Final Activity Report Summary - ISIS (Injecting spins into silicon)
Spin electronics or spintronics is at the interface between magnetism and electronics and concerns mainly fully metallic magnetic nanostructures. A challenging step in spintronics evolution consists in combining magnetic materials and semiconductor in the same "hybrid" structure. Storage, detection, logic and communication capabilities would then be combined on a single chip; such multifunctional devices would replace several components with evident gain of size and speed.
In order to develop these "hybrid" devices, spin polarised current injection into a non-magnetic semiconductor has to be achieved. In order to achieve room temperature operation, the injection electrode must be a ferromagnetic metal. However, it was demonstrated both experimentally and theoretically that spin injection was impossible through a ferromagnet/semiconductor ohmic contact, due to the so-called "conductivity mismatch" between the two materials. This problem can be solved by introducing a Schottky barrier or a tunnel junction at the interface between the ferromagnetic metal and the semiconductor.
In order to pave the way towards hybrid devices fully compatible with CMOS technology and easily implemented in an industrial process, we choose to use silicon as the semiconductor channel. A significant advantage of silicon is its long spin life-time due to weak spin-orbit coupling: the spin diffusion length at room temperature is of the order of a few microns, allowing operation in micron-sized devices.
Studying spin injection into silicon is therefore an ambitious project that involves microfabrication of test devices, electrical characterisations, physical characterisation and magneto-transport measurements.
The fabrication of ferromagnet/insulator/semiconductor (F/I/S) injector and collector junctions was realised on silicon wafers with a tunnel barrier similar to those used for magnetic tunnel junctions. The oxide layer was deposited through very well controlled and optimised growth procedures on a clean silicon surface. Different test-structures were realised in order to test different ferromagnetic metals (such as Co, Ni, NiFe) and tunnel barriers (SiO2, Al2O3, MgO). The objective was to achieve the combination of the most compatible materials with high-quality interfaces.
F/I/S diodes quality was tested by electrical characterisation (current-voltage and capacitance-voltage measurements) in order to address the question of interface charge density at the silicon/dielectric interface. Since silicon contamination by transition metals might lead to spin depolarisation, it is essential to find a structure free of this type of defects. Two different structures were thoroughly studied: NiFe/SiO2/Si and NiFe/Al2O3/Si. The major result of this work was the evidence of a correlation between the electric defects density at the silicon/insulator interface, obtained by electrical spectroscopy technique, and the silicon surface contamination by the transition metals as revealed by Time Of Flight-SIMS analysis. More precisely, this correlation have shown that the unusually high defects density observed at the SiO2/Si interface is related to enhanced metal diffusion through the SiO2 tunnel barrier. This leads to high contamination level in the silicon substrate. In contrast, TOF-SIMS analysis on NiFe/Al2O3/Si evidenced diffusion barrier behaviour of Al2O3 which turned out to be consistent with the good electrical properties of the Al2O3/Si interface. These results suggest that NiFe/Al2O3/Si tunnel diodes are well adapted for spin injection into silicon.
In order to develop these "hybrid" devices, spin polarised current injection into a non-magnetic semiconductor has to be achieved. In order to achieve room temperature operation, the injection electrode must be a ferromagnetic metal. However, it was demonstrated both experimentally and theoretically that spin injection was impossible through a ferromagnet/semiconductor ohmic contact, due to the so-called "conductivity mismatch" between the two materials. This problem can be solved by introducing a Schottky barrier or a tunnel junction at the interface between the ferromagnetic metal and the semiconductor.
In order to pave the way towards hybrid devices fully compatible with CMOS technology and easily implemented in an industrial process, we choose to use silicon as the semiconductor channel. A significant advantage of silicon is its long spin life-time due to weak spin-orbit coupling: the spin diffusion length at room temperature is of the order of a few microns, allowing operation in micron-sized devices.
Studying spin injection into silicon is therefore an ambitious project that involves microfabrication of test devices, electrical characterisations, physical characterisation and magneto-transport measurements.
The fabrication of ferromagnet/insulator/semiconductor (F/I/S) injector and collector junctions was realised on silicon wafers with a tunnel barrier similar to those used for magnetic tunnel junctions. The oxide layer was deposited through very well controlled and optimised growth procedures on a clean silicon surface. Different test-structures were realised in order to test different ferromagnetic metals (such as Co, Ni, NiFe) and tunnel barriers (SiO2, Al2O3, MgO). The objective was to achieve the combination of the most compatible materials with high-quality interfaces.
F/I/S diodes quality was tested by electrical characterisation (current-voltage and capacitance-voltage measurements) in order to address the question of interface charge density at the silicon/dielectric interface. Since silicon contamination by transition metals might lead to spin depolarisation, it is essential to find a structure free of this type of defects. Two different structures were thoroughly studied: NiFe/SiO2/Si and NiFe/Al2O3/Si. The major result of this work was the evidence of a correlation between the electric defects density at the silicon/insulator interface, obtained by electrical spectroscopy technique, and the silicon surface contamination by the transition metals as revealed by Time Of Flight-SIMS analysis. More precisely, this correlation have shown that the unusually high defects density observed at the SiO2/Si interface is related to enhanced metal diffusion through the SiO2 tunnel barrier. This leads to high contamination level in the silicon substrate. In contrast, TOF-SIMS analysis on NiFe/Al2O3/Si evidenced diffusion barrier behaviour of Al2O3 which turned out to be consistent with the good electrical properties of the Al2O3/Si interface. These results suggest that NiFe/Al2O3/Si tunnel diodes are well adapted for spin injection into silicon.