How Fast is too Fast?

Jon Naden

Robert Bodnar

Larry Meinert

Jamie Wilkinson

David Norman

Fons van den Kerkhof

Robert Bodnar

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

Hi all,

What heating rate should I use to measure phase changes? is a question that someone new to the game of microthermometry commonly asks. The answer is usually a somewhat nebulous -- " as slow as is practicably possible". For me, this usually means 5 deg C per min -- a compromise between accuracy and speed (it takes forever to measure phase changes at =< 1 deg C per min). I've looked around and there are no published data on the effect of heating rate on observed phase transition temperature.

Recently we acquired a new Linkam MDS 600 stage, and I thought before I used it in anger that I would try and answer this question. Below are some interesting results and I'd thought I'd post them to the list to see if they are reproducible.

Rather than taking the traditional approach to measuring phase changes, i.e heat the sample slowly (1 deg C per min or in my case 5 deg C per min) across the transition temperature and record the phase change by hitting the temperature store button, I took a slightly different approach. I used the limit facility to approach the phase transition in steps of 0.1 deg. I first set the upper limit 0.5 deg C below the phase transition temperature and then rapidly cooled the sample by several degrees (lower limit). The temperature was then cycled between the two limits, increasing the upper limit by 0.1 deg C until the phase change occurred. I recorded the phase transition from the limit temperature rather than hitting the temperature store. I also varied the rate at which the programmer approached the upper temperature limit.

The results were obtained using Bob Bodnar's synthetic FI standards with the stage in its default settings (i.e. uncalibrated):

-------------------------------------------------------------------

CO2 melting ("theoretical" temperature -56.6)

Initial cycling range -67 to -57 deg C; increment -57 deg C limit by 0.1 deg C after each cycle; cooling rate in all cycles 30 deg/min; heating rates and temperatures for each phase transition as listed below

Rate TmCO2

100 deg C/min: -56.7

50 deg C/min: -56.7

30 deg C/min: -56.7

10 deg C/min: -56.6

5 deg C/min: -56.6

-------------------------------------------------------------------

Clathrate melting ("theoretical" temperature +10.1 deg C)

Intial cycling range 7 to 9.5 deg C; increment 9.5 deg C limit by 0.1 deg C after each cycle; cooling rate in all cycles 30 deg/min; heating rates and temperatures for each phase transition as listed below

Rate Tmclath

100 deg C/min: 10.2

50 deg C/min: 10.2

30 deg C/min: 10.2

10 deg C/min: 10.2

5 deg C/min: 10.2

-------------------------------------------------------------------

ice melting ("theoretical" temperature +0.0 deg C)

Intial cycling range -3 to -0.5 deg C; increment -0.5 deg C limit by 0.1 deg C after each cycle; cooling rate in all cycles 100 deg/min; heating rates and temperatures for each phase transition as listed below

Rate Tmice

100 deg C/min: 0.1

50 deg C/min: 0.1

30 deg C/min: 0.0

10 deg C/min: 0.1

5 deg C/min: 0.1

-------------------------------------------------------------------

water critical point ("theoretical" temperature 374.1)

Intial cycling range 370 to 373.5 deg C; increment 373.5 deg C limit by 0.1 deg C after each cycle; cooling rate in all cycles 100 deg/min; heating rates and temperatures for each phase transition as listed below

Rate ThH2O(critical)

100 deg C/min: 375.1

50 deg C/min: 375.0

30 deg C/min: 374.9

10 deg C/min: 375.0

5 deg C/min: 374.9

As you can see, using this method, the phase change temperature is independent of heating rate within error (+/- 0.1 deg C) and, with the exception of the critical point of water, within 0.1 deg C of the theoretical temperature (bear in mind the stage was NOT calibrated and no correction factor was applied for the different heating rates).

For me these results show how stable the temperature control on the

stage is. Also, the accuracy of this "cycling" approach to measuring phase changes means that phase changes can be measured easily, rapidly and accurately if you have some sort of temperature programming facility.

Has anyone tried similar experiments, as it would be interesting to know how reproducible my results are?

Regards

Jon Naden

JNA@wpo.nerc.ac.uk

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

Jon et al,

A few comments to follow-up on your results about heating rates. Very commonly I have people come to my lab who say that they could not see any phase change during heating from low temperatures to room temperature. When I ask them what heating rate they used, they invariably say "as slow as possible". My reply is always "that is the reason why you didn't see a phase change". The phase change is often so subtle that with a very slow heating rate one does not perceive the change because it occurs over a long period of time. The instructions that I give to all new students, and the approach I use myself, is, after cooling the sample to liquid nitrogen temperatures, to heat the sample to room temperature as fast as possible (>100C minute) and to watch a single inclusion continuously during heating (don't take your eyes away even for a second to look at the temperature). By doing this, even the most subtle phase change (melting of very small amounts of carbon dioxide, initial melting near the eutectic, final ice melting of low salinity fluids) are very obvious. Then, once you know what to look for, you repeat the experiment using progressively slower heating rates until you are able to measure the phase change temperature to whatever precision you desire. This approach takes less time overall than trying to use a slow heating rate from the beginning, and usually helps to identify phase changes that would otherwise be missed with only a slow heating rate.

