Genius

Semiconductor Device Simulator

Version 1.7.4

Genius User's Guide

C genda Pte Ltd

 Copyright (c) 2008-2009 Cogenda Pte Ltd, Singapore.

All rights reserved.

License Grant Duplication of this documentation is permitted only for internal use within the
organization of the licensee.

Disclaimer THIS DOCUMENTATION IS PROVIDED BY THE COPYRIGHT HOLDERS

AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WAR-
RANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WAR-
RANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE
GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIA-
BILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (IN-
CLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBIL-
ITY OF SUCH DAMAGE.

This documentation was typed in DocBook XML format, and typeset with the
ConTEXt program. We sincerely thank the contributors of the two projects, for
their excellent work as well as their generoisty.
 Contents
1 Physics in Genius Device Simulator 1

1.1 Level 1 Drift-Diusion Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.2 Level 2 Drift-Diusion Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

1.3 Level 3 Energy Balance Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

1.4 Band Structure Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

1.4.1 Band structure of compound semiconductors . . . . . . . . . . . . 8
1.4.1.1 Band Structure of SiGe . . . . . . . . . . . . . . . . . . . . . 8
1.4.1.2 Band Structure of Tertiary Compound
Semiconductor . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.4.2 Schenk's Bandgap Narrowing Model . . . . . . . . . . . . . . . . . . 9

1.5 Carrier Recombination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

1.6 Mobility Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

1.6.1 Bulk Mobility Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.6.1.1 Analytic Mobility Model . . . . . . . . . . . . . . . . . . 13
1.6.1.2 Masetti Analytic Model . . . . . . . . . . . . . . . . . . . 13
1.6.1.3 Philips Mobility Model . . . . . . . . . . . . . . . . . . . . 14
1.6.1.4 Velocity Saturation . . . . . . . . . . . . . . . . . . . . . . . 16
1.6.2 Unied Mobility Models . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
1.6.2.1 Lombardi Surface Mobility Model . . . . . . . . . . . 19
1.6.2.2 Lucent High Field Mobility Model . . . . . . . . . . . 20
1.6.2.3 Hewlett-Packard High Field Mobility Model . . . 22
1.6.3 Mobility Models of Complex Compound Semiconductors 24
1.6.4 Carrier Temperature Based Mobility Model . . . . . . . . . . . . 25

1.7 Generation Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

1.7.1 Impact Ionization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
1.7.1.1 Selberherr Model . . . . . . . . . . . . . . . . . . . . . . . . 26
1.7.1.2 van Overstraeten - de Man model . . . . . . . . . . . . 27
1.7.1.3 Valdinoci Model . . . . . . . . . . . . . . . . . . . . . . . . . 28
1.7.2 Band-to-band Tunneling . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

1.8 High Energy Particles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

1.9 Fermi-Dirac Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

2 Genius Command Reference 35

2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.1.1 Format of Input Card . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.1.2 Statement Description Format . . . . . . . . . . . . . . . . . . . . . . 37

Genius Device Simulator 3

Contents

2.2 Global Specication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

2.2.1 GLOBAL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.2.1.1 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.2.1.2 Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.2.1.3 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

2.3 Mesh Generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

2.3.1 MESH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
2.3.1.1 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
2.3.1.2 Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
2.3.1.3 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
2.3.2 X.MESH Y.MESH and Z.MESH . . . . . . . . . . . . . . . . . . . . 43
2.3.2.1 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
2.3.2.2 Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
2.3.2.3 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
2.3.3 ELIMINATE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
2.3.3.1 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
2.3.3.2 Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
2.3.3.3 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
2.3.4 SPREAD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
2.3.4.1 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
2.3.4.2 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
2.3.5 SPREAD3D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
2.3.5.1 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
2.3.5.2 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
2.3.6 REGION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
2.3.6.1 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
2.3.6.2 Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
2.3.6.3 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
2.3.7 FACE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
2.3.7.1 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
2.3.7.2 Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
2.3.7.3 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
2.3.8 REFINE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
2.3.8.1 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
2.3.8.2 Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
2.3.8.3 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

2.4 External Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

2.4.1 VSOURCE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
2.4.1.1 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
2.4.1.2 Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
2.4.1.3 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
2.4.2 ISOURCE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
2.4.2.1 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
2.4.2.2 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

4 Genius Device Simulator

 Contents

2.4.3 PARTICLE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
2.4.3.1 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
2.4.3.2 Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
2.4.3.3 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
2.4.4 LIGHT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
2.4.4.1 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
2.4.4.2 Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
2.4.4.3 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
2.4.5 ENVELOP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
2.4.5.1 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
2.4.5.2 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
2.4.6 Apply Source to Device . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
2.4.7 CIRCUIT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
2.4.7.1 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

2.5 Boundary Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

2.5.1 Boundary and Contact . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
2.5.1.1 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
2.5.1.2 Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
2.5.1.3 external circuit . . . . . . . . . . . . . . . . . . . . . . . . . . 74
2.5.1.4 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
2.5.2 ATTACH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
2.5.2.1 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
2.5.2.2 Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
2.5.2.3 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77

2.6 Fabrication Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

2.6.1 DOPING . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
2.6.2 Analytic PROFILE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
2.6.2.1 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
2.6.2.2 Analytic Prole Calculation . . . . . . . . . . . . . . . . 82
2.6.2.3 Analytic2 Prole Calculation . . . . . . . . . . . . . . . 83
2.6.2.4 Custom Doping Prole . . . . . . . . . . . . . . . . . . . . 84
2.6.2.5 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
2.6.3 MOLE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
2.6.3.1 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
2.6.3.2 Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
2.6.3.3 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
2.6.4 REGIONSET . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
2.6.4.1 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
2.6.4.2 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87

2.7 Physical Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88

2.7.1 PMI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
2.7.1.1 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
2.7.1.2 Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
2.7.1.3 Mobility Models . . . . . . . . . . . . . . . . . . . . . . . . . 90

Genius Device Simulator 5

Contents

2.7.1.4 Impact Ionization Models . . . . . . . . . . . . . . . . . . 91

2.7.1.5 Trap Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
2.7.1.6 Incomplete Ionization Model . . . . . . . . . . . . . . . 93
2.7.2 MODEL ....................................... 95
2.7.2.1 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
2.7.2.2 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96

2.8 Numerical Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97

2.8.1 METHOD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
2.8.1.1 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
2.8.1.2 Nonlinear Solver Algorithms . . . . . . . . . . . . . . . 99
2.8.1.3 Nonlinear Convergence Criteria . . . . . . . . . . . . 100
2.8.1.4 Linear Solver Algorithms . . . . . . . . . . . . . . . . . 100
2.8.2 SOLVE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
2.8.2.1 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
2.8.2.2 Device DC Sweep Mode . . . . . . . . . . . . . . . . . . 105
2.8.2.3 Device Transient Mode . . . . . . . . . . . . . . . . . . . 106
2.8.2.4 Device Operation Point Mode . . . . . . . . . . . . . . 106
2.8.2.5 Device Small-Signal AC Sweep Mode . . . . . . . 108
2.8.2.6 Device/Circuit Mixed Operation Point Mode . . 108
2.8.2.7 Device/Circuit Mixed DC Sweep Mode . . . . . . 108
2.8.2.8 Device/Circuit Mixed Transient Mode . . . . . . . 108
2.8.2.9 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
2.8.3 EMFEM2D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
2.8.3.1 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
2.8.3.2 Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
2.8.4 RAYTRACE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
2.8.4.1 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
2.8.4.2 Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
2.8.4.3 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114

2.9 Input and Output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115

2.9.1 IMPORT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
2.9.1.1 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
2.9.1.2 description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
2.9.1.3 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
2.9.2 EXPORT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
2.9.2.1 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
2.9.2.2 description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
2.9.2.3 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
2.9.3 HOOK . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
2.9.3.1 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
2.9.3.2 Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
2.9.3.3 Default Hook Functions . . . . . . . . . . . . . . . . . . 117
2.9.3.4 CV Postprocessor . . . . . . . . . . . . . . . . . . . . . . . 118
2.9.3.5 Probe Postprocessor . . . . . . . . . . . . . . . . . . . . . 118

6 Genius Device Simulator

 Contents

2.9.3.6 Visualization and Animation Postprocessor . . . 118

2.9.3.7 Interface to Visit visualizer . . . . . . . . . . . . . . . . 119

3 Extending Genius Simulator 121

3.1 Custom Material Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121

3.1.1 Overview of PMI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
3.1.1.1 Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
3.1.1.2 Classes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
3.1.1.3 How a PMI Library is Loaded . . . . . . . . . . . . . 124
3.1.1.4 How a PMI Library is Used . . . . . . . . . . . . . . . 125
3.1.2 Procedure of Building a Custom Material Library . . . . . . 128
3.1.2.1 Building the Example . . . . . . . . . . . . . . . . . . . . 128
3.1.2.2 Create a New Material Library . . . . . . . . . . . . . 128
3.1.3 Class References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
3.1.3.1 PMI_Server Class . . . . . . . . . . . . . . . . . . . . . . . 130
3.1.3.2 PMIS_Server Class . . . . . . . . . . . . . . . . . . . . . . 130
3.1.3.3 PMIS_BasicParameter Class . . . . . . . . . . . . . . 131
3.1.3.4 PMIS_BandStructure Class . . . . . . . . . . . . . . . 131
3.1.3.5 PMIS_Mobility . . . . . . . . . . . . . . . . . . . . . . . . . 137
3.1.3.6 PMIS_Avalanche . . . . . . . . . . . . . . . . . . . . . . . 138
3.1.3.7 PMIS_Trap . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
3.1.3.8 PMIS_Thermal . . . . . . . . . . . . . . . . . . . . . . . . . 139
3.1.3.9 PMIS_Optical . . . . . . . . . . . . . . . . . . . . . . . . . . 140
3.1.3.10 PMII . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
3.1.3.11 PMII_BasicParameter . . . . . . . . . . . . . . . . . . . . 141
3.1.3.12 PMII_Thermal . . . . . . . . . . . . . . . . . . . . . . . . . 141
3.1.3.13 PMII_Optical . . . . . . . . . . . . . . . . . . . . . . . . . . 141
3.1.3.14 PMIC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
3.1.3.15 PMIC_BasicParameter . . . . . . . . . . . . . . . . . . . 141
3.1.3.16 PMIC_Thermal . . . . . . . . . . . . . . . . . . . . . . . . . 142
3.1.3.17 PMIC_Optical . . . . . . . . . . . . . . . . . . . . . . . . . . 142

3.2 Hooks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143

3.3 Python Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144

3.3.1 HTTP Frontend . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144

A References CXLV

B Index CXLVII

Genius Device Simulator 7

Contents

8 Genius Device Simulator

 CHAPTER

1 Physics in Genius Device Sim-

ulator
Since Gummel's original work, the drift-diusion model has been widely used in
the semiconductor device simulation. It is now the de facto industry standard in
this eld.

The original DD model can be achieved by following approximation from hydro-

dynamic model:

Light speed is much faster than carrier speed.

All the collision is elastic.
Bandgap does not change during collision.
Carrier temperature equals to lattice temperature and keeps equilibrium.
The gradient of driving force should keep small.
Carrier degenerate can be neglected.

Some improvements have been applied to DD model for extend its capability.
These "patches" of course make things complex, but they can deal with real prob-
lems.

This chapter describes the DD model and its variations used by GENIUS code for
describing semiconductor device behavior as well as physical based parameters
such as mobility, recombination rate and son on.

Level 1 Drift-Diusion Equation

Level 1 Drift-Diusion (DDML1) is the fundamental solver of GENIUS code for
lattice temperature keeps constant though out the solve procedure.

The primary function of DDML1 is to solve the following set of partial dierential
equations, namely Poisson's equation, along with the hole and electron continuity
equations:

Poisson's Equation

= ( + + ) (1.1)

where, is the electrostatic potential of the vacuum level. This choice makes
the description of metal-oxide-semiconductor contact and heterojunction easier.

Genius Device Simulator 1

Physics in Genius Device Simulator Level 1 Drift-Diusion Equation

and are the electron and hole concentration, + and are the ionized
impurity concentrations. is the magnitude of the charge of an electron.

The relationship of conduct band , valence band and vacuum level is:

= (1.2.a)
= + . (1.2.b)

Here, is the electron anity. is the bandgap of semiconductor. and

are the bandgap shift caused by heavy doping or mechanical strain.

Furthermore, the relationship of vacuum level and intrinsic Fermi potential

intrinsic is:

= intrinsic ln (1.3)
2 2 ( )

The reference 0 eV of energy is set to intrinsic Fermi level of equilibrium state in

GENIUS.

Continuity The continuity equations for electrons and holes are dened as follows:
Equations
1
= ( ) (1.4.a)

1
= ( ) (1.4.b)

where and are the electron and hole current densities, and are the
recombination and generation rates for both electrons and holes.

Drift-Diusion The current densities and are expressed in terms of the level 1 drift-diusion
Current Equations model here.

= + (1.5.a)
= (1.5.b)

where and are the electron and hole mobilities. = and =

are the electron and hole diusivities, according to Einstein relationship.

Eective and are the eective driving electrical eld to electrons and holes, which
Electrical Field related to local band diagram. The band structure of heterojunction has been taken
into account here [1].

1
= (ln( ) ln( 3/2 )) (1.6.a)

2 Genius Device Simulator

Level 1 Drift-Diusion Equation Physics in Genius Device Simulator

1
= + (ln( ) ln( 3/2 )) (1.6.b)

The lattice temperature keeps uniform throughout DDML1, the above temperature
gradient item takes no eect in fact.

By substituting drift-diusion model into the current density expressions, and

combining with Poisson's equation, the following basic equations for DDML1 are
obtained:

= + ( ) (1.7.a)
( )

= ( ) (1.7.b)
( )
= ( + + ) (1.7.c)

DDML1 is suitable for PN diode, BJT transistor and long gate MOSFET simulation.
It is robust, and runs pretty fast for real work. The detailed discretization scheme
can be found at [[TODO]].

Genius Device Simulator 3

Physics in Genius Device Simulator Level 2 Drift-Diusion Equation

Level 2 Drift-Diusion Equation

The Level 2 DD model considers the inuence of lattice temperature by solving
the extra thermal equation simultaneously with the electrical equations. Also, the
formula of drift-diusion equation should be modied according to [2].

The electron diusion current in DDML1 can be written as:

,di = = (1.8)

Temperature But for DDML2, it has the form of

Gradient Correction

,di = ( + ) (1.9)

The hole diusion current should be modied in the same manner.

,di = ( + ) (1.10)

Heat Flow Equation The following heat ow equation is used:

= + + ( + 3 ) ( ) (1.11)

where is the mass density of semiconductor material. is the heat capacity.

is the thermal conductivity of the material. is the joule heating of current.
( +3 )( ) is lattice heating due to carrier recombination and generation.

From above discussion, the governing equations for DDML2 are as follows:

= + + ( ) (1.12.a)
( )

= ( ) (1.12.b)
( )
= ( + + ) (1.12.c)

= + + ( + 3 ) ( ) (1.12.d)

This model can be used as power transistor simulation as well as breakdown sim-
ulation. Unfortunately, nearly all the physical parameters are related with temper-
ature. They should be considered during self consistent simulation, which greatly
slows down the speed. The DDML2 solver runs 50 70 % slower than DDML1.
However, it seems no convergence degradation happens in most of the case. The
discretization scheme can be found at [[TODO]].

4 Genius Device Simulator

Level 3 Energy Balance Equation Physics in Genius Device Simulator

Level 3 Energy Balance Equation

Energy Balance Model [3] is introduced into GENIUS code for simulating short
channel MOSFET. This is a simplication of full hydrodynamic (HD) model [4].
The current density expressions from the drift-diusion model are modied to in-
clude additional coupling to the carrier temperature. Also, reduced carrier energy
conservation equations, which derived from second order moment of Boltzmann
Transport Equation, are solved consistently with drift-diusion model. The sim-
plication from HD to EB makes sophisticated Scharfetter-Gummel discretization
still can be used in the numerical solution, which ensures the stability.

Current Equation The current density and are then expressed as:
for EBM
= + ( + ) (1.13.a)
= ( + ) (1.13.b)

where, $T_n$ and $T_p$ are electron and hole temperature, respectively. The dif-
ference between above equations and carrier density equations in DDML2 is lattice
temperature replaced by carrier temperature.

Energy Balance In addition, the energy balance model includes the following electron and hole
Equations energy balance equations:

( )
+ = + (1.14.a)

( )
+ = + (1.14.b)

where, and are electron and hole energy. For HD model, the carrier energy
includes thermal and kinetic terms = 32 + 12 * 2 , but only thermal energy
for EB model = 32 . Here stands for or . 0 = 32 is the carrier
equilibrium energy, for carrier temperature equals to lattice temperature.

and are the ux of energy:

= ( + ) (1.15.a)


= + ( + ) (1.15.b)

The heat conductivity parameter for carriers can be expressed as:

Genius Device Simulator 5

Physics in Genius Device Simulator Level 3 Energy Balance Equation

2
2
= ( + ) (1.16)
5

where stands for and , respectively. The constant parameter equals 0.7 in
the GENIUS software.

The and are the rate of net loss of carrier kinetic energy:

3 3
= (Aug ) + + dir )
( 2 ) 2 ( SHR
( 0 )
(1.17.a)

3 3
= (Aug ) + + dir )
( 2 ) 2 ( SHR
( 0 )
(1.17.b)

Lattice Heat At last, the lattice heat ow equation should be rewritten as:
Equation for EBM

= + (1.18)

where

3 3 ( 0 ) ( 0 )
= SHR + + + + (1.19)
( 2 2 )

The carrier energy is mainly contributed by joule heating term , and heating
(cooling) due to carrier generation (recombination) term. The carriers exchange
energy with lattice by collision, which described by energy relaxation term .
This model is suitable for sub-micron MOS (channel length 1 0.1 m) and
advanced BJT simulation. However, the computation burden of EB method is
much higher than DD. And the convergence of EB solver is dicult to achieve,
which requires more strict initial value and more powerful inner linear solver. The
discretization scheme can be found at [[TODO]].

From above discussion, all the governing equations of DD/EB method is elliptical
or parabolic. From mathematic point of view, does not like hyperbolic system 1, the
solution of elliptical or parabolic system is always smooth. The required numerical
technique is simple and mature for these systems. As a result, the DD and EB
method is preferred against full hydrodynamic method.

1
One have to face discontinuous problem, i.e. shock wave.

6 Genius Device Simulator

Band Structure Model Physics in Genius Device Simulator

Band Structure Model

The band structure parameters, including bandgap , eective density of states in
the conduction band and valence band , and intrinsic carrier concentration
, are the most important and fundamental physical parameters for semiconduc-
tor material [5].

Eective Density Eective density of states in the conduction and valence band are dened as fol-
of States lows:
3/2
*
2 (1.20.a)
( 22 )
3/2
*
2 (1.20.b)
( 22 )

The temperature dependencies of eective density of states is fairly simple:

1.5
( ) = (300 K) ( ) (1.21.a)
300 K
1.5
( ) = (300 K) ( (1.21.b)
300 K )

Bandgap The bandgap in GENIUS is expressed as follows:

2
( ) = (0)
+
3002 2
= (300) + (1.22.a)
[ 300 + + ]

Bandgap When bandgap narrowing eects due to heavy doping takes place [6], the band
Narrowing due edge shifts:
to Heavy Doping
bgn total 2
= ln + ln total + bgn . (1.23)
2 ref ( ref )

The intrinsic concentration should be modied:

= exp exp( ) (1.24)
( 2 )

Since the carrier current Equation (1.6), p. 2 involves the energy level of conduc-
tion band and valence band , the bandgap shift should be attributed to them.

Genius Device Simulator 7

Physics in Genius Device Simulator Band Structure Model

The bandgap narrowing is attributed half to the conduction band and another half
to the valence band as default:
1
= (1.25.a)
2
1
= + (1.25.b)
2

The parameters used in the default band structure model is listed in Table 1.1,
p. 8.

Symbol Parameter Unit Silicon GaAs

(300) EG300 eV 1.1241 1.424
EGALPH eV/K 2.73 104 5.405 104
EGBETA K 0 204
bgn V0.BGN eV 6.92 103 0
ref N0.BGN cm3 1.30 1017 1 1017
bgn CON.BGN - 0.5 0.5
ELECMASS 0 1.0903 0.067
HOLEMASS 0 1.1525 0.6415
(300) NC300 cm3 2.86 1019 4.7 1017
(300) NV300 cm3 3.10 1019 7.0 1018

Table 1.1 Parameters of the Default band structure model

Band structure of compound semiconductors

[[TODO]]

Band Structure of SiGe

[[TODO]]

Band Structure of Tertiary Compound Semiconductor

[[TODO]]

Bandgap [[TODO]]

Electron Anity [[TODO]]

Eective Mass [[TODO]]

Density of States [[TODO]]

8 Genius Device Simulator

Band Structure Model Physics in Genius Device Simulator

Schenk's Bandgap Narrowing Model

[[TODO]] Equations of Schenk's model

The Schenk's bandgap narrowing model is available for silicon, and can be loaded
with the option Schenk in the PMI command.

Genius Device Simulator 9

Physics in Genius Device Simulator Carrier Recombination

Carrier Recombination
Three recombination mechanisms are considered in GENIUS at present, including
Shockley-Read-Hall, Auger, and direct (or radiative) recombination. The total
recombination is considered as the sum of all:

= = = SRH + dir + Auger (1.26)

where SRH , dir and Auger are SRH recombination, direct recombination and
Auger recombination, respectively.

SRH Recombination Shockley-Read-Hall (SRH) recombination rate is determined by the following for-
mula:

2
SRH =

+ exp + + exp
[ ( )] [ ( )]
(1.27)

where and are carrier life time, which dependent on impurity concentration
[7].

= (1.28.a)
1 + total /

= (1.28.b)
1 + total /

Auger Recombination The Auger recombination is a three-carrier recombination process, involving ei-
ther two electrons and one hole or two holes and one electron. This mechanism
becomes important when carrier concentration is large.

Auger = (2 2 ) + (2 2 ) (1.29)

where AUGN and AUGP are Auger coecient for electrons and holes. The value of
Auger recombination Auger can be negative some times, which refers to Auger
generation.

Direct Recombination The direct recombination model expresses the recombination rate as a function of
the carrier concentrations and , and the eective intrinsic density :

dir = ( 2 ) (1.30)

The default value of the recombination parameters are listed in Table 1.2, p. 11:

10 Genius Device Simulator

Carrier Recombination Physics in Genius Device Simulator

Parameter Unit Silicon GaAs Ge

ETRAP eV 0 0 0
DIRECT cm3 s1 1.1e-14 7.2e-10 6.41e-14
AUGN cm6 s1 1.1e-30 1e-30 1e-30
AUGP cm6 s1 0.3e-30 1e-29 1e-30
TAUN0 s 1e-7 5e-9 1e-7
TAUP0 s 1e-7 3e-6 1e-7
NSRHN cm3 5e16 5e17 5e16
NSRHP cm3 5e16 5e17 5e16

Table 1.2 Default values of recombination parameters

Surface At semiconductor-insulator interfaces, additional SRH recombination can be in-

Recombination troduced. The surface recombination rate has the unit cm2 s1 , and is calculated
with

2
Surf = . (1.31)
1 1
+ ) + + )
( (

The surface recombination velocities, STAUN and STAUP, have the unit of cm/s,
and the default value of 0.

Genius Device Simulator 11

Physics in Genius Device Simulator Mobility Models

Mobility Models
Carrier mobility is one of the most important parameters in the carrier transport
model. The DD model itself, developed at early 1980s, is still being used today
due to advanced mobility model enlarged its ability to sub-micron device.

Mobility modeling is normally divided into: low eld behavior, high eld behav-
ior and mobility in the (MOS) inversion layer.

The low electric eld behavior has carriers almost in equilibrium with the lattice.
The low-eld mobility is commonly denoted by the symbol 0 , 0 . The value
of this mobility is dependent upon phonon and impurity scattering. Both of which
act to decrease the low eld mobility. Since scattering mechanism is depended on
lattice temperature, the low-eld mobility is also a function of lattice temperature.

The high electric eld behavior shows that the carrier mobility declines with elec-
tric eld because the carriers that gain energy can take part in a wider range of
scattering processes. The mean drift velocity no longer increases linearly with
increasing electric eld, but rises more slowly. Eventually, the velocity doesn't
increase any more with increasing eld but saturates at a constant velocity. This
constant velocity is commonly denoted by the symbol . Impurity scattering is
relatively insignicant for energetic carriers, and so is primarily a function of
the lattice temperature.

Modeling carrier mobilities in inversion layers introduces additional complica-

tions. Carriers in inversion layers are subject to surface scattering, carrier-carrier
scattering, velocity overshoot and quantum mechanical size quantization eects.
These eects must be accounted for in order to perform accurate simulation of
MOS devices. The transverse electric eld is often used as a parameter that indi-
cates the strength of inversion layer phenomena.

It can be seen that some physical mechanisms such as velocity overshoot and quan-
tum eect which can't be described by DD method at all, can be taken into account
by comprehensive mobility model. The comprehensive mobility model extends
the application range of DD method. However, when the EB method (which ac-
counts for velocity overshoot) and QDD method (including quantum eect) are
used, more calibrations are needed to existing mobility models.

