Professional Documents
Culture Documents
32/22Nm Metrology: European Smes Come On Board
32/22Nm Metrology: European Smes Come On Board
"
News from the European 32/22nm CMOS Research pool at month 21 April 2008
EDITORIAL VITeam
Pullnano has certainly led the 32nm progress in order to stabilize the process
CMOS battle for Europe. Our competi- window at smaller nodes. In order to face
tive 32nm SRAM cell demonstrator, pre-
this challenge, the Pullnano Consortium
sented at IEDM 2007, is the world’s first
with a Fully Depleted Silicon On Insula- asked three SMEs to come to the rescue.
tor (FDSOI)/High K dielectric/Metal Gate An overview with Martin Schellenberger,
architecture. At 32nm design rules this
memory cell translates into a density of coordinator of the equipment subproject
33 million transistors per mm2! for Pullnano at the Fraunhofer IISB.
The withdrawal of NXP and Freescale
from the Crolles2 Alliance (as of Decem-
ber 07) with implications on the Small is beautiful ... but tiny semiconduc-
Pullnano consortium partnership left Martin Schellenberger, coordinator
former Crolles 2 Alliance members in tors pushing back the limits of miniaturi-
for the SP7, Fraunhofer IISB : « Pull-
N a n o C M O S
search of a new world size alliance. zation are not only a challenge for the
nano tells the process directions and
Elsewhere large alliances are forming as fabrication process with smaller struc- contributes to the process specifica-
well. Nevertheless work towards the tures, new materials and new device ar- tions for SEA-NET. In this case, the
32 nm node continues and will accele-
chitectures. They are also a challenge for specification is the know-how co-
rate, since no single CMOS technology
can “do it all” given the variety of appli- the analysis and characterization equip- ming from Pullnano, and we will get
cations and markets accessible to large ment, a prerequisite to be able to develop the results feedback from SEA-NET. »
chip manufacturers. Know-how, soft- and produce the nanostructures down to
ware tools methodologies, models, cha-
the 32/22nm nodes. As Pullnano results
racterization techniques and equipment
improvements developed through are developed towards the industrial ex- developing the necessary process and
Pullnano will be transferred to/used in ploitation phase, the equipment forum, metrology equipment, with a special
additional technologies known as More active in subproject 7 (SP7) and coordina- focus on 32/22nm capabilities related to
o f
Than Moore technologies (RF, analog, ted by the Fraunhofer Institute of Integra- the Pullnano needs », points out Martin
embedded memories…), optoelectro-
nics or NEMS (NanoElectroMechanical ted Systems and Device Schellenberger
l i m i t s
Systems). This will allow a smoother Technology (IISB) in Erlan- from the Fraunhofer
transition to 32nm manufacturing in gen, Germany, scored a Institute. In Pull-
Europe. few important achieve- nano, the equip-
During the final year of the Pullnano ments. « It is the overall ob- ment activities are
project the consortium’s attention is jective of the SP7 especially suppor-
focusing more and more towards 22nm The NovaScan platform is a metrology
equipment forum to opti- ted by new part-
devices; preliminary work shows that solution for critical dimensions and
with creative and bold device architec- mize and smooth the pro- shape profiling applications down to ners, Cameca and
t h e
ture the 22 nm node can offer the cedure of designing and 32nm (doc. Nova) Imagine Optic from
performances expected in the ITRS
(International Technology Roadmap for
Semiconductors). Certainly one of the Table of contents
options studied within Pullnano will
P u l l i n g
emerge in the final industrial 22nm tech- MASTAR : A European Enrico Sangiorgi :
nology. Good news: at 22nm the
technical aspect seems to work; if the compact model enters “Electronic industry is in
economics also works, we should then the international stage p.3 the midst of revolutionizing
see the 16nm node happen …
the traditional CMOS device” p.4
Dr Gilles Thomas (STMicroelectronics) High-K material
Pullnano coordinator Will Gadolinium succeed Dissemination p.6
to Hafnium ? p.6
1
VITeam
M a r t i n S c h e l l e n b e r g e r W P 7 E q u i p m e n t s
2
achieved resolution and ac- capabilities to be made availa-
curacy of 1.1mm for lateral ble for fabrication of necessary
resolution and 50nm in verti- Georges Antier, Vice President and standards and establishment of
cal resolution. Business Unit Manager of Cameca : metrology methodologies in ad-
