Dear Franco,

Since you are interested in the triple point of H as a standard state
("defintion fixed point") then the use of a sample characterized for D/H by
high precision IRMS seems less than satisfactory.  IRMS instruments measure
at high precision, but not at correspondingly high accuracy.  The accepted
isotope ratios for standards like VSMOW are set at the best particular
values available at the time they are first distributed, but these are
deceivingly precise.  The example with which I am most familiar is the
carbon standard, PDB, with a 13C/12C ratio  accepted to 6 significant
figures (0.0112372) but probably not accurate past the third significant
figure.  I believe VSMOW is better characterized but still it is probable
that there is more uncertainly in the absolute concentration of D/H than in
the precision.  The use of a sample with VSMOW-range D/H will introduce
error in the measurement that you are not expecting, particularly since D
properties are so different than H properties.  Also, there is no guarantee
that standards will not drift over the decades.

I would suggest a different strategy for establishing this standard state:
basing the fixed on the triple point of protium (or diprotium).  H2
generated by electrolysis of water isotopically fractionates as HD/H2
greatly compared to the liquid. I do not recall the exact fractionation
factor but it may be around 75%.  I would suggest that a line coupling in
series electrolysis followed by combustion followed by electrolysis
followed by ... etc. could purify protium water or H2 to arbitrary levels.
D-depleted water is available cheaply from suppliers and for all I know
this is the way they prepare it; one could start with depleted water and
further purify it.

This strategy would eliminate the inherent variability of D content and
also the arbitrary terrestrial D/H value, basing the entire measurement on
a material that can be completed defined.  One may also be able to
extrapolate the fixed state of absolutely pure protium from a series of
measurements on protium samples with small, variable amounts of HD, to
yield a truly absolute measure.

Regards,

Tom Brenna
Cornell University
Ithaca, NY