>Hello everyone,
>
>Our lab has been asked to analyze the hydrogen/deuterium ratio
>of a water sample from a crater lake. We have tried 3 times 
>(zinc reduction method) but the results differ from each other by at
>least 10 per mil. The problem is that the water is very acidic
>(pH 1 or 2) and we can't raise the pH without adding an external
>source of hydrogen (hydroxide). Does anyone have an idea as to
>how we could obtain the hydrogen/deuterium ratio of this sample?
>

Dear Natalie and others interested

We have had similar problems with D/H analysis of water extracted from 
certain micas where the water is reduced using zinc reagent. The extracted 
water was found to react ("fizz") with the zinc at room temperature within 
the sealed silica tube. For such micas we commonly had incomplete reactions 
between zinc and water resulting in poor replication of samples. It is 
possible that this problem may be related to micas with high F or Cl content 
which, during extraction of the water, may form HCl or HF. These may in turn 
contaminate the zinc in some way. The problem was readily solved though by 
doubling the amount of zinc used for the samples (we used about 100mg of 
zinc per expected mg of water).

With regard to distilling the sample, care must be taken not to fractionate 
the sample during distillation since the residual salts are very hygroscopic 
and it may be difficult to remove all water from the hydrated salts. I 
suggest you use disposable glass pipettes (essentially small glass capillary 
tubes) which are filled by dipping them into the water and are then dumped 
into a small 6mm o.d. silica tube filled with Ar. Once attached to the 
vacuum line, the water can be frozen down, the Ar pumped off, the tube 
sealed and heated with excess zinc at 480 deg C for 10 to 20 minutes.

Greetings,
                        Torsten Vennemann.

  
*********************************
Dr. Torsten W. Vennemann
Institut fuer Geochemie
Wilhelmstr. 56
D-72074 Tuebingen
Germany
phone ++49-(0)7071-294992
Fax ++49-(0)7071-296870
e-mail: [log in to unmask]
********************************