« Ultra Low Impact Energy is a key vance of production. This re-
Finally, a Fraunhofer IISB factor to get access to the ultimate quires a greater attention to
team is also working toge- depth resolution in SIMS. This is a strong re- expanding close ties between
ther with STMicroelectro- metrology development and
quest from all Cameca customers and from
nics on developing an process development. The co-
automated fault detection semiconductor analytical teams in general. operation between the metro-
and classification solution The work done in Pullnano was very fruitful logy community and
for low-k (2.5 or less) pore for Cameca as we were able to demonstrate semiconductor fabrication is
led at the sealing processes. The extremely good data with both Cesium and mandatory to keep equipment
(doc. Cameca) control solution is based on Oxygen primary ion beam. » development in Europe. But,
residual gas analysis (RGA), when the metrology is well mat-
and can be used in R&D as well as ched to the process tools and
in production. control and X-ray metrology and origina- processes, ramping times for pilot lines
ting from ideas in the Pullnano domain, and factories are reduced. Later on, an ap-
« In parallel to the Pullnano initiative, the are funded via SEA-NET » propriate combination of well-engineered
equipment forum worked to establish tight tools and appropriate metrology is neces-
contacts to the independent european In general, the metrology tool development sary to maximize productivity while main-
project for the assessment of semicon- requires the availability of state-of-the-art taining acceptable cost of ownership ■
ductor equipment for nanoelectronic
technologies (SEA-NET), where the
activities focus more on equipment David Scheiner, CTO of Nova Measuring Instruments :
testing on a large scale, versus Pull- « After gaining access to cooperation with IMEC, we will have a
nano where the focus lies on the unique application to offer our customers and gain a certain advan-
materials and process technolo- tage in relation to our competitors from the USA if the project is
gies », explains Martin Schellenber-
successful, and we hope to continue cooperation with IMEC and possibly
ger. « Currently, two projects,
related to advanced overlay install a metrology system to continue the development of this capability. »
VIProduct
3
VIP
Enrico Sangiorgi :
“Electronic industry is in the midst
of revolutionizing the traditional
CMOS device”
The continued scaling of CMOS technology is reaching a point where, due to
physical limits, new materials and device structures increasingly need to be introdu-
ced, requiring an extension of devices and circuits models in order to provide
solutions in time for forthcoming technology generations. The Device Physics cluster Pioneering the modeling of
in Pullnano is in charge of this task. Explanations by Enrico Sangiorgi. nanoscale chips…
Enrico Sangiorgi has been invol-
Why does electrical modeling become important role by developing and testing ad- ved in several European projects
more important in the 32nm and below pro- vanced physical models addressing the future of the 5th, 6th, and now 7th FP
cesses? needs. In particular all the issues related to the with management responsibilities.
As pointed out in the ITRS roadmap, device usage of strained silicon, the adoption of dif- He is currently a member of the
Executive-Scientific Committee of
characterization, modeling and simulation will ferent crystal orientations for the MOSFET
the Nanosil Network of Excel-
face several challenges for the 32 and 22nm channel, the use of high-k gate dielectrics, the lence, managing the modeling
CMOS device development. In fact, the tech- fluctuations of dopants, the dielectric and sili- activities, and a member of the
nological innovations introduced by the 32nm, con layer thickness, and the effects of line- strategic board of the Integrated
and even more by the 22nm process, may edge roughness, as well as their impact on the Project Pullnano. Since 2006 he is
very well revolutionize the traditional CMOS electrical parameters of the devices must be also the Vice Chairman of the
device scenario. Today, the number of tech- investigated. Scientific Community Council
(SCC) of the Eniac Technology
nological options is astonishingly large and in-
Platform and since 2007 a mem-
cludes both new materials and new device What is the contribution of Pullnano to this ber of the Steering Board of the
structures, like high-k insulators in various activity ? Eniac-related Aeneas Association.
