33
Baker 1 The One-Transistor, One-Capacitor (1T1C) Dynamic Random Access Memory (DRAM), and its Impact on Society R. Jacob Baker Department of Electrical and Computer Engineering Boise State University 1910 University Dr., ET 201 Boise, ID 83725 [email protected] Abstract – Memory technology development, in particular dynamic random access memory (DRAM), has been the greatest driving force in the advancement of solid-state technology for integrated circuit development over the last 40 years. The origin of DRAM circuits and technology can be traced to Dr. Dennard’s Patent (Number 3,387,286) granted on June 4, 1968. This truly visionary work, using a single transistor and capacitor (the 1T1C), is one of the most manufactured devices in the history of mankind. This talk will review the impact of his invention and discuss the brilliance of Dr. Dennard for conceiving the invention of the 1T1C cell prior to the maturity of metal oxide semiconductor (MOS) technology and in the face of critics that may have likely asked “why in the world would we want a memory, DRAM, that forgets its’ contents!?”

The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Embed Size (px)

Citation preview

Page 1: The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Baker 1

The One-Transistor, One-Capacitor (1T1C) Dynamic Random Access Memory (DRAM), and its Impact on Society

R. Jacob BakerDepartment of Electrical and Computer Engineering

Boise State University1910 University Dr., ET 201

Boise, ID [email protected]

Abstract – Memory technology development, in particular dynamic random access memory (DRAM), has been the greatest driving force in the advancement of solid-state technology for integrated circuit development over the last 40 years. The origin of DRAM circuits and technology can be traced to Dr. Dennard’s Patent (Number 3,387,286) granted on June 4, 1968. This truly visionary work, using a single transistor and capacitor (the 1T1C), is one of the most manufactured devices in the history of mankind. This talk will review the impact of his invention and discuss the brilliance of Dr. Dennard for conceiving the invention of the 1T1C cell prior to the maturity of metal oxide semiconductor (MOS) technology and in the face of critics that may have likely asked “why in the world would we want a memory, DRAM, that forgets its’ contents!?”

Page 2: The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Baker 2

Let’s Step Back in Time

What are the following (hint: they were found in your local supermarket in the 50s, 60s, and early 70s)

Photos taken from ebay.com

Page 3: The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Baker 3

Semiconductor-based Electronics were still new during this period of time

Does anyone remember the birth of the transistor radio?Below is seen the 2-transistor Furtura transistor radio (ca. 1955)

Note the output is via an earphoneAlso note the schematic (not many, if any, consumer electronics products today come with a schematic)

Photos taken from ebay.com

Page 4: The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Baker 4

Evolution of the Transistor Radio

These were a big deal both because they were portable (something not practically possible with vacuum tube radios) and because they used a new technology (solid-state devices that didn’t wear out)

Photos taken from ebay.com

Page 5: The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Baker 5

Growing Complexity

Note the use of bipolar junction technology to replace vacuum tubes

Photo taken from ebay.com

Page 6: The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Baker 6

What was going on with memory storage during this time?

Dominated by the magnetic core memory seen belowDuring the 1960s solid-state memory was developed based on the BJT

Ultimately wasn’t successful (compared to MOS-based memory)

Image from wikipedia.org

1 mm

Page 7: The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Baker 7

So why is Dr. Dennard’s invention such a big deal?

During the excitement of bipolar technology replacing vacuum tubes in most electronic devices he filed an invention disclosure on July 14, 1967 that

Used metal-oxide-semiconductor (MOS) technologyProposed a memory, the one-transistor, one-capacitor (1T1C cell), that forgets its contents!

Why is using MOS technology such a big deal over bipolar?MOS technology is scalable!Another significant contribution from Dr. Dennard’s is scaling theory.

Why is a memory that forgets its contents such a big deal?It’s small!It’s fast!

