Cold stages, air jets and other anticontamination devices

[Anette von der Handt]

Dear all,

I am interested to hear about everyones opinions and experiences regarding anti-contamination devices available to microprobes.

From what I have seen/heard/talked to people, there are various lines of defense available:

Keeping the vacuum clean: Scroll pumps, cryo-baffles around diffusion pumps
Keeping the sample clean: Cold stage, Airjet, LN trap.
Starting out with a clean sample: Plasma cleaner, possibly attached to the air lock.

We have a LN trap right now but we never really use as it is awkward to use and you have to keep it going with Liquid Nitrogen or it dumps all the carbon back on. I am currently dreaming up the perfect new microprobe for light element work and beam sensitive samples, so I want to hear about successful or failed attempts (both over-the-counter solutions (by the vendors) as well as home made contraptions) of keeping it clean!

Thanks,
Anette

[Probeman]

I am interested to hear about everyones opinions and experiences regarding anti-contamination devices available to microprobes.

From what I have seen/heard/talked to people, there are various lines of defense available:

Keeping the vacuum clean: Scroll pumps, cryo-baffles around diffusion pumps
Keeping the sample clean: Cold stage, Airjet, LN trap.
Starting out with a clean sample: Plasma cleaner, possibly attached to the air lock.

We have a LN trap right now but we never really use as it is awkward to use and you have to keep it going with Liquid Nitrogen or it dumps all the carbon back on. I am currently dreaming up the perfect new microprobe for light element work and beam sensitive samples, so I want to hear about successful or failed attempts (both over-the-counter solutions (by the vendors) as well as home made contraptions) of keeping it clean!

Hi Anette,
Here is a cost effective and sustainable solution Cameca came up with for my SX100 bought in 2006 or so (see attached photos below- you'll need to be logged in to see them):

Basically, it's an air cooled "freon" compressor with the expansion valve attached to a set of vanes just above the diffusion pump. 

The unit is sold by PolyCold in Petaluma California, and Cameca engineered it to fit the instrument. I think they charged me about $28K for the entire system without the temperature readout and warmup/cooldown controller which I bought separately for another $4.5K.  This is significantly cheaper than the oxygen jet-cold finger solution usually offered by Cameca.

I find that not only does it just run without any LN2, but it also provides about the same level of hydrocarbon decontamination as the oxygen jet and cold finger system usually provided by Cameca. I had that oxygen jet and cold finger system on my SX51 at UC Berkeley and it was hard to regulate the oxygen flow and required refilling every 8 hours with LN2.  See here for some data from my instrument using a diffusion pump and another using a dry turbo pumped system:

http://probesoftware.com/smf/index.php?topic=140.0

One nice thing about diffusion pumps is that they don't fail in a spectacular fashion as turbos sometimes do, and they are much cheaper.  Though the newest turbo pumps are getting more reliable and less expensive.

The cold vanes run at around 100 degrees Kelvin and so not only pump hydrocarbons, but also water vapor!  About once a year we let it warm up with the gun valve closed and let the roughing pumps pump out the accumulated condensates.

That said, we recently did a complete column rebuild on the SX100 and found they were quite coated with a very tough film that was not affected by solvents, but cleaned off pretty easily using detergent.  We suspect it was vaporized epoxy from beam damage.
john

[Probeman]

Our engineer also makes the following observations for what they are worth:

What's amazing is that 100K baffle temp can be reached without any IR shielding from the hot DP zone below! Possible improvements for this design:

1) Slice that SS rod which supports the baffle and bridge the gaps with KEL-F couplings (that SS rod is acting as a thermal short, so good thing is that SS is a somewhat poor heat conductor)

2) Install a simple annular IR shield around the baffle to block IR from room temp metal nearest the baffle

3) Install an IR shield to block IR from DP zone below (most complicated but you yield best possible reduction in baffle temperature).

[Anette von der Handt]

John,

thanks for all the excellent information!

[Ben Buse]

Hi Anette,

Strange enough we were just discussing this at the UK Jeol Users meeting yesterday in Cornwall. As you point out the LN cold finger is a pain when it runs out. What we have been investigating and looking to developing is a pelter cooled cold finger. Hirsh et a. 1994 showed that only a small reduction in temperature is required on a cold finger. We ran a test in only SEM making a crude pelter cold finger and it showed a reduction in the contamination. I can send you the pictures and slides if you like.

We regularly use a LN cold finger when doing carbon in steels and low voltage analysis (where overvoltage very small). The LN finger greatly reducing amount of carbon depositing in vaccum. But as you say its a right pain when it warms up. With steels we find although we can prevent carbon intensity changing during analysis - the levels of contamination seem to vary from sample to sample (=/- 1000ppm) - despite using same method of polishing and cleaning. Suggusting need for plasma cleaning of samples prior to analysis - would work for steels - maybe not for carbon coated geological samples though!

Ben

[Probeman]

I recently ran some trace carbon analyses on a stainless steel. This is the first time since we did a complete cleaning of our sample chamber, including the Polycold cryo baffle described here supplied by Cameca on our SX100 instrument:

https://probesoftware.com/smf/index.php?topic=646.msg3823#msg3823

I don't think I've ever seen the system so clean:



It's getting *cleaner* over time!  Probably from burning off the native hydrocarbon layer.

And this is using an oil diffusion pump and a Alcatel oil mechanical pump!   I wonder does anyone else out there have a large (8") cryo pumping baffle in their EPMA vacuum system? 

