Está en la página 1de 172




 

! #"%$'&)(+*-,./#01#2!354638&7 /69:#2;,$=<-.),$=(+*&>4?,.)@,.A/B2C#DFE&)$=<#3-/
0G*HGIJ$'<KEL.)MN,./01#0G$=(+*  ,./2C#0PE5O QSR  /6!UTB3  3-/-V:UTW(+*% O QSRL/
T??T8,  4U,.S*-,.).A/6XY?TZQ[$\M Q[$'<-@,.A/-X#]^*  ._<-,.

<&aQ_$\Q  QS,bMNEL.PXc$=(d._EZ,  ,(dQS.SE<$=(&

],(+*<$=(3 He <$fg,._&)$Q ih $',<<#3

Contents

1 Introduction to MINIMOS

Hj jkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj
1.2 MINIMOS 2D Model Structure jljHjkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj
1.3 MINIMOS 3D Model Structure jljHjkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj
1.4 Transient Simulation jHjkjljHjHjHjljHjkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj
1.5 Monte Carlo Simulation of Carrier Transport jHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj
1.6 Small-Signal AC Analysis jHjHjHjljHjkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj
1.7 Nonplanar Geometries jkjljHjHjHjljHjkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj
1.8 Three-Dimensional Simulation jHjljHjkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj
1.9 MESFET Simulation jHjkjljHjHjHjljHjkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj
1.10 Gate Charge and Capacitance Calculation jHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj
1.11 Monitoring Program Execution jljHjkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj
1.1

The New Features of MINIMOS 6

2 The Physical Models of MINIMOS


2.1
2.2
2.3
2.4
2.5

jHjHjHjljHjkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj
Mobility Model for Gallium Arsenide jljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj
Generation-Recombination Modeling jljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj
The Local Carrier Heating Model jHjkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj
Modeling of Band-Structure Parameters jHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj
2.5.1 Models for Silicon jHjHjHjljHjkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj
Mobility Model for Silicon

1
1
2
4
5
8
9
10
11
12
13
14
17
17
20
21
23
24
24

jHjkjHjljHjHjljHjkjljHjHjHjljHjkjljHjHjHjljHjkjljHjHj
2.6 The Band-to-Band Tunneling Model jljHjkjHjljHjHjljHjkjljHjHjHjljHjkjljHjHjHjljHjkjljHjHj
2.7 The Dynamic Interface and Bulk Trap Model jHjHjljHjkjljHjHjHjljHjkjljHjHjHjljHjkjljHjHj
2.7.1 The Physical Model and Numerical Considerations jHjHjljHjkjljHjHjHjljHjkjljHjHj
2.7.2 Interface Traps jljHjkjHjljHjHjljHjkjHjljHjHjljHjkjljHjHjHjljHjkjljHjHjHjljHjkjljHjHj
2.7.3 Bulk Traps jHjHjljHjkjHjljHjHjljHjkjHjljHjHjljHjkjljHjHjHjljHjkjljHjHjHjljHjkjljHjHj
2.7.4 Examples jHjHjljHjkjHjljHjHjljHjkjHjljHjHjljHjkjljHjHjHjljHjkjljHjHjHjljHjkjljHjHj
2.8 Gate Depletion Analysis jHjkjHjljHjHjljHjkjHjljHjHjljHjkjljHjHjHjljHjkjljHjHjHjljHjkjljHjHj
2.8.1 The Full Two-Dimensional Model (MPOLY=1) jljHjHjHjljHjkjljHjHjHjljHjkjljHjHj
2.8.2 The Approximate One-Dimensional Model (MPOLY=0) jljHjkjljHjHjHjljHjkjljHjHj
2.8.3 Fixed Charge and Traped Charge at the Gate/Oxide Interface jljHjHjHjljHjkjljHjHj
2.8.4 Examples jHjHjljHjkjHjljHjHjljHjkjHjljHjHjljHjkjljHjHjHjljHjkjljHjHjHjljHjkjljHjHj
2.9 Power VDMOSFET Simulation jHjHjljHjkjHjljHjHjljHjkjljHjHjHjljHjkjljHjHjHjljHjkjljHjHj
2.10 The Monte Carlo Transport Module jljHjkjHjljHjHjljHjkjljHjHjHjljHjkjljHjHjHjljHjkjljHjHj
2.10.1 Band Structure jljHjkjHjljHjHjljHjkjHjljHjHjljHjkjljHjHjHjljHjkjljHjHjHjljHjkjljHjHj
2.10.2 Scattering Rates jHjkjHjljHjHjljHjkjHjljHjHjljHjkjljHjHjHjljHjkjljHjHjHjljHjkjljHjHj
2.10.3 Impact Ionization jkjHjljHjHjljHjkjHjljHjHjljHjkjljHjHjHjljHjkjljHjHjHjljHjkjljHjHj
2.10.4 Material Characteristics jHjHjljHjkjHjljHjHjljHjkjljHjHjHjljHjkjljHjHjHjljHjkjljHjHj
2.10.5 The Monte-CarloDrift-Diffusion Coupling Technique jljHjkjljHjHjHjljHjkjljHjHj
2.10.6 The Monte Carlo Windows jljHjkjHjljHjHjljHjkjljHjHjHjljHjkjljHjHjHjljHjkjljHjHj
2.10.7 The Self-Consistent Iteration Scheme jHjHjljHjkjljHjHjHjljHjkjljHjHjHjljHjkjljHjHj
2.10.8 Calculation of Averages jHjHjljHjkjHjljHjHjljHjkjljHjHjHjljHjkjljHjHjHjljHjkjljHjHj
2.10.9 The Particle Weighting Algorithm jHjljHjHjljHjkjljHjHjHjljHjkjljHjHjHjljHjkjljHjHj
2.10.10 The Self-Scattering Algorithm jHjkjHjljHjHjljHjkjljHjHjHjljHjkjljHjHjHjljHjkjljHjHj
2.5.2

Models for Gallium Arsenide

3 Reference of Input Directives

jkjHjljHjHjljHjkjHjljHjHjljHjkjHjljHjHjljHjkjljHjHjHjljHjkjljHjHjHjljHjkjljHjHj
DEVICE jljHjHjljHjkjHjljHjHjljHjkjHjljHjHjljHjkjljHjHjHjljHjkjljHjHjHjljHjkjljHjHj
BIAS

ii

25
26
31
32
33
35
37
40
41
42
44
44
47
50
50
51
54
55
55
56
58
58
59
60
61
62
64

jHjHjHjljHjkjljHjHjHjljHjkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj 68
FIMPLANT jHjkjljHjHjHjljHjkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj 71
GEOMETRY jkjljHjHjHjljHjkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj 73
GRID jHjHjljHjkjljHjHjHjljHjkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj 77
IMPLANT jljHjkjljHjHjHjljHjkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj 83
INTERFACE jHjkjljHjHjHjljHjkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj 85
MOBILITY jHjkjljHjHjHjljHjkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj 94
MONTE-CARLO jHjHjHjljHjkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj 96
OPTION jHjljHjkjljHjHjHjljHjkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj 103
OUTPUT jljHjkjljHjHjHjljHjkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj 109
PROFILE jljHjkjljHjHjHjljHjkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj 112
RECOMBINATION jHjHjljHjkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj 120
STEP jHjHjljHjkjljHjHjHjljHjkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj 125
TRANSIENT jkjljHjHjHjljHjkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj 128
END

4 Files
4.1

129

jljHjHjHjljHjkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj 129
The Input Files jkjljHjHjHjljHjkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj 129
One-dimensional Profiles (FILE=1-D) jHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj 130
Twodimensional Profiles (FILE=2-D) jHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj 131
The Main Ouput Files jHjljHjkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj 134
The Formatted Output File jHjkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj 134
The Reference File jHjHjHjljHjkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj 135
The Binary Output File jljHjkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj 135
The Formatted PIF Output File jHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj 138
Auxiliary Output Files jHjljHjkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj 139
Table of I-V Data jHjHjHjljHjkjljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj 139
The Monte Carlo Output File jljHjHjljHjHjkjljHjHjljHjHjkjljHjHjljHjHjljkjHjHjljHj 139

The MINIMOS 2D Files


4.1.1

4.1.2

4.1.3

iii

jHjljHjkjHjljHjHjljHjkjljHjHjHjljHjkjljHjHjHjljHjkjljHjHj 140
The Gate Charge File jljHjHjljHjkjHjljHjHjljHjkjljHjHjHjljHjkjljHjHjHjljHjkjljHjHj 141
Poly-Gate Voltage Drop jHjHjljHjkjHjljHjHjljHjkjljHjHjHjljHjkjljHjHjHjljHjkjljHjHj 141
Charge-Pumping Output jHjHjljHjkjHjljHjHjljHjkjljHjHjHjljHjkjljHjHjHjljHjkjljHjHj 142
Band-to-Band Tunneling Currents jHjljHjHjljHjkjljHjHjHjljHjkjljHjHjHjljHjkjljHjHj 143
The MINIMOS 3D Files jHjHjljHjkjHjljHjHjljHjkjljHjHjHjljHjkjljHjHjHjljHjkjljHjHj 143
The Binary Data Link File jHjljHjkjHjljHjHjljHjkjljHjHjHjljHjkjljHjHjHjljHjkjljHjHj 143
The Formatted Output and the Reference Files jljHjHjHjljHjkjljHjHjHjljHjkjljHjHj 143
The Binary Output File jHjHjljHjkjHjljHjHjljHjkjljHjHjHjljHjkjljHjHjHjljHjkjljHjHj 144
The Transient Output File

4.1.4

Bibliography

149

Index

155

iv

1
Introduction to MINIMOS

MINIMOS is a software tool for the numerical simulation of semiconductor field-effect transistors. The
first version was released over 10 years ago, and since then the code has undergone continuous extensions and improvements, culminating in 1991s Version 5.2, which is capable of modelling silicon
MOSFETs and SOI transistors as well as gallium arsenide MESFETs. The fundamental semiconductor
equations wich are solved in MINIMOS numerically comprise Poissons equation and two carrier continuity equations for electrons and holes. For the various coefficients such as carrier mobility, carrier
generation/recombination and impact ionization rates carefully chosen models are implemented. The
numerical schemes of MINIMOS are designed to handle planar and non-planar device structures in both
two and three space dimensions.

1.1 The New Features of MINIMOS 6


MINIMOS 6 is upward-compatible to all previous versions of MINIMOS and works with any input file
valid for a previous version. The major extensions of MINIMOS 6 are listed below:

m
m

Transient analysis.

Self-consistent inclusion of the trap-rate equations for dynamic of interface and bulk traps.

A Monte Carlo Module which couples self-consistently to the Poisson equation and replaces the
drift-diffusion approximation in critical device areas.

A band-to-band tunneling model.


Depletion analysis of poly-silicon gates.
Interface to the TCAD shell of Vista and a PIF-interface.

CHAPTER 1 INTRODUCTION TO MINIMOS

For instance, the transient facility in conjunction with the trap rate equations allow for an accurate
numerical analysis of the charge pumping-phenomenon.
The Monte Carlo Module available in this version makes MINIMOS 6 to an easy to use Monte Carlo
device simulator which describes hot electron transport and impact ionization in a fully non-local manner.
In the following some minor extensions are summarized:

Output data can be written in ASCII-PIF format (OUTPUT PIF=YES).

Power Vertical DMOS transistors can be handled (OPTION LVDM=YES).

P-i-n diodes can be handled (flag for p-i-n diode simulation is set if an NPP+ or PNN+ structure
is specified by appropriate profile statements).

If a STEP directive is specified the resulting I-V data are written automatically to a file
(unless OUTPUT IV=NO).

Gate charge can be calculated and written to a file (OPTION GCHC=YES).

Coupled solution of the AC small-signal device equations for more robustness at higher
frequencies (OPTION CP=YES)

A new current integration method for transient and steady-state conditions

1.2 MINIMOS 2D Model Structure


MINIMOS builds the consistent solution of the semiconductor equations in an hierarchical manner, starting
with a relatively simple model which is subsequently refined by taking into account more complicated
physical mechanisms. This degree of sophistication is controlled by the key MODEL on the OPTION
directive, see Page 103).
Presently the user may specify 6 distinct models. Apart from MODEL=THRES, which leads to a threshold
voltage calculation, the models 1-D, 2-D, AVAL, HOT and MC define a sophistication hierarchy.
As an example consider the case that a user has specified a MOSFET in a MINIMOS input file, and
that the MODEL key on the OPTION directive is set to MODEL=AVAL. The following is valid only for
MOSFETs, the differences in case of MESFETs are described later in this section.
MINIMOS at first tries to minimize the error norms iteratively by accounting for the simplest physical
model. Thus, after having made a guess for the initial solution, MINIMOS starts with the internal model
option 1-D. In this model the MINIMOS does not solve the majority continuity equations and generation
recombination effects are neglected. The Poisson equation is solved in the full simulation domain, whereas
the minority continuity equation is just solved in the partial domain from to gateinsulator to about the
half of the pnjunction depth. Within the deep bulk region the minorities are computed by extrapolation
assuming a constant quasi fermi level. This model offers fast execution since the computational demand
for the costly solution of the carrier continuity equations is minimized. Within model 1-D MINIMOS

1.2 MINIMOS 2D MODEL STRUCTURE

builds its grid through an adaptive grid refinement loop. Each refinement step consists of a grid update
followed by a Gummel iteration loop, which terminates when all error norms lie beyond the threshold set
by the key ERROR on the END directive. After completing the grid loop, that is when the mesh weighting
function (BWF) is smaller than unity for all mesh intervals, MINIMOS switches its internal mode to
2-D. Now the minority continuity equation is solved in the full simulation domain, for the majorities
there is still a constant quasi fermi level assumed. No grid updates are carried out. Since the deviation
from thermal equilibrium in the deep bulk is usually small, only few Gummel iteration steps should be
necessary.
Finally, MINIMOS executes in AVAL mode. The Poisson equation and both carrier continuity equations
are solved in the full domain. The recombination rate is computed at every Gummel step, the impactionization and tunneling generation rate, however, only after a sufficient decrease in the error norms. This
leads to an avalanche subcycle. MINIMOS completes its execution when the last carrier generation update
does not increase the error norms over the threshold set by ERROR.
For a MESFET MINIMOS proceeds slightly different. In the 1-D mode both carrier continuity
equations are solved, however generation/recombination is neglected. Then in 2-D mode the generation/recombination mechanisms, impact ionization excluded, are switched on. Finally, the AVAL mode
for the MESFET has the same meaning as the AVAL mode for the MOSFET.
In model HOT another cycle is superimposed on the previous model AVAL. To account for carrier heating
effects local carrier temperatures are introduced. After the completion of each AVAL cycle a carrier
temperature update is performed. MINIMOS in HOT mode terminates when the error norms in both
carrier temperatures have decreased beyond the error measure TCERR, specified within the END directive.
In model MC MINIMOS generates an initial solution for the subsequent Monte Carlo iteration by processing
the internal models 1-D through AVAL. The resulting distribution of the electrostatic potential serves as
input for the Monte Carlo simulation. As output the distributions of electron mobility, electron temperature
and ionization coefficient are obtained in critical device regions. With the updated mobility and temperature
distributions a gummel iteration is performed. After its termination an avalanche subcycle is imposed to
account for the altered ionization coefficients. This procedure yields the potential for the next Monte Carlo
iteration. All the above described steps which are needed to obtain the new potential from the old one
can be considered as one Monte Carlo Poisson iteration step. MINIMOS terminates after a user-defined
number of such iteration steps has been executed.
In the following we summarize the model hierarchy of MINIMOS. A model can be specified by the
MODEL key on the OPTION directive.

CHAPTER 1 INTRODUCTION TO MINIMOS

MODEL
1-D

THRES
2-D
AVAL
HOT
MC

kn oqp
nkoqp
nkoqp
nJoqp
nkoqp
nJoqp

Description
Bulk-MOSFET: Poisson, minorities, partial domain, no
SOI-MOSFET : Poisson, minorities, full domain, no
MESFET : Poisson, both carriers, full domain, no
Loop for threshold voltage calculation by model 1-D
MOSFET : Poisson, minorities, full domain, no
MESFET : Poisson, both carriers, full domain,
, no impact ionization
MOSFET, MESFET: Full equation set, full domain,
including impact ionization.
Local carrier tempearture model in addition to model AVAL.
Full set of equations with electron temperature, electron mobility and ionization
coefficient from a Monte Carlo simulation.

1.3 MINIMOS 3D Model Structure


MINIMOS 3D relies on a similar model structure as MINIMOS 2D. With the exception of MODEL=THRES,
MINIMOS 3D can execute in the same modes as MINIMOS 2D. In other words, the key MODEL on the
OPTION directive directs the computation of the three-dimensional solution in an equivalent way. However
an additional parameter, M3MODE is required to control MINIMOS 3D execution. The combination of
both parameters allows the setting of a variety of three-dimensional models. At first it is assumed that
M3MODE=1 was set on the OPTION directive. M3MODE=1 leads MINIMOS 3D to build the mesh in zdirection together with a coarse three-dimensional solution by solely solving the Poisson equation in three
dimensions. The carrier concentrations are calculated from their quasi fermi level, assuming vanishing
current densities in the z-direction. The solution of the previously solved two-dimensional problem,
corresponding to the setting of the MODEL key, is taken as a Dirichlet boundary condition in the first
x-y-plane. This first x-y-plane is assumed to be the center of symmetry with respect to the z-direction.
This model offers comparatively fast execution since no solution of the carrier continuity equations is
carried out.
After completion of the model M3MODE=1 MINIMOS 3D either terminates execution or continues with
M3MODE=2, if this was specified on the OPTION directive. With M3MODE=2 the grid adaptation is
deactivated. The carrier continuity equations are solved according to the setting of the MODEL key on the
OPTION directive. Thus, if MODEL=1-D or MODEL=2-D was specified within the MINIMOS input deck
the minority carrier continuity equation is solved and no avalanche and hot carrier effects are regarded. In
all other cases, namely MODEL=AVAL or MODEL=HOT, both carrier continuity are solved and avalanche
and hot carrier effects are taken into account analogously to the two-dimensional case. Generally, the first
x-yplane is treated as a Neumann boundary, thus indicating that the device is modelled symmetrically
with respect to the first x-yplane.
The second character in the M3MODE specification allows the use of a previous three-dimensional solution
of the same example as initial solution, which can be helpful and CPU time saving during testing,
development etc. For example consider the case that MINIMOS 3D has been run with the setting
of M3MODE=1. To obtain a solution with the setting M3MODE=2 change the setting of M3MODE to

1.4 TRANSIENT SIMULATION

M3MODE=2O, rename the binary outputfile of the previous MINIMOS 3D run appropriately and restart
MINIMOS 2D. See Chapter 4. MINIMOS 2D is terminated after having loaded the changed commonblock data. Now invoke MINIMOS 3D. If MINIMOS 3D finds a file with the old three-dimensional
solution of M3MODE=1 it continues immediately with M3MODE=2. See also the key M3MODE on the
option directive Page 103 to obtain further information.

Model Option Key Summary (MINIMOS 3D)

1N
1O
2N
2O

n:r;p

Poisson only, full domain, no


Same as 1N, starting with prev. initial solution
Equation set according to MODEL key
Same as 2N, starting with prev. initial solution

1.4 Transient Simulation


An efficient algorithm for solving the time-dependent semiconductor equations has been built into MINIMOS. The dynamics equations for arbitrary distributed interface and bulk traps are consistently solved
with the basic semiconductor equations (see Section 2.7).
The solution of the time-dependent semiconductor equations in MINIMOS employs a hybrid algorithm
using a fast decoupled solution of the semiconductor equations [1]. The algorithm proceeds in two
main steps: At first the successive procedure devised by Mock computes an initial solution [2]. This
initial solution phase is indicated by a minus sign preceeding the Poisson solver iteration count in the
simulation reference file. In the subsequent correction steps Gummels algorithm is applied to compute the
selfconsistent final solution. By this approach the stability can be achieved at very short time steps, while
the Poisson equation is fulfilled with desired (controlled) accuracy independent of the actual time step.
The algorithm has been tested on a wide range of MOSFETs under a manifold of operating conditions
including biasing and switching times, ranged from very short (1 ) to very long (1 ) and has proven to
be both, fast and stable.

sAt

After each time step a summary of terminal currents is printed in the reference file. These currents are
split up into the electron (ele), hole (hol) and displacement (dis) current, each of these components being
printed on a separate line. The last line of this current information record prints the sum of electron, hole
and displacement current, i.e. the total electrodynamic current for each individual contact. The last entry
in this line is the sum of all terminal currents. Due to the conservation law of charge this must be a small
number, which depends on the solution error of the semiconductor equations. After the last time step
average values of all terminal currents across the total simulation time are printed in the reference file.
For the calculation of the terminal currents in transient conditions we employed an accurate weighting
functions-based technique.

CHAPTER 1 INTRODUCTION TO MINIMOS

The evolution of the terminal voltages and currents are written to file unit NUMTRA.

Specification of the terminal voltages


Each transient simulation initially requires the computation of a steady-state solution. The terminal bias
for this steady-state is specified on the BIAS directive (see BIAS directive Page 62). After convergence
MINIMOS will start a transient analysis, if a TRANSIENT directive (see Page 128) is found in the
MINIMOS input deck. The TRANSIENT directive is used to specify all required input parameters to
build waveforms for voltage-controlled contacts.
By default the voltage waveform for an individual contact is independent from the waveforms on other
contacts. The default voltage waveform for each contact is a constant voltage equals to the steady state
bias. To specify a different waveform the waveform-switch Wi is used, where i stands for B (bulk contact),
D (drain contact), G (gate contact) or S (source contact). Wi must be an integer out of 1,2,3 , with 1 being
the default value. Wi=1 denotes a constant voltage on the contact i. The two further choices, Wi= 2,3 are
described in detail below.

u v

Wi=2 denotes a periodic trapezoidal voltage with the starting point at t=0 located at the beginning of
the falling edge of the trapeze (see Fig. 1.1). To specify parameters for this waveform the Tij and Uij
parameters are used, where i denotes the contact as outlined above and j enumerates the parameters of the
individual waveform. All T parameters are specified in seconds, all U parameters in volts.

{}|

wAxgy

{}~

{'

2 are the following: Ti1 is the period length of the trapeze pulse ( ),
The waveform parameters for
Ti2 is the fall time ( ), Ti3 the lower-plateau time ( ), Ti4 the rise time ( ) and Ti5 is the simulation end
time. Note that the upper-plateau time ( ) of the waveform needs not be specified. If Ti5 is larger then
Ti1, more than one period will be simulated. The parameter Ui1 is by default equal to the corresponding
voltage given in the bias directive and needs not be specified. Ui2 specifies the amplitude of the trapezoidal
waveform. The given value for the amplitude may be positive or negative. If a negative value is specified
for Ui2, the waveform is mirrored with respect to the voltage axis of the initial steady-state bias. In this
case the specifications falling edge and rising edge interchange their meaning and the transient analysis
starts at
0 with a rising edge. Clearly, Ti2 denotes the rising edge and Ti3 denotes the falling edge
when 2 0.

{}

{6y

The parameters Ui3 and Ui4 control the time-step lengths when a rising or falling voltage ramp is stepped
through during a transient simulation. Specifically, Ui3 denotes the voltage increment to be used for the
contact i for the falling edge (if 2 0, otherwise the rising edge), Ui4 denotes the voltage increment to
be used for the contact i for the rising edge (if 2 0, otherwise the falling edge).

wxy

{=|

With
3 three-level voltage pulses can be constructed (see Fig. 1.2). The following parameters
specify all required values: Ti1 is the waveform period ( ), Ti2 the fall time prime ( ), Ti3 the mid-level
duration ( ), Ti4 the fall time double prime ( ), Ti5 the rise time ( ), Ti6 the top-level duration ( ) and
Ti7 the end of the simulation. Specification of the low-level duration ( ) is not necessary. The waveform
is periodic with the period length Ti1. Ui1 equals the steady-state bias and needs not be specified, Ui2
is the waveform total amplitude and can be positive or negative. In the negative case the waveform is
mirrored with respect to the axis placed on the top level. Both falling edges change to rising edges, the

{'

{ |

{=

{}~

1.4 TRANSIENT SIMULATION

Figure 1.1: Trapezoidal waveform

{ |

rising edge changes to the falling edge and the simulation starts at the first rising edge ( ). Ui3 specifies
the mid-level voltage, which must range within [Ui1,Ui1-Ui2].
The parameters Ui4, Ui5 and Ui6 control the time step lengths for the three voltage ramps: Ui4 is the ramp
step voltage increment of fall time prime, Ui5 is the ramp step voltage increment of fall time double prime
and Ui6 is the ramp step voltage increment of the rising voltage ramp. For a negative amplitude Ui2 these
parameters change their meaning accordingly.

Figure 1.2: Three-level waveform


If no voltage waveform is specified on any contact, all terminal biases remain constant within the whole
simulation time. In this case the obligatory key TG1 determines the first and the obligatory key TG2 the
second time step. The following time steps are expanded using a multiplication factor of the ratio of the
first two time steps. The transient simulation is stopped, when the end time given by the obligatory key
TG3 is exceeded. This time-stepping mode is prospective to be employed in some simulation modes in
future.
Different waveforms can be applied on different terminals simultaneously. Each waveform must be
completely described, in other words, all parameters must be given for each particular waveform, independently of other waveforms and of the end time. End time of the simulation must be the same for all

CHAPTER 1 INTRODUCTION TO MINIMOS

waveforms. The waveform parameters for each specified contact will be checked for completeness and
physical consistency. End time can be any positive number, also less than the period. Therefore, one can
focus on a part of the specified waveform when analyzing only e.g. turn-on and turn-off of device.
The actual time stepping is controlled by waveforms on all contacts simultaneously, fulfilling the requirements specified by user for the particular waveforms.

1.5 Monte Carlo Simulation of Carrier Transport


Todays semiconductor devices are characterized by large electric fields in conjunction with steep gradients
of the electric field and the carrier concentrations. As these variations may occur over distances comparable
to the carriers mean free path the widely used drift-diffusion approximation is losing validity. Simulation
of charge transport based on a Monte Carlo particle model is physically much more accurate under such
conditions.
Currently, in MINIMOS a particle model is implemented which describes the motion of electrons in
silicon. The band-model of silicon accounts for the six equivalent minima near the Xpoints. Each
of these valleys is assumed non-parabolic and anisotropic. Models for phonon scattering, scattering at
ionized impurities and at rough surfaces, as well as impact ionization, are described later in this manual
in more detail.
MINIMOS utilizes a hybrid approach which combines the Monte Carlo and the drift-diffusion models
for the analysis of a single device. In critical device regions, position-dependent coefficients such as
mobility, carrier temperature and ionization coefficient are extracted from a Monte Carlo simulation.
These coefficients, after being extended analytically over the rest of the simulation domain, are then used
in a generalized drift-diffusion current equation to simulate the overall device. The above mentioned
coefficients are sometimes also referred to as Monte-Carlodrift-diffusion coupling coefficients. The
theoretical foundation of the hybrid approach is given by the Boltzmann transport equation [3] [4].
In MINIMOS self-consistency of the carrier distributions with the electrostatic potential is achieved by
iteratively updating the MonteCarlodriftdiffusion coupling coefficients. Each step of the iteration
scheme requires one Monte Carlo simulation and one self-consistent solution of the semiconductor
equations (see also Section 1.2).
Currently the Monte Carlo module in MINIMOS deals with electrons only. Therefore, holes are treated
by the drift-diffusion approximation even in model MC.
The analytic band model of MINIMOS allows the equations of motion to be integrated analytically. The
time of a carriers free flight is determined by an optimized self-scattering algorithm.
To deal properly with scarcely populated regions in both real and energy space a trajectory multiplication
scheme has been implemented. For this purpose the simulation domain can be divided in up to four
different regions, in each of which a unique weight is assigned to the carriers. The positions of the
boundaries of these regions and the corresponding weights are determined automatically by MINIMOS
In energy space a carrier trajectory is multiplied if the energy threshold for impact ionization is exceeded.

1.6 SMALL-SIGNAL AC ANALYSIS

During a Monte Carlo simulation statistical data, sometimes also referred to as raw-data, are recorded
on an auxiliary grid which is independent of the MINIMOS grid. After a Monte Carlo simulation
the raw-data are assigned to the grid points of the MINIMOS grid by means of a convolution method.
The parameters for the weighting function used for this step can either be specified in the input file or
determined automatically by MINIMOS.

1.6 Small-Signal AC Analysis


To perform AC analysis with MINIMOS an AC bias has to be specified within the BIAS directive. This
bias voltage is defined by its real part VR and its imaginary part VI. If one of these keys is missing a zero
value is used for it as default.
The bias is applied to the gate electrode by default. Any of the 4 terminals can be specified by assigning
the key JCAP on the BIAS directive one of the following five symbolic values: G (Gate), S (Source), D
(Drain), B (Bulk). Specifying A (all terminals) leads to the application of AC bias to all terminals one
after the other.
The absolute value of the bias is not significant due to the linearity of the problem, in other words, the result
of the simulation does not depend on the absolute value of VR and VI. However the iteration convergence
may be influenced by both the absolute value as well as the angle of the complex input voltage. The
frequency of the applied AC bias can be set by the key OMEGA within the OPTION directive. The iteration
of the AC-analysis can be controlled by the keys CAPERR and NAC on the END directive. The assigned
value of CAPERR is the desired final error of the AC-analysis and, if specified, the value of the key NAC
limits the number of AC-analysis iterations.
The AC-analysis is preceded by a normal run of MINIMOS 2D. AC-analysis is restricted to planar twodimensional devices only. MINIMOS runs according to the degree of sophistication specified by the
MODEL key on the OPTION directive. The specified model influences the result and the convergence
properties the AC-analysis. It is recommended to use MODEL=AVAL for the AC-analysis, see also Section
1.2. With MODEL=THRES no AC-analysis is performed.

The result of the AC-analysis is one row of the admittance matrix, if just one terminal bias was specified
by the JCAP key, or the whole 4 4 matrix if JCAP=A was declared. The admittance matrix data are
printed (together with the other output data) in the formatted output file. The left item is the real part in
Siemens ( ), the right item is the capacity in Farad ( ).

If convergence problems arise with the default settings one should invoke the coupled solution algorithm
with CP=YES on the the OPTION directive. To obtain accurate results it is recommended to explicitly
specify the length of the source and drain terminals by the LSOURCE and LDRAIN keys on the OPTION
directive.

CHAPTER 1 INTRODUCTION TO MINIMOS

10

AC Analysis Key Summary

Key
JCAP
VI
VR
CAPERR
NAC
CP
LDRAIN
LSOURC
OMEGA

Description
ACbias terminal(s)
Imaginary part of applied voltage
Real part of applied voltage
Final Error
Iteration limit
switch to coupled solution method
Drain contact length
Source contact length
ACbias angular frequency

Directive
BIAS
BIAS
BIAS
END
END
OPTION
OPTION
OPTION
OPTION

1.7 Nonplanar Geometries


MINIMOS is capable of simulating nonplanar structures. This is true for both the twodimensional and
the threedimensional version. The nonplanarity is given by the oxide body. To model the nonplanar oxide
body a set of parameters is provided within the GEOMETRY directive (see Page 73). These parameters may
be divided into length specifications of straight lines and lengths of transitions. All length specifications
. The horizontal lengths of the transitions are measured from 10% to 90% of the
are to be given in
absolute values of the vertical distances of the slopes. The other horizontal length specifications are
measured from the 50% values of the transitions. Two drawings of nonplanar crossections are inserted
to support the spatial imagination of the geometry. Fig. 1.3 is a crossectional view through the central
xy plane, the center of symmetry of the threedimensional transistor at
2 , where W denotes the
channel width. Fig. 1.4 is a crossection through the yz plane at half gate length
2.

1yr

Gy ~

MINIMOS 2D automatically switches to nonplanar mode if the GEOMETRY directive is present within
the input deck. MINIMOS 3D calculates always nonplanarly in the zdirection. It runs nonplanar in all
spatial directions only if the GEOMETRY directive is given on the input deck.
The nonplanarity is handled internally by a point characteristics and a box integration concept. Due to the
necessity of resolving the nonplanarities sufficiently accurate there will generally be increased mesh sizes
and thus higher computer resources will be required. This is particularly true for full twodimensional
and threedimensional nonplanar simulation.

1.8 THREE-DIMENSIONAL SIMULATION


SCGRD

11

G1CGRD

G2GRD

SGAP

DCGRD

OXIDEF/2
+TINS/2
SDREOX

DGAP

DDREOX

RECESS
TINS

Y
S1GAP

D1GAP
D2GAP

S2GAP
SRGRD

R1GRD

R2GRD

DRGRD

Figure 1.3: Geometry parameters in the xy plane.

1.8 Three-Dimensional Simulation


Three-dimensional simulation of MOSFETs is necessary to accurately account for the influence of the
transition from the gate oxide to the field oxide, the so-called birds beak. It has been shown in the
literature that those effects are not negligible for channel widths of roughly less than 5 .
A three-dimensional cartesian coordinate system is introduced. The x-direction points into the main
current flow direction, the y-direction from the surface into the bulk, same as in MINIMOS 2D. The
z-direction points into the channel accordingly. The coordinate system origin is assumed to be at the
intersection between the gate-oxide to field-oxide transition and the source-sided gate edge. Thus, the
symmetry plane (x-y plane) has the z-coordinate 2 .
The transition from gate-oxide to field oxide introduces a nonplanarity in z-direction. The form of this
nonplanarity can be modelled by the keys BEAKL and OXIDEF within the GEOMETRY directive. BEAKL
specifies the length of the birds beak, OXIDEF the thickness of the field oxide. Default values for those
parameters are supplied as outlined in Chapter 3. since the specification of the GEOMETRY directive is
optional. Thus, with no GEOMETRY directive present, MINIMOS 3D calculates fully planar in x- and
y-direction and nonplanar in zdirection, otherwise nonplanar in all three dimensions. See Section 1.7 for
a detailed explanation of nonplanar simulation. The layered model structure MINIMOS uses to obtain a
three-dimensional solution of the semiconductor equations is explained in Section 1.2.
Three-dimensional simulation is comparatively memory and CPU time expensive due to the large ranks

CHAPTER 1 INTRODUCTION TO MINIMOS

12
W/2

W/4
OXIDEF

Z
RECESS
TINS

OXIDEF/2
-TINS/2

Figure 1.4: Geometry parameters in the yz plane.


of the arising linear systems of equations. A high performance can be achieved by running MINIMOS 3D
on vector- or vector-concurrent computers.
MINIMOS 3D expects a binary linkfile with all required data as input. This linkfile is produced by a
previous run of MINIMOS 2D with the M3MODE key specified in the input deck. The various settings
of this key are explained in Section 1.2. For information of the MINIMOS 3D input and output files see
Chapter 4.

1.9 MESFET Simulation


MINIMOS is capable of simulating MESFETs on either silicon or GaAs substrates. In the program a
Schottky contact model is included which accounts for a current dependent surface recombination velocity
and a field dependent barrier height [5], [6].
The oxide thickness given by the key TINS on the DEVICE directive acts as the switch from MOSFET
to MESFET simulation. Setting TINS to zero forces the program to use a Schottky contact at the gate.
In this case the GATE key on the DEVICE directive is interpreted as the barrier height of the Schottky
contact, which has to be supplied by the user.
The specific differences for GaAs versus silicon as substrate materials and the facilities to specify various
MESFET geometries are explained in detail in the sections DEVICE, GEOMETRY, IMPLANT, MOBILITY,

1.10 GATE CHARGE AND CAPACITANCE CALCULATION

13

PROFILE and RECOMBINATION. For the foundation of the mobility model for GaAs in MINIMOS see
Section 2.2. Both, steady-state and transient simulation, self-consistently including transient rate equations
for (deep) bulk traps, can be normally performed with MESFETs, Section 2.7.

1.10 Gate Charge and Capacitance Calculation

MINIMOS offers a possibility to obtain the total gate charge


by calculating the flux of the electric
field through a contour in the oxide around the gate. The contour can be seen in Fig. 2.4 and Fig. 2.5
(dot-dashed curve). In particular, the charge assigned to the source-sided and drain-sided edges, to the
bottom of the gate, as well as the total charge are calculated and written to both the reference file and the
file unit NUMGC1 (see file cmsysx.inc).

The quasi-static gate-capacitance is defined as

^y H
1 j 1

where H is the total charge in the gate. Performing the simulation at two gate biases r o 2 and

gate capacitance may be calculated by
o 2 the quasi-static
ZZ k o 2Lr! H r Zo 2
1 j 2

The differentiation should be done externally after a MINIMOS run. Proper results can be obtained
assuming i 25 . The value of can be specified on the STEP directive by the VG key.

Note that the grid must be frozen


(OPTION GRIDFREEZE=YES). Comparing the numerical capacitance
with the analytical result, assuming long-channel devices with uniformly doped bulk, we found that this
simple technique to calculate is very accurate. The numerical error is small, whereas an influence of

the discretization error is quite reduced by using the same grid for both bias points. In Section 2.8 the
accuracy of this technique is demonstrated.

Keys for Gate Charge Calculation

Key
GCHC
GRidf
VG

Description
Invoke calculation of total gate charge
freeze the discretization mesh
for each gate bias UG two DC-calculations are performed
at UG-VG/2 and UG+VG/2

The calculation of the gate charge is possible only for planar device structures.