Page 8: The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Baker 8

Dennard’s Scaling Theory*

Predicted that MOS devices would continue to shrink in size (scale) over time

Higher densityFasterLow power

Also discussed how interconnect would scaleRC times of the lines don’t scale but as distances shrink delays drop

* Dennard, R.H.; Gaensslen, F.H.; Rideout, V.L.; Bassous, E., and LeBlanc, A.R., “Design of ion-implanted MOSFET's with very small physical dimensions,” IEEE Journal of Solid-State Circuits, Volume 9, Issue 5, Oct 1974, pp. 256 - 268

Page 9: The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Baker 9

Scaling Theory

Scaling from Dennard’s paperDevice and Circuit ScalingInterconnect ScalingScale parameter is about 1.4 (1/κ = 0.7)

Page 10: The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Baker 10

The One-Transistor, One-Capacitor (1T1C) Dynamic Random Access Memory (DRAM) cell

The array is formed with word (row) and column (bit) lines

Page 11: The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Baker 11

Folded and Open Arrays

The open array

Page 12: The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Baker 12

Folded and Open Arrays

The folded array

Page 13: The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Baker 13

The Folded Array

Page 14: The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Baker 14

Further Reduction in Cell Size

Sharing the bitline contact

Page 15: The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Baker 15

Mbit Pair: Folded Array

Page 16: The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Baker 16

Folded Array Cell Size

Page 17: The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Baker 17

Open Array: 6F2

Page 18: The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Baker 18

Array Block

Page 19: The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Baker 19

The 1T1C DRAM Cell

ACT

EQ

VDD/2

NLAT

I/O

I/O

Row 1

Row 2

Col 1

1-transistor, 1-capacitor (1T1C) DRAM Cell

Page 20: The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Baker 20

SEM Photo of a Modern DRAM Array

Page 21: The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Baker 21

WorkstationsUnits - 1.69MDRAM – 5,741MB/unit averageDRAM Revenue - $1,164MSource: iSuppli

Units - 51MDRAM – 510MBDRAM Revenue - $3,105MSource: iSuppli

NotebookComputers

Memory UpgradesUnits – 87MDRAM – 295MBDRAM Revenue - $3,060MSource: iSuppli

Personal Digital AssistantsUnits – 8.9MDRAM - 65MB averageDRAM Revenue - $698MSource: iSuppli

Desktop ComputersUnits – 150MDRAM – 560MB DRAM Revenue - $10,048MSource: iSuppli

PrintersUnits - 112MDRAM – 35MBDRAM Revenue - $156MSource: iSuppli

PC Servers, EnterpriseUnits – 0.91MDRAM – 24,581MBDRAM Revenue - $2,684MSource: iSuppli

How is DRAM used today?

Page 22: The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Baker 22

Units – 15.7MDRAM – 8MBFlash – 2MB

Cable Modems

Cellular Base StationsUnits - 0.35MDRAM - 256MB Flash – 5112MB

Units – 113.5MDRAM – 15MB

LAN Switches Low­End to Mid­Range RoutersUnits – 1.5MDRAM – 406MB

Communications and Networking Market

Units – 62MDRAM – 56MB

Set­Top Boxes Cellular PhonesUnits - 740MDRAM – 5MBFlash – 87MB

DSL ModemsUnits – 55MDRAM – 8MBFlash –2MB

Sources: iSuppli, Gartner, Portelligent, Instat

Page 23: The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Baker 23

DRAM Demand by End-Use Application

Desktop PC

Mobile/Notebooks

Upgrade Modules

Other Data Processing*

PC Graphics

Wired Communications

ConsumerIndustrial

0%

20%

40%

60%

80%

100%

2004 2005 2006 2007 2008 2009

Perc

ent

of T

otal

Meg

abit

s

Automotive

Mobile Communications

Entry ­Level Servers (x86)

*Other Data Processing includes mainframe servers, enterprise servers, workstations, handheld PCs, storage cards, printers, and internet appliances

