Nonvolatile Memory Based on Polymer-Suspended Graphene Nanoplatelets with Fractional and Integer Quantum Conductance at 300K and Zero Magnetic Field

Wednesday, October 14, 2015: 16:00
Borein B (Hyatt Regency)
Y. Kang (ECE Department Virginia Tech), H. Ruan (NanoSonic, Inc), J. J. Heremans (Virginia Tech Physics Department), and M. K. Orlowski (ECE Department Virginia Tech)
Metal-insulator-metal (MIM) structures with graphene submicron-sized nanoplatelets embedded in 3-hexylthiophene (P3HT) polymer were fabricated. Such organic memory devices exhibit reliable memory operation with an ON/OFF ratio exceeding 10. Surprisingly, quantized conductance is observed at zero magnetic field and 300K, in most devices at integer multiples of Go=2e2/h(=12.9 kΩ)-1, and in some devices also at fractional-quantized with v=m/7 filling factor.  For memory applications, the quantum conductance is attractive as no heat is dissipated in the cells proper even at 300K. At high GNP concentrations the MIM cell behaves as an excellent conductor, at low GNP concentrations as a stable insulator, and at medium concentrations displays a significant memory hysteresis effect. Graphene-based MIM devices were fabricated on thermally-oxidized silicon wafers using standard semiconductor processes.  Bottom Au and top Cu electrodes have been deposited using electron-beam physical vapor deposition. The graphene nanoplatelet (GNP) powder was dispersed into the toluene and further exfoliated by ultrasonication. GNP platelets comprise one or more layers of one-atom-thick planar sheets (5-30) of sp2-bonded carbon atoms densely packed in a honeycomb crystal lattice. The lateral dimension of the flakes ranges 100 nm-10 mm. Ultrasonication is effective method in overcoming the van der Waals forces between the individual carbon sheets leading to uniform dispersions of single flakes or very thin stacks of flakes. The resulting P3HT(GNP) dispersion was drop-deposited onto the bottom Au electrode and covered by islands of Cu electrodes.  The graphene concentrations in the P3HT(GNP) solutions vary from 0.05mg/ml to 0.2mg/ml. The P3HT solution is 10 ml. The thicknesses of the layers are 60 nm, 700 nm, 150 nm, for Au, P3HT(GNP), and Cu, respectively. For each concentration, ten devices with the same structure and the same active area were fabricated. For comparison and assessment of the role of carbon nanoplatelets, six Au/P3HT/Cu structures have been fabricated with no graphene nanoplatelets. All of the latter devices show neither integer nor fractional quantum conductance effects. Four types of GNP polymer samples have been manufactured: i) no GNP, ii) 0.05 mg/ml, iii) 0.1 mg/ml, and iv) 0.2 mg/ml of GNP concentration. From these four types of devices, only devices with 0.1 mg/ml show integer and fractional quantum conductance. Roughly 25% of the 0.1 mg/ml devices display integer quantum conductance, 25% fractional quantum conductance, and remaining 50% no quantum conductance. For electrical characterization the bias voltage is first swept along the positive x-axis from 0V to 2V. The current is observed to increase ohmically in the low-voltage region, and then abruptly to increase at three voltages, specifically at 0.75V, 1.2V and 1.87V, to 10mA, 162mA and 409mA. The initially high-resistance state of about 100kΩ is reduced during the voltage sweep to a final low-resistance state, about 8kΩ.  When the voltage is swept back from 2V to 0V, a clear hysteresis in the I-V characteristics is observed lending itself to memory applications. The extracted conductance is quantized in multiples of Go. This behavior has been observed repeatedly on several devices and for several voltage sweeps on the same device. The step-like conductance is sloped by the thermal broadening owing to ambient temperature, T=300K. This behavior is modeled consistently with one single suspended graphene ribbon bridging the gap between the Cu and Au electrodes. The conductance changes in quantum units of Go (Go=2e2/h=77.5µS), from 1Go to 2Go and from 2Go to 3Go. In addition, a sequence of smaller conductance steps within the 1Go interval is observed indicating that the fractional quantum conductance coexists with integer quantum conductance. The best fit to the fractional quantum conductance within the first Go step can be obtained by assuming a filling factor of v=1/7, 2/7, 3/7, 4/7, 5/7, and 6/7, although 3/7 step is slightly off the experimental value. The resolution of the fractional conductance between 1Go and 2Go is always much less pronounced and appears to follow the same filling factor of multiples of 1/7. Steps with 8/7, 9/7, 11/7, and 13/7 can be clearly identified. In the third conductance step unit from 2Go to 3Go only a rudimentary fractional quantum conductance can be discerned. This behavior has been observed for several devices and can be repeated on the same device with a slightly changed pattern and length of the steps. The highly conductive devices with GNP concentration of 0.2 mg/ml do not display any quantum conductance effects and are a good material candidate for stretchable transparent electrodes. The broad range of electric tunability of GNP films points to the universal potential of GNP depending on its concentration in a polymer for a wide spectrum of applications.