OLED Simulation using DSMC method · • DSMC-Neutrals is composed of WF-Geom, WF-Geom2D, DSMC-Neutrals-GUI including solver, WF-View. Furthermore, the output file is available for

OLED Simulation using DSMC method

2016/5/18

Copyright © 2016 Wave Front Co., Ltd. All Rights Reserved. 2

Name : Wave Front Co., Ltd.

Incorporation : March 1990

Head Office : Yokohama in Japan

Business Description

(1) CFD(Computational Fluid Dynamics)&Plasma Software, Sales &

Consulting Services

(2) CMMS(Computerized Maintenance Management System) Software,

Sales, Consulting and Implementation services

Number of Employee : 18

Software Products sold in Foreign Countries

(1) Particle-PLUS( Plasma) & DSMC-Neutrals( Rarefied Gas Flow)

(2) PM-Optimizer( CMMS) & FLIPS( Scheduler)

Our Company


Semiconductor

(1)Thin Film Fabrication: Plasma Enhanced Chemical Vapor Deposition,

Magnetron Sputtering, Ion Plating, Vacuum Evaporation

(2)Microfabrication: Plasma Etching

Surface Processing

Coating on Glass and Other Materials, Hydrophilic Surface Treatment

Plasma Carburizing, Plasma Cleaning

Vacuum Technology

Vacuum Pump

Electrical Appliance

Lighting, Plasma TV, Air Cleaner (Conditioner), Organic Electroluminescence

Energy

Nuclear Fusion Plasma, Solar Cell Fabrication

Internal Combustion Engine

Spark Ignition

Ecology

Exhaust Gas Disposal

Application field of our software


1. first Rarefied Gas Flow Simulation

based on Direct Simulation Monte Carlo (DSMC) method

2. DSMC-Neutrals is composed of pre, post, and solver.

pre ： WF-Geom and WF-Geom2D

post ： WF-View

solver ： DSMC-Neutrals-GUI

3. No Divergence for All Model including Bad Quality Mesh

User can obtain solution always.

4. Simulation of Thin Film Growth due to Chemical Reaction like CVD

• Gas Phase Chemical Reaction

• Surface Chemical Reaction

5. Detail Geometry Modeling Capability by Unstructured Mesh

Feature


Mesh Parallel processor

tetrahedron

hexahedron

pyramid

wedge

"DSMC-Neutrals" is commercial software that computes the behavior of the

rarefied gas using the DSMC method. It consists of pre-processor with

powerful mesher, post-processor and solver. DSMC-Neutrals is available for

Hex dominant mesh. Therefore, it is possible to simulate the complicated

geometry model. Chemical reactions in the volume and on the surface are

available. Parallel computing using MPI processor is also possible.

Feature


Making Grid

Initial Particle Setup

Particles Moving

Reflection on Boundary

Collision between Particles

One particle collides another

in the same cell.

Cell size

D x < l : mean free path

Time interval

D t < t : collision time

Represent Particle of Neutral Species

DSMC method


Elastic Collision

Inelastic Collision (for Molecule)

• Vibrational Excitation

• Rotational Excitation

• Dissociation

• Recombination

• Molecular(Atom) Exchange

Chemical Reaction ( Total Collision Energy model )

Collision


T

ETk a

f exp

1132605.11098.4 9

aE

74.027317.40.282N

TDMspecies refref

Reaction : N2 + M -> N + N + M

*1 Bird G. A., “Molecular Gas Dynamics and the Direct Simulation of Gas Flows”, 1st edition, Clarendon Press,

Oxford, New York, (1994)

*2 Yasunori Tanak, J. Phys. D: Appl.Phys. 37 (2004) 1190-1205

T

etemperaturreferenceT

diameterreferenceD

massatomicM

ref

ref

:

:

:

:

VHS parameter*1

Arrhenius parameter*2

Gas phase chemical reaction

DSMC-Neutrals adopts TCE (Total Collision Energy) model. TCE model reproduces

gas phase chemical reaction of Arrhenius type. TCE mode considers dissociation,

recombination, and atom exchange. The reaction includes forward and reverse

reactions, reverse reaction is obtained by partition function. The collision energy

conserves, where activation energy and inner energy is considered.


