M.Sc Thesis

M.Sc StudentIzraeli Tzipora Yael
SubjectNon-Volatile Memory based on MOS Capacitors and Metallic
DepartmentDepartment of Physics
Supervisors DR. Joan Adler
DR. Yuval Yaish


When nanoparticles are embedded in the insulating layer of a metal -- oxide -- semiconductor (MOS) capacitor, the capacitor exhibits hysteresis. The device shows different characteristics depending on whether or not there are charges stored in the nanoparticles, and thus can act as a memory cell. External driving power is only needed for the transition between the “charged” and “uncharged” states (write/erase), and therefore this memory is non-volatile.

Previous studies include experimental and theoretical projects that attempt to understand, quantify, and utilize this intriguing effect of charge trapping that enables memory. The objective of this study is to understand the dynamics of the charging and discharging of the traps, and to explore effects due to the discrete charge of the electrons. The research presented in this thesis spans experimental, theoretical, and numerical aspects of the subject.

Experimental evidence of the hysteresis effect caused by charge trapping in nanoparticles was obtained for MOS capacitors in which the oxide layers of hafnia (HFO?2)  were embedded with gold nanoparticles. The measured capacitance versus voltage (C-V) curves appear to be shifted with respect to each other, depending on the direction of the voltage sweep, forming a counterclockwise loop. The hysteresis loops grow as the sampling range is increased.

The characteristics for devices both with and without embedded nanoparticles were analyzed and compared, and the contribution of the nanoparticles to the hysteresis was confirmed. In addition, we examined the capacitors' performance over long periods of time, and verified that they can separately maintain each of the two values, and thus are acceptable as memory cells.

We developed a model to explain hysteresis based on rate equations for the occupation probability of charges in the nanoparticles. Our model employs theories of quantum dots and of charge tunneling in semiconductors. We derived expressions for four tunneling rates of electrons in and out of the nanoparticles, accounting separately for the electrons of the conduction band and of the valance band of the silicon semiconductor.

A versatile simulation program was written and implemented, in which the solutions of the rate equations were studied for various conditions and for different capacitor designs. First the ability of the simulation to capture the main physical properties of the system was demonstrated under the constraint of only one available electron occupation level in the nanoparticle. Later, we extended the calculation to allow many levels, both positive and negative. The influence of different possible approximations on the simulation, and the necessity of exact numerical calculation are discussed. In addition, we tested how slight changes in the set values affected the final results, and extrapolated to extreme ranges for better understanding of the model's behavior at the limits.

Our model provides a viable platform for basic understanding of the system. The trends we found in our calculated results are consistent with our group's experiments, and with those reported by others.