Directive
OPTION
OPTION
STEP

CHAPTER 1 INTRODUCTION TO MINIMOS

14

1.11 Monitoring Program Execution


To survey the MINIMOS execution the reference file (see also Chapter 4) collects important runtime data.
The reference file offers the user information concerning:

m
m

Input deck, MINIMOS 2D only.


The entire input deck is included.

Important numerical constants, MINIMOS 2D only.

Gummel iterations.

Grid updates.

Virtual memory statistics.


CPU time consumption.

The displayed data are grouped according to the active model 1-D, 2-D, THRES, AVAL or HOT for
MINIMOS 2D and 1, 3 for MINIMOS 3D. Grid information is given selectively when grid adaption is
carried out i.e. when MODEL=1-D and/or M3MODE=1.
A grid adaption record contains the following information:

m
m

The mesh coordinate, x, y or z.

The grid coordinate at which a meshline should be inserted.

The grid index at which a meshline is inserted.

The mesh spacing at this position.


The value of the grid weighting function at the inserted meshline.

A Gummel iteration record contains the following information:

The consumed CPU time since MINIMOS start in .

The error norm for the majorities (LMAJ).

The error norm for the space charge (LF).

The error norms for the carrier temperatures if the HOT model is active.

The iteration count per Gummel loop.

The error norm for the minorities (LMIN).

The error norm for the potential distribution (LPSI).

The inner iteration count for the Poisson equation (PO).

1.11 MONITORING PROGRAM EXECUTION

15

The inner iteration count for the majorities (MA), MINIMOS 3D only.

The drain current in .

The inner iteration count for the minorities (MI), MINIMOS 3D only.

Either the bulk current for MOSFETs in


for MESFETs in for all models.

if the model AVAL or HOT is active, or the gate current

Note: If the Gummel iteration count is preceded by a minus sign, MINIMOS tries to accelerate convergence.
When MODEL=HOT is active a carrier temperature subiteration is performed. At each carrier temperature
update an informational record is issued, which contains:

m
m

The hot subiteration iteration count, 1 item.

a item.
m The carrier temperature error norm for the minorities, 3 item.
The carrier temperature error norm for the majorities, 2

MINIMOS controls convergence by the values of certain error norms. A Gummel iteration or a HOT
subiteration loop is terminated if all required error norms are smaller than the threshold set by the key
ERROR on the END directive. All error norms are dimensionless real numbers.

16

CHAPTER 1 INTRODUCTION TO MINIMOS

2
The Physical Models of MINIMOS

2.1 Mobility Model for Silicon


In this subsection the expressions representing the MINIMOS mobility model for silicon are summarized.
A more detailed discussion can be found in [7] [8].
The temperature dependence of lattice mobility in pure silicon is modeled by simple power laws.

~ y

300K 
z
~ y 460 cmVs 300K

1430

cm2
Vs

2 j 1

2 j 2

2 18

To account for mobility reduction due to ionized impurity scattering the formulae of Caughey and Thomas
[9] is used in conjunction with temperature dependent coefficients.

~ 8
S~_ Cy W_ _ r S
1
!  8
z z 200 K
8 y 80 cmVs 300K

z z
8 y 80 cmVs 200K
200K
300K 
z z 200 K
8 y 45 cmVs 300K

0 45

0 15

0 45

2 j 3

0 45

17

200 K

2 j 4
2 j 5
2 j 6

CHAPTER 2 THE PHYSICAL MODELS OF MINIMOS

18

W y

45

5 # 200K  z

 | y 1 j 12 10 cm z

300K

z
 | y 2 j 23 10 cm 300K

z
_ y 0 j 72 300K

cm2
Vs

200K
300K

0 45

0 15

200 K

32

17

17

32

0 065

2 j 7
2 j 8
2 j 9
2 j 10

Surface scattering is modeled by the following empirical expression:

 | ~[ } |
_~[N y _ 1 b _ r; S o>  1 | rP\
_ S
z
 | y 638 cmVs 300K

z
 | y 240 cmVs 300K

 |
exp r
Zy 1 2 exp
r 2 o o } |

2

1 19

1 09

2 j 11
2 j 12
2 j 13
2 j 14

The pressing forces


and
in Equation 2.11 equal to the magnitude of the normal field strength at the
interface, if the carriers are attracted by the interface, otherwise they are zero.
Deviations from the ohmic low-field mobility are given for both electrons and holes by

~_N
2
[ ~ N S y
S


_ 1 

1 N } Z
y grad Pr 1 grad (5 )

y grad ^ s1 grad 5 sAS

2 j 15
2 j 16
2 j 17

2.1 MOBILITY MODEL FOR SILICON

19

Here
represent the driving forces for electrons and holes and
voltages.

are the carrier temperature

The saturation velocities are assumed temperature dependent.

\N y 1 j 45 10 cms S
\N y 9 j 05 10 cms
7

tanh

2 j 18

tanh

2 j 19

155K

312K

Several parameters in the MINIMOS mobility model can be weighted. The weight coefficients which can
be specified on the MOBILITY directive have the following meaning:

Key
ML
MI
MC
MR
MS
MT
MV
MB

Symbol
Lattice and Impurity Scattering
and
in Equation 2.1 and Equation 2.2, respectively.
weight for
weight for
and
in Equation 2.3.
Surface Scattering
weight for
and in Equation 2.11.
weight for
and
in Equation 2.11.
and
in Equation 2.14.
weight for
weight for
and
in Equation 2.11.
High Field Mobility
weight for
and
in Equation 2.15.
weight for the exponents
and in Equation 2.15.

~  |

 |
 |
 |

'

~  |
_
} |
 |
 |
'

 |
_

All coefficients multiply the coresponding parameters, except MS, which divides the default
. Those
weights should only be used for fine tuning, since the default mobility parameters in MINIMOS are chosen
carefully and agree rather well to experiments. The default parameters in MINIMOS are:

CHAPTER 2 THE PHYSICAL MODELS OF MINIMOS

20

W ~
 |
} ||

} |
'

o>t
o>t
j
j
o>t
j Boq
j

j oat

=
=
=
=
=
=
=
=
=
=

1430
80
1 12 1017
0 72
638
7 0 105
1 69
10 10 7
1 45 107
2

2
2

8 ~
} |
  ||

 |
'

=
=
=
=
=
=
=
=
=
=

All constants refer to a simulation temperature of 300 , except


2.18 and Equation 2.19

o>kt
o>kt
j
j
o>kt
j Boq
j

j oat
2

460
45
2 23 1017
0 72
240
2 7 105
1 00
10 10 7
9 05 106
1

_\N that obey the law given by Equation

2.2 Mobility Model for Gallium Arsenide


In this subsection the expressions representing the MINIMOS mobility model for gallium arsenide are
summarized. A more detailed discussion can be found in [10] [11].
The temperature dependence of lattice mobility in pure gallium arsenide is modeled by simple power laws.

z
_~ Cy _ 5 300K
#

2 j 20

Impurity scattering is modeleled by the Caughey and Thomas [9] expression Equation 2.3. For gallium
arsenide the asymptotic mobilities at high impurity concentrations,
and
, can be weighted. The
coresponding key is MB key on the MOBILITY directive. The default value of MB is zero in case of gallium
arsenide, so that the influence of
is suppressed. Note that the MB key acts differently for gallium
arsenide and silicon (see key description Page 94).

W_

The high field mobility for electrons is given by

~ _ y ~[ 1 F A ~[ N

1F
A ~[ N

' y 6 10 cms 8 1 ~_ r ~_
10
5 j 477 10

~[
~_ #
: y 10 0 j 6 exp 10 r 2 exp 7 r 285 j 7 2

2 j 21
2 j 22

4 cm2
Vs

6 cm2
Vs

3 cm2
Vs

cm2
Vs

2 j 23

2.3 GENERATION-RECOMBINATION MODELING

{Zy 4

1280 sinh

21

~_
250

2 j 24

cm2
Vs

For holes the high field mobility reads

~_
~_ y 1 q N

' y 1 j 5 10 cms

2 j 25
2 j 26

and
represent the driving forces for electrons and holes according to Equation 2.16 and Equation
2.17, respectively.
For gallium arsenide several parameters can be weighted in the mobility model. The weights which can
be specified on the MOBILITY directive have the following meaning:

Key
MB
MI
ML
MV

Symbol
weight for
weight for
weight for
weight for

8  | and W} | in Equation 2.3 (default MB=0).


in Equation 2.3.
~ andand ~ in_ Equation
2.2, respectively.
' and ' in Equation2.12.21andandEquation
Equation 2.25, respecitively.

All weights multiply the corresponding parameters. The default settings of the various parameters in the
gallium arsenide mobility model are:

8 ~
 |

=
=
=
=
=
=

>o Jt
>o Jt
>o Jt

8000
1
1500
8000
1017
05

2
2

8 ~

 |

=
=
=
=
=
=

380
2
50
380
3 232 1017
0 4956

j
j

>o Jt
>o Jt
>o Jt

2
2

2.3 Generation-Recombination Modeling


In this section we present models for the standard generation-recombination mechanisms which are
accounted for by MINIMOS in the AVAL mode. Dynamic generation-recombination via interface and

CHAPTER 2 THE PHYSICAL MODELS OF MINIMOS

22

bulk traps is described below in Section 2.7. A description of generation-recombination due to interband
tunneling as it is modeled in MINIMOS can be found in Section 2.6.
The parameters in the following expressions can be set on the RECOMBINATION directive.

Shockley-Read-Hall recombination rate:

p S y _ ( )s r [ (s )
2 j 27

where , s and are the local concentrations of electrons, holes and the intrinsic concentration, respec2

tively. The lifetimes TN and TP can be specified by the user.

Auger recombination rate:

plJy s r s
2 j 28

S and p  are independent of the bulk-trap


with CN and CP being the Auger coefficients. Both, p
2

specification and are superposed on the generation-recombination calculated for bulk traps.

Surface recombination at the gate-oxide/bulk interface:


p  | y s r s%
2 j 29

with SN and SP being the surface recombination velocity for electrons and holes, respectively at the
2

interface between substrate and gate insulator. This surface recombination is independent of terminal bias
and interface trap density. It is superposed to the surface recombination rate calculated for interface traps
as explained in the previous section. If
0 or
0 is specified this surface generation-recombination
vanishes. Note, default values are not 0.

Surface recombination at the silicon-film/bulk-insulator interface in SOI devices:


pH  | y s r s 
2 j 30



with VN and VP being the surface recombination velocity for electrons and holes, respectively. This
2

surface recombination is independent of terminal bias and interface trap density. It is superposed to
the surface recombination rate calculated for interface traps placed at the back interface. This surface
generation-recombination can be suppressed by
0.

2.4 THE LOCAL CARRIER HEATING MODEL

23

Impact Ionization
MINIMOS has been enhanced to include a depth dependent ionization model that has been verified against
n-channel substrate current measurements. The ionization coefficients are given by

Zy: exp r Z
2 j 31

Zy: exp r
2 j 32

with denoting the local electric field and the depth from the Si-SiO interface. The existence of a

was suggested by Slotboom [12] and is also used in a number of lucky


depth dependent factor
_

electron models of impact


ionization [13].

The depth dependence of


_ is given by the expressions
gy 1 exp r + c
2 j 33
Zy 5 1 exp r [ j
2 j 34

The form of the depth dependence is similar to that used in the above references. The effect of the
depth dependence is to increase the
factor near the surface, which is similar to decreasing the
S settings
mean free path near the surface. Default
for the electron coefficients have been derived from
experimental measurements of the substrate current over a number of n-channel devices with channel
length in micrometer and 3 o 4 micrometer range. Hole parameters have not been calibrated against local











experimental data and rather use the so-called dark space coefficients from MINIMOS 4. Since no
experimental data were available for checking a depth dependent ionization model for holes, the default
is to apply the electron depth dependence to holes analogously.
All the above described generation-recombination mechanisms are active in steady-state as well as in
transient conditions.

2.4 The Local Carrier Heating Model


In MINIMOS enhanced drift-diffusion equations are implemented which allows for elevated carrier
temperatures. The carrier temperatures are obtained from a local approximation of the energy balance
equations [14].

z S y 2 \N ~[1N) r ~[1N

S
5 3 S  _ _ _
2 j 35
The energy relaxation times
and are constant with respect to the carrier temperatures. They are

modeled by


CHAPTER 2 THE PHYSICAL MODELS OF MINIMOS

24

3
y 2 a5
y 32 )

~_N
'
~[N
'

2 j 36

2 j 37

The energy relaxation time weights UN and UP have a default value of 0 8 and can be specified on the
OPTION directive.

2.5 Modeling of Band-Structure Parameters


In this section the models for the effective masses of electrons and holes, band gap and intrinsic concentration in silicon and gallium arsenide are presented.

2.5.1 Models for Silicon


The temperature dependence of the effective masses is modeled by

Cy 1 j 045 4 j 5 10 Cz Z

yC 0 j 523 1 j 4 10 z r 1 j 48 10 z

where z is the absolute temperature in .


denotes free-electron mass in vacuum.


2 j 38
2 j 39





The intrinsic concentration, if not explicitly specified by the INTRINSIC key on the OPTION directive,
is modeled by
  
E

exp
2 40
2

r z
where E is the energy bandgap for silicon E r E according to
z r 6 j 05 10 z
E y 1 j 1700 1 j 059 10
C
C z
z r 3 j 05 10 z
EUy 1 j 1785 r 9 j 025 10

z
with E being obtained in q .


170

170



2 j 41
2 j 42

2.5 MODELING OF BAND-STRUCTURE PARAMETERS

and the effective masses is given by


z j
300K

2 j 43

 

The relationship between the density of states product 

y 2 j 51 10

 

19

25


3 2

2.5.2 Models for Gallium Arsenide


The band gaps between the valence band and the lowest three conduction bands are given by

y 1 j 519 r 5 j 405 10 z6oq z6^oq^ 204



z6oq^
E y 1 j 815 r 6 j 050 10

z6oq^ 204
#z oq
E y 1 j 981 r 4 j 600 10
z6oq 204 q
The electron effective mass is modeled by

yC exp r z exp r z
with
y 0 j 067
y 0 j 55
y 0 j 85
y E r E
y E r E






3 2

3 2



2 j 44
2 j 45
2 j 46



2 j 47
22 jj 4489
2 j 50
2 j 51
2 j 52

2 3

3 2







The hole effective mass is modeled by

!
with
y 0 j 076
y 0 j 50
The density of states of the conduction and the valence band is calculated by
y 8 j 63 10 z
z
z

1 r 1 j 93 10 C r 4 j 19 10 C 21 exp r zG 44 exp r z
y

"!$#

"





13

2 j 53
22 jj 5545

3 2 2 3

3 2



3 2



CHAPTER 2 THE PHYSICAL MODELS OF MINIMOS

26

y 2 2 z
%

2 j 56

3 2

& 2

Using these formulations the intrinsic carrier concentration will be determined by

 '(

exp r 2E%z j


2 j 57

2.6 The Band-to-Band Tunneling Model


Interband tunneling has recently become increasingly important for Si technology. It is a source of leakage
in memory cells ( [15, 16]) and MOSFETs ( [17, 18]). This effect can produce a degradation in static
( [19, 20]) and an enhanced degradation in transient bias conditions ( [21]). It is prospective to be used
to design programable memories with a high injection efficiency ( [22]), as a monitor for hot-carrier
degradation ( [23, 24, 25]) and to develop devices whose operation is based on this effect ( [26, 27]).
Tunneling is a sensitive function of the electric field, whereas the field depends on several parameters such
as oxide thickness, dopant concentration, trap density, geometry and terminal biases. As a consequence,
tunneling in MOSFETs can only be properly modeled by a numerical approach [28, 29, 30, 31].
We have developed a two-dimensional numerical model of the band-to-band tunneling. Two approaches
have been implemented in MINIMOS:
One-dimensional model, in which the tunneling path is searched from a particular point in the bulk in
direction perpendicular to the oxide/bulk interface, as often applied in the literature, e.g. [17, 32].
Two-dimensional model, where we search from a particular point in the bulk discretization mesh, which
represents the starting point, in two dimensions for the nearest endpoint whose potential difference is
larger than the characteristic tunneling band gap (says E ). The tunneling path found in this way is normal
to the corresponding equipotential lines assuming sufficiently fine local grid. Since the tunneling length
)
is shorter than the curvature of the equipotential lines, the tunneling between the starting point and the
endpoint may be treated like a planar one-dimensional problem.

The first integer number in the key MTUN in the RECOMB directive specifies which of these two approaches
is used, as explained bellow.

y o

To calculate the generation rate associated with an individual path, models for direct and indirect tunneling
)
E * and the field
as function of the starting field , an average field along the tunneling path
variable from
to
(endfield) are implemented. Which field-model is applied in the calculation can
be specified by the third integer number in the key MTUN.

The calculated generation rates of electron and holes are separated in the position space. They are coupled
with the continuity equations via the generation term in a selfconsistent manner after filtering. The total
charge in device is strictly conserved. Filtering is used for smoothing the distributions only, to allow an
efficient and automatic grid adaption in the critical areas. Standard deviations for filtering associated with
and directions can be scaled by the keys SXT and SYT, respectively in the GRID directive. The default

2.6 THE BAND-TO-BAND TUNNELING MODEL

27

values of the standard deviations are chosen carefully, being appropriate for the grid adaption when the
tunneling is localized in the gate/drain overlap region. Therefore, it is not recommended to change these
keys when analyzing the standard gate-induced-drain-leakage (GIDL) problem ( [17, 18, 23, 25]). It
is, however, necessary to increase SXT up to 5 when tunneling takes place along the whole MOSFET
channel, as in the BBISHE injection [22].

Since the tunneling rate is strongly dependent on the electric field, the tunneling analysis is done on the
grid which is MXT times finer for the -mesh and MYT times finer for the -mesh than the MINIMOS
mesh for solving the basic semiconductor equations in the bulk. Increasing MXT and MYT quite reduce
the discretization error in the terminal currents, but can prolong the computation time. The default values
of 2 should represent a proper choice.

Terminal currents are calculated by a very accurate technique based on the local concentration-dependent
weighting functions (a similar method is proposed in [33]).
After an individual bias point is calculated, some quantities characteristic for the band-to-band tunneling
analysis are printed in the reference file. These cover the number of active starting points in the fine
discretization mesh, the actual fine mesh for the analysis in the bulk, the minimal and an average tunneling
distance, the maximum electric field involved in the calculation of the tunneling rates (either starting
or an average field, dependent on the specification) and an average electric field across all paths found.
If the variable-field model is specified, the field variation along the shortest tunneling distance and an
average field variation across all paths are printed, as well. In addition, the total tunneling generation
rates of electrons and holes in the whole device are printed; they should be equal. Particularly, the carriers
0) and the drain-half of device are presented. Note that
generated in the source-half ( ,+ 2,
the tunneling can occur in both junctions at some bias conditions (sufficiently large bulk reverse bias).
After the tunneling rates the total avalanche generation rate in device, as well as the sum of the both, total
avalanche and tunneling rates are printed ( [29]). These rates should be compared with the electron and
hole components of the terminal currents, printed in the current-information record. Note that additional
generation-recombination processes also contribute to the terminal currents.

-o

Physical model for the tunneling rate


Direct tunneling is modeled by the Kane-Keldysh expression [34, 35, 36]


 E
exp r
2 j 58

18 E


The values of the physical parameters are given below. This model is derived for the internal-field emission
in an infinite material, assuming a constant field and two-band interacting over
perturbation [35].
Because of the later, it is convenient for narrow-gap semiconductors, but not for silicon. Regarding to


.-0/

 2
%

1 2

& 2

1 2

1 2

&

3 2

2143

silicon MOS devices, the second restriction is the assumption of an infinite medium. In MOSFETs,
the tunneling occurs very close to the interface, where the movement becomes quantized in the normal
direction. Third important restriction is a variation of the electric field along the individual tunneling
paths, an effect normally taking place in devices.

CHAPTER 2 THE PHYSICAL MODELS OF MINIMOS

28

In Equation 2.58 stands either for an average field E * , being the tunneling length, or for the field
at the starting point . For the gap E the user can choose both, indirect and direct silicon band gap, in
spite of the former being physically unreasonable. TLIF and TEXF are fitting parameters with a value 1
as default.

We observed significant variations of the electric field along the tunneling paths, from the starting field
to the endfield .
ranges from 1 5 to 5 for GIDL problem in common cases. This ratio can also
be smaller than 1, as for tunneling towards the interface in the BBISHE injection in n-channel devices and
for the GIDL problem in p-channel devices. To analyze the impact of the field variation on the tunneling
)
) )
current, we develop a model for linearly variable field:
. It is based on WKBJ
 6
and the two-band 5 5 dispersion relation. Assuming that
is independent of the energy associated with
the momentum perpendicular to the tunneling direction, after several approximations we arrive at [31]

8 y r r o

|y 181 E  1 3 o 16 exp r 2  E 1 3 o 32

2 j 59

o 2 and Jy r o . In the derivation of Equation 2.59, field


with Jy

variations less then four times are assumed. The WKBJ part was multiplied with o 9 1 j 1 in the
transmission coefficient in a non-constant field (cf.
[36]). Because of the later, this model reduces
exactly to Equation 2.58 in a constant field case. For moderate field variations | becomes close to the

constant-field model
which employs an average field y E o y o 2, Fig. 2.1.
Indirect phonon-assisted tunneling employs the model by Keldysh derived for a constant field [36, 37]
y 32 2 E 1
2 j 60


with phonon-absorption and emission related factors:
y exp r 43 2  E6r
2 j 61


-y exp r 43 2  EW
2 j 62



2-7/

 :9

 2 
& 2
1 2


&

3 2

2143

% 2

.-0/

 *

 5 2

 2
";
& 9 22< = 7 4<
3 4 % 5 2


7 4

@

&

3 2

&4C

3 2

2143

2143

BA

being Bose-Einstein population

@

and

&4C

1 2

@

?>

&

@

5 2

1 2

r 1
z

1

& C

exp 

2 j 63

the absolute temperature.

y o

If the indirect tunneling with the variable-field model is specified, the Equation 2.60 is used assuming an
average field
2. After extensive comparisons between Equation 2.58 and Equation 2.59
we found an average-field approach to be satisfactory for direct tunneling problems if the field variations
are not too large. We may expect an average-field approach together with Equation 2.60 Equation 2.62
may be sufficient for engineering use regarding to phonon-assisted tunneling, since other effects are so
and so not accounted for, as the realistic band structure, manifoldness of the phonon branches and the

2.6 THE BAND-TO-BAND TUNNELING MODEL

29

interface-vicinity effect.
An anisotropy in silicon has not been accounted for in the models implemented. The anisotropy effects
seem to be small at higher fields which are of interest [38].
Note that the starting-field model and the 1D approach for the tunneling direction are relevant for the
investigation purposes only.

List of the physical parameters:

y j

E
&



elementary charge; 
1 602 10 19
for indirect band gap of silicon see Section 2.5;
for direct tunneling band gap in silicon we assume E
3 42 
&
&
reduced Planks constant (Diracs constant);
2%
1 0544 10 34 D

Boltzmanns constant;
1 3804 10 23 D
2E
free-electron mass in vacuum; 
9 108 10 31

1
;
reduced mass in tunneling process;
1
1

F ! ;
0 26  ,

0 1  , is used for both direct and indirect tunneling.
0 588 

The effective masses in silicon are:

0 98 , conduction band electron longitudinal mass;
0 19 , conduction band electron transversal mass;

F !
0 16 , valence band light hole mass;
!"
!
0 49 , valence band heavy hole mass.

For a possible impact of the field on the hole mass see discussion in [38].
number of valence band electrons per unit cell, taking place in tunneling;
2
;
product of the volume of unit cell and the matrix element for the character
2
istic interband phonon scattering in indirect tunneling process; 
<
=
<
1 0 10 62 D 2 3 is roughly estimated according to electron-phonon scattering coupling constants extracted by MC simulation of high-field transport in
silicon [39].
&4C
optical-phonon energy; 
60

y
y

y
y
y
y

&4C

.< =

<

y j q
y o y j t
y j oq

#y j
#y o o

j Cy
j
j
j
j
j
j

y q

Model description
The model is specified by the key MTUN in the RECOMB directive. The input data consists of three integer

numbers GIH
with the meaning:

CHAPTER 2 THE PHYSICAL MODELS OF MINIMOS

30
G
H

1 one-dimensional model in direction perpendicular to the oxide/bulk interface


2 two-dimensional model for the tunneling path
1 direct tunneling model, but with indirect band gap of silicon
2 direct tunneling (with direct band gap of silicon)
3 indirect (phonon-assisted) tunneling in silicon
4 indirect (phonon-assisted) + direct (direct band gap) tunneling
(superposition of 2 and 3)
1 average electric field along the tunneling path
2 field at the starting point
3 field linearly variable along the tunneling path

As example we present the calculation of a characteristics which can serve as a good test for each bandto-band tunneling model in comparison with experiment [28]: generation rate, drain or bulk currents are
computed versus drain-bulk voltage, while the drain-gate bias is kept constant for individual curves and
bulk-source bias is set zero, Fig. 2.1.
A roll-off in the family of the characteristics reflects well the
10-5
U DG =6.5V

-6

10

bulk current (A)

10-7
10-8
10-9
U DG =5V
-10

10

-11

10

-12

10

tox=8nm

average field
starting field
variable field

10-13
0

drain-bulk bias (V)

Figure 2.1: MINIMOS numerical calculation of the GIDL current due to the band-to-band tunneling
versus drain bias. Gate-drain bias is held constant. Impact ionization increases the terminal currents less
than 5% in this case.
two-dimensional nature of the potential and field distributions in the critical area and the impact of the
field variations along the tunneling paths on the tunneling current [31]. One of the input files used in the
calculation is

2.7 THE DYNAMIC INTERFACE AND BULK TRAP MODEL

BBT rate;
*
DEVICE
+
*
BIAS
STEP
*
PROFILE
INTERF
*
GRID
+
+
+
*
RECOMB
+
*
OPTION
OUTPUT
END

31

2D model, linearly variable field; UDG=5V


CHANNEL=N GATE=-0.540 DGAP=0.40E-4
TOX=8.0E-7 W=11.5E-4 L=0.510E-4

SGAP=0.40E-4

UB=0 US=0 UD=3.0 UG=-2.0


ND=29 DD=-0.100 LUDG=YES
FILE=2D
NSS=6.0E10 *to fit the threshold
PGMI=0.050E-7 GGMI=0.1E-7 MXT=4
YS=0 YE=15E-7 DY=1E-7
XS1= 50E-7 XE1=150E-7 DX1=10E-7
XS2=410E-7 XE2=460E-7 DX2=5E-7
MTUNN=213
SN=0 SP=0

TN=1E9 TP=1E9

MODEL=AVAL TEMP=298.0
NONE=YES
BIN=NO ERR=1E-3

MYT=2

CN=0 CP=0

PHYS=NO

Several keys deserve attention here: LUDG which determines that both, gate and drain bias are changed
with DD simultaneously, MXT and MYT and MTUNN. PHYS=NO enables the negative gate bias to be applied
on n-channel device. With SN=0, SP=0, CN=0, CP=0 and large TN and TP we suppressed other
generation-recombination processes, except the avalanche generation, in order to check the accuracy of
the terminal current calculation.

2.7 The Dynamic Interface and Bulk Trap Model


The first part of this section is devoted to physical modeling and to the numerical implementation of the
trap-related equations which are coupled with the basic semiconductor equations. The second and third
parts describe the interface and bulk traps implemented in this version of MINIMOS. Some examples are
presented in the fourth part.

CHAPTER 2 THE PHYSICAL MODELS OF MINIMOS

32

2.7.1 The Physical Model and Numerical Considerations


The model is based on numerical solution of the time-dependent basic semiconductor equations
1JLK

NPO?QIRBSUT
div M

R
V

Z[ eff

O7QIRYS]\

QIRYS
N X
_ QIRYSUT
_XQIRBSc\
V `
a
Z [ eff
b
Q
S div M
R
Z
V
J
QedcK
QIRYSBSUT
K7f QIRYS
IQ RYS
QIRBSik
h jml
n
Z `
Z ?
div
grad
W
W g
_

Q
jUo
prq
Z

2 ^ 64

1J

J
[
where

QIRYS

VXW

2 ^ 65
S

2 ^ 66

and [ eff are the effective (net) generation rates for electrons and holes, respectively, and
is the total trapped charge. Trap-related quantities do not appear in the total-current-continuity
equation which is solved at the beginning of each time step (see Section 1.4):
K

W?g

QIRYS eff

div s

d]K

N O h
N _
V T
grad u
div M
div M ^
R t
Q

2 ^ 67

Our model allows arbitrary interface and bulk trap distributions in both the energy and position space.
jwv
For an individual single-level interface (or bulk) trap with density , located at the position xzy along the
v
channel, at energy level E , the trap dynamics is described by the equations
{ O T

jwv

Z W
| }O
~

v
\

{_wT

{ O

W
| }_
~

v
\

O T
[

W
| O
~

v
\

{_

_T

j'v

Z W
| _
~

v
\

2 ^ 68
_

where
is the electron capture rate,
the hole capture rate, [
the electron emission rate and [ is
v
the hole emission rate. The occupied trap density is denoted by W . The electron and hole capture and
emission time constants are given by
\
T v Y_ K"2_K
| }l_ 1 
| } l O 1 T v O K O K i
W
`7
v
v

\
T v Y_
K _K"j'K
E} Z E
E Z E
| l O 1 T v O K O K"j } K
| l_ 1 
exp s Z K~ t
exp s Z ~K t
v O

v_

2 ^ 69
S

2 ^ 70

2_

,
are the average thermal velocities towards the interface and ,
the average capture crossj
j'

sections for electrons and holes. } , E} ,


and E are effective density of states and energy of band edge
for conduction and valence band, respectively. Wi and `7 are the electron and hole surface concentrations
at the position x0y . In this model the conduction and valence band are reduced to single energy levels whose
population is governed by Maxwell-Boltzmann statistics. When applying Equation 2.68, Equation 2.69
and Equation 2.70 to non-equilibrium conditions, it is postulated that the capture cross-sections associated
with the emission and capture processes remain constant and equal to each other as in equilibrium. The
present model is known as Shockley-Read-Hall theory. An eventually present dispersion in the values of
the capture cross-sections associated with single-energy levels is currently not implemented.
The trap-dynamics equation is given by

T{rO

Z[

O
Z

{ _ h
[

_ T
Z

W
|

h*

K"j'v
|

2 ^ 71

where * being the steady-state occupancy function and | an effective time constant which determines the
actual trap dynamics
| l 1 T

| }lO 1 h

| l O 1 h

| }l_ 1 h

| l_ 1 \

| KQ | } l O 1 h

| l _ 1S ^

2 ^ 72

2.7 THE DYNAMIC INTERFACE AND BULK TRAP MODEL

33

The trap-trap transitions, when more trap levels exist in the forbidden gap, are not accounted for in
Equation 2.71 in the present implementation. Such an approach is proper for interface traps. In spite
of their possible close location in the energy space, interface traps are separated in the position space.
The electron wave functions are very localized at the trap sites. It is believed that the wave functions of
neighbour interface traps do not overlap. However, trap-trap transitions may take place in case of bulk
traps.
The trap-dynamics equations are discretized by a strongly implicit method. The net generation during an
J
J
individual time
step, associated with the trap,
is calculated consistently with the discretization. The charge
K v
h
KaQj'v
vS
Z W
becomes Z
for donor-like traps. To solve the rate equations the
W for acceptor-like and
R
time steps (say ) for solving the basic semiconductor equations (Equation 2.64, Equation 2.65, Equation
2.66 and Equation 2.67) are divided into NTSI subintervals. The key NTSI belongs to TRANS directive.
R
The carrier concentrations (W , ` ) are exponentially interpolated in the time interval . By this approach
we can reduce the time-discretization error in the rate equations for the capture-related processes. Of
course, this refinement is irrelevant for the non-steady-state emission. Finding a compromise between the
computation time and the accuracy, controlled by the time stepping (see Section 1.4) and the parameter
NTSI, requires experience and physical reasoning for each specific class of problems. Some directions
are given in [2]. In addition to the net generation and the trapped charge associated with the individual
traps we calculate the corresponding derivations of these quantities with respect to the local carrier
concentrations and the potential.
Both interface and bulk traps are discretized in the energy space as explained below. To calculate the
total contribution from traps at the particular position the contributions from all individual trap levels are
R
superposed. The total net generation during , the trapped charge and their derivatives at the particular
position are coupled with the basic semiconductor equations. At each time step the complete system is
solved self-consistently in the whole device.
To accelerate the convergence of the discretized basic equations the derivatives of the net generation
rates and the trapped charge are taken into account. This is indispensible to achieve convergence of the
iterative algorithm when high trap-densities are assumed in device. Within the limits specified on the
INTERFACE and PROFILE directives the convergence properities are quit satisfactorily in both steadystate and transient conditions and for donor-like as well as acceptor-like interface and bulk traps at room
temperature. These limits are much higher than densities commonly taking place in practice, but the
conditions become worse at low temperature.
A more detailed discussion about the time-discretization of the trap-dynamics equations and the basic
semiconductor equations, the time-discretization error and the numerical methods can be found in [2].
In the steady-state the implemented equations reduce exactly to the well known Shockley-Read-Hall
trap-assisted recombination model.

2.7.2 Interface Traps


Various distributions of interface traps at the oxide/bulk interface, which are accounted for in the transient
simulation, can be specified on the INTERFACE directive.

CHAPTER 2 THE PHYSICAL MODELS OF MINIMOS

34

AUUD, DUUD: Traps uniform in both the position and energy space.

ARUD1,ARUS1,ARUE1 and ARUD2,ARUS2,ARUE2: Traps rectangular in the position space


K~Q
and uniform in the energy space. The total amount of traps in the device is exactly 2.. 2 Z
SK
P2za
, independent of the x -discretization mesh.
AGUD1,AGUS1,AGUX1, AGUD2,AGUS2,AGUX2 and DGUD,DGUS,DGUX:
Gaussian distributed traps in the position space and uniform in the energy space.

vQ
x

ScT

exp Z

.a4

x Z
s

.aX?

2P0z

2 ^ 73

In MINIMOS the origin of the x -axis is the source-sided gate edge. Note that, although we called
AGUS the standard deviation, it is 2 times larger than this quantity in the mathematical sense.

AULD, DULD: Traps uniform in the position space and linearly distributed in the energy space.
The traps are distributed in the forbidden gap from 0 at the valence band to AULD or DULD at the
conduction band.
AUGD,AUGS,AUGX and DUGD,DUGS,DUGX: Traps uniform in the position and gaussian distributed in the energy space.

vQ

ScT

XP

E Z 2 .

exp 7Zs

The total amount of traps per unit area is exactly

2
vYQ

2 ^ 74

E E for all values of DUGX and DUGS,

independent of the energy discretization mesh. The location of the peak should be given with
respect to the valence band edge for both type of traps.

AGLD,AGLS,AGLX and DGLD,DGLS,DGLX: Gaussian distributed traps in the position space and
linearly in the energy space. The traps are distributed in the forbidden gap from 0 at the valence
band to AGLD or DGLD at the conduction band (for the peak in the position space) and analogous
to Equation 2.73 in the position space.

For SOI devices interface traps can be specified at the silicon-film/bulk-oxide interface.

BAUUD, BDUUD: Traps distributed uniformly along the back-interface in SOI devices, uniform in
the energy space.

For all traps the first letter determines the nature of the traps, A standing for acceptor-like and D for donorlike. The interface traps may be arbitrarily combined. The total trap density becomes a superposition of
the individual densities specified. Negative values are also allowed on input (they will be subtracted), but
the key PHYSCK=NO should be set on the OPTION directive.

2.7 THE DYNAMIC INTERFACE AND BULK TRAP MODEL

35

Ja
An equidistant discretization of the energy space has been implemented
for a continuum of interface
v
g between two adjacent energy
states from the valence to the conduction band, with spacing E
levels. With this choice the equilibrium Fermi-Dirac occupancy function is well resolved. Regarding
the non-steady-state emission occupancy function , in each time decade about 8 trap levels effectively
T
ve
contribute to the signal. The number of energy levels is E
E* E . Note that E is temperature
dependent, becoming large at low temperature. By the key MENErg on the INTERFACE directive the
user can specify E externally.

The capture cross-sections are specified on the RECOMB directive:

CSEA, CSHA cross-section for all acceptor-like interface traps for capture of electrons and holes,
respectively
CSED, CSHD cross-section for all donor-like interface traps for capture of electrons and holes,
respectively

The thermal velocities are given by


v O T yv KP

yv

107

QIO

vY_ T yv K

where
3 ^ 35
' , g is the thermal voltage and
models of which can be found in Section 2.5.

QI_
O

2 ^ 75

are the effective masses the

Other interface traps which can be specified by the keys FSS, PFSS, DA0 and DD0 on the INTERFACE
directive are populated in the transient conditions as in the steady-state.

