[PPT] - VICTORY PROCESS Full Physical 3D Semiconductor Simulator VICTORY PowerPoint Presentation

SLIDE 1

VICTORY PROCESS

Full Physical 3D Semiconductor Simulator

SLIDE 2

Full Physical 3D Semiconductor Simulator

VICTORY Process – 3D Process Simulator

VICTORY Process provides the capability to simulate

comprehensive full process flows

Etching, Deposition, Lithography
Oxidation, Stress
Implantation
Diffusion
Self explanatory process flow description
Based on ATHENA syntax
Open interface for modeling
Model parameters and functions can be accessed and modified
Open C-code is used to implement the model functions
Pre-compiled library is provided
Makefile to create an extend model library is provided
2 -

SLIDE 3

Full Physical 3D Semiconductor Simulator

VICTORY Process – Numerics

Level set based framework
for structure representation and
for interface propagation
Very stable surface propagation algorithms
Important for Etching, Deposition, Oxidation
Based on the multi-layer structure concept
Automatic void detection
Avoids the problem of loop creation and correction
3 -

SLIDE 4

Full Physical 3D Semiconductor Simulator

VICTORY Process – Meshes

VICTORY Process uses two types of meshes
Hierarchical Cartesian mesh for implicit geometry representation
Irregular Cartesian mesh for volume data representation
Doping, Stress, .....
Manual and automatic mesh refinement
Manual refinement of the geometry mesh
Automatic refinement of the geometry mesh
Automatic set-up of volume data mesh
Manual refinement of the volume data mesh
4 -

SLIDE 5

Full Physical 3D Semiconductor Simulator

VICTORY Process – Etching/Deposition/Lithography

Comprehensive mask support
GDSII – format masks
lay – format masks (MaskViews)
Definition of mask polygons inside the processing deck
Mask variations via the deck (shrink and expand)
Selection of a simulation window
Lithography
Calculation of aerial images
Pattern transfer of aerial images
5 -

SLIDE 6

Full Physical 3D Semiconductor Simulator

VICTORY Process – Etching/Deposition/Lithography (con’t)

Geometrical Etching
Idealized directional mask pattern or image transfer
Pattern transfer with tilted sidewalls and rounded corners
Idealized isotropic, dry and directional etching
Selective and non-selective mode
Fast structure prototyping
6 -

SLIDE 7

Full Physical 3D Semiconductor Simulator

VICTORY Process – Etching/Deposition/Lithography (con’t)

Geometrical Deposition
Idealized vertical resists or material regions defined by a mask
Idealized conformal deposition
Deposition of features with tilted sidewalls and rounded corners
Planarisation mode to partially fill holes
7 -

SLIDE 8

Full Physical 3D Semiconductor Simulator

VICTORY Process – Etching/Deposition/Lithography (con’t)

Geometrical CMP
Idealized planarization
Selective and non-selective mode
8 -

Selective Non-Selective

SLIDE 9

Full Physical 3D Semiconductor Simulator

VICTORY Process – Etching/Deposition/Lithography (con’t)

Physical Etching
Selective etching of several materials
Comprehensive set of default models
Selective Etching
Isotropic, Anisotropic, Directional
Crystal orientation dependent
Plasma Etching
Ion-Milling
Material dependent yield functions
Re-deposition (material dependent efficiencies)
Rotating beams, static beams, divergent beams
Reactive Ion Etching
Deep Reactive Ion Etching
Ballistic transport of reactants
High performance due to multi-threading
Near linear speedup on multi-core machines
9 -

SLIDE 10

Full Physical 3D Semiconductor Simulator

VICTORY Process – Etching/Deposition/Lithography (con’t)

10 -

Ion milling with re-deposition

SLIDE 11

Full Physical 3D Semiconductor Simulator

VICTORY Process – Etching/Deposition/Lithography (con’t)

