As I read in the text , we can calculate, the magnetic moment per particle by multiplying the magnetization at saturation with the volume of the particle. Anyone know of any other ideas?
Yes. I am having silver nano particles having ferromagnetic behavior at low temperature and at higher temp a super paramagnetic behavior. spin orbit interaction will give magnetic moment.
Sorry--no! Spin orbit interaction will not per se provide a magnetic ground state.
Moreover, the electrons which might form a magnetic moment derive from the atomic 5s (zero angular momentum) and 5p (l=1) states, mainly. S.o.c. is not very strong on the silver states (much less strong than in gold, for example).
In order for me to believe that silver gets ferromagnetic you would have to provide me with element specific evidence for this finding. Finding magnetic moment in a specimen which contains Ag nanoparticles is not sufficient. Unfortunately, way too many magnetometers and sample holders etc. in this world are contaminated with material which provides an erroneous signal. It suffices that someone uses the wrong tweezers to generate a ferromagnetic signal from a glass plate.
Read these carefully, please:
Greget et al.:
Magnetic Properties of Gold Nanoparticles: A Room-Temperature Quantum Effect
CHEMPHYSCHEM 13, 3092-3097 (2012)
DOI: 10.1002/cphc.201200394
Bartolome et al.:
Strong Paramagnetism of Gold Nanoparticles Deposited on a Sulfolobus acidocaldarius S Layer
If suppose we are using modified Langevin function to fit MH data and getting values of variables such as saturation magnetization, no. of particles contribution to magnetization, diamagnetic susceptibility etc. from this data we can calculate particle moment. We have already published one paper. The details are here:Jpn. J. Appl. Phys. 47 (2008) pp. 706-711 [DOI: 10.1143/JJAP.47.706]
I am not quite sure to get your point: the paper you mention is on Cu2O, not silver, - and you're not a coauthor?!
Nevertheless, I'd not expect Cu2O to be a ferromagnet neither (from simple electron counting rules) but I may be missing something here. I'd be interested in reading the paper in order to see how one can derive a surface anisotropy from measurements and a Heisenberg model. I'd love to be able to do that. However, I unfortunately have no institutional access to JJAP.
In the JJAP paper, magnetizations are given in emu/g. This is per gram of what and was determined in which way?
In what state was the sample at the time of magnetization measurement? The particles still on the stainless steel plate as collected in the first place?
... And they confirmed Cu2O composition from EDS? XRD diffraction peaks look too narrow to be from 8nm particles as well -- too many conclusions drawn from one XRD diffraction pattern. I don't think they did any ICP on these either before and after magnetization measurements -- or any other analytical assay to determine metal contamination!
magnetization given in emu/g is calculated by dividing amount of sample used for SQUID characterization to the total magnetization shown by sample. The sample is in powder bulk powder at the time of measurement. The particles are stripped off after experiment is completed.
As you can see from my publications I am interested in and working on magnetic nanoparticles. Our preferred method of investigation is based on x-ray spectroscopy ("XMCD") using soft x-rays. We detect our signal by "electron yield". The finite mean free path of excited electrons leads to the fact that this is a fairly surface sensitive technique, probing only few nm into a sample (we like this, but it can also be a disadvantage at times).
In one of the preparation methods, nanoparticles are obtained in a chemical route involving reverse micelles (DOI: 10.1002/adma.200601759). Later, a colleague of mine picked up the method to make pure Cobalt particles, which did work fine, in principle. We made a few XMCD measurements. Additionaly, SQUID measurements were undertaken. Since the SQUID is not working under vacuum, some coating was tried for the particles to survive transfer and measurement in the metallic state.
In the SQUID, at low temperature, an important additional amount of magnetic moment was suddenly visible. The temperature dependence suggested a paramagnetic contribution, roughly a few Bohr magnetons per unit.
The lesson learned: besides the nanoparticles the method produces lots of paramagnetic "things" (mostly Co atoms, presumably), buried in the substrate such that we won't detect any of them in our XMCD experiment. SQUID measurements, unfortunately, cannot characterize these particles at low temperature due to the additional magnetic moment detected by this volumetric method, since the contributions cannot be identified and separated reliably enough.
Did we publish this? No. Why: no substantial result for nanoparticle magnetism, unfortunately.