Bulk Mobility Models

The rst family of mobility models were designed to model the carrier transport
at low electric elds. They usually focus on the temperature and doping con-
centration dependence of the carrier mobilities. The surface-related or transverse
E-eld eects are not included in these models. On the other hand, in GENIUS,

12 Genius Device Simulator

Mobility Models Physics in Genius Device Simulator

these low-eld mobilities models are coupled to a velocity saturation model to

account for the carrier velocity saturation eect. This family of mobility models
are suitable for bulk device, such as bipolar transistors.

In brief, the low eld carrier mobility is rst computed, then a velocity satura-
tion formula is applied to yield the corrected mobility value. Three choices are
available for the low-eld mobility calculation, each described in one of the fol-
lowing sub-sections. The choices of velocity saturation is described in the last
sub-section.

Analytic Mobility Model

In the GENIUS code, the Analytic Mobility model [8] is the default low eld mo-
bility model for all the material. It is an concentration and temperature dependent
empirical mobility model expressed as:

max (
) min
0 = min + 300 (1.32)

total
1+(
300 ) ( ref )

where total = + is the total impurity concentration.

Default parameters for Si, GaAs and Ge are listed below:

Symbol Parameter Unit Si:n Si:p GaAs:n GaAs:p

min MUN.MIN / MUP.MIN cm2 V1 s1 55.24 49.70 0.0 0.0
max MUN.MAX / MUP.MAX cm2 V1 s1 1429.23 479.37 8500.0 400.0
NUN / NUP - -2.3 -2.2 -1.0 -2.1
XIN / XIP - -3.8 -3.7 0.0 0.0
ALPHAN / ALPHAP - 0.73 0.70 0.436 0.395
ref NREFN / NREFP cm3 1.072e17 1.606e17 1.69e17 2.75e17

Table 1.3 Default parameter values of the analytic mobility model

In GENIUS, the analytic model is the simplest mobility model, and is available
for a wide range of materials. For some materials, such as silicon, some more
advanced mobility models are available.

Masetti Analytic Model

The doping-dependent low-eld mobility model proposed by Masetti et al.[9] is an

alternative to the default analytic model. The general expression for the low-eld
mobility is

Genius Device Simulator 13

Physics in Genius Device Simulator Mobility Models

min2 1
dop = min1 exp + const (1.33)
( tot ) 1 + ( / )
1 + ( /tot )
tot

where tot is the total doping concentration. The term const is the temperature-de-
pendent, phonon-limited mobility

const = max ( (1.34)
300 )

where is the lattice temperature.

The parameters of the Masetti model is listed in Table 1.4, p. 14. The Masetti
model is the default mobility model for the 4H-SiC material.

Symbol Parameter Unit 4H-SiC:n 4H-SiC:p

max MUN.MAX / MUP.MAX cm2 V1 s1 947.0 124.0
MUN.ZETA / MUP.ZETA - 1.962 1.424
min1 MUN.MIN1 / MUP.MIN1 cm2 V1 s1 0 15.9
min2 MUN.MIN2 / MUP.MIN2 cm2 V1 s1 0 15.9
1 MUN1 / MUP1 cm2 V1 s1 0 0
PCN / PCP cm3 0 0
CRN / CRP cm3 1.94 1017 1.76 1019
CSN / CSP cm3 0 0
MUN.ALPHA / MUP.ALPHA - 0.61 0.34
MUN.BETA / MUP.BETA - 0 0

Table 1.4 Parameters of the Masetti mobility model

Philips Mobility Model

Another low eld mobility model implemented into GENIUS is the Philips Uni-
ed Mobility model [10],[11]. This model takes into account the distinct acceptor
and donor scattering, carrier-carrier scattering and carrier screening, which is rec-
ommended for bipolar devices simulation.

The electron mobility is described by the following expressions:

0, 1 = Lattice, 1 + ++ 1 (1.35)

where 0, is the total low eld electron mobilities, Lattice, is the electron mo-
bilities due to lattice scattering, ++ is the electron and hole mobilities due to
donor (D), acceptor (A), screening (P) and carrier-carrier scattering.

14 Genius Device Simulator

Mobility Models Physics in Genius Device Simulator

2.285
Lattice, = max ( (1.36)
300 )
sc, ref +
++ = 1, + 2, (1.37)
( sc,e, ) ( sc, ) ( sc,e, )

The parameters 1, and 2, are given as:

2
max 31.5
1, = ( 300 ) (1.38.a)
max min
max min 300 1.5
2, = (1.38.b)
max min ( )

where sc, and sc,e, is the impurity-carrier scattering concentration and eect
impurity-carrier scattering concentration given by:

sc, = * + * + (1.39.a)

sc,e, = * + * ( ) + (1.39.b)
( )

*
where and * take ultra-high doping eects into account and are dened by:

1
* = 1 + (1.40.a)
,ref 2
+
( )

1
* = 1 + (1.40.b)
,ref 2
+
( )

The screening factor functions ( ) and ( ) take the repulsive potential for
acceptors and the nite mass of scattering holes into account.

0.89233 0.005978
( ) = 1 +
0.28227 0.19778 0.72169 1.80618
0.41372 + ( 0 300 )
[ ] [ ( 0 300 ) ]
(1.41)

0.7643 0.6478 + 2.2999 + 6.5502

( ) = (1.42)
0.6478 + 2.3670 0.8552

The parameter that takes screening eects into account is given by:

Genius Device Simulator 15

Physics in Genius Device Simulator Mobility Models

1

2
= +
( 300 ) (1.43)
sc,ref sc, 2/3 c,ref

+ ( 0 )

Similar expressions hold for holes. The default parameters for Philips model are
listed in Table 1.5, p. 16:

Symbol Parameter Unit Si:n Si:p

min MMNN.UM / MMNP.UM cm2 V1 s1 55.24 49.70
max MMXN.UM / MMXP.UM cm2 V1 s1 1417.0 470.5
ALPN.UM / ALPP.UM - 0.68 0.719
ref NRFN.UM / NRFP.UM cm3 9.68e16 2.23e17
CRFD.UM - 0.21 0.21
CRFA.UM - 0.5 0.5
D,ref NRFD.UM cm3 4.0e20 4.0e20
A,ref NRFA.UM cm3 7.2e20 7.2e20
me_over_m0 0 1.0 -
mh_over_m0 0 - 1.258
- 2.459 2.459
- 3.828 3.828
sc,ref NSC.REF cm2 3.97e13 3.97e13
c,ref CAR.REF cm3 1.36e20 1.36e20

Table 1.5 Default values of Philips mobility model parameters

In the actual code, Philips model is corrected by Caughey-Thomas expression for

taking high eld velocity saturation eects into account. This model can be loaded
by Philips keyword in the PMI statements.

Velocity Saturation

Silicon-like For silicon-like materials, the Caughey-Thomas expression [12], is used:

materials
0
= 1/
(1.44)
0
1+
[ ( sat ) ]

where is the electric eld parallel to current ow. sat is the saturation velocities
for electrons or holes. They are computed by default from the expression:

16 Genius Device Simulator

Mobility Models Physics in Genius Device Simulator

sat0
sat ( ) = (1.45)

1 + exp ( 600 )

The parameters and the default values for silicon is listed in Table 1.6, p. 17.

Symbol Parameter Unit Si:n Si:p

sat0 VSATN0 / VSATP0 cm/s 2.4 107 2.4 107
BETAN / BETAP - 2.0 1.0
VSATN.A / VSATP.A - 0.8 0.8

Table 1.6 Velocity saturation parameters of silicon-like materials

GaAs-like materials For GaAs-like materials, another expression due to [13] is used to describe the
negative dierential resistance:

4
0 +
( 0 )
= 4
(1.46)

1+
( 0 )

where 0 is the reference eld, and the saturation velocity

( ) = vsat vsat (1.47)

The negative dierential property of carrier mobility is described in this model.

When electric eld increases in this model, the carrier drift velocity ( ) reaches
a peak and then begins to decrease at high elds due to the transferred electron
eect.

The parameters are listed in Table 1.7, p. 17.

Symbol Parameter Unit GaAs:n GaAs:p

vsat VSATN.A / VSATP.B cm/s 1.13 107 1.13 107
vsat VSATN.A / VSATP.B cm/s/K 1.2 104 1.2 104

Table 1.7 Velocity saturation parameters of GaAs-like materials

GaAs-specic When using this model for GaAs MESFET device simulation, the negative dif-
model ferential property may cause the drain output characteristics (current vs. voltage)
exhibit an unrealistic oscillation behavior. Another model to describe high eld
eects developed by Yeager [14] can be used.

sat 0
= tanh (1.48)
( sat )

Genius Device Simulator 17

Physics in Genius Device Simulator Mobility Models

This GaAs-specic model can be loaded by Hypertang keyword in PMI state-

ment.

4H-SiC-specic For 4H-SiC, the saturation velocity is calculated with the following formula
model

( ) = vsat vsat ( (1.49)
300 )

where the parameters are listed in

Symbol Parameter Unit 4H-SiC:n 4H-SiC:p

7
vsat VSATN.A / VSATP.B cm/s 1.07 10 8.37 106
vsat VSATN.A / VSATP.B cm/s 0 0

Table 1.8 Velocity saturation parameters of 4H-SiC

18 Genius Device Simulator

Mobility Models Physics in Genius Device Simulator

Unied Mobility Models

The other family of mobility models are the unied mobility models. The eect
of high transverse and parallel E-eld is an integral part in the design of these
mobility models. As a result, these models are recommended for silicon MOSFET
simulation. On the other hand, the availability of the unied models is limited to
a few materials, such as silicon and silicon-germanium.

Lombardi Surface Mobility Model

Along an insulator-semiconductor interface, the carrier mobilities can be substan-

tially lower than in the bulk of the semiconductor due to surface-related scattering.
If no surface degradation is considered, the drain-source current may exceed about
30 % for MOS simulation.

The Lombardi mobility model [15] is an empirical model that is able to describe
the carrier mobility in the MOSFET inversion layer. The Lombardi model consists
of three components

, the doping-dependent bulk mobility. This component mainly accounts for

the ionized impurity scattering.
ac , the mobility degradation due to acoustic phonon scattering in the inversion
layer. Due to the quantum connement in the potential well at the interface,
this mobility degradation is a strong function of the transverse electric eld.
sr , the mobility degradation due to the surface roughness scattering. This
component is also a strong function of the transverse electric eld.

To obtain the nal value of carrier mobility, the three components are combined
using the Matthiessen's rule:

s 1 = 1 + ac 1 + sr 1 . (1.50)

Bulk Component The bulk mobility component in Lombardi's model is similar to that of Masetti's
model, which reads

max 0 1
= 0 exp + (1.51)
( tot ) 1 + ( / )
1 + ( /tot )
tot

max = 2 ( (1.52)
300 )

Acoustic Phonon The acoustic phonon limited mobility component is

Component
total
ac = + (1.53)
3

Genius Device Simulator 19

Physics in Genius Device Simulator Mobility Models

where is the transverse component of the electric eld.

Surface Roughness Finally, the surface roughness limited mobility is expressed as

Component

sr = . (1.54)

The Lombardi model is uses the Caughey-Thomas model for velocity saturation
calculation, see Velocity Saturation, p. 16 for details.

The parameters used in the Lombardi model is summarized in Table 1.9, p. 20.

Symbol Parameter Unit Si:n Si:p

EXN1.LSM / EXP1.LSM - 0.680 0.719
EXN2.LSM / EXP2.LSM - 2.0 2.0
EXN3.LSM / EXP3.LSM - 2.5 2.2
EXN4.LSM / EXP4.LSM - 0.125 0.0317
EXN8.LSM / EXP8.LSM - 2.0 2.0
0 MUN0.LSM / MUP0.LSM cm2 V1 s1 52.2 44.9
1 MUN1.LSM / MUP1.LSM cm2 V1 s1 43.4 29.0
2 MUN2.LSM / MUP2.LSM cm2 V1 s1 1417.0 470.5
PC.LSM cm3 0 (xed) 9.23 1016
CRN.LSM / CRP.LSM cm3 9.68 1016 2.23 1017
CSN.LSM / CSP.LSM cm3 3.43 1020 6.10 1020
BN.LSM / BP.LSM cm/s 4.75 107 9.93 106
CN.LSM / CP.LSM - 1.74 105 8.84 105
DN.LSM / DP.LSM - 5.82 1014 2.05 1014
sat0 VSATN0 / VSATP0 cm/s 2.4 107 2.4 107
BETAN / BETAP - 2.0 1.0
VSATN.A / VSATP.A - 0.8 0.8

Table 1.9 Parameters of Lombardi mobility model

Hewlett-Packard mobility model can be loaded by Lombardi keyword in the PMI

statement.

Lucent High Field Mobility Model

The Lucent Mobility model [16] is an all-inclusive model which suitable for MOS
simulation. This model incorporates Philips Unied Mobility model and the Lom-
bardi Surface Mobility model, as well as accounting for high eld eects. For low
longitudinal eld, the carrier mobility is given by Matthiessen's rule:

20 Genius Device Simulator

Mobility Models Physics in Genius Device Simulator

1
1 1 1
0 = + + (1.55)
[ b ac sr ]

where is the bulk mobility comes from the Philips model, and ac and sr come
from the Lombardi model. The details of these models are described in Philips
Mobility Model, p. 14 and Lombardi Surface Mobility Model, p. 19, and
will not be repeated here.

There is however a modication to the surface roughness formula Equation (1.54),

p. 20. The constant exponent is replaced by the following function

( + )
=+ . (1.56)
tot

The Lucent model uses the Caughey-Thomas model for velocity saturation calcu-
lation, see Velocity Saturation, p. 16 for details.

The parameters of the Lucent model is listed in Table 1.10, p. 21.

Symbol Parameter Unit Si:n Si:p

2 1 1
min MMNN.UM / MMNP.UM cm V s 55.24 49.70
max MMXN.UM / MMXP.UM cm2 V1 s1 1417.0 470.5
ALPN.UM / ALPP.UM - 0.68 0.719
ref NRFN.UM / NRFP.UM cm3 9.68e16 2.23e17
CRFD.UM - 0.21 0.21
CRFA.UM - 0.5 0.5
D,ref NRFD.UM cm3 4.0e20 4.0e20
A,ref NRFA.UM cm3 7.2e20 7.2e20
me_over_m0 0 1.0 -
mh_over_m0 0 - 1.258
- 2.459 2.459
- 3.828 3.828
sc,ref NSC.REF cm2 3.97e13 3.97e13
c,ref CAR.REF cm3 1.36e20 1.36e20
sat0 VSATN0 / VSATP0 cm/s 2.4 107 2.47
EXN4.LUC / EXP4.LUC - 0.0233 0.0119
EXN9.LUC / EXP9.LUC - 0.0767 0.123
AN.LUC / AP.LUC - 2.58 2.18
BN.LUC / BP.LUC cm/s 3.61 107 1.51 107
CN.LUC / CP.LUC - 1.70 104 4.18 103
Table 1.10 Default values of Lucent mobility model parameters

Genius Device Simulator 21

Physics in Genius Device Simulator Mobility Models

Symbol Parameter Unit Si:n Si:p

DN.LUC / DP.LUC - 3.58 1018 3.58 1018
FN.LUC / FP.LUC - 6.85 1021 7.82 1021
KN.LUC / KP.LUC - 1.70 0.90
sat0 VSATN0 / VSATP0 cm/s 2.4 107 2.4 107
BETAN / BETAP - 2.0 1.0
VSATN.A / VSATP.A - 0.8 0.8

Table 1.11 Default values of Lucent mobility model parameters

The Lucent model can be loaded by Lucent keyword in the PMI statements. This
is an accurate model recommended for MOS devices. However, its computation
cost is higher than other mobility models. At the same time, it is also less numer-
ically stable.

Hewlett-Packard High Field Mobility Model

It is reported that Hewlett-Packard mobility model [17] achieves the same ac-
curacy as Lucent model with relatively small computational burden in the MOS
simulation.

This model also takes into account dependence on electric elds both parallel ( )
and perpendicular ( ) to the direction of current ow. The mobility is calculated
from

= (1.57)
2



1 + ( ) +


+

where the tranverse eld dependent component is given as:

0 if total > ref

0
=
1 +

ref

The low eld mobility 0 is calculated from the Analytic model, as described
in Analytic Mobility Model, p. 13.

The parameters of the Hewlett-Packard mobility model and its default values for
silicon is listed in Table 1.12, p. 23.

22 Genius Device Simulator

Mobility Models Physics in Genius Device Simulator

Symbol Parameter Unit Si:n Si:p

min MUN.MIN / MUP.MIN cm2 V1 s1 55.24 49.70
max MUN.MAX / MUP.MAX cm2 V1 s1 1429.23 479.37
NUN / NUP - -2.3 -2.2
XIN / XIP - -3.8 -3.7
ALPHAN / ALPHAP - 0.73 0.70
ref NREFN / NREFP cm3 1.072 1017 1.606 1017
0 MUN0.HP / MUP0.HP cm2 V1 s1 774.0 250
VCN.NP / VCP.HP cm s1 4.9 106 2.928 106
VSN.NP / VSP.HP cm s1 1.036 107 1.2 107
GN.HP / GP.HP - 8.8 1.6
ref NREFN / NREFP cm3 5 1017 5 1017
ref ECN.HP / ECP.HP V cm1 5.5 105 2.78 105

Table 1.12 Default values of HP mobility model parameters

The Hewlett-Packard mobility model can be loaded by HP keyword in the PMI

statement.

Genius Device Simulator 23

Physics in Genius Device Simulator Mobility Models

Mobility Models of Complex Compound Semiconductors

[[TODO]]

24 Genius Device Simulator

Mobility Models Physics in Genius Device Simulator

Carrier Temperature Based Mobility Model

We should notice here, all the above mobility models are developed under the
framework of DD method. Since DD is an approximate model for semiconductor,
the dierence between DD model and real device is corrected by mobility models!
These mobility model contains some physical model that DD does not consider.
For example, the high eld correction has already contains the eect of hot car-
riers. The surface mobility for MOSFET not only considers the mobility degrade
due to surface roughness, but also contains the eect caused by carrier concen-
tration decrease due to quantum well in inverse layer. These corrections extended
the application range of DD model, also make the mobility model rather complex.

When the physical model is more accurate, the carrier mobility model can be
less complicated. Thus, the mobility models suitable for DD model may NOT be
suitable for energy balance model. There are some mobility models developed
special for energy balance model [3]. However, they are yet to be implemented in
GENIUS.

Genius Device Simulator 25

Physics in Genius Device Simulator Generation Model

Generation Model

Impact Ionization
The generation rate of electron-hole pairs due to the carrier impact ionization (II)
is generally modeled as [5]:

| | | |
| | | |
= + (1.58)

where and are electron and hole ionization coecients, related with electri-
cal eld, material and temperature.

Three models are implemented in Genius to calculate the ionization coecient.

Selberherr Model

Selberherr gives an empirical formula [2], based on the expression derived by

Chynoweth[18]:
Crit ,
,

, = , ( ) exp (1.59)
[ ( , ) ]

where , is the magnitude of driving elds. When EdotJ model is used, ,

can be given by:

= (1.60.a)
| |
| |

= (1.60.b)
| |
| |

and for GradQf model:

= || || (1.61.a)
| |
= | | (1.61.b)
| |

Crit

where , = , for which , are the optical-phonon mean free paths for
,
electrons and holes given by:

26 Genius Device Simulator

Generation Model Physics in Genius Device Simulator

op
( ) = ,0 tanh (1.62.a)
( 2 )
op
( ) = ,0 tanh (1.62.b)
( 2 )

in the above expressions, is the optical-phonon energy. ,0 and ,0 are the

phonon mean free paths for electrons and holes at 300 K.

The temperature dependent factors and are expressed as:

= ,0 + ,1 + ,2 2 (1.63.a)
= ,0 + ,1 + ,2 2 (1.63.b)

The Selberherr model is the default avalanche model for many materials in Ge-
nius. One can also explicitly load it with the option Selberherr in the PMI
statements.The parameters used for Selberherr model are listed in Table 1.13,
p. 27.

Symbol Parameter Unit Silicon GaAs Ge

,0 LAN300 cm 1.04542e-6 3.52724e-6 6.88825e-7
,0 LAP300 cm 6.32079e-7 3.67649e-6 8.39505e-7
EXN.II - 1.0 1.6 1.0
EXP.II - 1.0 1.75 1.0
op OP.PH.EN eV 6.3e-2 3.5e-2 3.7e-2
,0 N.IONIZA cm1 7.030e5 2.994e5 1.55e7
,1 N.ION.1 cm1 0.0 0.0 0.0
,2 N.ION.2 cm1 0.0 0.0 0.0
,0 P.IONIZA cm1 1.528e6 2.215e5 1e7
,1 P.ION.1 cm1 0.0 0.0 0.0
,2 P.ION.2 cm1 0.0 0.0 0.0

Table 1.13 Default values of Impact Ionization model parameters

van Overstraeten - de Man model

The model also uses the ionization coecient derived by Chynoweth [18]

= exp , (1.64)
( )

where

Genius Device Simulator 27

Physics in Genius Device Simulator Generation Model

op
tanh
( 20 )
= . (1.65)
op
tanh
( 2 )

The van Overstraeten - de Man model uses two sets of values for parameter and
at high and low electric eld. The threshold for the switch is 0 .

The parameters are listed in Table 1.14, p. 28.

The van Overstraeten - de Man model is can be selected with the vanOverstraetendeMan
option in the PMI command. It is the default model for 4H-SiC, InN, InAs and
InSb materials.

Parameter Parameter Unit 4H-SiC:n 4H-SiC:p

AN.II.LO / AP.II.LO cm1 4.2 105 4.2 105
AN.II.HI / AP.II.HI cm1 4.2 105 4.2 105
BN.II.LO / BP.II.LO V/cm 1.67 107 1.67 107
BN.II.HI / BP.II.HI V/cm 1.67 107 1.67 107
0 E0N.II / EOP.II V/cm 4 105 4 105
op EN.OP.PH / EP.OP.PH eV 1.0 1.0

Table 1.14 van Overstraeten - de Man Impact Ionization model parameters

Valdinoci Model

GENIUS has another Valdinoci model for silicon device which has been reported
to produce correct temperature dependence of breakdown voltage of junction diodes
as high as 400 C [19]. It can be loaded with the Valdinoci option in the PMI
statements.

The electron impact ionization rate for Valdinoci model reads:

= (1.66)
( )
( ) + ( ) exp
( + ( ) )

where

( ) = + (1.67.a)
( ) = exp ( ) (1.67.b)
( ) = + + 2 (1.67.c)

28 Genius Device Simulator

Generation Model Physics in Genius Device Simulator

( ) = + + 2 (1.67.d)

Similar expressions hold for holes. The parameters for Valdinoci model are listed
in Table 1.15, p. 29.

Parameter Silicon:n Parameter Silicon:p Unit

A0N 4.3383 A0P 2.376 V
A1N 2.42 1012 A1P 0.01033 V K
A2N 4.1233 A2P 1.0 -
B0N 0.235 B0P 0.17714 V
B1N 0.0 B1P 2.178 103 K1
C0N 1.6831 104 C0P 0.0 V cm1
C1N 4.3796 C1P 9.47 103 V cm1 K
C2N 1.0 C2P 2.4924 -
C3N 0.13005 C3P 0.0 V cm1 K2
D0N 1.233735 106 D0P 1.4043 106 V cm1
D1N 1.2039 103 D1P 2.9744 103 V cm1 K1
D2N 0.56703 D2P 1.4829 V cm1 K2

Table 1.15 Default values of Valdinoci Impact Ionization model parameters

Genius Device Simulator 29

Physics in Genius Device Simulator Generation Model

Band-to-band Tunneling
Kane's model The carrier generation by band-band tunneling is also considered by Genius,
which can be expressed as [20][21]:

2 3/2
= A.BTBT exp B.BTBT (1.68)
( )

where is the magnitude of electrical eld.

30 Genius Device Simulator

High Energy Particles Physics in Genius Device Simulator

High Energy Particles

As a heavy ion passes through the device, it will interact with some silicon atoms
and transfer energy to the semiconductor lattice, which generates electron-hole
pairs along it trajectory. The simulation of high-energy particle and the energy
deposition in semiconductor can be simulated with the Geant4 or other Monte
Carlo simulation tool. On the other hand, the generation of electron-hole pairs
and the eects to the device behavior must be simulated in a 3D device simulator.

Assuming the proton hit the diode at = 0 s, and the electron-hole generation rate
follows a Gaussian time dependence, with a maximum at max , and a characteristic
time , the carrier generation rate can be calculated by

( max )2
= 0 exp . (1.69)
[ 2 2 ]

On the other hand, the Monte Carlo simulators such as Geant4 provides the total
energy deposition data, which relate to by

( max )2
= 0 exp d. (1.70)
[ 2 2 ]
0

where is the average energy loss for each generated electron-hole pair. We there-
fore have the normalization factor
2
0 = . (1.71)

Genius Device Simulator 31

Physics in Genius Device Simulator Fermi-Dirac Statistics

Fermi-Dirac Statistics
In general, the electron and hole concentrations in semiconductors are dened by
Fermi-Dirac distributions and density of states:

= 1/2 ( ) (1.72.a)
= 1/2 ( ) (1.72.b)

The and are dened as follows:

= = 1/2 1 (1.73.a)
( )

= = 1/2 1 (1.73.b)
( )

where and are the electron and hole Fermi energies. The relationship of
Fermi energy and Fermi potential is = , = .

