![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
kareinaLast night when I finished up my uni work for the evening I opened up LiveJournal to do my daily progress report, read the few posts that had come in, read some geology blogs on GoogleReader, shut down the computer, did yoga with ![[livejournal.com profile]](https://www.dreamwidth.org/img/external/lj-userinfo.gif)
clovis_t and went to bed. Somehow I forgot to actually do the report.
It strikes me as an odd time to forget too, since I actually have good news to report--you'd think it would be the "nothing much accomplished" days wherein I'd "forget". :-)
Yesterday turned out to be a very good day for uni work--I actually managed to keep my time spent reading e-mail/livejournal/blogs to a minimum and worked most of the day and evening (other than a bit of social time and a walk to the waterfall withbaronsnorri. In addition to getting stuff done, starting early in the day, with no tendency to procrastinate (as has happened too often recently), I also managed to get a difficult sample to actually work. This particular sample comes from the area with the highest grade metamorphic rocks in all Tasmania. Some of the garnets in this area grew to over two centimetres in length. Therefore it was one of the first samples I tried when I first obtained the program with which I'm calculating the temperature and pressure based upon the chemical composition of the entire sample. Alas, from the very beginning this sample caused me difficulties. When I first tried it I discovered that the calculations thinks that it is not possible to have a garnet as rich in iron as this sample is growing in a rock of this composition. Over the course of the last couple of years I have tried a variety of different adjustment trying to correctly model this sample.
The "standard" correction to the whole-rock composition before using it in calculations is to assume that 90% of the iron measured in the sample is in the form FeO and the rest is Fe2O3. Since one of the problems with this sample is "not enough FeO" to make the garnet, I tried assuming that 100% of the iron is FeO. While this helped, it didn't raise the level high enough.
Another problem with the sample was the CaO is very low. The "standard" correction to the whole rock composition includes the assumption that all of the phosphorous present in the sample is in the mineral apatite--therefore we subtract enough CaO from the total to account for the phosphorous present. This sample being so low in total CaO that the program was having problems modelling the calcium in the garnet I tried assuming that only 90% of the CaO needed to make the amount of apatite possible from that amount of phosphorous was actually used for phosphorous. This helped, but not enough.
Yet another difficulty this sample presented was that it contains tourmaline. Tourmaline is not one of the minerals normally included in the solution set used with Perple_X, in part because it is a very complex mineral in terms of its crystalline structure, and in part because it is one of the few "commonly" occurring minerals to contain the element boron. When doing this sort of modelling, one generally limits the calculations to a fairly short list of elements, usually described in terms of their oxides. My standard calculations consider only the list Na2O, MgO, AL2O3, K2O, CAO, TiO2, MnO, FeO, SiO2, and H2O, which is generally enough to describe the minerals present, and is short enough as to keep the calculation times down to something reasonable. However, reasonable early on in my project, when I was first trying to make this sample work, I stumbled upon a paper wherein they expanded the list of solution models for Perple_X to include B2O3 and the various end-members of tourmaline. So I e-mailed the authors, got a copy of their expanded solution set and gave it a try. Again, it was better, but still not giving a *good* match for my garnets.
I tried using all three of the above differences to the "standard" technique together in various combinations, and discovered that the closest match I could achieve was to use all at once. It wasn't great, but it was better than nothing at all. However, when I tried to use that as a starting point for garnet fractionation calculations, to see how the conditions would change over time as the garnet grew (and, given the size of these garnets, grow is something they did!) I ran into yet more problems--the calculations kept running out of CaO before the predicted garnet became as rich in CaO as the real garnet is (garnets, at least in my field area tend to start out their growth with very little CaO and increase their concentration over time, often reaching five times as much on the rim as it has in the core). Unfortunately, when the program ran out of CaO it tended to crash, and trying to start over from just before it ran out, but with CaO removed from the list of possible ingredients, simply didn't work. However, with none of them working, I'd long since given up on this sample and worked on the others. Over the past few weeks I've been having good luck with some of the samples which were having a *very* different problem (predicting the presence of a mineral which is not present) by telling it that instead of assuming that the sample is water-saturated that there is only a limited amount of H2O. Since the problem mineral is one which includes H2O in its crystal structure reducing the amount of H2O available to the calculations caused the program to quit predicting the mineral. Having gone through and gotten all of the other samples working, and compiling a table of which settings were tried for which sample and what the result were, I realized that I was down to just the one sample for which nothing had worked, but that I hadn't tried limiting the amount of H2O for that sample. Not expecting anything, I decided to give it a try (on the version with all three of the changes I detailed in the above cut), just to be complete. And it worked! It not only worked for the garnet core predictions, but when I tried the "path" calculations it worked there too! This time it did *not* run out of CaO, but actually predicted that the garnet increases in CaO to the level actually measured in the sample, at the same time as it reaches the correct level of FeO and MgO too.
