0% found this document useful (0 votes)
432 views6 pages

Flash Memory

This document provides an introduction to flash memory, including its history, operation, and scaling challenges. It describes how flash memory was invented in the 1980s at Toshiba and how the technology has improved over time. The key aspects of flash memory operation - programming, erasing, and reading - are explained through comparisons to MOSFET transistors and by modeling the floating gate capacitor system. Issues with further scaling flash memory below 10nm, such as direct tunneling and charge retention times, are also discussed.

Uploaded by

ianchaffin
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
432 views6 pages

Flash Memory

This document provides an introduction to flash memory, including its history, operation, and scaling challenges. It describes how flash memory was invented in the 1980s at Toshiba and how the technology has improved over time. The key aspects of flash memory operation - programming, erasing, and reading - are explained through comparisons to MOSFET transistors and by modeling the floating gate capacitor system. Issues with further scaling flash memory below 10nm, such as direct tunneling and charge retention times, are also discussed.

Uploaded by

ianchaffin
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 6

Ian$Chaffin$

ENGS$199$
Final$Report$
$

Flash$Memory:$A$Brief$Introduction$
$

Introduction$
$
$
The$goal$of$this$report$is$to$provide$a$brief$yet$sufficiently$thorough$
explanation$of$Flash$Memory$operation$so$that$a$student$with$some$background$in$
semiconductor$devices$could$read$the$report$and$have$a$good$understanding$of$how$
Flash$works.$It$is$important$that$students$in$the$semiconductor$field$have$an$
understanding$of$Flash$because$its$use$is$widespread.$Memory$is$now$an$
indispensable$part$of$cell$phones,$digital$cameras,$personal$computers,$automotive$
systems$and$many$others.$The$Flash$memory$industry$is$controlled$by$a$few$
semiconductor$superpowers$including,$Samsung,$Toshiba,$Micron$and$Intel.$In$2010,$
Samsung's$quarterly$flash$memory$revenue$was$$1.7$P$$2.1$billion.$Each$company$is$
pushing$towards$higher$density$and$faster$operation.$$An$explanation$of$the$devices$
operation$will$be$given$later$so$as$to$provide$an$understanding$of$the$challenges$
associated$with$making$Flash$smaller$and$faster.$$
$
$History$
$
$
Dr.$Fujio$Masuoka$invented$flash$Memory$while$working$for$Toshiba$in$the$
1980's.$At$first$the$device$was$known$as$simultaneously$erasable$EEPROM$
(Electrically$Erasable$Programmable$Read$Only$Memory).$$However$the$name$flash$
was$used$because$lots$of$memory$could$be$erased$"in$a$flash."$When$NAND$flash$was$
invented$in$1987,$the$name$flash$made$it$into$the$title$of$Dr.$Masuoka's$paper$and$it$
became$a$household$from$then$on$as$NAND$structure$devices$reduced$the$memory$
cell$area$to$achieve$a$lower$cost$per$bit.$These$devices$were$an$important$step$
forward$for$memory$since$they$could$retain$data$even$when$the$power$was$
switched$off,$unlike$the$dynamic$random$access$memory$(DRAM)$chips$that$were$
already$prevalent$in$the$early$1980's.$Despite$facilitating$the$invention$of$flash,$
Toshiba$has$not$been$the$only$contributor.$Intel$was$right$on$their$tale$from$the$
outset$being$the$first$to$commercialize$NOR$flash$in$the$late$1980's.$As$the$century$
turned,$Flash$became$a$staple$in$the$memory$industry$with$companies$like$IBM$
creating$new$applications$for$flash$like$the$USB$drive.$With$more$companies$
developing$flash$devices,$the$size$of$the$devices$followed$closely$with$Moore's$law$of$
scaling.$Figure$1$shows$the$scaling$timeline$from$the$invention$of$flash$to$present.$As$
of$2013,$Samsung$has$mass$produced$128Gb$3Pbit$MLC$NAND$flash$in$10$nm$class$
process$technology.$Many$challenges$arise$when$venturing$past$the$10$nm$mark$for$
flash$memory.$These$will$be$discussed$later,$but$for$now$the$operation$of$flash$
devices$will$be$discussed.$
$
$
$

Figure$1:$NAND$Flash$Device$Scaling$by$Industry$Leader$

$
$
Operation$
$
To$understand$the$method$by$which$a$single$flash$cell$works$it$is$important$to$draw$
some$comparisons$to$the$classic$MOSFET.$For$an$nMOS$transistor$a$voltage$is$
applied$at$the$gate$which$induces$a$flow$of$electrons$from$source$to$drain$in$the$
channel.$In$order$to$obtain$a$functioning$memory$device,$the$charge$that$is$
generated$in$the$device$through$the$gate$voltage$must$be$stored$in$some$capacity.$
This$is$accomplished$through$a$floating$gate$transistor,$in$which$an$additional$gate$
is$electrically$isolated$and$placed$between$the$control$gate$and$the$substrate.$The$
two$devices$are$shown$below$in$figure$2.$
$
$