Page 24: The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Baker 24

Worldwide DRAM Shipments by Type

FP/EDO16Mb SDRAM

64Mb EDO 64Mb SDRAM

64Mb DDR128Mb SDR

128Mb DDR

128MbRDRAM

256Mb SD

256Mb DDR

512Mb DDR

512 Mb DDR2

512MbDDR3

1GbDDR2 1Gb DDR3

2Gb DDR3

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

2002 2003 2004 2005 2006 2007 2008 2009

Perc

ent

of T

otal

Meg

abyt

es P

rodu

ced

Page 25: The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Baker 25

DRAM Technology Trends

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

2002 2003 2004 2005 2006 2007 2008 2009 2010

Uni

ts

DDR3

DDR2

DDR

SDRAM

Page 26: The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Baker 26

DRAM Density Trends

Source: IDC, Isuppli, Gartner, IC Insights Q106

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

2002 2003 2004 2005 2006 2007 2008 2009 2010

Uni

ts

4 Gb

2 Gb

1 Gb

512 Mb

256 Mb

128 Mb

64 Mb

16 Mb

Page 27: The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Baker 27

Historical PC Market Growth

0

10,000,000

20,000,000

30,000,000

40,000,000

50,000,000

60,000,0001Q

953Q

951Q

963Q

961Q

973Q

971Q

983Q

981Q

993Q

991Q

003Q

001Q

013Q

011Q

023Q

021Q

03

3Q03

1Q04

3Q04

1Q05

3Q05

E

Uni

ts i

n M

illi

ons

-15%-10%-5%0%5%10%15%20%25%30%35%

Year

ove

r Ye

ar G

row

th

Grand Total Grand Total

Page 28: The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Baker 28

Price per Bit Cycles

2001

1997

1974

1976

1977

19881986

1979

19801978

1981

1982

1983

19841985 1987

1999

2003

2004

20052002

20001998

1996

1994

1995

1992

19931991

1990

1989

1975

0.01

0.10

1.00

10.00

100.00

1,000.00

10,000.00

100,000.00

Pric

e pe

r Bi

t (M

ilic

ents

)

1

100

10,0

00

1,00

0,00

0

100,

000,

000

Cumulative Bit Volume (1012)

Source: Gartner, 05/05

Historically Price per Bit has declined by an average of 9%

every quarter

(1974-2005)

Page 29: The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Baker 29

Americas DRAM Unit Shipments by Density

200.0

400.0

600.0

800.0

1,000.0

1,200.0

1,400.0

1,600.0

1,800.0

2,000.0

2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

Mill

ions

of U

nits

1Mbit 4Mbit 16Mbit 64Mbit 128Mbit 256Mbit 512Mbit 1Gbit 2Gbit 4Gbit 8Gbit

1Gb

512Mb

256Mb

128Mb

64Mb

Page 30: The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Baker 30

256Mb Equivalent Unit DRAM Shipments

256Mb Equivalent Units Shipped(K Units)

0

100,000

200,000

300,000

400,000

500,000

600,000

700,000

Apr-

98Ju

l-98

Oct

-98

Jan-

99Ap

r-99

Jul-

99O

ct-9

9Ja

n-00

Apr-

00Ju

l-00

Oct

-00

Jan-

01Ap

r-01

Jul-

01O

ct-0

1Ja

n-02

Apr-

02Ju

l-02

Oct

-02

Jan-

03Ap

r-03

Jul-

03O

ct-0

3Ja

n-04

Apr-

04Ju

l-04

Oct

-04

Jan-

05Ap

r-05

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%Total 256Mb EU 12/12 Growth

Page 31: The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Baker 31

DRAM Applications withthe Highest Unit Growth Rates

0% 10% 20% 30% 40% 50% 60%

DVD Video Recorder

Digital TVs

Flash Storage Cards

MP3 player

SAN Switches

Wireless LAN Access Point/Bridge

Wireless LAN Client Access

Mobile PCs

Internet Appliances

Entry-Level PC Servers

(CAGR 2004 – 2008)

Page 32: The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Baker 32

Fastest Growing DRAM Consumption

0% 20% 40% 60% 80% 100%

Mobile Handsets

Remote Access RAC & RAS

Internet Appliances

Workstations

Entry-Level PC Servers

Deskbound PCs

Mobile PCs

Handheld Computers

Upgrade Module

Enterprise Servers(CAGR 2004 – 2009)

Page 33: The One-Transistor, One-Capacitor (1T1C) Dynamic Random

Baker 33

Summary Overview

DRAM chips are found in virtually every computer in use today. Looking at the simplicity of the 1T1C cell one might wonder about the significance of this invention by modern day standards. However, if it is remembered that this idea was conceived back in the 1960s before MOS technology had matured enough for production or the idea that circuits could be “dynamic” (only operating correctly for a short period of time) the significance of the invention becomes clear. What is usually not mentioned when talking about Dr. Dennard’scontributions to the 1T1C memory cell is his contributions of seeing this idea to product (the true test for any practical electrical engineer). The MOS process development (and the concerns for defects and reliability which have a drastic effect on the dynamic operation of MOS circuits) and supporting circuitry are also extremely important contributions that Dr. Dennard made while at IBM. In summary, Dr. Dennard cut the path for modern DRAM memory chip developments.