Ideal medium: single bit 1 single-domain particle 6.12J/3.155J Microelectronic processing Patterned 100nm storage media Each pillar is single-domain can be magnetized up'or 'down 200nm Each bit(occupying about100×100mm,≈100Gbin2) onsists of 1000 grains; media noise largely eliminated Dec.10,2003
D e c. 10 , 2 0 0 3 6.12J / 3.155J Microelectronic processing Ideal medium: single bit = 1 single-domain particle Patterned storage media 100 nm ‘1’ ‘0’ ‘1’ Each bit (occupying about 100 x 100 nm, ≈ 100 Gb/in2 ) consists of 1000 grains; media noise largely eliminated 200 nm Each pillar is single-domain, can be magnetized ‘up’ or ‘down’
How small must particle be to be in single-domain state? 6.12J/3.155J Microelectronic processing Exchange length, 1, over which the magnetization is approximately parallel 2)l/2 λ=(AM2) 20 nm for Ni) A= exchange constant(10-6 erg/cm); M=saturation magnetization Domain wall width, SE IA/K, )/2 Sd particles K= anisotrop thermally decouple Superparamagnetism eQg I Single domain nm scale No on-unitorm d<入 25k T Multi-domain K micron scale d>>λ,6 (Semiconductor memory problem: CV2< kT) Dec.10,2003
D e c. 10 , 2 0 0 3 6.12J / 3.155J Microelectronic processing How small must particle be to be in single-domain state? Exchange length, l, over which the magnetization is approximately parallel l = (A/Ms2)1/2 (20 nm for Ni) A = exchange constant (~10-6 erg/cm); Ms = saturation magnetization Single domain nm scale Non-uniform d < l d > l Domain wall width,d ≈ p(A/Ku)1/2, Ku = anisotropy. Multi-domain micron scale d >> l, d Vparticle £ 25kBT Ku SD particles thermally decouple: Superparamagnetism (Semiconductor memory problem: CV 2 < kBT )
Spin-based devices Metal length scales shorter 6.12J/3.155J Microelectronic processing Mobile carrier Scatters from ion 光 Spin-dependent resistivity Spin memory is lost over x, t M(x) Dec.10,2003
D e c. 10 , 2 0 0 3 6.12J / 3.155J Microelectronic processing Spin-based devices Metal length scales shorter Mobile carrier e- e- eScatters from ion Spin-dependent r < r < r resistivity x M (x) Spin memory is lost over x, t e- M (x) x e-
Spintronics spin(magnetism)-based electronic devices 6.12J/3.155J Microelectronic processing The smaller the better Ww→ MIX Spin memory Separator Device is lost over x, t non-mag metal = low-impedance spin valve or spin switch Unlike semiconductor devices insulator = high-impedance performance of spin-tunnel junction (Messervey and Tedrow, Phys. Rpts. 238, 174( 96) spin-based devices Moodera et al. Phys. Rev. Lett. 80, 2941 (98) improves as thickness decreases because screening lengths and spin diffusion lengths in metals < than in semiconductors Dec.10,2003
D e c. 10 , 2 0 0 3 6.12J / 3.155J Microelectronic processing Unlike semiconductor devices, performance of spin-based devices improves as thickness decreases because screening lengths and spin diffusion lengths in metals << than in semiconductors. Spintronics = spin (magnetism)-based electronic devices Separator Device non-mag metal => low-impedance metal => low-impedance spin valve or spin switch insulator => high-impedance insulator => high-impedance spin-tunnel junction (Messervey and Tedrow, Phys. , Phys. Rpts. 238, 174 (‘96); Moodera et al. Phys. Rev. et al. Phys. Rev. Lett. 80, 2941 (‘98)) The smaller the better x M (x) Spin memory is lost over x, t e-
6.12J/3.155J Microelectronic processing So it is worth exploring the limits of Nano-lithography, nano-patterning self-assembly not just for semiconductor devices but for magnetics as well Dec.10,2003
D e c. 10 , 2 0 0 3 6.12J / 3.155J Microelectronic processing So it is worth exploring the limits of Nano-lithography, nano-patterning, self-assembly not just for semiconductor devices but for magnetics as well…