Construction of a permanent magnet NMR spectroscope

[ChristofferB]

Can't really be done. It's the issue of signals physically overlapping, and thus becoming indecipherable. That's why professional instruments go UP in operating frequency. it gives more room between signals.

[T3sl4co1l]

Note that excitation frequency isn't very critical.

Consider the electrical circuit.  The nuclear spins manifest as resonances coupled to the coil.

It behaves much like a quartz crystal very weakly coupled to a coil.  You can cancel out the coil's parameters (reactance and loss) using an impedance bridge (first, a capacitor to cancel reactance, then a complementary arm of the bridge with the same L and C, but a variable resistor to adjust loss to match).  What's leftover is higher order effects, like skin effect, cable and chamber resonances, etc.  Most of which vary slowly with frequency, so are easily corrected for, over the narrow frequency range needed here.

The tiny peaks in that narrow ~Hz wide window are your signal: a measurement of atomic resonances.

To resolve the peaks, as a practical instrument, you need quite good resolution, at least 0.1ppm I would guess.

If you just want to see a peak, a somewhat more crude magnet and receiver can be used.  (This is definitely a development milestone!)

Conversely, the stimulus doesn't need to be very special.  The burst can be quite long in real time, as long as it's a suitable fraction of the FID time constant.  The Fourier transform of a rectangular windowed tone burst is a sinc(F) spectrum, centered at the tone's frequency.  More pulse duration means narrower bandwidth means more power at the frequency it's needed; shorter duration means more bandwidth and less precision needed in lining up stimulus and actual resonance.

Tim

[zappedy]

I was more thinking that you would use a couple of calibration / control samples and let an AI learn how to drive the various parameters to get a better signal than would have been available in the 60s, from a 60s type instrument. Replacing the poor grad student with a computer positioning the window 100000 times per second in just the right pattern. Or possibly letting the AI learn how to interpret data based on the temperature fluctuations, etc. Again based on known control samples.

This would probably be difficult to accomplish and I'm sure somebody who is experienced in AI stuff is laughing at me right now, it's just a half baked idea of again using modern computer technology to "boost" old tech while keeping it within the reach of DIY.

Anyway, I don't know anything about spectrometers except what I learned in high school and the odd article in science magazines so I'll quietly leave this thread...

[knotlogic]

I've been taking some courses on NMR spectroscopy, and I wonder- Is it possible to build an NMR machine yourself?

Possible, but a lot of work I would have thought.  I think a few years back there was a news article making the round about a guy who build a low-field imager at home and was just about able to image his torso.  I didn't see the images, and I don't know if you could actually make anything out, but still...

The Aberdeen Mark-I was the clinically used scanner (First clinical scanner?)  That was a water cooled electromagnetic system.  I also knew some guys doing low field work at university.  IIRC, one of them build a permanent magnet imager for his PhD.  You do want better field homogeneity for spectroscopy than for imaging though.

I was at a place that built a specialised 0.5T imaging unit.  Two permanent magnets and an iron frame to channel the field.   Samples had to be under an inch in size though.

[knotlogic]

I assume you'd always be able to ramp up the freq. and thus resolution with more advanced parts, reusing magnet and probe.

Since the magnetic field determines the NMR frequency, with a permanent magnet you have to stick with more or less the same frequency

Of course, this assumes that you stick with proton NMR. In principle, you can change to a different nucleus and hence work at a different frequency with the same magnet. I don't know if this is actually practical with our simple equipment, as it relies in part on the high concentration of protons in H2O etc to get strong enough signals.

But you can get combination probes, with outputs for multiple nuclei. I've seen a 1H, 2H, and 13C combi probe once, i think.
I actually assumed 13C would be more efficient, since they're usually taken at lower frequencies.

With the proton concentration in H2O, you could use another solvent to get a nice 13C peak. CDCl3 is sometimes used, TMS would be good too.

If you're imaging tissue (human or animal) there's a lot more hydrogen than carbon available, so a lot more signal there.  I suppose the same applies to organic chemistry and spectroscopy.  Also, the relative abundance of carbon-13 is 1.1%, which makes things a whole lot worse.  A chemist told me that it wasn't uncommon to do a ton of signal averaging with some samples just to be able to make out the spectroscopic peaks.

[ChristofferB]

I thought the RF burst had to be short, in order to uncertainty out over the entire spectrum? In earlier models, the magnetic field was varied. I was thinking maybe simplifying things with a tracking generator.

