The following discussion took place during January and February 1999.


Dear fluid inclusionists:

Apologies if this is a dumb question, but is anyone aware of anything that will RAISE the melting point of CO2? I have some little beggars melting at -56.3°C, but the SynFlinc CO2 standards clock right in at -56.6°C. Also ThCO2 = 31°C (critical) for the unknowns, which looks about right for CO2. Colligative properties should normally result in a freezing point depression when impurities are added, I thought. Any ideas?

FYI: The inclusions are essentially pure CO2 vapour, effervescing from a moderate salinity, relatively low-T,P liquid (deep epithermal environment). I'm using a Linkam TH600, with no other known foibles.

Thanks for the help,
Jeremy Richards

Jeremy P. Richards, PhD
Associate Professor, Economic Geology
Dept. Earth and Atmospheric Sciences
University of Alberta
Edmonton, Alberta
Canada, T6G 2E3
Tel: (780) 492-3430 (messages: 3265)
Fax: (780) 492-2030

Dear fluid inclusionists:

Thanks to those of you who made helpful suggestions about the odd CO2 inclusions that appear to melt at -56.3°C. I followed up one practical suggestion, that it may have been due to a dust particle between the wafer and the silver block on the Linkam stage, by inverting the wafer and running it again. Somewhat to my relief, this made no difference (the implication that this may have had such an effect would have been worrying -- how would you know there was a problem, unless you knew what result to expect and got a different one, as here?). I also checked that the block itself was clean (but the CO2 standard that gave -56.6°C was run on the same block anyway).
I'm also relieved that I didn't get flamed for asking a dumb question, and also that at least two other people reported seeing the same effect on occasion. These people did Raman analyses on their samples, and found nothing but CO2. One person suggested that a higher hydrocarbon might be to blame -- anyone in the oil business seen such an effect?

Best wishes,
Jeremy Richards

Dear Jeremy,

Notwithstanding the comments of other fluid inclusionists, I still suspect that you have a measurement problem with your inclusions. Does the solid CO2 melt over a temperature interval or instantaneously at exactly -56.3 °C? Is this temperature reproducible in individual inclusions and in many inclusions in the sample?

If you have three or more phases present at the melting point (e.g. solid + liquid + vapour + clathrate), then higher hydrocarbons will not be able to raise the melting point of solid CO2. You need to find a component that will form a solid solution with crystalline CO2. This is not inconceivable, but I can't imagine exactly what it could be.

I think it is more likely that either (1) you do not have the assemblage of phases at "TmCO2" that you think you do; (2) you have thermal gradients in the sample causing delayed melting; or (3) your pure CO2 standard is not pure CO2.

Good luck,
Larryn Diamond

Professor Larryn W. DIAMOND
Department of Mineralogy and Petrology
Institute of Earth Sciences
Montanuniversität Leoben
Peter Tunner Strasse 5
A-8700 Leoben

Telephone Office: ++ 43 3842 402 450
Telephone Secretary: ++ 43 3842 402 451
Fax: ++ 43 3842 470 16
e-mail Office:
e-mail Secretary:


Dear Jeremy,
I agree with some of what Larryn has said, although I am not as pessimistic as he is regarding a measurement problem. In ancient history, Barb Murck and I and Linc Hollister published melting temperatures of CO2 higher than -56.6 °C (Murck, et al., 1979, Am. Min., v. 63, p. 40, the fig. is reprinted in Burruss, 1981, MAC short course) in inclusions in ultramafic xenoliths. We didn't try to explain them, but they are as accurate and precise as we could do at that time with a Chaix-Meca stage.
The only way to get CO2 to melt at temperatures above the triple point is to form a solid solution with another component in the inclusions. As far as I know, there is very little published on solid solutions of CO2 and anything else, but it certainly would be worth a quick literature search to check. My guess is that the component is a small molecule like CO, SO2, or H2S, not a "higher" hydrocarbon. Hydrocarbons above C3 become increasingly more immiscible with CO2 liquid as carbon number increases and they are unlikely to be compatible in the solid state.
Larryn raised an interesting question about the phase assemblage present at CO2 melting. What is it? Is an aqueous phase present? Also, what is the Th to vapor?
Finally, you described the geologic environment as low T,P, epithermal. Is there anything in the rocks that would suggest a source for another fluid component, like sulfides, organic matter, or graphite?
Bob Burruss

Dr. Robert C. Burruss
U. S. Geological Survey, National Center MS956, 12201 Sunrise Valley Drive,
Reston, VA 20192 USA
telephone: 703-648-6144; fax: 703-648-6419; e-mail:



Just as an addendum, the sample I had with the apparently anomalously high CO2 melting was barite and not very high T (pretty weird in having 2- and 3-phase CO2 inclusions in the first place). In the light of Larryn/Rob's comments I wonder if the mystery component could be an oxidised sulphur species?

