Troilite analysis problems

[BenjaminWade]

Hi all
I am hoping that someone can give me some advice about analysing Troilite. What I thought should be very straight forward is turning me crazy. To give you some background I have a user who is attempting to analyse some troilite (amongst other sulfides which are coming out fine), and we are continuously overestimating the Fe content getting ~64.5% Fe (Stoichiometric troilite is Fe: 63.53% S: 36.47%). We are pretty much spot on with our S measurements. Beam conditions used are 20kV/20nA with a range of beam sizes from focussed to 10um. I am assuming 20nA is OK to use, perhaps I should be reducing it? I turned on TDI's at one point and had a look at the Fe signal and it is pretty rock solid.

Initially we were getting much too low Fe contents (~61%) until I discovered that for whatever reason (Fe-S vs Fe-S2 bonding?) compared to the Fe standard/s I peaked on (Astimex marcasite, P&H Chalcopyrite) there was quite a large peak shift in Fe towards lower values in the troilite. I checked the peak position for Fe in my 3 standards and they are within a few sintheta units of each other, but quite different to that in the troilite sample (~10 units). Following this discovery at the start of every troilite analysis I was peakscanning for Fe (and S) and was hoping this would fix my low Fe problem, but has seemed to now push it too far the other way with high Fe.

As such I am getting totals of ~101-101.5% with Fe anywhere from 64.5-65% no matter which of my limited FeS standards I use (Astimex marcasite, Astimex Pentlandite, P&H Chalcopyrite). I have had a literature scour and there are obviously a number of papers that have analysed troilite via EPMA, and its interesting to note is that in a lot of them Fe is always underestimated by ~1wt% with no explanation as to why (peakshift?).

Does anyone have any experience successfully analysing troilite and ideas of what I am doing wrong here. I am suspecting a standardisation problem but cant say for sure. Does a troilite standard exist? I know of the meteorite Canyon Diablo of which the troilite is used as a Sulfur isotope standard. Not sure if they are homogeneous in Fe/S though.

Cheers

[Probeman]

Hi Ben,
I might be remembering this incorrectly but I seem to remember that Troilite is very beam sensitive in that the electron beam can change the composition.

I would try two tactics: first try using a larger beam/lower beam current, second try the TDI correction.

I would be very interested in seeing what you find.
john

[BenjaminWade]

Hi John
Ok thanks. I had already tried the TDI to no real effect and the Fe counts looked pretty solid, but might try lowering the beam current to 10nA and trying an even bigger beam. Stay tuned.

Cheers

[Karsten Goemann]

I haven't done troilite myself (only pyrrhotite) but it sounds like you should definitely use the shifted Fe peak position for the troilite. Maybe it's not only the peak shift but also the shape? Have you tried integrated intensities?

[BenjaminWade]

Ok great thanks Karsten, and no I haven't actually investigated the peak shape. Good idea I will try the integrated intensities.

 Yes weird thing is they are also running pyrrhotite which is coming out fine...

[D.]

Have you played with different matrix corrections?

Deon.

[John Donovan]

I haven't done troilite myself (only pyrrhotite) but it sounds like you should definitely use the shifted Fe peak position for the troilite. Maybe it's not only the peak shift but also the shape? Have you tried integrated intensities?

The Fe Ka peak shape is unlikely to change, but sulfur Ka, absolutely yes.  You aren't using the Fe La line are you? 

But if you are using the Fe Ka line, I'm not sure why a sulfur peak shape change (actually it's a peak shift from would sulfate to pyrite to pyrrhotite), would cause the Fe intensity to be too high.

[Brian Joy]

I’ve gotten good results for troilite using synthetic pyrrhotite (Fe7S8) as a standard for Fe and S; I’m not aware of any beam sensitivity issues, nor have I ever noted any obvious shifts in Fe Ka peak position or change in shape amongst different sulfides, other minerals, or Fe metal.  Keep in mind that the Fe contents of the standards you are using (marcasite, chalcopyrite, pentlandite) are all much lower than that of troilite.  Note also that Fe in pentlandite is subject to fluorescence by the Ni K lines.  I tried once to analyze pyrrhotite using pyrite as a standard for Fe and S (taking the small shift in S Ka peak position into account) and got very poor results (low totals), though I can’t remember if one element or the other was more problematic.