2.7.3 Bulk Traps


All bulk traps are assumed to be uniformly distributed in the whole semiconductor. They are accounted
for in the Dirichlet boundary conditions at the ohmic contacts. Dynamics of all bulk traps is included in
the transient simulation. The following sets of keys can be specified on the PROFILE directive:
1. single-energy level deep donors and acceptors defined by:
CD concentration of deep donors
CA concentration of deep acceptors
ED activation energy of deep donors
EA activation energy of deep acceptors
2. donors and acceptors distributed uniformly in energy:
VAUD density of acceptors
VDUD density of donors

CHAPTER 2 THE PHYSICAL MODELS OF MINIMOS

36

3. donors and acceptors gaussian distributed in energy:


VAGD peak density of acceptors
VAGS standard deviation of the distribution for acceptors
VAGX peak position of the distribution for acceptors
VDGD peak density of donors
VDGS standard deviation of the distribution for donors
VDGX peak position of the distribution for donors
At present only traps from one set (either 1, 2 or 3) can be specified simultaneously. This for practice not
too strong restriction is motivated by minimization of the trap levels for the discretization in the energy
space, in order to reduce the computation time and memory consumption. The generated mesh in the
energy space is different for acceptor-like and donor-like traps in general. ED is measured with respect
to the conduction band edge, but EA with respect to the valence band edge. Both, VAGX and VDGX are
defined with respect to the valence band edge.
The single-energy level traps are particularly interesting for todays MESFETs, which are nearly exclusively fabricated on semi-insulating substrate containing a considerable concentration of deep donors and
deep acceptors. The corresponding lifetimes of these traps are
T

| O p
| O n

| _ p

K~ v O K

| _ n

K v O K

K~ vY_ K

K" vY_ K

2 ^ 76
S

2 ^ 77

with capture cross-sections introduced below. Typical values for deep-donors and acceptors in GaAs are:

2
T

4

EL2
Cr

0 ^ 76 J
0 ^ 76


4
T

5 10 16
K

l
1 ^ 2 10 19

.
T

l 18 2
17 2

2 10
K
l
5 10

Traps distributed uniformly in the energy are discretized equidistantly in the energy space according to
K
T
J
the key VDEN, which belongs
to PROFILE directive. We set ..
2 g as default. The number of

trap levels becomes E 4. , but with maximum limited to 25.

Gaussian distributed bulk traps are discretized with an equidistant energy-mesh placed symmetrically

around the peak position in the 3 interval. For the acceptor-like traps the difference between adjacent
energy-levels becomes
T min

2.
S

2^ 5

S
Z

4.

max 0 2. Z 2 ^ 5

2 ^ 78

g
.42 2 as default. In common cases traps are discretized with 10 levels.
with 42
min 2
For donor-like traps an analogous formula follows. The total amount of traps per unit volume is given

