IMPROVING THE HAFNIA-BASED RESISTIVE RANDOM-ACCESS MEMORY THROUGH MATERAL ENGINEERING

Abstract

RESISTIVE RANDOM-ACCESS MEMORIES (RRAMS) ARE EXTENSIVELY INVESTIGATED AS A REPLACEMENT FOR FLASH MEMORIES DUE TO ITS SUPERIOR CHARACTERISTICS: HIGHER SPEED, BETTER ENDURANCE, MORE SCALABLE DESIGN AND SIMPLER FABRICATION PROCESSES. HOWEVER, DUE TO THE STOCHASTIC NATURE OF SUCH RESISTIVE SWITCHING, A LARGE VARIATION IN THE RESISTANCE OF EACH STATE IS OFTEN OBSERVED FROM CYCLE TO CYCLE. THIS EFFECTIVELY REDUCES THE MEMORY WINDOW BETWEEN THE HIGH AND LOW RESISTANCE STATES AND COULD POSSIBLY RESULT IN BIT ERRORS. THIS THESIS AIMS TO ADDRESS THE ISSUE OF LARGE RESISTANCE VARIATION IN HFO2-BASED REDOX RRAMS BY INCREASING THE MEMORY WINDOW AND/OR REDUCING THE VARIATION. USING CMOS AVAILABLE PROCESSES LIKE SILICIDATION AND ATOMIC LAYER DEPOSITION (ALD), SYSTEMATIC INVESTIGATIONS ON BOTTOM ELECTRODES AND SWITCHING CHARACTERISTICS OF HFO2 WERE CARRIED OUT TO IDENTIFY THE BEST BOTTOM ELECTRODE MATERIAL TO GIVE A LARGE MEMORY WINDOW AND LOW VARIATION IN READ-WRITE CHARACTERISTICS. PROGRAMMING OR R