Thread by @manenbu

We just had a short paper out on the topic, so let's talk oxygen fugacity (ƒO₂), ok?
Thread.
doi.org/10.1093/petrol…

ƒO₂ is one of the most confusing concepts in petrology, and it's understandable. It is not real. It does not exist. I has zero physical meaning (most of the times). It's easier to say what it isn't instead of what it is. Let's start with some examples.

1. ƒO₂ is not 𝑝O₂! That is, ƒO₂ does not equal the pressure of oxygen gas. First, in most cases there is no oxygen gas at all, so it's can't be the pressure. Second, O₂ is not an ideal gas so the thermodynamic potential of O₂ isn't the same as the pressure.

Confusingly, ƒO₂ is usually measured in units of pressure (e.g. bar). Even if you take supercritical pure O₂ in geological pressures, the ƒO₂ will be an order of magnitude higher, when measured in the same units.

2. ƒO₂ is not measure of how oxidised or reduced a rock is. Frost (1991) has a wonderful example that refutes this. Just by changing the contents of non redox sensitive elements and keeping redox-sensitive elements fixed, ƒO₂ will change.

For example, adding Cr to a mantle rock will trap more Fe³⁺ in spinels, potentially lowering ƒO₂. In contrast, adding more Ca will add cpx which (if charge-balanced) will trap more Fe³⁺ in cpx, increasing ƒO₂.

Likewise, fixing ƒO₂ but changing composition will change the redox state of redox-sensitive elements. It is well known that adding alkalis (particularly Na) to melts at a constant ƒO₂ will cause oxidation of Fe to Fe³⁺.

This leads to number 3: Rocks are not buffered. Melts are not buffered. If a basalt assimilates felsic rocks with lots of feldspars, there is no free oxygen around to oxidise the iron in order to maintain ƒO₂. There is no reason to think ƒO₂ will be the same, or buffered.

If a rock sequence plots on a certain ƒO₂ buffer (FMQ, NNO, etc) it is ENTIRELY by coincidence. The rock/melt is not buffered.

4. ƒO₂ does not control anything (at least not directly). In most rocks and melts, it is the ferrous/ferric ratio that is important. Atoms don't care about a made up number, they care about the other atoms in the system.

ƒO₂ is an experimental trick. Because in natural system there is usually no gain or loss of oxygen, but in an experimental system oxygen can freely move (directly or indirectly) in most experimental setups, we needed a way to control it somehow.

This was why buffers were invented. NNO, FMQ, all those things you hear about all the time. ƒO₂ can be controlled and measured. Ferrous/ferric is much harder to control and measure.

This is why we calibrate things against ƒO₂ and not ferrous/ferric. Not because it's better, but because it's easier/possible. That's also why we care:

ƒO₂ is important for speciation of minor or trace elements in magmas an rocks (e.g. Ce, Eu, S, C, H, V, etc). ƒO₂ is an artificial mathematical construct invented by experimentalists that is determined (sometimes rather arbitrarily) by the ferrous/ferric ratio of the system.

However, this ƒO₂ is the controlling variable in various calibrations for things that are not ferric/ferrous ratios. It is an intermediate number that is "easy" to control in experiments and works out nicely in calculations.

In other words, ƒO₂ is a good measure of very-small-variations-that-make-a-difference™ in oxygen contents of a system that has oxygen as the most abundant element. It is a measure of the amount of oxygen (actually electrons but as geologists we don't talk about those) that...

...can move between redox-sensitive elements. ƒO₂ by itself is meaningless, just an arbitrary number. ƒO₂ relative to an experimental buffer (which has zero physical meaning in nature) is a useful reference point and a basis for common language.

5. ƒO₂ values of a source rock and its derived-melt are never the same, except by coincidence. ƒO₂ of cumulate rocks and residual melt are never the same except, except by coincidence. In an evolving system, ƒO₂ can go up, down, stay, or change trajectories several times.

That's why buffering ƒO₂ in modelling software is usually wrong, unless you have evidence that ƒO₂ remained the same (or nearly so) during the melt evolution. But even then, it wasn't because it was actually buffered in nature. It just happened, in this particular case.

Partial melt derived from rocks can be either more oxidised or more reduced that the source rocks can be either more oxidised or more reduced (in ƒO₂ terms), depending on rock composition an melting conditions.

Just thinking out loud here. That's enough ranting for today. I might add some more stuff if I think of something useful. For more information refer to the paper in the first tweet of this thread.