Way back when we first started doing synthetic fluid inclusion work (20 years ago), Mike Sterner and I measured the temperatures of various phase changes in inclusions where we knew ahead of time what the temperature should be. The bottom line was that for almost any homogenization temperature measurement or temperature of halite or sylvite dissolution, it is almost impossible to heat the sample too fast to get a reliable temperature. Of course, slower heating rates have to be used for ice melting, but even for this measurement relatively fast heating rates can be used. It is my experience that most fluid inclusionists waste far too much time trying to use slow heating rates and trying to reproduce the temperature 3 or 4 times. My recommendation is that fast is good and once is enough. There are very few studies in which one needs Th numbers more precise than +/- 5 C (not including low T gas homogenization) or ice melting temperatures more precise than +/- 0.5 C, and this level of reproducibility is obtainable with a very fast heating rate.

So, Jon, I agree with your conclusion that the heating rate is not important for most fluid inclusion measurements, and most workers could save lots of time by using faster heating rates than they currently do.

Bob

Dr. Robert J. Bodnar
University Distinguished Professor and
 C.C. Garvin Professor of Geochemistry
Department of Geological Sciences
4044 Derring Hall
Virginia Tech
Blacksburg, VA 24061-0420
Tel: (540) 231-7455 (O)
 (540) 953-2448 (H)
 (540) 353-2448 (Cellular)
Fax: (540) 231-3386
e-mail: bubbles@vt.edu
http://www.geol.vt.edu/profs/rjb/rjb.html

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

On the subjects of FI heating rates, I agree completely with Jon and Bob. I instruct students to first figure out what accuracy is needed for a particular study (typically, 5°C, 0.5°C are more than adequate for heating/melting, respectively) and then to NEVER measure more precisely than that. For some people this is very difficult to do as there apparently is some little voice in the back of our heads that whispers, "a more precise number is better". Furthermore, using the FLINC gas flow stage, that can drop the temperature very quickly, I measure homogenization temperatures in 5°C increments by looking at whether the vapor bubble "pops" back in or grows continuously upon cooling, rather than trying to observe the temperature at which the vapor bubble disappears. Thus, at a particular temperature, say 355°C, I look at all the inclusions of interest and if the bubble is still there, I move (quickly) to 360°C, etc. When I reach a 5°C increment, say 385°C, where I think the bubble is gone, then I drop the temperature (quickly) to see if the bubble grows back "continuously" or whether it "pops" back in after undercooling 10-20°C or more. If the latter, then I know the bubble was there at 380°C and had homogenized by 385°C. This is far quicker and allows work on far smaller/fuzzier inclusions than would be possible if one is trying to observe the actual homogenization at 383.7°C. Again, the critical step is in deciding what accuracy is necessary for a particular study. I review lots of manuscripts where the stage accuracy is reported as 3°C at 375°C and then the data are reported as 383.7°C, etc.

 

Larry Meinert
Department of Geology
Washington State University
Pullman, WA 99164-2812
Office: 509-335-2261
Sec:	509-335-3009
FAX:	509-335-7816
meinert@wsu.edu

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

Just to add a couple of comments to the discussion, I would in general agree with Bob and Jon, with the exception of measurement of hydrate melting and sometimes halite/sylvite dissolution. Particularly with hydrates, rates of 5 C/min would result in some degree of overshoot - this may be more a problem of chemical equilibration within the inclusion than thermal equilibration. The phase changes Jon reports (pure CO2 triple point, pure water triple point, pure CO2 clathrate, Th) are relatively "easy" transitions to observe, are not subject to dissolution kinetic problems and perhaps are the most likely to be insensitive to heating rate. In practice, I tend to measure phase changes by stepping the temperature up in increments (at whatever interval of desired precision) once I know the approximate transition temperature from a fast run, a la Bob. I'm not sure whether clathrate melting temps could be routinely measured accurately using fast heating rates, but this would save a lot of time especially where cycling is required. My gut feeling is that the difference would be not that great, but the problem may be in actually observing the final melting temperature during relatively fast heating as opposed to cycling since the continued presence of clathrate can only be confirmed by the distortion of the vapour bubble on recooling.

The Linkam stage (MDS600 and THMS600 varieties) has a small thermal mass and a very fast response to temperature changes - using a small sample chip (couple of mm, 100um thick), the inclusion response is virtually instantaneous. This can be seen when close to homogenization of the carbonic phases of a CO2-H2O inclusion (or a near critical density aqueous inclusion) when stepping the temperature up 0.1C immediately results in an increase (or decrease) in the bubble size. Indeed the pulsed heating effect of the THMS stage can be observed in the vapour bubble under such conditions. Furthermore, in such thin samples there is no discernible vertical thermal gradient through the sample (try measuring an inclusion near the top of a sample with the chip upside-down).

