Mutual replacement reactions in alkali feldspars II: trace element partitioning and geothermometry

When I was taking care of the upstairs ICPMS lab back in my former university life, I would occasionally pick up stray orphan scientists who were interested in some mass spectrometry. These are people collaborating with ANU scientists on non-mass spectrometric matters, who want to get the trace element composition of something, but aren’t hooked up with one of the analytical gurus.

Ian was one such fellow. I think he was mostly here doing TEM work or something, but he came into lab one day wanting to do some laser work- something on alkali in feldspars.

Feldspars contain weight percents of alkali, so this didn’t sound terribly interesting. I set the machine up, tuned it up. OK, is there anything in particular you’re interested in?” I asked.

“I’ll need the smallest spot you can give me.”

I wasn’t a big fan of our small spot- it had 1/14th the area, and thus 1/14th the signal, of the standard size. “Why so small?” I asked.

“I’m trying to analyse exsolution lamellae. Also, I wouldn’t mind getting good numbers on rubidium, cesium and thallium.”

To make matters worse, it was a evolved unit from a layered intrusion, so the trace element compositions were way lower than from a granite.

What had been a painfully routine day had just turned into a challenge.

Fortunately, it was a challenge for which I was prepared. Rb and Cs are notorious in laser systems for having high and variable backgrounds due to reionization off the back of the skimmer cone from ejected electrons. We had a grad student who was looking at Rb and Cs in primitive, depleted melts, so I had been working on this problem for over a year. Suddenly, I had an application for all the work I had been doing.

So, I blew off everything else I had planned to do on what was supposed to be a routine out-of-the-lab-by-morning-tea day, converted everything into low detection mode, put the machine back together, tuned it up, checked the backgrounds, and off we ran. With enough trial and error, We eventually got a machine state and a run table that got most elements detectable most of the time, and by the end of the day the signals were looking semi-respectable. Cs in the albite ended up being too low for us to see, but everything else we got decent detections for. The next day was productive, and by the time we finished it was almost routine. Charlotte did the data reduction and Ian interpreted and plotted all of it up, so I never actually saw what the results meant. When I finally got the draft, I was astounded at the outcome. It’s a natural system, and yet it behaves just like it ought to. Who’d a thunk that?

Of course, the papers describes the analytical procedure used to collect the counts. But counts don't make a signal, counts in excess of background do. And while I'd love to tell you how the background was reduced, that paper just got rejected.

The abstract and word cloud are below:

Perthitic alkali feldspar primocrysts in layered syenites in the Klokken intrusion in South Greenland, underwent dissolution–reprecipitation reactions in a circulating post-magmatic aqueous fluid at *450_C, and are to a large degree pseudomorphs. These ‘mutual replacement’ reactions provide a perfect natural experiment with which to study trace element partitioning between sodium and potassium feldspars growing simultaneously. The reactant ‘phase’ was a cryptoperthitic feldspar consisting of low albite and low microcline in a coherent sub-lm ‘braid’ intergrowth and the product phases were ‘strain-free’ incoherent subgrains of low albite and low microcline forming microporous patch perthites on scales up to 200 lm. The driving force for the reaction was reduction of coherency strain energy. The mechanisms of this process are described in Part I. Five mixed braid perthite–patch perthite crystals were analysed for major and trace elements using laser ablation-inductively coupled plasma mass spectrometry with a 19 lm beam diameter. This gave bulk analyses of the braid texture, which were in the range Ab73–54Or45–27An4.3–0.8, but could resolve Ab- and Or-rich patches in patch perthite. The major element bulk compositions of the crystals were retained during the replacement reactions. Major components in patches plot on tielines in the Ab–Or–An ternary system that pass through or very close to the parent braid perthite composition and indicate local equilibrium on the scale of a few tens of mm. Many trace elements, including REE, were lost to the fluid during the deuteric reactions, but the effect is large only for Fe and Ti. Cs, Pb and Sr were added to some crystals. Plots of log distribution coefficient D for Rb, Ba, Pb, Eu2?, La and Ce between Or- and Ab-rich patches against ionic radius are straight lines, assuming eightfold coordination, and to a first approximation are independent of ionic charge. K also lies on these lines, and the smaller ions Na and Ca lie close to them. The best linear fits were obtained using ionic radii for [8]K and [8]Ca, but there is ambiguity as to whether [7]Na or [5]Na is most appropriate. The linear relationship shows that the listed trace elements are in the feldspar M-site rather than in inclusions. Tl is in M although an exact D could not be obtained. The very large Cs ion partitions strongly into the Or-rich phase but its D value appears to be less than predicted by extrapolation. The near-linearity arises because partitioning is occurring between two solids into sites which have similar Young’s moduli, so that the parabolas that normally represent trace element partitioning between crystals and liquids (which have negligible shear strength) approximately cancel out. Ga and Be are in T-sites, as well as some of the Fe and Ti present, although part is in oxide inclusions. The site of Sc is unclear, but if structural it is likely to be T. Partitioning on M-sites is a potential geothermometer but because the effective size of the irregular M-site is defined by its K and (Na ? Ca) contents, which are controlled by ternary solvus relationships, its calibration is not independent of conventional two-feldspar geothermometers. Trace elements may however provide a useful means of confirming that feldspar pairs are in equilibrium, and of recognising feldspar intergrowths produced by non-isochemical replacement rather than exsolution. Two-feldspar geothermometry for the ternary phases in the low-albite microcline patch perthites gives temperatures above the stability range of microcline, markedly so if a correction is made for Si–Al ordering. This is probably because current geothermometers are too sensitive to low concentrations of An in ordered Or-rich feldspars. This interpretation is supported by two-feldspar assemblages growing at known temperatures in geothermal systems and sedimentary basins.

Parsons, I., Magee, C., Allen, C., Shelley, J., & Lee, M. (2008). Mutual replacement reactions in alkali feldspars II: trace element partitioning and geothermometry Contributions to Mineralogy and Petrology, 157 (5), 663-687 DOI: 10.1007/s00410-008-0358-1