Given the importance of this sample in terms of location and intensity of metamorphism, it is a relief to *finally* have some results estimating the conditions at which it formed!
clovis_t and went to bed. Somehow I forgot to actually do the report. It strikes me as an odd time to forget too, since I actually have good news to report--you'd think it would be the "nothing much accomplished" days wherein I'd "forget". :-)
Yesterday turned out to be a very good day for uni work--I actually managed to keep my time spent reading e-mail/livejournal/blogs to a minimum and worked most of the day and evening (other than a bit of social time and a walk to the waterfall with
baronsnorri. In addition to getting stuff done, starting early in the day, with no tendency to procrastinate (as has happened too often recently), I also managed to get a difficult sample to actually work. This particular sample comes from the area with the highest grade metamorphic rocks in all Tasmania. Some of the garnets in this area grew to over two centimetres in length. Therefore it was one of the first samples I tried when I first obtained the program with which I'm calculating the temperature and pressure based upon the chemical composition of the entire sample. Alas, from the very beginning this sample caused me difficulties. When I first tried it I discovered that the calculations thinks that it is not possible to have a garnet as rich in iron as this sample is growing in a rock of this composition. Over the course of the last couple of years I have tried a variety of different adjustment trying to correctly model this sample. The "standard" correction to the whole-rock composition before using it in calculations is to assume that 90% of the iron measured in the sample is in the form FeO and the rest is Fe2O3. Since one of the problems with this sample is "not enough FeO" to make the garnet, I tried assuming that 100% of the iron is FeO. While this helped, it didn't raise the level high enough.
Another problem with the sample was the CaO is very low. The "standard" correction to the whole rock composition includes the assumption that all of the phosphorous present in the sample is in the mineral apatite--therefore we subtract enough CaO from the total to account for the phosphorous present. This sample being so low in total CaO that the program was having problems modelling the calcium in the garnet I tried assuming that only 90% of the CaO needed to make the amount of apatite possible from that amount of phosphorous was actually used for phosphorous. This helped, but not enough.
Yet another difficulty this sample presented was that it contains tourmaline. Tourmaline is not one of the minerals normally included in the solution set used with Perple_X, in part because it is a very complex mineral in terms of its crystalline structure, and in part because it is one of the few "commonly" occurring minerals to contain the element boron. When doing this sort of modelling, one generally limits the calculations to a fairly short list of elements, usually described in terms of their oxides. My standard calculations consider only the list Na2O, MgO, AL2O3, K2O, CAO, TiO2, MnO, FeO, SiO2, and H2O, which is generally enough to describe the minerals present, and is short enough as to keep the calculation times down to something reasonable. However, reasonable early on in my project, when I was first trying to make this sample work, I stumbled upon a paper wherein they expanded the list of solution models for Perple_X to include B2O3 and the various end-members of tourmaline. So I e-mailed the authors, got a copy of their expanded solution set and gave it a try. Again, it was better, but still not giving a *good* match for my garnets.
I tried using all three of the above differences to the "standard" technique together in various combinations, and discovered that the closest match I could achieve was to use all at once. It wasn't great, but it was better than nothing at all. However, when I tried to use that as a starting point for garnet fractionation calculations, to see how the conditions would change over time as the garnet grew (and, given the size of these garnets, grow is something they did!) I ran into yet more problems--the calculations kept running out of CaO before the predicted garnet became as rich in CaO as the real garnet is (garnets, at least in my field area tend to start out their growth with very little CaO and increase their concentration over time, often reaching five times as much on the rim as it has in the core). Unfortunately, when the program ran out of CaO it tended to crash, and trying to start over from just before it ran out, but with CaO removed from the list of possible ingredients, simply didn't work. However, with none of them working, I'd long since given up on this sample and worked on the others. Over the past few weeks I've been having good luck with some of the samples which were having a *very* different problem (predicting the presence of a mineral which is not present) by telling it that instead of assuming that the sample is water-saturated that there is only a limited amount of H2O. Since the problem mineral is one which includes H2O in its crystal structure reducing the amount of H2O available to the calculations caused the program to quit predicting the mineral. Having gone through and gotten all of the other samples working, and compiling a table of which settings were tried for which sample and what the result were, I realized that I was down to just the one sample for which nothing had worked, but that I hadn't tried limiting the amount of H2O for that sample. Not expecting anything, I decided to give it a try (on the version with all three of the changes I detailed in the above cut), just to be complete. And it worked! It not only worked for the garnet core predictions, but when I tried the "path" calculations it worked there too! This time it did *not* run out of CaO, but actually predicted that the garnet increases in CaO to the level actually measured in the sample, at the same time as it reaches the correct level of FeO and MgO too.
Given the importance of this sample in terms of location and intensity of metamorphism, it is a relief to *finally* have some results estimating the conditions at which it formed!
