Analysis of Resistive Switching Characteristics in Surface Anchored Metal Organic Framework (SURMOF) Films
Device fabrication was carried out by depositing top electrodes from either Cu or Au using a commercial e-beam deposition system. Two different device stacks were fabricated and tested. The first device consisted of an Au/SURMOF/Cu/Au stack with SURMOF film thickness of around 75nm. The Cu/Au top electrode was deposited by e-beam evaporation with a thicknesses of 200nm and 25nm respectively and patterned with a shadow mask to produce disks of 200µm in diameter. The second device consisted of an Au/SURMOF/Al/Au stack with a SURMOF thickness of 10nm, 20nm and 50nm. The Al/Au top electrode was deposited by e-beam evaporation with a thicknesses of 700nm and 100nm respectively, and patterned with a shadow mask to yield disks of 150µm in diameter. A slow voltage ramp typically between -1.5V to +1.5V was used. Measurements have been performed on pristine samples directly after synthesis. Initially pristine SURMOF switching devices are in a high resistance state. In our device configuration the bottom electrode is grounded and the voltage is applied to the top electrode. In devices with Cu/Au as top electrode, ON was observed on negative voltage sweeps and OFF was observed on positive voltage sweeps. In devices with Al/Au top electrodes, ON was observed on positive voltage sweeps and OFF was observed on negative voltage sweeps. After increasing the voltage from 0 V to max. 0.75 V, a transition to a low resistance state is observed. The voltage was swept in the reverse polarity to switch from low to high resistance state. Current as high as 50mA was required to switch the device from low to high resistance state.
Measurements were also performed on the unloaded SURMOF devices. Compared to the loaded device, the switching characteristics are quite poor. The observed phenomena are crucially dependent on ferrocene loading. Ferrocene loading increased the conductivity within the MOF layer and therefore lowers the resistance. The devices could be switched using pulse width longer than 20 ns. Although the device switched within tens of nanoseconds, the time of switching could not be established due to large device capacitance values. The polarity of the SET and RESET in the devices with the Cu as the top electrode rules out the possibility of Cu ion migration being the reason for formation and dissolution of the low resistance path during switching. Additionally, devices with Al top electrode were also measured. Similar to the Cu top electrode, the Al top electrode devices also showed resistive switching.