Now my preliminary assessment of what I see in the JJAP paper:
There is a temperature dependent slope in the M(H) curves. The slope increases towards low temperature. This looks like a paramagnetic contribution to me due to some magnetic "impurity" or "contamination".
[The surface anisotropy invoked in the paper requires a valid mechanism first. The possible mechanisms relate to spin-orbit-coupling induced anisotropy and nonspherical spin density distribution at the magnetic sites. Both contributions can be enhanced at particle surfaces. Invoking the Berkowitz work on ferrites for another material is very dangerous. The spinels are known to produce strong spin canting/anisotropy effects at the nanoscale, which other materials will not so readily mimick - especially when moments are small].
The higher moment - Langevin-like looking - part of the M(H) curves are remarkably temperature independent. Nothing I'd normally expect from nanoparticle samples. (Others may have different opinions and experience - please share). As I understand it, anomalously rapid demagnetization by spin excitations is invoked in the paper to rationalize this finding. I would regard it as highly accidental if that effect could really account for a nearly temperature independent M(H) curve.
My first thought would be: "something's strange here".
Ag nanoclusters have been extensively studied theoretically, and as far as I know there's no evidence or reason why they should have a net magnetic moment. My (real) ab inito colleagues (especially prof. Kari Laasonen) have also done extensive ab initio on Cu2O, and there's again no trace of magnetism in this system.
Sorry if my professional jargon is incomprehensible. Ab initio (or first-principles) refers to a numerical, iterative solution of the quantum-mechanical Kohn-Sham equations for the electronic structure of a given system (metal cluster here).
Report of ferromagnetism in nanoparticles that are nonmagnetic in bulk is very common. But most of these are due to misinterprtation of the data or due to the presence of impurities as Kai has already pointed out. There had been an outragious report claiming that all materials are ferromagnetic in sufficiently small nanosizes!
In my case, I have characterized samples with EDS but there were no impurity traces found in it. Even, SRXRD also didnot detect any oxidized phases. Suppose observation of ferromagnetism in nanoparticles which are diamagnetic in bulk is because of misinterpretation, but few papers are published in Nature Science also.
Please give me the references. I shall read them and give you my comments. I must also add that papers published in Journals with high impact factors are not always scientifically correct. It is simply because you cannot publish in those journals unless you claim something extraordinary. This leads to over interpretation and even falsification!
Here I am giving some references in which they are saying about observation of ferromagnetism in Metal nanoparticles which are non magnetic in bulk. I am confusing with answers given earlier. Could you please help me to clear my question? Thank you!
DOI: 10.1103/PhysRevLett.93.116801
PRL paper entitled: Direct Observation of Ferromagnetic Spin Polarization in Gold Nanoparticles,
DOI: 10.1103/PhysRevB.69.174411
PRB Title: Diameter dependence of ferromagnetic spin moment in Au nanocrystals
DOI: 10.1103/PhysRevLett.93.087204
PRL Title: Permanent Magnetism, Magnetic Anisotropy, and Hysteresis of Thiol-Capped Gold Nanoparticles
DOI: 10.1103/PhysRevLett.91.237203
PRL title: Ferromagnetism in fcc Twinned 2.4 nm Size Pd Nanoparticles
I read some of the papers and I find them rather careful experimental works. It looks like that Gd and Pd nanoparticles do develop very small magnetism somehow. So your Ag nanoparticle may perhaps develop magnetism in the nanoscale. But you have convince by doing several careful experiments. You may do XMCD experiment at the Ag edge if possible. That may convince skeptical Referees.
You still have to answer the questions raised by Kai and also why ab-initio calculations on Ag nanoparticle do not show any magnetism? Kai is a specialist in this field. In your place I would try XMCD investigation. But of course sample preperation and handling are crucial in this research. Also can you give me the reference of the (your) publication that Kai referes to?
@Shrikrishna, a word of caution: Sensitivity of EDS (even wavelength dispersive ones) to detect trace impurities is few orders of magnitude worse than the level that could influence your magnetic / SQUID measurements. Also impurity does not always behave volumetric when it comes to influencing your signal. Don't get me wrong; I am not trying to say your signals are from impurities, but scanning via EDS does not prove that there are none. ICP scans before and after you do the magnetic measurements including the "non metallic" capsule / holder you used would get you closer to your answer. Last year we were scanning for 50+ metals using a third party double blind commercial ICP source to check on impurity related magnetism from another nanoparticle system -- nanodiamonds.
