Integration of ferrimagnetic CoFe2O4 epitaxial films with silicon

Abstract

Microelectronics is progressing continuously by the exponential growth with time of the number of transistors per integrated circuit, the popularly known as “Moore’s Law”. This law is still valid but it is approaching intrinsic limits. The “More than Moore” is a complementary approach based on the use of radically new concepts as well as on the use of new materials in existing devices to improve performance. In particular, functional complex oxides represent an opportunity to extend and develop new devices functionalities with a wide range of applications. This thesis presents a study on the integration of CoFe2O4 thin films with silicon. CoFe2O4 is ferromagnetic and electrically insulating at room temperature, the properties required to be used as tunnel barrier in a spin filter device. This device could permit the injection of spin polarized currents in silicon, as an alternative to the injection using ferromagnetic electrodes and passive tunnel barriers. However the spin filter requires a nanometric CoFe2O4 film, thinner than 4-5 nm to allow tunneling, and has to be epitaxial with high crystalline quality to preserve the ferromagnetism and tunneling transport. The thermodynamical instability between CoFe2O4 and silicon imposes the use of a buffer layer for its epitaxial integration. The challenging goal is therefore fabricating ultrathin epitaxial CoFe2O4/buffer bilayers on silicon. Investigating the possibility to achieve such goal has been the main objective of this thesis. The buffer layer is critical. Thus we have followed a strategy based on investigating in parallel several candidates. SrTiO3, which can be grown epitaxially on Si(001) with sharp interface and that has been already used as single crystal to deposited CoFe2O4, has been a natural option. We have used thick (around 17 nm) SrTiO3 buffers fabricated by collaborators at INL-Lyon to grow by pulsed laser deposition (PLD) CoFe2O4, which is epitaxial and ferromagnetic. However, there is diffusion of Ti into CoFe2O4 and the SrTiO3/Si(001) interface could be unstable. Yttria-stabilized-zirconia (YSZ) has been other investigated material. It is widely used to grow oxides on Si(001), but having the YSZ buffershigh thickness of tens of nm and presence of interfacial SiOx. Here we have investigated the mechanisms of YSZ epitaxy to determine the limits reducing the YSZ thickness and the interfacial layer. Ultrathin buffers around 2 nm thick, with less than 1 nm thick SiOx layer, can be fabricated by reflection high energy electron diffraction (RHEED) assisted PLD. Ultrathin CoFe2O4 films subsequently grown were epitaxial, although (111) oriented and with the SiOx layers more than 2 nm thick. The result is remarkable, but the total thickness of CFO/YSZ/SiOx is excessive for a tunnel device. We have used also Sc2O3 and Y2O3 buffers on Si(111), provided by collaborators at IHP-Frankfurt Oder. They are original candidates never combined with CoFe2O4. In spite of the huge lattice mismatch of around 15 and 20% CoFe2O4 grows epitaxially. Detailed transmission electron microscopy (TEM) has showed a mechanism of domain matching epitaxy. The films present magnetization close to the bulk value and without interfacial SiOx layer in the CoFe2O4/Y2O3/Si(111) sample. Thus Y2O3 appear as very promising buffer layer and maybe convenient for the nanometric structure required in a spin filter. We have demonstrated that ultrathin Y2O3 buffers, less than 2 nm thick, permit epitaxial growth of CoFe2O4, although the investigation of the interface stability has not been conclusive.

Date of Award	30 Sept 2013
Original language	English
Supervisor	Bénédicte Warot-Fonrose (Director), Florencio Sánchez Barrena (Director) & Javier Rodriguez Viejo (Tutor)