It’s reminiscent of the demise of the 52 year-old Energy Conversion Devices, the brain-child of the great Stan Ovshinsky, which was sold off earlier this year.
ECD spent half a century developing, among other things, phase change memories which never made it into the mainstream.
11 years ago it looked as if Ramtron was on the brink of success. In July 2001 a technological breakthrough gave FRAM the chance to become the ideal memory – fast to read, fast to write, non-volatile, durable and dense.
Of all these qualities, it was density which always eluded Ramtron.
The 2001 breakthrough was the reduction of its basic memory cell down from two transistors plus two capacitors to one transistor and one capacitor, so reducing an FRAM cell down to the same basics constituents as a DRAM memory cell and, consequently, matching DRAM density.
Of course DRAM cells manage to make further savings on chip size by piling transistors on top of the capacitors – either by burying the capacitors in trenches in the silicon substrate or stacking devices on top of each other.
So Ramtron had to find a way to stack.
“The stacked cell is an absolute requirement for doing big memories,” said Ramtron’s Mike Alwais in May 2001, “when we can perfect the stacked cell we can do dense memory.”
He had plenty of help. As well as at Ramtron, the pursuit of the stacked FRAM cell was pursued by Infineon, STMicroelectronics and Toshiba.
In 2001, Infineon actually announced it would have 8Mbit FRAM in Q3 2001, customersamples of 32Mbit FRAMs in 2002, and production of 64/128Mbit FRAMs in 2003.
In 2001, the highest selling, top density DRAMs were16Mbit and 64Mbits with the 64Mbit unit volume overtaking the 16Mbit unit volume for the first time during 2001. So those announced densities for FRAM from Infineon would have got FRAM pretty close to DRAM density.
Somehow it didn’t happen. FRAM trudged on as a niche market product in low density, low power, radiation resistant applications.
Now, with the rejection of an $87.5 million bid from Cypress, Ramtron is up for grabs.
In turning down the offer, the Ramtron board says it ‘intends to consider the full range of available options including, but not limited to, the potential sale of the company.’
Ramtron said it has invited Cypress to join in this process which looks rather like an auction. Atmel and Renesas are said to be potential bidders.
So FRAM will continue to rumble along down its own little by-way but, not having Ramtron as the industry’s FRAM cheer-leader, will somehow leave a hole.