Where graphene addresses magnonics memory

Electronic and spin transport of information has been extensively investigated in various materials, such as metals, magnetic insulators, and semimetallic systems, for storage, spin electronic devices, and memory applications [1]. At present, flash memory has been used in mobile application for its nonvolatility, but it suffers from slow write speed and low endurance. On the contrary, spin-transfer torque magnetic random access memory (STT-MRAM) is gaining popularity as a future nonvolatile memory because of its high speed and endurance. Recently, Everspin Technologies has started distributing a 64-Mbit DDR3 memory chip. In STT-MRAM, the information stored in the magnetic nanostructures is written and read electrically, which allows fast access, and is nonvolatile because of the high anisotropy energy of the magnetic nanostructures. However, there are important challenges associated with STT-MRAM in materials engineering, device optimization, and device performance, which should be addressed in order to use it for commercial purposes [2].

In a parallel world of research, quantum computing – which does not rely only on 1's and 0's, but on their superpositions – is being considered for next-generation computing. Recently, magnonic holographic memory (MHM) has been proposed as a possible candidate to produce high-speed quantum computing devices utilizing spin-wave interference for data processing in a medium with low damping constant, such as magnetic insulator yttrium iron garnet (YIG). In MHM devices, spin waves can transport information via spin angular momentum through the YIG over a wide range of distances from 40 μm to millimeters. In principle, MHM devices may be advantageous over optics-based holography, because of shorter wavelength and lower power consumption.

The main challenge of MHM development is the scaling of the operational wavelength, which requires the development of submicrometer-scale elements for spin-wave generation and detection. Moreover, the compatibility of MHM based on YIG with complementary metal-oxide semiconductor (CMOS) integration is still a limitation to make commercial MHM devices. Therefore, it is essential to develop new memory devices that combine important factors such as cost-effectiveness, nonvolatility, high data density, compatibility with CMOS technology, and fast access [3]. Graphene is being considered as the most promising medium for long-range spin-coherence propagating across graphene and compatible with CMOS, nanoelectronic, and spintronic devices. Its unique transport properties such as long spin-coherence lengths (100 μm), because of the absence of hyperfine interactions and weak spin–orbit coupling, enable it to play a fundamental role in the development of new technologies [4][5][6] and [7].

Pure spin transport in graphene and multilayer graphene on SiO2 substrate at room temperature has been previously shown [8] and [9]. Moreover, pure spin transport/spin wave and precession over long channel lengths extending up to 16 μm with a spin lifetime of 1.2 ns and spin diffusion length of 6 μm at room temperature have also been illustrated. These spin parameters are up to six times better than previously reported values and are the highest at room temperature for any form of pristine graphene from industry [7]. Furthermore, the possibility of inducing ferromagnetism in graphene with a large exchange interaction has been shown by proximity effect or doping graphene band gaps with magnetic materials to enhance spin–orbit coupling. These techniques can lead to spin-wave and spin-current transport phenomena such as the spin Hall effect (SHE), inverse spin Hall effect (ISHE), and anomalous Hall effect (AHE) for spintronics [10].

On the basis of the aforementioned advantages of graphene and its compatibility for CMOS, we propose a new class of memory that utilizes a combination of magnonics, graphene, and spin-wave propagation, which we call “Holographonics”. The schematic of the device is shown in the figure.

A novel generation of magnonic memory based on graphene.
A novel generation of magnonic memory based on graphene.

In this proposed configuration, a network of ferromagnetic graphene with tunable band gaps acts as a spin-wave bus or spin-current transport medium. The generating/detecting signal elements are placed at the edges of the network. The graphene network consists of junctions, each of which has a hologram element placed on its top. The hologram elements can be made of magnetic or nonmagnetic materials. The read-in and read-out operations of the holographonics device are performed via spin-wave interferences.

Heavy metals with strong spin–orbit coupling, such as Pt or Ta layers [11], micro-antennas [12], spin torque oscillators (STO) [13], spin Hall nano-oscillators (SHNO)[14] and [15], and multiferroic elements [16], may also be considered for converting the input electric signals into spin waves and the output spin waves into electrical signals. Spin waves generated by the edge elements are used for information read-in and read-out. The difference between the read-in and read-out modes of operation is in the amplitude of the generated spin waves. In the read-in mode, the elements generate spin waves of a large amplitude. In the read-out mode, the amplitude of the generated spin waves is much lower than the threshold value required to overcome the energy barrier between the states of magnetic hologram elements [3]. These elements act as memory bits containing information encoded in the different states. In graphene network junctions, the hologram elements can also manipulate the phase of hologram patterns based on element properties and spin-wave interferences for desired pattern recognitions. In summary, holographonics enables a new horizon for building scalable holographic devices with high memory density, compatible with CMOS, and lower power consumption and high speed.


We gratefully acknowledge Nanyang Technological University, Start-Up Grant for the funding of this research and E. Ebrahimshah for her support in designing the illustration.

Further reading

[1] L.J. Cornelissen, et al.
Nat. Phys., 11 (2015), p. 1022

[2] A.D. Kent, D.C. Worledge
Nat. Nanotechnol., 10 (2015), p. 187

[3] A. Khitun
J. Appl. Phys., 113 (2013), p. 164503

[4] M. Fujita, et al.
J. Phys. Soc. Jpn., 65 (1996), p. 1920

[5] S. Okada, A. Oshiyama
Phys. Rev. Lett., 87 (2001), p. 146803

[6] F. Muñoz-Rojas, J. Fernández-Rossier, J.J. Palacios
Phys. Rev. Lett., 102 (2009), p. 136810

[7] M.V. Kamalakar, et al.
Nat. Commun., 6 (2015), p. 6766 http://dx.doi.org/10.1038/ncomms7766

[8] N. Tombros, et al.
Nature, 448 (2007), p. 571

[9] T. Maassen, et al.
Phys. Rev. B, 83 (2011), p. 115410

[10] Z. Wang, et al.
Phys. Rev. Lett., 114 (2015), p. 016603

[11] L. Liu, et al.
Phys. Rev. Lett., 106 (2011), p. 036601

[12] M. Covington, T.M. Crawford, G.J. Parker
Phys. Rev. Lett., 89 (23) (2002), p. 237201

[13] S.M. Mohseni, et al.
Science, 339 (2013), p. 1295

[14] V.E. Demidov, et al.
Nat. Mater., 11 (2012), p. 1028

[15] M. Ranjbar, et al.
IEEE Magn. Lett., 5 (2014), p. 3000504

[16] A. Khitun, M. Bao, K.L. Wang
IEEE Trans. Magn., 44 (9) (2008), p. 2141

Read full text on ScienceDirect

DOI: 10.1016/j.mattod.2016.07.006