exactly by

E

0

.P

exp s Z

f E l ~~ 2
E for all values of VAGX and VAGS, independent of the energy
~~~ q t

discretization mesh.
The capture cross-sections of bulk traps are specified in the RECOMB directive:

2.7 THE DYNAMIC INTERFACE AND BULK TRAP MODEL

37

CAN, CAP cross-section for all acceptor-like bulk traps for capture of electrons and holes, respectively

CDN, CDP cross-section for all donor-like bulk traps for capture of electrons and holes, respectively

The thermal velocities are

v O T v K

with

v T

13 ^ 4 107

QI

vY_ T v K

QI _

2 ^ 79

 .

The bulk traps are resolved in time as the basic semiconductor equations, with other words the key NTSI
is set to 1 internally.

2.7.4 Examples
The present model is verified against several experimental data for the charge-pumping effect. Some
comparisons with experiment can be found in [40, 41, 42].
As an example for interface traps we analyze the numerical simulation of the charge-pumping experiment
used to extract the spatial distribution of interface states [43]. MINIMOS is employed to calculate the
family of the characteristics for an assumed trap distribution J localized in the drain junction, Fig. 2.2. The
l 2
l 1

input file for the curve related to the peak density 1012
is:

acceptor-like int-traps; peak of damage 1.0E12


*
DEVICE
CHANNEL=N GATE=-0.54 TOX=12E-7 W=100E-4 L=0.80E-4
+
DGAP=0.5E-4 SGAP=0.5E-4 YMAX=0.8E-4
*
BIAS
US=0 UD=0 UG=6.075 UB=0.175
STEP
NG=48 DG=-0.075 LUGB=YES
TRANS
TG1=1.0E-6 TG2=0.1E-6 TG3=0.4E-6 TG4=0.1E-6 TG5=1.0E-6
+
WG=2 UG2=9.0 UG3=0.16 UG4=0.16 NTSI=2
*
PROFILE
NB=9E15 ELEM=P DOSE=6E14 AKEV=60
+
TEMP=950 TIME=2200 TOX=25E-7
PROFILE
XOFF=0.1E-4 ELEM=AS DOSE=4E15 AKEV=160
IMPLANT
DOSE=1.7E12 ELEM=B TIME=2200 TEMP=950 AKEV=10
IMPLANT
DOSE=7E11 ELEM=B AKEV=30
*
GRID
PGMI=0.1E-7 GGMI=0.2E-7
+
XS1=0.060E-4 XE1=0.110E-4 DX1=4E-7
+
XS2=0.690E-4 XE2=0.740E-4 DX2=2E-7
*

CHAPTER 2 THE PHYSICAL MODELS OF MINIMOS

38

RECOMB
*
INTERF
+
*
OPTION
+
OUTPUT
END

AN=0 AP=0 SN=0 SP=0 CN=0 CP=0 TN=1E9 TP=1E9


NSS=0 AUUD=2.0E10 MENER=100
AGUD1=1.0E12 AGUS1=10E-7 AGUX1=720E-7
MODEL=AVAL TEMP=300 PHYS=NO
LSOURCE=0.5E-4 LDRAIN=0.5E-4
NONE=YES
BIN=NO ERR=1E-3 PACC=1E-8

GRIDFR=YES
LCPUMP=YES

Input file used to calculate the characteristics presented in Fig. 2.2 (from [43]).
The input file for the curve denoted as uniform in Fig. 2.2 is the same, but without the second line in the
INTERFACE directive.
The interesting keys in this example are:
LUGB
WG=2
TG1-TG5
UG2-UG4
NTSI=2
XS1,XE1,DX1
XS2,XE2,DX2
AN,AP,SN,SP
CN,CP,TN,TP
AUUD
AGUD1
AGUS1,AGUX1
MENER
MODEL
PHYS
GRIDF
LCPUMP
PACC

UG and UB are changed simultaneously with DG.


trapeze waveform on the gate contact. Bias at other contacts is kept constant
during each transient simulation.
time parameters of the trapeze gate waveform.
voltage parameters of the trapeze gate waveform.
to solve the trap-dynamics equations each time step is divided into 2 timesubintervals.
additional grid lines in the source junction and, particularly, in the drain junction where the damage is assumed.
suppress all additional generation-recombination processes
specifies uniform trap distribution in virgin device.
model for damaged region in stressed device.
number of energy-levels for interface traps.
AVAL model is necessary to activate the generation-recombination processes.
to allow positive bulk bias in n-channel device.
freeze the grid.
print all charge pumping relevant data in an external file for postprocessing.
increase the accuracy of solving the Poisson equation.

The next example assumes bulk traps in an SOI pin-diode made in polycrystalline thin-film technology.
In general, using the pin-diode structure we can avoid the convergence problems which may occur when
the Gummel successive algorithm is applied to floating-body SOI devices. Pin-diode as a test structure
is also very useful for measurements in SOI technology [44]. In this example we omitted interface
traps, which normally occur in real devices. Assuming surface-trap density at the grain boundaries in the
j v
T
l 2
T

polysilicon
1 ^ 5 1012
20W , it follows the equivalent
and the average grain size c

2.7 THE DYNAMIC INTERFACE AND BULK TRAP MODEL

charge-pumping current (nA)

39

Parameter: peak density (cm-2 eV -1 )

1012

f=1MHz
t r =t f =0.1 s

UG =9V
UGL =-3.1V

LG=0.8 m
t ox =12nm
W=100 m

11

10

uniform

1
0

0.5

1.0

1.5

2.0

2.5

3.0

drain-bulk voltage (V)

Figure 2.2: Calculated charge-pumping current


models of virgin and stressed device.
j v

Tj

}_

versus reverse junction bias

3.5

for assumed

3 . We model the polycrystalline silicon with two


7 ^ 5 1017
bulk-trap density

gaussian distributed traps in the energy space, located symmetrically with respect to the intrinsic level
(VAGX,VDGX). The upper distributions are acceptor-like and the lower donor-like traps. For the chosen
standard deviations VAGS
and VDGS the peak densities VAGD and VDGD are calculated such that the
j v
total density becomes per unit volume. Although empirical, this model reflects most experimental
observations in the polycrystalline silicon. We, however, do not account for the specifics of mobility in
the polysilicon. Since the bulk traps in MINIMOS are uniformly distributed in the film, our approach is

proper for small-grain film only (assumed 20W ).

SOI pin-diode; polysilicon film


DEVICE
CHANNEL=N GATE=NPOLY TOX=40E-7 W=100E-4 L=5.0E-4
+
FILM=0.15E-4 BULK=0.380E-4 KBULK=3.9
BIAS
UG=-3.0 UD=0 US=0.1 UB=20.0
STEP
NG=28 DG=0.5
TRANS
TG1=100E-6 TG2=1E-6 TG3=49E-6 TG4=1E-6 TG5=150E-6
+
WG=2 UG2=8.0 UG3=0.2 UG4=0.2
PROFILE
NB=5.0E16
+
TOX=30E-7 TEMP=980 TIME=9000
+
ELEM=P DOSE=7.0E15 AKEV=90
+
ASYM=Y
+
VAGD=3.0E18 VAGX=0.650 VAGS=0.070
+
VDGD=3.0E18 VDGX=0.450 VDGS=0.070
PROFILE
ELEM=B
DOSE=8.0E15 AKEV=35
OPTION
MODEL=AVA TEMP=293 PHYSCK=NO LCP=YES

CHAPTER 2 THE PHYSICAL MODELS OF MINIMOS

40

charge-pumping current (nA)

20

SOI pin-diode; polycrystalline film


LG=5m
W=100m
tox=40nm
tsoi=150nm
tbox=0.38m

15

Ur =0.1V
UB =+20V

UG =8V

10

f=10kHz
tf =1s
tr=1s
d.cycle 50%

0
-14

-12

-10

-8

-6

-4

-2

gate-base level (V)


Figure 2.3: Numerical calculation of the charge-pumping current versus gate-base level characteristics of
an SOI pin-diode made in the polycrystalline silicon film.
+
INTERF
OUTPUT
RECOMB
+
END

LSOURCE=1E-4 LDRAIN=1E-4
AUUD=0 BAUUD=0 PSS=4E11
NONE=YES
TN=1E9 TP=1E9 SN=0 SP=0 VN=0 VP=0 CN=0 CP=0 AN=0 AP=0
CAN=1E-15 CAP=1E-15 CDN=1E-16 CDP=1E-16
BIN=N ERR=1E-3 PACC=1E-6

Input file used to calculate the charge-pumping curve presented in Fig. 2.3.
The key ASYM=Y defines that the second PROFILE directive is used for the drain junction. On the
second PROFILE directive ELEM=B is specified, while on the first ELEM=P is specified for the drain and
the source, respectively. To suppress the MINIMOS profile-clash error we use PHYSCK=NO. The total

simulation time covers 1 ^ 5 signal periods. We calculate the charge-pumping currents in the time interval

K~\
K~
0^ 5
1^ 5
in a post-processing step from the data written to the file units NUMCP1 and NUMCP2.
Several examples for the transient simulation including interface traps are given in [2, 40, 41, 42, 43, 45].

2.8 Gate Depletion Analysis


Due to low chemical concentration of impurities and their partial activation (both caused by several
effects [2, 46, 47, 48]) the implanted gates, commonly used in submicrometer technology, can be

2.8 GATE DEPLETION ANALYSIS

41

nondegenerately doped. As a consequence of the band-bending, caused by the penetration of the electric
field, these gates can no longer be assumed to be an equipotential area. The effect may have a profound
impact on the device characteristics and performance.
Two approaches to analyze the gate depletion effect are implemented in MINIMOS. The first is a full
two-dimensional solution to the problem, which can be specified by the key MPOLY=1 in the OPTION
directive. The second is an approximative one-dimensional approach based on the gradual channel
approximation in the gate, specified by MPOLY=0. In both models the activated impurity concentration
in the gate is given by the key PDOP in the OPTION directive, which also activates the gate depletion
analysis. The gate is assumed to be uniformly doped with a level specified by PDOP.

2.8.1 The Full Two-Dimensional Model (MPOLY=1)


MINIMOS has been extended to solve self-consistently the basic semiconductor equations also in the
v
poly-gate area, (Fig. 2.4). Poissons equation is solved in the total simulation area (from to ).
Considering the continuity equations, we restricted ourselves to steady-state condition for this MINIMOS
version. In the steady-state the poly-gate is in thermodynamic equilibrium. In that case the leakage
currents are negligible and the net recombination vanishes. A unique and constant Fermi level exists in the
poly-gate, which enables the carrier concentrations to be calculated analytically as a function of the local
potential . We accounted for the band-gap narrowing and for Fermi-Dirac statistics. A rigid-parabolicband model is assumed to be valid at doping concentrations of interest. Assuming a uniformly doped gate
it follows
Q
W

where

STj } K

1 2 s

and `

STj'K

1 2 s

2 ^ 80

j } jw
,
are the effective density of states of the conduction and valence bands, respectively. It holds
J
h
Q
S
S
\
Q
T
}
Z EuZ E

2 ^ 81

where E is the ideal band gap, discussed in Section 2.5 and E is the total band-gap narrowing. The

quantities } and can be calculated by

K" l 1 j
j }Y \
K l 1 j
jw
Q
S
T

2 ^ 82
1 2
1 2
j
for N-type and P-type gates, respectively. is the activated impurity concentration in the gate, specified
T
}

by the key PDOP. At the gate contact neutral majority carrier concentration is assumed, which equals
the activated impurity concentration. The top gate potential with respect to the Fermi level in the source
which is imposed as boundary condition is given by

} Z

E}

hQ

E} Z

Ey

2 ^ 83

with E being the band-gap narrowing at the conduction band. is the terminal voltage and E} , Ey

denote the conduction band edge and the intrinsic level in the ideal silicon band. The Fermi integral 1 2
and its inverse are calculated by analytical approximations proposed in [49, 50].

CHAPTER 2 THE PHYSICAL MODELS OF MINIMOS

42

Figure 2.4: Schematic presentation of the simulation area for the two-dimensional model. The dot-dashed
curve is the contour referred to the field-flux calculation.
The band-gap narrowing in the gate is modeled by an empirical expression
E
j

32 ^ 0 ln 1

h
s

j
K

4 1018

h
s

j
K

9 1018

QI

S\

2 ^ 84

3 . Equation 2.84 is in agreement with rather scattered experimental data for the
where is given in
gap narrowing in the quasi-neutral regions from the literature. No temperature dependence of E has been
T
implemented. In absence of any experimental data a symmetrical gap narrowing is assumed: E }
E 2.
A criticism regarding the application of the model Equation 2.80 to heavy doped space-charge regions can
be found in [2, 51].

The height of the gate electrode can be specified by the key GCTHE in the GEOMETRY directive.
While adapting the mesh to the doping profile, before the model 1D is executed, MINIMOS will adapt
Qj S
. Additional mesh refinement in the gate
the y-mesh in the gate according to the Debye length
area takes place in the model 1D and is controlled by the maximum potential difference allowed between
adjacent grid lines, specified by the keys DPXP and DPYP in the GRID directive.

2.8.2 The Approximate One-Dimensional Model (MPOLY=0)


Let denote the difference between the potential at the gate/oxide interface and the top-gate
Q SrT
Q S
potential , which is determined by UG: x
Z x . It is obvious that the value of

can range from roughly Z 200 in the strong accumulation to 1 ^ 2 in the strong inversion, assuming
an N-type poly-gate. The maximum difference between at the source side and the drain side is
limited on about 1 ^ 4 , whereas the corresponding difference in the channel is roughly p . Therefore,
the poly-gate may be considered as a MOS device with a small drain-source bias. In such devices
the gradual-channel-approximation (G.C.A.) and the capacitance relation ( 0 ) are reasonable

assumptions near the interface. Let us assume that the MOSFET is in saturation. The G.C.A. is valid
then in the so-called source channel-part. We may expect that it holds in the corresponding part of the
poly-gate as well (just on the other side of the oxide). The G.C.A. does not hold in the drain channel-part
and probably vis-`a-vis in the poly-gate too, however the gate has a small influence on the bulk in this

2.8 GATE DEPLETION ANALYSIS

43

region. Consequently, where the gate controls the channel, the G.C.A. holds in the gate. Unlike in the
bulk, two-dimensional effects are of a second order in the poly-gate, except just at the corners and we can
assume the problem to be one-dimensional. The surface potential depends only on the local field in
the polysilicon, perpendicular to the gate/oxide interface (Fig. 2.5). Accounting for Fermi-Dirac
statistics, after one integration of the one-dimensional Poisson equation, we obtain

j }

h
} Z
}
t Z
t
3 2 s
3 2 s

g
g
T:

d_

j'

h
j

3 2 s

T


t Z
t
3 2 s

g
g
s

c
g

2 ^ 85

d_

with c
g
and } ,
defined previously [52]. For the permittivity of the single

crystal undoped silicon is assumed. The transcendental Equation 2.85 is solved by Newtons iterative
method, applying limiting and damping on the potential increment, which should converge absolutely.

Q S
For the Fermi integral 3 2 x we employed the approximation proposed in [50].
The relation between

and the normal field in the oxide is

d_

v
y
d _

v
y
\

2 ^ 86

where we accounted for the fixed charge and the trapped charge of both, donor y and acceptor
v

nature y at the gate/oxide interface. Assuming we have no interest in the gate-side effects (like

sidewall capacitances) we can neglect the actual gate thickness. The model reduces to the conventional
Q S]T
planar MOSFET simulation, but with the gate-contact potential variable along the x coordinate: x
Z . In MINIMOS, this is implemented by updating the gate potential boundary condition after each
Q S
Q S
Poisson/continuity equation solution-step by calculating x for a given local x . This calculation

is very fast compared to the two-dimensional solution of the Poisson equation in the gate. Moreover, the
additional x and grid lines necessary to resolve the gate/oxide interface, as well as grid adaptation are
avoided. This is an important savings, since the depletion region in the poly-gate is usually quite small,
due to the relatively large doping levels and requires a fine grid to resolve it. Damping of the potential
changes applied to the gate may employed to speed convergence of the simulation.

GO (x)

EGO

x
y
Figure 2.5: Simulation area for the one-dimensional model is the same as in the conventional calculation.
Q S
The boundary condition for the gate potential x is variable from 0 to . It depends on the local
Q S
normal field x in a selfconsistent manner. The dot-dashed curve refers to the field-flux calculation.

For both models MPOLY=0 and MPOLY=1, the type of the gate electrode is determined by GATE in the
DEVICE directive, while PDOP should be a positive number. All four cases, n- and p-channel devices
with N- and P-type of gate are supported.

CHAPTER 2 THE PHYSICAL MODELS OF MINIMOS

44

2.8.3 Fixed Charge and Traped Charge at the Gate/Oxide Interface


The strong influence of the charge at the polysilicon-gate/oxide interface on the field penetration into the
gate and, therefore, particularly on the flat-band and threshold voltages has been demonstrated [52, 53].
There is not much information about the nature of this charge in the literature. Since the polysilicon is
deposited over the oxide, it is believed that the gate/oxide interface is worse than the interface between
l 2
thermally grown oxide and bulk-silicon. A positive total interface charge of order 1012
has been
obtained experimentally in [54]. The acceptor type interface traps have been speculated in [55] and
the fixed charge, traps and dipole layers have been proposed for heavy doped gates in [56]. We have
implemented in MINIMOS:
PNss
PFss
DA0
DD0

fixed charge
traps uniformly distributed in the forbidden gap. Positive values implicitly
denote donor-like and negative values acceptor-like traps.
acceptor-like traps parabolically distributed in the energy space from 0 at the

0 at the conduction band edge.


valence band edge to

donor-like traps parabolically distributed from


0 at the valence band edge
to 0 at the conduction band edge.

For example, the density distribution for the latter trap type is given by

v Q

ST
y

v0 K
s

E} Z E
t
E} Z E

2 ^ 87

After considering several data sets, we have introduced the parabolic density-distributions to enable better
fitting of experimental data. An example is given in Fig. 2.8 (cf. in [51]).
The trap population in determined by the local Fermi level at the gate/oxide interface. Fast and stable
Gummels iterative procedure, when high trap densities are presented, is achieved by employing the
derivatives of the trapped charge with respect to the local potential to predict the new potential increment.
All charges are taken into account in both MPOLY=0 and MPOLY=1 model. For the one-dimensional
model the charge is uniformly distributed along the gate, from 0 to (Fig. 2.5). In the two-dimensional
model the charge is uniformly placed along the whole non-planar poly-gate/oxide interface (Fig. 2.4).
The traps at the grain boundaries in the polysilicon have not been taken into account in the present model.
A reason is given in [51].

2.8.4 Examples
In the following example the accuracy of the numerical technique for calculating the quasi-static gatecapacitance is demonstrated. The result obtained by MINIMOS is compared with an analytical (exact)
solution to the gate-depletion problem, Fig. 2.6. The analytical model is explained in [52]. This
comparison, of course, does not include any fitting. The input file used for this calculation reads:

2.8 GATE DEPLETION ANALYSIS

45

MINIMOS versus analytical result; Poly-gate depletion: mpoly=0


*
DEVICE
CHANNEL=N GATE=NPOLY TOX=7E-7 W=1000E-4 L=100E-4
+
DGAP=0.5E-4 SGAP=0.5E-4
BIAS
US=0 UB=0 UD=1E-3 UG=1.0
STEP
NG=25 DG=-0.100 VG=0.025
PROFILE
NB=1E17
+
ELEM=P DOSE=2E14 AKEV=60
+
TEMP=915 TIME=1800 TOX=25E-7
PROFILE
XOFF=0.10E-4 ELEM=AS DOSE=5E15 AKEV=150
GEOME
GCTHE=0.2E-4
INTERF
PNSS=2E11
OPTION
MODEL=AVAL TEMP=293 PHYS=NO GRIDF=YES
+
PDOP=8.0E18 MPOLY=0 GCHC=YES
OUTPUT
NONE=YES
END
BIN=NO ERR=1E-3

Input file used to calculate the characteristics shown in Fig. 2.6.


The keys particularly interesting in this example are:
GCHC
GCthe

PDop
MPoly
DG,NG
VG
GRidf

Invokes calculation of the total gate charge (OPTION directive)


Thickness of the gate contact (GEOMETRY directive). Since in this example
the one-dimensional model MPOLY=0 is specified, the gate-contact height is
not important.
activated dopant concentration in the gate (OPTION directive)
model for the gate depletion analysis (OPTION directive)
do step changes of the gate bias UG (STEP directive)
for each gate bias UG perform two DC calculations for UG-VG/2 and
UG+VG/2 (STEP directive)
freeze the discretization mesh

The voltage drop in the poly-gate is shown in Fig. 2.7. Because the model MPOLY=0 does agree well with
the analytical result, the discretization, as well as the numerical error are judged as very small in the oxide
and in the bulk. Differences between MPOLY=1 and the exact solution occur due to the discretization
error caused by a too coarse -mesh in the gate area. By reducing DPYP parameter the error at inversion
and accumulation in the gate can be suppressed.
In the next example the inversion part of a quasi-static gate-capacitance curve is calculated.
P-gate/P-channel device; inversion part
*
DEVICE
CHANNEL=P GATE=PPOLY TOX=12.5E-7 W=100E-4 L=99.5E-4
+
DGAP=0.5E-4 SGAP=0.5E-4
BIAS
US=0 UB=0 UD=-0.001 UG=-10.0
STEP
NG=20 DG=0.400 VG=0.025
PROFILE
FILE=2D
INTERF
NSS=1E10 PFSS=0 PNSS=1.0E12 DA0P=6.0E12 DD0P=0
MOBIL
MT=1E4 MC=2

CHAPTER 2 THE PHYSICAL MODELS OF MINIMOS

46

numerical calculation versus analytical solution

2
gate capacitance (fF/ m )

tox =7nm

17

ideal
exact
mpoly=0
mpoly=1

N B =10 cm
+
18
-3
N G =8 10 cm

-3

0
-1.4 -1.2 -1.0 -0.8 -0.6 -0.4 -0.2

0.2

0.4

0.6

0.8

1.0

gate-bulk voltage (V)

Figure 2.6: Comparison between numerically calculated quasi-static gate capacitance and analytical
j
T
K
l 3
8 1018
. Bulk is uniformly doped
(exact) result. Gate depletion is analyzed. Gate doping is
j T

l 3
. The ideal characteristics is the analytical results for a negligible gate depletion.
1017
with
numerical calculation versus analytical solution
1.2

tox =7nm

17

potential drop in the gate (V)

1.0

-3

N B =10 cm
+
18
-3
N G =8 10 cm

0.8
0.6

exact
mpoly=0
mpoly=1

0.4
0.2
0
-0.2
-5

-4

-3

-2

-1

gate-bulk voltage (V)

Figure 2.7: Voltage drop in the gate for the same device as in the previous figure. Note that the differences
between the solid-line (analytical results) and the dashed line (MPOLY=0 MINIMOS model) are negligible.

2.9 POWER VDMOSFET SIMULATION

OPTION
+
GEOM
GRID
+
OUTPUT
END

47

MODEL=AVAL TEMP=295 GRIDFREEZ=YES


PDOP=1.2E19 MPOLY=1 GCHCAL=YES
GCTHE=0.2E-4
XS1=0
XE1= 0.8E-4 DX1=0.1E-4
XS2=98.7E-4 XE2=99.5E-4 DX2=0.1E-4
NONE=YES
BIN=NO ERR=1E-3

Input file used to calculate the inversion part of the quasi-static gate-capacitance presented in Fig. 2.8
(solid-curve).
30

gate capacitance (pF)

25

20

15

10

experiment
calculation

tox=12.5nm

0
-10

-8

-6

-4

-2

gate-source voltage (V)

Figure 2.8: Quasi-static gate capacitance calculated by MINIMOS using MPOLY=0 compared with
experimental data.
The keys particularly interesting in this example are: Gate, PDop, MPoly, PFss, PNss, DA0p, DA0p
and GCthe.
Both the one-dimensional and two-dimensional models of the gate depletion effect can be combined with all
other MINIMOS features, as for instance transient simulation and band-to-band tunneling.

2.9 Power VDMOSFET Simulation


MINIMOS has been extended to be capable to account for power VDMOS structure shown in Figure Fig.
2.9.
The electrical characteristics of VDMOSFETs are determined by four characteristic parts: the channel
region formed by the double-diffusion, the distributed resistive layer close to the oxide/silicon interface,

CHAPTER 2 THE PHYSICAL MODELS OF MINIMOS

48

Figure 2.9: Two halfcell cross section of power VDMOSFET.


l

the parasitic JFET between the P junctions and the epi-region, the rest of the epi-region which acts
o
as a resistor and the N (buried) layer contacted by the drain. This complex structure posses severe
difficulties to analytical modeling approaches. In particular, the channel region alone determines the
threshold voltage of VDMOSFET and the saturation transfer characteristics at low gate biases. Unlike
the conventional MOSFETs, the channel doping decreases sharply along the interface towards the drain
in VDMOS transistors. Moreover, the channel of modern power VDMOSFETs is sufficiently short that
the mobility degradation and the drift velocity saturation due to high lateral fields must be accounted for
in modeling in addition to the surface mobility variation with the perpendicular field [57, 58]. The drift
velocity saturation also occurs within the parasitic JFET [59, 60] which has a non-uniform geometry
and the length/depth ratio so low that the conventional gradual-channel-approximation becomes invalid.
It has been proposed that the carrier velocity saturation is responsible for the quasi saturation effect in
the output characteristics of VDMOSFETs [60]. Recent studies have proposed that the quasi-saturation
effect even involves more complex phenomena [57, 61]. The current flow in the rest of the epi-layer
is strongly twodimensional [62, 63]. The distributed epi-layer close to the interface where the current
flow strongly changes in direction can be either accumulated by majority carriers (in the linear and the
quasi-saturation region) or free of them (in the saturation region), which depends on the bias conditions
[64, 57, 60, 65]. The rigorous numerical approach is the only way to account for these various effects
with sufficient accuracy.
In MINIMOS model of the VDMOS structure the basic semiconductor equations
1J

N O QIRYST
div M

div

Qed

QIRYS

VXW
V

grad

O QIRYS

Z[ eff
J
QIRBSBST
f

;
QIRYS

Z
`

1J
QIRYS

N X
_ QIRYScT
div M
Z

W?g

QIRYShkj

QIRYS

VP`
V
n

Z[ eff
R

l
Z

o
p q

_XQIRYS

2 ^ 88
2 ^ 89

are solved in the rectangular domain denoted by p in Figure Fig. 2.9 (dashed line). The solution of
the complete system of equations is obtained selfconsistently applying the successive Gummel-algorithm
for the steady-state problems [66]. In the transient simulations, the decoupling Mock-scheme [67] is
employed at the beginning of each time step, followed by the time-dependent Gummel-algorithm. These
solution approaches have shown absolute stability in all simulations of low-voltage VDMOS structures

2.9 POWER VDMOSFET SIMULATION

49

Figure 2.10: Doping profile of a low-voltage power VDMOSFET as interpolated onto MINIMOS grid.

carried out for different operating




we
regions,
including those of high terminal currents (large and
p ) and deep saturation biases ( p
100 ) [68]. Contrary to a general opinion that the decoupled
algorithms cannot be applied to simulate the power devices due to the strong coupling between the
equations (high current densities) see discussions in [65, 66] we have demonstrated the applicability
of the decoupled schemes on the low-voltage power VDMOSFETs in both, steady-state and transient
studies. A slow convergency in the very high-current conditions and divergency of the scheme for very
high drain biases (several hundreds volts) in the OFF-state conditions have still been observed.

MINIMOS is modified in several directions to account for VDMOS structures [68]:


A two-dimensional VDMOSFET doping profile is used directly as MINIMOS input profile; the key
FILE=2D must be presented in the PROFILE directive (Figure Fig. 2.10). In order to resolve the details
in the doping structure with minimal mesh-size without loosing the accuracy in the critical regions, a grid
which is very non-uniform in both x and -direction should be used for the input profile. The input profile
will be checked for the physical consistency.
The drain related quantities, i.e. the drain bias UD and STEP directive, the applied drain pulses (TRANS
directive) and the drain external resistor RD, are associated internally in the code with the bulk contact.
The source related bias US and STEP directive, and geometrical quantities SGAP, XLSOUR and XMIND
T
2. The
are placed on the drain side, which results in a symmetrical structure with respect to x
p

complete domain
is simulated, but not the half-domain denoted by
. Since the used x -mesh is not
T
symmetrical with respect to x
u 2, differences occur between the terminal currents computed for the
left and the right source contact as a consequence of the discretization error. These differences lay under
a few percent in common cases.
For the VDMOS structure a device-independent initial guess for the potential is applied, which is
Q
h
S
Q
ST

obtained by solving the equation div W


0 in the region p for the boundary
`7 grad Z y

50

CHAPTER 2 THE PHYSICAL MODELS OF MINIMOS

values at the contacts given by the external biases. The W and ` are the local neutral
carrier concentrations
T
Q
S
T
2
g ln W~Wy is the built-in
at equilibrium (W`0 W y ), which are determined by the net doping and y
potential. This equation follows from the steady-state carrier continuity equations assuming vanishing

W and `
`7 . For the initial
recombination rates and drift-diffusion transport model, after setting W
electron and hole concentration the neutral equilibrium values are assumed. We found this initial guess
very suitable for VDMOSFETs. Other an approach has been proposed in [65], which relies on solving
the Laplace equation for the initial potential.

Starting from the initial solution, MINIMOS solves the coupled Poisson and the carrier continuity equation

for the minority carriers in the complete domain p , while the concentration of the majority carriers is set
according to the local potential assuming a constant quasi-Fermi level. For VDMOSFETs this quasi-Fermi
level is governed by the bias at the source contact, but not by the bulk-contact bias as in simulation of bulk

MOSFETs. Note that the width of the analyzed domain p must be sufficiently large so that the source
o
o
contact which connects the N area also extends over the P region, as shown in Figure Fig. 2.9. After
a given accuracy is achieved, both carrier continuity equations are solved selfconsistently along with the
Poisson equation including generation-recombination processes. Since the majority carrier currents are
small in VDMOSFETs when the impact-ionization rate is low, the later solution step only requires a few
iteration cycles.
All physical models and features implemented in MINIMOS are normally applicable for VDMOSFET
simulation (e.g. AC analysis).
The MINIMOS approach provides a rigorous model for the terminal electrical characteristics of VDMOSFETs. It is convenient to study the resistivity in ON-state and its temperature dependence, the
quasi-saturation effect, the transfer characteristics including the short-channel effects, the switching characteristics and the dynamics effects of slow body-drain diode.
The VDMOSFET simulation feature is activated by key LVDM=YES on the OPTION directive.

2.10 The Monte Carlo Transport Module


In MINIMOS a single-particle Monte Carlo algorithm is included which allows a steady-state solution
of the transport problem of hot electrons to be computed. For each key on the MONTE-CARLO directive
a default setting exists such that for the user the invokation of the Monte Carlo module is considerable
eased.

2.10.1 Band Structure


The model for the conduction band of silicon accounts for six equivalent minima located close to the
X-points. Non-parabolicity is described by the band-form function
ec

 

2 90

2.10 THE MONTE CARLO TRANSPORT MODULE

51

  
   
    


ec

2  91
  
2  
For instance, if we consider a valley located on the  100 axis the effective masses are     
     
  

which has the non-parabolicity factor as the only parameter. The band-form function is related to the

wave vector through the effective masses


2

, with
and
as the longitudinal and transversal effective masses, respectively. In
MINIMOS, the Herring and Vogt transformation is introduced, which reduces the ellipsoidal equienergetic
surfaces to spheres:

2 92

!#" %$ & #" 



#-/.
2
0
1
2
0
1
2
0
1
$  " (')*,+ 0 10 00 3 4 $  " ()*' 10 0 -. 00 3 5 $  " (')* 10 01 00 - 3 

2  93
/.
+
0 0
0
0
1
0
0 1
+
$ &6"
Three transformation matrices
are required to account for the three possible orientations of the
ellipsoidal valleys. Consequently, the band-form function in the starred space becomes of spherical type,






2  94
2 
1

 

Note, that one is free in the choice of the reference mass occurring in the denominator of Equation 2.94.
such that in the transformation matrices (Equation 2.93) two
We have chosen this mass equal to
diagonal elements are always unity.
The parameters of the band-structure can be specified on the MONTE-CARLO directive by the following
keys:
Key
MLONG
MTRANS
NONPAR

 
 


Symbol

Description
Longitudinal effective electron mass
Transversal effective electron mass
Non-parabolicity factor

Default Setting
0 9163
0 1905
1
05




 798;:

2.10.2 Scattering Rates


In the following the scattering mechanisms included in MINIMOS are briefly described. Modeling pure
lattice scattering we closely follow the work of Jacoboni and Reggiani [39]. Additional scattering sources
we account for are ionized impurities, surface-roughness and impact ionization.

CHAPTER 2 THE PHYSICAL MODELS OF MINIMOS

52

Acoustic deformation potential scattering (ADP): For ADP scattering, which introduces intravalley
transitions only, the elastic approximation is assumed. The integrated scattering probability reads

 ACBEDGF
>= 2  ?I  @ J KML H
+

<

 N
2  95
 ? O  @ denotes the density-of-states effective mass, F H the acoustic deformation
where 
potential, J K the longitudinal sound velocity and L the semiconductor mass density. In the elastic
3 2

ec



2 1 3

approximation, ADP scattering is isotropic in the Herring-Vogt transformed space.


The parameters of ADP scattering can be specified on the MONTE-CARLO directive by the following
keys:
Key
DEFPOT
VSOUND
RHO
The lattice temperature

F H
JK
L

Symbol

BED

Description
Acoustic deformation potential
Longitudinal sound velocity
Semiconductor mass density

Default Setting
90
90 33 104
23 28 10 4

 7 8 Q 5RTS
 P V U R Q 
 P :

can be set by the TEMP key on the OPTIONS directive.

Intervalley phonon scattering: Electron transitions between states in two different equivalent valleys
can be induced by both acoustic and optical mode phonon scattering. In the following
denotes the
number of possible final equivalent valleys. For so-called g-type scattering this number equals to one

1 , while for f-type scattering four equivalent valleys are available


4 . The total scattering
rates read for phonon absorption

WX

WNX

<

ec

 ? @ F  Ym ZM[ ]
NW X 2 I  L  Z\[ ^] + X 1  X 1 2 X 
=
 ZM[ 

X
3 2

WX

2 96

2 97

and for phonon emission

 ? @ F _ Ym ZM[

NW X 2 I  L ` ZM[ ]
X 1  X 1 2 X a
1
2  98
+
=
Nb ZM[

X

2  99
F
`Z\[
In these equations, Y is the coupling constant, the phonon energy and X the electron energy after
\
Z
[
cZ\[
scattering. ] is the average phonon occupation number for the phonon ,
 Z [ b :
Z\[ ed
d



AMBED/f 1f 
exp
2  100
]
<

e]

3 2

Intervalley phonon scattering is isotropic in the Herring-Vogt transformed space.


In MINIMOS, six different intervalley phonons are accounted for. The coupling constants and the phonon
energies can be specified on the MONTE-CARLO directive by the following keys:

2.10 THE MONTE CARLO TRANSPORT MODULE


Key
DTK1
DTK2
DTK3
DTK4
DTK5
DTK6

F _Ym
F Y
_m
F Y
 m
F Y
_m
F Y
 m
F Y
_m

Symbol

1
2
3
4
5
6

798 RR QQ 
798 R Q 
798 R Q 
798 R Q 
798 R Q 
798

Default Setting
0 5 108
0 8 108
11 0 108
0 3 108
2 0 108
2 0 108

 P
 P
 P
 P
 P
 P

53

cZ
cZ
cZ
cZ
cZ
c Z

Key
EPH1
EPH2
EPH3
EPH4
EPH5
EPH6

Symbol
1
2
3
4
5
6

Default Setting
12 1 10 3
18 5 10 3
62 0 10 3
19 0 10 3
47 4 10 3
59 0 10 3

 P :
 P :
 P :
 P :
 P :
 P :

798
798
798
798
798
97 8

Mode
TA
LA
LO
TA
LA
TO

WNX

1
1
1
4
4
4

Ionized Impurity Scattering: Ionized Impurity Scattering is basically treated by the Brooks and Herring
approach. Additionally, the influence of scattering events having a large impact parameter is reduced
according to the reasoning of Ridley [69]. The starting point is the total scattering rate after Brooks and
Herring,

ih
jVk @ I e] [ l"7  ? m 1n 1  1 2 

2  101
4 1m 
+
2
=
1
n
m n   ?po 

2  102
q
2
[
m
Here, ] is the concentration of charged impurities, h
jVk is a dimensionless weighting factor and n is
defined through the screening length o q . The screening length equals to the Debye length weighted by a
dimensionless factor hsrVtvuwVw
x ,
VAMB
X
@


o q yhcrVtzu
w
wVx|~ {}} K
2  103


7
@
:

], @ X
2  104

B

from the electron concentration by solving
where
is the electron temperature, and is calculated
< Ag

ec

3 2

1 2

1 2

1 2

Equation 2.104.


1 b exp d`b < Ag e  fN 

2  105

yhVzrv 2I ] [ : @ 

2  106

In these equations, denotes the electron group velocity and the average distance between ions. The
latter can again be weighted by a dimensionless factor h
vrz .
Ionized impurity scattering behaves strongly anisotropic. In MINIMOS, the scattering angle is calculated
from a random number evenly distributed between 0 and 1 according to the following equations,


 ]b ln ^ 1 b exp d b < Ag e  fa 

2  107

Ag eb 
<

cos 1 b 2 < Ag
2  108
e
 4 1m n 

The weighting factors hj khcrVtzu


w
wVx and h
zrv can be specified on the MONTE-CARLO directive. Their
Following the reasoning of Ridley one obtains the following expression for the total scattering rate [70]

<

ec

1 3

default setting is unity.

CHAPTER 2 THE PHYSICAL MODELS OF MINIMOS

54

I ?P o  27
2
 K  ] [ ] ? 

Surface Roughness Scattering: Surface roughness scattering is currently modeled after [71]

<

2 109

] ?

where is the root-mean-square deviation of the surface from flatness and is the correlation length of
roughness. In MINIMOS the gradual channel approximation is employed to evaluate
in Equation
2.109, and additional fitting parameters are introduced such that we end up with the following expression,

I ?P o 7 d K_ f m d m f n


2  110
m

K

m max
b m P  0 

2  111
Equation 2.110 admits a general power law to be specified for the normal-field dependence. As a
consequence of the introduction of a general exponent one has to introduce some reference field strength
m . Note that for 2 the value assigned to m has no influence on the scattering rate. In Equation
2.111 is defined as unity vector normal to the semiconductor surface.
The lateral dimension of the region in which surface roughness scattering may occur is determined by the
m be positive. The vertical dimension of this region is a fixed distance
condition that the pressing force
<

2 2

ref

2
ref

ref

ref

which can be specified by the DEPSS key.


The parameters of surface roughness scattering can be specified on the MONTE-CARLO directive by the
following keys:
Key
DEPSS
DLSS
EXSS
EREF

Symbol

Description
Depth to which surface scattering may occur
Standard deviation times correlation length
Exponent of normal-field dependence
Reference normal-field for 2

P o
m

ref

P : QQ 
P :
 RQ
8

Default Setting
25 10 7
3 10 15 2
20
105

2.10.3 Impact Ionization


Impact ionization is treated as an additional scattering mechanism in the Monte Carlo simulation. The
related scattering rate, which is identical to the secondary pair genertion rate, is calculated by Keldyshs
empirical formula [72]

< [![ e]yh


j
z< e  d ab   fap 

2  112
 yh
w
m 

2  113

<z e  is the
Here, WBII is a constant prefactor,  is the threshold energy for impact ionization, and
m denotes the band-gap energy. The parameters of
phonon scattering rate at the threshold energy  .
impact ionization can be specified on the MONTE-CARLO directive by the following keys:
Key
EXII
WBII
WEII

Description
Exponent in Keldysh formula
Prefactor in Keldysh formula
Weight factor for threshold energy

Default Setting
20
3 9 10 2
10




 P :

2.10 THE MONTE CARLO TRANSPORT MODULE

55

2.10.4 Material Characteristics


In addition to full device simulation characteristics of bulk material can be computed by MINIMOS. Since
in this mode uniform conditions are assumed no device needs to be specified in the input deck. The
following example aims at the calculation of the electron ionization coefficient in a range of field strengts
of 180
to 650
. The result is written to the output file.

 8 RQ

*
OPTION
MONTE-CARLO
STEP
END

 8 RQ

MATERIAL CHARACTERISTICS SI (IMPACT IONIZATION)


MODEL=MC TEMP=300
MCMOD=2 NUMB=5E6 TRACE=NO WBII=3.9E-2 WEII=1
E1=180E3 E2=650E3 NE=20 ELOG=YES

10

Monte Carlo
Chynoweth
Ionization Coefficient (1/cm)

10

10

10

10

1.0

2.0

3.0

4.0

5.0

6.0

-6

1/E (10 cm/V)

Figure 2.11: Electron ionization coefficient from Monte Carlo simulation in comparison with Chynoweths
law.

 P Q  :  

P 8 RQ

Fig. 2.11 compares the results from a Monte Carlo simulation with the analytic Chynoweth model. The

1
7 03 105
1 231 106
[73].
parameters of the latter model are

2.10.5 The Monte-CarloDrift-Diffusion Coupling Technique


In MINIMOS a hybrid approach is implemented which couples the Monte Carlo and drift-diffusion
transport models. Both models can be simultaneously applied in different device regions. Drift zones and
other low-field regions eventually present in a device are treated by the cheaper drift-diffusion method,
whereas in high-field regions the physically more accurate Monte Carlo technique is used.
By means of coupling coefficients given below a Monte Carlo particle model can be coupled to a driftdiffusion based device model in a consistent way. The latter device model is represented by the following
equations:

div e grad u

7 b b ]

2 114

CHAPTER 2 THE PHYSICAL MODELS OF MINIMOS

56

 b^ 2
b^
1
7 div 

  7  7
 7 E  d`b grad 1 grad VAMB f
7B


M
A

7 s dcb grad b 1 grad


7 f
b 17

2 115

2 116

2 117

2 118

2 119

div






Three coefficients of this equation set, namely electron mobility, electron temperature and electron
