TC_CombTC_Comb is a wrapper for the THERMOCALC program of R. Powell and T. Holland running in multi-equilibrium geothermobarometry modes (only avPT+Reactions now). This program can make THERMOCALC geothermobarometry more effective and comfortable.
 The main program features are:

  • easy and fast entering of input data;
  • fast (re)calculation of end-member activities using AX2 program of T. Holland;
  • primitive but effective permutational approach for calculations of PT-parameters based on many sets of mineral analyses made from a single small area of thin section;
  • estimation of H2O and CO2 activities or mole fractions (since version 1.1);
  • additional functions: filtering of reactions by the presence of specified components, tracing of selected reaction line through a series of plots, etc.;
  • visualization (plotting of PT-averages, error ellipses, reaction curves and their uncertainty bands) makes results of THERMOCALC thermobarometry more transparent for users. TriQuick (another program which acts as a server and editor of diagram graphics) is required for working with diagrams;
  • output of all required information in tabular form.
TC_Comb window

 TC_Comb is a freeware application.

Read about important changes in TC_Comb version 1.1

 How to work with TC_Comb vers. 1.0 (basic operations demo):


Thanks for creating this software, it looks very promising!
Here's a first question for you: we can't find anything in the help menu... We see the topics, but no text. Any idea?


DimaDD's picture

This is a known issue with CHM help files related to evolution of Microsoft security politics (any downloaded CHM-files are regarded now as potentially dangerous objects and blocked by default). The solution is:

1. Right-click the .CHM file and choose "Properties"
2. On the "General" tab, click the button labeled "Unblock"
3. Click "OK"

It is from
Another source (with pictures and one more cause):

I'm also planning to duplicate documentation in PDF format.

I get this alert from TC-COMB : Error double calculation at the same calcTatP varable detected (T/P) possibly due to inconsistent file Rct.kbd (use/create another one for this system). calculations terminated.

How can I solve this problem. I am not expert on thermocalc.

DimaDD's picture

The simplest solution:
- place original file Kbd.lib from the distributive to TC_Comb directory (if this file has been modified while experimenting with interactive runs of Thermocalc in TC_Comb),
- press the "Kbd.lib" button in TC_Comb,
- select "Default" in the list (the first item),
- press the "Use" button (TC_Comb will show message about copying of avPT.kbd and Rct.kbd).

Some background info:
Thermocalc can calculate two kinds of tables with PT-coordinates of reaction curves: either T for the range of fixed P values (CalcTatP=Yes) or P for the range of fixed T values (CalcTatP=No). Default keyboard macro in TC_Comb (Rct.kbd) creates both of them one after another, then the program operates all the data simultaneously to avoid problems with sub-isobaric and sub-isothermal [parts of] curves.

While you can create/modify/select keyboard macros in TC_Comb, it's not necessary to use these features in regular work and the standard "Default" set of macros is enough. So, there is no need to use buttons "Kbd.lib" and "avPT.kbd" (or "Rct.kbd") and to uncheck "Batch mode" below the button "Thermocalc" (i.e. interactive runs of Thermocalc are not required usually in TC_Comb). I think I should add the option like "Expert mode" to the program to hide these buttons and checkbox...

DimaDD's picture

Reply to the question appeared on the TWQ_View page (


It seems that this is the problem in TC_Comb, not TWQ_View...;)

Anyway, yes: this is due to cordierite! This mineral requires input of H2O and CO2 activities in Thermocalc even if the system is "dry" (doesn't contain explicit fluid components in the tc-ax.txt file). Therefore the keyboard macro "Default" is not suitable in this case and you should try another one. Press the "Kbd.lib" button in TC_Comb, select the macro named "Cordierite (aH2O & aCO2)" in the opened window and press the "Use" button.

BTW, you can double-click on any record in the "Results" window (even with failed calculations) and look at the Thermocalc screen output in the separate window. Type of calculations (avPT or Rct) can be selected in the main TC_Comb window. This way may be more convenient than checking of tc-log.txt.


Julien's picture

Hi Dimitry,

First thank you very very much for your software that actually reconcile me with Thermocalc ;) So much easier to use it now, and the ability to plot reactions in just one click is just awesome! I do have a problem, however, for some calculation in upper amphibolite / lower granulite facies rocks. I am working with Al-rich pelite composed of quartz, plagioclase, K-feldspar, biotite, cordierite, and sillimanite. I successfully run AX and got the correct activity for all phases (and added as pure minerals quartz and sillimanite). Plagioclase activity has both an and ab components, and K-feldspar activity are "san" rich with some "ab" component. The problem is that I cannot use BOTH plagioclase AND K-feldspar in the same Thermocalc calculation. TC_Comb ask me to choose plagioclase OR K-feldspar, with the error message "Similar phases found: Pl and Kfs. Use Pl?"... Is this a bug or a limitation of the software?

