Dear Dr. Verma and Colleagues,

I appreciate Dr. Verma's recent response very much. Now we have a nice
debate and we can make Andrew happy!  He was right that such discussions
should be performed here, on this server. If we had this discussion a
year ago, there would have been no Comment in GCA (did I use the right
tense here?). Nancy - you wrote that chemists were so passionate. I am
not a chemist, but a physicist. We are even worse. Our brains shrink
when we think, leaving more space in the skull for passion and less for
reason.

OK, let us go back to the subject. Let me first summarize the discussion in
points, from the beginning. I shall list all Dr. Verma's critical
remarks (from his Comment and e-mails) and discuss them one by one. Than
comes one suggestion and one personal statement.

1. Discussion

a) Halas pH=10.3 and Verma's pH=10.85 in his Comment (GCA 64, 573-574,
2000) and 10.7 in his recent posting. Bad chemistry argument for the
first time.

I am glad to see you finally admitt that there ARE differences between
models, and calculation of pH does not always give the same value. Your
simplified calculations in the Comment gave pH=10.85, whereas in your
recent posting, calculated by activity approach for the same system, we
find different pH=10.71. Finally, if you perform pH calculations for
this system with the WEB-PHREEQ web-based solver
(http://www.ndsu.nodak.edu/instruct/sainieid/geochem/webphreeq/index.shtml)
you end up with pH=10.28 for Halas's values of X1, X2 and pCO2, and
10.38 for values given in your recent posting. Halas in his Reply gives
the revised value of 10.37!!!

source                                pH

Halas paper                        10.30
Halas measurement             10.30
Verma comment                 10.85
Halas reply                         10.37
Verma file                           10.71
PHREEQ (Halas's data)      10.28
PHREEQ (Verma's data)    10.38

The most important point here, missing in your Comment, is the possible
influence of changes in fractions of dissolved species on the calculated
DIC-CO2 fractionation factor. This factor is calculated by use of eqn
(15) from Halas's paper and the calculated fractions of HCO3- and CO32-.
These fractions in your most recent Excel file are 0.1833 and 0.8167,
respectively. Halas's values from his Reply are 0.1828 and 0.8172,
respectively. Room temperature fractionation factor calculated by your
values is 4.998 and by Halas's values is 5.001. In his paper he gives
5.0 +/- 0.2 permil. Where is the difference? Where is the bad chemistry?
Verma's and Halas's results are in PERFECT agreement here! And it is
well documented by your own postings now, that problems with pH
calculations are universal.
As for the error propagation: eroneous 0.35 difference in pH in the
discussed region would cause change of calculated fractionation factor
of about 0.15 permil. This is therefore safe to report the fractionation
factor with +/- 0.2 permil uncertainty, as is done by Halas.

Conclusion: There is no bad chemistry in Halas's paper. There are
universal problems with pH calculation, but Halas's pH calculated by 2
different programs gives values close to obtained by his direct
measurements. Verma's results and Halas's results for fractions X1 and
X2 produce the same value of isotopic fractionation. We can trust this
value.

b) No HCO3 and OH terms in the ionic strength formula (8) in Halas's
Reply (GCA, 64, 575-576, 2000). Bad chemistry argument for the second
and third time (2 postings)

You need to substitute left side of eqn(1) from the Reply to full eqn(7)
to find out that both HCO and OH terms ARE used in formula (8), although
indirectly. This is simple addition and multiplication, and Dr. Verma is
wrong here.

Conclusion: Eqn. (8) correct, no bad chemistry.

c) Water activity 0.98 impossible for this system
You kept repeating it untill you calculated it by yourself to be
0.98011.

Conclusion: no bad......

d) Ph measurement impossible with high precission.
To make it simple: there is a single measurement error, which is
directly related to instrumental precision. It is additionally enlarged
by random influences, like minute temperature changes. Let us assume
that the maximum single pH measurement error (uncertainty) is very high, as high as
0.1. We know that repetitive measurements of pH are independent.
According to the error propagation formula, we have the following
uncertainties for the mean value of pH:

number
of measurements                   uncertainty of mean
  1                                          0.100
  2                                          0.071
  5                                          0.045
10                                          0.031
50                                          0.014

It is threfore enough to perform 50 measurements of pH with a poor
precision instrument to obtain satisfactory precision of the mean value.
There is nothing shocking in reporting the pH value with 0.01
uncertainty.

Conclusion: You are, of course, right that there are numerous problems
associated with pH measurements. As it is with measurement of every
other physical property. Shall we stop measuring anything because of
associated errors? Halas gives a detailed error propagation discussion
and the wide +/- 0.2 permil uncertainty for his fractionation factor.
Your argument here is irrelevant.

e) New argument: no water activity in eqn. (5) in the Reply
Now, a new one. There was nothing about this in the Comment, and now it
is the only "basis mistake". Water activity in eqn(5) was obviously set
to 1 to simplify calculations. You must agree that dividing something of
the order of 10^(-15) by 1 or by 0.98 makes no important difference?