Thanks Kai. Meanwhile I read a paper in Nano letters. Ab-intio calculations on Au and Ag clusters show that these clusters shows magnetism. Have you seen any such evidence experimentally?
Actually no, but I also never tried. Au already was debatable for a long time and when I last was in Grenoble (2-2.5 yrs ago) I met Andrei Rogalev who critisized it all for being nuts (to me he is one of the top experts in the hard x-ray regime). Meanwhile he's got 2 papers on the subject himself and that is what I recommend for reading when the subject comes up.
Ag is even more difficult for me to believe. The 4d's are >3.5 eV below the Fermi energy. And it is harder to measure, because the energies for L2,3 excitation are in a range which is not readily accessible in most beam lines. And where it is, intensity is way off optimum.
Could you share that nano letters reference? I'd be curious for the mechanism and the magnitude of moments.
Thank you very much for your consistent comments and discussion on this topic. Here, I am giving some references from Nano Letters, which are talking about ferromagnetism in Silver/Gold nanoparticles.
s-Electron Ferromagnetism in Gold and Silver Nanoclusters
Thanks. With high degeneracy of states, as in these calculations, this is indeed a possibility I could think of. This is some ultimate form of small scale Stoner magnetism, then. But in how many real specimes would we have a significant amount of particles fulfilling such a requirement? (I remember a colleague suspecting something similar for Na_11 clusters, if I remember this correctly. I don't think this was ever published, though. Now, Na has a fairly large exchange constant. I could imagine this is partly due to the fact that it has low atomic number and therefore relatively few nodes in the radial wave function (3s). This is quite different in Ag (5s) and Au (6s)). Now I don't know whether the potentials used were full potentials or preudo-potentials. On this scale this could make a difference.
A side note: I don't actually know, whether the approximations built in to VASP leave such a result untouched or have a probability to produce them. It would be important to confirm such a finding with other implementations of DFT.
It would be good to get comments from knowledgeable theorists on the topic.
Let me add a few thoughts: the beautiful experiments by B. von Issendorff and collaborators using photoemission on mass selected gas phase clusters demonstrate impressively how well the super-atom picture actually works in the noble or coinage metal clusters. Also the increased spin orbit coupling in Au can be readily appreciated.
This means that in the cluster potental well the delocalized single electron orbitals do also acquire angular momentum etc. The associated phase changes (exp (im\phi)) now do not occur along the movement about a specific atom but about the cluster as a hole. then, of course, also spin orbit coupling occurs etc.
Remember, though, that all the energies underlying the first and second Hunds rules are Coulomb energies in fact. As such, they are MUCH smaller for delocalized states than in atomic volumina. I would therefore expect both, the stabilization of the lowest energy LS multiplett and/or the magnetic ground state against Stoner excitations to be MUCH weaker than in other cases. One would have to check at which temperatures these configurations can actually survive.
There is a comment about the disappearance of mgnetic moment of the Au cluster in the Nano Letters paper. "Because of small energy level separations near the HOMO , we fiind that a small Gaussian smearing energy (0.01 eV) is required...Larger Gaussian smearing destroys the magnetic moment, implying the disappearance of magnetic moment at higher temperatures.
And this is probably not talking about Stoner excitations, which change the multiplicity and therefore do not appear in the single particle description.
I remember that someone commented that his colleagues did ab-initio calculations (DFT?) and found no moments in Au and Ag nanoparticles? I don't see his comments anymore here. Do you remember that Kai?
I have found now the comments. Tapio, what do you say about the calculations published in Nano Letters? They claim that aAu and Ag cluster can develop magnetism.
I would suspect so, yes. In the calculations the "more spherical" icosahedral clusters get magnetic more readily due to the higher degeneracy.
I would think that you can view shape variations and surface chemistry in analogy to crystal field splitting for atomic states. Their effect is likely to lift degeneracies.