ionization coefficient, are extracted from Monte Carlo simulation. Monte Carlo generated averages
are converted into these coefficients as follows:

 7  E R9 z Y

B V A 1  [ [
] <c[O [[
e
   

2 120

2 121

2 122

In Equation 2.122

is the number of considered space dimensions (in MINIMOS

2).

The remaining coefficients in Equation 2.114 to Equation 2.119 are modeled analytically (Section Section
2.1). In particular, holes are in any case described by the conventional drift-diffusion current relation
(
).

B BD

2.10.6 The Monte Carlo Windows


The basic idea of the hybrid approach is to apply the Monte Carlo technique only to those device regions
where the validity of the drift-diffusion model becomes questionable. In this subsection it is described
how in a MOSFET the critical region is identified and how the boundaries are treated.

According to Fig. 2.12 the whole simulation domain is labeled 0 . The subdomain 1 is called the
Monte Carlo window. Within this window particle trajectories are calculated and averages are recorded.

The subdomain 2 is called coefficient-window, because only inside this window the Monte Carlo and
drift-diffusion models are actually coupled by means of the coefficients given by Equation 2.120 to
Equation 2.122.
The boundaries of the simulation domain
keys on the DEVICE directive.

can be specified by the XLEFT, XRIGHT ans YDEPTH

To impose safe boundary conditions the Monte Carlo window 1 has to be chosen such as to include
parts of the heavily doped source and drain areas. Electrons are randomly injected at the left boundary of

2.10 THE MONTE CARLO TRANSPORT MODULE

57

G
S

D2
D1

D0

Figure 2.12: Placement of Monte Carlo window


simulation domain 0 for a MOSFET.

and coefficient window

within the overall

1 according to an equilibrium distribution function. Electrons arriving at the right boundary are either
reflected or absorbed, depending on the BDR (back diffusion ratio) key on the MONTE-CARLO directive.
The distance between the left boundary of 1 and the vertical source junction can be specified by the
SOVL (source overlap) key. Analogously, the distance between the vertical drain junction and the right
boundary of 1 can be specified by the DOVL (drain overlap) key. Note that an increased overlap of the
Monte Carlo window with the source and drain areas increases physical accuracy but decreases efficiency,
and vice versa.

The coefficient window 2 is determined by certain criteria. Since for both mobility and temperature
different criteria are used, different windows are obtained. In what follows we shall call them mobilty
and temperature windows, respectively. A mesh point lies inside the mobility window if the average
momentum loss rate exceeds a given minimum value.

d p f

Vx

^
7
P
2  123
coll
The default setting of the key DFMIN is 2000 8 R Q  . A point lies inside the temperature window if the
carrier temperature exceeds the lattice temperature by a certain amount.

B * B D P Vx

2 124

UTMIN is a factor close to unity. Default setting of this key is 1 05. Additionally, only those points
identified as carrying statistically reliable averages can contribute to a coefficient window. If the number
of scattering events assigned to a certain mesh point is less than CNTMIN times the number of total
scattering events (given by NUM or NUM1,NUM2 ) then this point is excluded from any coefficient
window.

pp

Note that in any case the Monte Carlo window should be larger than the coefficient windows. In particular
the right boundary should lie deep enough in the drain such that the Monte Carlo window covers the whole
area in which according to Equation 2.124 elevated carrier temperatures exist.

CHAPTER 2 THE PHYSICAL MODELS OF MINIMOS

58

2.10.7 The Self-Consistent Iteration Scheme


For very short devices, an increasing fraction of the electron population is in non-equilibrium conditions.
As Monte Carlo treats the average motion of those electrons in a substantially different way than the
standard drift-diffusion model, the resultant spatial distribution of mobile charge will also differ. Therefore,
realistic results can only be expected by applying some sort of self-consistent iteration scheme. In
MINIMOS, this scheme is based on the Monte-Carlodrift-diffusion coupling technique.
In each cycle of the self-consistent iteration loop, a Monte Carlo simulation is performed with a potential
taken from the previous cycle. The spatial distributions of electron mobility, electron temperature and
electron ionization coefficient (Equation 2.120 to Equation 2.122) serve as result. Inside the coefficient
windows, identified by Equation 2.123 and Equation 2.124, the Monte Carlo generated coefficients are
used, whereas outside the windows the coefficients are calculated from analytical models. With coefficientprofiles assembled in such a way the coupled equation set consisting of Equation 2.114 through Equation
2.119 is solved. With the updated potential the iteration cycle is repeated until a given number of
iterations is done. The initial potential distribution is generated by a standard drift-diffusion simulation
(MODEL=AVAL of MINIMOS).
This novel iteration scheme converges very rapidly. Typical iteration numbers are of the order of 5. The
number of self-consistent iterations can be specified by the MCPIT key. The evolutions of drain current,
norms of potential update, electron concentration update and temperature update with the number of
iterations are printed to the reference file.

pp

The iterative procedure can by tuned by the keys NUM1 NUM5. Note that the very first Monte Carlo
iteration uses an initial potential profile which results from a standard drift-diffusion simulation. As the
initial potential may deviate from the self-consistent one the Monte Carlo results of this iteration may
deviate from the final results as well. Therefore, the results of the very first Monte Carlo iterations need
not be calculated as accurately as the results of the final iterations. Assume one specifies MCPIT 4 (four
self-consistent iterations) and NUMB 40 106. The total number of scattering events to be processed for
this bias point would be 160 106 . A similar accuracy can be achieved by processing 100 106 scattering
events if one assigns the numbers 15 106 15 106 30 106 40 106 to the keys NUM1 through NUM5,
respectively.

P  P  P  P

2.10.8 Calculation of Averages


During a Monte Carlo simulation statistical data are recorded on an auxiliary grid. This grid is independent
of the MINIMOS grid used to discretize the semiconductor equations. The single-particle Monte Carlo
algorithm used by MINIMOS allows the so-called before-scattering technique to be employed [39]. In

are the
the following denotes a generic physical quantity, the total scattering rate, and
electrons position and momentum immediately before the -th scattering event. If is the volume of the
-th cell of the auxiliary grid the average within this cell is defined by

< 

 [ < : [  [


<  [
l v : [

 [   [

2 125

2.10 THE MONTE CARLO TRANSPORT MODULE

59

Once the averages have been evaluated on the auxiliary grid they are assigned to the MINIMOS discretization grid by means of a convolution method. The assignment function used for this step is

  Xc No T1oN P 1
b oN oN 
2
2

I
2I
P

1
cos
o
o ; 
b oN o 
2
2

cos

2 126


o  of
Outside the interval given by Equation 2.127 the assignment function vanishes. The lateral width 
the assignment function can be specified by the XWIDTH key, the vertical width oN by the YWIDTH key.

2 127

In order not to introduce an additional error due to charge assignment the widths of the assignment
function needs to be chosen carefully. If a Monte Carlo generated profile is very smooth the filter widths
are probably too large. If there is much noise on such a profile the filter widths are too small with respect
to the number of processed scattering events. In this case one can either increase the filter widths, which
does not increase accuracy, or one can increase the number of scattering events.
The number of points of the auxiliary grid is predefined. The respective parameters are MAXNFX and
MAXNFY in the file mccom.inc. The left and right boundaries of the auxiliary grid can be specified by
the AUXS and AUXD keys, respectively. The auxiliary grid can be smaller than the Monte Carlo window.
However, one must be aware that no averages are recorded outside this grid. Therefore, the auxiliary grid
has to be larger than any of the coefficient windows. Otherwise, the latter are clipped by the boundaries
of the grid.

2.10.9 The Particle Weighting Algorithm


To deal properly with scarcely populated regions in both geometric and energy space a trajectory multiplication scheme is applied. The Monte Carlo window can be divided in up to four different regions, in
each of which a unique weight is assigned to the carriers. As an example Fig. 2.13 shows a device with
G
S

R1

R2

R1

Figure 2.13: Decomposition of the Monte Carlo window for the purpose of trajectory multiplication

the Monte Carlo window divided into three regions. Following the notation of [74] denotes the region
of common events, in 1 and 2 rare events occur. Note that regions and 1 are not contiguous.

If a particle moves from to 1 or from


the other hand, if carriers move from 2 to

to 2 then it is split into a certain number of light ones. On


then they are gathered to heavy ones. The
1 or from 1 to

CHAPTER 2 THE PHYSICAL MODELS OF MINIMOS

60

algorithm in MINIMOS ensures that in a certain region all particles have the same weight, regardless of
their history.
The user can specify the number of split boundaries by the NSB key. For example, in Fig. 2.13 the Monte
Carlo window is divided into three regions, hence the number of split boundaries is NSB=2. NSB can be
interpreted as that number of boundaries a particle crosses on its way from the outermost to the innermost
region. Setting NSB=0 switches off particle splitting in a simulation.
Once the number of split boundaries is given MINIMOS tries to find optimized locations for these
boundaries both on the source and drains sides of a device. The splitting or multiplication factors are
chosen by MINIMOS automatically, as well.
In energy domain a carrier trajectory is multiplied if the energy threshold for impact ionization is exceeded.
The multiplication factor in that case can be specified by the EMULT key. Setting EMULT=1 disables
particle splitting in the energy domain.
Generally speaking, Monte Carlo simulation is feasible only for devices in the on-state. The larger the
currents the better the Monte Carlo technique will perform. The closer a bias point lies to turn-off, e.g.
the closer gate voltage lies to threshold voltage, the more important is the role of the particle weighting
algorithm. This algorithm can be thought to increase the injection probability of electrons over the built-in
barrier present between source and channel. The algorithm can furthermore be thought to reduce the
fraction of time spent for the simulation of cold electrons in the heavily doped source and drain regions,
and to increase the fraction of simulation time spent for the hot electrons in the pinch-off region.

2.10.10 The Self-Scattering Algorithm


The analytic band model of MINIMOS allows the equations of motion to be integrated analytically. The
time of a carriers free flight is determined by an optimized self-scattering algorithm.

An upper bounding function to the physical scattering rate is constructed by a piece-wise linear function
on the 2 axis. In MINIMOS, the bounding function comprises three linear segments of the form

  

2 128

Due to impurity and surface roughness scattering the parameters and are chosen spatially dependent.
The choice of these parameters can be controlled by the E1SELF and E2SELF keys. At each mesh point
the parameters and for the first linear segment are determined from the physical scattering rate in the
energy interval 0
E1SELF. To set up the parameters for the third linear segment the energy interval
E1SELF E2SELF is considered.

3
Reference of Input Directives

The input parameters for MINIMOS are specified by a sequence of lines in a formatted data file. The first
72 characters are used to parse information and 80 characters are read and written to the output file to
offer the user the possibility of sequencing his/her statements. The first line in the data file is interpreted
as a title statement. All text on this line is used to identify the printout pages. This title line must not be
omitted. All the other statements conform to a simple syntax, namely:
DIRECTIVE1
DIRECTIVE2
..........
DIRECTIVEj

KEY1=VALUE1, KEY2=VALUE2 ...


KEY1=VALUE1, KEY2=VALUE2 ...
...........................
KEY1=VALUE1, KEY2=VALUE2 ...

DIRECTIVE may be any character string of the following set:


BIAS, DEVICE, END, FIMPLANT, GEOMETRY,GRID,
IMPLANT, INTERFACE, MOBILITY, MONTE-CARLO, OPTION,
OUTPUT, PROFILE, RECOMBINATION, STEP,TRANSIENT

Abbreviations are allowed. Valid delimiters for the keys within the directive are blanks, commas, or
asterisks. Any directive can be continued in the next line. In this case the continuation line has to start
with a plus in the first column. Comment information can be placed anywhere in the input stream.
Start of comment is indicated by an asterisk. Note that comments cannot be continued by a + sign on the
next line. Thus, a line starting with an asterisk is totally interpreted as a comment line and an asterisk in
any column of a directive indicates that the remainder of this line is comment. Comment is allowed within
continuation lines. All blank lines are ignored but not allowed within a continued directive. Blanks in
front of a directive are ignored.
The BIAS, DEVICE and PROFILE directive with appropriate keys are required in any MINIMOS input
file. All other directives are optional. The directives may appear in arbitrary order in the input file, except
the END directive which has to be the last statement to indicate the end of the input stream. One excetpion
is when material characteristics are calculated by the Monte-Carlo model (MONTE-CARLO MCMOD=2).
61

62

CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

Then no device,no bias and no profile have to be specified since uniform conditions are assumed. In that
case the OPTION and the STEP directive are sufficient for a valid input file.
The following sections are a comprehensive description of all possible directives and keys that may be
specified on the MINIMOS input deck.
The significant characters of the individual keys are printed enlarged. Variable types, default values,
ranges, units etc. are given in tables.
The variable types are
<C> Single character
<I> Integer Number
Note that MINIMOS reads all numbers integer or float as real numbers. Thus, an <I>
specification in the keydescription table is just a meaningful recommendation.
<L> Logical
Valid specifications are the character strings YES and NO only, of which only the first character is
significant.
<R> Real Number
<S> Character string

Directive BIAS

b r

rGb j

The BIAS directive is used to specify the operating point for which the simulation should be performed.
and
must not be
The input data will be checked for physical plausibility. Therefore
negative for nchannel devices and not positive for pchannel devices. This plausibility check can be
suppressed by setting PHYSCK=NO on the OPTION directive.

JCAP Node Index for Small Signal Bias


This keyspecifies the type of the electrode for which the admittances in case of AC analysis should be
computed. If A is specified, denoting all terminals, the complete 4 4 admittance matrix is calculated,
else only the specific row of the adnittance matrix
Type:
<I>
Default:
1
Units:
Range:
0 bias successively at all terminals
1 bias at gate
2 bias at source
3 bias at drain
4 bias at bulk

BIAS

63

UB Applied Bulk Voltage


Type:
Default:
Units:
Range:

<R>
US

8b

20

UB

UD Applied Drain Voltage


Type:
Default:
Units:
Range:

20

<R>
(obligatory key)

8b

20

UD

20

UG Applied Gate Voltage


This key must be omitted when MODEL=THRES is specified on the OPTION directive.
Type:
<R>
Default:
(obligatory key)
Units:
Range:
20 UG 20

8b

US Applied Source Voltage


Type:
Default:
Units:
Range:

<R>
0

8b

20

US

20

The following two keys refer to AC-analysis calculations. They specify the real and imaginary parts of
the voltage applied to the terminal in question. The magnitude should be unimportant due to the linear
dependence, but the angle between VR and VI may influence the iteration history. If both keys are omitted
no AC-analysis is carried out. At least one of the keys VR and VI must be specified, if AC-analysis is to
be carried out.

VI Imaginary Part of Small Signal Bias


Type:
Default:
Units:
Range:

<R>
0

VI

10

VR Real Part of Small Signal Bias


Type:
Default:
Units:
Range:

<R>
0

VR

10

CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

64

Directive DEVICE
The DEVICE directive is used to specify the principal physical device parameters.

BULK Thickness of Insulating Substrate


This key specifies the thickness of the insulating substrate for simulation of SOI transistors. If this key is
given, the FILM and the KBULK keys have also to be given.
Type:
<R>
Default:

Units:
Range:
10 5 BULK 2 10 3

Q

P :

CHANNEL Channel Type


This key specifies the channel type of the device.
Type:
<C>
Default:
(obligatory key)
Units:
Range:
N for a nchannel device
P for a pchannel device
DGAP Drain Gap Length
This key specifies the gap between the gate edge and the drain contact.
Type:
<R>
Default:
SGAP
Units:
Range:
10 6 DGAP 10 2

Q

E1 Thickness of First Epilayer


Type:
Default:
Units:
Range:

<R>
0

Q

10

:
6

10

10

E1

E2 Thickness of Second Epilayer


Type:
Default:
Units:
Range:

<R>
0

Q

10

:
6

E2

FILM Thickness of Semiconductor Film


This key specifies the thickness of the semiconductor film for the simulation of SOI transistors. If this
key is given, the BULK and the KBULK keys have to be given too.
Type:
<R>
Default:

Units:
Range:
2 10 6 FILM 5 10 4

Q

P :

P :

DEVICE

65

GATE Work Function of Gate Material


This key specifies the work function difference for the gate with respect to the intrinsic level in the substrate
in case of MOSFET simulation. Either a character string or a real number can be specified. In case of a
character string, only the first character of which is significant, AL indicates aluminium, NPOLY ndoped
polysilicon, PPOLY pdoped polysilicon as gate material, which results in a work function of 0 59 ,
0 55 or 0 55 respectively. For the poly-gate depletion analysis GATE determines only the gate type
(NPOLY or PPOLY). If GATE equals a real number an explicit work function is specified. If TINS on the
DEVICE directive equals zero, a Schottky gate contact is assumed and GATE now specifies the height of
the Schottky barrier. The GATE key must not be omitted, unless the UFB key on the OPTION directive is
given. In this case the GATE key has to be omitted.
Type:
<R> <S>
Default:
0 55
Units:
Range:
AL
NPOLY
PPOLY
1 5 GATE 1 5

b 8

b 8

b
7 8

INS Relative Permittivity of Gate Insulator


This key specifies the relative permittivity of the gate insulator. Either a character string or a real number
can be specified. In case of a character string, only the first character of which is significant, NITRIDE
( 3 4 ) or OXIDE (
2 ) indicate a relative permittivity of 7 2 or 3 9 respectively. In case of a real
number the relative permittivity is specified explicitly.
Type:
<R> <S>
Default:
39(
2)
Range:
1 INS 30
NITRIDE
OXIDE

KBULK Relative Permittivity of Insulating Substrate


This key specifies the relative permittivity of the insulating substrate for the simulation of SOI transistors.
If this key is given, the BULK and the FILM keys have also to be given.
Type:
<R>
Default:

Range:
1 KBULK 30

L Gate Length

This key specifies the gate electrode length.


Type:
<R>
Default:
(obligatory key)
Units:
Range:
10 5 L 10 2

Q

CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

66

MAT Substrate Material


Only the first character of the specified string is significant.
Type:
<S>
Default:
SI Silicon
Range:
SI Silicon
GAAS galliumarsenide

SGAP Source Gap Length


This key specifies the gap between the gate edge and the source contact.
Type:
<R>
Default:
10 TINS
Units:
Range:
10 6 SGAP 10 2

Q P
:

TINS Thickness of the Gate Insulator


If TINS equals zero a Schottky gate contact is assumed and the program switches to MESFET simulation
mode. Note that this key acts as the main switch between MOSFET and MESFET simulation.
Type:
<R>
Default:
(obligatory key)
Units:
Range:
2 10 7 TINS 10 3 for MOSFET
0 for MESFET

Q

P :

W Channel Width
Type:
Default:
Units:
Range:

<R>
(obligatory key)

Q

P :

5 10

XLEFT Minimum x-Coordinate of Simulation Domain


With this key one can define the minimal extent of the simulation domain from the source-sided gate
contact edge in the source contact direction.
Type:
<R>
Default:
0
Units:
Range:
10 2 XL 0

Q
b :

XRIGHT Maximum x-Coordinate of Simulation Domain


With this key one can define the minimal extent of the simulation domain from the drain-sided gate contact
edge in the drain contact direction.
Type:
<R>
Default:
0
Units:
Range:
0 XR 10 2

Q

DEVICE

67

YDEPTH Maximum y-Coordinate of Simulation Domain


With this key one can define the minimal extent of the simulation domain from the semiconductor surface
into the bulk.
Type:
<R>
Default:

Units:
Range:
10 5 Y 10 2

Q

CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

68

Directive END
The END directive is used to signal the end of the input file. This directive may be omitted. If present,
however, it must be the last directive.

BIN Save Results on Binary File


This key specifies if the simulation results are to be saved on an external binary file. This is true for both
MINIMOS 2D and MINIMOS 3D. The format of this file is explained in detail in Chapter 4.
Type:
<L>
Default:
NO

CAPERR ACAnalysis Error Limit


This key specifies the error norm threshold of the ACanalysis. See Section 1.11 for information about
error norms in MINIMOS. The ACanalysis iteration is terminated if the arithmetic mean value of the
error norms under consideration is smaller than the limit set by CAPERR. Error norms are computed from
the increments of both real and imaginary parts of the Poisson equation, the minority carrier continuity
equation and additionally for the majority carrier equation if MODEL=AVAL or MODEL=HOT is specified
on the OPTION directive.
Type:
<R>
Default:
10 4
Units:

Range:
10 6 CAPERR 10 2

CACCREL Error Norm Limit for Iterative Continuity Equation Solvers


Type:
Default:
Range:

<R>
10 8
10 10

CACCREL

ERROR General Error Norm Limit

10

This key specifies the error norm threshold of the simulation. See Section 1.11 for information about the
error norms in MINIMOS. To terminate a Gummel iteration loop the error norms under consideration have
to be smaller than ERROR. Error norms are computed from the increments of the Poisson equation, the
minority carrier continuity equation and additionally for the majority carrier equation if MODEL=AVAL
or MODEL=HOT is specified on the OPTION directive. Note that all above mentioned error norms are
dimensionless real numbers and are an integral error measure. Beside those norms the F-norm is used,
an error measure, which is computed as the local maximum of the relative space charge increment. If
MODEL=THRES is specified ERROR additionally specifies:

_ : 
2. The limit for the threshold voltage increment in 8 . The threshold iteration is terminated if any of

1. The limit for the relative threshold current error


the two limits is reached.

Type:
Default:
Units:
Range:

<R>
10 2

10 5

ERROR

0 5

END

69

NAC Maximum Iteration Count for AC Analysis


This key specifies the maximum number of iterations for any AC condition.
Type:
<I>
Default:
300
Range:
1 NAC 500

NCYCLES Maximum Iteration Count


This key specifies the maximum number of iteration cycles for each model.
Type:
<I>
Default:
300
Range:
1 NCYCLES 1000

NGEN Maximum Generation Subcycles


This key specifies the maximum number of generation subcycles. It applies for MINIMOS 2D only.
Type:
<I>
Default:
1
Range:
1 NGEN 20

NHOT Maximum Number of Hot Loops


This key applies for MINIMOS 2D only.
Type:
<I>
Default:
20
Range:
1 NHOT 500

NUPDATES Maximum Number of Grid Updates


Type:
Default:
Range:

<I>
20
1 NUPDATES

100

PACCREL Error Norm Limit for Iterative Poisson Equation Solvers


Type:
Default:
Range:

<R>
10 3
10 10

PACCREL

10

TCERR Carrier Temperature Error Norm Limit


This key specifies the limit for the error norms for the carrier temperatures of both the majorities and the
minorities. TCERR may only be given if MODEL=HOT is specified on the OPTION directive.
Type:
<R>
Default:
10 2
Range:
10 5 TCERR 0 5

3CYCLES Maximum Number of 3D Iteration Cycles


This key specifies the maximum number of iterations for each model in a three-dimensional simulation.
Type:
<I>
Default:
500
Range:
1 3C 1000

CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

70

3GEN Maximum Number of 3D Generation Cycles


This key specifies the maximum number of generation subcycles in a three-dimensional simulation.
Type:
<I>
Default:
1
Range:
1 3G 20

3HOT Maximum Number of 3D Hot Loops


Type:
Default:
Range:

<I>
20
1 3H

500

100

3UPDATES Maximum Number of 3D Grid Updates


Type:
Default:
Range:

<I>
20
1 3U

FIMPLANT

71

Directive FIMPLANT
The FIMPLANT directive is used to specify a field implantation at the channel edge in the channel-width
direction. The edge is assumed at the zcoordinate zero. The available keys are handled analogously to
the keys within the IMPLANT directive.
Note: This directive is meaningful only for threedimensional simulation.

AKEV Field Implantation Energy


The AKEV key must not be omitted, if any of the FIMPLANT keys is specified.
Type:
<R>
Default:

Unit:
Range:
10 AKEV 300

 78

DOSE Field Implantation Dose


The DOSE key must not be omitted, if any of the FIMPLANT keys is specified.
Type:
<R>
Default:

2
Unit:
9
Range:
10
DOSE 1018

Q :

ELEM Field Implantation Element


Valid elements are antimony (SB), arsenic (AS), boron (B) and phosphorus (P) for a silicon substrate.
For galliumarsenide substrates you may specify antimony (SB), arsenic (AS), beryllium (BE), boron
(B), oxygen (O), magnesium (MG), phosphorus (P), and silicon (SI). The first character of the element
specifications is significant, except antimony and beryllium, for which the first two characters have to be
specified. The ELEM key must not be omitted, if any of the FIMPLANT keys is specified.
Type:
<S>
Default:

Range:
AS Arsenic
B Boron
BE Beryllium
MG Magnesium
O Oxygen
P Phosphorus
SB Antimony
SI Silicon

TEMP Field Implantation Anneal Temperature


The TEMP key must not be omitted, if any of the FIMPLANT keys is specified. In case of a double
implanted channel, however, it may only be specified once.
Type:
<R>
Default:

Unit:
Range:
800 TEMP 1300

CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

72

TIME Field Implantation Anneal Time


The TIME key must not be omitted, if any of the FIMPLANT keys is specified. In case of a double
implanted channel, however, it may only be specified once.
Type:
<R>
Default:

Unit:
Range:
60 TIME 48 3600

GEOMETRY

73

Directive GEOMETRY
The GEOMETRY directive is used to specify various geometric parameters. This directive can be omitted.

BEAKL Lenght of Birds Peak


This key specifies the length of the birds beak, the length of the transition from the gate-oxide to the
field-oxide.
Type:
<R>
Default:

Units:
Range:
0 BEAKL 10 2

Q

DCGRD Slope of Drain Contact Edge


Type:
Default:
Units:
Range:

<R>

Q

DCGRD

10

DDREOX Depth of Drain Contact Reoxidation


Type:
Default:
Units:
Range:

<R>

Q

DDREOX

10

DRGRD Slope of Drain-Sided Reoxidation


Type:
Default:
Units:
Range:

<R>

Q

DRGRD

10

D1GAP Offset of Drain-Sided Recess Edge


This key specifies the offset of the drain-sided recess edge relative to the drain-sided gate contact edge.
Type:
<R>
Default:

Units:
Range:
0 D1GAP 10 2

Q

D2GAP Offset of Drain-Sided Oxidation


This key specifies the offset of the drain-sided reoxidation edge relative to the drain-sided gate contact
edge.
Type:
<R>
Default:

Units:
Range:
0 D2GAP 10 2

Q

CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

74

GCTHE Gate Contact Thickness

This key specifies the thickness of the gate contact. The default value for MOSFETs is TINS/4, whereas
for MESFETs 0 1 .
Type:
<R>
Default:
Built-in Model
Units:
Range:
0 GC 10 3

G1CGRD Slope of Source-Sided Drain Contact Edge


Type:
Default:
Units:
Range:

<R>

G1CGRD

10

G2CGRD Slope of Drain-Sided Drain Contact Edge


Type:
Default:
Units:
Range:

<R>

G2CGRD

10

LD Drain Contact Length for Capacity Calculations


Type:
Default:
Units:
Range:

<R>

10

LD

10

LS Source Contact Length for Capacity Calculations


Type:
Default:
Units:
Range:

<R>

10

LS

10

OXIDEF Field Oxide Thickness


Type:
Default:
Units:
Range:

<R>

OXIDEF

2 10

RECESS Gate Recess Depth


Type:
Default:
Units:
Range:

<R>

RECESS

10

R1GRD Slope of Source-Sided Recess Edge


Type:
Default:
Units:
Range:

<R>

R1GRD

10

GEOMETRY

75

R2GRD Slope of Drain-Sided Recess Edge


Type:
Default:
Units:
Range:

<R>

R2GRD

10

SCGRD Slope of Source Contact Edge


Type:
Default:
Units:
Range:

<R>

SCGRD

10

SDREOX Depth of Source Contact Reoxidation


Type:
Default:
Units:
Range:

<R>

SDREOX

10

SRGRD Slope of Source-Sided Reoxidation


Type:
Default:
Units:
Range:

<R>

SRGRD

10

S1GAP Offset of Source-Sided Recess Edge


This key specifies the offset of the source-sided recess edge relative to the source-sided gate contact edge.
Type:
<R>
Default:

Units:
Range:
0 S1GAP 10 2

S2GAP Offset of Source-Sided Reoxidation


This key specifies the offset of the source-sided reoxidation edge relative to the source-sided gate contact
edge.
Type:
<R>
Default:

Units:
Range:
0 S2GAP 10 2

TD Length of Drain-Sided T-Bar


Type:
Default:
Units:
Range:

<R>

TD

10

TGATE Thickness of T-Bar


Type:
Default:
Units:
Range:

<R>

TGATE

10

CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

76

TS Length of Source-Sided T-Bar


Type:
Default:
Units:
Range:

<R>

TS

10

T1GRD Slope of Source-Sided T-Bar Edge of the Thickness of T-Bar


Type:
Default:
Units:
Range:

<R>

T1GRD

10

T2GRD Slope of Drain-Sided T-Bar Edge of the Thickness of T-Bar


Type:
Default:
Units:
Range:

<R>

T2GRD

10

YMDEP Depth of Semiconductor


Type:
Default:
Units:
Range:

<R>

10

YMDEP

10

GRID

77

Directive GRID
The GRID directive is used to modify parameters for the initial grid and for the grid refinement. This
directive may be omitted.

DGMI Minimum Grid Spacing


This key specifies the minimum grid spacing for the weighting function from the doping profile.
Type:
<R>
Default:
20 10 7
Units:
Range:
10 8 DGMI 10 5

DJRX Maximum Relative xCurrent Difference


The maximum local contribution to the drain current for xdirection
Type:
<R>
Default:
5
Units:
%
Range:
1 DJRX 20

DJRY Maximum Relative yCurrent Difference


The maximum local contribution to the drain current for ydirection
Type:
<R>
Default:
5
Units:
%
Range:
1 DJRY 20

DOPX Doping Profile xWeighting Factor


This key specifies the weighting factor for the weighting function from doping profile in xdirection.
Type:
<R>
Default:
0 35
Range:
0 1 DOPX 1

DOPY Doping Profile yWeighting Factor


This key specifies the weighting factor for the weighting function from doping profile in ydirection.
Type:
<R>
Default:
0 35
Range:
0 1 DOPY 1

DOPZ Doping Profile zWeighting Factor


This key specifies the weighting factor for the weighting function from doping profile in zdirection.
Type:
<R>
Default:
0 35
Range:
0 1 DOPZ 1

CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

78

DPXC Maximum xPotential Difference (Channel)


This key specifies the maximum potential difference between adjacent gridlines in xdirection in the
channel region.
Type:
<R>
Default:
07
Units:
Range:
0 25 DPXC 5

DPXD Maximum xPotential Difference


This key specifies the maximum potential difference between adjacent gridlines in xdirection.
Type:
<R>
Default:
1 25
Units:
Range:
0 5 DPXD 10

DPXP Maximum xPotential Difference in the Polysilicon Gate


This key specifies the maximum potential difference between adjacent gridlines in xdirection in the
poly-gate area. This key is active for the gate depletion analysis.
Type:
<R>
Default:
03
Units:
Range:
0 1 DPXP 1

DPYC Maximum yPotential Difference (Channel)


This key specifies the maximum potential difference between adjacent gridlines in ydirection in the
channel region.
Type:
<R>
Default:
07
Units:
Range:
0 25 DPYC 5

DPYD Maximum yPotential Difference


This key specifies the maximum potential difference between adjacent gridlines in ydirection.
Type:
<R>
Default:
1 25
Units:
Range:
0 5 DPYD 10

DPYP Maximum yPotential Difference in the Polysilicon Gate


This key specifies the maximum potential difference between adjacent gridlines in ydirection for the
poly-gate area. This key is active for the gate depletion analysis.
Type:
<R>
Default:
0 15
Units:
Range:
0 050 DPYP 0 5

GRID

79

DPZ Maximum zPotential Difference


This key specifies the maximum potential difference in between adjacent gridlines in zdirection for the
weighting function.
Type:
<R>
Default:
05
Units:
Range:
0 1 DPZ 5

DP1D Maximum Potential Difference for MODEL=1D


This key specifies the maximum potential difference between adjacent gridlines for the weighting function
in xdirection after the initial guess at the beginning of the model 1-D.
Type:
<R>
Default:
25
Units:
Range:
0 5 DP1D 10

DX1 Grid Spacing (First Set) in xDirection


This key specifies the distance between two gridlines of the first set of userdefined supplementary
gridlines in xdirection.
Type:
<R>
Default:
10 4
Units:
Range:
10 8 DX1 10 4

DX2 Grid Spacing (Second Set) in xDirection


This key specifies the distance between two gridlines of the second set of userdefined supplementary
gridlines in xdirection.
Type:
<R>
Default:
10 4
Units:
Range:
10 8 DX2 10 4

DX3 Grid Spacing (Third Set) in xDirection


This key specifies the distance between two gridlines of the third set of userdefined supplementary
gridlines in xdirection.
Type:
<R>
Default:
10 4
Units:
Range:
10 8 DX3 10 4

DY Grid Spacing in yDirection


This key specifies the distance between two gridlines of userdefined supplementary gridlines in y
direction.
Type:
<R>
Default:
10 4
Units:
Range:
0 25 10 8 DY 10 4

CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

80

DZ Grid Spacing in zDirection


This key specifies the distance between two gridlines of userdefined supplementary gridlines in z
direction.
Type:
<R>
Default:
10 4
Units:
Range:
0 25 10 8 DZ 10 4

GFILL Number of Grid Sections


This key specifies the number of grid-sections under the gate electrode.
Type:
<I>
Default:
6
Range:
2 GFILL 20

GGMI Minimum Grid Spacing for Initial Grid


Type:
Default:
Units:
Range:

<R>
10 7

10

GGMI

10

MXTun x-Grid Refinement for Band-to-Band Tunneling


The analysis of band-to-band tunneling in the bulk is performed on x-grid which is MXT times finer
than grid used to solve the basic semiconductor equations. Each x-grid interval is subdivided into MXT
subintervals. Note that increasing MXT can prolong the computation time.
Type:
<I>
Default:
2
Range:
1 MXT 4

MYTun y-Grid Refinement for Band-to-Band Tunneling


The analysis of band-to-band tunneling in the bulk is performed on y-grid which is MYT times finer
than grid used to solve the basic semiconductor equations. Each y-grid interval is subdivided into MYT
subintervals. Note that increasing MYT can prolong the computation time.
Type:
<I>
Default:
2
Range:
1 MYT 4

PGMI Minimum Grid Spacing for Grid Updates


Type:
Default:
Units:
Range:

<R>
0 25 10

2 10
9

PGMI

10

SXTun x-Direction Standard Deviation in Grid Adaption for Tunneling


The standard deviation for the x-direction used to filter the band-to-band tunneling generation rates is
multiplied by SXT (cf. Section 2.6).
Type:
<R>
Default:
1
Range:
0 5 SXT 5

GRID

81

SYTun y-Direction Standard Deviation in Grid Adaption for Tunneling


The standard deviation
multiplied by SYT.
Type:
Default:
Range:

for the y-direction used to filter the band-to-band tunneling generation rates is
<R>
1
0 5 SYT

The following keys can be used for defining supplementary gridlines in certain areas of the device.

XE1 First End Value in xDirection


This key specifies the end value in xdirection of the first set of userdefined supplementary gridlines.
Type:
<R>
Default:

Units:
Range:
10 10 4 XE1 10 10 4

XE2 Second End Value in xDirection


This key specifies the end value in xdirection of the second set of userdefined supplementary gridlines.
Type:
<R>
Default:

Units:
Range:
10 10 4 XE2 10 10 4

XE3 Third End Value in xDirection


This key specifies the end value in xdirection of the third set of userdefined supplementary gridlines.
Type:
<R>
Default:

Units:
Range:
10 10 4 XE3 10 10 4

XS1 First Start Value in xDirection


This key specifies the start value in xdirection of the first set of userdefined supplementary gridlines.
Type:
<R>
Default:

Units:
Range:
10 10 4 XS1 10 10 4

XS2 Second Start Value in xDirection


This key specifies the start value in xdirection of the second set of userdefined supplementary gridlines.
Type:
<R>
Default:

Units:
Range:
10 10 4 XS2 10 10 4

CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

82

XS3 Third Start Value in xDirection


This key specifies the start value in xdirection of the third set of userdefined supplementary gridlines.
Type:
<R>
Default:

Units:
Range:
10 10 4 XS3 10 10 4

YE End Value in yDirection


This key specifies the end value in ydirection of the set of userdefined supplementary gridlines.
Type:
<R>
Default:

Units:
Range:
5 10 4 YE 10 3

YS Start Value in yDirection


This key specifies the start value in ydirection of the set of userdefined supplementary gridlines.
Type:
<R>
Default:

Units:
Range:
5 10 4 YS 10 3

ZE End Value in zDirection


This key specifies the end value in zdirection of the set of userdefined supplementary gridlines.
Type:
<R>
Default:

Units:
Range:
2 10 3 ZE 10 3

ZS Start Value in zDirection


This key specifies the start value in zdirection of the set of userdefined supplementary gridlines.
Type:
<R>
Default:

Units:
Range:
2 10 3 ZS 10 3

IMPLANT

83

Directive IMPLANT
The IMPLANT directive is used to specify a channel implantation. This directive can be omitted if a
homogeneous doped channel area is desired. One IMPLANT directive is necessary for a single channel
implanted device, and two IMPLANT statements are required for a double channelimplanted device.
However, only one set of anneal parameters is allowed, as both implants are always annealed together.
The channel and the sourcedrain temperature treatments are totally independent and do not at all affect
each other. The channel anneal parameters can be specified on the first or the second IMPLANT directive.
If FILE=2-D or FILE=1-D is specified on the PROFILE directive, no IMPLANT directive at all is
allowed.
The channel implantation is assumed to be performed through an insulator of material INS with thickness
TINS (cf. DEVICE directive). If a permittivity number is explicitly given with the INS key,
2 is
assumed for the insulator if INS=3 9 and 3 4 is assumed for any other value of INS.

ACTIVATION Activation of Dopant Atoms


The ACTIVATION key is used to calculate the activation of the dopants. This is done by using the
following formulation to calculate the active dopant concentration:

#2\

1


 



1 1018

If this key is omitted 100% activation will be assumed.


Type:
<R>
Default:

Units:

Range:
0 5 AC 1

AKEV Channel Implantation Energy


The AKEV key must not be omitted, if any of the IMPLANT keys is specified.
Type:
<R>
Default:


Units:
Range:
10 AKEV 300

DOSE Channel Implantation Dose


The DOSE key must not be omitted, if any of the IMPLANT keys is specified.
Type:
<R>
Default:

2
Units:
Range:
109 DOSE 1018

CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

84

ELEM Channel Implantation Element


Valid elements are antimony (SB), arsenic (A), boron (B) and phosphorus (P) for a silicon substrate.
For galliumarsenide substrates you may specify antimony (SB), arsenic (A), beryllium (BE), boron
(B), oxygen (O), magnesium (M), phosphorus (P), and silicon (S). The first character of the element
specifications is significant, except antimony and beryllium, for which the first two characters have to be
specified. The ELEM key must not be omitted, if any of the IMPLANT keys is specified.