Physical Deposition
Comprehensive set of default models
Conformal Deposition
Non-conformal Deposition
Directional Deposition
Selective Deposition
Empirical epitaxial growth
Ion beam deposition
User accessible yield functions
Rotating beams, static beams and divergent beams
Sputter deposition
Ion assisted sputter deposition
Ballistic transport of reactants
High performance due to multi-threading
Near linear speedup on multi-core machines
11 -

SLIDE 12

Full Physical 3D Semiconductor Simulator

VICTORY Process – Etching/Deposition/Lithography (con’t)

12 -

Ion beam deposition with tilted beam

SLIDE 13

Full Physical 3D Semiconductor Simulator

VICTORY Process – Etching/Deposition/Lithography (con’t)

Physical Etching and Deposition - Open modeling system

 Open C-function model library  User definable transport characteristics ( flux model )

 Define particles involved in the process  Determines how the reactants approach the wafer surface  Determines how the reactants are re-emitted from the surface

 User definable surface reaction ( surface reaction model )

 Determines how the local etch rate is calculated

 User definable etching / deposition models ( topography model )

 Defines the link between particle transport and surface reaction

 C-function models for material characteristics

 Etch rates, conformity, anisotropy, sticking efficiency  Technological models (e.g etch rate versus gas flow)

13 -

SLIDE 14

Full Physical 3D Semiconductor Simulator

Modes for Flux Calculation

Primary-Ballistic Mode:

Only particles approaching the surface

directly from the reactor chamber (primary particles) are taken into account

The velocity distribution function of the

primary particles in the vicinity of the wafer is a modeling function definable by the user

14 -

SLIDE 15

Full Physical 3D Semiconductor Simulator

Modes for Flux Calculation (con’t)

Reflective-Ballistic Mode:

Along with primary sources (reactor

chamber) also secondary sources (particles reflected or emitted from surface) are taken into account.

The velocity distribution function of the

primary and secondary particles are modeling functions definable by the user

15 -

SLIDE 16

Full Physical 3D Semiconductor Simulator

Modes for Flux Calculation (con’t)

Rate-Coupled-Reflective-Ballistic Mode:

Along with primary sources (reactor

chamber) also secondary sources (particles reflected

r

emitted from surface) are taken into account.

The velocity distribution function of the

primary and secondary particles are modeling functions definable by the user

The amount of secondary particles is a

function of the rate (e.g redeposition)

16 -

SLIDE 17

Full Physical 3D Semiconductor Simulator

VICTORY Process – Ion Implantation – Analytical

Analytical and Monte-Carlo methods for ion implantation
Analytical method uses a well established database for a wide

range of implantation conditions

Gaussian profiles, Pearson profiles and Dual Pearson profiles
User definable profiles
By means of moments of the profile
By means of data files
Range scaling techniques are applied to take into account multi-

material layers

High performance due to
Very efficient numerical algorithms
Efficient multi-threading
17 -

SLIDE 18

Full Physical 3D Semiconductor Simulator

VICTORY Process – Ion Implantation – Monte-Carlo

Monte Carlo method is more appropriate
For complex geometries
For large tilted implants

because many effects cannot be accounted for by even the most elaborated analytical procedure

Monte Carlo method takes into account all important implantation

effects:

Ion channelling
Ion dose dependency due to damage accumulation
Effect of multiple layers of different materials
Partial shadowing of ion flux for tilted implants
Multiple ion reflections within structures with complex 3D geometry
Lateral ion scattering from mask walls or so-called implantation proximity

effect

18 -

SLIDE 19

Full Physical 3D Semiconductor Simulator

VICTORY Process – Ion Implantation – Monte-Carlo (con‘t)

Is completely physically based
Uses best available electronic and nuclear stopping models
Acceleration techniques
Trajectory splitting
High performance due to efficient multi-threading
19 -

SLIDE 20

Full Physical 3D Semiconductor Simulator

VICTORY Process – Ion Implantation – Monte-Carlo (con‘t)

Implant at tilt 50 degrees and twist along structure’s diagonal.

5 million trajectories can be simulated in a few minutes on an 8 core machine
20 -

SLIDE 21

Full Physical 3D Semiconductor Simulator

Primary, i.e. direct impact implantation Shadowed, i.e. secondary impact implantation

VICTORY Process – Ion Implantation – Monte-Carlo (con‘t)

Implant at tilt 50 degrees and twist along structure’s diagonal.