So the icosahedral symmetry plays a big role for the appearance of magnetism. As far as I know you can get such metallic clusters in atomic beams. A colleague of mine in MPI Stuttgart used to produce them. But that was quite some time ago. Now how can we get such clusters fixed in some condensed substrate? This reminds me that such icosahedral clusters exist in solid intermetallic phases. I wonder weather we can get magnetism in such alloy phases containing non-magnetic atoms.
Now coming to experimental detection of small magnetism in such clusters I guess atomic physicists have enough tools at their disposal. But clusters in solid substrate (if at all that is possible to produce) is quite a different matter. Magnetic impurities are probably bound to creep in. The bulk magnetic properties will not help. So only technique that is left is the element-sensitive resonace X-ray absorption. What do you think Kai?
In 'real' solid phases I am skeptical about an expectation that this kind of magnetism would persist. These s,p electrons will be the first to engage in chemical bonding.
In my 'early years' I worked on organo-metallic Au clusters that were believed at the time to consist of 55-atom-fcc fragments. Their surfaces were decorated with phosphene ligands and Cl counter ions. These guys did'nt quite show a signature of the Au plasmon resonance. I am just mentioning this to indicate that the behavior of the delocalized electrons is readily changed in such small objects.
It seems that in later experiments (e.g. R. Whetten et al.) it was shown that using thiol terminated alkane ligands and excessive purification lead to optical properties with some structure to it. So may be that our earlier samples were just not sufficiently monodisperse (although there was reason to believe they were really good).
Whether or not you get magnetism does also depend on overall band filling, which you need to taylor for high degeneracy. Remember that e.g. in quasicrystals band filling is such as to yield a strongly reduced DOS at the Fermi energy - that is energetically favorable (these quasicrystals are the ones that form).
As to the detection with XMCD - yes, it is a technique with tremendous sensitivity. Nevertheless you need suitable absorption edges. In the noble metals it is not simple. With the proposed magnetism (Stoner magnetism of sp-type electrons) it will be very hard. L-edge absorption will be the only way to go to capture magnetic polarization of s-type electrons. It is a l->(l-1) transition though and as such carries relatively little oscillator strength (e.g. 1% compared to 2p->3d in the 3d metals).
In silver, the L edges are situated in a relatively unfavorable range.
With free clusters in the gas phase, Stern Gerlach experiments could represent an alternative (but need to be really well done). I have recently met W. de Heer, who reported on findings of both diamagnetic and paramagnetic gold clusters. Since bulk Au is diamagnetic, paramagnetism in clusters could hint at some moment formation.
Sure quasicrystals are made of such non space filling clusters, but there exist even close crystalline versions, albeit distorted, because perfect tetrahedrons don't fill up space.
I am curious to know whether such magnetic cluster with small moment can have any application at all! So it just of fundamental interest or there is some possible applications?
I can supply some know-how on the ab initio calculations here. The calculations by Luo et al. appear to be kosher to my eye. There are no real issues with PAW or pseudopotentials in this case, or indeed in any case, the basic formalisms are perfectly applicable everywhere, but many of the standard implementations come with major caveats when going to high pressures or otherwise extreme geometries. But that is not the case here.
As was already noted, Luo et al. state that a > 0.01 eV gaussian smearing destroys the magnetism. This is a very small smearing indeed, one can estimate the corresponding temperature by equating the full width at half maximum values of the gaussian (=2.4*sigma) and Fermi-Dirac distributions (=3.5*kB*T). A sigma of 0.01eV then corresponds to about 80K, so one would not expect to see this at room temperature. I would not expect to see it at 80K either, since the atomic vibrations should also contribute a smearing even to the highly symmetric clusters found to be magnetic. A simple standard estimate of the zero point motion for (bulk) silver from the Debye temperature is 9*Theta_D*k_B/8, which for silver (Theta_D=215K) would be 0.02 eV. Bashing out the naive estimates a bit more, that would be about 0.01Å minimum uncertainty for a harmonic oscillator. That is not so small, if I were to calculate phonons in silver, I would use distortions of about 0.02-0.04Å to perturb the lattice, so I suspect that random displacements of that size would spoil the degeneracy sufficiently to remove the magnetism.
One can probably raise objections to the above estimates, but, to my eye, the calculated results by Luo et al. should be interpreted as a strictly 0K result.