Type:
<S>
Default:

Range:
A Arsenic
B Boron
BE Beryllium
M Magnesium
O Oxygen
P Phosphorus
SB Antimony
S Silicon

LD Diffusion Length
The LD key can be used to determine the extent of profile broadening due to annealing. If this key
is specified the TEMP and the TIME key are not effective. This key effects each IMPLANT statement
separately.
Type:
<R>
Default:

Units:
Range:
0 LD 10 4

TEMP Channel Implantation Anneal Temperature


The TEMP key must not be omitted, if any of the IMPLANT keys is specified. In case of a double implanted
channel, however, it may only be specified once.
Type:
<R>
Default:

Units:
Range:
500 TEMP 1300

TIME Channel Implantation Anneal Time


The TIME key must not be omitted, if any of the IMPLANT keys is specified. In case of a double implanted
channel, however, it may only be specified once.
Type:
<R>
Default:


Units:
Range:
60 TIME 48 3600

INTERFACE

85

Directive INTERFACE
The INTERFACE directive is used to specify the charge and traps at the semiconductor/insulator interfaces.
MINIMOS offers different possibilities to specify interface states and fixed oxide charge. All charge
profiles are independent, in other words the total charge density at the interface is given by superposition
of all specified densities. One elementary charge is assumed per state. The iterative procedure should
converge when the total trap density lies in the proposed ranges in common cases at room temperature. For
traps which are accounted for in both, steady-state and transient analysis it is possible to specify negative
values for densities, but PHYSCK=NO should be used. These densities will be subtracted from the total
density. More explanation is given in Section 2.7.
The INTERFACE directive can be omitted.

Fixed charge at the gate-oxide/bulk interface:

CISS Peak Density of the Gaussian Distributed Fixed Charge


This key determines the peak density of the fixed (Gaussiandistributed) oxide charge along the interface
between substrate and gate insulator. If this key is omitted, no Gaussian density profile is taken into
account. Note that beside the CISS key, the SISS and XISS keys may be used to control the Gaussian
distribution shape. Furthermore, if the CISS key is given, the XISS key is obligatory.
Type:
<R>
Default:
0
2
Units:
13
Range:
10
CISS 1013

NRD Space-Rectangular Fixed Charge Density


Type:
Default:
Units:
Range:

<R>
0


10
2

13

NRD

1013

NRE Space-Rectangular Fixed Charge End Coordinate


Type:
Default:
Units:
Range:

<R>
0

10
3

NRE

10

NRS Space-Rectangular Fixed Charge Start Coordinate


Type:
Default:
Units:
Range:

<R>
0

10
3

NRS

10

CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

86

NSS Fixed Oxide Charge Density


This key specifies the fixed oxide charge density at the interface between the substrate and the gate
insulator. The charge is assumed to be uniformly distributed along the interface. This key must not be
specified, if the UFB key (cf. OPTION directive) is given.
Type:
<R>
Default:
0
2
Units:
13
Range:
10
NSS 1013

SISS Standard Deviation of Gaussian Distributed Fixed Charge


This key specifies the standard deviation of Gaussiandistributed fixed oxide charge at the interface
between the substrate and the gate-insulator.
Type:
<R>
Default:
10 5
Units:
Range:
10 7 SISS 10 3

XISS Peak Position of Gaussian Distributed Fixed Charge


This key specifies the peak position of Gaussian distributed fixed oxide charge. Remember that the
sourcesided gate edge is the origin of the MINIMOS coordinate system. If this key is omitted, no
Gaussian distributed density profile is taken into account. If the XISS key is given, the CISS key is
obligatory.
Type:
<R>
Default:

Units:
Range:
10 3 XISS 10 2

Fixed charge at the bulk/substrate-insulator interface (for SOI):

PSS BulkInsulator Fixed Charge Density


This key specifies the density of fixed charge uniformly distributed along the interface between semiconductor and insulating substrate. This key is only active for SOI simulations.
Type:
<R>
Default:
0
2
Units:
Range:
1014 PSS 1014

INTERFACE

87

Fixed charge at the poly-gate/gate-oxide interface (poly-gate depletion analysis):

PNSS Fixed Oxide Charge at Gate/Oxide Interface


This key specifies the fixed oxide charge density at the interface between the poly-gate and the gateinsulator. The charge is assumed to be uniformly distributed along the interface.
Type:
<R>
Default:
0
2
Units:
14
Range:
10
PNSS 1014

Interface traps at the gate-oxide/bulk interface:

FSS Interface Trapped Charge Density


This key specifies the trap density (fast or chargeable surfacestates) at the interface between the substrate
and the gate-insulator. Positive FSS values implicitly denote a donor type and negative values acceptor
type of traps. The traps are assumed to be distributed uniformly along the forbidden band with density
FSS. Their population is determined by the surface quasiFermi level.
Type:
<R>
Default:
0
2 
1
Units:
Range:
1014 FSS 1014

CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

88

Interface traps at the poly-gate/gate-oxide interface:

PFSS Interface Traps at the Poly-Gate/Oxide Interface


This key specifies the trap density at the interface between the gate and the gate-oxide. This key is
interesting for gate depletion analysis. Positive PFSS values implicitly denote a donor type and negative
values acceptor type of traps. The traps are assumed to be distributed uniformly along the forbidden band
with density PFSS. Their population is determined by the surface quasiFermi level.
Type:
<R>
Default:
0
2 
1
Units:
14
10
PFSS 1014
Range:



DA0 Acceptor-like Traps at the Poly-Gate/Oxide Interface Parabolic in Energy


This key specifies a uniform acceptor-like trap density at the interface between the gate and the gateoxide. This key is interesting for poly-gate depletion analysis. The traps are assumed to be distributed
parabolically in the forbidden band from 0 (valence band edge) to DA0 (conduction band edge). Their
population is determined by the surface quasiFermi level.
Type:
<R>
Default:
0
2 
1
Units:
Range:
0 DA0 1014

DD0 Donor-like Traps at the Poly-Gate/Oxide Interface Parabolic in Energy


This key specifies a uniform donor-like trap density at the interface between the gate and the gate-oxide.
This key is interesting for gate depletion analysis. The traps are assumed to be distributed parabolically
in the forbidden band from DD0 (valence band edge) to 0 (conduction band edge). Their population is
determined by the surface quasiFermi level.
Type:
<R>
Default:
0
2 
1
Units:
Range:
0 DD0 1014



INTERFACE

89

Interface traps at the gate-oxide/bulk interface accounted for in the transient analysis:

MENERG Number of Trap Levels for the Discretization in Energy Space


This key determines the number of trap levels. It is active for interface traps only.
Type:
<I>
 E 
Default:
Units:

Range:
1 MENE 350

AUUD Acceptor-like Traps; Space Uniform; Energy Uniform


Type:
Default:
Units:
Range:

<R>
0


0 AUUD
2



1014

ARUD1 Density of Acceptor-like Traps; Space Rectangular; Energy Uniform (first set)
Type:
Default:
Units:
Range:

<R>
0


0 ARUD1 2 10
2



13

ARUS1 Starting Point for Acceptor-like Traps; Space Rectangular; Energy Uniform (first set)
Type:
Default:
Units:
Range:

<R>

10
3

ARUS1

10

ARUE1 Endpoint for Acceptor-like Traps; Space Rectangular; Energy Uniform (first set)
Type:
Default:
Units:
Range:

<R>

10
3

ARUE1

10

ARUD2 Density of Acceptor-like Traps; Space Rectangular; Energy Uniform (second set)
Type:
Default:
Units:
Range:

<R>
0


0 ARUD2 2 10
2



13

ARUS2 Starting Point for Acceptor-like Traps; Space Rectangular; Energy Uniform (second set)
Type:
Default:
Units:
Range:

<R>

10
3

ARUS2

10

ARUE2 Endpoint for Acceptor-like Traps; Space Rectangular; Energy Uniform (second set)
Type:
Default:
Units:
Range:

<R>

10
3

ARUE2

10

CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

90

AGUD1 Peak Density of Acceptor-like Traps; Space Gaussian; Energy Uniform (first set)
Type:
Default:
Units:
Range:

<R>
0


0 AGUD1 2 10
2



13

AGUS1 Standard Deviation for Acceptor-like Traps; Space Gaussian; Energy Uniform (first set)
Type:
Default:
Units:
Range:

<R>

10

AGUS1

10

AGUX1 Peak Position for Acceptor-like Traps; Space Gaussian; Energy Uniform (first set)
Type:
Default:
Units:
Range:

<R>

10
3

AGUX1

10

AGUD2 Peak Density of Acceptor-like Traps; Space Gaussian; Energy Uniform (second set)
Type:
Default:
Units:
Range:

<R>
0


0 AGUD2 2 10
2



13

AGUS2 Standard Deviation for Acceptor-like Traps; Space Gaussian; Energy Uniform (second set)
Type:
Default:
Units:
Range:

<R>

10

AGUS2

10

AGUX2 Peak Position for Acceptor-like Traps; Space Gaussian; Energy Uniform (second set)
Type:
Default:
Units:
Range:

<R>

10
3

AGUX2

10

AULD Acceptor-like Traps; Space Uniform; Energy Linear


The traps are distributed in the forbidden band from 0 at the valence band edge to AULD at the conduction
band edge.
Type:
<R>
Default:
0
2 
1
Units:
Range:
0 AULD 1014



AUGD Peak Density of Acceptor-like Traps; Space Uniform; Energy Gaussian


Type:
Default:
Units:
Range:

<R>
0


0 AUGD
2



1014

INTERFACE

91

AUGS Standard Deviation in Energy-space for Acceptor-like Traps; Space Uniform; Energy Gaussian
Type:
Default:
Units:
Range:

<R>




10

AUGS

AUGX Peak Position in Energy Space for Acceptor-like Traps: Space Uniform; Energy Gaussian
This enery has to be specified relatively to the valence band edge.
Type:
<R>
Default:

Units:

Range:
1 AUGX 2

AGLD Peak Density of Acceptor-like Traps; Space Gaussian; Energy Linear


The traps are distributed in the forbidden band from 0 at the valence band edge to AGLD (for the peak in
position space) at the conduction band.
Type:
<R>
Default:
0
2 
1
Units:
Range:
0 AGLD 2 1013

AGLS Standard Deviation for Acceptor-like Traps; Space Gaussian; Energy Linear
Type:
Default:
Units:
Range:

<R>

10

AGLS

10

AGLX Peak Position for Acceptor-like Traps; Space Gaussian; Energy Linear
Type:
Default:
Units:
Range:

<R>

10
3

AGLX

10

BAUUD Acceptor-like Traps at the Back Interface of SOI Devices


The traps are distributed uniformly in real space and uniformly in energy space.
Type:
<R>
Default:
0
2 
1
Units:

Range:
0 BAUUD 1014



BDUUD Donor-like Traps at the Back Interface of SOI Devices


The traps are distributed uniformly in real space and uniformly in energy space.
Type:
<R>
Default:
0
2 
1
Units:

Range:
0 BDUUD 1014



CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

92

DUUD Donor-like Traps; Space Uniform; Energy Uniform


Type:
Default:
Units:
Range:

<R>
0


0 DUUD
2





1014

DGUD Peak Density of Donor-like Traps; Space Gaussian; Energy Uniform


Type:
Default:
Units:
Range:

<R>
0


0 DGUD 2 10
2





13

DGUS Standard Deviation for Donor-like Traps; Space Gaussian; Energy Uniform
Type:
Default:
Units:
Range:

<R>



10

DGUS

10

DGUX Peak Position for Donor-like Traps; Space Gaussian; Energy Uniform
Type:
Default:
Units:
Range:

<R>

10



DGUX

10

DULD Donor-like Traps; Space Uniform; Energy Linear


The traps are distributed in the forbidden band from 0 at the valence band edge to DULD at the conduction
band.
Type:
<R>
Default:
0
2 
1

Units:
Range:
0 DULD 1014

DUGD Peak Density of Donor-like Traps; Space Uniform; Energy Gaussian


Type:
Default:
Units:
Range:

<R>
0


0 DUGD
2





1014

DUGS Standard Deviation in Energy Space for Donor-like Traps; SpaceUniform; Energy Gaussian
Type:
Default:
Units:
Range:

<R>

9


10

DUGS

DUGX Peak Position in Energy Space for Donor-like Traps: Space Uniform; Energy Gaussian
This energy has to be specified with respect to the valence band edge.
Type:
<R>
Default:

Units:

Range:
1 DUGX 2

INTERFACE

93

DGLD Peak Density of Donor-like Traps; Space Gaussian; Energy Linear


Type:
Default:
Units:
Range:

<R>
0


0 DGLD 2 10
2





13

DGLS Standard Deviation for Donor-like Traps; Space Gaussian; Energy Linear
Type:
Default:
Units:
Range:

<R>



10

DGLS

10

DGLX Peak Position for Donor-like Traps; Space Gaussian; Energy Linear
Type:
Default:
Units:
Range:

<R>

10



DGLX

10

METRAP External Trap Distribution in the Position Space

At present the traps are uniform in energy, acceptor-like and will be placed at the gate-oxide/bulk interface.
&
The data are read from an external file (file unit NUMETP), which has two columns containing !#"%$ (' .
1 , respectively.
The units of the data have to be  and  2 
Type:
<I>
Default:
0
Range:
0 no external traps
1 acceptor-like traps; energy uniform

G

CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

94

Directive MOBILITY
The MOBILITY directive is used to specify parameters that selectively stress or attenuate the various
scattering mechanisms. The physical meaning of the parameters given below are explained in detail in
Section 2.1 and Section 2.2. We recommend to use those weights only for fine tuning, since the default
values in the code are chosen carefully and agree rather well to experiments. This is particularly true for
silicon devices. For galliumarsenide devices only the MB, MI, ML, MT and MV keys are meaningful due
to a different mobility model, Section 2.2.
This directive can be omitted. No key of this directive is required. All values are dimensionless real
numbers.

MB Generic Weight
For Silicon, this key specifies the weight for the exponent in the relation between the drift velocity and the
electric field. A higher value leads to a shorter transition from ohmic mobility to saturation velocity. For
gallium arsenide this key is used to weight the minimum mobility in the mobility impurity concentration
relation.
Type:
<R>
Default:
1 for Si devices; 0 for GaAs devices
Range:
0 MB 3

MC Normal Field Exponent Weight


This key specifies the weight for the normal field exponent in the surface mobility model. MC > 1 leads
to a lower mobility and viceversa.
Type:
<R>
Default:
1
Range:
0 2 MC 2

MI Impurity Scattering Weight


This key specifies the weight for the critical concentration in the mobility model due to impurity scattering.
MI > 1 leads to a higher mobility and viceversa.
Type:
<R>
Default:
1
Range:
10 3 MI 106

ML Lattice Mobility Weight


This key specifies the weight for the zero field lattice mobility. ML > 1 leads to a higher lattice mobility
and viceversa.
Type:
<R>
Default:
1
Range:
0 5 ML 2

MOBILITY

95

MQ Weight for the Coulomb Scattering due to Interface States


The total number of surface charged-sites determines the mobility reduction. All traps and fixed charges
from the INTERFACE directive contribute to the scattering. MQ > 1 leads to a lower mobility and
viceversa. If this key is omitted, no mobility reduction due to this type of scattering will be taken into
account.
Type:
<R>
Default:
0
Range:
0 MQ 103

MR Weight for the Mobility at the Surface


This key specifies the weight for the surface mobility at the zero normal field (max. mobility at the
surface). MR > 1 leads to a higher mobility at the surface and viceversa.
Type:
<R>
Default:
1
Range:
0 5 MR 2

MS Surface Scattering Distance Weight


This key specifies the weight of the characteristic distance for the transition from normalfield dependent
surface mobility to bulk mobility. For MS > 1 this distance is decreased and viceversa.
Type:
<R>
Default:
1
Range:
0 1 MS 10

MT Critical Normal Field Weight


This key specifies the weight for the critical normal field in the surface mobility model. MT > 1 leads to a
higher mobility and viceversa. For galliumarsenide substrates MT weights the critical field in driftfield
relation.
Type:
<R>
Default:
1
Range:
0 1 MT 102

MV Saturation Velocity Weight


This key specifies the weight of the saturation velocity. MV > 1 leads to a higher saturation velocity and
viceversa.
Type:
<R>
Default:
1
Range:
0 5 MV 103

CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

96

Directive MONTE-CARLO
AUXS Source Sided Extension of Auxiliary Grid
Type:
Default:
Units:
Range:

<R>
0 1 10



AUXS

0 5 10

AUXD Drain Sided Extension of Auxiliary Grid


Type:
Default:
Units:
Range:

<R>
0 2 10



AUXD

0 5 10

BDR Back Diffusion Ratio at Drain Contact


This key specifies the back diffusion ratio at the absorbing drain contact. A carrier is reflected at the
contact BDR/(1-BDR) times before it is absorbed. BDR=0 means that all arriving carriers are absorbed,
BDR=1 means that all arriving carriers are reflected.
Type:
<R>
Default:
05
Units:
1
Range:
0 BDR 0 999

CNTMIN Threshold for Statistical Significance


To identity an average at a given mesh point as statistical reliable more than a specified minimum number
of scattering events have to contribute to this average. This minimum number is obtained by multiplying
CNTMIN with the total number of simulated scattering events.
Type:
<R>
Default:
10 5
Units:
1
Range:
0 CNTMIN 1

DTK) , *,+ 1 - 2 - 3 - 4 - 5 - 6 Coupling Constants of the Intervalley Phonons


With these keys six phonon coupling constants can be specified. The ponons with ordering numbers 1 2 3
are g-type phonons (the number of final valley .0/ equals to 1), those with ordering numbers 4 5 6 are
f-type phonons ( .0/
4 ' . The program accounts for six intervalley phonon scattering mechanims in any
case, even if some of them should be physically disabled by setting DTK
0.
Type:
<R>
Default:
0 5 108
0 8 108
11 0 108
0 3 108
2 0 108
2 0 108
Units:
 1 
Range:
0 DTK
1010







9 9

MONTE-CARLO

97

DEFPOT Acoustic Deformation Potential


Type:
Default:
Units:
Range:

<R>
90

DEFPOT

102

DEPSS Depth to Which Surface Scattering is Active


Type:
Default:
Units:
Range:

<R>
25 10



DEPSS

0 1 10

DFMIN Threshold Parameter for Mobility Window


Type:
Default:
Units:
Range:

<R>
2000

9
0

1 

DFMIN

105

DLSS Surface Roughness Parameter


This key specifies the product of correlation length times standard deviation of surface roughness.
Type:
<R>
Default:
3 10 15
2

Units:
Range:
0 DLSS 10 12

DOVL Drain Sided Overlap of Monte Carlo Window


Type:
Default:
Units:
Range:

<R>
0 3 10

0 DOVL 2 10
4



E1SELF Energy Parameter for the First Self-Scattering Interval


Type:
Default:
Units:
Range:

<R>
01




10

E1SELF

1.0

E2SELF Energy Parameter for the Second Self-Scattering Interval


Type:
Default:
Units:
Range:

<R>
50

E2SELF

10

EMULT Trajectory Multiplication Factor in Energy Domain


Type:
Default:
Units:
Range:

<I>
10
1
1 EMULT

104

CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

98

EPH) , *(+ 1 - 2 - 3 - 4 - 5 - 6 Energies of the Intervalley phonons


With these keys the energies of the six intervalley phonons can be specified. See also the description of
the DTK keys.
Type:
<R>
Default:
12 1 10 3
18 5 10 3
62 0 10 3
19 0 10 3
47 4 10 3
59 0 10 3

Units:
Range:
10 3 EPH
05







9


EREF Reference Field for Surface Roughness Scattering


Type:
Default:
Units:
Range:

<R>
105

2 

EREF

106

EXII Exponent in Keldysh Formula


Type:
Default:
Units:
Range:

<R>
2
1
0 2 EXII

5 0

EXSS Exponent in Surface Roughness Scattering Model


This key specifies the power law for the normal-field dependence in the surface roughness scattering
model.
Type:
<R>
Default:
2
Units:
1
Range:
0 1 EXSS 10

MCMOD Monte Carlo Simulation Mode


This key has the following meaning:
Description
Perform Monte Carlo Device Simulation (default)
Calculate material characteristics under uniform conditions
Print debye-length versus doping to output file
Print scattering rates versus energy to output file
Print self-scattering rate to output file

MCMOD
1
2
3
4
5

In MCMOD=2 material characteristics are calculated for a range of electric field strengths and for a range
of doping levels. These ranges can be specified on the STEP directive.
Type:
<I>
Default:
1
Range:
1 MCMOD 5

MONTE-CARLO

99

MCPIT Number of Monte-CarloPoisson Iterations


This key specifies the number of self-consistent iterations to be executed. The program terminates when
the specified number of iterations are done. The convergence history is printed to the reference file.
Type:
<I>
Default:
1
Range:
1 MCPIT 20

MLONG Longitudinal Effective Electron-Mass


This key specifies the ratio of the longitudinal electron-mass to the free electron-mass.
Type:
<R>
Default:
0 9163
Units:
1
Range:
10 2 MLONG 10

MTRAN Transversal Effective Electron-Mass


This key specifies the ratio of the transversal electron-mass to the free electron-mass.
Type:
<R>
Default:
0 1905
Units:
1
Range:
10 2 MLONG 10

NONPAR Non-Parabolicity Factor


This key specifies the non-parabolicity factor of the conduction band of silicon.
Type:
<R>
Default:
05
1

Units:
Range:
0 NONPAR 10

NSB Number of Particle Split Boundaries


This key specifies how often the trajectory multiplication algorithm is applied in real space. Up to three
boundaries can be specified. That means that the simulation domain is divided into one common domain
and three rare domains. The positions of the multilication boundaries in real space and the corresponding
multiplication factors are automatically chosen by MINIMOS. In energy space a carrier trajectory is
multiplied if the energy threshold for impact ionization is exceeded.
Type:
<I>
Default:
3
Range:
0 NSB 3

NUMB Number of Scattering Events per Single MC-Poisson Iteration


Type:
Default:
Range:

<I>
20 106 , MCMOD=1
106 , MCMOD=2
105 NUMB 5 108

CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

100

NUM) , * =1,2,3,4,5 Number of Scattering Events for a Specific MC-Poisson Iteration


For the lowest five self-consistent iterations the number of scattering events can be specified individually.
The key NUM overwrites the key NUMB for the -th iteration. With these keys one can reduce the number
of scattering-events for the very first iterations, when the solution is far away from self-consistency. In
the final iterations that determine the final results the noise can be reduced by using a higher number of
scattering events, that might be specified by the NUMB key.
Type:
<I>
Default:
NUMB
Range:
105 NUM
108

RHO Semiconductor Mass Density


This key specifies the mass density of silicon
Type:
<R>
Default:
23 28 10 4
3
43 
Units:
4
Range:
10
RHO 10 2

SAV) , *(+ 1 - 2 - 3 - 4 - 5 Save Results after a Specific MC-Poisson Iteration


Whith this keys up to five iteration numbers can be specified, after which certain distributed quantities are
to be written to file unit NUMUAI.
Type:
<I>
Default:

MCPIT
Range:
1 SAV

SEED Initial Seed of the Pseudo Random Number Generator


Type:
Default:
Range:

<I>
78
1 SEED

168

SOVL Source Sided Overlap of Monte Carlo Window


Type:
Default:
Units:
Range:

<R>
0 3 10

0 SOVL 2 10
4



TRACE Trace Execution of MC-Modul


If TRACE=YES is specified addtitional information is written to the reference file. A message is written
if a particle is launced or absorbed, and if a particle is split into light ones or light particles are gathered
to a heavy one.
Type:
<L>
Default:
NO

UTMIN Threshold Parameter for Temperature Window


Type:
Default:
Units:
Range:

<R>
1 05
1
1 UTMIN

10

MONTE-CARLO

101

VSOUND Sound Velocity in the Semiconductor


Type:
Default:
Units:
Range:

<R>
90 33 104

10 VSOUND
5

106

WBH Weight of Brooks-Herring Scattering Rate


This key weights the Brooks-Herring scattering rate in Ridleys formula for ionized impurity scattering.
Type:
<R>
Default:
1
Units:
1
Range:
0 WBH 102

WBII Weight of the Prefactor in the Keldysh Formula


Type:
Default:
Units:
Range:

<R>
3 9 10 2
1
0 WBII

103

WDIST Weight of the Average Distance Between Ions


In Ridleys model for ionized impurity scattering the average distance between ions in conjunction with
the actual carrier velocity determines the upper limit for the scattering rate. This key allows to weight this
&
WDIST 2 87 :9 ' 1 ; 3 .
distance which is obtained from the local impurity concentration by 6
Type:
<R>
Default:
1
Units:
1
Range:
10 2 WDIST 102

WEII Weight of The Engergy Threshold for Impact Ionization


Type:
Default:
Units:
Range:

<R>
1
1
10 2

WEII

102

WINFREEZE Freeze Window for MC-Transport Parameters


If this key is set to YES the windows in which the analytical transport parameters are replaced by the
MC generated ones do not change between succsessive MC-Poisson iterations. Otherwise the criteria for
window boundary placement are evaluated after each MC-Poisson iteration.
Type:
<L>
Default:
NO

WSCREEN Weight of the Screening Length


With this key the influence of the screening length on the ionized impurity scattering rate can be weighted.
Type:
<R>
Default:
1
Units:
1
Range:
10 2 WSCREEN 102

CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

102

XWIDTH Lateral Width of Assignment Function


Type:
Default:
Units:
Range:

<R>
Minimum of <1 10 and 50 10

10 XWIDTH 0 5 10



YWIDTH Vertical Width of Assignment Function


Type:
Default:
Units:
Range:

<R>
Minimum of <

10 YWIDTH 0 5 10


50 and 10 10



OPTION

103

Directive OPTION
The OPTION directive serves as a directory for general purpose keys. Some physical device parameters
are collected herein as well as parameters to control MINIMOS execution. This directive can be omitted.
No key of this directive is required.

ACCELERATION Try to Accelerate Convergence


This key may be used to suppress the use of an acceleration algorithm in the iteration process.
Type:
<L>
Default:
YES

CCLP Critical Carrier Concentration for Determining the Capture Boundary


This key specifies the critical electron or hole concentration for finding the x-coordinate which determines the area for the total electron and hole capture at the specified terminal bias (charge-pumping
threshold and flat-band voltage). The output is written in the file *.LATPROF. The result is in a format
&
NMPO
=>?A@B?1?ACDFEH@ G FEHC G #<
IEJ@G FE5@ K FE5C K #<
LE5@ K
' , where EH@ G , EJCG , EQ@ K
and E5C K are the coordinates at the drain and source side for electrons and holes, respectively. < is the gate length (LGATE).
If the coordinate does not exist the output is the total left or right coordinate of the simulation domain.
Remember that the source-sided gate edge is the origin of the MINIMOS x-coordinate axis. Note that the
calculation assumes DC condition.
Type:
<R>
Default:

3
Units:

5
Range:
10
CCLP 1020

CP Flag for Coupled Solution of AC Equations


Type:
Default:

<L>
NO

CURRENT Drain Current for Threshold Calculation


This key is active only for threshold voltage calculation, i.e. if MODEL=THRES is specified on the OPTION
directive. In that case the CURRENT key determines the drain current for threshold condition.
Type:
<R>
Default:
10 7 RS for nchannel devices
10 7 RS for pchannel devices
Units:
T
Range:
2 10 3 CURRENT 2 10 3

EA Acceptor Activation Energy


This key specifies the difference between the impurity level of acceptors and the valence band edge
(ionization energy). If EA is not specified, total ionization of acceptors is assumed.
Type:
<R>
Default:
Total Ionization
Units:

Range:
0 001 EA 0 1
Typical: 0 045 for boron in silicon

CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

104

ED Donor Activation Energy


This key specifies the difference between the conduction band edge and the impurity level of donors
(ionization energy). If ED is not specified total ionization of donors is assumed.
Type:
<R>
Default:
Total Ionization

Units:
Range:
0 001 ED 0 1
Typical: 0 039 for antimony in silicon
0 045 for phosphorus in silicon
0 054 for arsenic in silicon

GCHCALC Calculation of the Total Gate-Charge

If GCHCALC is specified, the total charge in the gate will be calculated by evaluating the electric field flux

, the source-sided charge U DVWV ,
through a contour around the gate. The results are the total charge U 
the bottom charge U (X and the drain-sided charge U (YZV . The result is written to file unit NUMCG1 in the
&

form =>?A@B?2?AC(U[
U[DV\V8U (X U (YZV ' .
Type:
<L>
Default:
NO

GRIDFREEZE Freeze Grid


This key is active only if any of the step parameters has been chosen on the STEP directive. This key may
be used to determine whether grid modifications should be performed between voltage steps or not.
Type:
<L>
Default:
NO

IDEALITY Ideality Factor for Gate Diode


This key specifies an ideality factor for the gate diode of MESFETs. It affects the gate current in the
forward biased region. This factor is defined by
]

9 @

1


&`_ ]Qa
^


'

cb

3 

This key is only effective in case of MESFET simulation. If this key is omitted an ideal gate diode will be
assumed.
Type:
<R>
Default:
1
Units:

Range:
1 ID 5

INTRINSIC Intrinsic Concentration


This key specifies the intrinsic concentration at the simulation temperature. If this key is omitted, a
builtin model for the temperature dependence of the intrinsic concentration will be used.
Type:
<R>
Default:
Builtin Model
3
Units:

Range:
10 20 INTRINSIC 1015

OPTION

105

LCPUMP Charge Pumping Relevant Data on Output


If LCP is specified, all quantities necessary to perform postprocessing for the charge-pumping analysis
will be written on output after transient simulation. Independent of LCP all specified interface and bulk
traps are always taken into account in the selfconsistent calculation.
Type:
<L>
Default:
NO

LDRAIN Drain Contact Length


This key specifies the length of the drain contact in xdirection
Type:
<R>
Default:

Units:

Range:
10 6 LDRAIN 0 1

LSOURCE Source Contact Length


This key specifies the length of the source contact in xdirection.
Type:
<R>
Default:

Units:

Range:
10 6 LSOURCE 0 1

LVDMOSFET Device is Interpreted as Vertical DMOS Transistor


If this key is specified, doping has to be read in from a two-dimensional doping file. Furthermore, the
length of the source contact (see OPTION directive) and the depth of the simulation domain must be given
(see DEVICE directive).
Type:
<L>
Default:
NO

MODEL MINIMOS 2D Model Option


This key specifies which mode of calculation MINIMOS 2D will perform. For a detailed explanation of
the five modes see Section 1.2. Only the first character of the specified character string is significant.
Type:
<S>
Default:
2-D
Range:
1-D
THRES
2-D
AVAL
HOT
MPOLY Model for the Gate Depletion Analysis
This key is active only in the gate depletion analysis.
Type:
<I>
Default:
1
Range:
0 one-dimensional model
1 full two-dimensional solution in the poly-gate area

CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

106

M3MODE MINIMOS 3D Model Option


This key is required if MINIMOS 3D is to be invoked. It specifies the mode of calculation for MINIMOS 3D. For a detailed explanation of the four modes see section Section 1.2. Only the first character of
the specified string is obligatory. The second character N - new, O - old, is optional. Default is N.
Type:
<S>
Default:

Range:
1N
1O
2N
2O

OMEGA ACAnalysis Frequency


This key specifies the angular frequency for which ACanalysis should be carried out.
Type:
<R>
Default:
104 27
Units:
rad d 1
Range:
1 OMEGA 6 3 1010

PDOP Activated Doping in the Poly-Gate


This key specifies the activated impurity concentration in the poly-gate area. The type of impurity is
determined by the GATE directive. This key activates the gate depletion analysis. Thickness of the
poly-gate layer is given with the GCTHE key.
Type:
<R>
Default:

3

Units:
Range:
1014 PDOP 1021

PHYSCK Check Physical Error


This key specifies whether physical errors should be processed. Physical errors are out-of-range
errors of any numerical input value as well as sign errors within the BIAS directive. Another meaning
of this key is to decide whether execution should be terminated if any iteration seems to diverge (e.g.
threshold voltage infinite due to punch-through).
If physical errors are detected by MINIMOS with PHYSCK=NO, a warning message will be issued. If
this key is omitted, physical errors will be processed. For instance, if a positive bulksource voltage
should be specified for an n-channel device PHYSCK=NO has to be given. Since any guarantee of regular
execution is lost with PHYSCK=NO, use of this statement is not recommended.
Type:
<L>
Default:
YES

OPTION

107

RBULK Effective Bulk Resistor


This key specifies an effective bulk resistor. When this key is given, the effective substrate bias is calculated
M4i
Zj CNkXlV . For UB see Page 62. As a large value of RBULK can affect the computation time in
as ef hgfNe
an undesirable manner, it is recommended to determine the effective substrate bias by the UB key directly,
provided that j CmkXlV is known. This is not applicable if the STEP directive is given. This key is active only
if MODEL=AVAL or MODEL=HOT is specified.
Type:
<R>
Default:
0
Units:

Range:
0 RBULK 105

RDRAIN Effective Drain Resistor


This key specifies an effective drain resistor. When this key is given, the effective drain bias is calculated
qpsr
j8@ . For UD see Page 62. As a large value of RDRAIN can affect the computation time
as en ogPn4g
in an undesirable manner it is recommended to determine the effective drain bias by the UD key directly,
provided that j @ is known. This is not applicable if the STEP directive is given.
Type:
<R>
Default:
0
Units:

Range:
0 RDRAIN 100

RSOURCE Effective Source Resistor

This key specifies an effective source resistor. When this key is given, the effective source bias is calculated
xw
as eqt ugqt4veg
sj8C . For US see description of the BIAS directive. As a large value of RSOURCE can
affect the computation time in an undesirable manner it is recommended to determine the effective source
bias by the US key directly, provided that j8C is known. This is not applicable if the STEP directive is
given.
Type:
<R>
Default:
0
Units:

Range:
0 RSOURCE 100

TEMP Simulation Temperature


Type:
Default:
Units:
Range:

<R>
300
y

77

TEMP

450

UFB Flat-band Voltage


This key specifies the flatband voltage. If this key is omitted, the flatband voltage will be calculated
from the work function difference of the gate, (GATE key on the DEVICE directive) the Fermi barrier
deep in the substrate and the fixed oxide charges (cf. INTERFACE directive). If UFB is specified, the
GATE key on the DEVICE directive and the NSS key on the INTERFACE directive must be omitted.
Type:
<R>
Default:
Builtin Model
Units:
Range:
3 UFB 3

CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

108

UN Energy Relaxation Time Weight (Electrons)


This key specifies a weight for the energy relaxation time for electrons. These keys are active only if
MODEL=HOT on the OPTION directive is specified.
Type:
<R>
Default:
08
Units:

Range:
0 UN 1

UP Energy Relaxation Time Weight (Holes)


This key specifies a weight for the energy relaxation time for holes. These keys are active only if
MODEL=HOT on the OPTION directive is specified.
Type:
<R>
Default:
08
Units:

Range:
0 UP 1

OUTPUT

109

Directive OUTPUT
The OUTPUT directive is used to specify the physical quantities which are to be printed after the simulation
has terminated successfully. All keys possible on this directive are logical items. Thus, the only decision
requested is if some physical quantity is to be printed or not.

ALL Print All Quantities


This key specifies the printing of all available quantities regardless of the specifications for the individual
keys.
Type:
<L>
Default:
NO
AS Flag to Produce Formatted Output File
Type:
Default:

<L>
YES

AVAL Print Avalanche Generation Rate


The avalanche generation rate is available only if MODEL=AVAL or MODEL=HOT is given on the OPTION
directive.
Type:
<L>
Default:
NO

BI Flag to Produce Binary File


Type:
Default:

<L>
NO

CC Print Carrier Concentrations


Type:
Default:

<L>
YES

DC Print Doping Profile


Type:
Default:

<L>
NO

DIGITS Number of Decimal Digits written to PIF-File


With this key the floating point format used for ASCII-Pif output is controlled. DIGITS specifies the
number of decimal digits to be written. This key applies only if PIF=YES is set on the OUTPUT directive.
Type:
<I>
Default:
6
Range:
2 DIGITS 18

ELAT Print Lateral Electric Field


Type:
Default:

<L>
YES

ETRAN Print Transversal Electric Field


Type:
Default:

<L>
YES

IV Print Current Voltage Characteristics


Type:
Default:

<L>
YES

CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

110

JLAT Print Lateral Current Density


Type:
Default:

<L>
NO

JTRAN Print Transversal Current Density


Type:
Default:

<L>
NO

MAJ Print Majorities


Majority quantities are available only if MODEL=AVAL or MODEL=HOT is given on the OPTION directive.
Type:
<L>
Default:
NO
MIN Print Minorities
Type:
Default:

<L>
YES

MOB Print Mobility Distributions


Type:
Default:

<L>
NO

MTV Write output for PLOTMTV


For MTV=1 and MTV=2 the CONTOUR and CONTCURV format of PLOTMTV is used, respectively. For
MTV=0 no output is written.
Type:
<I>
Default:
0
Range:
0 MTV 2

NONE Print No Distributions


This key suppresses the printing of all available quantities regardless of the specifications for the individual
keys. If both, NONE=YES and ALL=YES, are specified, no distributions will be printed.
Type:
<L>
Default:
NO
OXIDE Print Quantities within the Oxide
This key specifies the printing of the electrostatic potential and the electric field components within the
insulator.
Type:
<L>
Default:
NO

PHI Print QuasiFermi Levels


This quantity is not available if MODEL=HOT is given on the OPTION directive.
Type:
<L>
Default:
NO
PIF Write an ASCII-PIF file
Quantities to be written to the PIF-file can be specified on the OUTPUT-directive
Type:
<L>
Default:
NO

PSI Print Electrostatic Potential


Type:
Default:

<L>
YES

OUTPUT

111

RHO Print Space Charge


Type:
Default:

<L>
NO

STRIPS Columns in Print File


This key specifies the number of columns in the print file.
Type:
<I>
Default:
12
Range:
4 STRIPS 20

TC Print Carrier Temperatures


This quantity is available only if MODEL=HOT is given on the OPTION directive.
Type:
<L>
Default:
NO

CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

112

Directive PROFILE
The PROFILE directive is used to specify the source and drain doping profiles, the bulk doping and the
bulk trap distribution. Either an external file, from which the doping profile for sourcedrain and the
channel is to be read, can be specified, or parameters for an internal process simulation that generates the
doping profile can be given. Both possibilities are mutually exclusive.
In case of FILE=NO or FILE=1-D the mask edge for source or drain profiles is assumed to be at the gate
contact edges for MOSFETs. XOFF specifies the offset from these mask edges. For MESFETs the mask
edges are assumed to be the source contact edge for the source profile and the drain contact edge for the
drain profile. These mask edges can be shifted with the DMOFF and the SMOFF key. To specify a constant
doping of either n or p type, e.g. for the simulation of a MESFET with constant substrate doping, the only
permitted keys on the PROFILE directive are the NB key together with the SUBS key. For MINIMOS 3D
FILE=NO must be in effect, which means that presently no external doping file for threedimensional
simulation can be used.

Possibilities to Define a SourceDrain Profile


There are four choices to define a source-drain profile. Once the user has chosen one of them, the
associated keys must appear. Keys in brackets may be omitted.

1. FILE=1-D [, ASYM=<L> ] [, LFIT=<R> ] [, XOFF=<R> ] [, YOFF=<R> ]


2. FILE=2-D
3. [ FILE=NO ,] [ ASYM=<L> ,] NB=<R> , NS=<R> , [ SUBS=<C> ,] TEMP=<R> , TIME=<R>
[, XOFF=<R> ] [, YOFF=<R> ]
4. [ FILE=NO ,] [ ASYM=<L> ,] AKEV=<R> , DOSE=<R> , ELEM=<S> , NB=<R> [, SUBS=<C> ],
TEMP=<R> , TIME=<R> [, TOX=<R> ] [, XOFF=<R> ] [, YOFF=<R> ]
If FILE=NO and ASYM=YES, or FILE=NO and XOFF is given, two PROFILE directives must be used.

ACTIVATION Activation of Dopant Atoms


The ACTIVATION key is used to calculate the activation of the dopants. This is done by using the
following formulation to calculate the active dopant concentration:

# \

1


l



1 1018

If this key is omitted 100% activation will be assumed.


Type:
<R>
Default:

Units:

Range:
0 1 AC 1

PROFILE

113

AKEV Implantation Energy


This key specifies the sourcedrain implantation energy. If the AKEV key is specified with a particular
PROFILE directive, the ELEM and DOSE keys have also to be specified and the NS key must be omitted
on this PROFILE directive . NS may appear on another PROFILE directive if two PROFILE directives
are required. This key must be omitted if FILE=2-D or FILE=1-D is specified.
Type:
<R>
Default:


Units:
Range:
10 AKEV 300

ASYM Asymmetric Profile


This key specifies a nonsymmetric sourcedrain profile. If ASYM=YES and FILE=NO is specified, two
PROFILE directives must appear. The profile given with the first PROFILE directive will be used for the
source region and the profile given with the second PROFILE directive will be used for the drain region.
If XOFF is also specified, the profile given with the first PROFILE directive will be used for source and
