Higher gain needed for PHA Xe counter over time?

[Julien]

Dear all,

Our relatively new JEOL-8230 (installed April 2019) has shown some "early sign of ageing" () just after 2-3 years... Whereas I got quite stable PHA bias value (at constant gain) on my P-10 counters, I realise quickly that the bias was strongly shifting in the course of just 1-2 years on PET monochromator especially. The effect is also visible on LiF (to some extend) so this suggest a counter problem, not a monochromator problem). The strange evolution of the bias / gain depict below occurred simultaneously on my two Xe counters (one is a L-type spectrometer, the other an H-type, both with PET + LiF).

At first, my optima for the two Xe counters on PET were at a gain of 16 to reach around 1700 V for a pulse at 4 V. Now, after 2 years, I have to use a gain of 32 (or 64 for lower X-ray energy lines beyond 140 mm spectrometer position). Same on the LiF, now my optimum bias is only achievable with a higher gain.

John D. suggested me that Xe-counter should be replaced every 3-5 years... But seeing that much change in just 2 years surprises me, and I didn't recall having seen this behavior on my other 8230 that was installed in Colorado. I cannot pinpoint the exact time I opt for higher bias on my newest probe, but it was in spring / summer last year, so definitely 2 years after installation.

Is this just a normal and unavoidable behaviour (i.e., loosing Xe-gas => need to increase the gain to get optimum bias around 1700 V)? Does anyone have any experience with that? My local (German) engineer has no idea why it occurs but they have seen it on other instruments...


Julien Allaz

[Probeman]

Our relatively new JEOL-8230 (installed April 2019) has shown some "early sign of ageing" () just after 2-3 years... Whereas I got quite stable PHA bias value (at constant gain) on my P-10 counters, I realise quickly that the bias was strongly shifting in the course of just 1-2 years on PET monochromator especially. The effect is also visible on LiF (to some extend) so this suggest a counter problem, not a monochromator problem). The strange evolution of the bias / gain depict below occurred simultaneously on my two Xe counters (one is a L-type spectrometer, the other an H-type, both with PET + LiF).

At first, my optima for the two Xe counters on PET were at a gain of 16 to reach around 1700 V for a pulse at 4 V. Now, after 2 years, I have to use a gain of 32 (or 64 for lower X-ray energy lines beyond 140 mm spectrometer position). Same on the LiF, now my optimum bias is only achievable with a higher gain.

John D. suggested me that Xe-counter should be replaced every 3-5 years... But seeing that much change in just 2 years surprises me, and I didn't recall having seen this behavior on my other 8230 that was installed in Colorado. I cannot pinpoint the exact time I opt for higher bias on my newest probe, but it was in spring / summer last year, so definitely 2 years after installation.

Is this just a normal and unavoidable behaviour (i.e., loosing Xe-gas => need to increase the gain to get optimum bias around 1700 V)? Does anyone have any experience with that? My local (German) engineer has no idea why it occurs but they have seen it on other instruments...

Hi Julien,
I was hoping that someone with more JEOL experience would chime in, but I would guess that you simply have a leaky Xe detector and that it's getting "pumped out" over time a little quicker than usual.

All sealed detectors leak over time (too bad we have to locate them in a vacuum chamber!), but some could be more leaky than others and therefore require quicker replacement.

Another reason I am hoping that JEOL and Cameca start offering solid state WDS detectors soon!

https://probesoftware.com/smf/index.php?topic=644.0

[Brian Joy]

Hi Julien,

The fact that you’re seeing the same behavior on two different X-ray counters suggests that the problem is not due to leaking counter gas.  Also, if the respective spectrometers are L- and H- type, then count rates are relatively high.  My guess is that you’re setting the anode bias at too high a value and that this (excessive avalanching), combined with the high count rates, is causing excessive generation of ionizing radiation within the counter and possibly causing premature breakdown of the quench gas and subsequent contamination of the anode.  I generally operate using an electronic gain of either 32 or 64, and I try to keep the anode bias at 1650 +/- 50 V.  Even under these “gentler” conditions, I see shifts in the pulse amplitude distribution after just a couple years on the L- and H-type spectrometers with sealed Xe counters.  I also eventually see broadening of the pulse amplitude distribution and other anomalies, as I’ve described here:  https://probesoftware.com/smf/index.php?topic=528.msg7232#msg7232.  Over a given period of time, the development of PHA shifts and anomalies occurs much more slowly on my one spectrometer with PETJ/LiF crystals.

Brian

[Karsten Goemann]

We have two Xe counters on our 8530F+ which are still original, now 5 years old. Both are on L type spectrometers.
I'm still using the same gain settings as initially and the change in bias over time to have the PHA peak at 4V is small, e.g. Fe K alpha with gain 32, bias at 1685 V after installation, now at 1694 V after 5 years.
The overall PHA width doesn't appear to have changed dramatically but I haven't looked at this as thoroughly as Brian.
Both counters always had a little hump on the high side of the main PHA peak that I've also seen on some but not all Xe counters on other 8530s.

