JOURNAL CLUB: SESSION 2

[Jose Luis Lado Villanueva]

Hello everyone

to continue with the journal club, I propose the following paper. The title is "Graphene spintronics", so I think that it should suit our interests! Lets define as deadline to read the paper the next Tuesday 12th May (one week from today), so that on Wednesday 13th we start the discussion.

You can download the paper from this post.

The original link is http://www.nature.com/nnano/journal/v9/n10/full/nnano.2014.214.html 

See you next week!
Jose

[Jose Luis Lado Villanueva]

Hi guys. To start with, I would to point out some questions that I had while reading the paper, so that we can share what we think about them.

• Regarding the nonlocal signal
  
• Which materials are better to be used as ferromagnetic electrodes, and why?
• How accurate do you think is the semiclassical picture of spin diffusion in graphene, are quantum effects important?
• Is there any upper bound to spin injection efficiency, and is it some simple way to improve the tunnel barrier result?
• Is it possible to measure the nonlocal signal in other 2D materials (MoS2, black phosporus)? How large has to be the mobility?

• Regarding magnetism in graphene
• Which would be the best way to measure local magnetism in a 2D material and how to distinguish between this local magnetism and magnetism created by impurities?
  
• Is it possible to create local magnetism in other 2D materials and which elements would chemically bond to them? 
• Could fluctuations completely destroy magnetism? How important is Kondo-like phenomena in graphene?

• Regarding spin orbit coupling
• Which 2D materials could be used to induce larger spin orbit coupling by proximity?
• Would it be realistic to use to use as substrate materials with large electric dipolar moment?
  

• Regarding spin relaxation
• What is the interplay between charge diffusion and spin diffusion?
• How important is the effect of substrate magnetic impurities and spin orbit coupling?
• Which would be characteristics of the "ideal" substrate to minimize spin relaxation? Is there any realistic material which fulfils that?
• Which mechanism would become dominant for other 2D materials?
  
• How does spin relaxation depend on the doping level?
• Does the spin relaxation change a lot if the experiments are done in different atmospheres (O2, N2, He...)?
  

[josepingla]

Hi all,

I think I can give answers to some of your questions:

• Which materials are better to be used as ferromagnetic electrodes, and why?  First of all, they should be ferromagnetic, then they should have a relatively high coercive field, so that you can measure Hanle (apply an out of plane magnetic field) without pulling the contacts out of plane at low fields below 0.2 T. We also want to have contacts with preferential magnetization directions. We call that shape anisotropy because it is determined by the shape (width, length and thickness) of the contacts.  Cobalt has high coercive fields and it works quite well. The only problem is that it oxidizes in air. 

• How accurate do you think is the semiclassical picture of spin diffusion in graphene, are quantum effects important? For the spin valves you can find in the literature, the mean free path is in the order of 0.1 um, no ballistic effects expected (I have measured on a sample with a mean free path  higher than 1 um but it still showed diffusive behavior over 15 um). The measurements are done at room temperature, where the phase coherence time is extremely short, and also at 4 K, where you still cannot see extremely sharp localization peaks, meaning that quantum corrections are not so important in these samples. Also we don't see UCF at 4 K, even in the best samples.  Apart from these points, the most convincing reason for me is that the Hanle curves we obtain fit perfectly to the diffusive model, which I think is also a good indication of diffusive transport (what a coincidence it would be that there is something different going on following exactly the same curves).

• Is it possible to measure the nonlocal signal in other 2D materials (MoS2, black phosporus)? How large has to be the mobility? Yes it is, the problem is the conductivity mismatch problem. Since these materials usually have high resistances, if the contact resistance is not high enough, the injected spins find it easier to come back to equilibrium by relaxing back to the contact reducing the amplitude of the spin signal drastically. I don't think the mobility is a problem, look at the mobilities of http://www.nature.com/nature/journal/v448/n7153/full/nature06037.html, I am sure that the mobilities achieved in MoS2 are better.

• What is the interplay between charge diffusion and spin diffusion? The only difference I now between them is that spin diffusion coefficient is affected by elastic scattering between electrons while the charge diffusion coefficient it is not (this mechanism is called spin Coulomb drag, there are quite some papers about it) but, in the systems we are using, the main scattering sources come from impurities attached to the flake, phonons at room temperature, etc... If the probability of electron-lattice (+crap) scattering events is much higher than the electron-electron one, then basically you will never see the electron-electron interactions. You can also think about it in terms of the Matthiessen rule: 1/T=1/T1+1/T2, if T1>>T2 then 1/T=1/T2 being T1 the scattering time for electron-electron processes and T2 the scattering time for electron-lattice (+crap) events.

Cheers,

Pep

[j.c.leutenantsmeyer]

Hoi hoi,

Additionally to Peps remarks, I would like to point out here that should also care about a high spin-polarization. For these you can still stick to Cobalt, but if you want to have a spin-polarization of more than 50%, you have to go to more advanced systems. Just to mention one: CoFeB/MgO junctions work and have demonstrated a spin polarization of around 50% at room temperature and around 60% at low temperatures. The idea is basically that you filter the band of the system by symmetry. If the interface quality is good, only the D1 band is dominating the transport through the barrier and that band is fully spin polarized at Fermi level in Co and Fe. So you end up with a high spin polarization. As far as I remember theory predicts here up to 80% spin polarization.

There was actually a publication earlier this year on graphene on YIG. They probed the Hall resistivity of the graphene and found besides the normal Hall effect an additional component to the Rxy which was following the perpendicular YIG magnetization. They attributed that to the anomalous Hall effect, which occurs in ferromagnets. It is definitely a nice way to study the ferromagnetic proximity effect. If you have no other magnetic impurity, you could still put the sample into a SQUID or VSM, but then already small contamination can mess up your measurement.