All geometrical effects are taken into account
21 -

Ion distribution in silicon

SLIDE 22

Full Physical 3D Semiconductor Simulator

VICTORY Process – Annealing/Diffusion

Based on an Open Modeling System
PDE system is extendable by the user
Modelling species can be added
Reactions between species can be extended and modified
Comprehensive set of default models
Direct, Fermi, FullCPL, Single-Pair, Five-stream
Dopant activation and solid solubility
Impurity segregation at material interfaces
Point defect trapping
Point defect clustering
Concurrent simulation of several impurities
22 -

SLIDE 23

Full Physical 3D Semiconductor Simulator

VICTORY Process – Annealing/Diffusion (con‘t)

 Finite difference scheme is applied to discretize the PDE‘s  Solved on a Cartesian mesh

 Guarantees high performance and high stability  Irregular Cartesian mesh is used to facilitate local refinement  Interface points are inserted to properly grasp material interfaces  Free line points are inserted to resolve high gradients near interfaces

 Special numerical discretization for

 interface points and  free line points

 Very efficient multi-threading

 High multi-core performance gain

23 -

SLIDE 24

Full Physical 3D Semiconductor Simulator

VICTORY Process – Annealing/Diffusion (con‘t)

24 -

Mesh for annealing simulation

SLIDE 25

Full Physical 3D Semiconductor Simulator

VICTORY Process – Annealing/Diffusion (con‘t)

Default models :
Direct model
Fermi model
FullCPL model
25 -

( )

D

J t C   ⋅ ∇ − = ∂ ∂ C D J

C D

∇ ⋅ − =         ⋅ ⋅ ⋅ ⋅ − ∇ ⋅ − = T k E q C Z C D J

C F

   ( )

F

J t C   ⋅ ∇ − = ∂ ∂ ( )

CPL

J t C   ⋅ ∇ − = ∂ ∂

∑

=

      ⋅ ⋅ ⋅         ⋅ −         ∇ ⋅ − =

V I X X X A X X A AX X CPL

T k E q C C C Z C C C D f J

, * *

  

SLIDE 26

Full Physical 3D Semiconductor Simulator

VICTORY Process – Annealing/Diffusion (con‘t)

Default models :
Five-Stream model
26 -

( ) ( ) ( )

AV AI A V I I I I AV I A V I I I

C C C C C R J R R R J t C

s

,

, , , + ⋅ ∇ − = + + + ⋅ ∇ − = ∂ ∂

↔ ↔ ↔

    ( ) ( ) ( )

AV AI A V I V V V AI V A V I V V

C C C C C R J R R R J t C

s

,

, , , + ⋅ ∇ − = + + + ⋅ ∇ − = ∂ ∂

↔ ↔ ↔

    ( )

AV AI A V I A I AV V AI AV AI A

C C C C C R R R R R t C

s s s

,

, , , = − − + = ∂ ∂

↔ ↔

( ) ( ) ( )

AV AI A V I AV AV I AV AV AV AV

C C C C C R J R R J t C

s

,

, , , + ⋅ ∇ − = + − ⋅ ∇ − = ∂ ∂

↔

    ( ) ( ) ( )

AV AI A V I AI AI V AI AI AI AI

C C C C C R J R R J t C

s

,

, , , + ⋅ ∇ − = + − ⋅ ∇ − = ∂ ∂

↔

   

All model species are defined in a C-function model file

SLIDE 27

Full Physical 3D Semiconductor Simulator

VICTORY Process – Annealing/Diffusion (con‘t)

Default models :
Five-Stream model
27 -

   

Reactions are defined in a C-function model file

SLIDE 28

Full Physical 3D Semiconductor Simulator

VICTORY Process – Annealing/Diffusion (con‘t)

Default models :
Five-Stream model
28 -

   

