SXFive Tactis - bad totals

[Miss_Boron]

Hi,

We have a SXFive Tactis since one year, and it worked great until early November. Since then, we get bad totals for oxides wt. % for some minerals (olivine ~ 104%, plagioclase ~ 106 %), while other minerals are ok (pyroxene, garnet).
We have obviously tried to recalibrate, and to calibrate with different standards with no results.
We also have tried different crystal configuration, or spectro configuration, with no success.

Did anyone already encountered this problem? Any idea on how to fix that?

Thanks for your help.
Celine

[Mike Jercinovic]

Hi Celine, That's pretty weird!  I'm sure you've already tried this, but I would first try to setup a new quanti settings file from scratch.  I assume you are setting these as GeoSpecies, right?  You might try processing with a different species and see what you get, or running as oxide without a species and see how that goes.  If you have other things that are running okay with the same standards, etc., it seems species specific.  If you calibrate on a plag (Ca, Si, Ca anyway), then run that plag as an unknown, what happens? 

[Miss_Boron]

Hi Mike,

Thanks for your answer. Yes, we tried to do a new setting from scratch, it's what I called recalibration.
We don't use the "geo" mode, just the standard one. But it worked well for years doing so (on the old SX100, and until last month on the Tactis).
We calibrated on San Carlos olivine then re-run the olivine as unknown. The totals (oxide wt. %) were at 103%. And we did that with different calibrations, and with different setting files too.

Thanks
Celine

[sem-geologist]

Hi Celine,
While we have no Tactis, we are running SxFiveFE (alongside SX100), and as far as I know the spectrometers are the same, and part of electronics are the same. We had seen this like behavior many times, and it can have different causes. Devil in the details.
I am going to give this in some ordered list, albeit it can be argued to be very subjective order.

• Old/too thin/thick coating of Sample and/or Standard. Old coating of sample (the older the worse) gives lower totals, old coating of standard will lead to lower counts of standard and thus higher totals of unknown. In particularly, rod-carbon coaters should be avoided at all cost, as it gives very poor coating quality when compared with superior coating done with carbon-thread method. The lower average mass of the unknown substance - the higher influence the thickness of carbon coating has to totals.
• Standard/Unknown measured with too high current which leads to unaccounted pile-up dead-time of spectrometer counting electronics; SX50;SX100;SXFive;(SXTactis?) use the same pre-amplifiers and same counting principles (albeit a few components and voltage levels had changed - but principles are still the same). Although, behind the window is 2020 (soon 2021) the new micro-probes still use the state-of-art from 80'ies. I would be not mad if there would be no alternatives, but already the few generations of pre-amplifiers had been developed, and so there is actually no excuse.
• Spot size is important for minerals which have significant part of light elements (Mg in olivine); We had seen countless times that olivines (especially Mg ones) do not like anything above 10 nA and often for totals to behave nicely some small defocusing of beam is required (2-4 um, it depends from olivine); It is easy to get 105% of totals when using 20nA and very finely focused beam on olivines with close to Forsterite composition (Fajelite is different storry).
• There is of course external stability issues, which can influence the analysis (temperature, humidity)
• The TAP crystal is often blanked (to get better spectral resolution and to blank tail-artifacts of peaks), from one side with copper plate, from other - with crappy plastic piece. That plastic piece can partially pop-out, and thus cast variable X-ray shadow onto crystal. We had such anecdotal situation: At morning everything nicely coming to ~100%; at afternoon 98%, at night 96% and 95%. After doing new calibration (after night run), totals back to ~100% until some longer pause. After pause, using new calibration, the totals were going up to ~104%. All these total variations were on the behalf of Al and Si (Na was not so affected). The problem was that lousy plastic piece - after prolonged run it was being heated up by x-rays and so the piece was bending more and casting the larger X-ray shadow on TAP. Analogically, during brake/pause it was cooling down, and so initially casting a smaller shadow (until heating up again).
  

Generally to troubleshoot You need to know not only totals, but how higher/ lower than expected counts/concentration are per element basis. If all elements are too high/ or too low - that's clear - it is most probably the coating. If only second and subsequent elements (per spectrometer) gets higher concentration, and first measured elements gets ok, or a bit lower - that is insufficient current reduction and insufficient spot size. If behavior changes with time on all spectrometers - that external problem. If variations are observed on a single spectrometer - that is spectrometer problem.

[theo_nt]

Hi Celine,

it sounds you have a hardware problem with your spectrometer. Try to use the same setup  but with another spectrometer with TAP (if you have one) to see if you have the same problem.

all the best
Theo

[jon_wade]

Celine - did you ever sort this?  if not, look back at old calibrations and see if the count rate has dropped off.  That some minerals come out 'ok' and others *high* suggests someone has  stuck a grubby finger on the standard block