[Jacob]

Hitachi recently introduced an electron flood based anti-contamination device: https://doi.org/10.1017/S1431927621002051

Sounds a lot like a charge neutralization flood gun from an XPS, actually.  Just with a different intention in mind.  Not sure how the fluxes would differ.

The advantage as I see it is that in a microprobe with the secondary electron detector bias and grid turned off the anti-contamination device could be left running during analysis if desired.  No degradation of the chamber vacuum is required for operation.  Most things that would be adversely affected by electrons are already shielded in same manner to prevent interference by secondaries.

[Probeman]

I think you could be correct, though on Cameca EPMA instruments there might be an issue with the 1 um polypropylene column separation windows...  I don't know.

Maybe they are out there on EPMA instruments, but I haven't heard of any. Has anyone?

[Probeman]

Here is a comparison of two line traverses. The upper one performed using a "dry pumped" EPMA and the lower one done this week on our cryo-pumped (100K) EPMA:



You can barely see where our traverse was acquired.  I have to say I am a little disappointed in the "turbo/scroll" pumped instrument carbon contamination rate!

[sem-geologist]

I never found that we would need any cryo-stuff on our SXFiveFE (turbo+scroll) or SX100 (diffusion+oil pumps). What kind of materials are those in picture? I am aware that too long a beam exposure would take coated carbon from sample (so there is thinning of the coating until it completely thins-out at hit spot) - that leaves mark in the coating. Carbon thinning is main factor limiting beam exposure on sample. At least I was so thinking, hitherto.

I am not doing metals often.

Your picture is suggesting that there is e-beam induced deposition of organics. I am aware that such deposition can be witnessed in its all gory details on apertures (for some apertures used for too many years it looks like shield volcano  ) or on sample stage beam parking position (our FEG instrument). I started to wonder why in one case we get coating disintegration and removal, and in other organic (hydrocarbon) deposition? But maybe I am completely wrong with a first, maybe there is no thinning of carbon coating at all, and those marks are deposited organic too? But why carbon coating quality then matters? considering same coating thickness of 20nm, with rod-based shitty coating the coating will go off in few minutes with 10nA; with carbon-thread single pulse coating coating stays for ten minutes with up to 100nA; with multiple-carbon thread composite layer coating 700nA for 10 minutes can be achieved. If there would be only the e-beam deposition of hydrocarbons, and no carbon removal by e-beam, there should be no such differences between coatings, beam dose for when absorption current abruptly drops, and charging starts should be similar for all case - but it is not.

So why in some cases do we observe contamination, while in other we get removal (cleaning) with same e-beam on same machine?

[Probeman]

I never found that we would need any cryo-stuff on our SXFiveFE (turbo+scroll) or SX100 (diffusion+oil pumps). What kind of materials are those in picture? I am aware that too long a beam exposure would take coated carbon from sample (so there is thinning of the coating until it completely thins-out at hit spot) - that leaves mark in the coating. Carbon thinning is main factor limiting beam exposure on sample. At least I was so thinking, hitherto.

I am not doing metals often.

Your picture is suggesting that there is e-beam induced deposition of organics. I am aware that such deposition can be witnessed in its all gory details on apertures (for some apertures used for too many years it looks like shield volcano  ) or on sample stage beam parking position (our FEG instrument). I started to wonder why in one case we get coating disintegration and removal, and in other organic (hydrocarbon) deposition? But maybe I am completely wrong with a first, maybe there is no thinning of carbon coating at all, and those marks are deposited organic too? But why carbon coating quality then matters? considering same coating thickness of 20nm, with rod-based shitty coating the coating will go off in few minutes with 10nA; with carbon-thread single pulse coating coating stays for ten minutes with up to 100nA; with multiple-carbon thread composite layer coating 700nA for 10 minutes can be achieved. If there would be only the e-beam deposition of hydrocarbons, and no carbon removal by e-beam, there should be no such differences between coatings, beam dose for when absorption current abruptly drops, and charging starts should be similar for all case - but it is not.

So why in some cases do we observe contamination, while in other we get removal (cleaning) with same e-beam on same machine?

Good questions. Yes, these are all uncoated metal specimens because the client was interested in trace carbon.

I really cannot speak about electron beam damage to carbon coats but I have observed a change in the coating coloration on coated metal specimens.

On uncoated samples Ben Buse and I (and others I am sure) have measured a decrease in the carbon signal in the first 5 to 10 seconds (or longer) of beam exposure  which I attribute to volatilization of the native hydrocarbon layer.

This by the way, is the rationale for the so called "decontamination time" option in the Acquisition Options dialog in Probe for EPMA. The idea being to delay the start of counting after the faraday cup is removed, until this native hydrocarbon layer is removed by beam exposure.

This same parameter can also be utilized for beam sensitive samples for the so called "incubation time", in this case waiting for the sample to heat up (and ion migration to commence), in order to ensure a linear (log) decay of the alkali intensities.

My observations on uncoated samples are that as (native) carbon is removed directly underneath the beam spot, elemental carbon is deposited from hydrocarbons in the vacuum system becoming "cracked", and then deposited adjacent to the heated beam spot where it is slightly cooler.  This is the genesis (I suspect) of the "carbon ring" we often observe after acquisition.

Of course these effects are extremely dynamic and depend entirely on the cleanliness of the vacuum system and sample, the beam size and current (electron dose areal density), and the thermal conductivity of the specimen itself.