Feature Article - A Brief History of Non-Volatile Memory, Part 1 - By Christopher Henderson

Non-volatile memory, which can hold data with power removed, has become critical to devices like PDAs, cell phones, and digital cameras. Over the next several issues, this three-part article series will discuss basic non- volatile memory technologies to give you a better understanding of their background and importance.

In Part 1, we will cover one of the original non-volatile memory technologies, the metal nitride oxide silicon, or MNOS technology. Part 2 will discuss an improvement over MNOS called silicon oxide nitride oxide silicon (SONOS) technology, and its predecessor, silicon nitride oxide silicon (SNOS). Next, we’ll discuss the floating gate avalanche injection MOS or FAMOS transistor, used extensively for years in ultraviolet erasable memory. One of the more popular variants of this technology is flash memory. Part 3 will cover the strengths and weaknesses of NAND and NOR, the two most common non-volatile memories in use today.

Two basic mechanisms store charge in non- volatile memory. The first method, which forms the basis for MNOS technology, utilizes traps in the insulator or at an interface. The second method, called floating gate technology, stores the charge on a conductive or insulator material sandwiched in between the control gate and the silicon channel.

In 1967, Richard Wegener and his colleagues at the Sperry Rand Research Center developed the metal nitride oxide silicon transistor (MNOS) while researching technologies to help store data and reduce power dissipation in military electronics. When they replaced the oxide layer in a standard MOS transistor with a nitride-oxide stacked layer, the bonds between the nitride and the oxide created an interface containing a high level of traps (unterminated bonds that attract positive or negative charge). Placing a high voltage on the gate forced charge onto these interface traps, while placing a high voltage of the opposite polarity on the gate removed the interface traps. One disadvantage of this technique is that trapped charge tends to disperse or recombine, limiting the storage time of the structure. The higher the storage temperature, the more quickly the charge recombines. Researchers have been able to improve the retention time by replacing the top metal gate with a silicon gate.

The figure on the left illustrates the basic MNOS cell structure and a graph of the structure’s charging and discharging times. The graph shows an example set of curves for a combined dielectric stack of 1000 angstroms and differing oxide thicknesses. For thinner oxides, the charging and discharging times decrease substantially. However, a MNOS structure requires a long charging time to retain trapped charge. Conversely, a short charging time leads to a short discharge time. Because of this problem, MNOS memory never became popular as a high-volume commercial technology. However, its greater tolerance of radiation made the MNOS suited for some niche applications, such as military and space.