[Kleinstein]

For a normal NMR the exciting pulses don't need to be that short. The length of the pulse determines the bandwidth that is excited. Usually the needed bandwidth is still relatively small, e.g. way less than 1% - so pulses can be 100 periods at least.  It does not make sense to make the pulses longer than the relaxation time though, but up to that limit the signal output gets higher, though it might mean one would need different excitation signals to cover the whole range, e.g. to find the peaks at first.

A more difficult part is that the excitation might need quite some power. So I would guess the main difficulty for excitation is the power amplifier and matching to the probe. Signal generation might be as simple as a digital pattern. Or at lower frequencies a DDS chip controlled by an µC. Today even the Pulse pattern for pulse echo measurements is not that difficult to produce.

For a permanent magnet version I would keep the field constant and adjust the frequency as needed. Adding an electromagnet generates heat and thus caused thermal drift. So if at all have an large external coil to compensate for an changing external (earth) field - no need to cancel out, just keep it constant. It's still a question if the sensor causes more noise than the natural changes in field. Changing temperatures can also influence flied homogeneity.

[ChristofferB]

Keeping the field constant is a good idea. I think my first experimental setup would be a permanent magnet, maybe with a small tuning coil, and then 3D Helmholtz coils (hula hoop ring based?) around it for homogeneity. Does Helmholtz coils need anything special to drive them? Or do you just throw some DC at them all? I've never actually handled them in practice.

[Wolfram]

There was an article in Scientific American from the '50s or '60s on building one using a magnetron magnet.  It was later republished in a book.  If you use the correct RF frequency, it should be possible using the earth's magnetic field.  The signal would be noisy though.

Edit:
Book:
The Scientific American book of projects for the amateur scientist.
Author:    Clair L Stong
Publisher:    New York, Simon and Schuster, 1960.

Article:
http://www.scientificamerican.com/article/the-amateur-scientist-1959-04/

The whole book is available for free on archive.org: https://archive.org/details/TheAmateurScientist

[KE5FX]

There was an article in Scientific American from the '50s or '60s on building one using a magnetron magnet.  It was later republished in a book.  If you use the correct RF frequency, it should be possible using the earth's magnetic field.  The signal would be noisy though.

Edit:
Book:
The Scientific American book of projects for the amateur scientist.
Author:    Clair L Stong
Publisher:    New York, Simon and Schuster, 1960.

Article:
http://www.scientificamerican.com/article/the-amateur-scientist-1959-04/

The whole book is available for free on archive.org: https://archive.org/details/TheAmateurScientist

Here's an actual decent scan.  (45 MB)  The  NMR article begins on page 355 of the .PDF.

[knotlogic]

I'm not sure about spectroscopy, but in imaging the length of the pulse affects the width of the slice (in the presence of a slice select gradient).

Keeping the field constant is a good idea. I think my first experimental setup would be a permanent magnet, maybe with a small tuning coil, and then 3D Helmholtz coils (hula hoop ring based?) around it for homogeneity. Does Helmholtz coils need anything special to drive them? Or do you just throw some DC at them all? I've never actually handled them in practice.

Say a pair of permanent magnets?  Sample placed in between them.  Some sort of thermal stabilisation for the magnets, otherwise the field drifts.  Field inhomogeneities are handled with shim coils, they're driven with DC I think.  It's not an area I'm familiar with either, but there are a few - XY, XZ, YZ or something like that, and more for higher orders.

[ChristofferB]

KE5FX: thanks for the scan! That's not bad reading, and not only for the NMR stuff!

I'm not sure about spectroscopy, but in imaging the length of the pulse affects the width of the slice (in the presence of a slice select gradient).

Keeping the field constant is a good idea. I think my first experimental setup would be a permanent magnet, maybe with a small tuning coil, and then 3D Helmholtz coils (hula hoop ring based?) around it for homogeneity. Does Helmholtz coils need anything special to drive them? Or do you just throw some DC at them all? I've never actually handled them in practice.

Say a pair of permanent magnets?  Sample placed in between them.  Some sort of thermal stabilisation for the magnets, otherwise the field drifts.  Field inhomogeneities are handled with shim coils, they're driven with DC I think.  It's not an area I'm familiar with either, but there are a few - XY, XZ, YZ or something like that, and more for higher orders.

That was something similar to what I was thinking. The largest problem remains in finding a good magnet. Some older homebrew's uses magnetron magnets, but I think microwave oven ones are too puny..

[KE5FX]

KE5FX: thanks for the scan! That's not bad reading, and not only for the NMR stuff!

Yeah, it's an awesome book.  Lots of politically-incorrect experiments for the doughty amateur.