Evaluate Inverse 1/2 1 is the inverse Fermi integral of order one-half. Joyce and Dixon gives its
Fermi Integral approximation analytic expression in the year of 1977 [22], which can be given
by:

log() + + 2 + 3 + 4 < 8.463

1/2 1 () = 3
3/4
2
1/2

otherwise
[( 4 )
6]

= 0.35355339059327379 (1.74.a)
= 0.0049500897298752622 (1.74.b)
= 1.4838577128872821 104 (1.74.c)
= 4.4256301190009895 106 (1.74.d)

In the GENIUS code, the and are derived from carrier concentration by
Joyce-Dixon expression.

For convenience, we introduce oowing two parameters as referred by [23]:

1/2 ( )
= (1.75.a)
exp ( )
1/2 ( )
= (1.75.b)
exp ( )

32 Genius Device Simulator

Fermi-Dirac Statistics Physics in Genius Device Simulator

The carrier concentration for Fermi statistics and Boltzmann statistics can be de-
scribed uniformly by:

= exp ( ) (1.76.a)
= exp ( ) (1.76.b)

where = = 1 for Boltzmann statistics, and less than 1.0 for Fermi statistics.

DD Equation with Consider the drift-diusion current Equation (1.5), p. 2 when the carrier satises
Fermi Statistics Fermi statistics and forces zero net current in equilibrium state, one can get the
modied current equation, for which the Einstein relationship:

= (1.77.a)


= (1.77.b)

should be replaced by:

= / (1.78.a)
1/2 ( ) 1/2 ( )

= 1/2 ( ) /1/2 ( ) (1.78.b)

where 1/2 is the Fermi integral of order minus one-half. The corresponding
current equation for electrons is

= ( + ) (1.79)

where

1/2 ( )
= (1.80)
1/2 ( )

The Fermi integral has an useful property:

d
() = 1 () (1.81)
d

From the above property, one can derive two useful derivatives:

() = (1.82.a)


() = 1 ) (1.82.b)
(

Genius Device Simulator 33

Physics in Genius Device Simulator Fermi-Dirac Statistics

With the two derivatives, Equation (1.79), p. 33 can be rewritten into the follow-
ing equivalent formula:

= ( + (ln )) (1.83)

The last term is the modication to Einstein relationship, which can be combinated
into potential term. As a result, the current Equation (1.5), p. 2 keeps unchanged,
but the eective driving force should be modied as:

1
= (ln( ) ln( 3/2 )) (ln ) (1.84)

The same formula exists for holes:

1
= + (ln( ) ln( 3/2 )) + (ln ) (1.85)

As a conclusion, when Fermi statistics is considered, the formula of DD method

keeps unchanged, only an extra potential term should be considered. However,
Fermi statistics also eect the implement of Ohmic boundary condition, please
refer to [[TODO]]

34 Genius Device Simulator

 CHAPTER

2 Genius Command Reference

Introduction

Format of Input Card

GENIUS code takes its input from a user specied card le. Each line of the
card le is a particular statement, identied by the rst keyword on the card. The
remaining parts of the line are the parameters of that statement. The statement has
the format as follow:

KEYWORD [parameters]

The words on a line should be separated by blanks or tabs. If more than one line
of input is necessary for a particular statement, it may be continued on subsequent
lines by placing a backslash sign \ as the last non-blank character on the current
line.

GENIUS code parses the input le by GNU ex and bison. It rst read the pattern
le located at $GENIUS_DIR/lib/genius.key which provides the keyword
names and parameter attributes before the parse work. After that, the grammar of
user provide le is matched by this pattern. With this exible mechanism, adding
new statement to GENIUS is fairly easy.

Parameters may be one of ve types: oat, integer, bool, string or enumeration.

The oat point number supports C style double precision real number. The integer
value also supports C style integer. The bool value can be True, On, False and O.
String value is made up of lower line, dot, blank, number and alpha characters.
The string should not begin with number; quotation marks are only needed when
string contains blank. The enumerate value is a string with its value in a limited
range which is predened in the pattern le. At last, the length of string (including
enumerate string) is limited to 80 characters.

All the parameter specication has the same format as

parameter_name = [number|integer|bool|string|enumeration]

In the card descriptions, keywords, parameters, enumerate strings are not case
sensitive. And their names does not need to be typed in full; only enough char-
acters to ensure unique identication is necessary. However user input strings is

Genius Device Simulator 35

Genius Command Reference Introduction

case sensitive, because le name may be specied by the string. Comments must
begin with '#' and can be either an separated line or locate at the end of current
statement.

The sequence Most of the cards GENIUS used are sequence insensitive. The order of occurrence
of input deck of cards is signicant in only two cases. The mesh generation cards must have the
right order, or it can't work properly. And GENIUS will execute the 'driven' cards
sequentially. So the placement order of 'driven' cards will aect simulation result.

36 Genius Device Simulator

Introduction Genius Command Reference

Statement Description Format

Syntax of The following special characters are used in the formatted parameter list:
Parameter Lists
slanted shape num - value of parameter in the indicated type
Square brackets [ ] - optional group
Vertical bar | - alternate choice
Parentheses ( ) - group hierarchy
Braces { } - group hierarchy with high level

Value Types Besides some string parameters which have xed values, most of the parameters
need a user dened value. A lower case letter in angle brackets represents a value
of a given type. The following types of values are represented:

num - double precision numerical value

int - integer value
bool - boolean value
str - string value
enum - enumeration value
a-prex - array of values

Genius Device Simulator 37

Genius Command Reference Global Specication

Global Specication

GLOBAL
The GLOBAL command denes some global simulation parameters such as unit
scale and ambient temperature.

GLOBAL
[ TExternal=num ] [ DopingScale=num ] [ Orientation=int ]
{ [ Z.Width=num ] | [ CylindricalMesh=bool ] }
[ ResistiveMetal=bool ]

Parameters

Parameter Type Description Default Unit

Global Simulation Parameters

TExternal numerical Ambient temperature external to the simulated device. Synonym: 300 K
LatticeTemp
DopingScale numerical Doping concentration normalization factor. Physical quantities are 1e18 cm3
normalized based on this factor to reduce numerical errors. Usually,
a good choice of this value is the highest doping concentration in the
device, although in some cases lower values are more desirable.
Z.Width numerical For a 2D mesh, the width of the device in the z-direction. Internally 1.0 m
calculated current densities are multiplied by this width to produce
the terminal currents. This parameter only aect 2D mesh.
CylindricalMesh bool For a 2D mesh, Genius will consider it as cylindrically symmetric. false none
This parameter only aect 2D mesh. The electrical current output is
integrated along the angular direction thus in the unit of Amperes.
Orientation integer The crystalline orientation of the silicon substrate. Only 100, 110, 100 none
and 111 are recognized.
ResistiveMetal bool Flag to enable resistive metal in device structure, which is introduced false none
from version 1.7.0

Description

When several GLOBAL commands are present in the input le, and a parameter is

38 Genius Device Simulator

Global Specication Genius Command Reference

set multiple times, only the last value takes eect.

Example

# device width in Z dimension. Unit:um

GLOBAL Z.Width = 2

# specify initial temperature of device. Unit:K

GLOBAL LatticeTemp = 3e2

# set carrier scale reference value

GLOBAL DopingScale = 1e16

# enable resistive metal

GLOBAL ResistiveMetal = true

Genius Device Simulator 39

Genius Command Reference Mesh Generation

Mesh Generation
GENIUS code has exible mesh data structure which support various shapes of
2D and 3D elements. The only limit is the element should have circum-circle
for 2D or circum-sphere for 3D to meet nite volume method used by GENIUS.
The supported element shapes can be triangle and quadrangle for 2D, tetrahedron,
prism and hexahedron for 3D. Please note that the triangle and tetrahedron can
be any shape; however, quadrangle, prism and hexahedron should meet the in
circuit or in sphere requirement. The GENIUS solver is designed as element shape
independent, which means using mixed shapes of elements in mesh is possible.

Figure 2.1 2D elements

Figure 2.2 3D elements

GENIUS oers several build-in mesh generators to build both 2D and 3D mesh for
semiconductor device. At present, they can only generate simulation mesh with
unique mesh element. However, it is possible to import device model girded with
hyper-mesh from third party mesh generator.

Coordinate System All the mesh generators use Cartesian coordinate system, in which the top horizon-
tal line has the minimal y coordinate, left vertical line has the minimal x coordinate
and the front vertical line has the minimal z coordinate. In general, y coordinate
is the direction vertical to silicon wafer.

Mesh Statements The device mesh is specied by a series of mesh statements. They are listed as
required input order:

40 Genius Device Simulator

Mesh Generation Genius Command Reference

MESH
X.MESH
Y.MESH
Z.MESH
ELIMATE (optional)
SPREAD or SPREAD3D (optional)
REGEION
FACE (optional)

Please note that the order in which statements appear is important. Changing the
order changes the results. Generally, a mesh is specied by the following steps:

Tell GENIUS which mesh generator will be used. (MESH)

The mesh begins as a set of (nonuniformly) spaced x-, y- and z- lines compris-
ing a simple hexahedron (X.MESH, Y.MESH and Z.MESH).
Mesh lines may be terminated inside the device, and redundant nodes removed
from the grid (ELIMINATE).
The hexahedron can be distorted to track non-planar geometry or match the
doping prole, although strongly non-planar structures are dicult to treat in
this way (SPREAD, SPREAD3D).
Material regions can be specied as a union of hexahedrons, completing the
mesh specication (REGION).
Boundary faces can be specied as surface rectangles when necessary (FACE).

MESH
This statement indicates the beginning of the mesh generation behavior.

MESH
Type=(S_Tet4|S_Prism6|S_Hex8|S_Tri3|S_Quad4)
[ Triangle=str ] [ Tetgen=str ]

Parameters

Parameter Type Description Default Unit

Type enum Indicates which mesh generator is to be used. It can be one of none none
S_Tet4, S_Prism6, S_Hex8, S_Tri3 and S_Quad4.
Triangle string The parameters passed to triangular mesh generator for S_Prism6 pzADq30Q none
and S_Tri3 meshes. The triangular mesher recognizes the same
set of parameters as the Triangle program. The detailed descrip-
tion of the available options can be found at Triangle's home page
http://www.cs.cmu.edu/~quake/triangle.html.

Genius Device Simulator 41

Genius Command Reference Mesh Generation

Parameter Type Description Default Unit

TetGen string The parameters passed to the tetrahedral mesh generator Tetgen pzAYq1.5 none
code, which is used for S_Tet4 meshes. The detailed descrip-
tion of this string can be found at Tetgen's home page http://tet-
gen.berlios.de.
File.Prefix string File name prex of the mesh les to be loaded in the simulator. Cur- none none
rently only works with S_Tet4 meshes generated by Tetgen. For a
le prex foo, the les foo.node, foo.face and foo.ele will
be read for node, face, and element data of the mesh, respectively.

Description

All simulations in Genius are performed in a discrete mesh. The available types of
meshes are enumerated above, and illustrated in Figure 2.1, p. 40 and Figure 2.2,
p. 40.

There are two alternative ways to generate the mesh structures. The user can use
the X.MESH, Y.MESH and Z.MESH commands to specify the position of mesh
lines, and have Genius to call the appropriate mesh generator programs behind
the scene.

Alternatively, the user can dene the device geometry with external tools, pro-
duce the mesh with an external mesh generator, and commands Genius to import
from les containing the mesh structure. Currently Genius can import mesh les
generated by Tetgen for tetrahedral S_Tet4 meshes.

Example

MESH Type=S_Tri3
MESH Type=S_Tet4 Tetgen="pzAq1.5"
MESH Type=S_Tet4 File.Prefix="foo"

42 Genius Device Simulator

Mesh Generation Genius Command Reference

X.MESH Y.MESH and Z.MESH

The X.MESH, Y.MESH and Z.MESH cards specify the location of lines of nodes in
a hexahedral mesh (rectangular mesh for 2D) . The original mesh can be modied
by following mesh cards like ELIMINATE and SPREAD or SPREAD3D.

X.MESH
{ Width=num | ( X.Min=num X.Max=num ) }
{ N.Spaces=int [Ratio=num] | H1=num [H2=num] }
[Min.Space=num]

Y.MESH
{ Depth=num | ( Y.Min=num Y.Max=num ) }
{ N.Spaces=int [Ratio=num] | H1=num [H2=num] }
[Min.Space=num]

Z.MESH
{ Width=num | ( Z.Min=num Z.Max=num ) }
{ N.Spaces=int [Ratio=num] | H1=num [H2=num] }
[Min.Space=num]

Parameters

Parameter Type Description Default Unit

Width numerical The distance of the grid section in x or z direction. This direction is none m
parallel to silicon wafer.
Depth numerical The distance of the grid section in y direction. This direction is none m
vertical to silicon wafer.
X.Min numerical The x location of the left edge of the grid section. synonym: none m
X.Left. When more than one X.Mesh statements exist, the X.Min
location of next mesh edge will follow most right of previous grid
section automatically.
X.Max numerical The x location of the right edge of the grid section. synonym: none m
X.Right.
Y.Min numerical The y location of the top edge of the grid section. synonym: Y.Top. none m
When more than one Y.Mesh statements exist, the Y.Min location
of next mesh edge will follow most bottom of previous grid section
automatically.
Y.Max numerical The y location of the bottom edge of the grid section. synonym: none m
Y.Bottom.
Z.Min numerical The z location of the front edge of the grid section. synonym: none m
Z.Front. When more than one Z.Mesh statements exist, the
Z.Min location of next mesh edge will follow most back of previous
grid section automatically.

Genius Device Simulator 43

Genius Command Reference Mesh Generation

Parameter Type Description Default Unit

Z.Max numerical The z location of the bottom edge of the grid section. synonym: none m
Z.Back.
N.Spaces integer The number of grid spaces in the grid section. 1 none
Ratio numerical The ratio between the sizes of adjacent grid spaces in the grid sec- 1.0 none
tion. Ratio should usually lie between 0.667 and 1.5.
H1 numerical The size of the grid space at the begin edge of the grid section. none m
H2 numerical The size of the grid space at the end edge of the grid section. none m
Min.Space numerical The size of the minimum allowed grid spacing in this grid section. 1e-4 m

Description

[give pictures here]

Example

X.MESH X.MIN=0.0 X.MAX=0.50 N.SPACES=8

Y.MESH DEPTH=0.1 N.SPACES=8 RATIO=0.8
Y.MESH DEPTH=0.1 N.SPACES=20
Y.MESH DEPTH=0.6 H1=0.005 H2=0.050
Z.MESH WIDTH=1.0 N.SPACES=5

44 Genius Device Simulator

Mesh Generation Genius Command Reference

ELIMINATE
The ELIMINATE statement eliminates mesh points along planes in a hexahedral
grid over a specied volume. This statement is useful for eliminating nodes in re-
gions of the device structure where the grid is more dense than necessary. Points
along every second line in the chosen direction within the chosen range are re-
moved, except the rst and last line. Successive eliminations of the same range
remove points along every fourth line, eighth line, and so on.

ELIMINATE
{ Direction = (XNorm | YNorm | ZNorm) }
[ {X.Min=num | IX.Min=int} ] [ {X.Max=num | IX.Max=int} ]
[ {Y.Min=num | IY.Min=int} ] [ {Y.Max=num | IY.Max=int} ]
[ {Z.Min=num | IZ.Min=int} ] [ {Z.Max=num | IZ.Max=int} ]

Parameters

Parameter Type Description Default Unit

Direction enum Species that lines of nodes in direction of X, Y or Z are eliminated. none none
X.Min numerical The minimum x location of the hexahedron volume in which nodes XMin m
are eliminated. The default value is the minimum x-coordinates in
the device. synonym: X.Left.
X.Max numerical The maximum x location of the hexahedron volume in which nodes XMax m
are eliminated. The default value is the maximum x-coordinates in
the device. synonym: X.Right.
IX.Min integer The minimum x node index of the hexahedron volume in which 0 none
nodes are eliminated. synonym: IX.Left.
IX.Max integer The maximum x node index of the hexahedron volume in which IXMax-1 none
nodes are eliminated. synonym: IX.Right.
Y.Min numerical The minimum y location of the hexahedron volume in which nodes YMin m
are eliminated. The default value is the minimum y-coordinates in
the device. synonym: Y.Top.
Y.Max numerical The maximum y location of the hexahedron volume in which nodes YMax m
are eliminated. The default value is the maximum y-coordinates in
the device. synonym: Y.Bottom.
IY.Min integer The minimum y node index of the hexahedron volume in which 0 none
nodes are eliminated. synonym: IY.Top.
IY.Max integer The maximum y node index of the hexahedron volume in which IYMax-1 none
nodes are eliminated. synonym: IY.Bottom.
Z.Min numerical The minimum Z location of the hexahedron volume in which nodes ZMin m
are eliminated. The default value is the minimum Z-coordinates in
the device. synonym: Z.Front.
Z.Max numerical The maximum z location of the hexahedron volume in which nodes ZMax m
are eliminated. The default value is the maximum z-coordinates in
the device. synonym: Z.Back.

Genius Device Simulator 45

Genius Command Reference Mesh Generation

Parameter Type Description Default Unit

IZ.Min integer The minimum z node index of the hexahedron volume in which 0 none
nodes are eliminated. synonym: IZ.Front.
IZ.Max integer The maximum x node index of the hexahedron volume in which IZMax-1 none
nodes are eliminated. synonym: IZ.Back.

Description

Not all the mesh generator support ELIMINATE. The S_QUAD4 and S_HEX8 mesh
generator can not do mesh line eliminate. User should take care of the elimination
process. The excessively elimination may cause bad element quality.

Example

ELIMINATE Direction=YNORM Y.TOP=-1.0

ELIMINATE Direction=XNORM IX.MAX=8

46 Genius Device Simulator

Mesh Generation Genius Command Reference

SPREAD
The SPREAD statements provide a way to adjust the y position of nodes along grid
lines parallel to the x-plan (the disturb is uniform in Z dimension) in a hexahedral
mesh to follow surface and junction contours. The statement is only supported by
S_Tri3, S_Tet4 and S_Prism6 mesh generator.

SPREAD
Location = (Left|Right) Width=num Upper=int Lower=int
[ Enchroach=num]
{ Y.Lower=num | (Thickness=num [Vol.Rat=num]) }
[ Grading=num ]

Parameters

Parameter Type Description Default Unit

Location enum Species which side of the grid is distorted. none none
Width numerical The width of the distorted region measured from the left or right 0 m
edge of the structure.
Upper integer The index of the upper y-grid line of the distorted region. 0 none
Lower integer The index of the lower y-grid line of the distorted region. 0 none
Encroach integer The factor which denes the abruptness of the transition between 1.0 none
distorted and undistorted grid. The transition region becomes more
abrupt with smaller Encroach factors. The minimum allowed value
is 0.1.
Y.Lower numerical The vertical location in the distorted region where the line specied none m
by Lower is moved. The grid line specied by Upper does not move
if this parameter is specied.
Thickness numerical The thickness of the distorted region. Specifying Thickness usu- none m
ally causes the positions of both the Upper and Lower grid lines to
move.
Vol.Rat numerical The ratio of the displacement of the lower grid line to the net change 0.44 none
in thickness. If Vol.Rat is 0, the location of the lower grid line does
not move. If Vol.Rat is 1, the upper grid line does not move.
Grading numerical The vertical grid spacing ratio in the distorted region between the 1.0 none
y-grid lines specied with Upper and Lower. The spacing grows
or shrinks by Grading in each interval between lines. Grading
should usually lie between 0.667 and 1.5.

Example

SPREAD Location=Left Width=0.625 Upper=0 Lower=2 \

Thickness=0.1

SPREAD Location=Right Width=0.625 Upper=0 Lower=2 \

Thickness=0.1

Genius Device Simulator 47

Genius Command Reference Mesh Generation

SPREAD3D
The SPREAD3D statements also provide a way to adjust the y position of nodes
along grid lines parallel to the xz-plan in a hexahedral mesh to follow surface and
junction contours. The statement is only supported by S_Tet4 mesh generator.

SPREAD3D
[ {X.Min=num | IX.Min=int} ] [ {X.Max=num | IX.Max=int} ]
[ {Z.Min=num | IZ.Min=int} ] [ {Z.Max=num | IZ.Max=int} ]
Upper=int Lower=int [ Enchroach=num]
{ Y.Lower=num | (Thickness=num [Vol.Rat=num]) }
[Grading=num]

Parameters

Parameter Type Description Default Unit

X.Min numerical The minimum x location of the area in which nodes are disturbed. XMIN m
synonym: X.Left.
X.Max numerical The maximum x location of the area in which nodes are disturbed. XMAX m
synonym: X.Right.
IX.Min integer The minimum x node index of the area in which nodes are disturbed. 0 none
synonym: IX.Left.
IX.Max integer The maximum x node index of the area in which nodes are disturbed. IXMAX-1 none
synonym: IX.Right.
Z.Min numerical The minimum z location of the area in which nodes are disturbed. ZMIN m
synonym: Z.Front.
Z.Max numerical The maximum z location of the area in which nodes are disturbed. ZMAX m
synonym: Z.Back.
IZ.Min integer The minimum z node index of the area in which nodes are disturbed. 0 none
synonym: IZ.Front.
IZ.Max integer The maximum z node index of the area in which nodes are disturbed. IZMAX-1 none
synonym: IZ.Back.
Upper integer The index of the upper y-grid line of the distorted region. 0 none
Lower integer The index of the lower y-grid line of the distorted region. 0 none
Encroach integer The factor which denes the abruptness of the transition between 1.0 none
distorted and undistorted grid. The transition region becomes more
abrupt with smaller Encroach factors. The minimum allowed value
is 0.1.
Y.Lower numerical The vertical location in the distorted region where the line specied none m
by Lower is moved. The grid line specied by Upper does not move
if this parameter is specied.
Thickness numerical The thickness of the distorted region. Specifying Thickness usu- none m
ally causes the positions of both the Upper and Lower grid lines to
move.

48 Genius Device Simulator

Mesh Generation Genius Command Reference

Parameter Type Description Default Unit

Vol.Rat numerical The ratio of the displacement of the lower grid line to the net change 0.44 none
in thickness. If Vol.Rat is 0, the location of the lower grid line does
not move. If Vol.Rat is 1, the upper grid line does not move.
Grading numerical The vertical grid spacing ratio in the distorted region between the 1.0 none
y-grid lines specied with Upper and Lower. The spacing grows
or shrinks by Grading in each interval between lines. Grading
should usually lie between 0.667 and 1.5.

Example

SPREAD3D X.MIN=1.0 X.MAX=2.0 Z.MIN=0.0 Z.MAX=1.0 \

Upper=0 Lower=2 Thickness=0.1

Genius Device Simulator 49

Genius Command Reference Mesh Generation

REGION
The REGION statement denes the location of materials in the mesh. Currently,
GENIUS supports following materials: Vacuum; semiconductor material includ-
ing Si, Ge, GaAs, 3C-SiC, Si 1-xGe x, Al xGa 1-xAs, In xGa 1-xAs, InP, InAs, InN,
InSb, Hg (1-x)Cd (x)Te; insulator material including Air, SiO 2, Si 3N 4 and electrode
region including Elec, Al, Cu, Au, Ag and PolySi.

REGION
Shape=(Rectangle|Ellipse) Label=str Material=str
[ {X.Min=num | IX.Min=int} ] [ {X.Max=num | IX.Max=int} ]
[ {Y.Min=num | IY.Min=int} ] [ {Y.Max=num | IY.Max=int} ]
[ {Z.Min=num | IZ.Min=int} ] [ {Z.Max=num | IZ.Max=int} ]

Parameters

Parameter Type Description Default Unit

Shape enum Species the shape of the region. Can be Rectangle or Ellipse. Rectangle none
Label string Species the identier of this region, limited to 12 chars. none none
Material string Species the material of the region. none none
X.Min numerical The minimum x location of the region. synonym: X.Left. XMIN m
X.Max numerical The maximum x location of the region. synonym: X.Right. XMAX m
IX.Min integer The minimum x node index of the region. synonym: IX.Left. 0 none
IX.Max integer The maximum x node index of the region. synonym: IX.Right. IXMAX-1 none
Y.Min numerical The minimum y location of the region. synonym: Y.Top. YMIN m
Y.Max numerical The maximum y location of the region. synonym: Y.Bottom. YMAX m
IY.Min integer The minimum y node index of the region. synonym: IY.Top. 0 none
IY.Max integer The maximum y node index of the region. synonym: IY.Bottom. IYMAX-1 none
Z.Min numerical The minimum z location of the region. synonym: Z.Front. ZMIN m
Z.Max numerical The maximum z location of the region. synonym: Z.Back. ZMAX m
IZ.Min integer The minimum z node index of the region. synonym: IZ.Front. 0 none
IZ.Max integer The maximum z node index of the region. synonym: IZ.Back. IZMAX-1 none

Description

Several regions can be dened one by one. But users should be careful that regions
can't get cross each other. The situations showed in Figure 2.3, p. 51(A) and (B)
are allowed, but (C) will break the mesh generator of genius.