Jamie Wilkinson

Dr Jamie Wilkinson
Fluid Processes and Mineralisation Research Group
Royal School of Mines, Imperial College
London SW7 2BP, U.K.
Tel: (44)(0)171-5946415 Fax: (44)(0)171-5946464 Email:




If I may add some remarks about CO2 melting point, I would say that the component should not be H2S as it will decrease the melting point until -85°C sometimes. I have published data about this system with H2S in Bull.Miner. 107, in 1984.

Another point is as I worked on numerous hydrocarbon fluid inclusions containing a lot of CO2 measured by infrared spectroscopy and in all the cases I studied, The melting point of CO2 was -56.6 typical of pure CO2. This will agree with the immiscibility of CO2 at low temperature with heavy hydrocarbons.

Nicole Guilhaumou
Université Pierre et Marie Curie
Département de Géotectonique
4 place Jussieu 75252 Paris Cedex 05
Tel: 33 (0)144275235
Fax: 33 (0)144275085


Thanks again to all those of you who made further suggestions about CO2-rich FIs with apparent melting points above that for pure CO2.

There seems to be sufficient consensus that this is not an artifact: thermal gradients in the new Linkam stages are minimal -- I tried inverting the chip, and the standard was run in exactly the same way.

As for the purity of the SynFlinc CO2 standard, I would have to say that I think it unlikely that that group would have produced dud standards. However, I'll call Jim Reynolds (the vendor) today and ask whether he is aware of any problems.

That leaves a component that can form a solid-solution with CO2 resulting in an elevation of m.p. The first thing I did was to check the melting points of H2S and SO2, but these are both lower than CO2 (-85°C and -73°C), so I would expect a mixture to depress the m.p. of CO2 unless they mix in a highly deviant way.


In answer to Larryn's and Bob's questions:
The -56.3°C m.p. is reproducible in a group of 5 FIs, three of which appear to be essentially pure CO2 (probably a bit of water, but not visible -- i.e., the vapor phase), and two that were heterogeneously trapped with some water. The co-existing liquid-phase FIs reveal the presence of minor amounts of CO2 in clathrate melting above zero (no liquid CO2 visible).
The solid CO2 phase recrystallizes a couple of degrees above the m.p., but melts convincingly in one step at -56.3°C.
As for "a source for another fluid component," previous bulk gas chromatograph analyses of these samples (not specifically including gas-rich FIs) show a small but significant amount of methane, plus a significant slug of sulfur of unknown speciation, and some trace higher hydrocarbons (Richards, J.P., Bray, C.J., Channer, D.M.DeR., and Spooner, E.T.C., 1997, Fluid chemistry and processes at the Porgera gold deposit, Papua New Guinea: Mineralium Deposita, v. 32, p. 119-132). This is a bonanza gold system, so one would expect at least a little sulfur (bisulfide) in solution to carry the gold about. I would love this m.p. elevation to be due to the presence of H2S, but somehow doubt it for the reason above. Given that the system was hovering near the sulfide/sulfate redox boundary, and small amounts of barite do drop out, SO2 could also be the candidate, but similar doubts apply.

So, I'll root around in the chemical literature to see if anything is known, and will let you know if I find anything.

Best wishes,


Dear inclusionists,

I read with attention the discussion started by Jeremy Richards, concerning TmCO2 lower higher than -56.6. A key question, as underlined by Bob Burruss, is the nature of the phases present at the melting temperature. In fact, another possibility for such a TmCO2 in the sub-system pure CO2 is to have only liquid at the melting temperature. As the slope of the S+L curve is positive in this system, higher values of TmCO2 than -56.6 are possible. This means that the internal pressure at TmCO2 is higher than about 5.1 bars at TmCO2, density higher than 1.18 (Angus et al., 1978).
I don't know if it can be an explanation for the phenomenon related by Jeremy, nor if such a phenomenon has been described, but it is theoretically possible.

Best wishes,

Michel Dubois

Universite des Sciences et Technologies de Lille
U.F.R. des Sciences de la Terre - SN5
URA 719, Sedimentologie et Geodynamique
Visitez notre site web :

Page personnelle :
Tel : (33) 03 20 43 65 47
Fax : (33) 03 20 43 49 10
E-mail :

Thanks to Michel and many others who have responded to this question. What Michel suggests is similar to the effect you see in monophase water inclusions when they melt without a bubble being present -- artificially high m.p.s are obtained, with no meaning for salinity determinations.
However, in the case of the inclusions studied, they are two- or three-phase at room temperature: liquid CO2, vapor CO2, +/- visible water (interestingly, the group studied preserve excellent evidence for heterogeneous trapping of a CO2-rich vapor phase and liquid aqueous phase: several inclusions appear to be purely carbonic with no visible aqueous phase, but some others contain variable amounts of water and homogenize at unreasonable temperatures; liquid-rich aqueous inclusions are also present, and homogenize around 160°C). During melting, these carbonic inclusions all have a vapor phase present, and melt smartly at the same T (-56.3°C according to my system).