Stream functions are selected in a C-function model file

SLIDE 29

Full Physical 3D Semiconductor Simulator

Default models :
Single-Pair model
Simplification of the Five-Stream model where one dopant defect-pair is

neglected

Special model variant for interstitial supersaturation where only dopant-

interstitial pairs are taken into account

Is about 30% faster than the full Five-Stream model

VICTORY Process – Annealing/Diffusion (con‘t)

29 -

( ) ( ) ( )

AV AI A V I V V V AI V A V I V V

C C C C C R J R R R J t C

s

,

, , , + ⋅ ∇ − = + + + ⋅ ∇ − = ∂ ∂

↔ ↔ ↔

    ( ) ( ) ( )

AV AI A V I I I I AV I A V I I I

C C C C C R J R R R J t C

s

,

, , , + ⋅ ∇ − = + + + ⋅ ∇ − = ∂ ∂

↔ ↔ ↔

    ( )

AV AI A V I A I AV V AI AV AI A

C C C C C R R R R R t C

s s s

,

, , , = − − + = ∂ ∂

↔ ↔

( ) ( ) ( )

AV AI A V I I AV I AV V I I AV I AV I AV I AV

C C C C C R J R R J t C

s

, / / / / / / /

, , , + ⋅ ∇ − = + − ⋅ ∇ − = ∂ ∂

↔

   

SLIDE 30

Full Physical 3D Semiconductor Simulator

Implant/Diffusion – Annealing/Diffusion (con‘t)

30 -

 Simulation time scales with the complexity of the model and with the number of nodes  Diffusion simulation may last only few minutes with simple models (Fermi)

Implant arsenic energy=30.0 dose=3e15 \ tilt = 0 plus.one \ cluster311 clust.fact=0.1 \ min.clust=1e16 max.clust=1e19 Method material=silicon model=fullcpl Diffuse temp = 1000.0 time = 10 sec Implant arsenic energy=30.0 dose=3e15 \ tilt = 0 plus.one \ cluster311 clust.fact=0.1 \ min.clust=1e16 max.clust=1e19 Method material=silicon model=fullcpl Diffuse temp = 1000.0 time = 10 sec

SLIDE 31

Full Physical 3D Semiconductor Simulator

VICTORY Process – Oxidation/Stress

Numerically very demanding process:
Involves different processes
Oxygen diffusion
Chemical reaction: Si + O2 = SiO2
Material deformation
Processes are interlinked
Diffusion and reaction depend on oxide flow (via stresses)
Oxide flow depends on diffusion
Mechanical properties of oxide change significantly with temperature
Complex behavior near corners
31 -

SLIDE 32

Full Physical 3D Semiconductor Simulator

VICTORY Process – Oxidation/Stress (con‘t)

Oxidation step is represented as a succession of 4 sub-steps :
1. sub-step - Diffusion of oxygen through the oxide
Solving diffusion equation
2. sub-step - Reaction at SiO2/Si interface
Calculating volume expansion velocity
Calculating reactive interface velocity
3. sub-step - Deformation velocity field according to the mechanical

behavior of each material

Viscous
Solving creep flow problem
4. sub-step - Deformation of the entire structure according to the

velocity fields

Interface propagation
Solving moving boundary problem
32 -

SLIDE 33

Full Physical 3D Semiconductor Simulator

VICTORY Process – Oxidation/Stress (con‘t)

1. sub-step - Oxygen transport models :
Obtains the oxygen concentration in oxide

Volume equation : Boundary conditions (oxygen flux across interfaces) : (supply) (reaction)

33 -

( )

xide

ambient

xide

ambient

C C h n n n C D J

/ /

− ⋅ ⋅ = ⋅ ∂ ∂ − =   

xide

silicon

xide

silicon

C k n n n C D J

/ /

⋅ ⋅ = ⋅ ∂ ∂ ⋅ − =   

= ⋅ ∇

Volume

J 

C D Jvolume ∇ ⋅ − =  

SLIDE 34

Full Physical 3D Semiconductor Simulator

VICTORY Process – Oxidation/Stress (con‘t)