50 Genius Device Simulator

Mesh Generation Genius Command Reference

Figure 2.3 Multi-Region denition.

When the background grid line is disturbed by SPREAD statement, it is something

dicult for specifying region bounding with grid line location. One can use grid
line index instead of grid line location here.

Example

REGION Label=Si1 Material=Si Y.TOP= 0.0 Y.BOTTOM=-0.100

REGION Label=SiGe1 Material=SiGe Y.TOP=-0.1 Y.BOTTOM=-0.125

Genius Device Simulator 51

Genius Command Reference Mesh Generation

FACE
Face is a group of facet elements (triangle and quadrangle for 3D, and edge for 2D)
which have the same attribute. This statement species the label of a special face.
User can assign the face with a special boundary condition type by BOUNDARY
statement.

FACE
Label=str
{ Location = (Top|Bottom|Left|Right|Front|Back) |
( Direction=(XNorm|YNorm|ZNorm) ( X=num|Y=num|Z=num ) )
}
[ {X.Min=num | IX.Min=int} ] [ {X.Max=num | IX.Max=int} ]
[ {Y.Min=num | IY.Min=int} ] [ {Y.Max=num | IY.Max=int} ]
[ {Z.Min=num | IZ.Min=int} ] [ {Z.Max=num | IZ.Max=int} ]

Parameters

Parameter Type Description Default Unit

Label string Species the identier of this face, limited to 80 chars. none none
Location enum Species which side the face lies along. none none
Direction enum Species the dimensional orientation of the face. none none
X numerical Species the X coordinate of the face parallel to YZ plane. 0.0 m
Y numerical Species the Y coordinate of the face parallel to XZ plane. 0.0 m
Z numerical Species the Z coordinate of the face parallel to XY plane. 0.0 m
X.Min numerical The minimum x location of the face. synonym: X.Left. XMIN m
X.Max numerical The maximum x location of the face. synonym: X.Right. XMAX m
IX.Min integer The minimum x node index of the face. synonym: IX.Left. 0 none
IX.Max integer The maximum x node index of the face. synonym: IX.Right. IXMAX-1 none
Y.Min numerical The minimum y location of the face. synonym: Y.Top. YMIN m
Y.Max numerical The maximum y location of the face. synonym: Y.Bottom. YMAX m
IY.Min integer The minimum y node index of the face. synonym: IY.Top. 0 none
IY.Max integer The maximum y node index of the face. synonym: IY.Bottom. IYMAX-1 none
Z.Min numerical The minimum z location of the face. synonym: Z.Front. ZMIN m
Z.Max numerical The maximum z location of the face. synonym: Z.Back. ZMAX m
IZ.Min integer The minimum z node index of the face. synonym: IZ.Front. 0 none
IZ.Max integer The maximum z node index of the face. synonym: IZ.Back. IZMAX-1 none

Description

The FACE statement can specify facet elements not only on the boundary of the
mesh, but also inside the mesh, i.e. at the interface of two regions. Here gives
the rules about how GENIUS dealing with of faces. The faces specied by FACE

52 Genius Device Simulator

Mesh Generation Genius Command Reference

statement always has the highest priority. They will have the label as FACE state-
ment declared. For the remaining faces, the interface faces between two regions
will be assigned by name1_to_name2 in which the name1 and name2 are the la-
bels of the two regions in alphabetic order. The remain faces of a region will be
assigned by name_Neumann and the name is the label of the region. There are
some limitations about S_Prism6 mesh generator. Face can not be dened along
Z direction, i.e. on front or back plane.

Example

FACE Label=Anode Direction=Horizontal X.MIN=0.0 X.MAX=1.0

\
Z.MIN=0.0 Z.MAX=3.0 Y=0.0

FACE Label=Cathode Direction=Horizontal X.MIN=0.0 X.MAX=3.0

\
Z.MIN=0.0 Z.MAX=3.0 Y=3.0

FACE Label=Anode Location=TOP X.MIN=0.0 X.MAX=1.0

FACE Label=Cathode Location=BOTTOM

Genius Device Simulator 53

Genius Command Reference Mesh Generation

REFINE
The REFINE statement allows renement of a coarse mesh. GENIUS supports
three types of mesh renement: uniform, conform and hierarchical mesh rene-
ment. At present, the solution is destroyed after the mesh renement. Thus user
should ensure that there are DOPING and probably MOLE statements exist in the
same input le so that device doping prole and mole distribution can be rebuild
on the new mesh.

REFINE.Uniform [Step=int]

REFINE.Conform
Variable = ( Doping | Potential | Electron | Hole |
Temperature | E.Temp | H.Temp | QFN | QFP |
E.Field | Net.Carrier | Net.Charge |
Optical.Gen | Particle.Gen )
Evaluation = ( Gradient | Quantity )
Measure=( Linear| SignedLog ) [ Region=str ]
{ error.fraction=num | cell.fraction=num |
error.threshold=num }

REFINE.Hierarchical
Variable = ( Doping | Potential | Electron | Hole |
Temperature | E.Temp | H.Temp | QFN | QFP |
E.Field | Net.Carrier | Net.Charge )
Evaluation = ( Gradient | Quantity )
Measure=( Linear| SignedLog ) [ Region=str ]
{ error.refine.fraction=num | cell.refine.fraction=num |
error.refine.threshold=num }
[ error.coarsen.fraction=num | cell.coarsen.fraction=num |
error.coarsen.threshold=num ]

Parameters

Parameter Type Description Default Unit

Uniform Mesh Rene

Step integer how many steps should the uniform renement repeats. none none

General Mesh Renement Parameters

54 Genius Device Simulator

Mesh Generation Genius Command Reference

Parameter Type Description Default Unit

Variable enum the mesh renement criteria variable. Can be one of: Doping, none none
Potential, Electron, Hole, Temperature, E.Temp, H.Temp,
QFN, QFP, E.Field, Net.Carrier and Net.Charge.
Evaluation enum Species whether the variable quantity or its gradient should be used Gradient none
in renement critiria.
Measure enum Species that renement is based on the original value or logarithm Linear none
of the specied quantity.
Region string The name of the region over which renement takes place. Elements none none
in other regions may be rened for mesh conform or well-shaped el-
ements. User can specify several regions by multi Region parame-
ters.

Conformal Mesh Renement

Error.Fraction numerical When the cell's error exceeds error.fraction*max_error, it 0.3 none
will be rened..
Error.Threshold numerical When the cell's error exceeds error.threshold, it will be re- 0.1 none
ned.
Cell.Fraction numerical The cell.fraction*(total cell number) of cells with most 0.3 none
error will be rened.
X.Min numerical The minimum x coordinate of the region that will be considered for 0 m
renement. synonym: X.Left.
X.Max numerical The maximum x coordinate of the region that will be considered for 0 m
renement. synonym: X.Right.
Y.Min numerical The minimum y coordinate of the region that will be considered for 0 m
renement. synonym: Y.Top.
Y.Max numerical The maximum y coordinate of the region that will be considered for 0 m
renement. synonym: Y.Bottom.
Z.Min numerical The minimum z coordinate of the region that will be considered for 0 m
renement. synonym: Z.Front.
Z.Max numerical The maximum z coordinate of the region that will be considered for 0 m
renement. synonym: Z.Back.

Hierarchical Mesh Renement/Coarsening

Error.Refine.Fr numerical When the cell's error exceeds error.refine.fraction*max_error, 0.3 none
it will be rened. synonym: Error.Rene.Fraction
Error.Coarsen.Fr numerical When the cell's error is less than error.coarsen.fraction*max_error0.0
, none
it will be coarsened. synonym: Error.Coarsen.Fraction
Error.Refine.Th numerical When the cell's error exceeds error.refine.threshold, it will 0.1 none
be rened. synonym: Error.Refine.Threshold
Error.Coarsen.Th numerical When the cell's error is less than error.coarsen.threshold, it 0.1 none
will be coarsened. synonym: Error.Coarsen.Threshold
Cell.Refine.Fr numerical The cell.refine.fraction*(total cell number) of cells 0.3 none
with most error will be rened. synonym: Cell.Refine.Fraction

Genius Device Simulator 55

Genius Command Reference Mesh Generation

Parameter Type Description Default Unit

Cell.Coarsen.Fr numerical The cell.coarsen.fraction*(total cell number) of cells 0.3 none
with least error will be coarsened. synonym: Cell.Coarsen.Fraction

Description

REFINE.Uniform is only intended for debugging. Only TRI3 and TET4 mesh
can do REFINE.Conform. And the conform rene process requires the mesh
should be generated by corresponding mesh generator which is dened in the
same input le. However, this is the most stable mesh renement procedure. The
REFINE.Hierarchical is the new mesh rene and coarsen mechanism intro-
duced into GENIUS. It supports all the elements and regardless mesh generators.
This method will generate hanging node on the side of elements with dierent
renement level. User should ensure that the hanging node be far away from pn
junction of the device. Or convergence problem may occur. Generally, rene by
potential has the best eect.

Example

REFINE.U Step=1

REFINE.C Variable=potential cell.fraction=0.3

REFINE.H Variable=potential cell.refine.fraction=0.3 \

cell.coarsen.fraction=0.1

56 Genius Device Simulator

External Sources Genius Command Reference

External Sources
For simulation the transient response of device, GENIUS supports several types
of voltage and current source. The original models of these sources come from
SPICE, a famous circuit simulation program. Several sources may be dened in
one disk le. And the placement of these denitions are not critical. The sources
can be assigned to electrode by ATTACH statement when needed.

VSOURCE
The VSOURCE command denes a voltage source.

VSOURCE
ID=str [ TDelay = num ]
[ Type = (VDC|VSin|VExp|VPulse|VShell) ]

DC Source
{ VConst=num } |

Sinusoidal Source
{ VAmp=num VFreq=num } |

Exponential Source
{ VHi=num VLo=num TRC=num TFD=num TFC=num } |

Pulse Source
{ VHi=num VLo=num Tr=num Tf=num Pw=num Pr=num } |

User-defined Source
{ DLL=str Function=str }

Parameters

Parameter Type Description Default Unit

General Parameters

Type enum Selects the type of voltage source. The following types are available: VDC none
DC source, sinusoidal source, exponential source, pulse source and
user-dened source.
ID string Declares the name of the source. none none

Genius Device Simulator 57

Genius Command Reference External Sources

Parameter Type Description Default Unit

TDelay numerical Delay time before the source voltage is applied. none none

DC Source Parameters

VConst numerical Constant voltage. 0.0 V

Sinusoidal Source Parameters

VAmp numerical Amplitude of the sinusoidal source. 0.0 V

VFreq numerical Frequency of the sinusoidal source none Hz

Exponential Source Parameters

VHi numerical Upper bound of exponential voltage waveform. 0.0 V

VLo numerical Lower bound of exponential voltage waveform. 0.0 V
TRC numerical Rising edge time constant. none s
TFD numerical Delay time before the falling phase starts. The length of the rising none s
phase is thus TFD-TDelay.
TFC numerical Falling edge time constant. none s

Pulse Source Parameters

VHi numerical High-voltage level of the pulses. 0.0 V

VLo numerical Lower-voltage level of the pulses. 0.0 V
Tr numerical Rising edge time constant. 1e-12 s
Tf numerical Falling edge time constant. 1e-12 s
Pw numerical Pulse width. 0.0 s
Pr numerical Repeating period of the pulse train. 0.0 s

User-dened Source Parameters

DLL string Name of the dynamically loadable library le that contains the none none
user-dened source.
Function string Function name of the source in the DLL le. none none

Description

GENIUS supports user dened voltage and current source by loading shared ob-
ject (.so) le. The le which contains a user dened voltage source should have
the function as follow. GENIUS will pass the argument time in the unit of second
to the function vsrc_name and get voltage value in the unit of volt. The current
source function is almost the same except the unit of current is A.

58 Genius Device Simulator

External Sources Genius Command Reference

double vsrc_name(double time)

{
/* calculate the voltage amplitude */
return vsrc_amplitude;
}
double isrc_name(double time)
{
/* calculate the current amplitude */
return isrc_amplitude;
}

The c code should be linked with -shared and -fPIC option as:

gcc -shared -fPIC -o foo.so foo.c -lm

The foo.so le should be put in the same directory as input le.

Example

vsource Type=VDC ID=GND Tdelay=0 Vconst=0

vsource Type=VDC ID=VCC Tdelay=0 Vconst=5

vsource Type=VSIN ID=Vs Tdelay=1e-6 Vamp=0.1 Freq=1e6

vsource Type=VEXP ID=V1 Tdelay=0 TRC=1e-6 TFD=1e-6 \

TFC=1e-6 Vlo=0 Vhi=1

vsource Type=VPULSE ID=V2 Tdelay=0 Tr=1e-9 Tf=1e-9 \

Pw=5e-6 Pr=1e-5 Vlo=0 Vhi=1

vsource Type=VSHELL ID=VGauss DLL=foo.so Func=vsrc_gauss

Genius Device Simulator 59

Genius Command Reference External Sources

ISOURCE
The ISOURCE command denes a voltage source.

ISOURCE
ID=str [ Type = (IDC|ISin|IExp|IPulse|IShell) ]
[ TDelay = num ]

DC Source
{ IConst=num } |

Sinusoidal Source
{ IAmp=num IFreq=num } |

Exponential Source
{ IHi=num ILo=num TRC=num TFD=num TFC=num } |

Pulse Source
{ IHi=num ILo=num Tr=num Tf=num Pw=num Pr=num } |

User-defined Source
{ DLL=str Function=str }

Parameters

Parameter Type Description Default Unit

General Parameters

Type enum Selects the type of current source. The following types are available: VDC none
DC source, sinusoidal source, exponential source, pulse source and
user-dened source.
ID string Declares the name of the source. none none
TDelay numerical Delay time before the source is applied. none none

DC Source Parameters

IConst numerical Constant current. 0.0 V

Sinusoidal Source Parameters

IAmp numerical Amplitude of the sinusoidal source. 0.0 V

IFreq numerical Frequency of the sinusoidal source none Hz

60 Genius Device Simulator

External Sources Genius Command Reference

Parameter Type Description Default Unit

Exponential Source Parameters

IHi numerical Upper bound of exponential current waveform. 0.0 V

ILo numerical Lower bound of exponential current waveform. 0.0 V
TRC numerical Rising edge time constant. none s
TFD numerical Delay time before the falling phase starts. The length of the rising none s
phase is thus TFD-TDelay.
TFC numerical Falling edge time constant. none s

Pulse Source Parameters

IHi numerical High-current level of the pulses. 0.0 V

ILo numerical Lower-current level of the pulses. 0.0 V
Tr numerical Rising edge time constant. 1e-12 s
Tf numerical Falling edge time constant. 1e-12 s
Pw numerical Pulse width. 0.0 s
Pr numerical Repeating period of the pulse train. 0.0 s

User-dened Source Parameters

DLL string Name of the dynamically loadable library le that contains the none none
user-dened source.
Function string Function name of the source in the DLL le. none none

Example

ISource Type=IDC ID=I1 Tdelay=0 Iconst=5

Isource Type=ISIN ID=I2 Tdelay=0 Iamp=0.1 Freq=1e6

Isource Type=IEXP ID=I3 Tdelay=0 TRC=1E-6 TFD=3E-6 \

TFC=1E-6 Ilo=0 Ihi=1

Isource Type=IPULSE ID=I4 Tdelay=0 Tr=1E-9 Tf=1E-9 \

Pw=5E-6 Pr=1E-5 Ilo=0 Ihi=1

Genius Device Simulator 61

Genius Command Reference External Sources

PARTICLE

PARTICLE

From scatter data file

Profile = ( FromFile2D | FromFile3D ) Profile.File=str
[Transform.xx=num] [Transform.xy=num] [Transform.xz=num]
[Transform.yx=num] [Transform.yy=num] [Transform.yz=num]
[Transform.zx=num] [Transform.zy=num] [Transform.zz=num]
[Translate.x=num] [Translate.y=num] [Translate.z=num]

From analytic expression

Profile = Expression LET=num

From particle track

Profile = Track Profile.File=str

[Energy.Eff=num] [t0=num ] [tmax=num] [ t.Char=num ]

Parameters

Parameter Type Description Default Unit

Profile enum Type of particle prole input. none none
Transform numerical The transformation matrix applied to the coordinates imported I none
from the input le. The nal coordinates are obtained from
Transform*location + Translate.
Translate enum The translation vector applied to the input coordinates. 0 none
Energy.Eff numerical The energy loss for each generated electron-hole pair. 3.6 eV
t0 numerical The starting time when high-energy particles penetrates the device. 0.0 s
tmax numerical The time instant when electron-hole pair generation rate reaches its 0.0 s
maximum.
t.Char numerical The variance () of the Gaussian decay of carrier generation rate. 1e-12 s

Description

There are several ways to specify the energy deposit of a particle: by scatter data
le, by analytic express and by particle track.

The input data le must be in plain text, with numerical values separated by white
spaces. Each line contains the coordinates and the energy deposition at that posi-
tion. For 2D prole, each line has 3 values, while for 3D prole, each line has 4

62 Genius Device Simulator

External Sources Genius Command Reference

values. The coordinates are in the unit of m, while the energy deposition is in
the unit of eVm3 .

Examples

PARTICLE Profile=FromFile3D Profile.File="particle.dat" \

tmax=2e-12 t.char=2e-12

Genius Device Simulator 63

Genius Command Reference External Sources

LIGHT
LIGHT
{ SpectrumFile=str | ( Lambda=num Intensity=num [Eta=num]
) }
Profile = ( FromFile2D | FromFile3D )
[ Profile.File=str ] [ SkipLine=num ]
[ LUnit = (m|cm|um|nm) ] [ FUnit = (m|cm|um|nm) ]
[Transform.xx=num] [Transform.xy=num] [Transform.xz=num]
[Transform.yx=num] [Transform.yy=num] [Transform.yz=num]
[Transform.zx=num] [Transform.zy=num] [Transform.zz=num]
[Translate.x=num] [Translate.y=num] [Translate.z=num]

Parameters

Parameter Type Description Default Unit

Profile enum Type of eld prole input. none none
Lambda numerical Wavelength of the monochromatic light. 0.532 m
Intensity numerical Intensity of the monochromatic light. 0.0 W/cm2
Eta numerical Quantum eciency of carrier generation for the monochromatic auto none
light. If not explicitly specied, Genius will automatically calcu-
late the quantum eciency using the bandgap of the material.
SpectrumFile string Filename of the spectrum le. This le should describe the wave- none none
length and intensity at each frequency in the spectrum.
Profile.File enum Filename of the eld solution data le. When a spectrum le is none none
provided, the provided lename extension is appended to the eld
solution lename.
SkipLine integer Skip the rst SkipLine number of lines in the eld data le. 0 none
LUnit enum Length unit used for node coordinates in the eld data le. um none
FUnit enum Length unit used in E-eld value. The voltage unit is always Volt. um none
Transform numerical The transformation matrix applied to the coordinates imported I none
from the input le. The nal coordinates are obtained from
Transform*location + Translate.
Translate enum The translation vector applied to the input coordinates. 0 none

Description

Spectrum le With the spectrum le we can load the eld data at multiple wavelengths in the
same LIGHT command. Each line in the spectrum le should contain the following
columns

A lename extension for the eld data le. Suppose Profile.File is diode1,
and the extension is 3290, Genius will try to load the eld data from the le
diode1.3290.
The wavelength in the unit of micro-meter.

64 Genius Device Simulator

External Sources Genius Command Reference

Intensity density (per unit wavelength) at this wavelength. The unit is W/cm2/m.
Note that this diers from the intensity parameter for monochromatic light.
(Optional) quantum eciency for carrier generation. If not explicitly spec-
ied, Genius will automatically calculate the quantum eciency using the
bandgap of the material.

A sample spectrum le is listed below.

# ext lambda intensity

3290 0.912 0.2
3145 0.954 0.5
3000 1.00 0.4
2855 1.05 0.3
2710 1.11 0.2
2565 1.17 0.2
2420 1.24 0.1

Optical Carrier HFSS assumes by default that the incident wave has E-eld strength of 0 =
Generation 1 V/m, which corresponds to the incident intensity of

1
0 = 0 0 0 2 . (2.1)
2

We need to scale the E-eld in the solution according to the user-specied incident
intensity.

If the user does not supply the quantum eciency value, Genius will calculate
with the following assumption

= oor( ). (2.2)

where the oor function round to largest integral value not greater than the argu-
ment. The carrier generation is nally calculated with

0 ( )
= 2. (2.3)

HFSS places a descriptive heading in each spectrum le, we shall skip that line.

Field Data File The input data le must be in plain text, with numerical values separated by white
spaces. Each line contains the coordinates and the E-eld strength of the EM wave
at that position. For 2D prole, each line has 3 values, while for 3D prole, each
line has 4 values. The unit of the coordinates is specied by LUnit, and the unit

Genius Device Simulator 65

Genius Command Reference External Sources

of the E-eld is specied by FUnit.

Examples

LIGHT SpectrumFile=diode1 Profile.File=diode1 \

Profile=fromfile3d LUnit=m FUnit=m SkipLine=1

66 Genius Device Simulator

External Sources Genius Command Reference

ENVELOP
The ENVELOP command denes an envelop in time, which is used to modulate
light source. The light source can be introduced by LIGHT command and RAYTRACE
command. User can specify the envelop of light source at the SOLVE command.

ENVELOP
ID = str
[ Type = (Uniform|Gaussian|Pulse|Expression|Shell) ]
[ TDelay = num ]

Uniform envelop
{ Amplitude=num } |

Gaussian envelop
{ Amplitude=num T0=num Tao=num} |

Pulse envelop
{ Amplitude.high=num Amplitude.low=num
Tr=num Tf=num Pw=num Pr=num } |

User-defined envelop by expression

{ Expression=str }

User-defined envelop by dynamic loadable library

{ DLL=str Function=str }

Parameters

Parameter Type Description Default Unit

General Parameters

Type enum Selects the type of envelop. The following types are available: Uni- Uniform none
form, Gaussian, Pulse and user-dened envelop.
ID string Declares the name of the envelop. none none
TDelay numerical Delay time before the envelop is applied. none none

Uniform Envelop Parameters

Amplitude numerical Constant amplitude. 1.0 none

Gaussian Envelop Parameters

Genius Device Simulator 67

Genius Command Reference External Sources

Parameter Type Description Default Unit

Amplitude numerical Amplitude of Gaussian waveform. 1.0 none

T0 numerical The time when Gaussian waveform reaches its max. 0.0 s
Tao numerical The characteristic time of Gaussian waveform. 1e-12 s

Pulse Envelop Parameters

Amplitude.high numerical High level of the pulses. 1.0 V

Amplitude.low numerical Lower level of the pulses. 0.0 V
Tr numerical Rising edge time constant. 1e-12 s
Tf numerical Falling edge time constant. 1e-12 s
Pw numerical Pulse width. 0.0 s
Pr numerical Repeating period of the pulse train. 0.0 s

User-dened Envelop Parameters

DLL string Name of the dynamically loadable library le that contains the none none
user-dened source.
Function string Function name of the source in the DLL le. none none

Examples

ENVELOP ID=gaussian_pulse Type=Gaussian amplitude=1e3 tdelay=0

t0=5e-12 tao=1e-12

68 Genius Device Simulator

External Sources Genius Command Reference

Apply Source to Device

Field source including particle and light, both generate electron and hole pairs in
the semiconductor device. The DDM solver, mixed type DDM solver and Half
Implicit solver support the simulation including the eects of eld source by set
parameter pat.gen and/or opt.gen to true. By default, Genius will do the cal-
culation of carrier generation only with pat.gen and/or opt.gen. However,
the SOURCEAPPLY command without any parameter force to calculate the carrier
generation even if not be used later.

Genius Device Simulator 69

Genius Command Reference External Sources

CIRCUIT
The CIRCUIT command denes a circuit netlist (in SPICE syntax) for mixed-mode
simulation.

CIRCUIT
Netlist = str

Parameters

Parameter Type Description Default Unit

Netlist string Path to the netlist le. The netlist should be in SPICE format. none none

70 Genius Device Simulator

Boundary Conditions Genius Command Reference

Boundary Conditions

Boundary and Contact

The BOUNDARY statement sets boundary information to representing faces of the
device mesh. GENIUS now fully support electrode region (the material of this
region should be metal or poly-Si). One should use CONTACT statement to specify
the electrode type of this region(s).