drain, whereas the profile given with the second PROFILE directive will be superimposed at offset XOFF
at the drain only. If FILE=1-D is specified, there have to be two profile slices on the external doping
file. See also section Chapter 4. The data of the first slice are then used for the source region and the data
from the second slice are used for the drain region. Similarly, if XOFF is also specified, the data of the
first slice will be used for source and drain, whereas the data from the second slice will be superimposed
at offset XOFF at the drain only. The ASYM key must be omitted if FILE=2-D is specified. The ASYM
key has to be specified if XOFF is negative.
Type:
<L>
Default:
NO

CA Concentration of Acceptor-like Deep (Bulk) Traps


This key specifies the concentration of the single-level, acceptor-like deep traps uniformly distributed in
the bulk. Related keys are EA (PROFILE directive) and CAN and CAP (RECOMBINATION directive).
These traps are accounted for in the transient simulation as well. They can be combined with CD, but not
with VAUD, VDUD, VAGD or VDGD.
Type:
<R>
Default:
0
3

Units:
Range:
0 CA 2 1018

CD Concentration of Donor-like Deep (Bulk) Traps


This key specifies the concentration of the single-level, donor-like deep traps uniformly distributed in the
bulk. Related keys are ED (PROFILE directive) and CDN and CDP (RECOMBINATION directive). These
traps are accounted for in the transient simulation as well.
Type:
<R>
Default:
0
3
Units:

Range:
0 CD 2 1018

CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

114

C1 Concentration of First Epilayer


This key specifies the net doping concentration of the first epilayer.
Type:
<R>
Default:
0
3
Units:

Range:
1019 C1 1019

C2 Concentration of Second Epilayer


This key specifies the net doping concentration of the second epilayer.
Type:
<R>
Default:
0
3
Units:

19
Range:
10
C2 1019

DMOFF Offset of the Drain Profile Mask


This key specifies the shift of the implantation mask from the gate contact edge to the drain contact.
Type:
<R>
Default:
DGAP
Units:

Range:
0 DM 10 2

DOSE Implantation Dose


This key specifies the source-drain implantation dose. If the DOSE key is specified with a particular
PROFILE directive, the ELEM and AKEV keys have also to be specified and the NS key must be omitted
on this PROFILE directive. This key must be omitted if FILE=2-D or FILE=1-D is specified.
Type:
<R>
Default:

2
Units:

10
Range:
10
DOSE 1017

EA Energy Level of Acceptor-like Deep (Bulk) Traps


This key specifies the single energy level (activation energy) for the acceptor-like deep traps CA, with
respect to the valence band edge.
Type:
<R>
Default:


Units:
Range:
0 EA 1 2

ED Energy Level of Donor-like Deep (Bulk) Traps


This key specifies the single energy level (activation energy) for the donor-like deep traps CD, with respect
to the conduction band edge.
Type:
<R>
Default:

Units:

Range:
0 ED 1 2

PROFILE

115

ELEM Implantation Element


This key specifies the sourcedrain implantation element. Valid elements are antimony (SB), arsenic (A),
boron (B) and phosphorus (P) for a silicon substrate. For galliumarsenide substrates you may specify
antimony (SB), arsenic (A), beryllium (BE), boron (B), oxygen (O), magnesium (M), phosphorus (P), and
silicon (S). The first character of the element specifications is significant, except antimony and beryllium,
for which the first two characters have to be specified. If the ELEM key appears together with a particular
PROFILE directive, the DOSE and AKEV keys have also to be specified and the NS key must be omitted.
For nchannel MOSFETs only P, SB and AS is allowed, for p-channel devices only B can be specified.
This key must be omitted if FILE=2-D or FILE=1-D is specified.
Type:
<S>
Default:

Range:
AS Arsenic
B Boron
BE Beryllium
MG Magnesium
O Oxygen
P Phosphorus
SB Antimony
SI Silicon

FILE Read Profile from External File


This key specifies if doping data are to be read from an external file. The format of the external file is
given in Chapter 4. The first character is of the specified character string is significant only. 1-D denotes
onedimensional external doping data and is equivalent with YES. 2-D denotes twodimensional external
doping data. In that case no other key must be given. With FILE=1-D only the ASYM, LFIT, XOFF or
YOFF keys are compatible. See also the explanation given at the beginning of this section.
Type:
<S>
Default:
NO
Range:
1-D Onedimensional doping data
2-D Twodimensional doping data
NO No external doping input
YES Same as 1-D

LD Diffusion Length
The LD key can be used to determine the extent of profile broadening due to annealing. If this key is
specified the TEMP and the TIME key are not effective. This key effects all PROFILE statements.
Type:
<R>
Default:


Units:
Range:
0 LD 10 4

LFIT Subdiffusion Fitting Parameter


This key specifies the fitting parameter for the subdiffusion in case of using doping data from an external
file. The LFIT key can be omitted but, if specified, FILE=1-D must have been given too.
Type:
<R>
Default:
07
Range:
0 1 LFIT 3

CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

116

NB Bulk Doping
The NB key has to be specified unless FILE=2-D or FILE=1-D has been given. In this case the NB key
must be omitted.
Type:
<R>
Default:
0
3

Units:
Range:
0 NB 5 1018

NS Predeposition Concentration
This key specifies the surface concentration for a sourcedrain predeposition. If the NS key is specified with
a particular PROFILE directive, the ELEM, DOSE, AKEV and TOX keys must not be specified. Phosphorus
and boron dopants are assumed for n-channel and p-channel devices, respectively. The specified value
must be at least 100 times larger than the bulk doping. The NS key must be omitted, if FILE=2-D or
FILE=1-D is specified or the ELEM, DOSE and AKEV keys are given. Note that predeposition is not
implemented for GaAs as substrate material.
Type:
<R>
Default:

3

Units:
Range:
1015 NS 5 1022

SMOFF Offset of the Source Profile Mask


This key specifies the shift of the implantation mask from the gate contact edge to the source contact.
Type:
<R>
Default:
SGAP

Units:
Range:
0 SM 10 2

SUBS Substrate Material Type


This key specifies the type of the substrate material. Allowed characters are N for ndoped substrates, or P
for pdoped substrates. On default a ptype substrate is assumed for nchannel MOSFETs and an ntype
substrate is assumed for pchannel MOSFETs. For nchannel MESFETs the substrate type defaults to
ntype and for pchannel MESFETs to ptype.
Type:
<C>
Default:
P for nchannel MOSFET
N for pchannel MOSFET
N for nchannel MESFET
P for pchannel MESFET
Range:
N for ndoped substrate
P for pdoped substrate

PROFILE

117

TEMP Diffusion Temperature


This key specifies the sourcedrain profile diffusion temperature. The TEMP key has to be specified unless
FILE=2-D or FILE=1-D has been given. In this case the TEMP key must be omitted. If two PROFILE
directives are required, the TEMP key may appear only once with either the first or the second PROFILE
directive.
Type:
<R>
Default:

Units:
Range:
100 TEMP 1300

TIME Diffusion Time


This key specifies the sourcedrain profile diffusion time. The TIME key has to be specified unless
FILE=2-D or FILE=1-D has been given. In this case the TIME key must be omitted. If two PROFILE
directives are required, the TIME key may appear only once with either the first or the second PROFILE
directive.
Type:
<R>
Default:


Units:
Range:
60 TIME 48 3600

TOX Implantation Isolation Oxide Thickness


This key specifies the sourcedrain implantation isolation oxide. This key can be omitted. If specified,
the ELEM, DOSE and AKEV keys have also to be specified and the NS key must be omitted. If not specified
and the ELEM, DOSE and AKEV keys are given, a default value of 0 is assumed, that means implantation
into the bare surface. If two PROFILE directives with implantation data are required, the TOX key may
appear only once with either the first or the second PROFILE directive. This key must be omitted if
FILE=2-D or FILE=1-D is specified.
Type:
<R>
Default:
0

Units:
Range:
0 TOX 10 3

XOFF Offset for Superposition of Two Profiles


This key specifies an offset for superposition of two source and/or drain profiles. If XOFF is specified,
either two PROFILE directives have to be given (FILE=NO) or two slices for a source-drain profile
have to be given with the external doping file (FILE=1-D). This key must be omitted if FILE=2-D is
specified. If a negative value is specified, the ASYM key must be given. For MESFET simulation XOFF
specifies the shift of the LDD-implant mask edge from the gate contact.
Type:
<R>
Default:
0

Units:
Range:
2 10 4 XOFF 2 10 4

CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

118

YOFF Profile Shift Offset


This key specifies a distance for shifting the source-drain profile into the bulk. Used cleverly, the YOFF
key allows to account to some extent for non-planar source-drain surfaces. This key must be omitted if
FILE=2-D is specified.
Type:
<R>
Default:
0 (No vertical shift)
Units:

Range:
0 YOFF 10 4

VAGD Peak Density of Acceptor-like Bulk Traps, Gaussian Distributed in Energy


The traps are distributed uniformly in the bulk. Related keys are VAGX, VAGS (PROFILE directive) and
CAN and CAP (RECOMBINATION directive). These traps are accounted for in the transient simulation as
well. They can be combined with VDGD, but not with CA, CD, VAUD or VDUD.
Type:
<R>
Default:
0
3
1

Units:
Range:
0 VAGD 1019



VAGS Standard Deviation of Acceptor-like Bulk Traps, Gaussian Distributed in Energy


Type:
Default:
Units:
Range:

<R>

9


10

VAGS

VAGX Energy Position of Acceptor-like Bulk Traps: Gaussian Distributed in Energy


This energy has to be specified relatively to the valence band edge.
Type:
<R>
Default:


Units:
Range:
1 VAGX 2

VAUD Concentration of Uniformly Distributed Acceptor-like Bulk Traps


The traps are distributed uniformly in both, energy and position space. Related keys are VDEN (PROFILE
directive) and CAN and CAP (RECOMBINATION directive). These traps are accounted for in the transient
simulation as well. They can be combined with VDUD, but not with CA, CD, VAGD or VDGD.
Type:
<R>
Default:
0
3
1

Units:
Range:
0 VAUD 1019

VDEN Discretization Step in the Energy Space


Bulk traps are discretized in the energy space according to their standard deviations VAGS and VDGS,
thermal voltage   and the parameter VDEN. An explanation is given in Section 2.7.
Type:
<R>
Default:

Units:

Range:
2 5 10 4 VDEN 0 5

PROFILE

119

VDGD Peak Density of Donor-like Traps: Gaussian Distributed in Energy


The traps are distributed uniformly in the bulk. Related keys are VDGX, VDGS (PROFILE directive) and
CDN and CDP (RECOMBINATION directive). These traps are accounted for in the transient simulation as
well. They can be combined with VAGD, but not with CA, CD, VAUD or VDUD.
Type:
<R>
Default:
0
3
1
Units:

Range:
0 VDGD 1019

G

VDGS Standard Deviation of Donor-like Bulk Traps, Gaussian Distributed in Energy


Type:
Default:
Units:
Range:

<R>




10

VAGS

VDGX Energy Position of Donor-like Bulk Traps: Gaussian Distributed in Energy


This energy has to be specified relatively to the valence band edge.
Type:
<R>
Default:


Units:
Range:
1 VDGX 2

VDUD Concentration of Uniformly Distributed Donor-like Bulk Traps


The traps are distributed uniformly in both, energy and position space. Related keys are VDEN (PROFILE
directive) and CDN and CDP (RECOMBINATION directive). These traps are accounted for in the transient
simulation as well. They can be combined with VAUD, but not with CA, CD, VAGD or VDGD.
Type:
<R>
Default:
0
3
1

Units:
Range:
0 VDUD 1019

CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

120

Directive RECOMBINATION
The RECOMBINATION directive is used to specify generation and recombination parameters. For a more
detailed explanation see Section 2.7. This directive can be omitted. No key of this directive is required.
Any of the following keys is active only if MODEL=AVAL or MODEL=HOT is specified (cf. OPTION
directive).

AN Impact Ionization Rate Parameter Electrons


Type:
Default:
Units:
Range:

<R>
9 20 105 in silicon
1 34 106 in galliumarsenide




0 AN
1



109

AP Impact Ionization Rate Parameter Holes


Type:
Default:

<R>
1 70 107 in silicon
1 34 106 in galliumarsenide

Units:
Range:






0 AP
1

109

BN Impact Ionization Rate Parameter Electrons


Type:
Default:
Units:
Range:

<R>
2 02 106 in silicon
2 03 106 in galliumarsenide



9
5 10

2 

BN

107

BP Impact Ionization Rate Parameter Holes


Type:
Default:

<R>
3 20 106 in silicon
2 03 106 in gallium arsenide

Units:
Range:

2 



9
5 10
5

BP

107

CAN Electron Trapping Cross-Section for Acceptor-like Bulk Traps


This key specifies the cross-section of the specified acceptor-like bulk traps (deep acceptors the keys
CA, VAUD and VAGD in the PROFILE directive) for electron trapping.
Type:
<R>
Default:
10 16
2
Units:

Range:
0 CAN 106

RECOMBINATION

121

CAP Hole Trapping Cross-Section for Acceptor-like Deep (Bulk) Traps


This key specifies the cross-section of the specified acceptor-like bulk traps (deep acceptors the keys
CA, VAUD and VAGD in the PROFILE directive) for hole trapping.
Type:
<R>
Default:
10 16
2

Units:
Range:
0 CAP 106

CDN Electron Trapping Cross-Section for Donor-like Deep (Bulk) Traps


This key specifies the cross-section of the specified donor-like bulk traps (deep donors the keys CD,
VDUD and VDGD in the PROFILE directive) for electron trapping.
Type:
<R>
Default:
10 16
2
Units:

Range:
0 CDN 106

CDP Hole Trapping Cross-Section for Donor-like Deep (Bulk) Traps


This key specifies the cross-section of the specified donor-like bulk traps (deep donors the keys CD,
VDUD and VDGD in the PROFILE directive) for hole trapping.
Type:
<R>
Default:
10 16
2

Units:
Range:
0 CDP 106

CN Auger Recombination Parameter for Electrons


Type:
Default:
Units:
Range:

<R>
2 8 10
5 0 10




0 CN


6

31
31

in silicon
in gallium arsenide

10

28

CP Auger Recombination Parameter for Holes


Type:
Default:
Units:
Range:

<R>
9 9 10
5 0 10



z
0 CP


6

32
31

in silicon
in galliumarsenide

10

28

CSEA Electron Trapping Cross-Section for Acceptor-like Interface Traps


This key specifies the cross-section of all specified acceptor-like interface traps for electron trapping.
Type:
<R>
Default:
2 10 16
2
Units:

Range:
10 20 CSEA 10 10

CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

122

CSED Electron Trapping Cross-Section for Donor-like Interface Traps


This key specifies the cross-section of all specified donor-like interface traps for electron trapping.
Type:
<R>
Default:
2 10 16
2
Units:

Range:
10 20 CSED 10 10

CSHA Hole Trapping Cross-Section for Acceptor-like Interface Traps


This key specifies the cross-section of all specified acceptor-like interface traps for hole trapping.
Type:
<R>
Default:
0 5 10 16
2
Units:

Range:
10 20 CSHA 10 10

CSHD Hole Trapping Cross-Section for Donor-like Interface Traps


This key specifies the cross-section of all specified donor-like interface traps for hole trapping.
Type:
<R>
Default:
0 5 10 16
2
Units:

Range:
10 20 CSHD 10 10

DN Impact Ionization Rate Parameter Electrons


Type:
Default:
Units:
Range:

<R>
1 5 10 6 in silicon
not used for galliumarsenide



10

DN

10

DP Impact Ionization Rate Parameter Holes


Type:
Default:
Units:
Range:

<R>
1 5 10 6 in silicon
not used for galliumarsenide



10

DP

10

EN Impact Ionization Rate Parameter Electrons


Type:
Default:
Units:
Range:

<R>
1 in silicon
2 in galliumarsenide

1 EN 2

EP Impact Ionization Rate Parameter Holes


Type:
Default:
Units:
Range:

<R>
1 in silicon
2 in galliumarsenide

1 EP 2

RECOMBINATION

123

FN Impact Ionization Rate Parameter Electrons


Type:
Default:
Units:
Range:

<R>
2 2 in silicon
0 in galliumarsenide

0 FN 10

FP Impact Ionization Rate Parameter Holes


Type:
Default:
Units:
Range:

<R>
2 2 in silicon
0 in galliumarsenide

0 FP 10

MTUN Model for Band-to-Band Tunneling

Model for tunneling can be specified with: {zx . For the explanation refer to Section 2.6.
Type:
<I>
Default:
0
Range:
0 MTUN 1000

SN Surface Recombination Parameter for Electrons


This key specifies the surface recombination velocity for electrons at the interface between substrate and
gate insulator. This surface recombination is independent of terminal bias and interface trap density. It
r
0 is specified
will be superposed to the surface recombination rate calculated for interface traps. If t
this surface generation-recombination vanishes.
Type:
<R>
Default:
100
1


Units:
Range:
0 SN 107

SP Surface Recombination Parameter for Holes


This key specifies the surface recombination velocity for holes at the interface between substrate and gate
insulator. Additional explanation is given with SN.
Type:
<R>
Default:
100
1
Units:


Range:
0 SP 107

TEXF Exponential Factor for the Band-to-Band Tunneling Current


The exponential factors in the BBT models will be multiplied by TEXF.
Type:
<R>
Default:
1
Units:

Range:
0 1 TEXF 10

CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

124

TLIF Pre-exponential Factor for the Band-to-Band Tunneling Current


The pre-exponential factor in the BBT models will be multiplied by TLIF.
Type:
<R>
Default:
1
Units:

Range:
10 6 TLIF 106

TN Electron Lifetime
Type:
Default:
Units:
Range:

<R>
10 6

10

TN

109

TP Hole Lifetime
Type:
Default:
Units:
Range:

<R>
10 6

10

TP

109

VN Back-Interface Recombination Velocity for Electrons


This key specifies the interface recombination velocity at the bulkinsulator interface for electrons. The
surface recombination is independent of terminal bias and interface trap density. It will be superposed to
the surface recombination rate calculated for the traps placed at the back interface. This key is active only
for SOI simulations.
Type:
<R>
Default:
100
1
Units:

|
Range:
0 VN 107

VP Back-Interface Recombination Velocity for Holes


This key specifies the interface recombination velocity at the bulkinsulator interface for holes. This key
is active only for SOI simulations.
Type:
<R>
Default:
100
1

|
Units:
Range:
0 VP 107

STEP

125

Directive STEP
The STEP directive is used to specify a step sequence for any applied terminal voltage. The input data
will be checked for physical plausibility unless PHYSCK=NO is set on the OPTION directive. Therefore

ePn

eqt

?nPn

?nf

oeqt~}(s

oef}(s

1
1

r
n
r

have to be within the allowed range.

DB Bulk Voltage Increment


This key specifies the incremental bulk voltage. The DB key can be omitted but, if specified, the NB key
must also be given.
Type:
<R>
Default:

Units:
Range:
5 DB
0 00999
0 00999 DB 5

DD Drain Voltage Increment


This key specifies the incremental drain voltage. The DD key can be omitted but, if specified, the ND key
must also be given.
Type:
<R>
Default:

Units:
Range:
5 DD
0 00999
0 00999 DD 5

DG Gate Voltage Increment


This key specifies the incremental gate voltage. The DG key can be omitted but, if specified, the NG key
must also be given. However, if the MODEL=THRES option on the OPTION directive is used, neither the
DG key nor the NG key may appear.
Type:
<R>
Default:

Units:
Range:
5 DG
0 00999
0 00999 DG 5

C1 Lower Value of Impurity Concentration Range


This key specifies the lower value of the impurity concentration range, for which material characteristics
are calcualted by the Monte Carlo method. This key applies only if MONTE-CARLO MCMOD=2 is set.
See also the keydescription of NC.
Type:
<R>
Default:
0
3
Units:

Range:
0 C1 1022

CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

126

C2 Upper Value of Impurity Concentration Range


This key specifies the upper value of the impurity concentration range, for which material characteristics
are calcualted by the Monte Carlo method. This key applies only if MONTE-CARLO MCMOD=2 is set.
See also the keydescription of NC.
Type:
<R>
Default:
0
3
Units:

Range:
0 C2 1022

E1 Lower Value of Electric Field Range

This key specifies the lower value of the electric field range, for which material characteristics are
calcualted by the Monte Carlo method. If E1 0, a special algorithm for zero field strength is invoked.
This key applies only if MONTE-CARLO MCMOD=2 is set. See also the keydescriptions of NE and ELOG.
Type:
<R>
Default:
0
Units:
2 
Range:
0 E1 107

E2 Upper Value of Electric Field Range


This key specifies the upper value of the electric field range, for which material characteristics are
calcualted by the Monte Carlo method. This key applies only if MONTE-CARLO MCMOD=2 is set. See
also the keydescriptions of NE and ELOG.
Type:
<R>
Default:
0
2 
Units:
Range:
500 E2 107

ELOG Logarithmic Field Variation for Material Characteristics


This key applies only if MONTE-CARLO MCMOD=2 is set.
Type:
<L>
Default:
NO

LUDG Change Drain and Gate Voltage Simultaneously


This logical key specifies that the gate voltage is changed simultaneously with the drain voltage, such that
 @ remains constant. The step of both is determined by ND and DD. This key is useful in the analysis of
band-to-band tunneling.
Type:
<L>
Default:
NO
LUGB Change Bulk and Gate Voltage Simultaneously
This logical key specifies that the bulk voltage is changed simultaneously with the gate voltage, such that
 D= remains constant. The step of both is determined by NG and DG. This key is useful in charge-pumping
A@= ) is variable.
simulation when the gate pulse is fixed and the reverse bias (e.g. AC4=
Type:
<L>
Default:
NO

STEP

127

NB Number of Bulk Voltage Steps


This key specifies the number of bulk voltage steps to be performed. The NB key may be omitted, but, if
specified, the DB key must also be given.
Type:
<I>
Default:
1
Range:
1 NB 100

NC Number of Impurity Conctentration Steps


This key applies only if MONTE-CARLO MCMOD=2 is set. Material characteristics are calculated for NC
values of the impurity concentration within the range specified by the C1 and C2 keys.
Type:
<I>
Default:
1
Range:
1 NC 100

ND Number of Drain Voltage Steps


This key specifies the number of drain voltage steps to be performed. The ND key may be omitted, but, if
specified, the DD key must also be given.
Type:
<I>
Default:
1
Range:
1 ND 100

NE Number of Electric Field Steps


This key applies only if MONTE-CARLO MCMOD=2 is set. Material characteristics are calculated for NE
values of the electric field within the range specified by the E1 and E2 keys. Stepping through the field
interval is performed either logarithmic or linearly depending on the ELOG key.
Type:
<I>
Default:
1
Range:
1 NE 100

NG Number of Gate Voltage Steps


This key specifies the number of gate voltage steps to be performed. The NG key may be omitted, but, if
specified, the DG key must also be given.
Type:
<I>
Default:
1
Range:
1 NG 100

VG Small Differential Gate Voltage Increment


This key specifies the differential gate voltage increment. If specified, the calculation will be performed in
two points for UG-VG/2 and UG+VG/2, where UG is the gate voltage. This option can be combined with
the gate voltage step DG and NG. This key can be used to calculate the quasi-static small signal parameters
(e.g. gate QS capacitance from the total gate charge). For this purpose the GRIDFREEZE should be
specified, else the results could be wrong. VG cannot be used when either LUGB or LUDG are specified.
Type:
<R>
Default:

Units:
Range:
2 10 3 VG 1

CHAPTER 3 REFERENCE OF INPUT DIRECTIVES

128

Directive TRANSIENT
The TRANS directive is used to specify the parameters for the transient simulation. The specification
includes type of terminal and waveform applied to this terminal, as well as time and voltage parameters
of the waveform. Different waveforms can be applied on different terminals simultaneously. The detailed
description of the waveforms is given in Section 1.4. If not specified, a constant bias which equals to
steady-state value (cf. BIAS directive) is assumed at the particular terminal.

NTSI Number of Time-Subintervals for the Trap Equation


The dynamics equations for interface traps are solved in time steps which are NTSI times shorter than
time steps used to solve for the basic semiconductor equations. For NTSI=1 (default) the trap-dynamics
equations are solved with time resolution as the basic semiconductor equations (cf. Section 2.7).
Type:
<I>
Default:
1
Range:
1 NTSI 5

T Time Parameters of Terminal Pulses


The meaning of T{z , z 1 2 3 for terminal 8fn(#qt
waveform. A complete explanation is given in Section 1.4.
Type:
<R>

Units:

U Voltage Parameters of Terminal Pulses

The meaning of Uz ,
Type:
Units:
Range:

1 2 3
<R>

zu

20

U
{z

for terminal

08f,n,#t

is explained in Section 1.4.

20

W Waveform Specification

is different for each particular

W specifies the waveform applied for terminal


1.4.
Type:
<R>
Default:
1
Range:
1 constant voltage
2 trapeze pulse
3 three-level pulse

0u8fn(#qt

. A detailed explanation is given in Section

4
Files

4.1 The MINIMOS 2D Files


MINIMOS gains access to its input and output files via logical units. The user has to assign filenames to
those unit numbers, normally done at command level, to specify the files of his choice. The numerical
values of the logical units are coded in the include file CMSYSX.INC.

4.1.1 The Input Files


MINIMOS 2D requires an input control file, also termed input deck.
Unit:
Format:
Usage :
Contents:

NUMINP
formatted
obligatory
MINIMOS directives

Another input file is the doping file which may optionally be given.
Unit:
Format:
Usage :
Contents:

NUMDOP
binary
optional
Doping profile of a device

For a complete overview how to specify an appropriate doping profile see the PROFILE directive
(Page 112). The format of the doping file depends on the setting of the keys ASYM, FILE and XOFF on
the PROFILE directive. One- or two-dimensional doping data may be read depending on the setting of
the key FILE on the PROFILE directive. MINIMOS reads an external doping file in FORTRAN binary
mode.

129

CHAPTER 4 FILES

130

One-dimensional Profiles (FILE=1-D)


FILE=1-D must be specified on the PROFILE directive. The doping profile has to be given in onedimensional profile slices, which are artificially extended to two dimensions. The format of the file
depends on the settings of the keys ASYM and XOFF. If ASYM=YES and/or XOFF is given on the
PROFILE directive, three profile slices are required on the doping file (NSLICE=3). In all other cases
only two profile slices are required (NSLICE=2). One slice consists of MAXDO1 records of doping data.
The parameter MAXDO1 is set in the include file cmos02x.inc. The first slice is a source-drain profile
slice and the second slice is a channel profile slice. In case of NSLICE=3 i.e. when ASYM=YES and/or
XOFF are given the third slice is a second source-drain profile slice.
The first record of the file is expected the following way:
READ (NUMDOP) DUMMY,YINC

YINC is the distance increment for the doping table. The next MAXDO1 records of the doping file are
interpreted as the acceptor and donor concentration and their derivatives (one slice):

DO 1 I=1,MAXDO1
READ (NUMDOP) (AC(I,J),DACDY(I,J),DC(I,J),DDCDY(I,J),J=1,NSLICE)

where the arrays AC, DACDY, DC, DDCDY denote:


AC(I,J)
Acceptor concentration in 

DACDY(I,J)
Spatial derivative of the acceptor concentration in 
DC(I,J)
Donor concentration in 

DDCDY(I,J)
Spatial derivative of the donor concentration in 

The index I denotes the location, where I=1 is the surface and I=MAXDO1 denotes the depth
(MAXDO1-1) YINC. If MINIMOS needs the doping at a deeper distance the bulk value will be assumed.
The index J denotes the specific slice:

J = 1
first source and drain doping slice.
J = 2
channel profile extended homogeneously all over the device.
J = 3
second source and drain profile slice.

4.1 THE MINIMOS 2D FILES

131

In the case NSLICE=3 three different options are admitted:


ASYM=YES and XOFF not given
J=1 denotes the source profile only, J=3 only the drain profile.
ASYM=YES and XOFF given
J=1 denotes the source profile, whereas the drain profile is comprised from both the J=1 profile
slice and the J=3 profile slice with an offset XOFF.
ASYM=NO and XOFF given
J=1 denotes the inner source and drain profile and J=3 the outer source and drain profile, which is
assumed to be shifted with the distance XOFF.
For an LDD application J=1 gives the profile in the lowly doped region, J=3 in the heavily doped region
at offset XOFF.

Twodimensional Profiles (FILE=2-D)


FILE=2-D must be specified on the PROFILE directive, and the external doping file must be a fully
twodimensional doping profile in the following manner:
SUBROUTINE DOP2D (CDON,CACC,C,DCDL,DCDV,PL,PV)
***
**
*
*

**
*
**

DOP2D - SETUP DOPING FROM TWO-DIMENSIONAL DATA


READ (NUMDOP) NPIECE
READ SOURCE PROFILE
DONOR CONCENTRATION
CALL DOP2DR (C,DCDL,DCDV,PL,PV,1,1,NL,NV,NPIECE)
ACCEPTOR CONCENTRATION
CALL DOP2DR (C,DCDL,DCDV,PL,PV,2,1,NL,NV,NPIECE)
IF (ABS(NPIECE).EQ.1) THEN
REWIND NUMDOP
READ (NUMDOP) NPIECE
ENDIF
READ DRAIN PROFILE
DONOR CONCENTRATION
CALL DOP2DR (C,DCDL,DCDV,PL,PV,1,-1,NL,NV,NPIECE)
ACCEPTOR CONCENTRATION
CALL DOP2DR (C,DCDL,DCDV,PL,PV,2,-1,NL,NV,NPIECE)
END

CHAPTER 4 FILES

132

SUBROUTINE DOP2DR (C,DCDL,DCDV,PL,PV,NPASS,INC,NL,NV,IDCDX)


***
*
*
*
*
*
*

*
*

1
*

DOP2DR - READ TWO-DIMENSIONAL DOPING PROFILE


NPASS.EQ. 1 DONOR
CONCENTRATION
NPASS.EQ. 2 ACCEPTOR CONCENTRATION
INC .EQ. 1 SOURCE
PROFILE
INC .EQ.-1 DRAIN
PROFILE
IDCDX.GT. 0
DERIVATIVES ON INPUT
IDCDX.LE. 0 NO DERIVATIVES ON INPUT
IF (NPASS.EQ.1) THEN
SOURCE PROFILE
READ (NUMDOP) NL,NV
IF (INC.EQ.1) THEN
DONOR
JS=1
JE=NL
ELSE
ACCEPTOR
JS=NL
JE=1
ENDIF
READ (NUMDOP) (PL(I),I=JS,JE,INC)
READ (NUMDOP) (PV(I),I=1,NV)
ELSE
DRAIN PROFILE
IF (INC.EQ.1) THEN
DONOR
JS=1
JE=NL
ELSE
ACCEPTOR
JS=NL
JE=1
ENDIF
ENDIF
JJ=0
DO 1 I=1,NV
READ (NUMDOP) (C(JJ+J),J=JS,JE,INC)
JJ=JJ+NL
DERIVATIVES
IF (IDCDX.GT.0) THEN
JJ=0
DO 2 I=1,NV
READ (NUMDOP) (DCDL(JJ+J),J=JS,JE,INC)
JJ=JJ+NL
JJ=0
DO 3 I=1,NV
READ (NUMDOP) (DCDV(JJ+J),J=JS,JE,INC)
JJ=JJ+NL
ENDIF
END

4.1 THE MINIMOS 2D FILES

133

The absolute value of NPIECE has the following effect:


ABS(NPIECE) = 1
The same two-dimensional profile piece is used for source and drain.
ABS(NPIECE) > 1
Two different but matching profile pieces are used for the source and the drain region.
The sign of NPIECE has the following effect:
SIGN(NPIECE) = 1
The spatial derivatives of the two-dimensional profile are expected in the doping file.
SIGN(NPIECE) = -1
No spatial derivatives of the two-dimensional profile are expected in the doping file.
NL and NV are the number of points of the external profile array in lateral (parallel to the interface) and vertical (perpendicular to the interface) direction. NL and NV have to be in the range 10

(NHS,NVS) MAXDO2. The parameter MAXDO2 can be found in the include file cmos02x.inc.
PL(I) and PV(J) are the coordinates of the point (I,J) in centimeters, PL(I) in lateral and PV(J)
in the vertical direction.
There has to be an index K between 1 and NL such that PL(K)=0. This location is interpreted as the mask
edge of the two-dimensional profile. PL(I) less than zero is the free surface and PL(I) greater than
zero denotes the masked surface. There must also be an index L between 1 and NV such that PV(L)=0.
This location is interpreted as the gate-insulator/semiconductor interface. Positive PV(L) denote the
semiconductor area, values at negative PV(L) are ignored.
The following arrays denote the two-dimensional source doping data. The derivatives are only required if
NPIECE > 0 :
CDS(I,J)
The donor concentration for the source profile in  3 .
DCDSDH(I,J)
The partial spatial derivative of the donor concentration with respect to the horizontal direction for
the source profile in  4 .
DCDSDV(I,J)
The partial spatial derivative of the donor concentration with respect to the vertical direction for the
source profile in  4 .
CAS(I,J)
The acceptor concentration for the source profile in  3 .
DCASDH(I,J)
The partial spatial derivative of the acceptor concentration with respect to the horizontal direction
for the source profile in  4 .

CHAPTER 4 FILES

134

DCASDV(I,J)
The partial spatial derivative of the acceptor concentration with respect to the vertical direction for
the source profile in  4 .
The following arrays denote the twodimensional drain doping data. The derivatives are only required if
NPIECE > 0 :
CDD(I,J)
The donor concentration for the drain profile in  3 .
DCDDDH(I,J)
The partial spatial derivative of the donor concentration with respect to the horizontal direction for
the drain profile in  4 .
DCDDDV(I,J)
The partial spatial derivative of the donor concentration with respect to the vertical direction for the
drain profile in  4 .
CAD(I,J)
The acceptor concentration for the drain profile in  3 .
DCADDH(I,J)
The partial spatial derivative of the acceptor concentration with respect to the horizontal direction
for the drain profile in  4 .
DCADDV(I,J)
The partial spatial derivative of the acceptor concentration with respect to the vertical direction for
the drain profile in  4 .
If ABS(NPIECE) > 1, which means that two separate profile pieces for source and drain are given, the
profiles have to match for the middle of the channel. If the simulation domain generated by MINIMOS
is larger than the domains described by the external doping arrays, the profile is simply extended in the
respective direction.
Note: The doping data read from the doping file are not physically verified.

4.1.2 The Main Ouput Files


The Formatted Output File
MINIMOS 2D produces a formatted output file, which contains integral results as well as distributions of
various quantities. The contents of the output file mainly the types of the distributions is controlled by
the OUTPUT directive of the input deck. For further information see Page 109.
Unit:
Format:
Contents:

NUMOUT
formatted
MINIMOS 2D results

4.1 THE MINIMOS 2D FILES

135

The Reference File


During execution MINIMOS provides the user with systematic information how the iterative solution of
the problem is proceeding. These monitoring data can either be assigned to a terminal channel mainly for
interactive execution or to a reference file channel in the case that MINIMOS is run as a batch job. Those
data can be valuable for error logging and convergence tracing. See Section 1.11 to obtain information of
the structure of the reference file.
Unit:
Format:
Contents:

NUMREF
formatted
MINIMOS 2D execution history

The Binary Output File


Optionally MINIMOS produces a binary output file containing all distributed quantities. This file is aimed
mainly to serve as input for postprocessing.
Unit:
Format:
Usage :
Contents:

NUMBIN
binary
optional
MINIMOS 2D binary results

The binary output file is written in FORTRAN binary mode. The main purpose of this file is to store the
output data for subsequent postprocessing e.g. plotting. The length of this file depends on the specification
of the MODEL key (see Page 103). The various distributed quantities are given in modified j units
( ,  , , T ). To enable access to these data by user-written programs at first the successive WRITE
statements of the FORTRAN routine BINOUT are given, to highlight the exact order and type of the data
written.

M s

136

**

**
**
**
**
**
**

**
**
**
**

**

**
**
**

**

**
**
**

CHAPTER 4 FILES

FIRST WRITE TITLE AND SOME PHYSICAL CONSTANTS


WRITE (NUMBIN) HTITLE
WRITE (NUMBIN) CHCOD(J),(DBLE(HLP1(I)),I=1,4)
WRITE BIAS CONDITION
WRITE (NUMBIN) (DBLE(HLP1(I)),I=1,7)
WRITE COUPLE OF PHYSICAL CONSTANTS
WRITE (NUMBIN) DBLE(XYNOR/SQRT(CINTR)),DBLE(UT),DBLE(CINTR*CNOR)
WRITE GRID SIZE COUNTERS
WRITE (NUMBIN) NX,NT,NY,NYOX-1,NYTOT,NXS+1,NXD-1
WRITE LATERAL SPACING
WRITE (NUMBIN) (DBLE(X(I)*XYNOR),I=1,NX)
WRITE TRANSVERSAL SPACING
WRITE (NUMBIN) (DBLE(Y(I)*XYNOR),I=1,NYTOT)
SAVE INFORMATIONS OF INTERFACES
IF (LOXBOD)
WRITE (NUMBIN) NINT,NXR,1,IMDINT
WRITE (NUMBIN) HTITLE
WRITE (NUMBIN) (DBLE(HLP1(I)),I=1,NXR)
WRITE (NUMBIN) HTITLE
WRITE (NUMBIN) (DBLE(HLP2(I)),I=1,NXR)
ENDIF
SAVE NET DOPING CONCENTRATION
WRITE (NUMBIN) (DBLE((CDON(I)-CACC(I))*CNOR),I=1,NXY)
SAVE POTENTIAL DISTRIBUTION
WRITE (NUMBIN) (DBLE((PSI(I)+US)*UTCH),I=1,NXYTOT)
SAVE MINORITY CARRIER CONCENTRATION
WRITE (NUMBIN) (DBLE(CMIN(I)*CNOR),I=1,NXT)
SAVE X/Y-MINORITY MOBILTY
WRITE (NUMBIN) (DBLE(HLP1(I)*DIFNOR),I=1,NXT)
WRITE (NUMBIN) (DBLE(HLP1(I)*DIFNOR),I=1,NXT)
SAVE X/Y-MINORITY CURRENT DENSITY
WRITE (NUMBIN) (DBLE(HLP1(I)*CURNOR),I=1,NXT)
WRITE (NUMBIN) (DBLE(HLP1(I)*CURNOR),I=1,NXT)
FROM NOW ON AVAL MODEL
IF (MODEL.GE.4) THEN
SAVE MAJORITY CARRIER CONCENTRATION
WRITE (NUMBIN) (DBLE(CMAJ(I)*CNOR),I=1,NXT)
SAVE X/Y-MAJORITY MOBILITY
WRITE (NUMBIN) (DBLE(HLP1(I)*DIFNOR),I=1,NXT)
WRITE (NUMBIN) (DBLE(HLP1(I)*DIFNOR),I=1,NXT)
SAVE X/Y-MAJORITY CURRENT DENSITY
WRITE (NUMBIN) (DBLE(-HLP1(I)*CURNOR),I=1,NXT)
WRITE (NUMBIN) (DBLE(-HLP1(I)*CURNOR),I=1,NXT)
SAVE AVALANCHE GENERATION RATE
WRITE (NUMBIN) (DBLE(HLP1(I)*CNOR),I=1,NXT)
FROM NOW ON HOT MODEL OR MONTE-CARLO MODEL
IF (MODEL.GE.5) THEN
SAVE MINORITY/MAJORITY CARRIER TEMPERATURES
WRITE (NUMBIN) (DBLE(HLP1(I)*SIMTEM),I=1,NXT)
WRITE (NUMBIN) (DBLE(HLP1(I)*SIMTEM),I=1,NXT)
ENDIF
ENDIF

4.1 THE MINIMOS 2D FILES


1

HTITLE

CHCOD

3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22

UFB
TINS
XLGEO
XWGEO
US
UG
UD
UB
ID
IS
IB
LI
UT
NI
NX
NT
NY
NYOX-1
NYTOT
NXS+1

23

NXD-1

24
25

X
Y

137

string of 80 characters. The first line of the input file. The last 4 characters of HTITLE
always contain the word -2D. If MINIMOS has calculated in nonplanar mode
the character substring HTITLE(68) to HTITLE(76) equals to NONPLANAR.
Thus, it is easy to determine if MINIMOS calculated planarly or not. Note that the
distinction between both calculation modes is essential, because the structure of the
binary file varies with both modes.
1 character. This character may be N, or P and specifies the channel type of the
simulated transistor (see Page 64).
1 floating point word. The flatband voltage (see Page 103).
1 floating point word. The thickness of the gate insulator (see Page 64).
1 floating point word. The gate length < (see Page 64).
1 floating point word. The channel width (see Page 64).
1 floating point word. Source terminal voltage (see Page 62).
1 floating point word. Gate terminal voltage (see Page 62).
1 floating point word. Drain terminal voltage (see Page 62).
1 floating point word. Bulk terminal voltage (see Page 62).
1 floating point word. Drain terminal current.
1 floating point word. Source terminal current.
1 floating point word. Bulk terminal current.
1 floating point word. Intrinsic Debye length.
1 floating point word. Thermal voltage.
1 floating point word. Intrinsic concentration.
1 integer word. Number of mesh lines in x-direction.
1 integer word. Number of mesh lines of the active semiconductor in y-direction.
1 integer word. Number of mesh lines of the semiconductor in y-direction.
1 integer word. Number of mesh lines in the gate insulator in y-direction.
1 integer word. Total number of mesh lines in y-direction.
1 integer word. NXS denotes the x-index at the end of the source contact (planar
mode).
1 integer word. NXD denotes the x-index at the beginning of the drain contact (planar
mode).
NX floating point words. x-coordinates of the mesh.
NYTOT floating point words. y-coordinates of the mesh. Note that the mesh lines
in y-direction are in the oxide from index 1 to NYOX and in the semiconductor from
index NYOX+1 to NYTOT.
In case of nonplanar calculation the following geometry data will be included:

26

NINT

27
28

NXR
1

29

IMDINT

1 integer word. Number of interfaces stored on this file. Currently 2 interfaces, FOXO
and FOXU, are available.
1 integer word. Number of interface points in lateral (x) direction.
1 integer word. Number of interface points in transversal (z) direction. (1 in
MINIMOS 2D).
1 integer word. The mode how the interfaces are stored on this file. IMDINT=1
means that the interface information is stored on the original mesh, IMDINT=2
means that the interface information is stored on an equidistant mesh. IMDINT=3
covers all other cases. IMDINT=1 is default.

CHAPTER 4 FILES

138

30
31
32
33

HTITLE
FOXO
HTITLE
FOXU

string of 80 characters. Name of the upper interface FOXO.


NXINT NZINT floating point words. Contact (upper) interface.
string of 80 characters. Name of the lower interface FOXU.
NXINT NZINT floating point words. SemiconductorInsulator (lower) interface.
Distributed quantities for both planar and nonplanar geometries.

34
35
36
37

CDOP
PSI
CMIN
MOMINL

38

MOMINT

39
40

JLMIN
JTMIN

NX NY floating point words. Doping concentration.


NX NYTOT floating point words. Electrostatic potential.
NX NT floating point words. Minority carrier concentration.