That was something similar to what I was thinking. The largest problem remains in finding a good magnet. Some older homebrew's uses magnetron magnets, but I think microwave oven ones are too puny..

I wouldn't mess with those old alnico jobs unless there's some other reason to (like field homogeneity?)  Modern Nd magnets are 10x stronger than the WWII-era magnetron magnet in the Amateur Scientist article.

[ChristofferB]

Nah, you're very right, you can get a lot juicier magnets today, but if one were to use a stack of common modern neodymium's, i'd just think they would be more poorly made, and be harder to get to behave homogeneously. Maybe if you mounted them on a large block of soft iron...

[Akra]

If you use the magnets with an iron yoke (or other ferromagnetic material), you probably can get a much better and more homogeneous field.
Don't know if this would get homogeneous and stable enough...
But even a lab NMR needs to lock the frequency to an known standard (solvent signal) for every new sample.

If i remember correctly, than there are handheld NMR devices. They should be low resolution, so you don't see coupling,
but you can determine what elements are in your sample.

The high resolution NMR used in the lab gets really fast really complicated...

PS: Would really love to see someone do something this complicated (but really awesome) at home.

[RandallMcRee]

Neat project. I've built ribbon speakers using magnets which could be used for this project.

To take this to a realistic level you will need to model the magnetic field. I suggest;

http://www.femm.info/wiki/HomePage Its, good, free software.

If you can come up with a design using some number of Neodymium N42 rectangular magnets of size 0.5 inch x 0.5 x 2 inches, I will donate them to you(*). The scientific american article calls for 1450 gauss which should be practical. The hard part is getting the field uniform using obtainable materials. Usually I use 1018 steel or similar. Of course, constructing something using individual magnets is pretty difficult and dangerous. (Two magnets of this size in free air can amputate small digits when they come together). Be aware that is very difficult to construct a design where a N-N field is in close proximity along the long edge of a magnet. The 0.5 edge is do-able.
 
Magnets i have are like these:

http://www.kjmagnetics.com/proddetail.asp?prod=BY088

As you can see, not cheap. Obviously, bigger magnets are better but also very much more expensive.

Randy

(*) Reasonable restrictions apply. i have about 25 available and will not pay shipping costs. Donation obligates recipient to do good documentation!

[LaserSteve]

Early References:


1.  N. Bloembergen, E. M. Purcell, and R. V. Pound, Phys. Rev. 73, 679 (1948). 

2.  G. E. Pake, J. Chem. Phys. 16, 327 (1948).

3.  P. Grivet, M. Soutif, and R. Gabillard, Physica (Utr.) 17, 420 (1951).
 

4.  F. Bloch, W. W. Hansen, and M. Packard, Phys. Rev. 70, 474 (1946). 
 

5.  N. J. Hopkins, Rev. Sci. Instrum. 20, 401 (1949).  [ CAS ?]
 

6.  R. V. Pound and W. D. Knight, Rev. Sci. Instrum. 21, 219 (1950).
 

7. E. Lustig and W. B. Moniz, Anal. Chem. 38, 331R (1966). 
 


8. B. V. Rollin, Nature (Lond.) 158, 669 (1946). 
 

9.  Gabillard, C. R. Acad. Sci. (Paris) 232, 324 (1951).

Steve

[ale500]

TMS, tetramethylsilane is only used as a reference for 0 ppm shifts, not as a solvent. As a solvent you need something without H, if you want to measure protons, to avoid having a strong H signal. Anyways modern machines use the residual H in deuterated solvents to calibrate the scale, TMS is not required anymore.
I used to work with a 400 MHz Bruker machine (like 15 years ago), it needed manual shimming, lower frequency machines had automatic shimming.

[Kleinstein]

The old AlNiCo Material has 3 good properties: It has lower temperature dependence than other modern magnets. But I think by keeping the temperature reasonable constant this is not a big problem. Also the mechanical strength is quite good, but usually the magnets are under mechanical compression, so not a problem anyway. The higher (differential)  permeability of Alnico (about 3 compared to about 1) can be an advantage when, combining it with an electromagnet - however this might not be a good idea anyway, if one needs to keep the temperature constant. I would consider extra coils only for shimming and maybe compensation of changes in the external field - so more a low power part.

Using the FEM simulation is a good idea, as the geometry can be different from old designs using AlNiCo or ferrite magnets. This is because the Nd magnets are usually magnetized along there short axis.

[ChristofferB]

Thanks for all the feedback! there's more than enough to go on!

- RandallMcRee: That's very generous, but this is more of a "collect parts and ideas over 10 years" project for me. There'll probably be a loong time before I start on anything actually resembling a project.