Cotunneling enhancement of magnetoresistance in double magnetic tunnel junctions with embedded superparamagnetic NiFe nanoparticles

Temperature and bias voltage-dependent transport characteristics are presented for double magnetic tunnel junctions (DMTJs) with self-assembled NiFe nanoparticles embedded between insulating alumina barriers. The junctions with embedded nanoparticles are compared to junctions with a single barrier of comparable size and growth conditions. The embedded particles are characterized using x-ray absorption spectroscopy, transmission electron microscopy, and magnetometry techniques, showing that they are unoxidized and remain superparamagnetic to liquid helium temperatures. The tunneling magnetoresistance (TMR) for the DMTJs is lower than the control samples, however, for the DMTJs an enhancement in TMR is seen in the Coulomb blockade region. Fitting the transport data in this region supports the theory that cotunneling is the dominant electron transport process within the Coulomb blockade region, sequential tunneling being suppressed. We therefore see an enhanced TMR attributed to the change in the tunneling process due to the interplay of the Coulomb blockade and spin-dependent tunneling through superparamagnetic nanoparticles, and develop a simple model to quantify the effect, based on the fact that our nanoparticles will appear blocked when measured on femtosecond tunneling time scales.

Figure 1

(Color online) XAS data collected at room temperature in the vicinity of the $\text{Fe}{L}_3$ and $L_2$ edges (tabulated values marked with ticks). Samples measured: $\left[\text{wafer}\left(\text{Si}/{\text{SiO}}_2\right)\right]/\text{Ru}\left(10\right)/{\text{AlO}}_x\left(1\right)/x$, where $x$ is shown as the legend. The NiFe thickness given is for nominal film growth thickness rather than particle size. The curves have been normalized and offset for clarity.

Figure 2

Cross-sectional TEM images of the full junction stack with the 0.3 nm NiFe layer grown in between the ${\text{AlO}}_x$ layers. The metallic layers are labeled, and the double barrier with nanoparticles appears as the white stripe in the middle of the image. The inset shows the same image with the contrast adjusted to make the presence of the nanoparticles in the barrier layer more clear.

Figure 3

(Color online) (a) Typical TMR curves at 3 K for SMTJ and DMTJ, taken with 10 mV applied bias. The arrows define the relative alignment the outer electrodes. (b) Current-voltage $\left(I\text{−}V\right)$ characteristics for the junctions with NiFe (0.3 nm layer) embedded. The open symbols represent the data; the solid lines represent the fits to the data using Eq. (2). Inset: comparison between the SMTJ and DMTJ at 3 K. All curves in (b) were measured at $H=+150\text{Oe}$, which corresponds to the outer electrodes being magnetized parallel, as shown in (a). The data measured at 3 K shown in panels (a) and (b) are replotted from Ref. 16.

Figure 4

(Color online) (a) Typical differential resistance vs bias voltage ($R$ vs $V$) data for the DMTJs in the antiparallel outer electrode arrangement, at various temperatures. Inset: $R$ vs $V$ for the SMTJ at two temperatures. (b) Typical $R$ vs $V$ data for the DMTJs in the parallel outer electrode arrangement, various temperatures. Inset: $R$ vs $V$ for the SMTJ at two temperatures. (c) TMR vs bias voltage and temperature, derived from the full set of $I\text{−}V$ curves. The data measured at 3 K shown in panels (b) and (c) are replotted from Ref. 16.

Figure 5

(Color online) (a) Variation in the linear $\left(A\right)$ and (b) cubic $\left(B\right)$ conductance coefficients with temperature $T$ from fits to the DMTJ $I\text{−}V$ data. P and AP denote parallel and antiparallel alignments of the outer electrodes, respectively.

Figure 6

(Color online) Schematic of two-state model of tunnel current flow through the DMTJ; the bottom outer electrodes (the DMTJ free layer) will either both be parallel or both antiparallel with respect to the top outer electrodes (the DMTJ pinned layer).