Tunnel transport through CoFe2O4 barriers investigated by conducting atomic force microscopy
Résumé
Conducting atomic force microscopy has been used to monitor the quality of spin-filtering CoFe 2 O 4 tunnel barriers by mapping current as a function of their thickness. We show that appropriate film annealing leads to a substantial improvement of their tunnelling properties. The contact force between tip and sample was identified to have a determining influence on the width of the distribution P(I) in current maps, thus precluding its reliable use to infer barrier characteristics. Therefore, assessment of tunnel transport should be done by means of the typical current which is a well defined parameter at a given contact force, rather than by the current distribution width. 2 1.Introduction Spin filters, formed by a ferromagnetic insulating layer sandwiched between two normal metal electrodes, are expected to constitute spin polarized sources of relevance in future spintronics devices [ 1 ]. Spinel ferrites such as NiFe 2 O 4 and CoFe 2 O 4 are being considered for this purpose and recently, spin filtering effects have been demonstrated using these materials [ 2-5 ]. However, as the reported spin filtering efficiency is much below expectations, questions concerning the quality/homogeneity of the spinel barriers and on their effective height arise. Characterization of tunnel devices is far from simple, as local variations of the tunnel barrier properties may produce large variations in conductance due to its extremely non-linear dependence on the barrier characteristics. Therefore the knowledge of barrier properties at submicronscale is of the highest relevance. Atomic force microscopy with a conducting tip (C-AFM) is a suitable tool to analyze electric transport across nanometric barriers. Current mapping at a given bias voltage (V) has been used to determine the dependence of the conductance of different barriers as a function of their thickness (t), eventually confirming tunnel transport across them [ 2, 3 ]. Also attempts have been made to infer barrier characteristics from current maps recorded at different voltages, e.g. the altitude of the effective tunnel barrier [ 6 ] or the nanoscopic roughness of the barriers [ 7 ]. Typically, current maps contain a distribution of current values P(I) whose width reflects the homogeneity of the barrier properties [ 7-10 ]. For instance, it has been predicted that a Gaussian distribution of thicknesses of width σ should produce a log-normal P(I) distribution whose width is directly related to the ratio σ/λ, where λ is the electronic attenuation length in the barrier [ 8 ]. This relation is remarkable as it reveals that a tunnel barrier extremely homogeneous in thickness (σ/t <>1. Therefore, the analysis of the P(I) distribution may provide a deeper insight into the properties of tunnel barriers. We strengthen that reliable analysis of I(V) characteristics, as measured by C-AFM, requires the use of a good metallic contact between the tip and the probed material, which cannot be commonly achieved unless the film surface is metal-capped [ 11 ]. In this work we present a detailed analysis of current maps of CoFe 2 O 4 (CFO) barriers as a function of thickness, growth and measuring conditions, namely the applied voltage and the contact 3 force F m between the tip and the sample. We show that a suitable annealing process allows improvement of their tunnel properties. Using such improved CFO barriers, we critically revise the analysis of the P(I) distribution as a tool to determine barrier properties and demonstrate the crucial influence of the tip-sample contact force F m not only on the absolute current scale of P(I) but also on its width. Therefore, when comparing films of different thicknesses, assessment of tunnel transport should be done by means of the typical current which is a well defined parameter at a given contact force, rather than by the current distribution width. 2. Experimental details CoFe 2 O 4 thin films of thicknesses ranging from t = 2 nm to 8 nm were grown on a SrRuO 3 -bottom electrode (25 nm) on SrTiO 3 (111) single crystalline substrates by rf-sputtering. Growth conditions and structural and magnetic characterization of the films will be reported elsewhere [ 12 ]. Two series of films, denoted " as-prepared " and " annealed " where grown under the same nominal conditions. While as-prepared films were cooled down to room-temperature after growth, annealed films were kept at growth temperature (450 ºC) in 350 mTorr oxygen for 2 h before cooling down. The C-AFM measurements have been performed using a Nanotec Cervantes microscope and Nanosensors CDT-NCHR conducting tips. All scans were done with the same lateral tip speed of about 3 µm/s under N 2 atmosphere. The SrRuO 3 electrode was positively biased (V) and the tip was grounded. The cantilever force constant was determined from the frequency response of the free oscillation [ 13 ]. For each scan we calibrated the cantilever deflection by a deflection-displacement F(z) curve and adjusted the contact force F m. The simultaneously recorded topographic images (see inset of figure 1d for a typical image) indicate a rms roughness below 0.2 nm for all films, slightly smaller than that measured in dynamic mode (σ ~ 0.3 nm), reflecting their extreme flatness (the unit cell of CoFe 2 O 4 is ~0.8 nm). 4 Figure 1: Current maps of films of as-prepared (left) and annealed (right) samples of various thicknesses measured at V = 0.6 V and 1 V respectively. The different contrast observed in each figure corresponds to the changes in current flow after consecutive current mapping scans as explained in the text. Insets: (b) difference of the current distribution maximum between central square and outer area vs. scan sequence ; (c) illustrative I-V characteristic indicating tunnel transport ; (d) typical topographic image displaying substrate induced steps. 3. Results and discussion In figure 1 we show typical current maps obtained for annealed samples (right panels, measured at 1 V and 300 nN) and for the as-prepared samples (left panels, 0.6 V and 600 nN respectively). The colour contrast denotes the changes of conductance. Using a new batch of tips for the characterization of the annealed series, the contact force F m had to be reduced to 300 nN since samples showed signs of indentation after applying 600 nN. In all images two well defined regions can be distinguished: The central square corresponds to the area that had been scanned by the tip in a preliminary scan, while the complete image is the current map measured during a subsequent larger-area scan. We ascribe the difference in conduction between the two areas to a cleaning of the sample surface by the tip during the first scan. Note that in the measurements used for current maps, there are no marks in the corresponding topography image. Further, this difference is independent of the voltage applied during the first scan (also for 0 V) and disappears after 2-3 subsequent scans (see inset in figure 1b), ruling out indentation or electronic modification of the surface as possible origin for such difference. All data 2 nm 6 nm 2.5 nm 6 nm 10 -7 A 10 -10 A 0 1 2 3 4 5 0, 0 0, 2 0, 4 ∆ Log (R) # scan -1 0 1 -100 -50 0 50 100 Current (nA) Voltage (V) 0, 5nm -0, 5nm (a) (c) (d) (b) 2 nm 6 nm 2.5 nm 6 nm 10 -7 A 10 -10 A 0 1 2 3 4 5 0, 0 0, 2 0, 4 ∆ Log (R) # scan -1 0 1 -100 -50 0 50 100 Current (nA) Voltage (V) 0, 5nm -0, 5nm (a) (c) (d) (b) 5 presented in this work were obtained from only the cleaned, central area. I-V curves recorded in a fixed point (see inset in figure 1c) show the characteristic shape of tunnel transport. Figure 2: Normalized probability distribution P(I) vs log (I) of local currents in: (a) annealed films of different thicknesses (F m = 300 nN, V = 1 V) and (b) as-prepared films (F m = 600 nN) of various thicknesses and different measuring bias voltage.
Fichier principal
PEER_stage2_10.1088%2F0022-3727%2F43%2F29%2F295001.pdf (487.45 Ko)
Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...