Conclusion: Bad idea, to rise a new ad-hoc argument when the others
failed.

f) New argument: CO2 fractionation during cryogenic trapping in presence
of water vapour.
This is all described in the paper. CO2 and vapour are separated in the
upper glass ampoule without cooling, in 25C (see Fig. 1 in GCA 61, 2691
- 2695, 1997) by turning the stopcock. Than both CO2 and H2O are trapped
cryogenically in a  small ampoule. Than CO2 is released for measurement
and water remains as a solid in the ampoule. This is all done fast. How
could there be any significant dissolution of CO2 in water in this
experiment? There is supposed to be NO LIQUID WATER there, only water
vapour and frozen solid. Presence of trace amounts of liquid water that
could form on the walls due to temperature instabilities in high
temperature range, could possibly explain variation of results obtained
for high temperatures (together with the stopcocks problems reported in
the paper). But we have to understand one thing - this is not any major
drawback, but quite contrary, this was done with great care and reported
honestly.

Conclusion: same as in point e)

g) New argument: there should be no "fundamental experiment (like in
discussion) where there is need of
correcting isotopic measurements with chemical analysis"

Wrong! All the isotope exchange experiments are like this. What was
measured in this experiment? Weight of reagent, volume of water, pH,
temperature, isotopic compositions of reagent, DIC and CO2. Shall we
stop doing such experiments because they are difficult? This is the only
way to do this. As Dr. Jedrysek (my friend isotope geologist from
Wroclaw) used to say to his students: in this lab you shall find pain,
work, sweat, tears and kicked ass. This is science, I presume.

2. Suggestion
Having boxed Halas in public for a while, you say in your recent
posting: "my comments are just some suggestions,
which may be considered during their work". This is true, but the
meaning of your Comment and postings was quite different. I suggest you,
Dr. Verma, to send a letter to The Editor, GCA, stating that you were
wrong in raising the "bad chemistry problem" in the Halas's paper. Or at
least admit this here.

3. Personal
Nothing personal! I learned a lot last days and I am grateful for this
possibility. It looks like I will stay in New Mexico for a while. Being
so close to each other, we will probably have a possibility to meet and
have a couple of beers. Or better some Polish Bison Vodka, really cold,
not shaken. No hangover guaranteed.

With best regards,

Tomasz Durakiewicz


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

Mahendra Pal Verma wrote:

> Dear Colleagues,
>
> I don t know whether I should continue on this matter or not. It looks that there are
> some who are interest in knowing it. In my opinion, too, the Isogeochem Server is a
> good opportunity to discuss our viewpoints and improve our understanding with the
> comments of others. Let me write first that I am a regular reader of GCA and it
> publishes excellent material. Of course, it is well established. My comments were only
> that the reviewers should put more attention when a mistake had been pointed out.
> People at least I use the information published there as granted.
>
> I worked out the problem using the activity approach. The data are given in the
> attached file. I use  Error and Trail method  to solve a non-linear equation. I do it
> in excel without writing any programming code. I don t think that anyone is
> interested in that. The basis mistake in the reply by Halas et al. (GCA, 64, 575-576,
> 2000) was that their equation 5 should be K3=a(H)a(OH)/a(H2O). The values are given
> for their system (10 g Na2CO3+CO2 from 1g of Na2CO3). Again, there could be small
> differences if you use different constants. I tried to use the same constants as were
> given by Halas et al. I did the calculation without re-check. So,
> there could be some mistakes. I think it could be better if Halas et al. change
> equation 5 and re-calculate the parameters.
>
> Recently, we have started a project to study rainwater acidity around geothermal
> systems. We observed that it was really difficult to measure pH of pure water like
> rainwater with a pH-meter (rainwater characterization is based on pH around the world).
> Similarly there are problems in measuring pH of highly
> acidic or alkaline (more in case of alkaline) waters. (we are working to understand this).
> Additionally, the IAEA has
> conducted three interlab calibrations (Giggenbach, IAEA-TECDOC-641, Vienna,
> p.439-456, 1989 and two are very recent). If you look these reports, you will find
> that there is some improvement in the analysis after 10 years; but the quality of analysis
> of
> especially SiO2, HCO3 and CO3 is far bellow the expectations. That is different topic
> how to improve it. But the results indicate that in my viewpoint one should not
> design a fundamental experiment (like in discussion) where there is need of
> correcting isotopic measurements with chemical analysis. As I tried to explain it in
> my comments. The calculated  ratio of HCO3:CO3 is 18:82. So there is need of correcting the
> HCO3 contribution. One can calculate the propogation of error in their isotopic data due to
> the error in the chemical analysis.
>
> Secondly, the partial pressure of CO2 in vapor phase is very low (much less than
> 10^-4 bars). If you are extracting 10 micromoles, you can calculate how much volume
> should be of the experimental vessel and how the vapor phase CO2 can be extracted
> only without changing its isotopic composition. The amount of water vapor will be
> much higher than that of CO2. Check the steam tables. When you condense this vapor
> some part of CO2 will get dissolved in the water. So there will be some fractionation
> in this process, too. So, there are many points one has to consider for this work. I
> do agree that some thing is better than nothing. This way the calculated values (with
> spectroscopic data) could be more accurate than the experimentally measured.  The
> measurement of isotopic
> fractionation for CO2(g) and carbonate is a challenging subject. Definitely, it is
> hard job to work on this type of problems. So, we should do encourage the people who are
> working such types of problems. My comments are just some suggestions,
> which may be considered during their work.
>
> Best regards
>
> Mahendra
>
>   ------------------------------------------------------------------------
>                  Name: calFnl.xls
>    calFnl.xls    Type: Download File (application/x-msexcel)
>              Encoding: base64