Transcript Greatest challenges of the 21st Century:

Greatest challenges of the 21st Century:
To create computing capability that can operate with THz speed
with Terabits/cm2 information storage, and to apply this
technology in biotechnology, business, and education
•Speed drives technology
•Technology drives society
“Terascale electronics---endless quest
for IC speed”
Toh-Ming Lu
[email protected]; www.rpi.edu/~lut
Director
Center for Advanced Interconnect Science and Technology
(RPI, SUNY-Albany, MIT, UT-Austin, N. Texas, Texas Tech., Cornell,
UC Berkeley, Columbia, Georgia Tech, Rochester, U. of Maryland)
Outline
•End of scaling
•Systems technologies: on-chip/off-chip interconnect
•Nanoelectronics
Information age?
Execution, storage, and transmission of massive information
What technology drives the information age?
Hardware in Computer:
Chips, hard drives, display…….
-----Microelectronics technology
Technology for Information age
--------Microelectronics
 Electronics industry: driving force of the information age.
largest manufacturing industry in the United States and in
the developed world
 Over ~14% per year growth in the last 30 years --continue to grow in the next few decades
 Will need a continuing supply of BS, MS, and Ph.D
Who are chip makers?
•Intel, IBM, Motorola, AMD, DEC, LSI Logic,
National Semiconductor, Lucent, TI, HP…..
•DELL, Compaq, Gateway…don’t make chips!
History
Invention of solid state electronics (40’s)
--------The transistors
Then IC
Then Mainframe (60’s)----execution and
storage
Then PC (70’s)----execution and storage
Then PC plus internet plus WWW
-----execution, storage, and transmission
Why so exciting?
• Intellectually stimulating
• Impact: changes the society in major way
– Business: creates enormous wealth
– Education: fundamentally change the way
we learn
– Medicine: will change the way we treat
diseases
“Turmoil and opportunities at the dawn of the 21st Century
---the road of an academic department in higher education”
(Toh-Ming Lu, amazon.com, 2000)
Computer logic
---a series of on and off operation (clicks)
time
 Imaging a super fast telegraph!
(1GHz: 1000 million clicks per second; 1THz = 1000GHz)
 “Fast” means: more clicks per second
 The narrower the “click” the faster you get
 The shorter the device the narrower the click
MOSFET Transistor
Key questions in the industry
Technical:
---Is there an end to increase IC speed?
Business:
---Is there a market for super fast ICs?
Some key technological challenges
---Limit on device dimension
---Limit on interconnect speed
Recent news:
•Intel: 1 THz FET, 25nm channel length
•IBM: 210 GHz HBT, base 100 atoms
Limits on patterning: diffraction
resist
metal
Limit on RC delay
Chip cross section
Interconnect (RC) delay
Through wires
time
To avoid overlapping
Reduce the number of “clicks” per second
---separate the “clicks” apart
Therefore reduce the speed
time
“Terascale electronics---endless quest
for IC speed”
Toh-Ming Lu
[email protected]; www.rpi.edu/~lut
Director
Center for Advanced Interconnect Science and Technology
(RPI, SUNY-Albany, MIT, UT-Austin, N. Texas, Texas Tech., Cornell,
UC Berkeley, Columbia, Georgia Tech, Rochester, U. of Maryland)
Outline
•End of scaling
•Systems technologies: on-chip/off-chip interconnect
•Nanoelectronics
3D heterogeneous systems: bonding, alignment, via etching/filling
High bandwidth:
to optoelectronics
systems (THz)
?
Mitsubishi Electronics America: ADVANCED PACKAGING
June/July 2000 issue.
A/D, sensors, IP cores
--killer tech
memory
PC, communications,
internect…
GaAs/Si?
Logic layers
I/O, passives, power
Heat extractors
--Shorter wires, higher density, more functionalility—
Beyond Roadmap
Opportunities for more Si mainstream technologies:
---Decades beyond the Roadmap
Stacked chip assemblies (logic, memories, interposer for
passives);
Heterogeneous systems for sensors and MEMS;
Hard IP core-based SOC designs (including mixed signal);
High speed processors;
LAN architectures (for wireless applications and/or for
multiplexed interconnects).
Gutmann et al (2001)
J. Lu et al
Pictorial Representation of 3D Integration Concept
using Wafer Bonding,
Via Bridge
Via Plug
Substrate
Device
Surface
Third Level
(Thinned
Substrate)
Bond
(Face-to-back)
Substrate
Second Level
(Thinned
Substrate)
Device
Surface
Bond
(Face-to-face)
First Level
Device
Surface
Substrate
* Figure adapted from IBM Corporation and used with permission.
•Processing issues:
bonding-alignment
through wafer via etching
barrier and metallization
•Reliability:
thermo and mechanical stability
electromigration
heat extraction
Broad band interconnect technology
---high speed data transfer
Or: wireless!
