A STUDY OF INDIVIDUAL ELECTRON TRAPPING CENTERS AND 1/F NOISE IN METAL-INSULATOR-METAL TUNNEL JUNCTIONS

I have used resistance fluctuations of small-area (A < 100 (mu)m('2)) metal-insulator-metal tunnel junctions to study rates of electron capture and emission by, and interactions between, individual electron traps within the insulating barrier material. I worked in the temperature range 2K < T < 300K, at junction bias voltages (VBAR)V(VBAR) < 250m V, and with Nb(,2)O(,5) and a-Si barriers.;These data suggest a charge trapping model based on coupling to motion of the ions which form the trap. Within the model, thermally activated rates are due to the activated hopping of the ions between two nearly energetically equivalent spatial configurations; the low temperature saturation is due to finite zero-point motion of the ions which leads to a non-zero probability of transitons even at zero temperature.;Interactions between traps appear as variations in the magnitude of the resistance change (delta)R(,J), and the capture and emission rates for a given trap upon charging of a second trap. The defects appear to interact via both their electrostatic and lattice deformation potentials. The range of the interaction ((TURN)100(ANGSTROM)), together with the observed (delta)R(,J), suggests that the major interaction is via the lattice deformation.;Further, each trap contributes a single Lorentzian spectrum to the total resistance power spectrum S(,R)(f); the observed range of E(,B) and (tau)(,0) above leads to the 1/f spectrum observed in large area devices where the discrete Lorentzian contributions overlap.(, )For T > 150K, an understanding of the interactions becomes essential for a full understanding of 1/f noise. Thus, I have been able to demonstrate that 1/f noise in tunnel junctions arises from a superposition of many distinct electron trapping events and have used the noise itself as a sensitive probe to reveal the detailed kinetics of the individual electron traps.;For Nb(,2)O(,5), above (TURN)15K both the emission and capture rates are of the form (tau)(,0)('-1)exp(-E(,B)/k(,B)T). The range of parameters I observe is: 10('-9)sec < (tau)(,0) < 10('-13)sec and 10meV < E(,B) < 500meV. Below 15K the rates deviate dramatically from thermal activation and saturate at a value which is exponentially voltage dependent. a-Si yields similar results except that there is a marked dip in the rates at (TURN)9K just prior to the low temperature saturation.