Title: PIDS TWG Report on the ITRS Roadmap issues December 6, 2000
1PIDS TWG Report on the ITRS Roadmap
issuesDecember 6, 2000
Participants J. P. Bruines (Philips-Europe), M.
Yoshimi (Toshiba-Japan), D. H. Lee
(Samsung-Korea), P. M. Zeitzoff (International
SEMATECH-USA), J. Y. C. Sun (TSMC-Taiwan)
2Outline
- 2000 update
- Novel Devices Table
- Key 2001 issues
- Summary
32000 Update, PIDS Modifications
- Sped up scaling of MPU physical gate length (Lg)
- Lg specification pulled in by one year, but only
through 2001 - Other rows (Vdd, Tox, Ioff, etc.) pulled in by
one year, for consistency - In response to IRC recommendations and products
shipped/announced - IRC recommendation one year pull-in til 2005
and beyond - PIDS did not do pull-in beyond 2001
- Too many issues for 2000 update, particularly for
gate dielectric - ASIC Lg same as MPU
- 1999 ITRS MPU Lg lt ASIC Lg
4Key PIDS Table Modifications 2000
5Impact of Lg Pull In Beyond 2001
62000 Update, PIDS Modifications (cont.)
- DRAM specifications unchanged (DRAM half
pitchgttechnology node) - Inadequate time to make changes per IRC
recommendations to scale more rapidly - Key issue for 2001 ITRS sped up scaling
- Other Modifications
- MPU/ASIC Physical Gate Length (Lg) defined
as-etched, final length at bottom of gate
electrode - New separate ORTC Litho specifications for
gate length in photoresist - Relatively minor clarifications, corrections,
updates
7Novel Devices Tables
- Novel Devices tables added to Roadmap in 2000
- Novel devices when scaled bulk CMOS becomes
inadequate - Lists devices, description, potential advantages,
major challenges, and possible insertion node - Different novel device types
- Non-bulk CMOS are MOS type devices
- Novel devices are electronic devices, but
beyond MOS type - Novel storage are novel storage devices
- Novel systems are new systems, requiring novel
architectures
8Key PIDS Issues for 2001 ITRS
- Memory and logic
- Impact of accelerated scaling
- MOSFET scaling, process, other issues
- Particularly, availability of gate dielectric for
lt65 nm Lg - DRAM cell size, equivalent Tox, leakage, etc.
- Circuit, systems issues (with Design TWG)
- Impact of gate leakage on logic, including
dynamic logic - Gate resistance
- Static and overall power dissipation
- Meeting device requirements at Low Vdd
- Develop new rows for Flash and FERAM (with FEP
TWG) - Expand low power logic requirements (with FEP
TWG) - New, extremely low power (handheld, wireless)
category - Impact on gate leakage, gate dielectrics
9IRC/Litho. Scaling Proposals for 2001 ITRS
ITRS Roadmap Acceleration Continues...
(Including MPU/ASIC
Physical Gate Length
Proposal
)
500
500
1994
Note MPU/ASIC Physical Bottom Gate Length
350
350
still under discussion.
250
250
(DRAM Half Pitch)
Technology Node
1997
180
180
1998/1999
MPU/ASIC
130
130
2000/2001
Gate Length
Feature Size (nm)
100
100
XX
Technology Node - DRAM Half-Pitch (nm)
Minimum
90
70
XX
70
Feature Size
65
50
XX
50
45
35
XX
35
MPU Gate
1999 ITRS In Resist 2000 ITRS
33
25
XX
25
23
16
.7x per
Year of Production
technology
2001 Renewal Period
node (.5x
REV 1kg_g - 10/20/00
per 2 nodes)
10Key PIDS Issues for 2001 ITRS (cont)
- Analog/RF
- General review of process/integration issues
- Issues with high K coupling, bypass, and other
capacitors (with Interconnect TWG) - Need for precision passive elements
- Impact of gate leakage
- SiGe
- Reliability
- Review possible burn-in requirements and SOC
requirements - Management of reliability issues with many new
materials
11Summary
- 2000 update gate length scaling accelerated
through 2001 - Other, minor updates and changes
- Novel device table added
- 2001 ITRS
- Further acceleration of MPU and DRAM scaling
proposed - Difficult issues, especially with gate dielectric
- Many important issues need attention
- Inter-TWG cooperation is vital
- Add FERAM, flash memory, low standby power,
possibly MRAM
12FOLLOWING ARE BACKUP FOILS
13Issues with Accelerated Scaling, 2001 ITRS
- Proposed by IRC Litho further acceleration in
MPU scaling compared to 2000 ITRS - DRAM 0.7X per node (every 3 years)
- Results slightly faster than one year pull in
from 99 ITRS - MPU gate length (Lg)
- In photoresist Lg 0.7X DRAM half pitch
- Lg (physical) lt Lg (in photoresist) gt etch bias
- Result two year pull in from 99 ITRS, one year
from 2000 ITRS - Issue 1 early onset of process and control
difficulties, including - MPU CD error budget, control of etch bias
- DRAM litho, etch, and difficult high K dep.
(high A/R) - Very shallow, heavily doped junctions
14IRC/Litho. Scaling Proposals for 2001 ITRS
ITRS Roadmap Acceleration Continues...
(Including MPU/ASIC
Physical Gate Length
Proposal
)
500
500
1994
Note MPU/ASIC Physical Bottom Gate Length
350
350
still under discussion.
250
250
(DRAM Half Pitch)
Technology Node
1997
180
180
1998/1999
MPU/ASIC
130
130
2000/2001
Gate Length
Feature Size (nm)
100
100
XX
Technology Node - DRAM Half-Pitch (nm)
Minimum
90
70
XX
70
Feature Size
65
50
XX
50
45
35
XX
35
MPU Gate
1999 ITRS In Resist 2000 ITRS
33
25
XX
25
23
16
.7x per
Year of Production
technology
2001 Renewal Period
node (.5x
REV 1kg_g - 10/20/00
per 2 nodes)
15Accelerated Scaling Proposals for 2001 ITRS
16Issues with Accelerated Scaling, 2001 ITRS (cont)
- Issue 2 early onset of difficult MOSFETscaling
issues, including - High K gate dielectric may be required by
2003-2004 - Especially problematical for low power technology
- Low contact resistance, low sheet resistance
silicide - Channel engineering controlling short channel
effects - Control of Vt statistical variation
- Issue 3 early onset of difficult DRAM scaling
issues, including - New High K dielectric materials
- Leakage current
- Major re-evaluation of scaling, process, and
process control issues will be needed
17Tox Scaling and High K Gate Dielectric