212018201717

212018181717

1065.21073.61048.11018.61031.11014.2

1012.21000.61043.11011.61036.11020.2

][7500][10000][15000][20000][25000][30000

NeutralsDSMC

Arrhenius

KKKKKK

Initial density : N2 1x1020 [ / m3 ]

Validation of TCE model

TCE model reproduces well rate

coefficient which is input data of

Arrhenius format.


CAD WF-Geom DSMC-Neutrals-GUI WF-View

EnSight Desktop

1. Geometry 2. Mesh 3. Solver setting 4. Visualization

• DSMC-Neutrals is composed of WF-Geom, WF-Geom2D, DSMC-Neutrals-GUI

including solver, WF-View. Furthermore, the output file is available for EnSight

Goldd format.

Create geometry ⇒ Make mesh⇒ Set physical parameters ⇒ Visualize results

Pre processor Solver setting Post processor

Package of DSMC-Neutrals


1. Typical OVPD (Organic Vapor Phase Deposition)

2. OVPD with Carrier Gas

In this simulation, the separate domain model is

presented. The results of low pressure area

tends to be very rough. The high precision

results are obtained using the separate domain

model.

In this simulation, it is introduced how to determine evaporation

rate from enthalpy. Furthermore, DSMC-Neutrals’ results is

compared with experimental data.

Application Example 1


3. MBE (Molecular Beam Epitaxy) : nozzle shape effect

with cap

without cap

Experimental Data DSMC-Neutrals

Application Example 2

4. OVPD with N2 adsorption on wall : surface reaction

Step 60000 Step 80000 Step100000


The number of simulated particles is almost proportional to pressure. Therefore, the

number of simulated particles tends to be few and rough results on low pressure area when

the pressure difference between the shower head part and the chamber part is very large.

In OLED simulation, pressure difference between the chamber (1x10-6 Pa) and the source

cell (10 – 100 Pa) exist. Therefore, the numerical error in the chamber is very large. The

results including deposition rate profile are too rough to estimate the results.

DSMC-Neutrals can avoid above problem by using the Separate model. The separate

model computes automatically the low pressure area with many simulated particles. The high

precision results in the chamber including deposition rate are obtained.

1. Typical OVPD


Low pressure area

(chamber)

High pressure area

(source cell)

Separate model

• first computation

compute for whole computation domain

• second computation

compute separate domain (low pressure area)

The high precision results

are obtained by using

separate domain model.

conventional modelwith separate domain model

1. Separate domain model


500

[mm]

500

T ：300 [ K ]

full adsorption

Glass

T ：300 [ K ]

full adsorption

source cell

species：Alq3

T ：700 [ K ]

Evaporation rate : 1.0×1022 /m2s

T ：700 [ K ]

1. Typical OVPD model


adsorption profile on substrate

[ #/m2 s ]density profile [ #/m3 ]

The accurate result is obtained short time by using separate

domain model.

1. Results (Typical OVPD)


4

1.5

45

20 10 91

1.25

4.25

[ cm ]

Species

• carrier gas : N2

• organic material : Alq3

Temp: 269 ℃ or 276 ℃

Substrate300 K

LP-OVPD reactor*1

*1 J. Appl. Phys., Vol. 89, No.2, 1470 (2001)

*1 J. Appl. Phys., Vol. 113, 154503.1-6 (2013)

~3 Pa

Wall 550 K

2. OVPD with carrier gas


Evaporation rates are estimated with enthalpy in this simulation model. An evaporation rate

with carrier gas flow rate and a source cell temperature are 30 sccm and 276 degree,

respectively. The procedure of estimating evaporation rate is as follows;

1. Deposition rate on the substrate is computed with expected evaporation rate.

2. The evaporation rate is modified by comparing the DSMC-Neutrals’ deposition rate with

experimental that.

