Phase-change RAM (or PRAM) is non-volatile memory that does not require a power source to store data. Its working principle is to heat a special material to change it between two different physical states: crystalline and amorphous (glassy). Let’s take the crystalline state to be 0 and the glassy state to be 1. Reversing the state means that the data has been written. If you flip it back, the data will be deleted.
The secret weapon here is the chalcogenide glass. It is a mixture of elements in groups 15 and 16 of the periodic table, including germanium, antimony, and tellurium. When heated, a phase change occurs. This is not a new concept. The same physical method has been used on rewritable CDs and DVDs for decades. These “RW” optical discs use a laser to melt or heat treat the material to form pits and lands that represent music and movies. The drive reads data this way because the laser reflects differently than shiny crystal and dull glassy parts.
A PRAM does the same thing, but uses electricity instead of light.
Phase change RAM cells deal with nanoscale structures. To write data, the material must be blasted with short, strong electrical pulses. This heats and melts the material quickly, then cools it quickly and freezes into a disordered, glassy state. This high resistance state typically represents a 1. To write a zero, a smaller, longer pulse is used to heat the material and rearrange the atoms into a low-resistance crystal structure. Reading it is simple: just apply a small voltage, not enough to change the state, and measure the resistance. High resistance means 1, low resistance means 0.
This is considered a successor technology to flash memory.
In 1970, three researchers, R.G. Neale, D.L. Nelson and Gordon Moore published a paper in Electronics magazine proposing this type of amorphous semiconductor memory. Moore, who later founded Intel and give us his famous law, recognized this opportunity early on. However, it took nearly 40 years for this technology to mature to the point where it was commercially viable.
By the mid-2000s, big players were circling. Intel, Samsung and IBM have all started showing off PRam prototypes. The pitch was compelling. Samsung claims that PRAM offered write speeds thirty times faster than a standard NAND flash memory. It also promises a 10x longer life. With each writing cycle, Flash memory deteriorates as the oxide layer it uses to capture electrons wears out. PRam does not use electrons trapped in an oxide well. It takes advantage of physical phase changes and is inherently more durable.
In February 2008, Intel and STMicroelectronics made headlines by introducing a PRam device that can store four states per cell. These multi-level cells are able to distinguish not only binary zeros and 1s, but also different resistance levels that can double or quadruple the storage density.
Despite the hype, pram never became the dominant storage technology we had imagined. Flash memory, especially NAND, is getting cheaper and denser much faster than expected. DRAM (dynamic RAM)
