
for instance Jean Morrisons communication to Isogeochem dated 10/12/95 and John
Morrisons note dated 10/18/95).
There is an instrument bias in each and every case. It may be small for some
instruments but it still needs to be corrected for. In order to do so there is a
need for certified reference H2 gases that must be available to every lab for
the forseeable future. It is not imperative that these reference gases be
identical over a long time, they 'only' must be available and their (large)
differences be reliable. It would be an important service to the community if
they could be provided by IAEA, NIST, or USGS.
Table 2, D/H results versus VSM
MS / year Instr # delA/ delB/ delC/ H3+ SL/VSM Method
VSM VSM VSM Factor
MAT252 / 95 1 -9.83 -415.04 -704.48 8.00 -428.70 Cr
MAT252 / 94 6 12.36 -410.00 Zn
MAT252 / 90 3 -9.10 -420.00 -716.00 11.41 -421.00 Zn
MAT252 / 90 8
MAT252 / 94 32 9.48 -422.00
MAT252 / 89 35 13.92
MAT252 / 93 37 -3.04 -410.30 -701.64 9.17 -414.90 Zn
Delta S / 92 4 -12.10 -414.60 -702.60 14.70 -403.80 Zn
Delta S / 95 5 -9.50 -417.10 -708.80 5.58 -426.30 Cr
Delta S / 93 18 6.63
Delta S / 88 23 -2.10 -415.50 -712.00 9.40 -417.00
Delta S / 91 25 20.26 -417.90
Delta S / 93 29 -417.70
Optima / 96 10 -5.90 -427.30 -729.30 3.40
Optima / 92 11 -2.60 -420.80 -720.40 5.95 -418.00
Optima / 92 13 9.50 -427.00
Optima / 96 15 7.30
ES 2020 30 -17.90 -420.60 -710.10 3.05 -424.00 U
Sira II / 91 7 -10.50 -433.20 -742.20 8.45 -427.10
SIRA / 88 9 -5.50 -425.10 -725.40 6.60 -386.00 Zn
Sira 12 / 86 20
Sira 9 / 85 22 -10.40 -421.80 -717.90 3.38 -416.70 U
Sira II / 89 39 9.26 -384.00
Delta E / 86 12 38.20 -414.00
Delta E / 87 24 -6.84 22.84
Delta E / 86 27 27.25
Delta E / 88 33 155.8
MAT251 / 86 16 -7.19 -404.09 -689.87 -396.92 U
MAT251 / 82 26 13.78
MAT251 / 81 28 9.70 -419.40
MAT250 / 78 31 -5.60 -421.10 -718.40 14.95 -417.00
MAT250 / 80 34 -8.77 -410.76 -698.42 7.87 -424.6
MAT251 / 82 36 -423.00
VG 602 E/ 84 14 5.59
VG 602 D 19 5.64 -427.00
VG 602 C / 7x 21 14.49 [-519.5]
VG 602 D / 78 38 3.15 -424.00
VG 602 / 75 40 154.20 -412.90
MAT230C / 77 2 -7.86 -413.46 -704.23 9.10 Zn
VG Prism II /89 17 -0.70 -409.30 -701.40 38.20
Mean values [6-52]* -7.93 -418.17 -712.61 19.42 -416.19
std.dev. 3.88 7.13 13.03 35.16 11.76
Mean values [6-26]* -8.01 -417.92 -711.70 9.38 -419.10
std.dev. 5.18 4.90 9.02 4.44 6.99
*These numbers refer to the lines in my spreadsheet. [6-52] are all instruments
except 17, which
was treated as an outlier. [6-26] are the 'modern' instruments from the top to
Instr.# 30
___________________________________________________________________________
The results in table 2 are not as complete as the ones in table 1. There are two
reasons for this. First, I labeled these results optional, second, it is not
necessarily a straightforward procedure to relate a hydrogen gas to VSMOW. In
between there is the sample preparation of the VSMOW and SLAP waters which may
produce hydrogen that exhibits the correct difference but still may be off the
absolute value by some per mil. This certainly depends on the reduction method
or other means of sample preparation. Given these constraints the results are
remarkably close for all gases, however with a large error margin.
The values are also entirely consistent with the averages for the gases when a
stretching factor of 1.016 is applied.
For your interest I also have listed the H3+ factors as reported and the
SLAP/VSMOW results prior to stretching. The H3+ factor varies over a wide range
but it seems that there is no correlation between instrument performance for
determining the differences and the magnitude of this correction. Probably,
stability of the H3+ production is more important than size.
The raw SLAP/VSMOW values confirm that stretching is needed. In some cases the
offset for this difference correlates with the measurement of gases B and C
versus A, in others not. This finding supports the idea for employing separate
scaling for D/H measurements: One for the instrumental effects and the other for
the sample preparation effects.
Acknowledgements
I would like to thank all participants for their efforts. In particular Ty
Coplen deserves a hand for meticulously confirming the homogeneity of the gases
with a total of 97 (!) measurements. Last, but not least, I want to thank Renate
Schmidt and Messer Griesheim for making this test possible by producing the
gases and shipping them throughout the world.
Bremen, October 8th, 1996 Willi Brand
|
