Electrochemical metallization ReRAMs (ECM) - Experiments and modelling: general discussion

Abstract

Ella Gale opened discussion of the introductory lecture by Rainer Waser: At the end of your talk you introduced complementary resistance switches, which you used for binary pattern matching. Do you have any insight into how you might use these systems to do computation in the real number space? Rainer Waser answered: We are currently working on this. (2:[2]2) Ilia Valov said: From your point of view, at the moment, what are the hottest points of research in memristive cells? What are the main challenges? Rainer Waser answered: The hot points of research are the physical understanding of the reliability-limiting processes in conjunction with the material's treasure map and, from the application side, the development of a convincing neuromorphic circuit that shows clear benet compared to standard CMOS solutions (4:[4]4) Hans Hilgenkamp asked: Most memristor embodiments shown in the papers are based on stacked devices with the current ow perpendicular to the plane. Do all applications necessarily favour this type of conguration, or are there also applications for which an in-plane conguration would be preferred? DIS C8FD90059K

Effects of top electrode material in hafnium-oxide-based memristive systems on highly-doped Si

“…For instance, the electroforming process for the valence change memory (VCM) based bipolar devices may be polarity dependent due to the built-in asymmetry of the contacts at the top and bottom electrode interfaces as determined by the device design and fabrication procedure 33 . In the case of chemical metallization memory (ECM) 34,35 based devices, the formation of a filament requires the electrochemical dissolution of an active electrode (usually Ag or Cu), followed by the drift of the injected metal cation clusters toward the counter electrode by a positive voltage applied to the active electrode 26,36,37 . Thus, the comparison of the response of these devices to different biasing schemes may provide valuable information for deciphering the nature of resistive switching in a material system.…”

Electro-thermal transport in disordered nanostructures: a modeling perspective

“…with I being the identity matrix of appropriate size. In eqn (8) and (9) the GFs are different, namely retarded (G R ), advanced (G A ), lesser (G < ), and greater (G > ), and the inuence of I nj is indirectly accounted for in the lesser and greater boundary selfenergy S +,B (E). The advanced GF is the Hermitian transposition of the retarded GF, i.e.…”

“…Eqn (8) and (9) are known as the non-equilibrium Green's function (NEGF) formalism. 12 Intuitively, the lesser (greater) boundary self-energy, S B,+ (E), indicates the probability that a state gets lled (S B,< (E)) or emptied (S B,> (E)) through interactions with the contact.…”

“…The off-diagonal elements of the lesser and greater GF, G + (E), describe the correlation between the involved basis functions, whereas the diagonal entries contain the probability that a state n is occupied (G < nn (E)) or unoccupied (G > nn (E)). Therefore, it is not necessary to convert back the results of eqn (8) and (9) to a wave function j(E). All observable quantities such as the charge density r(r) at position r and the electrical current I d can be directly derived from selected entries of G + (E).…”

“…Injecting electrons into the device domain drives it out of equilibrium, which changes the distribution of electrons and modies the charge density r(r). This in turn affects the Hamiltonian H through the rst two terms in the effective potential V eff (r) in eqn (3), giving rise to a mutual dependence of eqn (2), (8) and (9). It must be resolved in a self-consistent manner until r(r) is converged.…”