Annealed cast iron as magnet pole pieces (for low-field NMR)

[ChristofferB]

Hi!

I'm still working on a homogenous-ish magnet to do NMR and EPR spectroscopic experiments with, and have settled on some 10cm diameter ferrite puck magnets as my magnets (found a paper doing just that, it's nice having a reference to compare your design to).

I'm looking into how to make the puck-shaped pole pieces for it, to homogenize the field, and of course, discs of soft iron would be best, but going to specialized magnet alloy companies is slightly (=completely extremely) over budget for this project. I've found an ebay seller that sells 10 cm cast iron rod in slices, and was wondering if these would be a suitable(ish) substitute if heated to anneal them.

What do you think? Would I be better off without pole pieces entirely, or what?


I've added a schematic of the planned magnet layout, not that it's really needed but posts with visuals are just nicer aren't they?

Thanks!

--Chris

Thanks in advance!

[RandallMcRee]

What are the dimensions? (10cm diameter? what about thickness?)

McMaster-Carr sells iron discs that might work.

I have lots of magnets left over from years of experiments with planar-magnetic speakers if you need magnets...just ask. No iron, though. I used some kind of steel (forget exactly what now, but cheap) which was listed in the femm database.

Do you know about FEMM?

[ChristofferB]

I didn't know about FEMM but I do now! Thanks a bunch!

The magnets I've settled on are 100 mm in diameter, 15 mm thick, and I want my pole pieces to be the same.

I'm in Europe, so McMaster-Carr is pretty much out of the question, unfortunately.

 I was looking at this auction: eBay auction: #https://www.ebay.com/itm/100mm-approx-4-dia-meehanite-cast-iron-round-bar-1-12-lengths/263704103005

1 inch pole pieces would be fine, too, and I'll find a local machine shop and ask if they could face them on the lathe to get flat parallel surfaces.

I'm not sure annealed cast iron has high enough permeability to work in this application, though.

thanks!

[T3sl4co1l]

Annealed mild steel should be just as good, and may be easier to get?

TLDR: I forget the exact magnetic properties (if such a term isn't a contradiction to begin with ) of common iron products... but I also suspect that there's more variation between samples, and their condition, than might be expected.  I'd prefer the simple solution (mild over cast), which is probably just as easy to obtain?


Soft gray iron should be more-or-less pure ferrite* in the metallic phase, with graphite flakes throughout (which of course don't have any magnetic effect).  Avoid anything of higher strength, including ductile iron (the softest grades of which are probably fine still, but the stronger ones not so much).

*Metallurgical "ferrite", alpha iron.  Soft, mechanically and magnetically.  Stronger alloys, treatments even, will contain more pearlite, which is magnetically hard.

Beware that annealing cast iron is... kind of complicated.  It's full of carbon, so can accidentally be hardened very easily.

Mild steel is very easy to anneal, just heat to orange-hot, then cool slowly.  The slowness doesn't even really matter (as long as the carbon and alloy contents are low) -- it's hard to cause much change even water-quenching it.

Mind that cheap steels are not created equal.  "A36", "rebar" and so on, are -- like many semiconductors, actually -- defined not by composition or structure, but by overall properties.  As long as they meet the requirements on strength and flexibility and whatever, they're fit for that purpose; but that says nothing about their magnetic or metallurgical behavior.  Preferably, stick with known 1020 mild steel (or less; the "20" means 0.20% carbon).  Or 12L14 is probably fine too, and better to machine.

Silicon steel and nickel steel (possibly austenitic, I forget?) of course are better, but if you don't have machining or purchasing capability for these, well then that's that.

Tim

[Circlotron]

In your diagram shouldn't you have another piece of ferrous material going from one outer magnet face to the other outer magnet face?

[ChristofferB]

Tim, thanks! Noted! I'll look for 1020 round mild steel slices or similar. I got it wrong, I had the idea that cast iron was very carbon-deficient, but I must've misremembered.

Should be obtainable, i hope.

Circlotron:

I think it's called a flux return, and I absolutely could, but for some reason, most of the tabletop NMR magnets I've seen in recent litterature have omitted them.

Maybe they make it harder to shim, I don't know.

Thanks!

[Circlotron]

Having a flux return will make a BIG improvement to the magnetic field strength where your object under test is situated. Or allow you to use smaller magnets if it was already sufficient.

[duak]

Here's a link to a graduate student's thesis for an NMR instrument: http://web.mit.edu/~20.309/www/Zimmerman2010.pdf   She chose 1018 mild steel for the pole pieces and Yoke or backing iron (flux return).  In my part of North America, 1018 is more commonly available than 1020.  Anyway, look up AISI 1018 and 1020 steel, there isn't a big difference.

I don't know if she used hot rolled or cold rolled steel or if it will make a difference.  I do know that cold rolled steel has internal stresses that indicate it has not been annealed after rolling and this may affect magnetic uniformity.  ie., it may not be isotropic.

[T3sl4co1l]

HRS (hot rolled) should be preferred as it's more or less in annealed condition; CRS's internal stresses will trap magnetic fields (again, in most steels, metallurgical "hard" is also magnetically "hard", as far as I know).