fashion with metal gate, ultra-thin Schottky The Pullnano project has devoted an unpre- He is currently in charge of the
contact Source and Drain, uniaxial or biaxial cedented effort toward the development of nano-microelectronics group of
strained silicon channel, Single or Double- advanced physical models, with many acade- the University of Bologna –
Gate SOI devices, FINFET, Omega-Gate, and mic teams grouped together to form a “De- Campus of Cesena. He is also the
Multi-Gate FET in general. The introduction of vice Physics” Cluster with ambitious yet clear Director of the Italian Universities
Nanoelectronic Team (IU.NET),
some of these highly innovative structures at goals. Once implemented in device simula-
which is a legal consortium grou-
the 32 and 22nm nodes will be a key element tors, the developed models will allow evalua- ping eight university groups active
to master the development of a competitive ting and comparing different device in the field of nanoelectronics. His
new CMOS process by boosting performance architectures and technological options, as research interests include the
and cost reduction at the same time. well as permit the calibration of the new com- physics, characterization, mode-
However, in the attempt to reach this goal, pact models needed for engineer-oriented ling, and fabrication of silicon
tools like MASTAR. solid-state devices and integrated
circuits. In order to tackle and
At approximately two thirds of the way along
eventually overcome the hurdles
the 30-months duration of the project, we
of device scaling, down to the ulti-
have carried out a thorough fundamental work mate physical and technological
to understand effects like the electron mobility limits, he has devised and develo-
in presence of a high-k interface and different ped several original concepts and
strained silicon channels, the degree of bal- methods in the characterization
listic transport, which is an important source and modeling of nanoscale silicon
of performance boost, as a function of the devices.
A grain boundary parallel to the source/drain Enrico Sangiorgi coauthored 33
junction creates a potential barrier that in- channel material and device structure, the im-
creases the threshold voltage right across the IEDM papers and published more
pact of detrimental effects like off-currents of than 150 papers in international
width of the device. (doc. University of Glasgow)
various origins, gate current, band-to-band journals and conference procee-
technologists face a double threat : the num- tunneling, etc., as a function of the different dings. He is a Fellow of the IEEE,
ber of available technological choices is too technological options, and, last but not least, a Distinguished Lecturer of the
wide to allow a “try and repeat” approach and the modeling of the holes in the channel of the Electron Device Society and
the physical mechanisms behind the possible p-MOSFETs, taking into account the quantum Chairman of the Electron Device
advantages of new materials and new struc- mechanical effects induced by quantization in Society TCAD Technical Com-
mittee. He has been Editor of
tures are largely unknown. Here is where phy- the channel well.
IEEE ELECTRON DEVICE
sically based electrical modeling can play an Another area where the Pullnano modeling LETTERS since 1994.
4
E n r i c o S a n g i o r g i
team has made impor-
tant progress, is the
study of the effects of
statistical device variabi-
lity due to intrinsic
sources, e.g. discrete
dopant distribution, line
edge roughness, etc.
This issue is becoming crucial because at na-
A 3-D analysis shows the impact of surface potential pinning and doping nonuniformity in
noscale dimensions the matter itself become
the gate and the dependance on grain boundary orientation (Poly-Si gate variability) :
discrete! The team at the University of Glas- (a) pinning in the middle of the band gap ; (b) increased doping concentration around the
gow, a world authority in this field and part of grain boundaries ; (c) combination of (a) and (b). (doc. University of Glasgow)
the Pullnano Device Physics Cluster, has des-
cribed for the first time in two recent publica- the best technological options (materials and Could you give details about the impact of
tions originated by Pullnano, the effect of the device architecture) available for the 32 and IUNET* on Pullnano ?
Poly-Si gate related variability in decanano- 22nm nodes. Last but not least, the results of The IUNET scientists are strongly contributing
metre MOSFETs with conventional architec- the device simulation will be “lumped” in up- to most aspects of the modeling effort in the
ture and the quantitative evaluation of graded compact models, starting with MA- Pullnano project and share the same spirit of
statistical variability sources in a 45nm tech- STAR which is a handy tool for engineers who cooperation which has been the key aspect
nology. want to compare the influence of different sustaining the success of IUNET with all the
At the same time, we are working on the stan- technological choices on device performance. other members.
dardization of the so-called “template de-
vices”, i.e. “ideal” structures, used by all
Unstrained silicon Strained silicon
partners to implement and test the models
and at the same time fabricated by the main
technological Pullnano partners to provide the
necessary experimental feedback to the theo-
retical modeling effort.