Jamie Wilkinson

j.wilkinson@ic.ac.uk

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

I have students perform experiments like the one you summarize at the bottom of your email for my fluid inclusion class. Generally I have them try rates of 0.1, 0.3, 0.5, 1, 5, 10 and 20 C/min. As a result we have lots of data on heating rates and the reproducibility of phase-change temperatures. We use a Linkam stage. The following affects the required heating rate:

1.The distance of an inclusion to the lower surface of the wafer

2.Quality of the wafer-stage contact

3.Themal conductivity of the mineral

4.Specific heat, heats of vaporization and melting of the inclusion liquid

5.Size of the inclusion

It is good to be aware of the variables that affect heating rates, but practically, empirical data obtained by repeating measurements using various rates works best.

Our guidelines are that work for thick and thin wafers, flat and rounded sections, small and 100+ micron inclusions are:

For measurements below 30 C use a maximum rate of 0.3 C/min, preferably 0.2 or 0.1 C/min.

For measurements above 30 C use a maximum rate of 1 C/min, preferably 0.5 C/min

Heat at 2 C/ min from -30 to ice melting in order to obtain reproducible clathrate melting temperatures.

Small CO2-filled inclusions in thin wafers will reproducibly melt and homogenize at higher heating rates, but my advice is to take measurements with patience. This is a small time-sacrifice with the Linkam stage because a high heating-rate can be used up to a few degrees below the phase change.

David I. Norman
Professor of Geochemistry
Dept. of Earth and Environmental Science
New Mexico Tech
Socorro, NM 87801
phone: office 		 505-835-5404
		 home 		 505-835-3004
		 fax office	 505-835-6436
email: office	dnorman@nmt.edu
 home 	dnorman@rt66.com

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

Dear all,

As a reaction to the discussion going on about stage calibration I agree with Jon Naden, Bob Bodnar and others that the heating rate should not (at least not always) be taken very slow to measure phase transition temperatures and that the effect of the heating rate on the measurement accuracies are minor. We are satisfied with the results by using rates of 1 o/min for CO2 and water melting, 5 o/min for CO2 homogenization, 10 o/min for water homogenization (Linkam stage). Only in "special cases" (e.g. sluggish reactions like recrystallization, dehydration) slower heating may be required. Most important is to keep a routine procedure in order to get best temperature reproducablities. Of course the same procedure (heating rate) should also be used for measuring the STANDARDS; the error made by making corrections using standard values is mostly larger than the error caused by different heating rates. As pointed out by Dave Norman the quality of the sample may have some effect on the temperature calibration: these factors include sample size, the polishing quality, mineral content (biotite!), contact with the stage and particularly fractures in the sample. Fractures may function as important "barriers" for heat transfer. Care should be taken when measuring highly fractured quartz! Some years ago I made some experiments with melting compounds (KNO3; Merck40; K2Cr2O7) in between cover (and sapphire) plates and found significant temperature gradients between the center and the rim of the opening (2.5 mm wide) of the stage: the gradients appeared to be up to 3 degrees centigrade for 398 oC (melting K2Cr2O7) by using heating rate of 1 o/min. That means that the position of a fluid inclusion in the sample (and shifting the sample) are expected to be important when striving for highest accuracies. However, within the accuracies normally required for getting geologically meaningful conclusions, these variations are not essential. According to our experience heating rates do not significantly effect the temperature accuracies, but other factors considering the sample quality and the stability of the electronics (!) have larger influence. These effects are however small and within the accuracy normally required and the temperature accuracy is not (any more) a big problem in microthermometry. A careful observation of the phase transitions is more important than reaching the highest possible accuracy in the measurement.

Regards,

Fons van den Kerkhof
IGDL - University of Göttingen
Goldschmidtstr. 3
37077 Göttingen
Germany
akerkho@gwdg.de

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

Just a short follow-up to the discussion of heating rates.

Much of the information that has been distributed during the past few days is "instrument specific", and some newcomers to the fluid inclusion game may not be aware of this. For example, Fons commented about thermal contact and the effect of fractures. These are only important if one is using the Linkam or, to a lesser extent, the ChaixMeca stages. Fractures and thermal contact are not important if one is using the USGS gas-flow stage. In that case. the location of the thermocouple relative to the inclusion(s) being measured is a more important factor. So, when discussing what might affect the accuracy or reproducibility of flinc measurements, it is important to indicate which equipment is being used to make those measurements.

Bob

Dr. Robert J. Bodnar
University Distinguished Professor and
 C.C. Garvin Professor of Geochemistry
Department of Geological Sciences
4044 Derring Hall
Virginia Tech
Blacksburg, VA 24061-0420
Tel: (540) 231-7455 (O)
 (540) 953-2448 (H)
 (540) 353-2448 (Cellular)
Fax: (540) 231-3386
e-mail: bubbles@vt.edu
http://www.geol.vt.edu/profs/rjb/rjb.html

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