Mill surface won't be very flat though.  You'll presumably need to machine something, either way, so that shouldn't be a problem, but it hasn't been mentioned if you have that capability?

Yeah, flux return is great, look at the classic pictures of, well, circlotrons for example, as our fellow forum member so kindly is -- often a boxy figure-eight shape, steel looping around the outside, with an electromagnet on the middle limb (you'd stack your magnet(s) here), and the pole pieces and gap on it as well.

Welding shouldn't be too much of a problem, by the way; even less so if you have the capability to anneal the assembly once finished.  That'll also relieve internal stress, which will definitely be present after welding.

I don't know how much this kind of on-the-fly redesign matters to a magnet; I would think, if you have the means to probe its homogeneity, and the ability to shim accordingly, you should be able to do so for any reasonably similar design.  The significance of, for example, the flux return, should be of more than doubling the flux density in the gap, which is definitely a worthwhile gain!

Magnets I think, tend to be discouraged because of their drift and tempco?  I know the original experiments (Purcell, et al.) were done with electromagnets ran from lead acid batteries (and, no lie, thermally regulated via grad-student-controlled windows), but we probably don't have to go quite to such lengths nowadays.  A nice strong water reference and some kind of tracking PLL may be good enough instead?

In any case, will be interesting to see if you get any detail out of this; not sure if chemical structure (frequency shifts of ~ppm) will be resolvable at these field strengths, or with reasonable acquisition times (or at all), but if so, that'll be freaking awesome.

Tim

[pwlps]

You might also have a look at those papers (unless you already found them):

https://doi.org/10.1016/j.jmr.2010.04.003
https://doi.org/10.1063/1.2713438

They discuss the geometry of permanent magnet-based shimming systems.

[ChristofferB]

Great! 1018/1020 mild steel, preferably hot rolled it is! I'll see what's available.

I have no machining capabilities myself but facing two steel rounds on the lathe is a pretty small operation, I'm pretty sure I can find a local machine shop or engineering college shop willing to do it for cheap

I think I will stay off of the flux return for now, I want to assess how homogenous a field I can get before I go too overboard.

Tim,  I think permanent magnets was disfavored over electromagnets because early nmr used a gradient magnetic field, which is pretty easy to do by just modulating the AC to the magnet with a slow ramp.

Temp likely also was a factor.

I agree it's a bit iffy if I get any sort of resolution out of this thing. Either way, project plan is:

1. Produce a homogenous, shimmable magnet putting the hydrogen larmor freq between 2 and 20 mhz

2. Produce a probe accommodating excitation (saddle) and pickup coils, 5mm nmr tube and maybe a shim coil.

3. Find H2O peak by rf sweep

4. Rf pulse gen for excitation

5. Lock-in amplifier

6. Adc and dsp to handle the entire mess

Thanks for the interest!

--chris

[Ian.M]

A lot of late 19th century experimental electromagnetic apparatus used bundled soft iron wire for cores.   If you don't have access to a machine shop, its still a viable method of building the yoke.   The ends of the bundle can be cropped one wire at a time to the desired plane, bonded with epoxy then finished by grinding or filing to flatness, limited only by the amount of time you put in to the work, bluing the face and checking it against a reference surface.

[ChunkyPastaSauce]

A lot of late 19th century experimental electromagnetic apparatus used bundled soft iron wire for cores.   If you don't have access to a machine shop, its still a viable method of building the yoke.   The ends of the bundle can be cropped one wire at a time to the desired plane, bonded with epoxy then finished by grinding or filing to flatness, limited only by the amount of time you put in to the work, bluing the face and checking it against a reference surface.

0.9T gap with wire yoke

1.2T gap box frame

[pwlps]

2. Produce a probe accommodating excitation (saddle) and pickup coils, 5mm nmr tube and maybe a shim coil.

I think nobody uses separate coils for excitation and measurement for pulsed NMR, except in some double-resonance probes. A single coil setup will save a lot of space so that the probe can be thinner and magnet poles closer,  the field homogeneity improving accordingly.   Using the same coil it is also easier to get a more homogeneous H1.

[RoGeorge]

3. Find H2O peak by rf sweep

If you are interested for the water's NMR, I've seen many DYI projects of NMR done in the Earth's magnetic field.

Earth's field is pretty low (about 50uT) but the advantage is that it has a very, very good uniformity, which is crucial for postprocessing.  Another big advantage of that low 50uT value is that the corresponding precession (Larmor) frequency of the H is in the audio spectrum, so very easy to amplify, then sample the echo with a soundcard.

There are open source programs to process the captured sound, and prepare a 3D NMR image from the spin echoes.

[ChristofferB]

Earth field NMR is a very cool but completely different area to work in, I think something looking a bit more like 60's nmr is my prefered way to go.

I've modeled my proposed magnet in FEMM (thanks again for the suggestion) and the field strenght in the gap is almost unchanged between soft iron and 1020 steel.

I'll go with the steel for now. attached is the output. Now there's something to put on your wall!