Spin FETish: Alternative transistors become more appealing

Spin FETish: Alternative transistors become more appealing

Pick of the week is back after shutting down for festive period.  This week’s pick is not actually from this week’s literature, and in fact dates back to 22nd December 2014 (just as I set off for my Christmas break), but the paper is significantly important and I feel that it must be covered here.  So impressed was I by “All-electric all semi-conductor spin field-effect transistors” by Chuang et al. in Nature Nano that I even consider it a very late runner for my favourite paper of the year (the main competition being “Direct determination of spin-orbit interaction coefficients and realisation of the persistent spin helix symmetry“).


Working Datta Das transistor

So, what makes Chuang’s effort so impressive?  Well, the title of the paper is pretty much the reason, assuming the content lives up to its billing.  The Datta Das spin transistor was proposed in 1990, and like GMR, is one of the cornerstone clichés in any general spintronics overview articles or PhD thesis.  A Datta Das transistor could offer several major advantages over a conventional charge-based transistor, including lower power consumption and cost, as well as being a possible component in an MRAM device.  The problem is that since 1990 a working Datta Das transistor has yet to be realised, with short spin lifetimes and low spin injection efficiently being key limiting factors.  Yes, there have been several demonstrations of spin transistors, but these are not devices that are compatible with existing electronic hierarchy.  This is because their operation requires a component that is not common place in large scale integrated circuit in order to achieve spin injection and detection, such as spin-polarised light or magnetic metallic contacts.  For Chuang’s spin transistor, all that is needed is semiconducting layers, a dielectric layer, and metallic gates, all of which can be found in a charge-based field-effect transistor.


Quantum point contacts

What separates Chuang’s transistor from all previous efforts is the use of quantum point contacts (QPCs) as the spin injector and detector.  QPCs, which are essentially a pair of laterally split metallic gates through which the source current must pass and across which a voltage is applied, have traditionally been used to generate an electrically one-dimensional current with a quantised conductance.  However, as has previously been noted, QPCs can also affect the spin degree of freedom of a charge carrier as the lateral voltage they apply in the plane of the sample can be considered as a lateral inversion asymmetry (sort of like a lateral version of the vertical electric potential gradient responsible for Rashba spin-orbit coupling in multilayer structures).  Due to spin-orbit coupling in the semiconductor layer, this lateral asymmetry causes the charge carriers to experience an out of plane effective magnetic field, along the direction of which their spin align parallel/anti-parallel.  Thinking of this in terms of a band-diagram, the out of plane effective magnetic field shifts the spin split carrier bands in energy, and so if the Fermi energy is tuned so that it crosses only one spin sub-band then the current passing through the QPC is completely spin polarised.  The same principle is true for current approaching a QPC (i.e. the detector in the spin-transistor) – it is an excellent spin filter.

The functionality of the transistor requires well-deifined on/off states, and for Chuang, these states are obtained in the traditional Datta Das method.  The spin polarised current from the detector passes through a region with a top gate separated from the semiconductor (InGaAs 2DEG in this case) channel by a dielectric layer.  Applying a voltage between the top gate and the substrate creates an electric field that acts out of the sample plane, and this is another form of structural asymmetry.  Therefore, as this is the orthogonal case of the QPC lateral inversion symmetry, the charge carriers (which are initially spin-polarised out of the sample plane) now experience an in-plane effective magnetic field, again as a result of spin-orbit coupling.  As the in-plane effective field is orthogonal to the spin orientation, it leads to a precession of the spin.  The orientation of the carrier spin at a given instant now depends on its location within the conduction channel (essentially how long it has been precessing for), and more importantly the strength of the in-plane magnetic field, which can be tuned by varying the top gate voltage.  Assuming the detector QPC is set at the same voltage as the injector QPC, if a carrier spin has precessed to its initial orientation by the time it reaches the detector it will be allowed through, but if it has precessed to an anti-parallel state it will not, and so therefore the drain current can be controlled by the gate voltage, just like in a charge-based transistor.


Well defined states

The results the paper presents are the most promising yet for any spin transistor.  The current passing through the detector is shown to vary as a function of gate voltage in an oscillatory manner as would be expected for a spin-precession based effect, with the minimum (off) current being more than 500 times smaller than the maximum (on) current.  This is considerably the best defined on/off states for any spin transistor.  The origin of the effect is further confirmed by varying conductance and temperature.  The 500x effect is achieved when the conductance is set just above the threshold conductance value so that only one spin sub-band is occupied.  As conductance is increased (by changing the QPC voltage), the on/off states become less well defined, and this is due to more than one spin sub-band becoming occupied and allowed past the injector QPC, thus less efficient spin injection.  Similarly, as temperature is increased the on/off states are also shown to become less well defined, again as (due to thermal considerations) a second spin sub-band begins to become occupied.


The heat is on

The sheer simplicity of the physics behind Chuang’s spin transistor is what makes this paper so appealing to me.  For nearly 25 years a good Datta Das transistor has eluded researchers, and this paper demonstrates the best attempt yet a realising just that.  Also, the paper introduced me to the idea of lateral QPCs as a spin-injector.  I wonder if such a structure could be used for other experiments investigating spin-orbit coupling phenomena?  For example, the in-plane effective magnetic fields created by structural inversion asymmetry (associated with Rashba spin-orbit coupling) can lead to carrier spins torqueing magnetic moments in magnetic materials.  It would be interesting if such QPCs could be applied to a magnetic semiconductor with out of plane magnetic anisotropy to lead to magnetic switching.  For now though I hope the focus is on improving Chuang’s spin transistor, most importantly to operate at significantly higher temperatures.  As Chuang suggests, changing the physical structure of the device is on approach to this.


Chuang, P., Ho, S., Smith, L., Sfigakis, F., Pepper, M., Chen, C., Fan, J., Griffiths, J., Farrer, I., Beere, H., Jones, G., Ritchie, D., & Chen, T. (2014). All-electric all-semiconductor spin field-effect transistors Nature Nanotechnology, 10 (1), 35-39 DOI: 10.1038/nnano.2014.296


Image: photognome, some rights reserved