Figure$2:$MOSFET$and$Floating$Gate$Transistor$Diagrams$$

Charge$is$added$to$the$floating$gate$when$electrons$cross$from$the$channel$to$the$
floating$gate$via$quantum$tunneling.$When$the$floating$gate$is$uncharged$the$cell$is$
given$a$bit$value$of$1$and$when$the$cell$is$charge$it$is$given$a$value$of$0.$The$reason$
being$that$when$there$is$charge$on$the$floating$gate,$it$impedes$the$flow$of$electrons$
in$the$channel$and$thereby$increases$the$threshold$voltage$by$some$VT.$The$
method$used$to$read$a$cell$is$simple.$An$intermediate$voltage$is$applied$to$the$
control$gate$and$the$drain$to$source$current$is$measured.$If$there$is$current$flowing$
then$the$device$is$"ON"$and$there$is$no$charge$on$the$floating$gate.$If$there$is$no$
current$flowing$then$the$device$is$"OFF"$corresponding$to$a$bit$value$of$0$since$the$
floating$gate$is$charged.$The$change$in$the$threshold$voltage$from$the$additional$
charge$on$the$floating$gate$is$given$by$the$following$equation:$
$
!!"
$
(1)$
!!! =
$
!!"
$
$
In$order$to$obtain$an$expression$for$the$charge$on$the$floating$gate$as$a$
function$of$the$voltage$on$the$drain$and$the$floating$gate,$the$device$can$be$modeled$
using$capacitive$coupling.$From$the$simple$charge$expression$Q$=$CV,$each$node$
interface$with$the$floating$gate$can$be$modeled$as$a$capacitor$and$added$together$to$
obtain$an$expression$for$the$charge.$Using$some$algebra,$the$desired$expression$is$
obtained$and$illustrated$in$Figure$3.$
$

Figure$3:$Capacitive$Coupling$Diagram$and$GCR$Derivation$$

$
The$equation$shown$in$Figure$3$is$important$because$it$provides$a$way$of$relating$
the$voltage$on$the$floating$gate$to$the$voltage$on$the$control$gate$via$the$GateP
Coupling$Ratio$(GCR).$Typically$a$GCR$of$more$than$0.6$is$essential$for$satisfactory$
operation.$This$means$that$the$thickness$of$the$blocking$oxide$must$be$at$least$1.67$
times$the$length$of$the$tunnel$oxide.$$
$
So$how$does$charge$build$up$on$the$floating$gate$in$the$first$place?$The$
charge$build$up$is$achieved$through$quantum$tunneling.$In$order$to$program$the$

device$very$quickly$a$specific$type$of$tunneling$know$as$FowlerPNordheim$(FPN)$
tunneling$is$employed,$which$uses$high$voltages$to$cause$electrons$to$tunnel$with$
high$current$density.$The$expression$for$the$current$density$as$a$function$of$electric$
field$of$injection$is$given$below:$$
$
!
$
(2)$
! !
!
!!!! ! = !!!!"#
! !!"# $
$
!/!
!! !
4 2! !!
!! = !
!!!!!! = !
$
8!!! !
3!"
$
Where$h$is$Plancks$constant,$b$is$the$energy$barrier$at$injecting$interface,$Einj$is$the$
electric$field$at$the$injecting$interface,$q$is$the$charge$of$an$electron,$m$is$the$free$
electron$mass,$m*$is$the$effective$mass$of$an$electron$in$SiO2$bandgap$and$$=$h/2.$
The$equation$can$be$rearranged$to$draw$out$the$energy$barrier$term$b.$$
$
!
!!"#
!!!! ! = ! !!
!
!!

!/!

!
!!! !