BOUNDARY
ID=str
Type = ( OhmicContact | SchottkyContact | SolderPadContact
|
GateContact | SimpleGateContact |
InsulatorInterface|
HeteroJunction | NeumannBoundary )

Ohmic Contact
[ external circuit ]
[ Heat.Transfer=num ] [ Ext.Temp=num ]

Schottky Contact
[ WorkFunction=num ]
[ external circuit ]
[ Heat.Transfer=num ] [ Ext.Temp=num ]

Gate Contact
[ WorkFunction=num ]
[ external circuit ]
[ Heat.Transfer=num ] [ Ext.Temp=num ]

SolderPad Contact
[ external circuit ]
[ Heat.Transfer=num ] [ Ext.Temp=num ]

Simple Gate Contact

[ WorkFunction=num ] [ QF=num ]
[ external circuit ]
[ Thickness=num ] [ eps=num ]
[ Heat.Transfer=num ] [ Ext.Temp=num ]

Insulator Interface
[ QF=num ]

Genius Device Simulator 71

 Genius Command Reference Boundary Conditions

HeteroJunction
[ QF=num ]

NeumannBoudary
[ Heat.Transfer=num ] [ Ext.Temp=num ]

Contact
ID=str Type = (OhmicContact|SchottkyContact|
GateContact|FloatMetal)

Ohmic Contact
[ external circuit ]
[ Heat.Transfer=num ] [ Ext.Temp=num ]

SchottkyContact
[ WorkFunction=num ]
[ external circuit ]
[ Heat.Transfer=num ] [ Ext.Temp=num ]

Gate Contact
[ WorkFunction=num ]
[ external circuit ]
[ Heat.Transfer=num ] [ Ext.Temp=num ]

Floating Metal
[ QF=num ]

Parameters

Parameter Type Description Default Unit

Type enum This parameter declares which type of boundary condition is dened none none
here.
ID string A unique string which identies the corresponding face. none none
WorkFunction numerical The workfunction of the Schottky contact or gate material. 0.0 eV
QF numerical For InsulatorContact and InsulatorInterface bc: The surface charge 0.0 cm2
density of semiconductor-insulator interface.
QF numerical For Heterojunction bc: The surface charge density of heterojunction. 0.0 cm2
QF numerical For FloatMetal bc: The free charge per micron in Z dimension. 0.0 C m1
Thickness numerical The thickness of SiO 2 layer. 2e-7 cm
eps numerical The relative permittivity of SiO 2 layer. 3.9 none
Heat.Transfer numerical The thermal conductance of the surface. see below W cm2 K1
Ext.Temp numerical The external temperature at the surface.Ele see below K

72 Genius Device Simulator

Boundary Conditions Genius Command Reference

Description

Five "electrode" boundary conditions are supported by GENIUS. The names are
ended with Contact. The OhmicContact and SchottkyContact electrodes
have current ow in both steady state and transient situations. While GateContact
and SimpleGateContact(a simplied MOSFET Gate boundary condition) only
have displacement current in transient situation. SolderPadContact also has
current ow in steady state and transient situations. The OhmicContact, SchottkyContact
and SimpleGateContact are boundaries on surface of semiconductor region.
GateContact should be the surface of insulator region and SolderPadContact
is the surface of a metal region.

Genius supports ve interfaces which can be set automatically: semiconductor-in-

sulator interface (InsulatorInterface), semiconductor-electrode interface(set
to OhmicContract as default), interface between dierent semiconductor ma-
terial (Heterojunction) and interface between same semiconductor material
(Homojunction). These boundaries can be set automatically by Genius if user
didn't set them explicitly. However, the electrode-insulator interface, may have
several situations: Gate to Oxide interface, FloatMetal to Oxide interface or
Source/Drain electrode to Oxide interface. As a result, this interface can only be
set correctly when electrode type is known. Please refer to the following CONTACT
statement.

GENIUS can build region with metal or poly-Si material to form an electrode.
Which means, i.e. for OhmicContact bc, one can simply specify a segment
as Ohmic bc or build an electrode region as Ohmic electrode. Since Version
0.45.03, GENIUS considers electrode region, semiconductor region and insula-
tor region during calculation. As a result, GENIUS added CONTACT statement
for fast boundaries specication of electrode region. At present, GENIUS sup-
port electrode with the type of Ohmic, Schottky, Gate and Floating-Metal. All
the electrode should be specied explicitly and GENIUS will set corresponding
boundaries automatically.

The ID parameter of BOUNDARY statement is limited to face label. And The ID

parameter of CONTACT statement is limited to region name.

The NeumannBoundary, which is the default boundary type for all the non-inter-
face segments, are set automatically at the outer surface of a region.

The thermal boundary conditions at outer surfaces of the device and the interface
with electrode regions are specied by the Heat.Transfer and Ext.Temp pa-
rameters. The default external temperature at the boundary is the external temper-
ature set in the GLOBAL command. The default value for heat conductance varies
with the types of boundary conditions. For OhmicContact, SchottkyContact
and Gatecontact, the default heat conductance is 103 W cm2 K1 . For NeumannBoundary

Genius Device Simulator 73

Genius Command Reference Boundary Conditions

the heat resistance is default to zero.

external circuit

Electrode boundaries can own its external circuit. Genius oers some commonly
used circuit type for convenient. At present, RCL model, PI model and RC trans-
mission line mode are supported. For complex external circuit support, user should
consider to use mixed model simulation.

External circuit
ext.circuit = ( RCL | Pi | Tline )

RCL external circuit

[ res=num ] [ cap=num ] [ ind=num ]

Pi type external circuit

[ res=num ] [ cap1=num ] [ cap2=num ] [ res.load=num ]

RC transmission line
[ res.um=num ] [ cap.um=num ] [ res.load=num ]
[ tl.length=num ] [tl.div=int ]

Parameter Type Description Default Unit

RCL model

res numerical The lumped resistance for the electrode. 0.0

cap numerical The lumped capacitance for the electrode. 0.0 F
ind numerical The lumped inductance for the electrode. 0.0 H

PI model

res numerical The resistance of PI model 1e-6

cap1 numerical The capacitance 1 of PI model 0.0 F
cap2 numerical The capacitance 2 of PI model 0.0 F
res.load numerical The terminal resistance. 1e-6

RC transmission line mode

res.um numerical Resistance per micron for the RC transmission line. 1e-12 m1
cap.um numerical Capacitance per micron for the RC transmission line. 0.0 Fm1
res.load numerical The terminal resistance. 1e-6
tl.length numerical The length of RC transmission line none m

74 Genius Device Simulator

Boundary Conditions Genius Command Reference

Parameter Type Description Default Unit

tl.div integer The discretization of the transmission line. 100 none

Example

BOUNDARY Type=SimpleGateContract ID=GATE Res=0 Cap=0 Ind=0 \

Thickness=1e-6 Eps=3.9 WorkFunction=4.7 QF=0

BOUNDARY Type=InsulatorInterface ID=IFACE QF=0

BOUNDARY Type=GateContract ID=GATE ext.circuit=RCL \

Res=0 Cap=0 Ind=0 WorkFunction=4.7

BOUNDARY Type=NeumannBoundary ID=WALL Heat.Transfer=0 \

EXT.Temp=300

BOUNDARY Type=SchottkyContract ID=sgate Res=10 Cap=0 \

VBarrier=0.8

BOUNDARY Type=OhmicContract ID=OMANODE

BOUNDARY Type=SolderPad ID=CATHODE Res=10 Cap=0 Ind=0

Genius Device Simulator 75

Genius Command Reference Boundary Conditions

ATTACH
The ATTACH command is used to add voltage or current sources to the electrode
boundary. The statement rst clears all the sources connected to the specied
electrode and then adds source(s) dened by VApp or IApp parameter. If two or
more sources are attached to the same electrode, the net eect is the superposition
of all sources. However, the sources attached to one electrode must have the same
type.

ATTACH Electrode=str { VApp=str | IApp=str }

[ { VApp=str | IApp=str } ... ]

Parameters

Parameter Type Description Default Unit

Setup Loadable Physical Models

Electrode string Electrode where the external source is attached. Must be a valid none none
ID of a electrical contact specied by the BOUNDARY or CONTACT
commands.
VApp string ID of the voltage source. none none
IApp string ID of the current source. none none

Description

If electrode is attached with voltage source(s), the external circuit dened in BOUNDARY
or CONTACT statement will aect later simulation. But solver will ignore R and
L when the electrode stimulated by current source(s). Please refer to Figure 2.4,
p. 77.

Several voltage sources or current sources can be attached to the same electrode
by multi- Vapp or multi- Iapp parameter. However, both the voltage source and
current source attach to the same electrode is invalid. Only Ohmic, Schottky and
SolderPad electrodes can be attached by current source(s). And the positive di-
rection of current is ow into the electrode.

If no source attached explicitly, the electrode is set to be attached to ground (0V

VDC).

76 Genius Device Simulator

Boundary Conditions Genius Command Reference

name: dummy
file: plotCmdRef/electrode_rcl.png
state: unknown

Figure 2.4 Voltage and current boundary for RCL type external circuit

name: dummy
file: plotCmdRef/electrode_pi_s.png
state: unknown

Figure 2.5 Voltage and current boundary for PI type external circuit

Example

ATTACH Electrode=Anode Vapp=V5V Vapp=Vs

ATTACH Electrode=Base Iapp=Ib

Genius Device Simulator 77

Genius Command Reference Boundary Conditions

name: dummy
file: plotCmdRef/electrode_tl.png
state: unknown

Figure 2.6 Voltage and current boundary for external transmission line

78 Genius Device Simulator

Fabrication Process Genius Command Reference

Fabrication Process

DOPING
The DOPING command selects the mode of doping prole specication. Only
analytic doping prole is currently implemented.

Doping Analytic

Genius Device Simulator 79

Genius Command Reference Fabrication Process

Analytic PROFILE
The PROFILE command species the doping prole. Uniform and analytic dop-
ing proles are supported.

PROFILE[.DOPING]
[ ID=str ]
Type = (Uniform|Analytic|Analytic2|File) Ion = (Donor|Acceptor|Cu

Uniform Profile Parameters

N.Peak=num
[ x.min=num ] [ x.max=num ] [ y.min=num ] [ y.max=num ]
[ z.min=num ] [ z.max=num ]

Anaylitic Profile Parameters

{ N.Peak=num | Dose=num }
[ x.min=num ] [ x.max=num ] [ y.min=num ] [ y.max=num ]
[ z.min=num ] [ z.max=num ]
{ Y.Char=num | Y.Junction=num } [ Y.ErfC=bool ]
{ X.Char=num | XY.Ratio=num } [ X.ErfC=bool ]
{ Z.Char=num | ZY.Ratio=num } [ Z.ErfC=bool ]

Anaylitic2 Profile Parameters

{ N.Peak=num }
{
mask.polygon=anum |
{ mask.x=num | mask.y=num | mask.z=num }
[ mask.xmin=num ] [ mask.xmax=num ]
[ mask.ymin=num ] [ mask.ymax=num ]
[ mask.zmin=num ] [ mask.zmax=num ]
}
[ implant.rmin=num ] [ implant.rmax=num ]
[ implant.theta=num ] [ implant.phi=num ]
[ depth.char=num ] [ lateral.char=num ]
[ resolution=num ]

Data File Parameters

File=str [ SkipLine=num ]
[ Axes= (x|y|xy|xz|yz|xyz) ]
[ LUnit = (m|cm|um|nm) ]
[Transform.xx=num] [Transform.xy=num] [Transform.xz=num]
[Transform.yx=num] [Transform.yy=num] [Transform.yz=num]
[Transform.zx=num] [Transform.zy=num] [Transform.zz=num]
[Translate.x=num] [Translate.y=num] [Translate.z=num]

80 Genius Device Simulator

 Fabrication Process Genius Command Reference

Parameters

Parameter Type Description Default Unit

General Parameters

Type enum Selects the type of doping prole. Doping prole can be Uniform, Uniform none
specied by Analytic expression, or loaded from a data File.
Ion enum Selects the doping species of the prole. Acceptor, donor and cus- none none
tom species are available. Custom species must bear an ID, so that
it can be referenced in other commands (e.g. in PMI trapping model
and incomplete ionization model).
ID string Identier of a custom doping prole. none none
x.min numerical Left boundary of the prole. Synonym: x.left. 0.0 m
x.max numerical Right boundary of the prole. Synonym: x.right. 0.0 m
y.min numerical Top boundary of the prole. Synonym: y.top. 0.0 m
y.max numerical Bottom boundary of the prole. Synonym: y.bottom. 0.0 m
z.min numerical Front boundary of the prole. Synonym: z.front. 0.0 m
z.max numerical Back boundary of the prole. Synonym: z.back. 0.0 m

Uniform Prole Parameters

N.Peak numerical Doping concentration. 0.0 cm3

Analytic Prole Parameters

N.Peak numerical Peak doping concentration. In the box dened by x.min, x.max, 0.0 cm3
y.min, y.max, z.min, z.max, the doping is uniformly N.Peak.
Outside the box, the doping concentration decays according to the
analytic function.
Dose numerical The dose of the doping concentration, assuming the resulting prole 0.0 cm2
is 1D Gaussian.
Y.Char numerical Characteristic length in the y direction. 0.0 m
Y.Junction numerical Target junction depth. 0.0 m
X.Char numerical Characteristic length in the x direction. 0.0 m
XY.Ratio numerical The ratio of characteristic length in the x direction against that in the 0.0 none
y direction.
X.Erfc bool Selects whether to use erfc instead of Gaussian function in the x false none
direction.
Z.Char numerical Characteristic length in the z direction. 0.0 m
ZY.Ratio numerical The ratio of characteristic length in the z direction against that in the 0.0 none
y direction.
Z.Erfc bool Selects whether to use erfc instead of Gaussian function in the z false none
direction.

Analytic2 Prole Parameters

Genius Device Simulator 81

 Genius Command Reference Fabrication Process

Parameter Type Description Default Unit

mask.poly numerical-array A numerical array species vertex location of the polygon mask. - $\uMeter$
Each vertex takes three number as its coordinates. The vertex should
be ordered along the loop of polygon. Polygon should be simple, i.e.
no self intersection.
mask.x numerical Indicate that doping mask plane is the X=num, for rectangle mask. 0.0 m
mask.y numerical Indicate that doping mask plane is the Y=num, for rectangle mask. 0.0 m
mask.z numerical Indicate that doping mask plane is the Z=num, for rectangle mask. 0.0 m
mask.xmin numerical Left boundary of the rectangle mask 0.0 m
mask.xmax numerical Right boundary of the rectangle mask 0.0 m
mask.ymin numerical Top boundary of the rectangle mask 0.0 m
mask.ymax numerical Top boundary of the rectangle mask 0.0 m
mask.zmin numerical Front boundary of the rectangle mask 0.0 m
mask.zmax numerical Back boundary of the rectangle mask 0.0 m
implant.rmin numerical The minimal depth of implant. Doping concentration keeps constant 0.0 m
between min and max doping depth.
implant.rmax numerical The maximal depth of implant. Doping concentration keeps con- 0.0 m
stant between min and max doping depth.
implant.theta numerical Dened as angel between +Y direction and implantation direction. 0.0 degree
implant.phi numerical Dened as angel between +X direction and the projection of im- 0.0 degree
plantation direction onto XZ plane.
depth.char numerical Characteristic length along the implantation direction. 0.1 m
lateral.char numerical Characteristic length lateral to the implantation direction. 0.1 m
resolution numerical The space resolution factor, minimal space equals to lateral.char 4.0 none
divide by this factor.

Doping Data File Parameters

File string Name of the doping concentration data le. none none
Transform numerical The transformation matrix applied to the coordinates imported I none
from the input le. The nal coordinates are obtained from
Transform*location + Translate.
Translate enum The translation vector applied to the input coordinates. 0 none
SkipLine integer Skip the rst SkipLine number of lines in the data le. 0 none
LUnit enum Length unit used for node coordinates in the data le. um none

Analytic Prole Calculation

When the Type parameter is set to Analytic, the doping prole is calculated
with the expression

(,) = . () () (), (2.4)

where N.PEAK is the peak doping concentration, and the functions (), ()
and () describes the variation of doping concentration in the x-, y- and z-di-
rections, respectively.

82 Genius Device Simulator

Fabrication Process Genius Command Reference

Two options are currently available for the variation functions, namely the Gauss
prole function and the Erfc prole function. One can choose the prole func-
tion for the x-, y- and z-directions independently. When the X.Erfc, Y.Erfc or
Z.Erfc parameter is enabled, the Erfc function is used in that direction.

The Gauss prole function is dened as

exp . 2 y < D.MIN

[ ( . ) ]

(,) = 1 D.MINy D.MAX .
. 2
exp [ ( . ) ] y>D.MAX

where = ,,.

The Erfc prole function is dened as

erfc ( .
. )
erfc ( .
. )
(,) = . (2.5)
2

If the XY.Ratio parameter is specied instead of the X.Char parameter, X.Char

is calculated with . = . .. Similar calculation applies to
ZY.Ratio and Z.Char.

Implant Dose The Dose parameter can be used instead of the N.Peak parameter to specify the
dose of impurities per unit area. In this case, the peak concentration is calculated
with

. = . (2.6)
.

Analytic2 Prole Calculation

When the Type parameter is set to Analytic2, the doping prole is calculated
with a given mask and the implant parameters. A planer rectangle mask or poly-
gon mask should be dened as the window of implantation. The implantation
is dened by incident angle given by implant.theta and implant.phi and
ranges given by implant.rmin and implant.rmax, which relative to the mask
plane. The doping concentration equals to N.Peak within the range between
implant.rmin and implant.rmax, and variated as Gauss prole function with
parameter depth.char along the doping line outside the range. For lateral dis-
tribution, a Gauss prole with parameter lateral.char is assumed. The Ge-
nius use a ray shooting method to calculate the nal doping concentration. The

Genius Device Simulator 83

Genius Command Reference Fabrication Process

accurate is depended on density of doping lines which controlled by parameter

resolution.

Custom Doping Prole

When a custom dopant species is desired, the user should set Ion to Custom, and
supply an identier ID.

Field Data File The eld data le must be in plain text, with numerical values separated by white
spaces. For 3D doping prole, each line contains four values (x-, y-, and z-co-
ordinate and the doping concentration). For 2D doping prole, the z-coordinate
is not present, and each line contains three values. The unit of the coordinates is
specied by LUnit, and the unit of the concentration is cm3 . Positive doping
concentration indicates Donor ion, while negative doping concentration indicates
Acceptor ion.

Example

Profile.Doping Type=Analytic Ion=Acceptor \

N.Peak=1e18 Y.Min=0.0 Y.Max=0.1 \
Y.Char=0.05 XY.Ratio=0.8

Profile.Doping Type=Analytic2 Ion=Acceptor N.PEAK=1E19 \

Mask.poly=[1,0,2,3,0,3,5,0,1,5,0,5,1,0,4] \
Implant.rmin=0.0 Implant.rmax=0.3 Implant.theta=0

Profile.Doping Type=Uniform Ion=Acceptor N.Peak=1e16

Profile.Doping Type=File File="halo.txt"

84 Genius Device Simulator

Fabrication Process Genius Command Reference

MOLE
The MOLE command species the mole fraction prole of compound materials.
For binary compound material A xB 1-x, the mole fraction prole x is needed. For
tertiary compound material A xB yC 1-x-y, both x and y prole are needed.

MOLE
Region=str
[ x.min=num ] [ x.max=num ] [ y.min=num ] [ y.max=num ]
[ z.min=num ] [ z.max=num ]

Binary Compound Material

X.Mole=num ( X.Mole.Slope=num | X.Mole.End=num )
[ X.Mole.Grad = (X.Linear|Y.Linear|Z.Linear) ]

Tertiary Compound Material

X.Mole=num ( X.Mole.Slope=num | X.Mole.End=num )
[ X.Mole.Grad = (X.Linear|Y.Linear|Z.Linear) ]
Y.Mole=num ( Y.Mole.Slope=num | Y.Mole.End=num )
[ X.Mole.Grad = (X.Linear|Y.Linear|Z.Linear) ]

Parameters

Parameter Type Description Default Unit

General Parameters

Region string Identier of the compound material region. none none

x.min numerical Left boundary of the prole. Synonym: x.left. 0.0 m
x.max numerical Right boundary of the prole. Synonym: x.right. 0.0 m
y.min numerical Top boundary of the prole. Synonym: y.top. 0.0 m
y.max numerical Bottom boundary of the prole. Synonym: y.bottom. 0.0 m
z.min numerical Front boundary of the prole. Synonym: z.front. 0.0 m
z.max numerical Back boundary of the prole. Synonym: z.back. 0.0 m

Binary Compound

X.Mole numerical Starting mole fraction x of compound material A xB 1-x. 0.0 none
X.Mole.Slope numerical The slope of mole fraction x. 0.0 m1
X.Mole.End numerical The mole fraction x at the end of the prole. 0.0 none
X.Mole.Grad enum The direction of mole fraction gradient. Y.Linear none

Tertiary Compound

Genius Device Simulator 85

Genius Command Reference Fabrication Process

Parameter Type Description Default Unit

X.Mole numerical Starting mole fraction x of compound material A xB yC 1-x-y. 0.0 none
X.Mole.Slope numerical The slope of mole fraction x. 0.0 m1
X.Mole.End numerical The mole fraction x at the end of the prole. 0.0 none
X.Mole.Grad enum The direction of mole fraction gradient. Y.Linear none
Y.Mole numerical Starting mole fraction y. 0.0 none
Y.Mole.Slope numerical The slope of mole fraction y. 0.0 m1
Y.Mole.End numerical The mole fraction y at the end of the prole. 0.0 none
Y.Mole.Grad enum The direction of mole fraction gradient. Y.Linear none

Description

Only linear mole fraction prole with its gradient aligned with one of the three
Cartesian axes is currently supported. There are two alternative ways to spec-
ify the expression of the prole. One can either specify the slope explicitly with
X.Mole.Slope and the mole fraction at the starting end with X.Mole. Start-
ing end is dened as the one with smaller coordinate value in the direction of
X.Mole.Grad. Alternatively, one can set the mole fraction at both ends of the
region with X.Mole and X.Mole.End, and let GENIUS calculate the gradient.

Example

MOLE region=SiGe X.Mole=0.4 X.Mole.End=0.7 \

X.Mole.Grad=Y.Linear

MOLE region=SiGe X.Mole=0.4 X.Mole.Slope=0.1 \

Y.Min=0 Y.Max=0.3 X.Mole.Grad=Y.Linear

86 Genius Device Simulator

Fabrication Process Genius Command Reference

REGIONSET
The REGIONSET command reset the doping and mole value of a given region.
Only uniform value can be specied. The given value will overwrite any previous
value.

REGIONSET
Region=str

Doping profile
[ doping.na=num ] [ doping.nd=num ]

Binary Compound Material

[ X.Mole=num ]

Tertiary Compound Material

[ X.Mole=num ] [ Y.Mole=num ]

Parameters

Parameter Type Description Default Unit

Region specify

Region string Identier of the compound material region. none none

Doping Parameters

doping.na numerical Set region (uniform) acceptor density 0.0 cm3

doping.nd numerical Set region (uniform) donor density 0.0 cm3

Binary Compound

X.Mole numerical Set mole fraction x of compound material A xB 1-x. 0.0 none

Tertiary Compound

X.Mole numerical Set mole fraction x of compound material A xB yC 1-x-y. 0.0 none
Y.Mole numerical Set mole fraction y. 0.0 none

Example

REGIONSET region=SiGe X.Mole=0.4

Genius Device Simulator 87

Genius Command Reference Physical Models

Physical Models

PMI
In Genius, the physical models of each material is loaded to the simulation system
through the physical model interface (PMI). The models can be selected through
the PMI command. The model parameters are also specied here. The general
syntax of the PMI command is

PMI
REGION=str
[ Type= (Basic|Band|Mobility|Impact|Thermal|Optical|Trap)
]
[ Model=str ] [print=int]
[ real<str>=num [ real<str>=num [...] ] ]
[ string<str>=str [ string<str>=str [...] ] ]

Parameters

Parameter Type Description Default Unit

Setup Loadable Physical Models

Region string The region that the loaded model would apply to. A valid region none none
name must be supplied, otherwise the simulation will abort. The
region string can be given as a regular expression to match a number
of regions.

88 Genius Device Simulator

Physical Models Genius Command Reference

Parameter Type Description Default Unit

Type enum The type of model to be loaded. Seven types are currently supported Mobility none
in Genius, though some of them are not applicable to all material
regions.

A Basic model species the basic physical parameters such as

permittivity and electron anity, etc.
A Band model species parameters related to the band struc-
ture of semiconductors, which include energy bandgap, eective
density of states, eective masses, etc. It also calculates rates of
some fundamental processes such as carrier recombination rates.
Band models are applicable to semiconductor regions only.
A Mobility model calculates the mobility of carriers, taking
into consideration the eect of temperature, doping, electric
eld and adjacent interfaces, etc. A range of mobility models
are provided in Genius, which will be discussed in more detail
in Mobility Models, p. 90. Mobility models apply to semi-
conductor regions only.
An Impact model calculates the carrier generation rates due to
the impact ionization process. A later section will describe im-
pact ionization models in detail. Impact models apply to semi-
conductor regions only.
An Thermal model describes the thermal conduction property
of materials.
An Optical model describes the optical propagation property
of materials.
An Trap model calculates the rate of carriers capture and emis-
sion at traps. Trap models currently apply to semiconductor
regions and semiconductor/insulator interfaces.
Model string Name of the physical models to be loaded. The list of available Default none
models for each material and each type of models is summarized in
later sections.
Print integer If greater than 0, print the current values of all the material parame- 0 none
ters.
real<str> numerical Additional custom numerical parameters accepted by the loaded none unknown
model. The name of the parameter is specied by str.
string<str> string Additional custom string parameters accepted by the loaded model. none none
The name of the parameter is specied by str.