Thanks in advance for your help!

Julien Allaz

DimaDD's picture

Hi Julien,

Thank you for the high appreciation of the program! Yes, it is the sad THERMOCALC (and TC_Comb) limitation: we can not use several phases consisting of the same components in a single avPT session. Thus, for example, two-felspar thermometry is not available. However, it's possible to get around this problem.

First of all, this system (Crd+Bt+Pl+Kfs+Qtz+Sil) has only one Ca-bearing component "an": it means that no an-bearing reactions can be written, therefore "an" should be excluded. The same situation is with "ab": it is the only Na-bearing component of the system. So, plagioclase is a useless phase at all ;) and you can select Kfs only.

Though, "an" is important for garnet-bearing rocks (due to reactions with grossular). If "an" is a required component, you can select plagioclase for "an" activities and add "san" + its activity to the list of pure minerals together with quartz and sillimanite, i.e. something like this:

q sill san 0.93 H2O

You can try different activities (e.g. extremal "san" activities taken from analytical data) to see dependence of results on them.

NB! It's important that all reactions with "san" in these systems are associated with the decomposition of biotite (the second K-bearing phase), that is why I've added also H2O to the list of pure phases. Another important feature of this Grt-absent system: there are only 3 linearly independent reactions, or 2 IR if water is excluded :(

If cordierite and H2O are used, you should select the proper keyboard macro (use the "Kbd.lib" button to do this). "H2O and CO2 activities" is a good choise in this case. Then you can set some fixed aH2O and aCO2 values in the "Settings" window and perform calculations with them. After this you can select e.g. "Fixed: aCO2" (under the "aH2O" and "aCO2" input fields), check the "Calc fluid" checkbox instead of "Fluid from the table" (at the bottom of the "Results" window) and press the "Recalculate" button for selected successful combinations which demonstrate some PT results. TC_Comb will try to find the most suitable aH2O value for fixed aCO2. You can also try to select "Fixed: none" (to find aH2O and aCO2 simultaneously). However, I've found that THERMOCALC fails at some aCO2 values for this system... :(

Julien's picture

Great suggestion! Thanks for this quick and more than complete response.

Yes, the rock are Grt-bearing (an omission in my previous post). Typical new garnet growth after Bt-related melting (however, some biotite remain stable after this reaction). I already have beautiful PT-estimate using TWQ, and I was looking for a comparison of the absolute PT-data with Thermocalc. To my surprise, the temperature estimate with Thermocalc and only Pl OR Kfs yield slightly lower temperature than TWQ, but roughly the same (within the error). I will try the trick with san 0.93, and also make sure to include H2O (I can only assume pure water, although some CO2 might be present. I might try also an evaluation of the "optimum" H2O-CO2. My experience - 10 years ago! - with Thermocalc and amphibolite facies pelite yielded relatively coherent aH2O values, around 0.9, which is what I expected with Al-rich pelite. On the contrary, TWQ was often giving me abnormal results (0.3 xH2O or less).

You also have a good point about Na... I always keep it thinking about the paragonite component in muscovite and the albite, but of course muscovite is long gone in this sample, and therefore there is no more Na-exchange possible... Except maybe with cordierite, but there is very small amount of Na in these cordierite (around 0.1 wt-%) and I doubt there are any good thermodynamic data for Na-cordierite, yet.



DimaDD's picture

A small remark about water: water activity may be substantially reduced not only because of addition of CO2 but also due to other fluid components (like NaCl, etc.) or due to melting. Thus, just H2O-CO2 binary is a too simple model and it's better to work only with activities of H2O and CO2 rather than with mole fractions. aCO2 in this system is required only for "tuning" of cordierite activity and its value may be too far from reality. Therefore "inconsistent" aH2O/aCO2 values are quite possible and should not be surprising.

Though, I have also obtained unrealistic aH2O values with TWQ (near zero!) but for "true" Hbl amphibolites (using database with amphiboles from Mader & Berman, 1992). According to our experience, TWQ with BA96A.DAT (we call it "granulitic database":) usually calculates reasonable water activities, especially if reactions in the "dry" subsystem are well converged.

ttmuedi's picture


Guys i need assistance on pasting my electron microprobe data on a software platform. I get the message saying ID number is absent.