Bob Bodnar and I have gone a round on this, and we plan to check the stage calibration with a new CO2 standard. I think that in the end this will turn out to be a combination of a stage calibration problem, and possibly a thermocouple hysteresis efect. We observed the latter phenomenon once with new thermocouples on the USGS stage, and Jim Reynolds determined that if you take the stage down to too low a temperature (<-120°C) and bring it up quickly, even if you do the final melting measurement at a slow rate, the thermocouple can have some memory of its recent refrigeration and give a measurement off by several tenths of a degree. I played with the CO2 standard some more last week, and found that if I hovered around -56.7 for a while, cycling up and down by 0.1°C, that I could actually get the standard to melt at -56.5°C (not -56.6°C that I get with a direct approach). Creeping towards -56.3, but not there yet. Maybe the additional 0.1--0.2°C difference is down to the standard, but we hope to determine this shortly. I'll report back then.
Didn't expect this query to generate so much interest -- in fact I expected a sharp slap for not being aware of some obvious explanation. Maybe the list has got more forgiving since Jamie Wilkinson got flamed a couple of years ago for not having committed to memory the entire contents of the FIR volumes!
Thanks again for the help,


Jeremy et al.:

What you describe below concerning the melting of water inclusions without a bubble, and what Michel suggested are, I believe, two different phenomena. I believe that what Michel is referring to is EQUILIBRIUM melting, along the solid-liquid boundary, of solid carbon dioxide in high density carbon dioxide inclusions (density greater than the liquid density at the triple point), which gives useful and correct density data. What you were referring to (I think) with the water inclusions is the common problem of low Th inclusions (Th less than about 75°C) which lose their bubbles upon freezing owing to the approximately 10 volume percent increase in going from liquid water to ice. Then during heating to measure ice melting, the bubble fails to nucleate and melting occurs along a metastable extension of the solid-liquid curve into "negative pressure space". This often results in ice melting temperatures which are greater than zero degrees. In the latter case, the ice melting temperatures cannot be used to infer a salinity, whereas with the carbon dioxide inclusions described above the ice melting T can be used to determine inclusion density.

Cheers, Bob Bodnar


Dear inclusionists,

As a part of the discussion about CO2-melting temperatures started by Jeremy Richards, I would like to add the following:
As Michel Dubois pointed out pure CO2 along the melting curve melts at higher temperatures than the triple point temperature. This may explain high melting temperatures in several cases e.g. high-density CO2 inclusions in mantle xenoliths (Frezzotti et al. 1992, Eur.J.Mineral,4,1137-1153) and in migmatites (vdKerkhof & Olsen 1990, GCA, 54, 895-901). Melting temperatures as high as -50.8 °C are described here. However, these "superdense" inclusions (d > 1.178 g/cm3) can be easily indentified by their unique melting behavior, as, on warming, the bubble disappears in the triple point and the solid "dissolves" into the liquid at higher temperatures. The phase sequence on melting is S+G --> S+L+G --> S+L --> L. As the reflective index of the high-density liquid and solid are close it may be difficult to observe the solid phase in these cases. High-density CO2 can also be easily identified by its Raman spectroscopic properties.

The melting temperature of -56.3 °C is not that high and most likely due to calibration problems. Even when the reproducibility is quite good (say within 0.2 °C), the accuracy may not be better than 0.5 °C. As suggested also by Larryn and others temperature gradients in the sample may be important. The Linkam stage is based on heat transfer in the sample and therefore the properties of the sample (both of the standard and study sample) effect the temperature readings: the thickness of the sample, the polishing quality, the type of mineral!! (quartz, carbonate, halite etc.), and particularly textural features like grain size and cracks, may have effect. Even when temperature gradients in the sample are minimal, they are NOT zero. By studying the melting of K2Cr2O7 (at 398°C) I found gradients of up to 5 degrees. At lower temperatures gradients are expected much less (probably within some tenths of a degree), but the effect of larger cracks are amazing!! I wonder if Jeremy´s sample has very different geometric or textural properties compared to his standard sample. In my opinion this would be a more plausible explanation than additional components. H2S as suggested possibility can be excluded as it results only in lower Tm with eutecticum at -95.6 °C (Sobocinski & Kurata, 1959, A.I.Ch.A. Journal 5-4, 545-551).


Fons van den Kerkhof
IGDL- University of Göttingen
Goldschmidtstr. 3
37077 Göttingen /Germany


Dear all,

I would like to mention a point which I haven't read yet about these strange temperatures and CO2 standards. Actually, we never measure the true triple point of CO2, if certain amounts of H2O are present (which is the case for most standards): we are measuring the famous Q3 point. Q3 is the quadruple point 3 where CO2 (solid-liquid-vapour) occurs with clathrate. The TP conditions of this point is close to the triple point, but exact numbers are not yet estimated. I can imagine that a difference of 0.3 degree may be valid. Furthermore, clathrates are known for their metastabilities and inconsistent behaviour, therefore I can imagine that this phase is not always present causing this apparent range of CO2 melting temperatures.


Ronald J. Bakker