Spin transport in benzofurane bithiophene based organic spin valves

In this paper we present spin transport in organic spin-valves using benzofurane bithiophene (BF3) as spacer layer between NiFe and Co ferromagnetic electrodes. The use of an AlO x buffer layer between the top electrode and the organic layer is discussed in terms of improvements of stacking topology, electrical transport and oxygen contamination of the BF3 layer. A study of magnetic hysteresis cycles evidences spin-valve behaviour. Transport properties are indicative of unshorted devices with non-linear I-V characteristics. Finally we report a magnetoresistance of 3% at 40 K and 10 mV in a sample with a 50 nm thick spacer layer, using an AlO x buffer layer. C (cid:2) 2014 Author(s). All article content,

In this paper we present spin transport in organic spin-valves using benzofurane bithiophene (BF3) as spacer layer between NiFe and Co ferromagnetic electrodes. The use of an AlO x buffer layer between the top electrode and the organic layer is discussed in terms of improvements of stacking topology, electrical transport and oxygen contamination of the BF3 layer. A study of magnetic hysteresis cycles evidences spin-valve behaviour. Transport properties are indicative of unshorted devices with non-linear I-V characteristics. Finally we report a magnetoresistance of 3% at 40 K and 10 mV in a sample with a 50 nm thick spacer layer, using an AlO x buffer layer. C  Organic semiconductors (OSC) have drawn increasing interest in spintronics since the first evidence of magnetoresistance in sexithiophene (6T) 1 and tris(8-hydroxyquinoline) aluminum (Alq 3 ) 2 based spin-valves. The choice of an organic semiconductor instead of an inorganic one is driven by the amount of interesting properties OSCs feature: potentially long spin relaxation time, 3 chemical tuning of electronic functionality, easy structural modification, possibility of self-assembly and mechanical flexibility. Though magnetoresistance (MR) seems to be obtained in most organic spinvalves (OSV), injection and transport of spin polarized current in OSC, and the role played by the interfaces between the organic layer and the ferromagnetic (FM) electrodes are still matters of discussion. 4,5 As reviewed recently, 6 most organic spintronics devices are based on Alq 3 7-10 as transport layer for spin-polarized charge carrier although many other π -conjugated systems have been synthesized by chemists. Among the results obtained with Alq 3 , a large dispersion of MR values, from -300% to +8%, has to be explained. 2,8,9,11 Alq 3 is well known as an electron transport OSC with low electron mobility, about 10 −6 cm 2 V −1 s −1 and also lower hole mobility 12  levels can involve an important energy injection barrier for electrons ( Fig. 1(a)). In this figure, the displayed experimental highest occupied molecular orbital (HOMO) and LUMO values for Alq 3 were determined by Hill et al. 13 by ultraviolet phoelectron spectroscopy (UPS) and inverse photoelectron spectroscopy (IPES) and give transport gap larger than the usually encountered optical gap obtained by absorption measurements. Although it has been pointed out that interface dipoles lower the potential barrier height of the Alq 3 -FM electrodes junction, 14 the question of holes instead of electrons injection and transport in Alq 3 based OSVs has been raised. 15 In this paper we present investigations of OSV based on benzofurane bithiophene (BF3) sandwiched between NiFe and Co electrodes. The chemical structure of BF3 is shown in inset of figure 1. BF3 is an ideal choice, among OSCs since it exhibits high hole mobility in air at room temperature, 16 up to 9.3 10 −3 cm 2 V −1 s −1 , and good thermal stability with a decomposition temperature of 293 • C. The HOMO and LUMO energy levels have been calculated by DFT methods and are found to be 5.3 eV and 2.3 eV respectively 16 (UPS and IPES data not available for BF3 at present time). This result shows that BF3 HOMO matches quite well with energy Fermi levels of the electrodes chosen here (Fig. 1).
The devices typical structure consists of glass/NiFe/BF3(/AlO x )/Co, patterned for a current perpendicular to plane (CPP) geometry with or without optional AlO x buffer. The upper Co and bottom NiFe FM electrodes are prepared by dc sputtering with 5 × 10 −3 mB Ar pressure through appropriate contact masks, resulting in devices with junction areas of 200 × 200 μm 2 . BF3 films are grown by Joule evaporation in a high vacuum (10 −7 mB) chamber connected to the sputter chamber. Masks are changed in situ using a micromanipulator and a homemade sample holder, preventing the sample from atmospheric exposure during mask changing. As direct Co sputtering is harmful for the organic layer, we add a 2 nm thick AlO x protective buffer layer between the upper electrode and the organic film, to prevent Co atoms to penetrate into the organic film. 8,10,17 Note that this aluminum oxide layer is not insulating enough for blocking carrier injection in the OSC.
This oxide buffer layer is obtained by O 2 exposure (30 mB) of two consecutive 1 nm sputtered Al films. Sample topological properties are studied on each successively grown layer by atomic force microscopy (AFM) using a Dimension 3100 Nanoscope AFM in tapping mode. For a more representative area investigation of our samples junctions, complementary study is required with Veeco optical profilometer (124 × 94 μm 2 ). Magnetic properties are investigated with a commercial Quantum Design Physical Properties Measurement System (PPMS) with a VSM module (Vibrating Sample Magnetometer). Transport and magnetotransport measurements are performed using a Keithley 2400 sourcemeter in 2 probes mode connected to the PPMS for temperature control, from 300 to 5 K. Magnetic field is applied within ± 10 kOe range in the plane of the substrate.
