Добавил:
Upload Опубликованный материал нарушает ваши авторские права? Сообщите нам.
Вуз: Предмет: Файл:

Книги2 / 1993 P._Lloyd,__C._C._McAndrew,__M._J._McLennan,__S._N

.pdf
Скачиваний:
51
Добавлен:
11.03.2016
Размер:
14.08 Mб
Скачать

270

K. Nishi et a1.: Technology CAD at OK!

Fig.19:Simulated structure of SATURN by OPUS

Fig.20:Current distribution simulated by ODESA.

lower in 3D simulation. These kind of 3D simulation is possible only by a coupled 3D process/device simulation.

4.3 Unified simulations of deep-submicron CMOS circuits

 

 

 

As

 

a final example, we will describe application of UNISAS

to the

design of half-micron

CMOS technology. Main purpose

was

to investigate the effects of narrower spacer[14J to tpd,

and

also

to

 

get an optimal process conditions with respect to

gate

oxide

thickness(toxl

and VT .

Table

3 summarizes the

simulated

samples.

First, process conditions are roughly adjusted to

obtain

desired VT .

Both shallow and

deep implantation

are

done

for VT control. Care is taken so as to reduce VT implantation conditions. Simulated and measured VT are also shown in Table 3,

and show

good agreements.

Fig.21 shows simulated and measured

saturation currents (IDS l .

Significant improvements in IDS are

observed

for narrower spacers compared to a wider spaqer. This

is due to

smaller resistance

in LDD.

K. Nishi et al.: Technology CAD at OK!

 

 

 

 

 

 

 

271

 

Finally, tpd' s of an 2-NAND

gate

are shown

in

Fig.

22.

tpd

is usually deeply correlated with IDS'

For different t ox ,

howev-

er,

higher

gate

and junction

capacitance may

cause

tpd

reduction

in

thinner

t ox ,

which appear

in

sample

Band

E.

In

any

case, we

see an excellent agreement between simulations and experiments. CPU time for the simulation of one sample on 30-MIPS machine

is 2 hours for process simulation and 2.5 hours for device simulations. CPU time for circuit simulation of a simple 2-NAND circuits is negligible. Noted is that these simulations can be done without any interruption by users. Thus, only one day was necessary in order to get the thorough results. Excellent agreement between experiments and simulations and reasonable CPU time indicate the practicability of the system, and can be the key

t~chnology to minimize

the

development cost of future VLSI's.

 

Table 3:Simulated samples.

 

 

 

 

VT(sim.)

VT(experi.)

SAMPLE

SPACER

tOX(A)

nMOS

pMOS

nMOS

pMOS

A

WIDE

120

0.59

0.72

0.58

0.77

B

NARROW

120

0.53

0.53

0.52

0.53

C

NARROW

120

0.60

0.68

0.60

0.70

D

NARROW

120

0.65

0.81

0.70

0.84

E

NARROW

100

0.52

0.65

0.51

0.66

F

NARROW

100

0.57

0.82

0.60

0.82

500

 

nMOS

 

 

E

400

 

-. EXPERI.

 

 

 

 

 

~

 

 

 

~

300

 

-0- SIM.

-~

200~

-+- EXPERI.

o

CI)

 

+

 

 

 

 

10:~

 

-<>-- SIM.

 

 

 

 

ABC D E

F

 

 

SAMPLE

 

 

Fig.21:Simulated and measured saturation currents (IDS)

272

K. Nishi et al.: Technology CAD at OK!

-CI)

..e:

"0

....a..

140

130

120

110

100

A

B

C

0

E

F

SAMPLE

Fig.22:tpd's of an 2-NAND gate.

5.Summary

TCAD system at OKI, UNISAS, was described. UNISAS is a unified process/device/circuit simulation system, and offers a user-friendly simulation environment. Multi-dimensional process and device simulators, OPUS and ODESA constitutes the core of UNISAS and can be used for versatile device structures for various purposes. Physics-based simulators constitute another part of UNISAS, and are used for modeling purpose. Some applications are described which show good examples of the effectiveness of

UNISAS.

UNISAS is

already used quite extensively by engineers

for

actual device

development and is now an indispensable· tool

for

low-cost, fast

TAT VLSI development.

Acknowledgements

 

 

The

authors would

like to thank S.Kuroda, K.Fukuda, K.Kai

for

their

simulations.

The authors also would like to J.Ida for

his

experiments.

 

K. Nishi et al.: Technology CAD at OKI

273

 

References

[l]:D.A.Antoniadis and R.W.Dutton, IEEE Trans. Electron Devices, ED-26, 490 (1979)

[2]:K.Sakamoto, K.Nishi, T.Yamaji, T.Miyoshi and S.Ushio, J.Electrochem. Soc. 132, 2457 (1985). First report appeard in The Electrochem. Soc. Extended abstracts, Vol.81-2, 411, 1981 by K.Nishi, K.Sakamoto, and S.Ushio.