[Brian Joy]

Just for comparison, these are the Xe counter bias settings (with gain set at 32) that I’ve used to keep the Fe Ka pulse amplitude distribution centered at 4 V over time using LiFL and LiFH.  The counters have been in use since December, 2015:

Dec. 2015:  LiFL:  1640 V, LiFH:  1648 V
June 2018:  LiFL:  1664 V, LiFH:  1674 V
July 2021:  LiFL:  1674 V, LiFH:  1670 V

By the time I had the old counters replaced (at 4.5 years old), I had to use a gain of 64 and anode bias of 1644 and 1646 V, respectively, to get the distribution centered at 4 V (for Fe Ka).

The smaller change since June, 2018 could be due to the fact that I’ve generally tended to work at lower beam current in recent years, though I have no proof of cause/effect.

When the replacement counters were new, using gain = 16 would have required the bias on the anode wire to be set about 80 V higher to get a distribution centered at 4 V.

[Probeman]

Based on the above information from Karsten and Brian, I would guess again that Julien has had the bad luck to have two leaky sealed detectors that will need to be replaced.

Or maybe it's an electronic issue common to both, but again that would also be a coincidence.

[Brian Joy]

Based on the above information from Karsten and Brian, I would guess again that Julien has had the bad luck to have two leaky sealed detectors that will need to be replaced.

Or maybe it's an electronic issue common to both, but again that would also be a coincidence.

Or the anodes are contaminated.  It would be great to get more input from additional JEOL users.

[Probeman]

That is another possibility for sure. But why would Julien's anodes be so contaminated, but yours and Karsten's aren't after a much longer period of time?  Maybe a bad batch of Xe for some of these sealed detectors?

I agree more info from JEOL users would be very helpful.  Since Julien's engineer said he has seen this on other instruments maybe they share a similar manufacturing period or not? Either way, they probably have to be replaced sooner or later. Julien: do you see an increase in noise as you bump up the gain?

[Brian Joy]

That is another possibility for sure. But why would Julien's anodes be so contaminated, but yours and Karsten's aren't after a much longer period of time?  Maybe a bad batch of Xe for some of these sealed detectors?

I agree more info from JEOL users would be very helpful.  Since Julien's engineer said he has seen this on other instruments maybe they share a similar manufacturing period or not? Either way, they probably have to be replaced sooner or later. Julien: do you see an increase in noise as you bump up the gain?

It depends on patterns of use of the Xe counters when paired with LiFL/PETL or LiFH/PETH.  If they are used for mapping or wavelength scans at high current, then breakdown of the quench gas could be an issue.  Though I can't prove it, I believe that I prematurely "aged" my own Xe counters that were replaced in December, 2015.

In my experience, amplification of electronic noise is only an issue at gain = 128.  When operating at that gain, I find that I need to set the baseline voltage at 1 V rather than using the default 0.7 V.

[Karsten Goemann]

We did a lot of wavescans on our standards initially (even though not at very high current) after installation. Not so much these days.
We do some high current mapping for trace elements (but count rates there wouldn't be high overall), but primarily point analyses.

[Brian Joy]

I used to do a lot of work on uraninite, especially during the period 2011-2014.  This work involved point analyses, X-ray maps, and abundant wavelength scans to determine trace element contents and interferences (as you can find virtually anything in uraninite).  I typically ran the wavelength scans at a microamp of current.  The Th Ma map of a sector-zoned uraninite below, collected at 250 nA (in March, 2014), measures 1550 x 900 pixels and required about 32 hours to complete.  In the red/bright areas, ThO2 content ranges up to about 10 wt%, while in the blue/dark areas, it ranges down to about 6 wt%.  I used the PETH crystal for the map; Th Ma X-ray count rates within the mineral grain would have varied between about 10 and 20 kcps.  I covered the epoxy around the grain with copper tape.

By the time I had the LiFL/PETL and LiFH/PETH Xe counters replaced in December, 2015 (a month before the service contract ended), I would have needed to set the anode bias at about 1740 V for Fe Ka on either while using gain = 32 in order to get a pulse amplitude distribution centered at 4 V.  These counters had only been in service since April, 2011.  The “new” ones have been in use for more than six years (and currently require bias = 1670 V at gain = 32 for Fe Ka).

That’s my dirty laundry for the day.

[ericwgh]

That is another possibility for sure. But why would Julien's anodes be so contaminated, but yours and Karsten's aren't after a much longer period of time?  Maybe a bad batch of Xe for some of these sealed detectors?

I agree more info from JEOL users would be very helpful.  Since Julien's engineer said he has seen this on other instruments maybe they share a similar manufacturing period or not? Either way, they probably have to be replaced sooner or later. Julien: do you see an increase in noise as you bump up the gain?

The bias voltage for both Xe-sealed detectors at Utrecht University has been stable since 2015. I was going to replace them bc I remember seeing the 3-5 yr lifetime in this thread a while ago, but I currently don't see the need. Countrates have been stable as well. (PHA data can be retrieved easily with the "drift" program)
Eric