1- sub-step - Oxygen transport models :
Linear transport model :
34 -

dependent rature

nly tempe

are .... , , h k D         ⋅ − ⋅ = T k E D T D

B D

exp ) (         ⋅ − ⋅ = T k E k T k

B k

exp ) (         ⋅ − ⋅ = T k E h T h

B h

exp ) (

.... Oxygen diffusivity .... Reaction coefficient .... Transport coefficient

SLIDE 35

Full Physical 3D Semiconductor Simulator

VICTORY Process – Oxidation/Stress (con‘t)

1. sub-step - Oxygen transport models :
Stress-dependent transport model :
35 -

ture

n tempera

and pressure c hydrostati local

n the

depends .... D         ⋅ − ⋅ = T k E D T D

B D

exp ) (      ≤ >         ⋅ ⋅ − ⋅ = ) ( .... ) ( ) ( .... ) ( exp ) ( ) , ( x p T D x p T k V x P T D T x D

B d

  

SLIDE 36

Full Physical 3D Semiconductor Simulator

VICTORY Process – Oxidation/Stress (con‘t)

1- sub-step - Oxygen transport models :
Stress-dependent reaction model :
36 -

ture

n tempera

and stress interface

n the

depends .... k         ⋅ − ⋅ = T k E k T k

B k

exp ) (      ≤ >         ⋅ ⋅ − ⋅ = ) ( .... ) ( ) ( .... ) ( exp ) ( ) , ( x T k x T k V x T k T x k

n n B k n

σ σ σ   

j ij i n

n n ⋅ ⋅ = σ σ

.... Force in the direction of the normal

SLIDE 37

Full Physical 3D Semiconductor Simulator

VICTORY Process – Oxidation/Stress (con‘t)

3. sub-step - Deformation models :
Obtains the deformation velocity of all materials above the reactive

surface

Solution variables :
37 -

equation ive Constituit .....         ∂ ∂ + ∂ ∂ ⋅ = ∂ ∂ +

i j j i ij ij

x V x V t G µ σ µ σ tensor Stress .....

ij

σ field n velocity Deformatio .....

i

V equation flow Creap .....

j ij i

x x P ∂ ∂ = ∂ ∂ σ pressure c Hydrostati ..... P

SLIDE 38

Full Physical 3D Semiconductor Simulator

VICTORY Process – Oxidation/Stress (con‘t)

3.sub-step - Deformation models :
Linear incompressible viscous flow model :
Viscoelastic term is neglected (assumes infinite shear modulus G)
38 -

        ∂ ∂ + ∂ ∂ ⋅ = ∂ ∂ +

i j j i ij ij

x V x V t G µ σ µ σ condition ibiliy incompress ..... = ⋅ ∇ V   dependent rature

nly tempe

is ..... µ Viscosity ..... exp ) (         ⋅ − ⋅ = T k E T

B µ

µ µ

SLIDE 39

Full Physical 3D Semiconductor Simulator

VICTORY Process – Oxidation/Stress (con‘t)

Oxygen transport, material flow and moving boundary problem are

solved on the same Cartesian mesh

High stability
Special boundary nodes are added to implement proper boundary

conditions (special method developed in-house)

39 -

Formulate standard equations for regular nodes Introduce cross-points with the interface Formulate standard equations for irregular nodes (considering cross-points) Formulate cross-point equation (tangential equation)

n u t t t t u t t u        ∂ ∂ × × = ⋅ ∂ ∂ − ⋅ ∂ ∂ ) (

β α α β β α

SLIDE 40

Full Physical 3D Semiconductor Simulator

VICTORY Process – Oxidation/Stress (con‘t)

 Multiple operational modes  Analytical mode

 Grows a layer of oxide of the requested thickness

 Empirical mode

 Calculates the oxide thickness on the basis of the Massoud model  Can be applied to oxide layers thinner than the mesh size

 Full physical mode

 Full solution of the transport and the flow equations in oxide  Requires appropriate high mesh resolution in oxide

