As for NPD, I am not sure whether it's possible. It could be it's different for single crystals, I'll try to check. But your surrounding at ILL should be one of the places with the highest probabilities, worldwide, to find someone who can answer accurately, shouldn't it. You may also consider going next door, the 'X-ray' store at ESRF does have world famous beam lines where you can study orbital moments using XMCD. Try meeting F. Wilhelm (he is on RG) for lunch.

What you say is right. In principle you cannot separate from neutron diffraction data orbital and spin parts of the magnetic moment. However flipping ratio measurements (polarized neutron diffraction) from single crystals with appropriate modeling can sometime achieve this. The spin and orbital parts of magnetic moment can be separated by X-ray scattering in principle. XMCD is a very good technique for ferromagnets but has problems with antiferromagnets. Also they get only Q = 0 data. Single crystal X-ray magnetic diffraction can be used for such study and can also give Q dependence (see the classic paper W. Neubeck, C. Vettier, F. de Bergevin, F. Yakhou, D. Manix, L. Ranno and T. Chatterji, J. Phys. Chem. Solids 62, 2173 (2001)). However all these fancy and difficult experiments (mind that the relevant X-ray magnetic scattering beam line ID20 has been killed by the ESRF) do not help for ordinary magnetic structure determination with NPD data. What we need is to refine the magnetic structure using appropriate magnetic form factor. We have to use NPD data because new samples come (at least at first) as powders. Surely XMCD helps but I guess not very much for antiferromagnets and helimagnets and we don't get Q dependence.

With single crystals, you still have linear x-ray dichroism for the antiferromagnets. As for the restriction Q=0, you're entirely right, of course. But then I'd say Q-dependency is already an advanced piece of information and you might be willing to go for a fancy experiment to get it...

So, what is the progress you imagine? Would it consist of better calculating the spatial spin density distribution (appropriately taking into account crystal field splitting and spin orbit coupling) and getting modified form factors on this basis? What does it take to discriminate such modifications from other parameters entering the evaluations?

I could imagine CoO to be a good test candidate. It destorts tetragonally and has been reported to possess quite sizeable an orbital moment (I can't find the reference I thought of quickly, but I see that there's been a lot of activity in recent years - you will know that, I guess). Powders and crystals are readily available. So, maybe one can produce some experimental evidence for how to improve on form factors?

Alternatively we could try to invent high energy (very) high resolution inelastic electron exchange scattering. Maybe it exists already...

I guess we the problem of orbital moment n a magnetic solid (especially with transition metal ions with unquenched orbital moment) is a difficult one. We have problem to refine magnetic structure because we do not know which magnetic form factor to use. This results in people published wrong magnetic moments by using Fullprof as a black box. The single crystal polarized neutron diffraction has often the problem of extinction/absorption and also structural/magnetic domain populations (not easy to get a single domain single crystal). This is about static magnetism. Things are much worse when we go to spin dynamics. The neutron scattering cross sections become horribly complex and therefore people use just the spin part yielding wrong results. X-ray scattering hardly helps. You mention "high energy high resolution electron exchange scattering". I never thought about this possibilities and I shall be much interested to know more about it. Has anyone really employed such technique to probe orbital magnetism. What is the principle? Do you exploit electron spin? Can one do such experiment on a magnetic material such as CoO?

Dear Tapan, concerning the hypothetical electron experiment, I don't really know, it was a spontaneous product of my fantasy. Not without background but nothing serious (yet). Elextron exchange scattering has been done in the past at low energies using EELS spectrometers. I believe to remember H.-J. Freund's group did such experiments on TM-oxides before he moved to Berlin (so that's a loooong time ago). then there have been claims on exchange scattering in LEED with different branches, one going for spin polarization (e.g. Feder) and others claiming to see AFM superstructures in LEED (put into question later by yet others).

I wrote about high energies because in order to achieve sensitivity at the powder level, you need high energy to reach sufficient penetration depth. But then, of course, you need very high energy resolution to get to the magnetic energy scales. I don't think that this is trivial by any means.

{and not to mention that the signals may be entirely dominated by other scattering mechanisms!! But maybe - just as with RIXS - it is possible to conceive well-suited scattering geometries that let you see what you want to see.}

Tapan, concerning single phase single crystals: I could imagine that with all the recent progress in TM oxide PLD it might be possible after all to grow strained CoO such that you have a homogeneous easy axis over large areas. I am not enough into neutron scattering to imagine what thickness would ultimately be required and whether that can be achieved with sufficient quality. But maybe a thin, large area single phase crystal could help circumventing some of the difficulties with crystals you mentioned before.

Also, while bulk CoO goes cubic at the Néel temperature (and I'd imagine this goes along with some change in the local orbital moment) the strain induced tetragonal distortion will remain there (while most probably T_N changes). Maybe this is useful.

The best way to split magnetic moments into orbital and spin contribution is XMCD. Look at this link http://neel.cnrs.fr/spip.php?article798&lang=fr

Paolo, Tapan knows this well, he is located at ILL and has worked with both, neutrons and x-rays. He has elaborated above on why and what for he is seeking for alternatives.

Ok, pardon me. I was just looking at the first question and didn't follow the whole discussion. Anyway, it's a difficult problem, and on my opinion it is mandatory to handle it using several techniques. I would go to have a talk with N. Brookes at the ESRF, for discussing both XMCD and RIXS. X-ray magnetic scattering can be of help as well, although "it's a difficult experiment". The L(Q)/S(Q) ratio is obtained by the intensity of the magnetic diffraction effects in non resonant conditions. Such an experiment would work on antiferromagnets and it should still be possible to perform it at ESRF ....

Thanks Kai and Paolo. I shall certainly talk to Nick.