Interplay of bulk and interface effects in the electric-field-driven transition in magnetite

Contact effects in devices incorporating strongly correlated electronic materials are comparatively unexplored. We have investigated the electrically driven phase transition in magnetite (100) thin films by four-terminal methods. In the lateral configuration, the channel length is less than $2\mu \text{m}$, and voltage-probe wires $\sim 100\text{nm}$ in width are directly patterned within the channel. Multilead measurements quantitatively separate the contributions of each electrode interface and the magnetite channel. We demonstrate that on the onset of the transition contact resistances at both source and drain electrodes and the resistance of magnetite channel decrease abruptly. Temperature-dependent electrical measurements below the Verwey temperature indicate thermally activated transport over the charge gap. The behavior of the magnetite system at a transition point is consistent with a theoretically predicted transition mechanism of charge gap closure by electric field.

Figure 1

(Color online) (a) SEM image of the device for four-probe measurements showing source and drain leads and two pairs of voltage probes within the channel. (b) Colored SEM image demonstrating electrical contacts to micron-sized Au pads with further In soldering to attach Au wires. (c) Schematics of electrical circuit of four-probe measurements. Letters S and D denote source and drain contact, respectively. Contacts are made of 6 nm Cu adhesion layer (reddish, below) and 10–20 nm cover layer of Au (yellow, above). (d) Temperature dependence of the low-bias resistance of magnetite channel $\left(R_{\text{DEV}}\right)$ and corresponding contact resistances $\left(R_{\text{C}}\right)$ at source and drain electrodes.

Figure 2

(Color) (a) Examples of two-terminal resistance dependence on the channel length at three different temperatures (85, 90, and 95 K) with corresponding linear fits. Extrapolation to zero channel length gives the total contact resistance, $R_{\text{C}}\left(\text{total}\right)$, at each temperature. (b) and (c) show SEM images of devices with different channel lengths. (d) and (e) plot calculated values of device resistances, $R_{\text{DEV}}$, and contact resistances, $R_{\text{C}}\left(\text{source}\right)$ and $R_{\text{C}}\left(\text{drain}\right)$, around zero source voltage for devices in (b) and (c), respectively.

Figure 3

(Color online) Examples of $I\text{−}V$ and corresponding $I\text{−}\Delta V$ curves at 80 and 85 K at the voltage ranges above switching voltage at each temperature. Arrows indicate the direction of the voltage sweep. The pulse sweep parameters are those to demonstrate the small hysteresis in forward and reverse voltage sweeps. Top inset shows $I\text{−}\Delta V$ curves at the same temperatures but at the voltage range below switching voltage. Bottom inset zooms in to $I\text{−}\Delta V$ curve at 80 K around transition point to demonstrate the discontinuity in measured $\Delta V$ value.

Figure 4

(Color online) (a) Schematic diagram of voltage distribution along the channel at a transition point (blue open squares) and right after transition (red closed squares). $V_{\text{C}}$ denotes the voltage drop at the interfaces. (b) A fragment of $I\text{−}V$ curve at 85 K in the vicinity of the transition. Blue open square marks the transition point and red closed square shows a point right after transition; blue and red squares in (a) correspond to the voltage distribution over the channel at these points. (c) and (d) are the voltage dependences of $R_{\text{C}}\left(\text{source}\right)$, $R_{\text{C}}\left(\text{drain}\right)$, and $R_{\text{DEV}}$, respectively, demonstrating the abrupt decreases (jumps) in all three resistances at the transition point.

Figure 5

(Color online) Temperature dependences of (a) the jumps in $R_{\text{C}}\left(\text{source}\right)$, $R_{\text{C}}\left(\text{drain}\right)$, and $R_{\text{DEV}}$ at a transition point; each point represents an average over eight independent measurements (left/right voltage pairs, different grounds, positive/negative switching voltages), standard deviation is within the symbol size (b) jumps in $\Delta V$ (squares) at a transition point and its exponential fit (solid line) (c) $V_{SW}$ and $\Delta V_{SW}$ with corresponding exponential fits and (d) conductance $\left(1/R_{\text{DEV}}\right)$ of magnetite channel in Arrhenius coordinates and its linear fit.