[3]:J.Ueda, Y.Namba, K.Sakamoto, T.Miyoshi and S.Ushio, J.lnst. Elec.Commun.Eng.Japan, J67-C, 825, (1984)

[4]:K.Nishi, K.Sakamoto, S.Kuroda, J.Ueda, T.Miyoshi and S.Ushio, IEEE Trans. Computer-Aided Des., CAD-8, 23, 1989

[5]:H.Kajitani N.Shimizu, K.Fukuda, S.Baba, K.Nishi and J.Ueda, NUPAD IV, p.245, 1992

[6]:S.Ushio, K.Nishi, S.Kuroda, K.Kai and J.Ueda, IEEE Trans. Computer-Aided Des., CAD-9, 745 (1990)

[7]:K.Kai, S.Kuroda and K.Nishi, VPAD'93, p.66, 1993

[8]:S.Kuroda and K.Nishi, IEICE Trans. Electron. E75-C, 145, 1992 [9]:S.Kuroda, Y.Kawai, H.Onoda and and K.Nishi, IEDM Tech.Dig., p.713, 1991

[10]:F.Fukuda, S.Baba, T.Miyoshi and J.Ueda, NASECODE VI, p.422, 1989

[ll]:C-Y Lu and J.M.Sung, IEEE Electron dev. Lett., 10, 446, 1989

[12]:C.Mazure

and M.Orlowski, IEEE Electron dev. Lett., 10, 556,

[13]:Y.Okita,

M.Shinozawa,

A.Kawakatsu, A.Umemura,

K.Yamaguchi

and K.Akahane, IEEE Proc. CICC, p.22.4, 1988

 

[14]:J.lda,

S.Ishii, Y.Kajita,

T.Yokoyama, and

M.lno, IE ICE

Trans. Electron. E76-C, 525,

1993

 

 

TECHNOLOGY CAD SYSTEMS

275

Edited by F. Fasching, S. Halama, S. Se1berherr - September 1993

The MASTER Framework

P.J. Hopper and P.A. Blakey

Silvaco International,

4701 Patrick Henry Drive, Bldg. 3, Santa Clara, CA 95054, USA

Abstract

The MASTER Framework has a two level architecture. The first level consists of a standard structure format and a set of 'MASTER Tools'. This provides a high quality utility based framework that supports highly interactive use. The second level is a 'Virtual Wafer Fab' that adds capabilities for large-scale simulation-based design and experimentation. Powerful semiconductor technology CAD systems are constructed by populating The MASTER Framework with conforming process and device simulators.

1. Introduction

Many simulation tasks, especially those performed as part of semiconductor technology development, require multiple simulators to be used in combination. For example, process simulators are used with device simulators to predict the influence of process parameters on electrical behavior. Semiconductor technology CAD (5-TCAD) frameworks and systems make it possible, and even convenient, to use multiple tools to perform simulation tasks.

The papers in these proceedings demonstrate that the philosophies, architectures, features, and uses of different 5-TCAD systems differ considerably. The MASTER Framework which is described in this paper is oriented towards the needs of industrial users. It can be configured to meet the needs of a wide range of users, it supports highly interactive use, and it supports exceptionally high levels of task integration.

The organization of the paper is as follows. The first section provides some background material. The second section reviews strategic issues that are of concern to both implementors and users of 5-TCAD frameworks. The general architecture of The MASTER Framework is reviewed, and the component parts are then described in detail. A brief description of the presently available range of MASTER-conforming simulators is provided in an Appendix.

276

P.J. Hopper et al.: The MASTER Framework

1.2. The Evolution of S-TCAD Use

In the early days of S-TCAD it was very difficult to use multiple simulators to perform a task. The first problem was portability: implementors wrote code for the operating system and language dialects of their local mainframe computer, and other users were expected to port the code to their own system. Significant impediments remained after tools were ported. Each simulator used a different input syntax and a different data representation; and each simulator required users to perform generic operations (such as visualization) in different ways. Users were responsible for data transfer between tools, and learning to use multiple simulators was unnecessarily complicated. Tool quality suffered in various ways as implementors used scarce resources to replicate generic functionality.

The situation improved greatly during the 1980's. Workstations became widely available and very affordable, thereby providing engineers with powerful, convenient, interactive computing environments. Standards for operating systems (Unix), programming languages (C and C++), and graphical display across networks (X-Wmdows) emerged. Competition among commercial developers of S-TCAD software led to improved tools, greater affordability, and multi-platform support.