Replacing electrical connection by optics:
•Modulators/switches: electro-optic, optic-optic
•Optical waveguides
•Data compression (software)
Modulators guide
switches
light
fiber
Chip stack
Nonlinear EO
Electro-optic
modulator
Modulated
light
light
Electrical signal
A
B
heterostructure
layer
C
substrate
d
R. Kersting, G. Stasser, and K. Unterrainer,
Terahertz phase modulator, Electr. Lett., 36,
1156 (2000)
modulating signal: B
from TH
z
source: A
A
0
0
1
1
B
0
1
0
1
C
0
0
0
1
read out: C
Mach-Zehnder R
ing
Optical switches:
•MEMS---mirror switches: D. Bishop et al, Physics Today Oct 2001 (Lucent)
•Nanotube switches: Zao et al (2001)---THz speed
•Quantum dots switches: Dutta et al (2001)---THz speed
MEMS
Potentially viable optical interconnect schemes—Dr. Persans
waveguide
CMOS circuits and
metallization
optoelectronic transceivers
metal or multilayer dielectric mirror
cladding
waveguide
via
receiver
• Bump-bond optoelectronic chip on top of
complete CMOS package
• Grow optoelectronic components
monolithically; local microphotonic waveguides
grown and patterned; polymer waveguide
layers for off-chip and longer distance
• Monolithic optoelectronic components;
incorporate longer waveguides into metal
interconnect package
• Use waveguides within sensor-chip or systemon-a-chip paradigm
Agarwal, Ponoth, Plawsky, Persans: Appl. Phys. Lett. 78, 2294 (2001)
Mainstream computer/communication technology:
•Strong industrial/State/Federal partner support
•Enormous employment opportunities
•Decades of growth---expected more growth in decades
End of device scaling does not
imply end of Si technologies!
Emerging technologies
•Nano-scale electronics: very rich and unexplored science
•Strong Government support
•Long term benefits (not likely mainstream computing
in at least 20 years)
---The greatest and immediate impact may not be in electronics,
but in biomedical applications
“Terascale electronics---endless quest
for IC speed”
Toh-Ming Lu
[email protected]; www.rpi.edu/~lut
Director
Center for Advanced Interconnect Science and Technology
(RPI, SUNY-Albany, MIT, UT-Austin, N. Texas, Texas Tech., Cornell,
UC Berkeley, Columbia, Georgia Tech, Rochester, U. of Maryland)
Outline
•End of scaling
•Systems technologies: on-chip/off-chip interconnect
•Nanoelectronics
Interconnects via Terahertz
Receiver-transmitter pairs
Deep sub-0.1
Ug
2D
electron
fluid
Gate
Source
Gate
Insulator
Drain
Plasma wave
• ULSI chip divided in tiles
• Communicate via plasma wave electronics receiver-transmitter pairs
Michael S. Shur
http://nina.ecse.rpi.edu/shur/
Room temperature single electron
transistor using nanotube
Smalley group (2001)
Oriented & interconnected nanotube networks—Ajayan et al
Focused Ions
Catalyst
Junctions
– Local modification and Junction formation
– Termination (cutting of structures)
Fantastic opportunities in applied and basic
science research
Examples:
New materials synthesis: polymers; nitrides, carbides
Novel polymer-metal, polymer-cermic, polymer-polymer composites:
Novel phase separation, crystallization, dynamic growth phenomena
Novel interfacial diffusion, reactions, and transformations
Novel nano-structure science; light emitting nano semiconductors
Novel non-linear thin film materials; high electro-optic coefficient
materials
Novel opto-electronics materials, layered structures
Quantum effect on narrow lines
Materials response under extreme speed and frequency
Real time atomic scale microscopies
THz Bio-Chip for Sensitive Detection
Nano-Si or nano-C layer
reference chip
sample chip
THz grating
helps coupling
THz wave
THz signature or fingerprint of genetic materials: DNA, RNA or
Protein attach to nano-layer in sample chip, from 10 GHz to 10 THz
frequency range. (Zhang, Kersting)
•Welcome students doing PhD at Rensselaer
•Welcome visiting scholars and collaboration
Yesterday
Today
“Norton”
Facility
(IC
Laboratory)
Tomorrow
Terascale
5”-8” 2µ CMOS
electronics?
MCR
Bio-devices?
(Wafer Processing
R&D)
8” State-of-the-Art Wafer Fab
Electrical Storages
•Memory:Trenches/Stacked capacitors
•Passives capacitors
Magnetic storage
---towards terabits/in2
C. Ross, Annu. Rev. Mater.
Res. 2001. 31:203-235.
Three strategies:
• exchange-decoupled grains
(conventional)
•In-plane patterned media
•Perpendicular patterned media
Limits on magnetization:
---Nayak/Wang/Korniss (Physics)
?
P
H
P
H
Molecular memories:
Materials Today
(Feb 2002)