Digit Geek
Digit Geek > Recent Articles > Technology > The evolution of NAND flash memory

The evolution of NAND flash memory

From humble floating gates to powerful 3D stacks

Storage devices are part and parcel of our electronic lives today. And among them, flash memory holds its own important niche. Present in our USB flash drives, among other things, flash memory enables non-volatile solid state storage that can be electrically erased and reprogrammed. That makes them similar to the HDD disks that we see in our PCs but with the difference being that they’re very portable. Broadly, flash memory is of two types – NAND and NOR, and yes, the primary difference between them is the logic gate used in their design. While NOR memory also has its uses, NAND is what we see in most everyday devices like the USB flash drive, cameras, memory cards, SSDs and other devices that require portable and easily accessible storage. But there’s a lot more to how NAND flash memory came to be as ubiquitous as it is today.

Originpoint

The invention of flash memory can be traced as far back as the 1960s to D. Kahng and S.M. Sze’s floating gate concept. NAND-type flash, if the popular consensus is to be believed, didn’t come around until the efforts of Fujio Masuoka were demonstrated at IEDM in 1987. Toshiba launched the NAND-type flash memory at IEEE International Electron Devices Meeting in 1987, held in San Francisco. The interesting thing about this is that the term ‘flash’ might have been coined by Mr Masuoka’s colleague Mr Sho-ji Ariizumi – because the erasure process of the NAND flash memory reminded him of a camera’s flash.

Dr Fujio Masuoka

The development of NAND flash has its precursor in EEPROM. Along with the floating gate concept, the application of Fowler-Nordheim tunnelling for both program and erase operations led to the development of the area efficient NAND chip. Soon after its introduction, NAND flash, along with flash memory overall, saw a significant rise. There were primarily three reasons behind it.

First, NAND flash’s high area efficiency due to its sparing use of metal contacts. Additionally, around when NAND was introduced, a method was developed to store more than one bit in a cell by controlling the tunnelling of electrons into the floating gate, increasing the digital data density in the NAND chip. On top of that, to mitigate the memory cell damage that arose from the application of Fowler-Nordheim tunnelling, techniques were introduced at the system level.

Others join in

Intel saw the potential in flash memory and launched NOR-type flash the very next year. The long read-write times made it unsuitable for heavy usage but the availability of random access made it perfect for system storage. It was in the same year that Eli Harari, inventor of the Floating Gate EEPROM which proved the practicality, reliability and endurance of semiconductor-based data storage, along with Sanjay Mehrotra and Jack Yuan, established SunDisk (now known as SanDisk) for the development of a new ‘system flash’ architecture for the emulation of disk storage using embedded controller, firmware and flash memory. Along with the establishment of JPEG and MPEG standards, over the next couple of years, the digital camera industry saw a boom with products from Fuji, Kodak and more. By 1990, the flash memory industry was growing like never before.

The digital camera industry saw a boom with developments in NAND flash

The golden age begins

In 1991, Toshiba announced the development of the first 4 Mb NAND type electrically erasable programmable read-only memory (EEPROM). These were specifications to achieve a system that was completely different from the memory and storage protocols of the time. While developments were also happening on the NOR flash side of the storage game, Toshiba and IBM, in a strategic partnership, also started developing 16-megabit solid state disks to replace hard disks. Within a year, Toshiba had announced the commercialisation and subsequent mass production of 16-megabit NAND flash-memory products. NAND flash was now a major portion of the company’s business. By the time the mid-90s came around, the flash memory industry had clocked a growth of more than 150% in 10 years.

In 1996, Toshiba had launched the SmartMedia memory card products with 16-megabit and 32-megabit variations, competing against CompactFlash introduced by SunDisk a couple of years ago. There was severe competition between digital cameras using SmartMedia and those using the Compact Flash standard. In the next few years, flash memory made its way to portable music players and cellphones of the time. Although, the popularisation of NAND flash was about to happen – right at the turn of the millennium.

Flash drives

In the year 2000, the commercial sale of USB flash drives was kicked off by Trak Technology and IBM. Until then, beyond cameras, flash memory was being used through encumbering floppy drive adapters to ensure backwards compatibility. As a result, the applications just couldn’t take advantage of the speed that NAND flash provided. While Trak Technology called the new product the ‘ThumbDrive’, IBM marketed the same in the US under the moniker of “DiskOnKey”, which was developed and manufactured by the Israeli company M-Systems. Within a year, the revenue from NAND flash had crossed $1 billion. Established companies like Toshiba and SanDisk announced 1 Gb MLC NAND. SanDisk even introduced its own system flash product.

USB flash drives were responsible for the growing popularity of NAND storage before smartphones and popular portable multimedia devices came along

Storage capabilities increased significantly over the next few years. While the 1Gb MLC NAND announced earlier came into being by 2005, around the same time, Samsung announced the development of a 2GB NAND flash storage. The price of NAND flash had already dropped below DRAM for the same density. Hynix and ST Micro formed a joint venture and Hynix NAND product was introduced. Infineon NAND product based on Saifun Charge Trap Flash was announced. But even with all these developments, bigger upheavals were headed towards NAND memory.

Flash as we know it

The first gen iPhone holds significant credit for generating a huge demand and popularity for NAND flash

On January 9, 2007, the iPhone was announced by Steve Jobs at the Macworld convention. Yes, it did come with NAND flash storage, as have many Android-powered and other smartphones since then. The growing popularity of iPhones, along with categories like iPods that were popular at the time, the demand for NAND chips soared. In 2009, Apple signed a long-term contract with Toshiba for NAND chips, cementing future demand for NAND chips.

A startup in Salt Lake City, Utah was also trying something different on the enterprise front at this point. Fusion.io was “trying to accelerate the other side of the internet”, as Steve Wozniak (who eventually joined the company as Chief Scientist in 2009) would put it, by putting NAND flash on servers using PCIe.

On the enterprise side of things, EMC (currently Dell EMC) introduced new 73GB and 146GB SSDs as options for their high-end Symmetrix DMX-4 systems. Additionally, while NAND flash memory was always heading towards smaller and smaller dimensions, it was in 2010 that Intel and Micron sampled 25 nm NAND products. Later that year, and next year, Samsung and Sandisk started shipping 27 nm and 24 nm NAND products respectively.

Meanwhile, the way NAND chips are built was due for a change. Instead of building out, NAND architects began to build up, stacking chips vertically. This led to 3D NAND, or as Samsung calls it, V-NAND. 3D NAND allows greater density and better performance without an increase in size while overcoming the interference and manufacturing challenges which had previously been limiting NAND technology. Further developments led to an increasing number of layers, greater capacities, better performance.

Samsung started off V-NAND with its its 850 Pro SSD

Quite recently, Samsung announced the mass production of its 5th gen V-NAND memory. According to Samsung, the new 256Gb or gigabit V-NAND flash memory has reached a speed of 1.4Gbps, or gigabits per second, resulting in a 40% increase from its 64-layer predecessor. From this point onwards, things can only get faster, smaller and more reliable.

Arnab Mukherjee

Arnab Mukherjee

A former tech-support desk jockey, you can find this individual delving deep into all things tech, fiction and food. Calling his sense of humour merely terrible would be a much better joke than what he usually makes.