 Hybrid mode

 Low mesh resolution can be applied in planar regions

 Empirical solution is calculated in these planar regions

 Coupled with full physical solution in curved regions  Significantly reduces the number of required mesh nodes  Automatic switching between modes

40 -

SLIDE 41

Full Physical 3D Semiconductor Simulator

VICTORY Process – Oxidation/Stress (con‘t)

41 -

LOCOS: Long simulations are stable without meshing issues

SLIDE 42

Full Physical 3D Semiconductor Simulator

VICTORY Process – Oxidation/Stress (con‘t)

42 -

Simulation is stable for complex structures (multi-layer concept) STI liner oxidation Buffered field oxidation

SLIDE 43

Full Physical 3D Semiconductor Simulator

VICTORY Process – Oxidation/Stress (con‘t)

43 -

Stress-dependent oxygen transport is taken into account

stress independent stress dependent

SLIDE 44

Full Physical 3D Semiconductor Simulator

VICTORY Process – Interface to VICTORY Device

A special device meshing engine is

used to export structures to VICTORY Device

Supports different base meshes

 Process - geometry mesh  Process - volume data mesh

Provides comprehensive automatic and

manual mesh adaptation features

 Refinement boxes  Refinement lines  Refinement near interfaces  Refinement according to doping

44 -

SLIDE 45

Full Physical 3D Semiconductor Simulator

VICTORY Process – Interface to VICTORY Device (con't)

A special device meshing engine is

used to export structures to VICTORY Device

Supports different base meshes

 Process - geometry mesh  Process - volume data mesh

Provides comprehensive automatic and

manual mesh adaptation features

 Refinement boxes  Refinement lines  Refinement near interfaces  Refinement according to doping

45 -

SLIDE 46

Full Physical 3D Semiconductor Simulator

VICTORY Process – Interface to VICTORY Device (con't)

A special device meshing engine is

used to export structures to VICTORY Device

Supports different base meshes

 Process - geometry mesh  Process - volume data mesh

Provides comprehensive automatic and

manual mesh adaptation features

 Refinement boxes  Refinement lines  Refinement near interfaces  Refinement according to doping

46 -

SLIDE 47

Full Physical 3D Semiconductor Simulator

VICTORY Process – Interface to VICTORY Device (con't)

Structure mirroring capability
Process simulation can make use of any geometrical symmetry
Allows for faster process simulation of symmetrical devices
47 -

SLIDE 48

Full Physical 3D Semiconductor Simulator

Applications: Re-Oxidation - STRESS

48 -

SLIDE 49

Full Physical 3D Semiconductor Simulator

Applications: Re-oxidation – Regions

49 -

SLIDE 50

Full Physical 3D Semiconductor Simulator

Applications: Re-oxidation – Stresses in Oxide

50 -

SLIDE 51

Full Physical 3D Semiconductor Simulator

Applications: Planar MOS Technology

51 -

Boron profile Phosphorus profile Arsenic profile

SLIDE 52

Full Physical 3D Semiconductor Simulator

Applications: FinFET

52 -

SLIDE 53

Full Physical 3D Semiconductor Simulator

Applications: Tri-gate FET with Contact Area

53 -

Donor profile Acceptor profile

SLIDE 54

Full Physical 3D Semiconductor Simulator

Applications: MEMS Cantilever Structure

54 -

SLIDE 55

Full Physical 3D Semiconductor Simulator

Current Development

Stress-dependent viscous oxidation
Performance improvement of the oxidation module
Extended default set of diffusion models
Extended default set of etching and deposition models
Improvement of the device meshing export
55 -

SLIDE 56

Full Physical 3D Semiconductor Simulator

Conclusion

VICTORY Process is the next generation tool for the investigation
f 3D process and device effects
Suitable for many semiconductor technologies such as
Planar MOS, FinFET, Power devices
Sensor and LED’s,
MEMS and
Complex quantum structures
Also suitable for non-semiconductor applications such as
Hard coating and
Mass storage applications
Very high stability due to
Multi-layer concept and
Robust and accurate methods for moving interfaces
Flexible due to open interface for modeling
56 -