Dealing With Magnetic Specimens

[Probeman]

Hi Keith,
The latest "magnetic nightmare" sample in question is the diffusion study sample. It's the one that has a piece of pure Fe in a horseshoe shape(!), wrapped around a piece of V in the first case and another horseshoe shaped piece of Fe wrapped around a piece of Nb in the second case.  I posted a picture of it above.  I have no idea how they get magnetized. I know they run these diffusion experiments in a furnace for hundreds of hours at very high temperatures.

Both samples show highly variable beam deflections depending on the exact sample area on the order of 5 to 50 microns or so.  I worry that dipping the mount in LN2 could crack the epoxy. In any case I'd have to ask the researcher first.

I'll look around for a stronger demagnetizer when I attempt these samples again.  Thanks for all of your and Dave's suggestions.
john

[JonF]

At risk of sounding a bit daft, but have you tried just dropping the sample on to the desk a few times?

[Probeman]

At risk of sounding a bit daft, but have you tried just dropping the sample on to the desk a few times?

Hi Jon,
You know, that's actually not a bad idea at all.  Certainly worth trying!   

I'll let you know if it works as soon as I get a chance to try it.  Hopefully next week or the week after.
john

[Probeman]

I apologize for resurrecting this topic but I am really at my wits end and about to give up trying to quantify WDS elements on these highly magnetic samples.

Here's what I've learned over the last year:

1. One can completely demagnetize these magnetic samples and they will re-magnetize in the instrument.

I checked this by demagnetizing the samples in a de-magnetizer setup as shown here:

https://probesoftware.com/smf/index.php?topic=354.msg6656#msg6656

Then once the newly demagnetized sample is in the instrument, I first navigate over to the steel sample (H13 and 4140 in the most recent effort), and place the beam over epoxy but just adjacent to the steel sample and when I unblank the beam it burns a spot right where it is supposed to.

2. The sample remagnetizes over time. But here's the interesting part: the re-magnetization appears to occur *not* from the sample sitting in the instrument (though there may be some effect from this), but mostly from the time when the beam is impinging on the steel sample. 

My febrile hypothesis at the moment is that at these currents (~50nA for getting good quant for light and trace elements), it is the current *flow* that is that is re-magnetizing the sample.

I have two line of evidence for this.

First when I am imaging the sample, the image is slowly drifting as soon as I am imaging the sample. The longer the beam is imaging the sample, the greater the drift. But if I blank the beam and wait a minute, then unblank the beam, the image drift starts again as before, but it starts *exactly* where it previously left off when the beam was blanked!

Second, I noticed that on these samples I had applied conductive tape on the top of the sample, on the edge away from the edge I am analyzing.  And the beam is being deflected only in the Y direction and towards the copper tape. My completely crazy idea that the current flow of electrons towards the grounded side of the sample is inducing a magnetic field that deflects the beam in the direction of the current flow.

So now the even more crazy idea is that instead of grounding the sample on the opposite side, perhaps I could ground *both* adjacent sides of the sample and the current flow might be balanced enough that they might cancel each other out? 

Well after an hour of acquisition I checked the beam, again by unblanking it on some nearby epoxy, and indeed the deflection was now much smaller and only in the direction of one of the adjacent sides.

Crazy I know.

But here's my question: can anything else be done? I mean besides running the sample above it's Curie temperature!

The beam deflection we are seeing in these 5 mm sized samples is indeed smaller than the 5 centimeter sized sample we were previously struggling with, so would cutting the sample down even further help?  What about plasma fibbing them out?  We only need a 600 um traverse distance from the edge.  Has anyone tried plasma fibbing out small samples to reduce induced magnetism?

What do people in the steel industry do?  Where is Les Moore when you need him?

[Les Moore]

I've never looked at magnets. Low core loss and magnetisable steels yes but not something  magnetised already.

How about a string of microhardness indents across the sample.  This will give you a static datum. you can then park the beam beside them.   
If the flow is towards the conductor, could you out the conductor on the surface - a thin film of some highly conductive metal (Al perhaps)
The latter will absorb your light trace elements a bit but it might be modellable in PENEPMA.

Perhaps you could use the new fandangled sputter coaters that deposit only a couple of nm of tantalum (is that the element).

