Skip to main content
European Commission logo print header

MIcrowave emission LinEwidth of Spin-Torque-Oscillators and associated phase NoisE

Final Report Summary - MILESTONE (Microwave emission linewidth of spin-torque oscillators and associated phase noise)

Spin torque oscillators (STOs)

Spintronics offers new possibilities to realise nanoscale microwave components due to the spin momentum transfer in ferromagnetic (FM1) / nonmagnetic (NM) / ferromagnetic (FM2) heterostructures (1,2). Here the spin polarised direct current (DC) polarised by FM1 interacts with the local magnetisation of FM2 and transfers spin angular momentum. This interaction represents an energy feedback which, for large enough current amplitudes, can overcome the energy loss through natural damping, resulting in self-sustained steady-state oscillations of the FM2 magnetisation. Such large amplitude oscillations are unachievable by conventional methods. These magnetisation oscillations, combined with the giant or the tunnelling magnetoresistance effect, give then rise to an oscillating output voltage, at frequencies that lie in the range of Gigahertz. Such a STO is a non-linear dynamical system characterized by a strong non-linear coupling between the oscillation amplitude and frequency (phase) (3). This provides the possibility to tune the oscillation frequency by altering the DC current and to realise nanoscale tuneable microwave generators that will find applications in telecommunications.

However, one of the major bottlenecks for the implementation of STOs is their relatively large microwave emission linewidth with values of about 10 - 100 MHz. The linewidth is directly related to the fluctuations of the amplitude and the phase around the stable trajectory. Understanding the emission linewidth and the physical mechanisms that lead to linewidth broadening is a crucial scientific and technological issue, if not the most crucial one, for the development of such microwave components. The scientific aim of this project was to gain a better understanding on the parameters that determine the emission linewidth.

A first important challenge was to establish real-time measurement techniques using a single-shot oscilloscope to extract the amplitude and phase fluctuations that at the beginning of the project had been little explored. These measurements were possible, due to a joint collaboration between SPINTEC UMR 8191 in Grenoble, France (host laboratory), CEA/LETI (Grenoble, France) and HITACHI GST, USA who provided high quality magnetic tunnel junction devices. The important results are:

(i) For STOs, the frequency fluctuations dominate the noise properties.
(ii) Frequency fluctuations are white Gaussian and occur on all time scales experimentally accessible (going from ms to ns). This is in contrast to what has been deduced prior to the start of the project from frequency-time spectrograms.
(iii) Comparison of the amplitude and phase noise power spectral density to theory provides a very useful tool to extract two non-linear parameters that determine all the properties of STO devices (4).

These parameters are the amplitude-phase coupling and the amplitude relaxation rate. With this we have demonstrated a powerful tool that can be applied to different configurations, excitation geometries and experimental conditions to extract the amplitude and phase fluctuations as well as non-linear parameters. In a further study we have provided an alternative, original approach to extract these non-linear parameters, which is based on the study of higher harmonics excitations (5).

These new time domain techniques were used to study the role of thermal fluctuations on the emission linewidth in planar magnetic tunnel junction devices (magnetisation oriented parallel to the layer plane). Prior to the studies carried out in the project, different groups performed temperature dependent studies of the linewidth broadening with controversial results. The project clearly demonstrated the role of the thermal fluctuations for the in-plane precession (IPP) mode once the experimental conditions have been chosen carefully. More specifically it was shown that the linewidth decreases linearly upon reducing temperature (6). The experimental results confirmed the analytical theory of non-linear auto-oscillators (3).

To reduce linewidth different approaches can be followed. One of them is to use phase-locked arrays of STOs. To optimise this mutual phase-locking of different STOs in an array it is necessary to clearly understand the process of injection-locking of an STO to an external driving force. Within the project phase locking experiments of magnetic tunnel junction nanopillars to an external force were performed. Such experiments demonstrated that if the driving signal has the form of a microwave current, the locking effect is well-pronounced when the driving frequency is twice the frequency of the free running oscillator, while the locking is almost completely absent when both frequencies are equal. This result also confirmed the predictions of analytical theory. Finally, it was shown that the noise also plays an important role in the locking process. While the frequency can be locked to the external source, the phase suffers phase slips, resulting in an emission linewidth that exceeds the one of the driving signal (7).

Another approach for linewidth reduction has been investigated via numerical simulations. Here it was shown that the emission linewidth can be strongly reduced in coupled FM systems (7). For instance, the free layer can couple for specific current and field values to the polariser (made of a synthetic anti-FM layer). It was shown that for the coupled system the linewidth is strongly reduced and that this reduction results from a reduction of the amplitude phase coupling and an enhancement of the amplitude relaxation rate. These simulations show new possibilities to reduce the linewidth using novel magnetic stacks.

Contact details: ursula.ebels@cea.fr

References:
(1) J. Slonckzewski, J. Magn. Magn. Mater. 159, L1 (1996); L. Berger, Phys. Rev. B 54, 9353 (1996).
(2) S. I. Kiselev, J. C. Sankey, I. N. Krivorotov, et al Nature (London) 425, 380 (2003).
(3) A. Slavin, and V. Tiberkevich, IEEE Trans. Magn. 45, 1875 (2009).
(4) M. Quinsat, D. Gusakova, J. F. Sierra et al. App. Phys. Lett. 97, 182507 (2010).
(5) To be published
(6) J. F. Sierra, M. Quinsat, U. Ebels, et al. arXiv:1112.2833v1.
(7) M. Quinsat, J. F. Sierra, I. Firastrau et al. App. Phys. Lett. 98, 182503 (2011).
(8) D. Gusakova, M. Quinsat, J.F. Sierra et al. ibid. 99, 052501 (2011).