NX NT floating point words. (Lateral) Minority carrier mobility parallel to the gate
insulator interface.
NX NT floating point words. (Transversal) Minority carrier mobility perpendicular
to the gate insulator interface.
NX NT floating point words. Minority carrier current density (x-component).
NX NT floating point words. Minority carrier current density (y-component).
EOF for MODEL=1-D, MODEL=THRES and MODEL=2-D

41
42

CMAJ
MOMAJL

43

MOMAJT

44
45
46

JLMAJ
JTMAJ
AVAL

NX NT floating point words. Majority carrier concentration.


NX NT floating point words. Majority carrier mobility parallel to the gate insulator
interface.
NX NT floating point words. Majority carrier mobility perpendicular to the gate
insulator interface.
NX NT floating point words. Majority carrier current density (x-component).
NX NT floating point words. Majority carrier current density (y-component).
NX NT floating point words. Avalanche (impact ionization) generation rate.
EOF for MODEL=AVAL.

47
48

TCMIN
TCMAX

NX NT floating point words. Minority carrier temperatures.


NX NT floating point words. Majority carrier temperatures.
EOF for MODEL=HOT and MODEL=MONTE-CARLO

The Formatted PIF Output File


Optionally MINIMOS produces a formatted PIF output file.
Unit:
Format:

NUMPIF
formatted

If PIF=YES is set on the OUTPUT directive an ASCII-PIF file is written, but no formatted output written
to unit NUMOUT.

4.1 THE MINIMOS 2D FILES

139

4.1.3 Auxiliary Output Files


Table of I-V Data
By default MINIMOS produces a formatted file which contains a table of the simulated bias points.
Unit:
Format:
Column
1
2
3
4
5
6
7
8

NUMIVC
formatted
Symbol
A=
A@
A
AC
j8=
j8@
j 
j8C

Description
Bulk voltage (V)
Drain voltage (V)
Gate voltage (V)
Source voltage (V)
Bulk current (A)
Drain current (A)
Gate current (A)
Source current (A)

The Monte Carlo Output File


Distributed Monte Carlo data can be written to a file:
Unit:
Format:

NUMUAI
binary

A quantitiy is appended to the file by calling the subroutine


SUBROUTINE USEOUT (TEXT,A,ASCAL,X,XSCAL,Y,YSCAL,NX,NY)
DIMENSION A(NX,NY),X(NX),Y(NY)
CHARACTER*(*) TEXT
CHARACTER*(80) TITLE
TITLE=TEXT
WRITE (NUMUAI)
WRITE (NUMUAI)
WRITE (NUMUAI)
WRITE (NUMUAI)
WRITE (NUMUAI)

TITLE
NY,NX
(Y(I)*YSCAL,I=1,NY)
(X(I)*XSCAL,I=1,NX)
((A(I+J*NX)*ASCAL,I=1,NX),J=0,NY-1)

END

Quantities can not only be written after termination of the self-consistent iteration loop but also after completion of intermediate iterations. Up to five iteration numbers can be specified by the keys SAV1 SAV5.
For each specified iteration number a set of quantities is appended to the file:

pp

CHAPTER 4 FILES

140

Quantity
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27

Symbol
]
s @@

X`$
s

q


q 
&

6qxs6N#'
&

s6N#'

Hb
H

&

9#9

9#9

#
/

6Ds6N#'


WZ

\
 

6s6x
6!As6N


window
 window

Description
&
Electron concentration from solution of PDEs  3 '
&
Xvelocity from solution of PDEs 5s'
&
Potential update %'
&
Negative xelectric field 1s5'
&
Yelectric field 2s5'
Number of assigned scattering events
Normalization quantity (= carrier conc. in arbitrary units)
&
Xvelocity 5'
&
Yvelocity 5'
&
XX-component of thermal voltage tensor %'
&
YY-component of thermal voltage tensor %'
&
Xmomentum loss rate 1s5'
&
Ymomentum loss rate 1s5'
&
Average energy s%'
&
Impact ionization rate 1 s'
&
Ionization coefficient 1 s5'
&
Magnitude of average velocity 5s'
&
Magnitude of average momentum loss rate 1s5'
&
Mobility from MC  2 m:'
&
Temperature voltage for transport from MC :'
&
Overlay of MC and analytical mobilities  2 m'
&
Overlay of MC and equilibrium   %'
&
Xenergy gradient field 2s5'
&
Yenergy gradient field 2s5'
&
Magnitude of energy gradient field 1s5'
&
Mobility window 1 '
&
Temperature window 1 '

The Transient Output File


In transient mode MINIMOS writes the evolution of terminal currents and voltages to a file.
Unit:
Format:
Column
1
2
3
4
5
6
7
8
9

NUMTRA
formatted
Symbol
t
A=
j =
@
j8@
A
j8
C

j8C

Description
Time (s)
Bulk voltage (V)
Bulk current (A)
Drain voltage (V)
Drain current (A)
Gate voltage (V)
Gate current (A)
Source voltage (V)
Source current (A)

4.1 THE MINIMOS 2D FILES

141

The Gate Charge File


Optionally MINIMOS writes gate charge to a formatted file.
Unit:
Format:

NUMCG1
formatted

The two-dimensional cross-section of the gate contact is assumed to be of rectangular shape. Not only
the total gate charge is written to the file but also the components located a the source-sided edge, at the
interface to the gate-oxide and at the drain-sided edge. The meaning of the columns is:
Column
1
2
3
4
5
6
7
8

Symbol
=

A@
A
AC

U
U

U

/
dW

U $Z

Description
Bulk voltage (V)
Drain voltage (V)
Gate voltage (V)
Source voltage (V)
Total gate charge (As)
Part of gate charge at the source-sided edge (As)
Part of gate charge at the lower edge (As)
Part of gate charge at the drain-sided edge (As)

Poly-Gate Voltage Drop


Optionally MINIMOS writes the voltage drop ocurring in a poly-silicon gate to a formatted file.
Unit:
Format:

NUMCG2
formatted

The voltage drop is evaluated at a coordinate half-way between the source and drain edges of the gate.
The meaning of the columns is:
Column
1
2
3
4
5

Symbol
A=
A@



AC

A

Description
Bulk voltage (V)
Drain voltage (V)
Gate voltage (V)
Source voltage (V)
Voltage drop in the gate (V)

CHAPTER 4 FILES

142

Charge-Pumping Output
In charge-pumping mode additional data regarding the components of the terminal currents is written to
a file.
Unit:
Format:
Column
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19

NUMCP3
binary
Symbol
t
A=
j =
K

j8=A

G
j = Y
@
j8@1
K

j @1

G
j @1 Y
A
j8!
K

j8!

G
j ! Y
AC
j CN
j8Cm

G
j Cm Y
:

64T
K

64T

Description
Time (s)
Bulk voltage (V)
Electron bulk current (A)
Hole bulk current (A)
Displacement bulk current (A)
Drain voltage (V)
Electron drain current (A)
Hole drain current (A)
Displacement drain current (A)
Gate voltage (V)
Electron gate current (A)
Hole gate current (A)
Displacement gate current (A)
Source voltage (V)
Electron source current (A)
Hole source current (A)
Displacement source current (A)
Integrated electron generation rate at front interface (A)
Integrated hole generation rate at front interface (A)

Information on the area actively contributing to the charge-pumping effect is written to a file as well.
Unit:
Format:
Column
1
2
3
4
5
6
7
8
9
10
11

NUMCP2
formatted
Symbol
=

@
A
AC
4G Y

 GV
u G Y

<2

KY


5
qK V

<1

u
$

KY

Description
&
Bulk voltage '
&
Drain voltage %'
&
Gate voltage '
&
Source voltage '
&
Drain sided location where the critical hole concentration occurs 5'
&
Source sided location where the critical hole concentration occurs 5'
&
Distance between location  G Y and drain edge 5'
&
Drain sided location where the critical electron concentration occurs 5'
&
Source sided location where the critical electron concentration occurs 5'
&
Distance between location  KY and drain edge 5'
&
Critical carrier concentration 1 s 3 '

4.1 THE MINIMOS 2D FILES

143

Band-to-Band Tunneling Currents


In band-to-band tunneling mode a formatted file is produced.
Unit:
Format:
Column
1
2
3
4
5
6
7

NUMBBT
formatted
Symbol
A


j = X X
j = X X
j Cm X X
j @1 X X

j = 9#9

G
K
K

Description
Gate voltage (V)
Drain voltage (V)
BBT electron bulk current (A)
BBT hole bulk current (A)
BBT electron source current (A)
BBT electron drain current (A)
Bulk current due to impact ionization (A)

4.1.4 The MINIMOS 3D Files


The Binary Data Link File
If the key M3MODE is given on the OPTION directive, MINIMOS 2D assumes that a subsequent run of
MINIMOS 3D is intended and produces a binary linkfile.
Unit:
Format:

NUMBI3
binary

MINIMOS 3D has no specific formatted input file as MINIMOS 2D. The input deck of the corresponding
two-dimensional example is fully valid. MINIMOS 3D requires a binary input file that serves as a link to
the corresponding two-dimensional solution.
Instead of a binary link file to MINIMOS 2D, MINIMOS 3D allows the use of a previous three-dimensional
solution as initial input. This may avoid repeated calculation of costly three-dimensional Gummel
iterations. The user has to specify an O as second character for the M3MODE key to direct MINIMOS 3D
to read such a file when invoked. Note that MINIMOS 3D does not check the existence of such a file and
crashes if the previous solution file is not found. Note also that MINIMOS 3D produces a binary output
file only if BIN=YES is set on the END directive. Furthermore, MINIMOS 2D has to be invoked first to
preset various common block variables.

The Formatted Output and the Reference Files


Like MINIMOS 2D, MINIMOS 3D produces a formatted output file. For further information see section
Page 109.

CHAPTER 4 FILES

144

Unit:
Format:
Contents:

NUMOUT
formatted
MINIMOS 3D results

The reference file for MINIMOS 3D is fully analogous to the reference file of MINIMOS 2D.
Unit:
Format:
Contents:

NUMREF
formatted
MINIMOS 3D execution history

The Binary Output File


Optionally, i.e. when BIN=YES is specified on the END directive, MINIMOS 3D produces a binary output
file containing all distributed quantities. This file is aimed mainly to serve as input for postprocessing. If
used as a feedback file for another MINIMOS 3D run this file should be renamed or copied appropriately.
Unit:
Format:

NUMBIN
binary

The binary output file is written in FORTRAN binary mode. The main purpose of this file is to store the
output data for subsequent postprocessing. The length of this file depends on the setting of the MODEL
and the M3MODE keys (see Page 103). The various distributed quantities are given in modified j units
( ,  , , T ). The exact order and type of the written data can be found by inspection of the FORTRAN
subroutine BINAT3.

M c

To enable access to these data by user-written programs at first the successive WRITE statements of the
FORTRAN routine BINAT3 are given, to highlight the exact order and type of the data written:
**

**
**
**
**
**
**
**
**

FIRST WRITE TITLE AND SOME PHYSICAL CONSTANTS


WRITE(TITLE(77:80),(A4)) --3D
WRITE (NUMBIN) TITLE
WRITE (NUMBIN) CHCOD(J),(HLP1(I),I=1,4)
WRITE BIAS CONDITION AND CONTACT CURRENTS
WRITE (NUMBIN) (HLP1(I),I=1,7)
WRITE COUPLE OF PHYSICAL CONSTANTS
WRITE (NUMBIN) XYNOR,UT,CNOR
WRITE GRID SIZE COUNTERS
WRITE (NUMBIN) NX,NT,NY,NYOX,NYTOT,NXS,NXG1,NXG2,NXD,NZ,NZSD,NZA,NYFOX
WRITE LATERAL SPACING
WRITE (NUMBIN) (X(I)*XYNOR,I=1,NX)
WRITE TRANSVERSAL-Y SPACING
WRITE (NUMBIN) (Y(I)*XYNOR,I=1,NYTOT)
WRITE TRANSVERSAL-Z SPACING
WRITE (NUMBIN) (Z(I)*XYNOR,I=1,NZ)
SAVE INFORMATIONS OF INTERFACES
WRITE (NUMBIN) NINT,NX,NZ,IMDINT
WRITE INTERFACE 1

4.1 THE MINIMOS 2D FILES

**

**

1
**

2
**

3
**

4
**

6
**

8
*

9
**

10

11
**

12

WRITE (NUMBIN) TITLE


WRITE (NUMBIN) (HLP1(I),I=1,NX*NZ)
WRITE INTERFACE 2
WRITE (NUMBIN) TITLE
WRITE (NUMBIN) (HLP2(I),I=1,NX*NZ)
SAVE DOPING CONCENTRATION
DO 1 IZ=1,NZ
JZ=(IZ-1)*NXYTOT
WRITE (NUMBIN) (HLP1(JZ+I),I=1,NXY)
SAVE POTENTIAL DISTRIBUTION
DO 2 IZ=1,NZ
JZ=(IZ-1)*NXYTOT
WRITE (NUMBIN) ((PSI(JZ+I)+US)*UTCH,I=1,NXYTOT)
SAVE MINORITY CARRIER CONCENTRATION
DO 3 IZ=1,NZ
JZ=(IZ-1)*NXY
WRITE (NUMBIN) (CMIN(JZ+I)*CNOR,I=1,NXY)
SAVE MAJORITY CARRIER CONCENTRATION
DO 4 IZ=1,NZ
JZ=(IZ-1)*NXY
WRITE (NUMBIN) (CMAJ(JZ+I)*CNOR,I=1,NXY)
SAVE X/Y-MINORITY MOBILTY
DO 5 IZ=1,NZ
JZ=(IZ-1)*NXY
WRITE (NUMBIN) (HLP1(JZ+I)*DIFNOR,I=1,NXY)
DO 6 IZ=1,NZ
JZ=(IZ-1)*NXY
WRITE (NUMBIN) (HLP1(JZ+I)*DIFNOR,I=1,NXY)
SAVE X/Y/Z-MINORITY CURRENT DENSITY
DO 7 IZ=1,NZ
JZ=(IZ-1)*NXY
WRITE (NUMBIN) (HLP1(JZ+I)*CURNOR,I=1,NXY)
DO 8 IZ=1,NZ
JZ=(IZ-1)*NXY
WRITE (NUMBIN) (HLP1(JZ+I)*CURNOR,I=1,NXY)
Z - DIRECTION
DO 9 IZ=1,NZ
JZ=(IZ-1)*NXY
WRITE (NUMBIN) (HLP1(JZ+I)*CURNOR,I=1,NXY)
IF (MODEL.GE.4) THEN
SAVE X/Y=MAJORITY MOBILITY
DO 10 IZ=1,NZ
JZ=(IZ-1)*NXY
WRITE (NUMBIN) (HLP1(JZ+I)*DIFNOR,I=1,NXY)
DO 11 IZ=1,NZ
JZ=(IZ-1)*NXY
WRITE (NUMBIN) (HLP1(JZ+I)*DIFNOR,I=1,NXY)
SAVE X/Y/Z-MAJORITY CURRENT DENSITY
DO 12 IZ=1,NZ
JZ=(IZ-1)*NXY
WRITE (NUMBIN) (-HLP1(JZ+I)*CURNOR,I=1,NXY)
DO 13 IZ=1,NZ
JZ=(IZ-1)*NXY

145

CHAPTER 4 FILES

146

13

14
**

15
**

16

17

WRITE (NUMBIN) (-HLP1(JZ+I)*CURNOR,I=1,NXY)


DO 14 IZ=1,NZ
JZ=(IZ-1)*NXY
WRITE (NUMBIN) (-HLP1(JZ+I)*CURNOR,I=1,NXY)
SAVE AVALANCHE GENERATION RATE
DO 15 IZ=1,NZ
JZ=(IZ-1)*NXY
WRITE (NUMBIN) (HLP1(JZ+I)*CNOR,I=1,NXY)
IF (MODEL.GE.5 .AND. M3MODE .GE. 2) THEN
SAVE MINORITY/MAJORITY CARRIER TEMPERATURES
DO 16 IZ=1,NZ
JZ=(IZ-1)*NXY
WRITE (NUMBIN) (HLP1(JZ+I)*SIMTEM,I=1,NXY)
DO 17 IZ=1,NZ
JZ=(IZ-1)*NXY
WRITE (NUMBIN) (HLP1(JZ+I)*SIMTEM,I=1,NXY)
ENDIF
ENDIF

4.1 THE MINIMOS 2D FILES

147

Explanation of the written data:


1

TITLE

CHCOD

3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27

UFB
TINS
XLGEO
XWGEO
US
UG
UD
UB
ID
IS
IB
LI
UT
NI
NX
NT
NY
NYOX
NYTOT
NXS
NXG1
NXG2
NXD
NZ
NZSD

28
29
30
31

NZA
NYFOX
X
Y

32
33

Z
NINT

34
35
36

NX
NZ
IMDINT

string of 80 characters. The first line of the input file. The last 4 characters of TITLE
always contain the word -3D.
1 character. This character may be N, or P and specifies the channel type of the
simulated transistor (see section Page 64)
1 floating point word. The flatband voltage (see section Page 103)
1 floating point word. The thickness of the gate insulator (see section Page 64)
1 floating point word. The gate length L (see section Page 64)
1 floating point word. The channel width W (see section Page 64)
1 floating point word. Source terminal voltage (see section Page 62)
1 floating point word. Gate terminal voltage (see section Page 62)
1 floating point word. Drain terminal voltage (see section Page 62)
1 floating point word. Bulk terminal voltage (see section Page 62)
1 floating point word. Drain terminal current.
1 floating point word. Source terminal current
1 floating point word. Bulk terminal current.
1 floating point word. Intrinsic Debye length.
1 floating point word. Thermal voltage.
1 floating point word. Intrinsic concentration.
1 integer word. Number of mesh lines in xdirection.
1 integer word. Number of mesh lines of the active semiconductor in ydirection.
1 integer word. Number of mesh lines of the semiconductor in ydirection.
1 integer word. Number of mesh lines in the gate insulator in ydirection.
1 integer word. Total number of mesh lines in ydirection.
1 integer word. Xindex of the last source point
1 integer word. X-index of the first gate point (drain side)
1 integer word. X-index of the last gate point (drain side)
1 integer word. Xindex of the first drain point
1 integer word. Number of meshlines in zdirection
1 integer word. Number of meshlines intersecting the source and drain contacts in
zdirection
1 integer word. Number of meshlines od the active semiconductor in zdirection
1 integer word. Yindex of the mesh line inserted at depth OXIDEF
NX floating point words. X-coordinates of the mesh
NYTOT floating point words. Y-coordinates of the mesh. Note that the mesh lines in
ydirection are in the oxide from index 1 to NYOX and in the semiconductor from
index NYOX+1 to NYTOT.
NZ floating point words. Z-coordinates of the mesh
1 integer word. Number of interfaces stored on this file. Currently 2 interfaces, FOXO
and FOXU, are available.
1 integer word. Number of interface points in lateral (x) direction.
1 integer word. Number of interface points in transversal (z) direction
1 integer word. The mode how the interfaces are stored on this file. IMDINT=1
means that the interface information is stored on the original mesh, IMDINT=2
means that the interface information is stored on an equidistant mesh. IMDINT=3
covers all other cases. IMDINT=1 is default.

CHAPTER 4 FILES

148

37
38
39
40
41
42
43
44
45

TITLE
FOXO
TITLE
FOXU
CDOP
PSI
CMIN
CMAJ
MOMINL

46

MOMINT

47
48
49

JLMIN
JTMIN
JZMIN

string of 80 characters. Name of the upper interface FOXO.


NX NZ floating point words. Contact (upper) interface.
string of 80 characters. Name of the lower interface FOXU.
NX NZ floating point words. Semiconductorinsulator (lower) interface.
NX NY NZ floating point words. Doping concentration
NX NYTOT NZ floating point words. Electrostatic Potential
NX NT NZ floating point words. Minority carrier concentration
NX NT NZ floating point words. Majority carrier concentration
NX NT NZ floating point words. (Lateral) Minority carrier mobility parallel to the
gate insulator interface
NX NT NZ floating point words. (Transversal) Minority carrier mobility perpendicular to the gate insulator interface.
NX NT NZ floating point words. Minority carrier current density (x-component)
NX NT NZ floating point words. Minority carrier current density (y-component)
NX NT NZ floating point words. Minority carrier current density (z-component)
EOF for MODEL=2-D

50

MOMAJL

51

MOMAJT

52
53
54
55

JXMAJ
JYMAJ
JZMAJ
AVAL

NX NT NZ floating point words. Majority carrier mobility parallel to the gate insulator
interface
NX NT NZ floating point words. Majority carrier mobility perpendicular to the gate
insulator interface
NX NT NZ floating point words. Majority carrier current density (x-component)
NX NT NZ floating point words. Majority carrier current density (y-component)
NX NT NZ floating point words. Majority carrier current density (z-component)
NX NT NZ floating point words. Avalanche (impact ionization) generation rate.
EOF for MODEL=AVAL

56
57

TCMIN
TCMAX

NX NT NZ floating point words. Minority carrier temperatures.


NX NT NZ floating point words. Majority carrier temperatures.
EOF for MODEL=HOT

Bibliography

[1] O. Heinreichsberger. Transiente Simulation von Silizium-MOSFETs. Dissertation, Technische Universitat Wien, 1992.
[2] P. Habas. Analysis of Physical Effects in Small Silicon MOS Devices. Dissertation, Technische
Universitat Wien, 1993.
[3] S. Bandyopadhyay, M.E. Klausmeier-Brown, C.M. Maziar, S. Datta, and M.S. Lundstrom. A Rigorous Technique to Couple Monte Carlo and Drift-Diffusion Models for Computationally Efficient
Device Simulation. IEEE Trans.Electron Devices, ED-34(2):392399, 1987.
[4] H. Kosina and S. Selberherr. A Hybrid Device Simulator that Combines Monte Carlo and DriftDiffusion Analysis. IEEE Trans.Computer-Aided Design, 13(2):201210, 1994.
[5] J.O. Nylander, F. Masszi, S. Selberherr, and S. Berg. Computer Simulations of Schottky Contacts
with a Non-Constant Recombination Velocity. Solid-State Electronics, 32(5):363367, 1989.
[6] Ph. Lindorfer and S. Selberherr. MESFET Analysis with MINIMOS. In Heuberger et al. Your
command was ignored. Type I <command> <return> to replace it with another command,or
<return> to continue without it. [75], pages 9296.
[7] S. Selberherr. MOS Device Modeling at 77K. IEEE Trans.Electron Devices, 36(8):14641474,
1989.
[8] S. Selberherr, W. Hansch, M. Seavey, and J. Slotboom. The Evolution of the MINIMOS Mobility
Model. Solid-State Electronics, 33(11):14251436, 1990.
[9] D.M. Caughey and R.E. Thomas. Carrier Mobilities in Silicon Empirically Related to Doping and
Field. Proc.IEEE, 52:21922193, 1967.
[10] J. Xu and M. Shur. Velocity-Field Dependence in GaAs. IEEE Trans.Electron Devices, ED34(8):18311832, 1987.
[11] M. Hirose, J. Yoshida, and N. Toyoda. An Improved Two-Dimensional Simulation Model (MEGA)
for GaAs MESFET Applicable to LSI Design. IEEE Trans.Computer-Aided Design, 7(2):225230,
1988.
[12] J.W. Slotboom, G. Streutker, G.J.T. Davids, and P.B. Hartog. Surface Impact Ionization in Silicon
Devices. In Proc.Int.Electron Devices Meeting, pages 494497, 1987.
149

150

BIBLIOGRAPHY

[13] T. Thurgate and N. Chan. An Impact Ionization Model for Two-Dimensional Device Simulation.
IEEE J.Solid-State Circuits, SC-20(1):320324, 1985.
[14] W. Hansch, M. Orlowski, and W. Weber. The Hot-Electron Problem in Submicron MOSFET. In
Nougier and Gasquet Your command was ignored. Type I <command> <return> to replace it
with another command,or <return> to continue without it. [76], pages 597606.
[15] S. Banerjee, D. Coleman, W. Richardson, and A. Shah. Leakage Mechanisms in the Trench Transistor
DRAM Cell. IEEE Trans.Electron Devices, 35(1):108116, 1988.
[16] T. Ozaki, A. Nitayama, T. Hamamoto, K. Sunouchi, and F. Horiguchi. Analysis of Band-toBand Tunneling Leakage Current in Trench-Capacitor DRAM Cells. IEEE Electron Device Lett.,
12(3):9597, 1991.
[17] T. Hori. Drain-Structure Design for Reduced Band-to-Band and Band-to-Defect Tunneling Leakage.
In VLSITECH90 Your command was ignored. Type I <command> <return> to replace it with
another command,or <return> to continue without it. [77], pages 6970.
[18] C. Chang and J. Lien. Corner-Field Induced Drain Leakage in Thin Oxide MOSFET.
Proc.Int.Electron Devices Meeting, pages 714717, 1987.

In

[19] Y. Igura, H. Matsuoka, and E. Takeda. New Device Degradation Due to cold Carriers Created by
Band-to-Band Tunneling. IEEE Electron Device Lett., 10(5):227229, 1989.
[20] G.Q. Lo, A.B. Joshi, and D.-L. Kwong. Hot-Carrier-Stress Effects on Gate-Induced Drain Leakage
Current in n-channel MOSFETs. IEEE Electron Device Lett., 12(1):57, January 1991.
[21] T. Tsuchiya. A New Enhanced Degradation Phenomenon in MOSFETs Under AC Stress: The Effect
of Band-to-Band Tunneling. In Symposium on VLSI Technology, pages 7980, Kyoto, Japan, 1989.
[22] I.-C. Chen and C.W. Teng. A Quantitative Physical Model for the Band-to-Band Tunneling-Induced
Substrate Hot Electron Injection in MOS Devices. IEEE Trans.Electron Devices, 39(7):16461651,
1992.
[23] Y. Tada, C. Nagata, and M. Iwahashi. Band-to-Band Tunneling Current as a Probe for Hot Carrier
Effects. In Ext. Abstr. of the 22nd Conf. on Solid State Devices and Materials, pages 315318,
Sendai, 1990.
[24] C. Duvvury, D.J. Redwine, and H.J. Stiegler. Leakage Current Degradation in N-MOSFETs Due to
Hot-Electron Stress. IEEE Electron Device Lett., 9(11):579581, 1988.
[25] A. Acovic, M. Dutoit, and M. Ilegems. Characterization of Hot-Electron-Stressed MOSFETs
by Low-Temperature Measurements of the Drain Tunnel Current. IEEE Trans.Electron Devices,
37(6):14671476, 1979.
[26] H.S. Momose, S. Kambayashi, H. Kawaguchi, I. Mizushima, S. Onga, and J. Matsunaga. A Bandto-Band Tunneling MOSFET using a Thin Film Transistor. In Ext. Abstr. of the 22nd Conf. on Solid
State Devices and Materials, pages 365368, Sendai, 1990.
[27] E. Takeda, H. Matsuoka, Y. Igura, and S. Asai. A Band-to-band Tunneling MOS Device (B2 TMOSFET) A Kind of Si Quantum Device . In Proc.Int.Electron Devices Meeting, pages
402405, 1988.

BIBLIOGRAPHY

151

[28] Y. Odake, K. Kurimota, and S. Odanaka. Three-Dimensional Numerical Modeling of the Indirect
Band-to-Band Tunneling in MOSFETs. In Ext. Abstr. of the 22nd Conf. on Solid State Devices and
Materials, pages 131134, Sendai, 1990.
[29] M. Orlowski, S.W. Sun, P. Blakey, and R. Subrahmanyan. Two-Dimensional Simulation of Band-toBand Tunneling in an LDD-MOSFET: Explanation of Experimental Results and Prediction of New
Phenomena. In VLSITECH90 Your command was ignored. Type I <command> <return> to
replace it with another command,or <return> to continue without it. [77], pages 6768.
[30] I. Nedev, A. Asenov, and E. Stefanov. Experimental Study and Modeling of Band-to-Band Tunneling
Leakage Current in Thin-Oxide MOSFETs. Solid-State Electronics, 34(12):14011408, 1991.
[31] P. Habas, A. Lugbauer, and S. Selberherr. Two-Dimensional Numerical Modeling of Interband
Tunneling Accounting for Nonuniform Electric Field. In Workshop on Numerical Modeling of
Processes and Devices for Integrated Circuits NUPAD IV, pages 135140, Seattle, 1992.
[32] T. Endoh, R. Shirota, M. Momodomi, and F. Masuoka. An Accurate Model of Subbreakdown Due
to Band-to-Band Tunneling and Some Applications. IEEE Trans.Electron Devices, 37(1):290296,
1990.
[33] G. Nanz, P. Dickinger, and S. Selberherr. Calculation of Contact Currents in Device Simulation.
IEEE Trans.Computer-Aided Design, 11(1):128136, 1992.
[34] L.V. Keldysh. Behavior of Non-Metallic Crystals in Strong Electric Field.
6(33)(4):763770, 1958.

Sov.Phys.JETP,

[35] E.O. Kane. Zener Tunneling in Semiconductors. J.Phys.Chem.Solids, 12:181188, 1959.


[36] E.O. Kane. Theory of Tunneling. Journal of Applied Physics, 32(1):8391, 1961.
[37] L.V. Keldysh. Influence of the Lattice Vibrations of a Crystal on the Production of Electron-Hole
Pairs in a Strong Electrical Field. Sov.Phys.JETP, 34(7)(4):665669, 1958.
[38] A. Schenk. Rigorous Theory and Simplified Model of the Band-to-Band Tunneling in Silicon.
Solid-State Electronics, 36(1):1934, 1993.
[39] C. Jacoboni and L. Reggiani. The Monte Carlo Method for the Solution of Charge Transport in
Semiconductors with Applications to Covalent Materials. Rev.Mod.Phys., 55(3):645705, 1983.
[40] O. Heinreichsberger, P. Habas, and S. Selberherr. Analysis of Geometric Charge-Pumping Components in a Thin-Film SOI-Device. In Maes et al. Your command was ignored. Type I <command>
<return> to replace it with another command,or <return> to continue without it. [78], pages
819822.
[41] P. Habas, O. Heinreichsberger, and S. Selberherr. Analysis of the Degradation of n-Channel LDD
MOSFETs by Numerical Simulation of the Charge-Pumping Experiment. In Proc.ICSICT, pages
691693, Beijing, China, 1992.
[42] P. Grubmair, P. Habas, O. Heinreichsberger, H. Kosina, C. Sala, and S. Selberherr. Recent Advances
in Device Simulation at the TU-Vienna. In Annual Semiconductor Conf. CAS 92, Sinaia, Romania,
October 1992.

BIBLIOGRAPHY

152

[43] P. Habas and S. Selberherr. A Closed-Loop Extraction of the Spatial Distribution of Interface Traps
Based on Numerical Model of the Charge-Pumping Experiment. In Ext. Abstr. of the Int. Conf. on
Solid State Devices and Materials, pages 170172, Tsukuba, 1992.
[44] T. Ouisse, S. Cristoloveanu, T. Elewa, H. Haddara, G. Borel, and D.E. Ioannou. Adaptation of
the Charge Pumping Technique to Gated p-i-n Diodes Fabricated on Silicon on Insulator. IEEE
Trans.Electron Devices, 38(6):14321444, 1991.
[45] P. Habas, O. Heinreichsberger, and S. Selberherr. Transient Two-Dimensional Numerical Analysis of
the Charge-Pumping Experiment. In Maes et al. Your command was ignored. Type I <command>
<return> to replace it with another command,or <return> to continue without it. [78], pages
687690.
[46] M.M. Mandurah, K.C. Saraswat, and C.R. Helms. Dopant Segregation in Polycrystalline Silicon.
Journal of Applied Physics, 51(11):57555763, 1980.
[47] J.Y.-C. Sun, R. Angelucci, C.Y. Wong, G. Scilla, and E. Landi. Rapid Thermal Processing of
Arsenic-Implanted Polysilicon on Very Thin Oxide. In Nougier and Gasquet Your command was
ignored. Type I <command> <return> to replace it with another command,or <return> to
continue without it. [76], pages 401404.
[48] C.L. Chu, G. Chin, K.C. Saraswat, S.S. Wong, and R. Dutton. Technology Limitations for N /P
Polycide Gate CMOS due to Lateral Dopant Diffusion in Silicide/Polysilicon Layers. IEEE Electron
Device Lett., 12(12):696698, 1991.
[49] N.G. Nilsson. Empirical Approximations for the Fermi Energy in a Semiconductor with Parabolic
Bands. Appl. Phys. Lett., 33(7):653654, 1978.
[50] X. Aymerich-Humett, F. Serra-Mestres, and J. Millan. An Analytical Approximation for the Fermi&
Dirac Integral 3 ; 2 ' . Solid-State Electronics, 24(10):981982, 1981.
[51] P. Habas and J.V. Faricelli. Investigation of the Physical Modeling of the Gate-Depletion Effect.
IEEE Trans.Electron Devices, 39(6):14961500, 1992.
[52] P. Habas and S. Selberherr. On the Effect of Non-Degenerate Doping of Polysilicon Gate in Thin
Oxide MOS-Devices Analytical Modeling. Solid-State Electronics, 33(12):15391544, 1990.
[53] P. Habas and S. Selberherr. Numerical Simulation of MOS-Devices with Non-Degenerate Gate. In
Heuberger et al. Your command was ignored. Type I <command> <return> to replace it with
another command,or <return> to continue without it. [75], pages 161164.
[54] G. Yaron and D. Frohman-Bentchkowsky. Capacitance Voltage Characterization of Poly
Structures. Solid-State Electronics, 23:433439, 1980.

M
2

[55] N. Lifshitz and S. Luryi. Influence of a Resistive Sublayer at the Polysilicon/Silicon Dioxide Interface
on MOS Properties. IEEE Trans.Electron Devices, ED-30(7):833836, 1983.
[56] T.W. Hickmott and R.D. Isaac. Barrier Heights at the Polycrystalline Silicon-SiO2 Interface. Journal
of Applied Physics, 52(5):34643475, 1981.
[57] Y.-S. Kim and J.G. Fossum. Physical DMOST Modeling for High-Voltage IC CAD. IEEE
Trans.Electron Devices, 37(3):797803, 1990.

BIBLIOGRAPHY

153

[58] P. Habas. A Physically Based Analytical MOSFET Model with Accurate Field Dependent Mobility.
Solid-State Electronics, 33(7):923933, 1990.
[59] J. Rebollo, E. Figueras, J. Millan, E. Lora-Tamayo, and F. Serra-Mestres. Analysis of the QuasiSaturation Region of High Voltage VDMOS Devices. Solid-State Electronics, 30(2):177180, 1987.
[60] M.N. Darwish. Study of the Quasi-Saturation Effect in VDMOS Transistors. IEEE Trans.Electron
Devices, ED-33(11):17101716, 1986.
[61] J. Paredes, S. Hidalgo, F. Berta, J. Fernandez, J. Rebollo, and J. Millan. A Steady-State VDMOS
Transistor Model. IEEE Trans.Electron Devices, 39(3):712719, 1992.
[62] C. Hu, M.-H. Chi, and V.M. Patel. Optimum Design of Power MOSFETs. IEEE Trans.Electron
Devices, ED-31(12):16931700, 1984.
[63] K. Board, D.J. Byrne, and M.S. Towers. The Optimization of On-Resistance in Vertical DMOS
Power Devices with Linear and Hexagonal Surface Geometries. IEEE Trans.Electron Devices,
ED-31(1):7580, 1984.
[64] J. Fernandez, S. Hidalgo, J. Paredes, F. Berta, J. Rebollo, J. Millan, and F. Serra-Mestres. An
ON-Resistance Closed Form for VDMOS Devices. IEEE Electron Device Lett., 10(5):212215,
1989.
[65] A.F. Franz and G.A. Franz. BAMBIA Design Model for Power MOSFETs. IEEE Trans.ComputerAided Design, CAD-4(3):177189, 1985.
[66] S. Selberherr. Analysis and Simulation of Semiconductor Devices. Springer, Wien, 1984.
[67] M.S. Mock. Analysis of Mathematical Models of Semiconductor Devices. Boole Press, Dublin,
1983.
[68] P. Habas, D. Pantic, Z. Prijic, and N. Stojadinovic. Technology CAD of Low-Voltage Power
VDMOSFET. In Proc. CAS94, October 1994.
[69] B.K. Ridley. Reconciliation of the Conwell-Weisskopf and Brooks-Herring Formulae for ChargedImpurity Scattering in Semiconductors: Third-Body Interference. J.Phys.C, 10:15891593, 1977.
[70] T.G. Van de Roer and F.P. Widdershoven. Ionized Impurity Scattering in Monte Carlo Calculations.
Journal of Applied Physics, 59(3):813815, 1986.
[71] C. Hao, J. Zimmermann, M. Charef, R. Fauquembergue, and E. Constant. Monte Carlo Study of TwoDimensional Electron Gas Transport in Si-MOS Devices. Solid-State Electronics, 28(8):733740,
1985.
[72] L.V. Keldysh. Concerning the Theory of Impact Ionization in Semiconductors. Sov.Phys.JETP,
21:11351144, 1965.
[73] R. VanOverstraeten and H.J. DeMan. Measurement of the Ionization Rates in Diffused Silicon p-n
Junctions. Solid-State Electronics, 13:583608, 1970.
[74] A. Phillips and P.J. Price. Monte Carlo Calculations on Hot Electron Energy Tails. Appl. Phys. Lett.,
30(10):528530, 1977.

154

BIBLIOGRAPHY

[75] A. Heuberger, H. Ryssel, and P. Lange, editors. Proc. 19th European Solid State Device Research
Conference - ESSDERC89, Berlin, 1989. Springer.
[76] J.-P. Nougier and D. Gasquet, editors. Proc. 18th European Solid State Device Research Conference
- ESSDERC 88, volume 49 of Journal de Physique, Les Ulis Cedex, France, 1988. les e ditions de
physique.
[77] Proc. Symposium on VLSI Technology, Honolulu, 1990.
[78] H.E. Maes, R.P. Mertens, and R.J. Van Overstraeten, editors. Proc. 22nd European Solid State
Device Research Conference - ESSDERC92, volume 19 of Microelectr.Eng., Leuven, Belgium,
1992. Elsevier.

Index

Abbreviation
Simulator Key
3C, 69
3G, 70
3H, 70
3U, 70
AC, 83, 103, 112
AK, 71, 83, 113
AL, 109
AS, 113
AV, 109
B, 64, 68
BE, 73
C, 64, 68
CA, 68
CI, 85
CNT, 96
CU, 103
D, 64
D1, 73
D2, 73
DC, 73
DD, 73
DEF, 97
DEPS, 97
DFM, 97
DI, 109
DLS, 97
DM, 114
DO, 71, 83, 97, 114
DR, 73
DTK , 96
E, 68
E1S, 97
E2S, 97
EL, 71, 84, 109, 115

EMU, 97
EPH , 98
ERE, 98
ET, 109
EXI, 98
EXS, 98
F, 64
FI, 115
FS, 87
G, 65
G1, 74
G2, 74
GC, 74
GCHC, 104
GFIL, 80
GR, 104
I, 65
ID, 104
IN, 104
JC, 62
JL, 110
JT, 110
K, 65
LCP, 105
LD, 105
LF, 115
LS, 105
LVDM, 105
M, 66
M3, 106
MA, 110
MCMO, 98
MCP, 99
MENE, 89
METR, 93
MI, 110

155

INDEX

156

ML, 99
MO, 105, 110
MP, 105
MT, 99, 110
MXT, 80
MYT, 80
NA, 69
NC, 69
NG, 69
NH, 69
NO, 110
NONP, 99
NS, 86
NU, 69
NUM , 100
OM, 106
OX, 74, 110
PA, 69
PD, 106
PF, 88
PH, 106, 110
PI, 110
PN, 87
PS, 86, 110
R1, 74
R2, 75
RB, 107
RD, 107
RE, 74
RH, 111
RS, 107
S, 66
S1, 75
S2, 75
SAV , 100
SC, 75
SD, 75
SI, 86
SM, 116
SO, 100
SR, 75
ST, 111
SU, 116
SXT, 80
SYT, 81
T, 66, 69
T1, 76

T2, 76
TE, 71, 84, 107, 117
TG, 75
TI, 72, 84, 117
TO, 117
TRAC, 100
UF, 107
UTM, 100
VSO, 101
WBI, 101
WD, 101
WEI, 101
WI, 101
WS, 101
XI, 86
XL, 66
XO, 117
XR, 66
XW, 102
Y, 67
YM, 76
YO, 118
YW, 102
Simulator Directive
BIAS, 62
DEVICE, 64
END, 68
FIMPLANT, 71
GEOMETRY, 73
GRID, 77
IMPLANT, 83
INTERFACE, 85
MOBILITY, 94
MONTE-CARLO, 96
OPTION, 103
OUTPUT, 109
PROFILE, 112
RECOMBINATION, 120
STEP, 125
TRANSIENT, 128
Simulator Key
3CYCLES, 69
3GEN, 70
3HOT, 70
3UPDATES, 70
Abbreviation

INDEX

157

3C, 69
3G, 70
3H, 70
3U, 70
AC, 83, 103, 112
AK, 71, 83, 113
AL, 109
AS, 113
AV, 109
B, 64, 68
BE, 73
C, 64, 68
CA, 68
CI, 85
CNT, 96
CU, 103
D, 64
D1, 73
D2, 73
DC, 73
DD, 73
DEF, 97
DEPS, 97
DFM, 97
DI, 109
DLS, 97
DM, 114
DO, 71, 83, 97, 114
DR, 73
DTK , 96
E, 68
E1S, 97
E2S, 97
EL, 71, 84, 109, 115
EMU, 97
EPH , 98
ERE, 98
ET, 109
EXI, 98
EXS, 98
F, 64
FI, 115
FS, 87
G, 65
G1, 74
G2, 74
GC, 74

GCHC, 104
GFIL, 80
GR, 104
I, 65
ID, 104
IN, 104
JC, 62
JL, 110
JT, 110
K, 65
LCP, 105
LD, 105
LF, 115
LS, 105
LVDM, 105
M, 66
M3, 106
MA, 110
MCMO, 98
MCP, 99
MENE, 89
METR, 93
MI, 110
ML, 99
MO, 105, 110
MP, 105
MT, 99, 110
MXT, 80
MYT, 80
NA, 69
NC, 69
NG, 69
NH, 69
NO, 110
NONP, 99
NS, 86
NU, 69
NUM , 100
OM, 106
OX, 74, 110
PA, 69
PD, 106
PF, 88
PH, 106, 110
PI, 110
PN, 87
PS, 86, 110

INDEX

158

R1, 74
R2, 75
RB, 107
RD, 107
RE, 74
RH, 111
RS, 107
S, 66
S1, 75
S2, 75
SAV , 100
SC, 75
SD, 75
SI, 86
SM, 116
SO, 100
SR, 75
ST, 111
SU, 116
SXT, 80
SYT, 81
T, 66, 69
T1, 76
T2, 76
TE, 71, 84, 107, 117
TG, 75
TI, 72, 84, 117
TO, 117
TRAC, 100
UF, 107
UTM, 100
VSO, 101
WBI, 101
WD, 101
WEI, 101
WI, 101
WS, 101
XI, 86
XL, 66
XO, 117
XR, 66
XW, 102
Y, 67
YM, 76
YO, 118
YW, 102
ACCELERATION, 103

ACTIVATION, 83, 112


AGLD, 91
AGLS, 91
AGLX, 91
AGUD1, 90
AGUD2, 90
AGUS1, 90
AGUS2, 90
AGUX1, 90
AGUX2, 90
AKEV, 71, 83, 113
ALL, 109
AN, 120
AP, 120
ARUD1, 89
ARUD2, 89
ARUE1, 89
ARUE2, 89
ARUS1, 89
ARUS2, 89
AS, 109
ASYM, 113
AUGD, 90
AUGS, 91
AUGX, 91
AULD, 90
AUUD, 89
AUXD, 96
AUXS, 96
AVAL, 109
BAUUD, 91
BDR, 96
BDUUD, 91
BEAKL, 73
BI, 109
BIAS
JCAP, 62
UB, 63
UD, 63
UG, 63
US, 63
VI, 63
VR, 63
BIN, 68
BN, 120
BP, 120
BULK, 64

INDEX

C1, 114, 125


C2, 114, 126
CA, 113
CACCREL, 68
CAN, 120
CAP, 121
CAPERR, 68
CC, 109
CCLP, 103
CD, 113
CDN, 121
CDP, 121
CHANNEL, 64
CISS, 85
CN, 121
CNTMIN, 96
CP, 103, 121
CSEA, 121
CSED, 122
CSHA, 122
CSHD, 122
CURRENT, 103
D1GAP, 73
D2GAP, 73
DA0, 88
DB, 125
DC, 109
DCGRD, 73
DD, 125
DD0, 88
DDREOX, 73
DEFPOT, 97
DEPSS, 97
DEVICE
BULK, 64
CHANNEL, 64
DGAP, 64
E1, 64
E2, 64
FILM, 64
GATE, 65
INS, 65
KBULK, 65
L, 65
MAT, 66
SGAP, 66
TINS, 66

159

W, 66
XLEFT, 66
XRIGHT, 66
YDEPTH, 67
DFMIN, 97
DG, 125
DGAP, 64
DGLD, 93
DGLS, 93
DGLX, 93
DGMI, 77
DGUD, 92
DGUS, 92
DGUX, 92
DIGITS, 109
DJRX, 77
DJRY, 77
DLSS, 97
DMOFF, 114
DN, 122
DOPX, 77
DOPY, 77
DOPZ, 77
DOSE, 71, 83, 114
DOVL, 97
DP, 122
DP1D, 79
DPXC, 78
DPXD, 78
DPXP, 78
DPYC, 78
DPYD, 78
DPYP, 78
DPZ, 79
DRGRD, 73
1 2 3 4 5 6, 96
DTK ,
DUGD, 92
DUGS, 92
DUGX, 92
DULD, 92
DUUD, 92
DX1, 79
DX2, 79
DX3, 79
DY, 79
DZ, 80
E1, 64, 126

INDEX

160

E1SELF, 97
E2, 64, 126
E2SELF, 97
EA, 103, 114
ED, 104, 114
ELAT, 109
ELEM, 71, 84, 115
ELOG, 126
EMULT, 97
EN, 122
END
3CYCLES, 69
3GEN, 70
3HOT, 70
3UPDATES, 70
BIN, 68
CACCREL, 68
CAPERR, 68
ERROR, 68
NAC, 69
NCYCLES, 69
NGEN, 69
NHOT, 69
NUPDATES, 69
PACCREL, 69
TCERR, 69
EP, 122
1 2 3 4 5 6, 98
EPH ,
EREF, 98
ERROR, 68
ETRAN, 109
EXII, 98
EXSS, 98
FILE, 115
FILM, 64
FIMPLANT
AKEV, 71
DOSE, 71
ELEM, 71
TEMP, 71
TIME, 72
FN, 123
FP, 123
FSS, 87
G1CGRD, 74
G2CGRD, 74
GATE, 65

GCHCALC, 104
GCTHE, 74
GEOMETRY
BEAKL, 73
D1GAP, 73
D2GAP, 73
DCGRD, 73
DDREOX, 73
DRGRD, 73
G1CGRD, 74
G2CGRD, 74
GCTHE, 74
LD, 74
LS, 74
OXIDEF, 74
R1GRD, 74
R2GRD, 75
RECESS, 74
S1GAP, 75
S2GAP, 75
SCGRD, 75
SDREOX, 75
SRGRD, 75
T1GRD, 76
T2GRD, 76
TD, 75
TGATE, 75
TS, 76
YMDEP, 76
GFILL, 80
GGMI, 80
GRID
DGMI, 77
DJRX, 77
DJRY, 77
DOPX, 77
DOPY, 77
DOPZ, 77
DP1D, 79
DPXC, 78
DPXD, 78
DPXP, 78
DPYC, 78
DPYD, 78
DPYP, 78
DPZ, 79
DX1, 79

INDEX

DX2, 79
DX3, 79
DY, 79
DZ, 80
GFILL, 80
GGMI, 80
MXTun, 80
MYTun, 80
PGMI, 80
SXTun, 80
SYTun, 81
XE1, 81
XE2, 81
XE3, 81
XS1, 81
XS2, 81
XS3, 82
YE, 82
YS, 82
ZE, 82
ZS, 82
GRIDFREEZE, 104
IDEALITY, 104
IMPLANT
ACTIVATION, 83
AKEV, 83
DOSE, 83
ELEM, 84
LD, 84
TEMP, 84
TIME, 84
INS, 65
INTERFACE
AGLD, 91
AGLS, 91
AGLX, 91
AGUD1, 90
AGUD2, 90
AGUS1, 90
AGUS2, 90
AGUX1, 90
AGUX2, 90
ARUD1, 89
ARUD2, 89
ARUE1, 89
ARUE2, 89
ARUS1, 89

161

ARUS2, 89
AUGD, 90
AUGS, 91
AUGX, 91
AULD, 90
AUUD, 89
BAUUD, 91
BDUUD, 91
CISS, 85
DA0, 88
DD0, 88
DGLD, 93
DGLS, 93
DGLX, 93
DGUD, 92
DGUS, 92
DGUX, 92
DUGD, 92
DUGS, 92
DUGX, 92
DULD, 92
DUUD, 92
FSS, 87
MENERG, 89
METRAP, 93
NRD, 85
NRE, 85
NRS, 85
NSS, 86
PFSS, 88
PNSS, 87
PSS, 86
SISS, 86
XISS, 86
INTRINSIC, 104
IV, 109
JCAP, 62
JLAT, 110
JTRAN, 110
KBULK, 65
L, 65
LCPUMP, 105
LD, 74, 84, 115
LDRAIN, 105
LFIT, 115
LS, 74
LSOURCE, 105

INDEX

162

LUDG, 126
LUGB, 126
LVDMOSFET, 105
M3MODE, 106
MAJ, 110
MAT, 66
MB, 94
MC, 94
MCMOD, 98
MCPIT, 99
MENERG, 89
METRAP, 93
MI, 94
MIN, 110
ML, 94
MLONG, 99
MOB, 110
MOBILITY
MB, 94
MC, 94
MI, 94
ML, 94
MQ, 95
MR, 95
MS, 95
MT, 95
MV, 95
MODEL, 105
MONTE-CARLO
AUXD, 96
AUXS, 96
BDR, 96
CNTMIN, 96
DEFPOT, 97
DEPSS, 97
DFMIN, 97
DLSS, 97
DOVL, 97
1 2 3 4 5 6, 96
DTK ,
E1SELF, 97
E2SELF, 97
EMULT, 97
EPH ,
1 2 3 4 5 6, 98
EREF, 98
EXII, 98
EXSS, 98
MCMOD, 98

MCPIT, 99
MLONG, 99
MTRAN, 99
NONPAR, 99
NSB, 99
NUM , =1,2,3,4,5, 100
NUMB, 99
RHO, 100
SAV ,
1 2 3 4 5, 100
SEED, 100
SOVL, 100
TRACE, 100
UTMIN, 100
VSOUND, 101
WBH, 101
WBII, 101
WDIST, 101
WEII, 101
WINFREEZE, 101
WSCREEN, 101
XWIDTH, 102
YWIDTH, 102
MPOLY, 105
MQ, 95
MR, 95
MS, 95
MT, 95
MTRAN, 99
MTUN, 123
MTV, 110
MV, 95
MXTun, 80
MYTun, 80
NAC, 69
NB, 116, 127
NC, 127
NCYCLES, 69
ND, 127
NE, 127
NG, 127
NGEN, 69
NHOT, 69
NONE, 110
NONPAR, 99
NRD, 85
NRE, 85
NRS, 85

INDEX

163

NS, 116
NSB, 99
NSS, 86
NTSI, 128
NUM , =1,2,3,4,5, 100
NUMB, 99
NUPDATES, 69
OMEGA, 106
OPTION
ACCELERATION, 103
CCLP, 103
CP, 103
CURRENT, 103
EA, 103
ED, 104
GCHCALC, 104
GRIDFREEZE, 104
IDEALITY, 104
INTRINSIC, 104
LCPUMP, 105
LDRAIN, 105
LSOURCE, 105
LVDMOSFET, 105
M3MODE, 106
MODEL, 105
MPOLY, 105
OMEGA, 106
PDOP, 106
PHYSCK, 106
RBULK, 107
RDRAIN, 107
RSOURCE, 107
TEMP, 107
UFB, 107
UN, 108
UP, 108
OUTPUT
ALL, 109
AS, 109
AVAL, 109
BI, 109
CC, 109
DC, 109
DIGITS, 109
ELAT, 109
ETRAN, 109
IV, 109

JLAT, 110
JTRAN, 110
MAJ, 110
MIN, 110
MOB, 110
MTV, 110
NONE, 110
OXIDE, 110
PHI, 110
PIF, 110
PSI, 110
RHO, 111
STRIPS, 111
TC, 111
OXIDE, 110
OXIDEF, 74
PACCREL, 69
PDOP, 106
PFSS, 88
PGMI, 80
PHI, 110
PHYSCK, 106
PIF, 110
PNSS, 87
PROFILE
ACTIVATION, 112
AKEV, 113
ASYM, 113
C1, 114
C2, 114
CA, 113
CD, 113
DMOFF, 114
DOSE, 114
EA, 114
ED, 114
ELEM, 115
FILE, 115
LD, 115
LFIT, 115
NB, 116
NS, 116
SMOFF, 116
SUBS, 116
TEMP, 117
TIME, 117
TOX, 117

INDEX

164

VAGD, 118
VAGS, 118
VAGX, 118
VAUD, 118
VDEN, 118
VDGD, 119
VDGS, 119
VDGX, 119
VDUD, 119
XOFF, 117
YOFF, 118
PSI, 110
PSS, 86
R1GRD, 74
R2GRD, 75
RBULK, 107
RDRAIN, 107
RECESS, 74
RECOMBINATION
AN, 120
AP, 120
BN, 120
BP, 120
CAN, 120
CAP, 121
CDN, 121
CDP, 121
CN, 121
CP, 121
CSEA, 121
CSED, 122
CSHA, 122
CSHD, 122
DN, 122
DP, 122
EN, 122
EP, 122
FN, 123
FP, 123
MTUN, 123
SN, 123
SP, 123
TEXF, 123
TLIF, 124
TN, 124
TP, 124
VN, 124

VP, 124
RHO, 100, 111
RSOURCE, 107
S1GAP, 75
S2GAP, 75
1 2 3 4 5, 100
SAV ,
SCGRD, 75
SDREOX, 75
SEED, 100
SGAP, 66
SISS, 86
SMOFF, 116
SN, 123
SOVL, 100
SP, 123
SRGRD, 75
STEP
C1, 125
C2, 126
DB, 125
DD, 125
DG, 125
E1, 126
E2, 126
ELOG, 126
LUDG, 126
LUGB, 126
NB, 127
NC, 127
ND, 127
NE, 127
NG, 127
VG, 127
STRIPS, 111
SUBS, 116
SXTun, 80
SYTun, 81
T, 128
T1GRD, 76
T2GRD, 76
TC, 111
TCERR, 69
TD, 75
TEMP, 71, 84, 107, 117
TEXF, 123
TGATE, 75
TIME, 72, 84, 117

INDEX

TINS, 66
TLIF, 124
TN, 124
TOX, 117
TP, 124
TRACE, 100
TRANSIENT
NTSI, 128
T, 128
U, 128
W, 128
TS, 76
U, 128
UB, 63
UD, 63
UFB, 107
UG, 63
UN, 108
UP, 108
US, 63
UTMIN, 100
VAGD, 118
VAGS, 118
VAGX, 118
VAUD, 118
VDEN, 118
VDGD, 119
VDGS, 119
VDGX, 119
VDUD, 119
VG, 127
VI, 63
VN, 124
VP, 124
VR, 63
VSOUND, 101
W, 66, 128
WBH, 101
WBII, 101
WDIST, 101
WEII, 101
WINFREEZE, 101
WSCREEN, 101
XE1, 81
XE2, 81
XE3, 81
XISS, 86

165

XLEFT, 66
XOFF, 117
XRIGHT, 66
XS1, 81
XS2, 81
XS3, 82
XWIDTH, 102
YDEPTH, 67
YE, 82
YMDEP, 76
YOFF, 118
YS, 82
YWIDTH, 102
ZE, 82
ZS, 82

166

INDEX