DimaDD's picture

Tables for TC_Comb should contain columns with mineral abbreviations and with IDs (numbers) of analyses. These identifiers are required to know which analysis of some mineral is used in particular combination. TC_Comb determines the column with IDs of analyses by its header. You can find list of possible headers in TC_Comb.ini, line #44:

Analysis=Point Spot Probe Analysis An

You can freely edit this list and add headers used in your own microprobe data tables, but make sure that it is not intersected with the other editable lists with headers (for columns with mineral abbreviations, area ID, totals and selection flags).

DimaDD's picture

The program has been updated to

  1. A sad bug causing hanging of AX_2.exe at repeated calculation of activities (for updated PT conditions) has been fixed. This bug has been revealed on new versions of Windows only.
  2. Another problem is the error message while launching Thermocalc from TC_Comb on Windows 10 (64-bit). The solution is the running of TC_Comb and/or Thermocalc in compatibility mode: right-click on an exe-file in and select "Properties", then select the "Compatibility" tab, check the box next to "Run this program in compatibility mode for:" and select Windows 8 or 7 from the drop-down list.
  3. A new experimental option "No avPT" has been added under the table with results of calculations. Checking of this option suppresses avPT mode of Thermocalc at re-calculation of results and performs simple geometric analysis in the way similar to INTERSX program of the TWQ package (but not the same): TC_Comb calculates so-called "simple averages" and some measure of scatter of reaction curves (RMSD or RMSWD, depending on state of the "Settings" window → "Reactions" tab → "Weighted Simple Averages" checkbox). PAvg and TAvg values are placed to the "P" and "T" columns of the table and RMS(W)D values – to the "fit" column. In the most cases, these "simple averages" should not be used as replacements of avPT results; however, you can use them in the case of only two linearly independent reactions (when avPT is not applicable). A more detailed description of "simple averages" will be published in the nearest future.
EricKelly's picture

Hi Dmitry,

Thank you for making TC_Comb. I have been using it with Windows 10 for a couple of weeks and it's very helpful and robust.

I have a question about your calculation of error ellipses. In Holland and Powell (1988, page 179) they describe using a value for n of ~2.5 for an error ellipse of 95% confidence. However, it seems that you use ~1.25 in TC_Comb. Can you explain why you use 1.25? I know others do this too, but I don't see the reason.


DimaDD's picture

Hi Eric,

Thank you for the interesting and important question!
It's not so easy to answer it in two words, but I will try...

First of all, I've implemented plotting of error ellipses in TriQuick, where ellipses may by described in two forms: pure geometric notation (center coordinates, width, height, slope) and statistical parameters (center coordinates, 2 sigmas on X and Y, correlation coefficient, variance, confidence level). TriQuick re-calculates statistical parameters to geometry before drawing. You can find and edit error ellipse statistical parameters in TriQuick (with a loaded TC_Comb diagram): open the "Editor" window, select error ellipse in the tree of objects and press the button "Stat.". TriQuick can also calculate error ellipses for clouds of points on XY scatter diagrams (plotted using its "Table" and "Variables" windows) for desired confidence levels. I've tested this functionality using independent statistical software (e.g. freeware PAST3) and I'm satisfied with it. I've placed the working algorithm to the Excel workbook ErrorEllipse.xls (you can download it from GoogleDrive or YandexDisk), it may be helpful.

The next difficult question was: what are these sd(T) and sd(P) values in the THERMOCALC output in fact? I've tried to reproduce error ellipse on Fig. 3 from Powell and Holland (1994, page 124) using data on activities from the Table 4 (the same paper, page 128), old THERMOCALC v.2.4 and old database TH.PD (generated 23.05.90), supposing the ellipse was calculated for conf. level 95%. Almost the same ellipse (but at somewhat lower PT) can be reproduced only if sd(T) and sd(P) are treated as 2 sigmas, not 1 sigma. Therefore I decided to place sd(T) and sd(P) values to TVL files "as is" without doubling them. I've put parameters calculated for this ellipse to the Excel workbook avept_DDD.xlsx (GoogleDrive,YandexDisk) as the first sample. This file is primarily the avept.xls of David Waters but with some my additions (just to compare results): Bi-scalar function approximation used in TriQuick (rough but enough precise for ellipses, I think) and drawing of better ellipse.

I'm not sure (as usual))) that this way is entirely correct but I hope that is so.