[Probeman]

Hi Les,
These are *not* magnets. But I do often de-magnetize them if I see big (~20 um or more) beam deflection effects when they are loaded the first time.

The particular samples we're running now are H13 and 4140 steels (we've seen the same thing on many other compositions that contain Fe). The thing is, they seem to *become* magnetic in the instrument (after de-magnetization!), but apparently only when exposed to a relatively high beam current (~30 nA or more) over time.

Haven't you (or anyone else!) seen this effect?  My lab manager Julie confirms that she sees this on many cm sized Fe alloy samples.  On these particular samples the customer wants several hundred 1 um points analyzed with C, N and O, so using reference points is OK to locate the beam position, but then the Bragg defocus brings the totals down to 90% or less. And to make things even worse the beam deflection is quite variable and depends on the actual stage position!  Not to mention the change in magnetism as the sample continues to magnetize.

It's an analytical nightmare!

We have a Cameca SX100, but I find it hard to believe it's only our instrument that has this issue!
john

[Probeman]

To continue Probeman's sad (magnetic) sample saga, we had Julie Chouinard cut down the 4140 and H13 tool steel samples in an attempt to reduce the amount of magnetic material. I noted that when mounting the samples, they were strongly magnetic and so ran them through our demagnetizer unit as shown here:

https://probesoftware.com/smf/index.php?topic=354.msg6656#msg6656

Once the samples were in the instrument again, I aligned the light optics to the beam, then I went near the edge of the steel samples and unblanked the beam over some nearby epoxy, and the beam burn showed the beam to be aligned perfectly on the beam centered position.

I then re-peaked all the spectrometers and acquired the standards. Then (~2 hours later) I went back to the steel samples, and alas, the beam was now deflected anywhere from 10 to 30 um depending on the distance from the edge of the samples.  In both cases the deflection was towards the sample edges and varied with the distance to the edges (beam was pulled more towards the sample, as I approached the edge).

It really does appear that just sitting in the instrument for a few hours re-magnetizes the samples and causes significant beam deflection.  This is a Cameca Sx100, but has anyone seen this effect in their instruments?

[AndrewLocock]

Hello,
In the course of analysis of olivine in pallasite (Fe-Ni metal and olivine) meteorites, we found a serious problem with one of the samples.
It was very strongly magnetic (see attached PDF that shows a pin sticking to the vertical polished surface).
As Mg and Si were measured on separate spectrometers (with TAP crystals), the results were terrible (because of beam deflection on the order of dozens of microns).
Initial results gave totals of 88 wt%, and ratios of (Mg+Fe):Si of up to 2.6!

The use of a demagnetizer (see attached PDF) that consists of an AC electric motor coil controlled by a Variac autotransformer solved the magnetism problem.
The use of the Variac is critical; we used a 5 - 10 - 5 second cycle (turn up power, hold, turn down power).
The totals are now around 100 wt% and the cation:Si ratio is 2.

Note that the AC motor coil should not be run for more than 30 seconds - it heats up!

All the best, Andrew

[Probeman]

Hello,
In the course of analysis of olivine in pallasite (Fe-Ni metal and olivine) meteorites, we found a serious problem with one of the samples.
It was very strongly magnetic (see attached PDF that shows a pin sticking to the vertical polished surface).
As Mg and Si were measured on separate spectrometers (with TAP crystals), the results were terrible (because of beam deflection on the order of dozens of microns).
Initial results gave totals of 88 wt%, and ratios of (Mg+Fe):Si of up to 2.6!

The use of a demagnetizer (see attached PDF) that consists of an AC electric motor coil controlled by a Variac autotransformer solved the magnetism problem.
The use of the Variac is critical; we used a 5 - 10 - 5 second cycle (turn up power, hold, turn down power).
The totals are now around 100 wt% and the cation:Si ratio is 2.

Note that the AC motor coil should not be run for more than 30 seconds - it heats up!

All the best, Andrew

Cool.

I would be very interested in hearing if, after sitting in the instrument for a few hours, you see any re-magnetization effects with this sample as described in the previous posts?

[AndrewLocock]

Hello,
No change after 7 hours of analysis of olivine included in the Fe-Ni metal.
That is, magnetism did not reappear during/after analysis of the olivine (which now gives good results).
However, we were not irradiating the metal directly....

Thanks, Andrew