First, AFM and optical profilometer studies are done on samples from the same batch of NiFe/BF3 (50 nm) with or without an AlO x barrier (Fig. 2) for comparison. They reveal a low roughness of the bottom NiFe electrode (rms=1 nm). The observed organic layer rms roughness is 5 nm for both samples, with or without an AlO x barrier. In samples including AlO x , this buffer layer grows conformal to the OSC, and non significant roughness changes are observed after Co deposition. Roughness is more important in samples without an AlO x barrier, and a blistered surface evidences the damage caused to the organic layer during Co sputtering (see Fig. 2(b)). All samples without an AlO x barrier are also found to be electrically shorted. These results are proofs that the AlO x barrier we added provides an efficient protection for the organic layer. The achievement of smooth morphological properties allows to investigate magnetic properties of full plane stacking on Si substrates.
VSM measurements between ± 1 kOe reveal uncoupled magnetic electrodes, a critical property for achieving spin valve behaviour and magnetotransport (Fig. 3). The influence of temperature on the switching fields of the Co layer is also to be noted. We observe an increase from 20 Oe and -20 Oe at 300 K to 200 Oe and -750 Oe at 10 K. The asymmetrical dependence of the increase of the switching field when temperature decreases can be explained by the formation of a CoO x antiferromagnetic layer on top of the Co layer due to air exposure of the top layer of the device after sample preparation. This leads to exchange coupling between ferromagnetic Co and antiferromagnetic CoO x of the hysteresis cycle of the Co layer after a positive field cooling. Nevertheless, this phenomenon does not constitute a spin valve behaviour issue.
Transport measurements between ± 10 V show non-linear I-V characteristics that obey a power law (inset Fig. 4). Data analysis allows us to exclude injection-limited device operation in favour of a mechanism governed by transport. Temperature dependence is also to be noted: the hybrid junction resistance displays a ten-fold increase between room temperature and 10 K, indicative of a semiconductor behaviour, and excluding tunnel MR in our samples. 18  the organic layer. 20 Further experiments are in progress to analyze temperature dependence of the resistance at higher temperatures. Magnetotransport is then studied on a sample of glass/NiFe/BF3 (50 nm)/AlO x /Co, in the range of ± 5 kOe, applying bias voltage from 70 mV to -70 mV. It reveals a MR of 2-3 % at 40 K using a bias voltage of 10 mV on the Co top electrode (Fig. 4). For T larger than 40 K MR is not detected in our sample. We attribute that to the existence of other transport mechanisms based on hoping that increase spin relaxation and induce a loss of spin-polarized transport in our sample. At lower temperature, our device became too resistive (R > 20 MOhm) and signal was too noisy for proper MR measurement. A MR decrease is noted when increasing voltage and MR gets down to low values at 70 mV (Fig. 4). This behaviour is symmetrical, applying bias voltage either on the NiFe or Co electrode. Our VRH modeling of R(T) characteristics show a neat voltage dependence of the transport mechanism. This is indicative of the creation of new current channels in the BF3 layer with multistep tunneling that will have an effect on the loss of spin polarization of the current and consequently on the observed voltage decay of the MR. We also observe ≈ 1 % MR for another sample with 110 nm BF3 layer. We notice that all samples display poor time stability. This phenomenon can be attributed to the oxidative degradation of the OSC during the fabrication of the AlO x barrier. In order to clarify this point, low temperature photoluminescence (low-T PL) experiments are performed on pristine BF3 films with and without oxygen exposure (two steps 25 min controlled oxidation, at room temperature and O 2 pressure = 30 mB, similar to AlO x buffer layer preparation). Indeed, it has been shown that low-T PL is very sensitive to OSC oxidation damage. 21 PL is excited by a 40 mW laser diode at λ ex = 405 nm (peak of BF3 absorption). Spectra were analyzed by a 1 m focal length grating monochromator and detected by a R636 Hamamatsu photomultiplier and standard lock-in techniques. The sample was cooled at 12 K using a closed cycle He cryostat. As shown in Fig. 6, slight peak shift (under 0,05 eV) suggests OSC layers inhomogeneities, but no effect of  O 2 exposure is observed on PL spectra. This is evidence of no oxidative defects formation in BF3 thin films. Consequently, we assume that, either thin Al layer provides efficient protection against oxidation, or the BF3 molecules are very stable when exposed to O 2 (as air mobility measurements tend to show 16 ). So we conclude that our devices time instability is not due to oxygen contamination of BF3 but rather to OSC layer inhomogeneities or metallic contamination from FM electrodes. STEM-EELS (Scanning Transmission Electron Microscopy-Electron Energy Loss Spectroscopy) experiments are in progress in order to confirm this issue.
In summary, we have prepared organic spin valves with BF3 spacer layer and we observed up to 3% MR in samples with 50 nm BF3 thickness topped with 2 nm AlO x . In this case, aluminium oxide layer is an efficient protection of BF3 prior to the deposition of the top cobalt film of the structure. Our results demonstrate the stability of BF3 to oxygen, its ability as a spin transport material with a relatively large spin diffusion length since we observed MR up to 110 nm organic layer.
This work was partly supported by the French RENATECH network.