[jon_wade]

the biggest problem is FeS is rarely bang-on stoichiometric. Additionally, depending on formation conditions, it can also contain appreciable oxygen. 
This has some consequences for estimating the chemical activity of Fe in FeS, but I won't bore you senseless!

[BenjaminWade]

Hi all
Awesome, thankyou all for the comments.

Definitely using the Fe Ka line. I am a bit puzzled now then. I definitely did see a peakshift in Fe on the users troilite sample compared to my Fe standards. I haven't had time yet to properly check the peak shapes but that is the next thing I will do. I will also check S Ka.

Brian can I ask where you got your synthetic pyrrhotite from? I have attached a copy of the paper they are trying to duplicate the methods of in which the authors are attempting to very accurately measure the Fe/S ratio in troilite, from which they calculate structural information and verify by HR-TEM. You can skip down to page 719 where they have their microprobe data in which they also mention they use an in-house natural Ni-free pyrrhotite from Dalengorsk mine in Russia. I can find many mineral specimens of pyrrhotite from Dalengorsk on ebay etc but quite reticent to spend a couple of hundred dollars on a specimen that may well be useless.

They also they use 15kV for their troilite. I guess there is no real problem with that for the levels of the elements and lines they are measuring. Could be another thing to try as well.

cheers

[Brian Joy]

Hi Ben,

The synthetic pyrrhotites that I use were made by R.G. Arnold (see Econ. Geol. 57:72-90) and Louis Cabri in the 1960s and 1970s.  Unfortunately I don’t have any loose material.

Natural terrestrial, essentially stoichiometric troilite can be obtained from some localities in Oregon and far northern California (Del Norte County), where it occurs in serpentinite in the Josephine Ophiolite.

Have you thought about trying to synthesize pyrrhotite/troilite yourself?  I haven’t done it, but it should be relatively easy to do by sealing Fe powder/sponge and sulfur in an evacuated silica glass tube and then heating in a muffle furnace.  It ought to be possible to produce stoichiometric material by saturating the system in Fe (and so then weighing the starting materials precisely wouldn’t be critical).  I can send phase diagrams if you’d like.

Brian

[BenjaminWade]

Hi Brian
That would be great if you could send the phase diagrams through if you have them on hand. This student is heading back to China and he says they can attempt to create it in his lab back home as they have had some experience synthesising Fe sulfides pretty much exactly as you describe. If I can give him the phase diagrams that would be great. I will also see if I can chase up some hand specimens of serpentinite from those localities.

I am going to try a recalibration at 15kV with peak integration today but not holding out much hope. I also noticed that using XPP over Armstrong/Love alleviated the problem a little but don't think I am justified in forcing the data to fit that way, I still think its a standard problem. Will let you all know how it goes..

Cheers

EDIT: Hi all. I am in the process of trying the integrated intensities, and noticed that the wavescan it is doing on Fe and S is very fast. I can change the number of steps no problems in the Elements/Cations box, but not matter what count times I change for Wave/Peak scan I cant get it to slow down. Is there some way I can change the count times of the wavescan it does during the Peak integration scan?

[Brian Joy]

Hi Ben,

I've attached a paper showing details of the Fe-S phase diagram.  Hope this is helpful.

Brian

[BenjaminWade]

Excellent, thanks Brian.

Cheers

[BenjaminWade]

Just to reply to my own thread...
John in an earlier post I edited I was having a problem slowing the peak scanning down for integrated intensities:

EDIT: Hi all. I am in the process of trying the integrated intensities, and noticed that the wavescan it is doing on Fe and S is very fast. I can change the number of steps no problems in the Elements/Cations box, but not matter what count times I change for Wave/Peak scan I cant get it to slow down. Is there some way I can change the count times of the wavescan it does during the Peak integration scan?

Can this be done or is this a fixed speed? I imagine it can be done given you can change the step sizes for the integration scan.

cheers