25 Apr
2012
25 Apr
'12
1:52 a.m.
Jochen Kunz wrote about MLC flash memory cells:
Unfortunately reading the contents of a cell draws a
few electrons of
its charge.
I'd be interested in references for that, as it contradicts everything
I've seen in the published literature on the theory of operation of
flash memory. In normal flash memory, even MLC, reading the cell does
not take any of its charge. The charge is on a floating gate,
electrically insulated from the conductors used to read it. As in a
normal MOSFET, there is no DC current from the floating gate to the
source or drain. The oxide for the floating gate has to be a *very*
good insulator under normal conditions, including reading; this is why
the flash part can store the data for years. Without using the erase
mechansim, the leakage of electrons is measured as a very small (<20)
number of electrons per year.
The write and erase processes, on the other hand, use high potential on
an additional gate to cause tunnelling of electrons through the oxide.
This is an entirely different mechanism from how the cells are read.