Normal and inverse spin-valve effect in organic semiconductor nanowires and the background monotonic magnetoresistance

We have observed both peaks and troughs in the magnetoresistance of organic nanowires consisting of three layers—cobalt, 8-hydroxy-quinolinolato aluminum $\left(\mathrm{Al}{\mathrm{q}}_3\right)$, and nickel. They always occur between the coercive fields of the ferromagnetic layers, and we attribute them to the normal and inverse spin-valve effect. The latter is caused by resonant tunneling through localized impurity states in the organic material. Peaks are always found to be accompanied by a positive monotonic background magnetoresistance, while troughs are accompanied by a negative monotonic background magnetoresistance. This curious correlation suggests that the background magnetoresistance, whose origin has hitherto remained unexplained, is probably caused by the recently proposed phenomenon of magnetic-field-induced enhancement of spin-flip scattering in the presence of spin-orbit interaction [Cahay and Bandyopadhyay, Phys. Rev. B 69, 045303 (2004)].

Figure 1

(Color online) Atomic force microscopy (AFM) image of the top surface of the alumina template formed by anodization using 3% oxalic acid at $40\mathrm{V}$ dc. Pore diameter $\sim 50\mathrm{nm}$.

Figure 2

(Color online) AFM image of the bottom surface of the alumina template after removing bulk alumina. We observe ‘‘through pores’’ as a result of reverse polarity etching.

Figure 3

Transmission electron micrograph of a single nanowire showing the $\mathrm{Al}{\mathrm{q}}_3$ layer sandwiched between the cobalt and nickel electrodes. This image was produced by releasing the nanowires from the alumina host by dissolution of alumina in dilute phosphoric acid and capturing the nanowires on TEM grids for imaging.

Figure 4

(Color online) Schematic representation of the nanowire organic spin-valve device. The nanowires are hosted in an insulating porous alumina matrix and are electrically accessed from each end. The magnetic field is applied along the axis of the wire.

Figure 5

(Color online) Magnetoresistance trace of the control sample, consisting of $\sim 500$ Ni-Co bilayered nanowires (no $\mathrm{Al}{\mathrm{q}}_3$).

Figure 6

(Color online) Histogram showing the distribution of spin-valve signal strength collected over 90 samples. Some of these samples show a positive spin-valve signal and the rest a negative spin-valve signal. All of these data were collected at a bias current of $10\mu \mathrm{A}$.

Figure 7

(Color online) Inverse spin-valve effect and background negative magnetoresistance in $\mathrm{Ni}\text{−}\mathrm{Al}{\mathrm{q}}_3\text{−}\mathrm{Co}$ nanowires at four different temperatures and fixed bias $\left(10\mu \mathrm{A}\right)$.

Figure 8

(Color online) Inverse spin-valve effect and background negative magnetoresistance in $\mathrm{Ni}\text{−}\mathrm{Al}{\mathrm{q}}_3\text{−}\mathrm{Co}$ nanowires at four different bias values and fixed temperature $\left(1.9\mathrm{K}\right)$.

Figure 9

(Color online) Explanation of the relation between the sign of the background magnetoresistance and the sign of the spin-valve peak. (a) Consider the case when $H>\mid H_{\mathrm{Co}}\mid >\mid H_{\mathrm{Ni}}\mid$ and inversion of injected spin polarization has taken place due to resonant tunneling through an impurity somewhere in the channel. We assume that the impurity is closer to the Ni contact so that the spin polarization of the Ni contact has been effectively reversed as shown in the figure. Owing to the high magnetic field, the spin relaxation rate is high (Ref. 23) and the injected spins are completely depolarized by the time they reach the $\mathrm{Co}∕\mathrm{Al}{\mathrm{q}}_3$ interface. Therefore, on the average, 50% of the spins have their polarizations aligned along the magnetization of the Co contact and are transmitted by the Co contact. The device $\text{resistance}=R_1$ (say). (b) When the magnetic field is decreased so that $H=0+$, the spin relaxation rate falls. Only partial depolarization of the spins occurs as the carriers traverse the channel, so that fewer than 50% of the spins have their polarizations aligned along the magnetization of the Co contact and are transmitted. In this case, the device resistance is $R_2$ which is larger than $R_1$. This explains why the background magnetoresistance is negative whenever there is resonant inversion and a resulting negative spin-valve peak. (c) Again, consider the case when $H>\mid H_{\mathrm{Co}}\mid >\mid H_{\mathrm{Ni}}\mid$ but no resonant inversion of injected spin polarization takes place. The high magnetic field completely depolarizes the injected spins by the time they reach the Co contact (owing to the high spin-flip rate) and again 50% of the carriers are transmitted. The device resistance is $R_1^{\prime }$. (d) When the field is reduced to $H=0+$, only partial depolarization takes place and more than 50% of the spins are aligned along the Co contact’s magnetization. Therefore more than 50% of the spins are transmitted and the device resistance is smaller than $R_1^{\prime }$. This explains why the background magnetoresistance is positive whenever there is no resonant inversion so that the spin valve peak is positive. Note that this physics is somewhat counterintuitive. At high magnetic fields, one would expect that the spins would remain aligned along the field and more of them will transmit through the Co contact. Just the opposite happens because the magnetic field increases the spin-flip rate and contributes to depolarization.