Description

The core program of Genius access the numerous material-specic models via the
Physical Model Interface (PMI).

In Genius, each supported material has a corresponding dynamic loadable library

(DLL) le located in the $(GENIUS_DIR)/lib/ directory. Each DLL le con-
tains all models applicable to that material. For instance, all physical models for
Si material are stored in $(GENIUS_DIR)/lib/libSi.so.

Genius Device Simulator 89

Genius Command Reference Physical Models

For each type of models, a default model will be loaded in the absence of a PMI
command. When the user supplied one or more PMI commands, Genius sequen-
tially loads the specied model and assign it to the specied region. The user
can optionally supply a list of custom numerical and string parameters to the PMI
model. For instance, the following command loads the trap model for Silicon
region, and setup the parameters of the trap model.

# set trap parameter to region Silicon

PMI Region=Silicon Type=Trap string<Type>=Bulk \
string<Profile>=dopingA string<ChargeType>=Acceptor \
double<Energy>=0.1

# set all the regions which has a name as active_num to use

Lucent mobility model
PMI Region="active_[0-9]+" Type=mob Model=Lucent

Each PMI model has its own list of accepted parameters, and the user should con-
sult the description of the individual PMI models for details.

If the model specied by a PMI command has already been loaded, the previously
loaded one is not replaced. The latter PMI command are useful in adjusting the
custom parameters of the PMI model.

Custom PMI Model Advanced users of Genius can compile and produce dynamic libraries that contain
custom PMI models. The easiest way of writing custom PMI model is to extend
from the source code of existing models. See [[TODO]] for a tutorial guide and
detailed references.

Mobility Models

The mobility model can be selected in the Model parameter of the PMI command:

PMI
REGION=str Type=Trap [ Model=str ]
[ param=val [ param=val [ ... ] ] ]
%

The variety of available mobility models and the default model for each material is
summarized in Table 2.1, p. 91. Each model has a set of its own parameters, which
can be adjusted with the PMI command. The description of each mobility model
and the corresponding parameters are detailed in Mobility Models, p. 12.

90 Genius Device Simulator

Physical Models Genius Command Reference

Material Models Default

3C-SiC Analytic Analytic
4H-SiC Masetti Masetti
AlGaAs Analytic Analytic
GaAs Analytic, Hypertang Analytic
Ge Analytic Analytic
HgCdTe Analytic Analytic
InAs Analytic Analytic
InGaAs Analytic Analytic
InN Analytic Analytic
InP Analytic Analytic
InSb Analytic Analytic
S-SiO2 Constant Constant
Si Constant, Analytic, Philips, HP, Lucent, Lombardi Analytic
SiGe Constant, Analytic, Philips, HP, Lucent, Lombardi Analytic
Table 2.1 Summary of Available PMI Mobility Models

The following example shows how one can load the HP mobility model and adjust
the default parameter.

PMI Region=Silicon Type=Mobility Model=HP \

double<MUN.MAX>=1200 double<MUP.MAX>=400

Impact Ionization Models

The mobility model can be selected in the Model parameter of the PMI command:

PMI
REGION=str Type=Impact [ Model=str ]
[ param=val [ param=val [ ... ] ] ]

For silicon and many other materials, the Default avalanche generation model
is the Selberherr model. The models for various materials are summarized in
Table 2.2, p. 92.

Genius Device Simulator 91

Genius Command Reference Physical Models

Material Models Default

3C-SiC Selberherr Selberherr
4H-SiC vanOverstraetendeMan vanOverstraetendeMan
AlGaAs Selberherr Selberherr
GaAs Selberherr Selberherr
Ge Selberherr Selberherr
HgCdTe - -
InAs vanOverstraetendeMan vanOverstraetendeMan
InGaAs Selberherr -
InN vanOverstraetendeMan vanOverstraetendeMan
InP Selberherr Selberherr
InSb vanOverstraetendeMan vanOverstraetendeMan
S-SiO2 - -
Si Selberherr, Valdinoci Selberherr
SiGe Selberherr Selberherr
Table 2.2 Summary of Available PMI Impact Ionization Models

Trap Model

Although the charge trapping model is automatically loaded, the user needs to
setup trap parameters with PMI command for it eect. The syntax and parameters
are as follows.

PMI
REGION=str Type=Trap Model=Default
[string<Type> = (Bulk|Interface)]
[string<ChargeType> = (Acceptor|Donor)]
[real<Energy>=num] [real<SigmaN>=num] [real<SigmaP>=num]

Bulk Trap
{ string<Profile>=str [ real<Prefactor>=num ] |

Interface Trap
string<Interface>=str real<IF.Density>=num }

Parameter Type Description Default Unit

Common Trap Parameters

92 Genius Device Simulator

Physical Models Genius Command Reference

Parameter Type Description Default Unit

Type enum Selects whether the current command species Bulk or Interface Bulk none
traps.
ChargeType enum Selects whether the traps is Acceptor or Donor. Acceptor none
Energy numerical Energy of the trap level relative to the intrinsic Fermi level of the 0 eV
material.
SigmaN numerical Capture cross-section for electrons. 4e16 cm2
SigmaP numerical Capture cross-section for holes. 4e16 cm2

Bulk Trap Parameters

Profile string The concentration of bulk traps is derived from a custom doping none none
prole dened by a PROFILE command. ID of the custom prole
should be specied here.
Prefactor numerical The concentration of the doping prole specied by Profile is 1.0 none
scaled by this Prefactor. This is useful in generating a number of
traps with the same spatial distribution but dierent trap energies.

Interface Trap Parameters

Interface string Interface traps are attached to the insulator/semiconductor inter- none none
face with the ID specied here. When an semiconductor region
and an insulator region share a common face, an interface is au-
tomatically generated. The ID of the interface is in the format of
RegionName1_to_RegionName2. Alternatively, one can use a
Face command to dene an interface with a user-specied ID.
IF.Density numerical The interface density of traps. 0.0 cm2

Incomplete Ionization Model

User can setup incomplete ionization parameters with PMI command. Please note,
for active this model, use also have to set IncompleteIonization to true in
the MODEL statement. Also, it is recommend to use Fermi statistics with Incom-
pleteIonization. The syntax and parameters are as follows.

PMI
REGION=str Type=band
[string<species>=str] [int<ion>=(+1|-1)]
[real<EB0>=num] [real<GB>=num]
[real<alpha>=num] [real<beta>=num] [real<gamma>=num]
%

Genius Device Simulator 93

Genius Command Reference Physical Models

Parameter Type Description Default Unit

Common Trap Parameters

species string Specify prole species. The string parameter is the id of prole with none none
Custom ion type.
ion integer Selects whether the species is Acceptor(-1) or Donor(+1). 0 none
EB0 numerical The constant activation energy. 0 eV
GB numerical The band degeneracy factor none none
alpha numerical The prefactor for the doping dependent term used in the calculation none eVcm1
of the band ionization energy.
beta numerical The prefactor the temperature dependent term used in the calculation none none
of the band ionization energy.
gamma numerical The exponent of temperature used in the calculation of the band ion- none none
ization energy.

94 Genius Device Simulator

Physical Models Genius Command Reference

MODEL
The MODEL command sets the switches that controls the behavior of the physical
equations.

MODEL
Region=str
[ Fermi=bool ] [IncompleteIonization=bool]
{ H.Mob=bool [ Mob.Force = (EJ|ESimple|EQF) ] }
{ ImpactIonization = (Local|No)
[ II.Force = (EdotJ|ESide|EVector|GradQF) ] }
[ BBT = (Local|No) ]
[ Optical.Gen = (true|false) ]
[ Particle.Gen = (true|false) ]
[ EB.Level= (None|Tl|Te|Th|TeTh|TeTl|ThTl|ALL) ]

Parameters

Parameter Type Description Default Unit

Region string Select models in the specied region. Regular expression can be none none
used to match a number of regions.
Fermi boolean Turns on Fermi-Dirac carrier statistics. false none

IncompleteIonization

H.Mob boolean Synonym to HighFieldMobility. Activate high-eld mobility true none

calculation. When H.Mob is true, parallel and transverse elds are
calculated and passed to the mobility model, so that high-eld ef-
fects on carrier mobility are taken into account.
ESurface boolean Use eective surface electric eld for carrier mobility calculation at true none
insulator-semiconductor interface.
Mob.Force enum Synonym to Mobility.Force. When H.Mob is enabled, this pa- ESimple none
rameter selects the driving eld used in high-eld mobility calcula-
tion.

When EJ is selected, the dot product between electric eld and

current density ( /| |) is used as the parallel eld. The cross
product ( /| |) is used as the transverse eld.
When ESimple is selected, the electric eld as the parallel
eld.
When EQF is selected, the gradient of the quasi-fermi level (
and ) as the parallel eld. The transverse eld is calculated
with /| |, as in EJ method.
ImpactIonization enum Activates the impact ionization (II) generation, and selects whether No none
the II generation should be calculated with the local model or the
non-local model. Only the local model is currently implemented.

Genius Device Simulator 95

Genius Command Reference Physical Models

Parameter Type Description Default Unit

II.Force enum When ImpactIonization is enabled, this parameter selects the No none
driving eld used in impact ionization coecient calculation.

When EdotJ is selected, the dot product between electric eld

and current density ( /| |) is used as the driving eld.
When EVector is selected, the magnitude of the electric eld
is used as the driving eld.
When ESide is selected, the E-eld component along the edge
of the mesh element is used as the driving eld.
When GradQf is selected, the gradient of the quasi-fermi level
( and ) as the driving eld.
BBT enum Activates the Band-to-Band Tunneling (BBT) generation, and se- No none
lects whether the BBT generation should be calculated with the lo-
cal model or the non-local model. Only the local model is currently
implemented.
EB.Level enum When the Energy-Balance equation solver is selected in the METHOD None none
command, this parameter selectively enables the equations for lattice
temperature, electron temperature and hole temperature.

Example

MODEL Region=emitter Fermi=true

MODEL Region=silicon H.Mob=true Mob.Force=EQF
MODEL Region=silicon EB.Level=TeTl

96 Genius Device Simulator

Numerical Solution Genius Command Reference

Numerical Solution

METHOD
The METHOD command species the parameters of the numerical solvers.

METHOD

Equation Selection
[ Type = (Poisson|DDML1|DDML1Mix|DDML2|DDML2Mix|
EBML3|EBML3Mix|HalfImplicit|Stress) ]

Numeric Solver Selection

[ NS = (Newton|LineSearch|TrustRegion) ]
[ LS = (CGS|BICG|BCGS|GMRES|TFQMR|LU|UMFPACK|SuperLU|
MUMPS|SuperLU_Dist) ]
[ PC = (Identity|Jacobian|BJacobian|ASM|ILU) ]
[ Damping = (No|Potential|BankRose) ]
[ Truncation = (Always|Boundary|No) ]
[ Halfimplicit.let = num ]

Convergence Criteria
[ MaxIteration=int ]
[ Relative.Tol=num ] [ Poisson.Tol=num ]
[ Elec.Continuity.Tol=num ] [ Hole.Continuity.Tol=num ]
[ Elec.Energy.Tol=num ] [ Hole.Energy.Tol=num ]
[ Latt.Temp.Tol=num ] [Electrode.Tol=num]
[ Toler.Relax=num ] [ QNFactor=num ] [ QPFactor=num ]

Parameters

Parameter Type Description Default Unit

Equation Selection

Genius Device Simulator 97

Genius Command Reference Numerical Solution

Parameter Type Description Default Unit

Type enum Selects the set of partial dierential equations to be solved. DDML1 none

Poisson selects the Poisson's equation solver.

DDML1 selects the basic level-one drift-diusion equation solver.
DDML1Mix selects the device-circuit mixed-mode solver with
the level-one drift-diusion equation solver.
DDML2 selects the level-two drift-diusion equation solver,
which includes the lattice temperature variation in device.
DDML2Mix selects the device-circuit mixed-mode solver with
the level-two drift-diusion equation solver.
EBML3 selects the energy-balance equation solver which in-
cludes carrier temperature variation in device. Users can selec-
tively activate the equations for electron temperature, hole tem-
perature and lattice temperature with the EB.Level parameter
in the MODEL command.
EBML3Mix selects the device-circuit mixed-mode solver with
the energy-balance equation solver.
HalfImplicit selects the fast half implicit solver. This solver
only support transient and OP simulation. And it runs typically
5x faster than DDM solver.
Stress selects the mechanical stress solver.

Numerical Solver Selection

NS enum Selects the Newton nonlinear solver algorithm. LineSearch none

Newton (synonym: Basic) selects the basic nonlinear update

algorithm with full Newton steps.
LineSearch selects the cubic line-search update algorithm.
TrustRegion selects the trust-region update algorithm.
LS enum Selects the linear solver algorithm. BCGS none
PC enum Selects the linear preconditioner algorithm. ILU none
Damping enum Selects the nonlinear update damping scheme. Potential none
Truncation enum Selects the element truncation strategy used in simulation. Always none
Halmplicit.let numerical Linearize error threshold of half implicit method. A larger threshold 1.0 Vt
will speedup the simulation, but introduce more error.

Convergence Criteria of Nonlinear Solver

MaxIteration integer Sets the number of maximum nonlinear iterations before the solver 30 none
reports failure.
Relative.Tol numerical Sets the relative update tolerance. 1e-5 none
Poisson.Tol numerical Sets the absolute tolerance for the residue norm of Poisson's equa- 1e-26 none
tion.
Elec.C.Tol numerical Sets the absolute tolerance for the residue norm of the electron con- 5e-18 none
tinuity equation. Synonym: Elec.Continuity.Tol.

98 Genius Device Simulator

Numerical Solution Genius Command Reference

Parameter Type Description Default Unit

Hole.C.Tol numerical Sets the absolute tolerance for the residue norm of the hole continu- 5e-18 none
ity equation. Synonym: Hole.Continuity.Tol.
Elec.E.Tol numerical Sets the absolute tolerance for the residue norm of the electron-en- 1e-18 none
ergy balance equation. Only applicable when the EBM solver is
used and the electron energy-balance equation is activated. Syn-
onym: Elec.Energy.Tol.
Hole.E.Tol numerical Sets the absolute tolerance for the residue norm of the hole en- 1e-18 none
ergy balance equation. Only applicable when the EBM solver is
used and the hole energy-balance equation is activated. Synonym:
Hole.Energy.Tol.
Latt.Temp.Tol numerical Sets the absolute tolerance for the residue norm of the lattice heat 1e-11 none
equation. Applicable when the DDML2 solver is selected; or when
the EBM solver is used and the lattice heat equation is activated.
Electrode.Tol numerical Sets the absolute tolerance for the residue norm of at the electrode 1e-9 none
boundaries.
Toler.Relax numerical Sets the tolerance relaxation factor for convergence on relative tol- 1e4 none
erance criteria. When the relative error is below Relative.Tol,
the solver checks the absolute residue norms of all equations before
it reports relative convergence. The absolute residue norms must
be below Toler.Relax*aTol, where aTol is the various absolute
tolerance described above.

Convergence Criteria of Linear Solver

KSP.RTol numerical Relative tolerance of convergence criterion when ksp methods are 1e-8 none
use drelative tolerance of convergence criterion when ksp methods
are used.
KSP.ATol numerical Absolute tolerance of convergence criterion when ksp methods are 1e-20 none
used.
KSP.ATol.FNorm numerical Absolute tolerance convergence criterion, when the residue norm of 1e-7 none
ksp iteration less than .. * ().

Nonlinear Solver Algorithms

GENIUS solves the coupled Poisson-Drift-Diusion equations in semiconductor

devices with Newton's method. The system of nonlinear equations is written as
() = 0, where is the nonlinear function and is the solution. Starting from an
initial guess to the solution, one can solve the nonlinear equation iteratively. The
update to the solution in every iteration is computed from the following linear
system of equations,

() = () (2.7)

where is the computed update, and is the Jacobian of . In the basic Newton
algorithm, the new solution is

Genius Device Simulator 99

Genius Command Reference Numerical Solution

= + . (2.8)

This iterative process continues until the convergence criteria is met. Apart from
the basic algorithm described above, two variants of Newton solution update algo-
rithms are provided. The cubic linesearch algorithm and the Trust-Region update
algorithms can be selected with the NS parameter.

Update damping algorithms can be used with Newton methods to improve stabil-
ity of the iterative process. All updates are damped so that the carrier densities
are positive. When the Basic Newton algorithm is selected, one can turn on the
Potential damping or the Bank-Rose damping, which sometimes improve conver-
gence.

Nonlinear Convergence Criteria

The nonlinear solver checks for convergence after Newton iteration. The following
sets of criteria are checked in sequence, if any one set of criteria is met, the solution
is determined to have reached convergence.

Absolute Residue The residue of each device equation is evaluated. If the norm of each residue is
Norm Convergence less than the respective absolute tolerance, convergence is reached.

Relative Residue The residue of all device equations is evaluated, and its norm compared against the
Norm Convergence residue norm before the nonlinear iterations. If the ratio between the two norm
is less than the relative tolerance, the absolute residue norm criteria is further
checked with the absolute tolerance relaxed by the factor Toler.Relax. Con-
vergence is reached if both tests passes.

Relative Update The norm of the update in the Newton iteration is evaluated, if this norm is less
Norm Convergence than the relative tolerance, the absolute residue norm criteria is further checked
with the absolute tolerance relaxed by the factor Toler.Relax. Convergence is
reached if both tests passes.

Linear Solver Algorithms

Equation (2.7), p. 99 can be solved by direct method based on LU factorization.

Direct linear solvers usually gives the exact solution of (2.7). A few direct linear
solvers based on LU decomposition are also provided. The default LU solver only
supports serial execution. The UMFPACK solver is a fast serial direct solver,
while MUMPS supports parallel computation. The serial and parallel version of
SuperLU are also available.

Alternatively, the more ecient iterative methods based on Krylov subspace (KSP)
theory can be used. However, iterative methods can only give approximate re-
sults. Among the choices of linear system solvers, the CG, BCGS, GMRES methods

100 Genius Device Simulator

Numerical Solution Genius Command Reference

are KSP iterative solvers. A preconditioner is necessary for eective operation of

iterative solvers. The ILU algorithm usually provides good performance. All the
iterative solver and preconditioners listed here support parallel computation. The
KSP iteration stops when any of the two criteria is met:

the NORM change between two iteration less than ksp.rtol.

residue NORM less than max (..*(()), .).

The iterative linear solver may fail to reach convergence in some cases, which
cause the nonlinear solver fail with the error DIVERGED_LINEAR_SOLVER. In
this case one may try relaxing the three linear solver tolerance parameters. On the
other hand, if the error norm of the nonlinear solver stagnates when a KSP linear
solver is in use, it may be caused by inexact linear solution. Tightening the KSP
convergence criteria may help.

Genius Device Simulator 101

Genius Command Reference Numerical Solution

SOLVE
The SOLVE command instructs Genius to solve the semiconductor device equa-
tions.

SOLVE
[Type=(Equilibrium|SteadyState|DCSweep|Op|ACSweep|Transient)]
[Predict=bool]

DCSweep Simulation Parameters

{ VScan=str VStart=num VStep=num VStop=num
[ VStepMax=num ] } |
{ IScan=str IStart=num IStep=num IStop=num
[ IStepMax=num ] }
[ Optical.Gen=bool ] [ Optical.modulate=str ]
[ Particle.Gen=bool ]

Operator Point Parameters for Mixed-type Simulation

{ [ rampup.steps=int ] [ rampup.vstep=num ]
[ rampup.istep=num ]
[ gmin=num ] [ gmin.init=num ]
}

Operator Point Parameters for Device Simulation

{ [ pseudotime=bool ] [ pseudotime.step=num ]
[ pseudotime.iteration=int ]
[ VStepMax=num ] [ IStepMax=num ]
}

Transient Simulation Parameters

{ TS=(ImplicitEuler|BDF1|BDF2)
TStart=num TStep=num TStop=num
[ AutoStep=bool ] [ TStepMax=num ]
[ TS.rTol=num ] [ TS.aTol=num ]
[ VStepMax=num ] [ IStepMax=num ]
[ Optical.Gen=bool ] [ Optical.modulate=str ]
[ Particle.Gen=bool ] [Source.coupled=bool]
[ tran.op=bool ]
}

ACSweep Simulation Parameters

{ VScan=str f.start=num f.stop=num
[ f.multiple=num ]
[ vac=num ]
}

102 Genius Device Simulator

 Numerical Solution Genius Command Reference

Input/Output Parameters
[ Hook=str ] [ Out.Prefix=str ]

Parameters

Parameter Type Description Default Unit

General Parameters

Type enum Selects the mode of the simulation. Equilibrium none

Equilibrium sets all external sources to zero, and timestep

to innity, so that the device equations yield equlibrium-state
solution.
SteadyState sets all external sources to their respective value
at time zero, and timestep to innity.
DCSweep scans the external voltage (or current) source at one
of the contacts. At each voltage (or current) step, the device
equations are solved in steadystate.
OP solves the operator point of device or circuit at time zero,
some convergence acceleration techniques such as gmin step in
circuit simulation and pseudo time method for device can be
used.
ACSweep applies an small-signal stimulus at one of the contacts,
and scans the frequency of the signal.
Transient solves the device equation with physical quantities
in the device and external sources vary as time progresses.
Predict bool When Predict is enabled, the solver use predictions based on pre- true none
vious solutions as the initial guess to the new solution. When three
previous solutions are available, quadratic projection is used. When
there are two previous solutions, linear projection is used.
When Predict is disabled, the solver uses the previous solution as
the initial guess.
Predict is applicable to DC-sweep and transient modes only.
Optical.Gen boolean Consider optical carrier generation in the simulation. Applicable in false none
DC- and transient-mode simulations only.
Optical.modulate string Optical wave can be modulated by an envelope, which is specied none none
by ENVELOP statement. When this parameter is omitted, constant
envelop with amplitude 1.0 will be used. For DC simulation, value
at transient time 0 will be used.
Particle.Gen boolean Consider carriers generated by high-energy particles in the simula- false none
tion. Applicable in DC- and transient-mode simulations only.
Source.coupled boolean Indicate that the light absorption or particle generation is coupled false none
with semiconductor device. Thus, Genius will re-calculate carrier
generation on each device state.

DC Sweep Simulation Parameters

Genius Device Simulator 103

 Genius Command Reference Numerical Solution

Parameter Type Description Default Unit

VScan string The electrode on which voltage sweep is applied. none none
VStart numerical Start voltage. none V
VStep numerical Initial voltage step. The actual voltage steps may change when there none V
is convergence problem.
VStop numerical Stop voltage. none V
VStepMax numerical Maximum voltage step. VStep V
IScan string The electrode on which current sweep is applied. none none
IStart numerical Start current. none A
IStep numerical Initial current step. The actual current steps may change when there none A
is convergence problem.
IStop numerical Stop current. none A
IStepMax numerical Maximum current step. IStep A

Circuit Operation-Point Simulation Parameters

for Mixed Simulation

rampup.steps integer The number of ramp-up steps for the voltage and current sources in 1 none
the circuit to their DC value.
rampup.vstep numerical The maximum changes of voltage in any ramp up step. 0.25 V
rampup.istep numerical The maximum changes of current in any ramp up step. 0.1 A
gmin.init numerical The initial value of gmin in gmin step-down. A large value may help 1e-12 1
convergence for dicult circuits.
gmin numerical The target value of gmin step-down. This value should be su- 1e-12 1
ciently small to avoid error in circuit simulation.

Pseudo-Time DC Simulation Parameters

pseudotime bool When pseudotime is set to true, Genius will use pseudo-time false none
method to calculate the steady-state solution of the device. Cur-
rently available in pure device simulation, with type=OP.
pseudotime.step numerical Time step for the pseudo-time method. Small time step will make 1e-6 s
the system more stable but require more time steps to drive the sys-
tem to steady state.
pseudotime.iter numerical Maximum number of pseudo-time iteration steps for driving the sys- 20 none
tem to steady state.
VStepMax numerical Maximum voltage update in any pseudo-time step. 0.1 V
IStepMax numerical Maximum current update in any pseudo-time step. 0.1 A

Transient Simulation Parameters

TS Enum Sets the time-discretization scheme. Available choices are BDF1 BDF2 none
(aka. Implicit Euler), BDF2 and TRBDF2.
TStart numerical Initial time. 0.0 s

104 Genius Device Simulator

 Numerical Solution Genius Command Reference

Parameter Type Description Default Unit

TStep numerical Initial time step. The actual time steps is adaptively determined by none s
the solver if AutoStep is enabled.
TStop numerical Stop time. none s
TStepMax numerical The maximum allowed time step when the solver automatically de- 0.0 s
termines time steps. If TStepMax is zero, there is no upper bound
for time steps.
AutoStep bool Sets if the solver can adaptively change the time steps. true none
TS.rTol numerical Sets the relative truncation error tolerance in time-discretization. 1e-3 none
When the tolerance criteria is not met, the current time step is re-
jected, and time step is reduced.
TS.aTol numerical Sets the absolute truncation error tolerance in time-discretization. 1e-4 none
When the tolerance criteria is not met, the current time step is re-
jected, and time step is reduced.
VStepMax numerical Limit the time step so the maxim change of electrode voltage is less 1.0 V
than this value
IStepMax numerical Limit the time step so the maxim change of electrode current is less 1.0 A
than this value
tran.op bool Sets if operator point should be computed before transient simula- true none
tion start. Only valid in Mixed-type Simulation. This option should
be set to false when OP already calculated.

