Dealing With Magnetic Specimens

[Probeman]

Hi all,
We are running some Fe-Ni-Cr alloys and some of the samples can cause the beam to deflect from the intended position by 50 um or more due to the magnetic field of the sample. Besides not analyzing where we want to, the Bragg defocus effects are killing our quant.

I tried the hysteresis button but that didn't help, so we are looking for a demagnetizer of some type, but wonder if anyone has a good solution for demagnetizing magnetic samples prior to loading them... or will the specimen magnetic field reappear once inside the instrument or the beam is on the sample?

[Karsten Goemann]

How thick are the samples? We used to encounter problems with massive magnetite, but when we prepare them as petrological polished thin sections (30 microns thick) on glass slides, the problem usually goes away.

[Probeman]

How thick are the samples? We used to encounter problems with massive magnetite, but when we prepare them as petrological polished thin sections (30 microns thick) on glass slides, the problem usually goes away.

Good question: unfortunately these are massive samples (1.25" to 1.5" diameter and at least a cm thick), and they aren't ours so they cannot be modified.

[Gseward]

I have similar issues with Fe and Co alloys. It took me along time to convince the researchers that this was the issue. Thankfully I have a BIG demag unit that is very effective at randomizing the magnetic domains within these samples. I don't remember the numbers exactly, but as I recall a particularly 'bad' sample would have a field of 100's of gauss in some directions; after 3 cycles of demag in orthogonal directions, I might get it down to 0.1 . Obviously the direction and magnitude of deflection can be very anisotropic, both macroscopically and, I suspect in my samples, on a phase by phase basis.

The equipment is a relict of our old geomag lab and was built by Bob Dunn, a former geomag wizard in our department. So, I can demag them for you, but you'll have to send them. I think I could do 1.5" but I'll check tomorrow.

Perhaps an nth order effect, but out of interest, I wonder if the angle of deflection could be severe enough to also affect the depth of e- interaction and the absorption path length?

[Gseward]

John,

Did you have any success? Where you able to measure the field strength before/after demag?

Gareth

[John Donovan]

Did you have any success? Where you able to measure the field strength before/after demag?

Did not find a way to measure the field, but I think we're seeing a magnetic sample beam deflection (in scan mode and in spot mode) which, after degaussing and inserting in the instrument, starts at zero deflection (relative to the optical image which does not drift at all) and increases to about 50 um over 10 minutes. It is definitely not charging. Seems to settle eventually. See images below.

Running the sample using *only* EDS elements now... to avoid Bragg defocusing.

Is it possible a magnetically susceptible specimen can be re-magnetized by the electron beam? As evidence I offer the observation that the beam deflection only drifts when the faraday cup is out. If we blank the beam and wait a few minutes and then unblank, the beam deflection as observed on the electron image, has not shifted relative to how it last looked, but then immediately and slowly starts shifting again, as long as we leave the beam on the sample.
john

[Les Moore]

A few thoughts...
1. Could it be a heating effect?
     Is the effect the same on all corners of the sample?
2. Could it be a charging effect?
     I recall that someone saying 40% of the 2nd'y electron yield comes from the polepiece via BSE. 
     So, these BSE could be BSE'd back onto the sample somewhere else to possibly charge.
     I would imagine that this would be cylic though.
3. Lastly, the sample should not be magnetised by the beam unless the beam creates a magnetic field.
    M=BILsin(theta).  'I' you have, the rest is anyones guess.
4. if the beam is magnetising the sample, it should be 'I' dependant. 
    But then again, so should heating and charging.

Now really pressing hard...
5. Local heating may allow the magnetic domain boundaries to migrate.  This may cause localised magnetic variations but this is just a guess. 

There's a lot of work being done on domain boundaries in TEM using low loss EELS and other techniques that are beyond my ken.

 

[Probeman]

A few thoughts...
1. Could it be a heating effect?
     Is the effect the same on all corners of the sample?