!!"#

$
Now$the$expression$shows$the$relationship$between$the$energy$barrier$and$the$
injecting$field.$When$Einj$increases$the$current$density$increases,$when$b$increases$
the$current$density$decreases.$As$long$as$the$voltage$drop$across$the$tunnel$oxide$is$
greater$than$b$then$the$cell$will$undergo$FPN$tunneling.$$
$
When$the$device$is$being$programed$the$initial$charge$on$the$floating$gate$is$
zero$and$drain$is$held$at$ground.$As$an$example$the$control$gate$is$held$at$15$V$and$
the$GCR$is$0.6.$At$first$the$electrons$tunnel$across$the$tunnel$oxide$onto$the$floating$
gate.$As$charge$begins$to$build$on$the$floating$gate,$the$voltage$on$the$floating$gate$is$
lowered$by$the$presence$of$negative$charge$according$to$the$following$equation:$$
$
!!"
$
(3)$
!!" ! = !"# !!" +
$
!!
$
As$the$voltage$on$the$floating$gate$is$lowered$the$voltage$difference$between$the$
voltage$on$the$control$gate$and$voltage$on$the$floating$gate$increases.$Because$of$the$
increased$voltage$difference$between$the$control$gate$and$the$floating$gate,$some$
tunneling$begins$to$occur$across$the$thicker$blocking$oxide.$Eventually$enough$
charge$builds$on$the$floating$gate$that$the$two$current$densities$equilibrate.$At$this$
point$the$total$current$tunneling$onto$the$floating$gate$is$equal$to$the$total$current$
tunneling$out.$Thus$the$charge$on$the$floating$gate$saturates$at$the$equilibrium$
value.$Figure$4$illustrates$the$program$process.$
$

Figure$4:$Band$Diagrams$of$the$Program$Process$$

$
$
Erasing$is$almost$identical$to$the$programming$a$cell,$only$a$negative$voltage$
is$applied$to$the$control$gate$instead$of$a$positive$voltage$and$the$process$runs$in$
reverse.$Charge$is$retained$on$the$cell$by$simply$setting$the$control$gate$voltage$to$
zero$so$that$the$electrons$cannot$overcome$the$barrier$of$the$tunnel$oxide.$If$the$
oxides$are$sufficiently$thick$and$there$are$few$defects$then$the$charge$can$remain$on$
the$floating$gate$for$up$to$10$years.$$
$
Multiple$bits$can$be$stored$in$a$single$cell$through$a$type$of$flash$known$as$
MuliPLevel$Cell$(MLC)$flash.$MLC$devices$control$the$amount$of$charge$stored$on$the$
floating$gate$through$specified$control$gate$voltages.$Thus$a$single$cell$can$achieve$
several$different$charge$states,$which$lowers$the$required$number$of$cells$per$data$
volume.$One$of$the$main$challenges$to$MLC$is$read$disturb.$Because$each$charge$
state$is$closer$together$than$before,$applying$an$intermediate$voltage$to$read$a$cell$
can$affect$the$charge$stored$on$an$adjacent$cell.$Also$each$cell$contains$more$data$
and$is$therefore$written$to$and$erased$more$often.$As$electrons$tunnel$with$high$
energy$they$collide$with$the$oxide$lattice$and$cause$defects$that$degrade$the$oxide.$
Eventually$the$oxide$wears$down$and$the$device$no$longer$functions$properly.$MLC$
can$typically$handle$around$10,000$program/erase$cycles$while$singlePlevel$cell$
(SLC)$can$handle$100,000$cycles.$Because$MLC$is$much$cheaper$to$produce,$many$
algorithmic$solutions$have$been$devised$to$increase$the$number$of$available$cycles.$
$
Conclusion3
3
$
Because$of$the$high$density$capabilities$of$flash$it$will$continue$to$be$
enhanced$in$the$years$to$come.$However$as$devices$get$smaller,$some$barriers$to$
scaling$arise$as$well.$When$tunnel$oxides$drop$below$10$nm,$the$devices$fall$prey$to$
direct$tunneling$and$trap$assisted$tunneling$that$cut$retention$from$years$to$days.$
New$methods$for$better$retention$will$have$to$be$developed$to$produce$reliable$
single$digit$nanometer$flash$memory$technologies.$

References$
$
Brewer,$Joe,$and$Manzur$Gill.$Nonvolatile*Memory*Technologies*with*Emphasis*on*
Flash:*A*Comprehensive*Guide*to*Understanding*and*Using*NVM*Devices.$
Hoboken,$NJ:$Wiley,$2008.$Print.$
Koh,$Yohwan.$"NAND$Flash$Scaling$Beyond$20nm."$2009*IEEE*International*Memory*
Workshop$(2009):$n.$pag.$Web.$
Zhao,$Chun,$Ce$Zhao,$Stephen$Taylor,$and$Paul$Chalker.$"Review$on$NonPVolatile$
Memory$with$HighPk$Dielectrics:$Flash$for$Generation$Beyond$32$Nm."$
Materials$7.7$(2014):$5117P145.$Web.$
Figures$
$
https://en.wikipedia.org/wiki/File:NAND_scaling_timeline.png*
*
All*other*figures*were*designed*by*the*author.*
$

You might also like