AC Sweep Simulation

VScan string The electrode on which small AC signal is applied. none none
f.start numerical Start frequency for AC sweep 1e6 Hz
f.stop numerical Stop frequency for AC sweep. 10e9 Hz
f.multiple numerical Species multiplicative factor for incrementing frequency. 1.1 none
vac numerical The magnitude of AC signal. 0.0026 V

Input/Output

Output.Prefix string Sets the prex to the lename used for output. The internal output none none
routine and the loadable hook functions will append their respective
postx and extensions to form the complete output lename.

Device DC Sweep Mode

The DC Sweep Mode is selected by setting type=DCSweep. In this mode, Genius

scans the external voltage (or current) source at the contact selected in the VScan
parameter. At each voltage (or current) step, the device equations are solved in
steadystate.

Failure Recovery When the numerical solver fails to converge at a particular step, Genius recovers
to the state of the previous converged step, reduces the step, and attempts to solve
the equations again. In DCSweep mode, the voltage or current step is recursively
halved when the solver diverges. When the solver converges at some reduced

Genius Device Simulator 105

Genius Command Reference Numerical Solution

step, it tries to progressively double the step it recovers to the initial step (VStep
or IStep).

Device Transient Mode

The Transient Mode is selected by setting type=Transient. In this mode, ex-

ternal sources (electrical or optical sources) vary as time progresses, and Genius
solves the device equation with the time partial-derivative included.

Initial State An initial device state must be prepared prior to the transient simulation, otherwise
the device state at time zero is undetermined. Either a DC Sweep, an Operation
Point or a Transient simulation can fulll this requirement.

Adaptive Time Step In the Transient mode, one can specify whether adaptive time step is used
with the option AutoStep. When adaptive step is turned o, the initial timestep
TStart is used throughout the simulation. With AutoStep turned on, the ini-
tial time step TStep is used in the rst few steps, and the time-domain truncation
error is evaluated. If the truncation error exceeds that specied in TS.rTol and
TS.aTol, the timestep is reduced to 0.9 of the previous value. On the other hand,
when the truncation error meets the tolerance criteria, time step is increased by
the factor of 1.1, until the maximum timestep TStepMax is reached.

One can use the VStepMax and IStepMax parameters to limit the maximum
change in voltage (current) of any voltage (current) source during any time step.
When any driving source has a sudden change, the time step will be reduced ac-
cordingly.

As in the DC Sweep mode, Genius will recover from convergence failure in the
Transient mode, cut the time step to half, and continue solving.

Device Operation Point Mode

The Operation Point (OP) mode is selected by setting type=OP, and is commonly
used to bring the device to a steady-state condition, as the starting state for a DC
Sweep or Transient simulation. For better convergence, Genius will ramp up the
voltage and current sources from equilibrium (zero) to the specied source values.
To limit the maximum change in source values in each ramp-up step, one can set
the VStepMax and IStepMax parameters.

Pseudo-Time New in 1.7.2. In OP mode for device simulation, pseudo-time method can be
Method used to better handle ill-conditioned problems, e.g. devices with oating regions,
which are traditionally dicult to converge and requires a direct linear solver.
With the more numerically stable pseudo-time method, such dicult problems
can converge with iterative linear solvers such as BCGS, drastically reducing mem-
ory consumption of large-scale problems. In benchmark tests, the pseudo-time

106 Genius Device Simulator

Numerical Solution Genius Command Reference

method (with BCGS) is modestly faster than traditional steady-state method with
the MUMPS direct linear solver.

The owchart of the pseudo-time method is shown in Figure 2.7, p. 107, with
comparison to the normal steady-state method.

Inputs:
previous state S0 ,
target state S1

Compute NSteps
according to
VStepMax/IStepMax

Y N
Is pseudo-time mode?

= n/NSteps
tramp = t NSteps
S = (1 ) S0 + S1

= t/tramp
Solve PDE for state S
S = (1 ) S0 + S1

Y n = n + 1;
Solve PDE for (S, t)
n < NSteps?

Y t = t + t;
End
t < tramp ?

Solve PDE

Y t = t + t;
x < ?

End

Figure 2.7 Floatchart of the Device Operation Point

Genius Device Simulator 107

Genius Command Reference Numerical Solution

Under the pseudo-time scheme, an articial time-derivative term is added to the

electron and hole continuity equations in semiconductor regions (/ and /)
. At the same time, an articial capacitance ( /) is added between any resistive
metal region and the ground. The system of equations is thus solved in transient
mode with the nominal time step specied by the pseudotime.step parameter.

The transient simulation consists of two stages, as shown in Figure 2.7, p. 107.
In the rst stage, voltage and current sources are ramped up. After the target
voltage/current are set, Genius solves the equations for a few further steps in the
second stage. As time progresses, the device settles down to its steady state, and
the change of internal variables (e.g. , , and ) between subsequent steps di-
minishes. After this change drops below the error threshold, the device is deemed
to have reached steady-state, and the pseudo-time algorithm completes.

As in the transient and DC sweep modes, in the OP mode, Genius will recover
from convergence failure, cut back ramp step or time step, and continue solving.

Device Small-Signal AC Sweep Mode

The AC Sweep Mode is selected by setting type=ACSweep. In this mode, a

small-signal sinusoidal sources is applied on the contact specied by the VScan
parameter, and the AC response of the device is simulated. Before the AC simu-
lation, A steady-state device state must be prepared by means of OP or DC Sweep
simulation.

Device/Circuit Mixed Operation Point Mode

When a circuit netlist is present, the device/circuit mixed OP Mode is selected by

setting type=OP.

Device/Circuit Mixed DC Sweep Mode

When a circuit netlist is present, the device/circuit mixed DC Sweep Mode is se-
lected by setting type=DCSweep.

Device/Circuit Mixed Transient Mode

When a circuit netlist is present, the device/circuit mixed Transient Mode is se-

108 Genius Device Simulator

Numerical Solution Genius Command Reference

lected by setting type=Transient.

Example

SOLVE Type=Equilibrium

SOLVE Type=DCSWEEP Vscan=Anode Vstart=0.0 Vstep=0.05 \

Vstop=0.8 out.prefix=diode_iv

SOLVE Type=Transient TStart=0 TStep=1e-12 TStepMax=1e-6 \

TStop=100e-6

Genius Device Simulator 109

Genius Command Reference Numerical Solution

EMFEM2D
The EMFEM2D command instructs the 2D nite-element EM wave solver to solve
the Maxwell equations.

EMFEM2D
[ WaveLength=num ] [ Intensity=num ]
[ Quan.Eff=num ] [ SpectrumFile=str ]
[ wTE=num ] [ wTM=num ] [ phase.TE=num ] [ phase.TM=num ]
[ Angle=num ]
[ ABC.Type= ( FirstOrder | SecondOrder | PML ) ]
[ ABC.Shape= ( Circle | Ellipse | Unknow ) ]
[ LS = (CGS|BICG|BCGS|GMRES|TFQMR|LU|UMFPACK|SuperLU|
MUMPS|SuperLU_Dist) ]
[ PC = (Identity|Jacobian|BJacobian|ASM|ILU) ]

Parameters

Parameter Type Description Default Unit

WaveLength numeric The Wave length of the simulation. Synonym: Lambda. 0.532 m
Intensity numeric Name of the TIF le to import from. 0.0 W/cm2
Quan.Eff numeric Quantum eciency for photo-generation. 1.0 none
SpectrumFile string Read wavelength, intensity and quantum eciency from a text le. none none
wTE numeric The weight of TE mode in the total intensity. 0.5 none
wTM numeric The weight of TM mode in the total intensity. 0.5 none
Phase.TE numeric The initial phase of the TE mode. 0.0 degree
Phase.TM numeric The initial phase of the TM mode. 0.0 degree
Angle numeric The direction of incident plane wave, measured in degrees from the 90 degree
+x axis to the wave vector.
ABC.Type enum The type of absorbing boundary condition used at the outer bound- SecondOrder none
ary of the simulation.
ABC.Shape enum Inform Genius the shape of absorbing boundary. When the shape Uknow none
is known to be Circle or Ellipse, the calculation of boundary
curvature can be accelerated.
LS enum Selects the linear solver algorithm. BCGS none
PC enum Selects the linear preconditioner algorithm. ILU none

Description

The spectrum le is a plain text le with each line for one frequency point in the
spectrum. Each line contains 2 or 3 numerical values separated by whitespaces:

<wavelength> <intensity> [quantum efficiency]

110 Genius Device Simulator

Numerical Solution Genius Command Reference

Genius will perform FEM EM simulation at each frequency point, calculate the
optical generation at each frequency, and sum up the total carrier generation.

When quantum eciency is given, Genius will use this value to calculate the op-
tical carrier generation rate. Otherwise, Genius will use the following quantum
eciency

(/ )
= (2.9)

The sum of weights wTE and wTM must be 1. When either wTE or wTM is given,
Genius will calculate the other automatically. If both values are given, Genius will
check if they add up to unity.

Absorbing The absorbing boundary condition can be rst-order or second-order, and can be
Boundary selected by the ABC.Type parameter. In general, the second-order boundary is
preferred. When possible, a circular shape absorbing boundary is highly recom-
mended. The ABC should be at least one wave-length away from the scatterer.

Mesh Size In FEM EM simulation, the mesh size should be 1/16 to 1/20 of the optical wave-
length in the material. Note that wavelength decreases for materials with relative
permittivity greater than 1. As a result, the mesh in dielectric materials (such as
Si) should be ner.

Genius Device Simulator 111

Genius Command Reference Numerical Solution

RAYTRACE
The RAYTRACE command instructs the ray-tracing solver to calculate the light
propagation.

RAYTRACE
{
SpectrumFile=str |
( Lambda=num Intensity=num [Quan.Eff=num] )
}
[ Theta=num ] [ Phi=num ] [ Ray.Density=num ]

Parameters

Parameter Type Description Default Unit

SpectrumFile string Filename of the spectrum le. This le should describe the wave- none none
length and intensity at each frequency in the spectrum.
Lambda numerical Wavelength of the monochromatic light. Synonym: Wavelength 0.532 m
Intensity numerical Intensity of the monochromatic light. 0.0 W/cm2
Quan.Eff numerical Quantum eciency of carrier generation for the monochromatic auto none
light. If not explicitly specied, Genius will automatically calcu-
late the quantum eciency using the bandgap of the material.
Theta numerical Azimuth angle of the wave vector of the incident light. 0 degree
Phi numerical Inclination angle of the wave vector of the incident light, measured 0 degree
from the y-axis.
E.Theta numerical Azimuth angle of the E-vector of the incident light. 0 degree
E.Phi numerical Inclination angle of the E-vector of the incident light, measured from 90 degree
the y-axis.
Ray.Density numerical Density of light rays. In 2D structures, each mesh element is ex- 10 none
pected to have at least Ray.Density number of rays. In 3D struc-
tures, each mesh element has at least Ray.Density*Ray.Density
number of rays.

Description

The direction of the incident wave is specied by the azimuth and inclination angle
and , and the direction unit vector can be expressed as

sin cos
cos
= . (2.10)
sin sin

The default values of Theta and Phi corresponds to incidence along the positive
y-axis direction.

112 Genius Device Simulator

Numerical Solution Genius Command Reference

Ray Seeding Genius seeds the incident light rays automatically. The density of rays is deter-
mined by the smallest element in device mesh and the parameter ray.density.
The ray to ray distance is set to D=min /ray.density. In general, Genius guar-
antees that each mesh element has at least ray.density rays in 2D structures,
and ray.density*ray.density rays in 3D structures, respectively.

Boundaries When a ray incidents on Ohmic boundaries and Schottky boundaries, the trac-
ing terminates and the incident ray is removed. If a Neumann boundary has the
Reflection parameter set to true (default is false), the above rule also ap-
plies.

On the other hand, the interfaces with metallic (electrode) regions are not aected
by the above rule, and refraction/reection rays are generated.

Absorption When the ray passes though a mesh element, the optical power associated to the
ray deposits in this element according to the absorption coecient. Genius will
sum the power deposited by all the rays and calculate the optical carrier generation
rate at each node.

Spectrum le With the spectrum le we can specify multi- spectrums of light with dierent
intensity. Spectrum le should contain at lease two spectrums. Each line in the
spectrum le should have the following columns

The wavelength in the unit of micro-meter.

Intensity density (per unit wavelength) at this wavelength. The unit is W/cm2/m.
Note that this diers from the intensity parameter for monochromatic light.
(Optional) quantum eciency for carrier generation. If not explicitly spec-
ied, Genius will automatically calculate the quantum eciency using the
bandgap of the material.

A sample spectrum le is listed below.

# lambda intensity
0.912 0.2
0.954 0.5
1.00 0.4
1.05 0.3
1.11 0.2
1.17 0.2
1.24 0.1

Genius Device Simulator 113

Genius Command Reference Numerical Solution

Example

RayTrace wavelength=0.4 intensity=0.1 k.phi=0 ray.density=10.0

RayTrace SpectrumFile=am15a.txt k.phi=0 ray.density=10.0

114 Genius Device Simulator

Input and Output Genius Command Reference

Input and Output

IMPORT
The IMPORT command loads mesh structures and solutions from a le, and builds
the simulation system based on it.

IMPORT
{ CGNSFile=str | TIFFile=str | TIF3DFile=str |
SilvacoFile=str | ISEFile=str }

Parameters

Parameter Type Description Default Unit

CGNSFile string Name of the CGNS le to import from. none none
TIFFile string Name of the TIF le to import from. none none
TIF3DFile string Name of the TIF3D le to import from. none none
SilvacoFile string Name of the Silvaco str le to import from. none none
ISEFile string Name of the DF-ISE le to import from. none none

description

Genius can load device denition and data from many dierent les. The CFD
General Notation System (CGNS) is the native le supported by Genius. User can
save the current state of device under simulation by EXPORT command and then
load it again by IMPORT command. Besides the native CGNS format. Genius
can also load device structure from Technology Interchange Format (TIF) format
which is commonly used by Synopsys medici and tsuprem, DF-ISE format used
by Synopsys Sentaurus, structure format used by Silvaco, and a extended TIF3D
le generated by Gds2mesh tool.

Example

IMPORT CGNSFile = "foo.cgns"

IMPORT TIFFile = "foo.tif"

Genius Device Simulator 115

Genius Command Reference Input and Output

EXPORT
The EXPORT command saves mesh structures and solutions to a le.

EXPORT
[ CGNSFile=str ] [ VTKFile=str ] [ ISEFile=str ]
[ BCInfo=str ]
{ [ NodeInfo=str ] [ LUnit = (m|cm|um|nm) ]
[ Numbering=bool ] }

Parameters

Parameter Type Description Default Unit

CGNSFile string Export the mesh and data in the CGNS le with the supplied le none none
name.
VTKFile string Export the mesh and data in the VTK le with the supplied le name. none none
ISEFILE string Export device mesh and doping information in the DF-ISE format none none
into two les with supplied le name and .grd and .dat le extension,
respectively.
BCFile string Export the boundary condition denition of the structure to the le, none none
which can be loaded in another Genius input le with an INCLUDE
command.
NodeFile string Export the coordinates of the mesh points to a plain text le with the none none
supplied le name.
LUnit enum Length unit used for node coordinates, when exporting to NodeFile. um none
Numbering bool Whether to print node number in NodeFile. true none

description

Genius can save device denition and data to les in the CFD General Notation
System (CGNS) which is natively supported. Genius also saves mesh and solu-
tion data with the Visualization ToolKit (VTK) formats which is supported by
VisualTCAD and many other post-process tool such as Paraview and Visit. Since
version 1.7, Genius can export device mesh and doping information in the DF-ISE
format.

Example

EXPORT CGNSFile = "foo.cgns"

EXPORT CGNSFile = "foo.cgns" VTKFile="bar.vtk"
EXPORT ISEFile = "foo"

116 Genius Device Simulator

Input and Output Genius Command Reference

HOOK
The HOOK command loads or unloads hook functions, which are called before
and after solving the device equations. Various hook functions are available that
serve as pre-/post-processors, input/output formatters and interface with other
softwares.

HOOK
[ Load=str [ id=\GSyntax{{\ttsl str }}] [ real<str>=num [...]
] [ str<str>=str [...] ]
[ Unload=str ]

Parameters

Parameter Type Description Default Unit

Load string Name of the hook function to be loaded. none none
id string Give an identical id to hook. User can load several hooks with the none none
same type by given each hook a dierent id.
real<str> numerical Custom numerical parameters for the loaded hook function. none none
string<str> string Custom string parameters for the loaded hook function. none none
Unload enum Name of the hook function to be unloaded. none none

Description

The HOOK command should be invoked before the SOLVE command to be eective.
Some hooks are only compatible with some of the analysis modes (e.g. with DC
Sweep mode only). After a hook is loaded, it is eective in subsequent SOLVE
commands as well if the hook function is compatible with the analysis mode, until
it is unloaded.

Some hooks accepts custom parameters when the user loads them. If one tries
to load a previously loaded hook function again, the parameters specied during
previous loading are all erased.

Default Hook Functions

The following hook functions are loaded by default.

gnuplot. Save the terminal voltages and currents from the simulation result
in plain text format, which is compatible with the gnuplot ploting program.
This hook is

Genius Device Simulator 117

Genius Command Reference Input and Output

raw. Save the terminal voltages and currents from the simulation result in the
SPICE raw format. The output le has the extension .raw.

Additionally, a few optional hook functions are available, and the user can load
some of them with the HOOK command. In the following sections, we briey
describe each of them.

CV Postprocessor

The cv hook function extracts the capacitance of gate electrodes. One can load
this postprocessor with the command

HOOK Load=cv

and the extracted capacitance-voltage data will be saved in output le with the
extension .cv. The electric charge in the gate electrodes are rst integrated, and
capacitance is then extracted by dierentiating with respect to the scanning volt-
age. It works in the DC voltage sweep mode only and the extracted capacitance
is quasi-static. For capacitance analysis at high frequency, one should use the AC
analysis mode.

Probe Postprocessor

The probe hook function allows the user to monitor the potential and carrier
concentration at a particular point in the device. It can be loaded with the following
command

HOOK Load=probe id=probe1 real<x>=1.0 real<y>=1.2 real<z>=0.0

HOOK Load=probe id=probe2 real<x>=1.5 real<y>=1.2 real<z>=0.0

where the following custom parameters are recognized.

Parameter Type Description Default Unit

x numerical The x-coordinate of the probe. 0 m
y numerical The y-coordinate of the probe. 0 m
z numerical The z-coordinate of the probe. 0 m

Visualization and Animation Postprocessor

The vtk hook function allows the user to save VTK le at each step in DC and
transient simulation. The serial of VTK les which contains eld evolving infor-

118 Genius Device Simulator

Input and Output Genius Command Reference

mation can be viewed by VisualTCAD, Paraview or Visit. Furthermore, user can

make animation with the serial of VTK les with Paraview or Visit.

Interface to Visit visualizer

Genius can feed solution to the visualizer program Visit through the visit hook
function.

Genius Device Simulator 119

Genius Command Reference Input and Output

120 Genius Device Simulator

 CHAPTER

3 Extending Genius Simulator

The Genius device simulator can be extended by users in several ways.

Through the physical model interface (PMI), new materials can be added and
existing materials can be modied.
Pre-processing and post-processing steps can be added through the hook in-
terface.
The simulator is programmable through its Python interface.

This chapter outlines the procedure of extending Genius with the various inter-
faces. This chapter is intended for advanced readers who intend to write programs
to customize or extend the Genius simulator. An intermediate level of knowl-
edge about the C++ programming language and the development tools in Linux
or Windows.

Custom Material Models

Overview of PMI
Each material in Genius is packaged as a dynamically linked library (DLL). With
this design, new materials can be added, and existing materials modied, without
the need of re-building the entire simulator.

Files

The material library les are located in the directory $GENIUS_DIR/lib. The
lename for material foo is libfoo.so.

Material In addition to the material's DLL, an entry in the catalog le $GENIUS_DIR/lib/material.d

Catalog File is needed for the material model to function properly in Genius. Below is a few
entries in material.def.

Genius Device Simulator 121

Extending Genius Simulator Custom Material Models

Si
{
property = Semiconductor
alias = Silicon
color = 0xffb6c1ff
}

GaAs
{
property = Semiconductor
color = 0xeaeaeaff
}

AlGaAs
{
# mole dependence: Ga(1-x)Al(x)As
property = SingleCompoundSemiconductor
color = 0xa6d6d6ff
}

SiO2
{
property = Insulator
alias = Ox
alias = Oxide
color = 0x7d0505ff
}

Each entry corresponds to a material. An entry starts with material's ID and the
content of the entry is contained in the pair of curly braces. The content is a list of
parameter=value pairs, each in its own line. The parameter property species
the type of the material, the possible values are

Semiconductor: Elemental and simple compound semiconductors with xed

compositions.
SingleCompoundSemiconductor: Compound semiconductors with a sin-
gle mole fraction number.
ComplexCompoundSemiconductor: Complex compound semiconductors,
with more than one mole fraction number.
Conductor: metals and metallic poly-silicon
Vacuum: only used in optical simulation
PML: only used in optical simulations

Each material can have several aliases in addition to its ID. In the Material
parameter of the REGION command, one can use either the aliases or the ID of a

122 Genius Device Simulator

Custom Material Models Extending Genius Simulator

material.

Classes

PMIS_Mobility GSS_SiC4H_Mob_Analytic

PMIS_Avalanche GSS_SiC4H_Avalanche_Default

PMIS_Server
(Semiconductor)

PMIS_Trap GSS_SiC4H_Trap_Default

PMIS_Thermal GSS_SiC4H_Thermal

PMIS_Optical GSS_SiC4H_Optical

PMII_BasicParameter

PMII_Thermal

PMII_Server
(Insulator)

PMII_Optical

PMIC_BasicParameter

PMIC_Server
PMI_Server PMIC_Thermal
(Conductor)

PMIC_Optical

PMIV_Server
PMIV_BasicParameter
(Vacuum)

PMIP_Server
PMIV_Thermal
(PML)

PMIV_Optical

Figure 3.1 Class hierarchy of the PMI

The class hierarchy of the PMI interface is shown in Figure 3.1, p. 123. The
abstract base class PMI provides the basic facilities for all physical models. For
each category of materials, semiconductor, conductor or insulator, there is a class
derived from PMI (in the second column). For a semiconductor material, there

Genius Device Simulator 123

Extending Genius Simulator Custom Material Models

are several 7 aspects of its physics. Every aspect is described in a class, as shown
in the third column. For insulators and conductors, the physics are simpler. The
physics covered by each class is summarized below.

BasicParameter: mass density, dielectric permittivity, magnetic permeability

and electron anity.
BandStructure: bandgap, carrier eective mass, eective density of states,
generation/recombination, thermoionic emission and (temporarily) band-to-band
tunneling.
Mobility: carrier mobility, include its degradation at surfaces and carrier ve-
locity saturation.
Avalanche: impact ionization generation.
Trap: carrier capture and emission at trap states.
Thermal: thermal conductivity and thermal capacity.
Optical: refraction index (real and imaginary parts) at dierent wavelengths.

All the above classes are abstract and can not be instantiated. When one imple-
ments a material, one must derive from each of the above classes, and provide
a concrete implementation. For example, for the 4H-SiC material, the classes in
the last column of Figure 3.1, p. 123 are implemented and packed in the dynamic
library libSiC4H.so.

Since dynamic libraries has a C interface (instead of a C++ one), a creator function
with C-style name and calling convention, must accomany each concrete class in
the dynamic library. For example, the C function PMIS_SiC4H_BandStructure_Default
in the library libSiC4H.so is responsible for creating an instance of the class
GSS_SiC4H_BandStructure.

How a PMI Library is Loaded

We shall describe the loading sequence of physical models, assuming the follow-
ing two commands appear in the input le to Genius.

# MESH commands
# ...
REGION Label=Substrate Material=Silicon
REGION Label=GOX Material=SiO2
#----------------

# ...
PMI Region=Substrate Type=Mobility Model=HP
PMI Region=Substrate Type=Avalanche Model=HP

124 Genius Device Simulator

Custom Material Models Extending Genius Simulator

The REGION commands are read in rst, together with the meshing commands.
After all the structure specication commands are read, the initial simulation sys-
tem is built. For the region Substrate, the default material models for silicon
are loaded. For the region GOX, the default models for SiO 2 are loaded.

After the simulation system is built, the rest of the commands, including those
PMI commands are executed sequentially. When the rst PMI command is ex-
ecuted, the region Substrate is looked up and found to be of material sili-
con. The dynamic library libSi.so is loaded. From the material name, the
model type Mobility, the model name HP, we assemble the creator function's
name PMIS_Si_Mob_HP. This creator function is called, which returns an object
instance of the class GSS_Si_Mob_HP. This GSS_Si_Mob_HP class, as a derived
class of PMIS_Mobility, contains routines for calculating carrier mobility and
velocity saturation.