Hi Les,
Thanks for chiming in.

I don't see how it could be a heating effect... that would *reduce* the magnetism (if we could get beyond the Curie temperature, which we can't in a thermally conductive material).

The effect (degree of beam deflection) is different in every part of the sample, but it is a complex sample of cm sized blocks of various pure metals including Ni, Fe and Cr...), so that would be expected?

2. Could it be a charging effect?
     I recall that someone saying 40% of the 2nd'y electron yield comes from the polepiece via BSE. 
     So, these BSE could be BSE'd back onto the sample somewhere else to possibly charge.
     I would imagine that this would be cylic though.

No, it is not. The degree and direction of beam deflection goes with the sample rotation- yes, we tested that on Monday!

Also the SE image shows zero charging.

3. Lastly, the sample should not be magnetised by the beam unless the beam creates a magnetic field.
    M=BILsin(theta).  'I' you have, the rest is anyones guess.

Yes, that is the question.

4. If the beam is magnetising the sample, it should be 'I' dependant. 
    But then again, so should heating and charging.

We have not tried different beam currents, but it is a good suggestion... we'd have to demag the sample each test and carefully time the degree of beam deflection.  It is very strange...

Now really pressing hard...
5. Local heating may allow the magnetic domain boundaries to migrate.  This may cause localised magnetic variations but this is just a guess. 

Again I don't think so. See above, #1

[Probeman]

I'm posting this response (with his permission) from one of our physics faculty who had some thoughts on this issue:

At 11:15 PM 11/20/2014, Benjamin McMorran wrote:
Hi John,

Interesting!. Yes, it sounds like Lorentz deflection of the beam to me. The question is why is there is a dose dependence? If it's only Fe-Ni-Cr in the sample, then I'd be surprised if it's heating up and switching domains significantly enough to deflect the beam by that amount. The azimuthal magnetic field from the beam itself is pretty small and drops off linearly with distance from the beam, so I don't think this is changing the magnetization (though that would be really cool!).

Maybe it's the fringing fields from the objective lens? A lot of SEMs can have a pretty high magnetic field from the OL, even if it's not purposefully using an immersion lens design. This might appear to depend on the scan range just because of the domain structure - different initial trajectories of the beam would get deflected different amounts.

For the demag, are you using a tabletop one, or are you talking about the instrument's demag? A tabletop one can be made with just a coil (I have one in the lab) and a variac.

Again, interesting problem!
Ben

Dr. Benjamin McMorran
Assistant Professor of Physics
Materials Science Institute and Oregon Center for Optics
Department of Physics
1274 University of Oregon
Eugene, OR 97403-1274
office phone:‚ 541-346-8624

My response was:

Hi Ben,
Thanks for the interesting and informative response.

The dose dependence could also be described as a time dependence as well, as it appears to increase the beam deflection with a constant beam current over time, and not while the beam is blanked.

We tried a hysteresis cycle several times on the objective lens, and to a first order it made no difference to the beam deflection degree or rate of change.

The demag is a table top device that Julie bought recently.

If you'd like to discuss this further just drop by the lab Mon through Wed when I'm here.  It's causing us a lot of problems- basically we had to give up on WDS for these samples!
john

[Les Moore]

Clutching at straws...

Could you cover all but a small window on the sample with Al tape.
It might shield the interaction volume from bulk sample effects.

Les

[John Donovan]

Could you cover all but a small window on the sample with Al tape.
It might shield the interaction volume from bulk sample effects.

Hmmm... an interesting idea I must say.

Although using PFE for automated EDS acquisition works fine because there is no Bragg defocusing and and long as we plan the traverese coordinates using the BSE image (which already contains the beam deflection effects) it all seems to work pretty well.

So I know that Al shielding can be used for EMI because that is what we did for our power panels in the equipment gallery behind our instruments. We put 1/4" Al between the wall and the back of the power panel and it produces at least a 20 db drop in EMI into the instrument room as seen here:



However, I wonder if Al would work for magnetic fields?  I know people usually employ "mu metal" which is a soft iron for magnetic shielding when I worked at LBL.

