Abstract
The elimination of extrinsic sources of spin relaxation is key in realizing the exceptional intrinsic spin transport performance of graphene. Towards this, we study charge and spin transport in bilayer graphene-based spin valve devices fabricated in a new device architecture which allows us to make a comparative study by separately investigating the roles of substrate and polymer residues on spin relaxation. First, the comparison between spin valves fabricated on SiO2 and BN substrates suggests that substrate-related charged impurities, phonons and roughness do not limit the spin transport in current devices. Next, the observation of a 5-fold enhancement in spin relaxation time in the encapsulated device highlights the significance of polymer residues on spin relaxation. We observe a spin relaxation length of ~ 10 um in the encapsulated bilayer with a charge mobility of 24000 cm2/Vs. The carrier density dependence of spin relaxation time has two distinct regimes; n<4 x 1012 cm-2, where spin relaxation time decreases monotonically as carrier concentration increases, and n>4 x 1012 cm-2, where spin relaxation time exhibits a sudden increase. The sudden increase in the spin relaxation time with no corresponding signature in the charge transport suggests the presence of a magnetic resonance close to the charge neutrality point. We also demonstrate, for the first time, spin transport across bipolar p-n junctions in our dual-gated device architecture that fully integrates a sequence of encapsulated regions in its design. At low temperatures, strong suppression of the spin signal was observed while a transport gap was induced, which is interpreted as a novel manifestation of impedance mismatch within the spin channel.
Highlights
Enhanced spin-relaxation times have been reported for bilayer graphene-based devices compared with those based on a single layer, the relatively low spin diffusion constants overall yield a lower spin-relaxation length of only 1–2 μm,[9,11] far below the theoretical predictions.[16]
One approach suggested for achieving a longer distance spin communication is to increase the spin diffusion constants by fabricating higher mobility devices.[8,17]
The fabrication is completed by forming MgO/Co/Ti (2.2 nm/30 nm/5 nm) electrodes on top of both the BN strips and the non-encapsulated regions of the graphene strip (Figure 1e); these serve as top gate electrodes and direct-contact electrodes to bilayer graphene, respectively
Summary
Graphene is considered to be a promising spin-channel material for future spintronics applications[1] because of its high-electronic mobility,[2] weak spin–orbit coupling[3,4] and a negligible hyperfine interaction.[5,6] The initial spin transport studies were mainly performed on single-layer[7,8,9,10] and bilayer exfoliated graphene,[9,11] and large-area graphene[12,13,14,15] deposited on conventional SiO2 substrates. It has been shown that the carrier mobility of graphene devices on SiO2 is mainly limited by interfacial charged impurities, surface roughness, and phonons.[18,19,20] The demonstration of an order-of-magnitude improvement in the mobility of graphene encapsulated between atomically flat, charge trap free boron nitride crystals[21,22] has triggered the recent spin transport studies in encapsulated single layer and recently bilayer graphene-based spin valves, where a spin-relaxation length of up to ~ 12 μm and ~ 24 μm have been observed, respectively.[23,24] For the case of bilayer graphene, the initial experiments on SiO2 revealed an inverse scaling between spin and momentum relaxation times, for example, the longest spinrelaxation times were observed in the lowest-mobility devices.[9,11]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
