CL of uncoated samples?

[pgopon]

I realize that this is a EPMA forum, but here is a quick question from the imaging side of things. 

I have always been told to never try CL imaging of uncoated samples (we have a Hitachi S-3400 with VP mode), because the air bled into the chamber to keep the charging down would kill your CL signal and maybe break your CL detector.  This is generally what I tell people when they come to me for CL imaging, and I turn them away if their samples are uncoated.  Today however, I had a particularly insistent user who insisted that we try CL imaging with uncoated samples because they will be measuring trace C with SIMS in this sample. 

We devised an experiment were we took one of his less precious samples and coated 1/3 of it with the standard 20nm carbon, 1/3 we left uncoated, and 1/3 we coated with 0.6nm iridium.  What we found is that carbon was actually produced by far the worst images out of the three (both in terms of overall signal and ability to see zonations), and best images were taken on the uncoated sample in VP mode (with the iridium coated samples being almost as good as the uncoated). 

This goes against everything I have been told about CL.  Has anyone else ever been told to not CL image uncoated samples, or I did I just get bad advice from someone?  Now that I think about it again, it makes sense to me that uncoated would yield the best images since there will be no attenuation of the CL signal by the coating.  Can anyone else come up with a better reason why this work?  We have tried this is a few different types geologic samples now, and they all seem to show the same results.

I have attached three images taken under the exact same brightness/contrast conditions with the only difference being the coating and using VP mode (the fourth image is just a repeat of the third image with the brightness turned out as the original was too bright).

Cheers,

phil

[Mike Jercinovic]

Fascinating.  What current and voltage were you using?  Also, what is the target VP pressure in the chamber?  As CL is mostly visible light, I guess we wouldn't necessarily expect chamber pressure to be a big effect on emission intensity (don't think my monitor here would look that much brighter under hi vac), but you would certainly expect the electron absorption to be problematic in terms of the efficiency of CL generation.  If I put a 20nm carbon coat on my screen, it would certainly dull things just the way your images show for CL due to both absorption of electrons as well as emitted photons.  We generally use about a 4nm carbon coat for high vacuum CL for this reason.

[Probeman]

This goes against everything I have been told about CL.  Has anyone else ever been told to not CL image uncoated samples, or I did I just get bad advice from someone?  Now that I think about it again, it makes sense to me that uncoated would yield the best images since there will be no attenuation of the CL signal by the coating.  Can anyone else come up with a better reason why this work?  We have tried this is a few different types geologic samples now, and they all seem to show the same results.

I have attached three images taken under the exact same brightness/contrast conditions with the only difference being the

I know that Zeiss uses the CL emission the from the VP gas to electron image at low vacuum, and maybe Hitatchi does too?  I think this is because the FEI VP imaging mode patents are still in effect.

I know that Chi Ma at Cal Tech has told me that because they are using this gas CL emission to electron image (on their Zeiss VP SEM) it becomes problematic when the sample is highly CL emitting.

As for CL imaging, I could imagine that if the VP gas "flashed over" when the CL PMT is on, that could saturate the PMT, but it will eventually recover though it might take 12 to 24 hours to discharge internally.

Are you using a PMT or a CL spectrometer?

[pgopon]

Thanks for the responses,

Mike, we were using 10 keV, with a Hitachi current of 80 (on the unitless 0-100 scale, since we don't have a faraday cup on our SEM), and experimented pressures ranging from 20-100 pascal with little change in the image quality (the attached image used 30 pa).  I might play around with different coating thicknesses (I choose 20nA C and 0.6nA Ir since those are standard thicknesses that can be used in the EPMA and SIMS afterwards) to see what the least carbon and iridium we can coat with and still get good enough conduction for CL imaging and see if the VP image is still better than a thin coated one. 

John, our Hitachi has a ESED detector (environmental secondary electron) which Hitachi won't really tell us how it works or what it detectors, but it does require a fair amount of gas pressure to work properly (>80pa).  We have noticed that we do see some of the same zonations when use this detector as we see in CL.  We are using the Gatan monoCL (no spectrometer), just a parabolic mirror that sends the light onto a photomultiplier.  Maybe staying at lower pressures is what kept us from having the VP gas flash over.  I should try to actually see if I can induce this behavior with the max gas pressure to see if it is even possible.  I agree that the worst that should happen is that the detector would need a day to discharge, so shouldn't be too big of a deal ( i hope).  The detector has a autotrip in case it gets oversaturated anyways. 

[qEd]

I have heard of such a warning but assumed it was directed at FEI VP/ESEM users who have Gatan parabolic mirrors. The large metal object below the pole piece interferes with the electric field used by FEI to collect VP secondary equivalent images LVD? or whatever the detector is called. Since the BSE detectors are annular they are blocked when the mirror is inserted and hence those with this set-up must coat insulating samples.

I am not surprised the uncoated specimens give the best results for some reasons listed above. Luminoscope samples (predates SEM-based methods) are not coated.  The Gas flash could be an issue but as mentioned the HT trip should protect the CL PMT.

[John Donovan]

Ok, since we're discussing CL, I have another question: I am adding the capability of full spectrum CL acquisition to Probe for EPMA:

http://probesoftware.com/smf/index.php?topic=529.msg2913#msg2913

Which can occur simultaneously with the EDS full spectrum acquisition since they just both need the faraday cup out, but...  the dark spectrum acquisition needs to occur with the faraday cup in. That means that some time has to be spent waiting for the dark spectrum acquisition to complete before the cup is pulled out and the WDS and EDS and CL spectrum acquisitions begins... so I added a Dark Spectra Count Time Fraction parameter in Acquisition Options that will be based on the specified CL spectrum acquisition time...

And I'm storing the CL spectra in raw counts, but with the acquisition time so they can be displayed as raw intensities or normalized to cps and I assumed that so long as the CL and dark spectra are normalized to cps, I can just subtract the channels from each other to get the net CL spectrum intensities...

And I set the default dark spectrum acquisition time fraction to 0.1 because I figured that most people won't want to, for example, double their acquisition time just to get a CL spectrum...

But then a guy (Norm Hoffer) at Ocean Optics said that the dark spectrum should be acquired for the same amount of time as the CL spectrum because "The dark spectrum changes with changes to integration time" and I'm trying to understand how important this is, if the CL acquisition time is long (WDS acquisition time long).

So my question is: why does the dark spectrum change over time and how important is it that we acquire the dark spectrum for as long as the CL spectrum?