But I guess it's worth a try and heck, at the very least it sounds a like a great idea for a student project! Just get some tiny magnets and mount them one each in epoxy pucks and try different orientations and so forth.

[Les Moore]

As I said, clutching at straws.

Could you cover all but a small window on the sample with Al tape.
It might shield the interaction volume from bulk sample effects.

But I guess it's worth a try and heck, at the very least it sounds a like a great idea for a student project! Just get some tiny magnets and mount them one each in epoxy pucks and try different orientations and so forth.

How cruel. 

Especially if you ask for high mag pictures.  I once tried to get a SEM picture of the fracture surface on a hard magnet; not possible.

Les

[Gian Colombo]

I recently encountered this issue as well, also on metallic specimens primarily containing iron-chromium-nickel.  The job request was to perform multiple point analyses in a square grid pattern that effectively covered the entire specimen face.  The analytical totals shifted wildly (~99wt% to ~55wt%) and displayed a repeating periodic pattern as the grid passed through regions of magnetic domains with different orientations that would shift the beam different amounts and in different directions which took different spectrometers in and out of focus depending on sample location.

Here's a summary of the few things that we tried to address the problem:

-De-magnetizing helped reduce the overall magnitude of the beam displacement, but prior regions that showed severe deflection still pushed the beam far enough off axis that analytical totals were not recovered.

-Working at a higher accelerating voltage (25-30kV) had a centering effect on the beam.  It was still not quite enough to eliminate the deflection in all areas, but analytical totals were recovered over a significant area of the specimen.

-The best results we achieved were accomplished by "re-aligning" the beam with a user-defined beam offset (much like one would use to analyze different features in a SEM field of view without moving the stage) to counteract the deflection induced by the specimen.  The optical image was centered on an easily recognizable feature in the SE image which was used as a target to define the beam offset such that the beam would now be hitting the correct spot on the specimen.  This worked beautifully over small areas where the magnetic effects were relatively consistent, but it became very tedious as I worked across the specimen because I had to change the offset multiple times to account for the changes in the magnetic beam deflection.

[Probeman]

Hi Gian,
That is very interesting (and tedious sounding!).

On our magnetic samples we tried higher keV which also didn't help enough, but because the customer only wanted specific traverses in specific locations on the sample, we ended up using PFE in EDS acquisition mode (without any WDS elements). (We have a Thermo NSS system on our SX100 interfaced to PFE for full spectra acquisition and quantification).

So, we would get to the correct sample locations using BSE electron imaging (which would always be off by some amount from the optical image), digitize the stage positions from the BSE image using the Digitize Image feature, and then just run them with the "Acquire EDS spectra for each point option. That worked pretty well as the EDS system doesn't have Bragg defocus issues of course.
john

[Probeman]

We had a client in our lab recently wanting to look at very large (4.5 cm diameter) mild steel samples. In fact we had to make a special sample holder for our Sx100 to accommodate them.

Unfortunately the sample was highly magnetized, so much that when the optics and electron image were the same FOV, the electron image was so deflected that the central feature (as viewed in the light optics), was barely visible on the edge of the electron image.

We tried a "de-gausser" device (Kandrick & Davis, model 19-217, see attached image below) which we had purchased last time for some Ohio State Fe samples, but it only slightly decreased the magnetic effect.

We wanted to run WDS on these samples for the oxygen and carbon contents, but had to utilize EDS instead because of the magnetic deflection from the sample.  Unfortunately the EDS sensitivity for oxygen and carbon in steel is pretty awful due to the large Fe L emission peaks, though I'm interested in discussing that issue in a separate topic.

In the meantime does anyone who works with steel samples in EPMA have a suggestion for a more effective de-magnetizer that we can buy/build for such large samples?
john