During the mid to late 1980's it became clear that the remaining impediments to the more widespread use of simulation would be eliminated if data representations were standardized, and if special software were written to perform generic operations required by many simulators. This led to the present interest in S-TCAD frameworks and systems.

1.3. Frameworks and Systems

A framework is a software environment that supports the use of multiple simulators, while working independently of any particular simulator. Frameworks normally provide convenient data transfer between simulators, a uniform user interface, and well defined procedures for adding new tools. A system is a collection of simulators and interfaces that allows users to perform tasks that involve the use of multiple simulators. It is not necessarily flexible or extensible. A framework-based system is obtained when a framework is populated with simulators.

Virtually all frameworks provide the benefits that result from the standardization of data and generic operations. Beyond this base level of functionality frameworks differ markedly with respect to the additional capabilities they provide. A basic choice is between adding features that are useful to developers, or features that are useful to users. 'Developers frameworks' are designed to make it easy for developers to add new features. They supply capabilities that allow developers' subtasks to be performed in a

P.J. Hopper et al.: The MASTER Framework

277

uniform, consistent manner. 'Users frameworks' focus on making it convenient for users to perform engineering tasks. The MASTER Framework is a users framework.

2. Strategic Design Issues

The strategic issues in the design (or assessment) of a semiconductor technology CAD system center on defining: the intended users; the system architects/designers; the system implementors; maintenance and support responsibilities; and how the system is financed. These questions can be investigated before any specific technical details are considered. The answers will often give a lot of insight into the nature of the system under consideration, and its probable strengths and weaknesses.

2.1. Organizational Strengths and Weaknesses

Several types of organization are involved in the design, implementation, financing and use of S-TCAD systems and frameworks. These organizations include: companies that manufacture semiconductor products; companies that develop S-TCAD software; university based research groups; government funding agenices; and centralized research consortia. In many cases organizations drawn from one or more of the above categories co-operate in a decentralized manner.

Each category of organization has characteristic strengths and weaknesses with respect to financial resources, personnel resources (number, experience, skill mix, and motivation level), continuity, and accountability. Understanding these strengths and weaknesses is helpful when assessing the nature and long term viability of a particular development effort.

Companies that produce semiconductor products generally have significant financial resources. The issues they face are: is it appropriate to start a peripheral development activity; can an internal development group be staffed adequately; compared to what is available commercially, can an internal group develop unique capabilities or provide significant cost advantages; and are internal developers willing to provide adequate levels of support to users?

Companies that develop S-TCAD software products can provide continuity and customer support. They are faced with the issues of recruiting personnel with the appropriate mix of skills necessary to design and implement products; funding development prior to the occurrence of sales; and making enough sales to ensure long term commercial viability.

Universities have a culture of innovation, and access to graduate student labor that is typically cheap, flexible, and enthusiastic. The issues faced by universities are: can inexperienced graduate students implement· capabilities

278

P.J. Hopper et al.: The MASTER Framework

with adequate speed and quality; are routine development tasks an appropriate part of graduate student research; can continuity of effort be maintained after students graduate; who is responsible for support and maintenance; can funding be obtained for work that is not in the mainstream; and can funding be retained for work that competes directly with commercial organizations?

For government funding agencies the issues are: how can certain organizations be selected for subsidy without undermining unsubsidized organizations; how can self-selection by commercially unviable organizations be avoided; how can progress be monitored and evaluated effectively; and will the work continue after government funding ceases?

Centralized research consortia can spread development costs over their membership. The issues faced by centralized research consortia include: developing an initial concensus; acquiring and retaining funding; recruiting staff; setting detailed technical directions; monitoring progress; establishing technology transfer mechanisms; and providing maintenance and support.

Co-operating groups that work in a decentralized fashion may appear to have the resources and skills necessary to perform a task. However, they usually find it difficult to overcome the problems of coordination and information transfer. The associated inefficiencies are usually much greater than anticipated, and the diffusion of accountability often means that few worthwhile results are produced.

2.2. The Rationale For Commercial Frameworks

Only a few of the very largest microelectronics companies can now consider developing and supporting S-TCAD frameworks and systems internally. A decision to support a major in-house development activity can no longer be justified on the grounds of cost, and only very seldom can it be justified on the grounds of capabilities. However some organizations continue to place a significant value on ownership and control of what they perceive as a strategic resource.

The vast majority of the microelectronics industry prefers to purchase affordable, high quality, well supported S-TCAD software from a commercial developer. The MASTER Framework was developed to meet the needs of this market. Silvaco management realized several years ago that there were several factors operating in favor of starting to implement a commercial framework. Silvaco was not locked in to a traditional batch-oriented way of using process and device simulation. Silvaco was also in a position to hire experienced development engineers and programmers in the right mix, and in sufficient numbers, to complete the project in a timely fashion. The company has a support and applications organization that is highly regarded by