How a PMI Library is Used

When the Genius simulator assemble the Poisson and continuity equations set out
in Level 1 Drift-Diusion Equation, p. 1, material properties, such as doping
concentration and mole fraction are available only at each of the grid nodes. Many
derived physical quantities, such as carrier mobility and SRH recombination rate,
are evaluated at each mesh node as well. This is best illustrated with an example.

Suppose when we assemble the carrier continuity equation, we need to calculate

the bandgap narrowing due to heavy doping at a grid node. We know the material
of the current mesh node is 4H-SiC, and from Figure 3.1, p. 123, the PMI class
that handles the bandgap narrowing physics is GSS_SiC4H_BandStructure. In
fact, as a sub-class of PMIS_BandStructure, it has a method EgNarrow that
calculates bandgap narrowing. The denition of the method is shown below, also
shown is the method nie that calculates the eective intrinsic carrier concentra-
tion.

Genius Device Simulator 125

Extending Genius Simulator Custom Material Models

class GSS_SiC4H_BandStructure : public PMIS_BandStructure

{
//...

public:
double EgNarrow(const double &p, const double &n,
const double &Tl)
{
double Na = ReadDopingNa();
double Nd = ReadDopingNd();
double N = Na+Nd+1.0*std::pow(cm,-3);
double x = log(N/N0_BGN);
return V0_BGN*(x+sqrt(x*x+CON_BGN));
}

double nie (const double &p, const double &n,

const double &Tl)
{
double bandgap = Eg(Tl);
double Nc = NC300*std::pow(Tl/T300,NC_F);
double Nv = NV300*std::pow(Tl/T300,NV_F);
return sqrt(Nc*Nv)*exp(-bandgap/(2*kb*Tl))
*exp(EgNarrow(p, n, Tl));
}

// ...
};

It is obvious that this few lines implements the Slotboom model of bandgap nar-
rowing

bgn total 2
= ln + ln total + bgn . (3.1)
2 ref ( ref )

In this simple model, neither the electron, hole concentration nor the lattice tem-
perature aects the bandgap narrowing. Therefore the arguments are ignored.
The model chiey relies on the doping concentration. The ReadDopingNa and
ReadDopingNd methods, inherited from the PMIS_Server class, returns the ac-
ceptor and donor concentrations at the present mesh node, respectively.

It is also necessary to evaluate the partial derivatives of the bandgap narrowing,

with respective to all unknowns in the system of equations. This may sound an in-
volved task. However, Genius uses the automatic dierentiation technology which
hides most of the complications from the user. The following snippet shows the
method that calculates the partial derivatives. It is also named EgNarrow, but

126 Genius Device Simulator

Custom Material Models Extending Genius Simulator

note the dierence in the type of arguments.

class GSS_SiC4H_BandStructure : public PMIS_BandStructure

{
//...

public:

AutoDScalar EgNarrow(const AutoDScalar &p, const AutoDScalar

&n,
const AutoDScalar &Tl)
{
double Na = ReadDopingNa();
double Nd = ReadDopingNd();
double N = Na+Nd+1.0*std::pow(cm,-3);
double x = log(N/N0_BGN);
return V0_BGN*(x+sqrt(x*x+CON_BGN));
}

AutoDScalar nie (const AutoDScalar &p, const AutoDScalar &n,

const AutoDScalar &Tl)
{
AutoDScalar bandgap = Eg(Tl);
AutoDScalar Nc = NC300*adtl::pow(Tl/T300,NC_F);
AutoDScalar Nv = NV300*adtl::pow(Tl/T300,NV_F);
AutoDScalar ret=sqrt(Nc*Nv)*exp(-bandgap/(2*kb*Tl))
*exp(EgNarrow(p, n, Tl));
return ret;
}
// ...

};

It is clear that the "derivative" version of EgNarrow and nie are almost identical
to the "normal" version. In almost all situations, it is sucient to replace the scalar
type double with the class type AutoDScalar.

Genius Device Simulator 127

Extending Genius Simulator Custom Material Models

Procedure of Building a Custom Material Library

The procedure of building a custom material library is outlined in this section.

Building the Example

Step 1 Run the congure script.

cd examples/Material
./configure.sh

Step 2 Enter the material directory and modify the source les as needed.

cd 4H-SiC

Step 3 Build the material library.

make

Step 4 Copy the library le libSiC4H.so to $GENIUS_DIR/lib/, backing up the old

le is strongly recommended. The new material library is ready for use now.

Create a New Material Library

Step 1 Make a copy of some existing model.

cp -R 4H-SiC 6H-SiC

Step 2 Replace the material name string in the all source les, with the new material
name, and rename the les.

sed -i -e "s/SiC4H/SiC6H/g" makefile

sed -i -e "s/SiC4H/SiC6H/g" *.cc
rename "s/4H-SiC/6H-SiC/" *.cc

Step 3 Edit the source, and build the material library.

cd 6H-SiC
make

Step 4 Copy the library le libSiC4H.so to $GENIUS_DIR/lib/, and edit the catalog
le $GENIUS_DIR/lib/material.def

128 Genius Device Simulator

Custom Material Models Extending Genius Simulator

SiC6H
{
property = Semiconductor
alias = 6HSiC
color = 0xffb6c1ff
}

The new material library is ready for use now.

Genius Device Simulator 129

Extending Genius Simulator Custom Material Models

Class References
The class reference presented here is a selected and adapted version of the classes
declared in the PMI development kit. Some macros are expanded and some inter-
nal details are omitted, so that this guide is easier to follow.

PMI_Server Class

PMI_Server(const PMI_Environment &env);

Constructor.

void ReadCoordinate (double& x, double& y, double& z) const

Return the spatial coordinates of the mesh node.

double ReadTime() const

Return the current time in simulation. This is only meaningful in transient-mode

simulations.

double ReadUserScalarValue (std::string &name) const

Return the scalar value with the given name at the current node. An examples of
the scalar values at a node is the density of trap state.

PMIS_Server Class

double ReadxMoleFraction () const

Return the mole fraction x at the current node. Only used in compound semicon-
ductor with variable compositions.

double ReadxMoleFraction (const double mole_xmin,

const double mole_xmax) const

Return the mole fraction x at the current node, with upper and lower limits.

130 Genius Device Simulator

Custom Material Models Extending Genius Simulator

double ReadyMoleFraction () const

Return the mole fraction y at the current node. Only used in compound semicon-
ductor with variable compositions.

double ReadyMoleFraction (const double mole_ymin,

const double mole_ymax) const

Return the mole fraction y at the current node, with upper and lower limits.

double ReadDopingNa () const

Return the acceptor dopant concentration.

double ReadDopingNd () const

Return the donor dopant concentration.

PMIS_BasicParameter Class

virtual double Density (const PetscScalar &Tl) const = 0

Return the mass density g/cm3 of the material.

virtual double Permittivity () const = 0

Return the relative permittivity of the material.

virtual double Permeability () const = 0

Return the relative permeability of the material.

virtual double Affinity (const double &Tl) const = 0

Return the anity energy [eV] of the material.

PMIS_BandStructure Class

Unless otherwise stated, the arguments of the following methods are

Genius Device Simulator 131

Extending Genius Simulator Custom Material Models

p hole concentration
n electron concentration
Tl lattice temperature
Tp hole temperature
Tn electron temperature

virtual double Eg (const PetscScalar &Tl) = 0

virtual AutoDScalar Eg (const AutoDScalar &Tl) = 0

Return the band gap of semiconductor.

virtual double EgNarrow (const double &p,

const double &n,
const double &Tl) = 0
virtual AutoDScalar EgNarrow (const AutoDScalar &p,
const AutoDScalar &n,
const AutoDScalar &Tl) = 0

Return the band gap narrowing due to heavy doping.

virtual double EgNarrowToEc (const double &p,

const double &n,
const double &Tl) = 0
virtual AutoDScalar EgNarrowToEc (const AutoDScalar &p,
const AutoDScalar &n,
const AutoDScalar &Tl) = 0

Return the conduction band shift due to band gap narrowing.

virtual double EgNarrowToEv (const double &p,

const double &n,
const double &Tl) = 0
virtual AutoDScalar EgNarrowToEv (const AutoDScalar &p,
const AutoDScalar &n,
const AutoDScalar &Tl) = 0

Return the valence band shift due to band gap narrowing.

virtual double EffecElecMass (const double &Tl) =

0
virtual AutoDScalar EffecElecMass (const AutoDScalar &Tl) =
0

132 Genius Device Simulator

Custom Material Models Extending Genius Simulator

Return the eective mass of electrons (DOS mass).

virtual double EffecHoleMass (const double &Tl) =

0
virtual AutoDScalar EffecHoleMass (const AutoDScalar &Tl) =
0

Return the eective mass of holes (DOS mass).

virtual double Nc (const double &Tl) = 0

virtual AutoDScalar Nc (const AutoDScalar &Tl) = 0

Return the eective density of states in the conduction band.

virtual double Nv (const double &Tl) = 0

virtual AutoDScalar Nv (const AutoDScalar &Tl) = 0

Return the eective density of states in the valence band.

virtual double ni ( const PetscScalar &Tl) = 0

Return the intrinsic carrier concentration.

virtual double nie (const double &p,

const double &n,
const double &Tl) = 0
virtual AutoDScalar nie (const AutoDScalar &p,
const AutoDScalar &n,
const AutoDScalar &Tl) = 0

Return the eective intrinsic carrier concentration.

virtual PetscScalar TAUN (const double &Tl) = 0

virtual PetscScalar TAUP (const double &Tl) = 0

Return the minority carrier life time due to SHR recombination.

Genius Device Simulator 133

Extending Genius Simulator Custom Material Models

virtual double R_Direct (const double &p,

const double &n,
const double &Tl) = 0
virtual AutoDScalar R_Direct (const AutoDScalar &p,
const AutoDScalar &n,
const AutoDScalar &Tl) = 0

Return the direct (interband) recombination rate.

virtual double R_Auger (const double &p,

const double &n,
const double &Tl) = 0
virtual AutoDScalar R_Auger (const AutoDScalar &p,
const AutoDScalar &n,
const AutoDScalar &Tl) = 0

Return the Auger recombination rate.

virtual double R_Auger_N (const double &p,

const double &n,
const double &Tl) = 0
virtual AutoDScalar R_Auger_N (const AutoDScalar &p,
const AutoDScalar &n,
const AutoDScalar &Tl) = 0
virtual double R_Auger_P (const double &p,
const double &n,
const double &Tl) = 0
virtual AutoDScalar R_Auger_P (const AutoDScalar &p,
const AutoDScalar &n,
const AutoDScalar &Tl) = 0

Return the rates of electron and hole initiated Auger recombination.

virtual double R_SHR (const double &p,

const double &n,
const double &Tl) = 0
virtual AutoDScalar R_SHR (const AutoDScalar &p,
const AutoDScalar &n,
const AutoDScalar &Tl) = 0

Return the rate of SRH recombination.

134 Genius Device Simulator

Custom Material Models Extending Genius Simulator

virtual PetscScalar R_Surf (const double &p,

const double &n,
const double &Tl,
const double &len) =0;
virtual AutoDScalar R_Surf (const AutoDScalar &p,
const AutoDScalar &n,
const AutoDScalar &Tl,
const PetscScalar &len) =0;

Return the rate of SRH recombination.

virtual double R_SHR (const double &p,

const double &n,
const double &Tl) = 0
virtual AutoDScalar R_SHR (const AutoDScalar &p,
const AutoDScalar &n,
const AutoDScalar &Tl) = 0

Return the rate of SRH and surface recombination at semiconductor-insulator in-

terface.

virtual double Recomb (const double &p,

const double &n,
const double &Tl) = 0
virtual AutoDScalar Recomb (const AutoDScalar &p,
const AutoDScalar &n,
const AutoDScalar &Tl) = 0

Return the total recombination rate of all recombination mechanisms.

virtual double ElecEnergyRelaxTime(const double &Tn,

const double &Tl)
= 0
virtual AutoDScalar ElecEnergyRelaxTime(const AutoDScalar &Tn,
const AutoDScalar &Tl)
= 0

Return the electron energy relaxation time. Used in EBM simulations.

Genius Device Simulator 135

Extending Genius Simulator Custom Material Models

virtual double HoleEnergyRelaxTime(const double &Tn,

const double &Tl)
= 0
virtual AutoDScalar HoleEnergyRelaxTime(const AutoDScalar &Tn,
const AutoDScalar &Tl)
= 0

Return the hole energy relaxation time. Used in EBM simulations.

virtual double SchottyJsn (double n,

double Tl,
double Vb) = 0
virtual AutoDScalar SchottyJsn (AutoDScalar n,
AutoDScalar Tl,
AutoDScalar Vb) = 0

Return the electron current density at Schottky contacts. The argument Vb is the
electron barrier height.

virtual double SchottyJsp (double n,

double Tl,
double Vb) = 0
virtual AutoDScalar SchottyJsp (AutoDScalar n,
AutoDScalar Tl,
AutoDScalar Vb) = 0

Return the hole current density at Schottky contacts. The argument Vb is the
electron barrier height.

virtual double SchottyBarrierLowerring (double eps,

double E) = 0

Return the total recombination rate of all recombination mechanisms.

virtual double ARichN() = 0

virtual double ARichP() = 0

Return the Richardson constant for electrons and holes, respectively.

virtual double ThermalVn (double Tl) = 0

virtual AutoDScalar ThermalVn (AutoDScalar Tl) = 0

136 Genius Device Simulator

Custom Material Models Extending Genius Simulator

Return the electron thermal emission velocity at hetero-junctions

virtual double ThermalVp (double Tl) = 0

virtual AutoDScalar ThermalVp (AutoDScalar Tl) = 0

Return the Richardson constant for electrons and holes, respectively.

virtual double BB_Tunneling(const double &Tl,

const double &E) = 0
virtual AutoDScalar BB_Tunneling(const AutoDScalar &Tl,
const AutoDScalar &E) = 0

Return the band-to-band tunneling generation rate. The argument E is the magni-
tude of electric eld.

PMIS_Mobility

Unless otherwise stated, the arguments of the following methods are

p hole concentration
n electron concentration
Tl lattice temperature
Tp hole temperature
Tn electron temperature

virtual double ElecMob (const double &p,

const double &n,
const double &Tl,
const double &Ep,
const double &Et,
const double &Tn) const=0
virtual AutoDScalar ElecMob (const AutoDScalar &p,
const AutoDScalar &n,
const AutoDScalar &Tl,
const AutoDScalar &Ep,
const AutoDScalar &Et,
const AutoDScalar &Tn) const=0

Return the electron mobility. The argument Ep and Et are the parallel and trans-
verse components of the electric eld, respectively.

Genius Device Simulator 137

Extending Genius Simulator Custom Material Models

virtual double HoleMob (const double &p,

const double &n,
const double &Tl,
const double &Ep,
const double &Et,
const double &Tp) const=0
virtual AutoDScalar HoleMob (const AutoDScalar &p,
const AutoDScalar &n,
const AutoDScalar &Tl,
const AutoDScalar &Ep,
const AutoDScalar &Et,
const AutoDScalar &Tp) const=0

Return the hole mobility. The argument Ep and Et are the parallel and transverse
components of the electric eld, respectively.

PMIS_Avalanche

Unless otherwise stated, the arguments of the following methods are

Tl lattice temperature
Tp hole temperature
Tn electron temperature
Ep driving force of impact ionization (parallel electric eld).
Eg bandgap

virtual double ElecGenRate (const double &Tl,

const double &Ep,
const double &Eg) const
= 0
virtual AutoDScalar ElecGenRate (const AutoDScalar &Tl,
const AutoDScalar &Ep,
const AutoDScalar &Eg) const
= 0

Return the generation rate of electron-initiated impact ionization.

138 Genius Device Simulator

Custom Material Models Extending Genius Simulator

virtual double HoleGenRate (const double &Tl,

const double &Ep,
const double &Eg) const
= 0
virtual AutoDScalar HoleGenRate (const AutoDScalar &Tl,
const AutoDScalar &Ep,
const AutoDScalar &Eg) const
= 0

Return the generation rate of hole-initiated impact ionization.

virtual double ElecGenRateEBM (const double &Tn,

const double &Tl,
const double &Eg) const
= 0
virtual AutoDScalar ElecGenRateEBM (const AutoDScalar &Tn,
const AutoDScalar &Tl,
const AutoDScalar &Eg) const
= 0

Return the generation rate of electron-initiated impact ionization, calculate from

the electron temperature. Used in EBM simulation only.

virtual double HoleGenRateEBM (const double &Tp,

const double &Tl,
const double &Eg) const
= 0
virtual AutoDScalar HoleGenRateEBM (const AutoDScalar &Tp,
const AutoDScalar &Tl,
const AutoDScalar &Eg) const
= 0

Return the generation rate of hole-initiated impact ionization, calculate from the
electron temperature. Used in EBM simulation only.

PMIS_Trap

[[TODO]]

PMIS_Thermal

The argument Tl is the lattice temperature.

Genius Device Simulator 139

Extending Genius Simulator Custom Material Models

virtual double HeatCapacity (const double &Tl) const

= 0
virtual AutoDScalar HeatCapacity (const AutoDScalar &Tl) const
= 0

Return the heat capacity JK1 cm3 of the material.

virtual double HeatConduction(const double &Tl) const

= 0
virtual AutoDScalar HeatConduction(const AutoDScalar &Tl) const
= 0

Return the heat conduction W/cm/K of the material.

PMIS_Optical

The argument lambda is the wavelength, Eg the bandgap, and Tl the lattice tem-
perature.

virtual std::complex<double> RefractionIndex(double lamda,

double Eg,
double Tl) const
= 0

140 Genius Device Simulator

Custom Material Models Extending Genius Simulator

Return the heat capacity JK1 cm3 of the material.

PMII

empty

PMII_BasicParameter

See PMIS_BasicParameter Class, p. 131.

PMII_Thermal

See PMIS_Thermal, p. 139.

PMII_Optical

See PMIS_Optical, p. 140.

PMIC

empty

PMIC_BasicParameter

virtual double Density (const PetscScalar &Tl) const = 0

Return the mass density g/cm3 of the material.

virtual double Permittivity () const = 0

Return the relative permittivity of the material.

virtual double Permeability () const = 0

Return the relative permeability of the material.

Genius Device Simulator 141

Extending Genius Simulator Custom Material Models

virtual double Affinity (const double &Tl) const = 0

Return the anity energy [eV] of the material.

virtual double Conductance () const = 0

Return the electrical conductance of the material.

PMIC_Thermal

See PMIS_Thermal, p. 139.

PMIC_Optical

See PMIS_Optical, p. 140.

142 Genius Device Simulator

Hooks Extending Genius Simulator

Hooks
[[TODO]]

Genius Device Simulator 143

Extending Genius Simulator Python Interface

Python Interface
[[TODO]]

HTTP Frontend
[[TODO]]

144 Genius Device Simulator

 A
References
[1] U. Lindefelt. Current-Density Relations for Nonisothermal Modeling of De-
generate Heterostructure Devices. J. Appl. Phys., 75(2):958966, Jan. 1994.
[2] S. Selberherr. Analysis and Simulation of Semiconductor Devices. Springer-Ver-
lag, 1984.
[3] Yu Zhiping, D. Chen, L. So and R. W. Dutton. PISCES-2ET and Its Applica-
tion Subsystems. Technical Report, Integrated Circuits Laboratory, Stanford
University, 1994.
[4] Andreas Aste and R. Vahldieck. Time-domain simulation of the full hydro-
dynamic model. Int. J. Numer. Model., 16:161174, 2003.
[5] S. M. Sze. Physics of Semiconductor Devices, 2nd edition.. New York: John
Wiley & Sons, 1981.
[6] J. W. Slotboom. The pn Product in Silicon. Solid-State Electronics,
20:279283, 1977.
[7] D. J. Roulston, N. D. Arora and S. G. Chamberlain. Modeling and Mea-
surement of Minority-Carrier Lifetime Versus Doping in Diused Layers
of + Silicon Diodes. IEEE Trans. Electron Devices, ED-29:284291,
1982.
[8] S. Selberherr. Process and Device Modeling for VLSI. Microelectron. Re-
liab., 24(2):225257, 1984.
[9] G. Masetti, M. Severi and S. Solmi. Modeling of carrier mobility against car-
rier concentration in arsenic-, phosphorus-, and boron-doped silicon. IEEE
Trans. Electron Devices, 30(7):764-769, 1983.
[10] D. B. M. Klaassen. A Unied Mobility Model for Device Simulation - I.
Solid-State Electronics, 35:953959, 1992.
[11] D. B. M. Klaassen. A Unied Mobility Model for Device Simulation - II.
Solid-State Electronics, 35:961967, 1992.
[12] D. M. Caughey and R. E. Thomas. Carrier Mobilities in Silicon Empirically
Related to Doping and Field. Proc. IEEE, 55:21922193, 1967.
[13] J. J. Barnes, R. J. Lomax and G. I. Haddad. Finite-Element Simulation of
GaAs MESFET's with Lateral Doping Proles and Sub-Micron Gates. IEEE
Trans. Electron Devices, ED-23:10421048, 1976.
[14] H.R. Yeager and R.W. Dutton. Circuit-simulation models for the high elec-
tron-mobility transistor. IEEE Trans. Electron Devices, 33:682692, 1986.
[15] C. Lombardi, S. Manzini, A. Saporito and M. Vanzi. A Physically Based
Mobility Model for Numerical Simulation of Nonplanar Devices. IEEE
Trans. Computer-Aided Design, 7(11):11641170, 1988.
[16] M. N. Darwish, J. L. Lentz, M. R. Pinto, P. M. Zeitzo, T. J. Krut-
sick and H. H. Vuong. An Improved Electron and Hole Mobility Model
for General Purpose Device Simulation. IEEE Trans. Electron Devices,
44(9):15291538, 1997.
[17] K.M. Cham, S-Y Oh, D. Chin and J.L. Moll. Computer-Aided Design and
VLSI Device Development. Kluwer Academic Publishers, 1986.

Genius Device Simulator 145

References Python Interface

[18] A.G. Chynoweth. Ionization rates for electron and holes in silicon. Physical
Review, 109(5):1537-1540, 1958.
[19] M. Valdinoci, D. Ventura, M.C. Vecchi, M. Rudan, G. Baccarani, F. Illien,
A. Stricker and L. Zullino. Impact-ionization in silicon at large operating
temperature. In International Conference on Simulation of Semiconductor
Processes and Devices, 1999.
[20] E.O. Kane. Zener tunneling in semiconductors. J. Phys. Chem. Solids,
12:181188, 1959.
[21] J. J. Liou. Modeling the Tunneling Current in Reverse-Biased p/n Junctions.
Solid-State Electronics, 33:971972, 1990.
[22] W. B. Joyce and R. W. Dixon. Analytic Approximation for the Fermi Energy
of an Ideal Fermi Gas. Appl. Phys. Lett., 31:354356, 1977.
[23] Yu Zhiping and R. W. Dutton. SEDAN III - A Generalized Electronic Mate-
rial Device Analysis Program. Technical Report, Stanford Electronics Lab-
oratory Technical Report, July 1985.

146 Genius Device Simulator

 B
Index
G
A GLOBAL command 38
ATTACH command 76
Auger recombination 10 H
Heat ow equation
B DDML2 4
Bandgap narrowing EBML3 6
Slotboom model 7 Hook command 117
Band-to-band tunneling
Kane's model 30 I
BOUNDARY command 71 Impact ionization 26
Selberherr model 26
C Valdinoci model 28
Carrier life-time 10 van Overstraeten - de Man model
CONTACT command 71 27
Continuity equation IMPORT command 115
DDML1 2 ISOURCE command 60
ENVELOP command 67
D SOURCEAPPLY command 69
DDML1 CIRCUIT command 70
Equation 1
DDML2 4 L
Density of states Light command 64
eective 7
Direct recombination 10 M
DOPING command 79 MESH command 41
Drift-diusion current METHOD command 97
DDML1 2 mobility
DDML2 4 Analytic model 13
EBML3 5 HP model 22
Lombardi model 19
E Lucent model 20
EBML3 Masetii model 13
Equation 5 Philips model 14
ELIMINATE command 45 MODEL command 95
EMFEM2D command 110 MOLE command 85
Energy-balance equation REGIONSET command 87
EBML3 5
EXPORT command 116 P
PARTICLE command 62
F PMI command 88
FACE command 52 Poisson's Equation 1
Fermi-Dirac distribution 32 PROFILE command 80

Genius Device Simulator 147

Index Python Interface

R Caughey-Thomas model 16
RAYTRACE command 112 GaAs-like 17
REFINE command 54 hyper-tangent model 17
REGION command 50 VSOURCE command 57
Material Parameter 122
X
S X.Mesh command 43
Shockley-Read-Hall recombination
see SRH recombination Y
SOLVE command 102 Y.MESH command 43
SPREAD command 47
SPREAD3D command 48 Z
SRH recombination 10 Z.MESH command 43

V
velocity saturation 16

148 Genius Device Simulator