3. Repeat above steps till the DSMC-Neutrals’ deposition rate agrees with experimental

deposition rate.

The evaporation rate with other source cell temperature is obtained from following equation.

Equilibrium vapor pressure and proportional to evaporation rate. In above

equation, the condition on the evaporation surface assumes to be equilibrium

that gas flow velocity is zero.

Enthalpy obtained experimental data.

Kinetic factor for evaporation and regarded as fitting parameter in this model.

2. Evaporation rate


F : 20 [ sccm ]

T : 269 [ ℃ ]

F : 20 [ sccm ]

T : 276 [ ℃ ]

F : 40 [ sccm ]

T : 269 [ ℃ ]

F : 40 [ sccm ]

T : 276 [ ℃ ]

Alq3 number density [m-3]

DSMC-Neutrals’ results reproduce well experimental data

reverse of deposition rate

vs

root of carrier gas flow rate

F : carrier gas flow rate

T : source cell temperature

2. Results (OVPD with carrier gas)


The nozzle with cap makes film thickness profile uniform. By comparing the models

between with a cap and without it, the geometry dependence is investigated. Furthermore,

DSMC-Neutrals' results are compared to experimental results which are presented on Web

site and in patent information.

cap

without cap with cap

＊1.

http://www.ekouhou.net/%E5%88%86%E5%AD%90%E7%B7%9A%E6%BA%90%E3%82%BB%E3%83

%AB/disp-A,2009-40615.html

Computation time：1.3 hour

No. of mesh：20000, No. of sample particle：80000

OS：Windows Vista, CPU：intel 2.66GHz, Memory：8GByte

3. Nozzle shape effect with cap


substrate

T : 300[K]

adsorp. : 1.0

chamber wall

T : 300[K]

adsorp. : 1.0

symmetry

[mm]

evaporation surface

species : Alq3

T : 700[K]

adsorp. : 1.0

without cap

with cap

40

40

40

3. Nozzle shape effect with cap


with capwithout cap

DSMC-Neutrals experimental data

3. Results (Nozzle shape effect with cap)


4. OVPD with N2 adsorption

It is reported that the nitrogen adsorption on the wall affects the film formation. The

coverage effect of nitrogen molecule on chamber wall is computed, where nitrogen

molecules are adsorbed on a surface layer. DSMC-Neutrals can consider the surface

site which shows the coverage rate for each species. Left figure shows the

dependence of adsorption probability of nitrogen on the wall upon the open site. The

open site means dangling bond.


図 1-1 図 1-2

inflow

flow rate ：Alq3 : 1.0×1021 [/m2s]N2 : 200 [sccm]

flow velocity ： 10 [m/s]

diameter of ： 6 [mm]

substrate

T : 300[K]

substrate side wall

T : 300[K]

chamber wall

T : 300[K]

outlet

T : 300[K]

P : 0.01[Pa]500

[mm]

500

500

450

450

100

4. OVPD with N2 adsorption


*1 A User’s Guide to Vacuum Technology (3rd ed.) (2003)

*2 Open site rate is 1 – N2 adsorption rate.

Open*2 adsorption

rate

0.57 0.0018

0.67 0.0034

0.75 0.0081

0.87 0.026

0.962 0.13

0.982 0.24

0.989 0.3

OPEN(S) + N2 -> N2(S)

surface reaction equation

（Ｓ） show surface site.

4. N2 adsorption on Ti model


substrate

Alq3 adsorption profle

Step 5000 Step 15000 Step 25000

Step 50000 Step 100000

4. Results (OVPD with N2 adsorption)


wall

N2 coverage profile

Step 0 Step 20000 Step 40000

Step 60000 Step 80000 Step100000

4. N2 Coverage profile on wall


• Four simulation models are presented.

• The high precision results are obtained by using the separate

domain model.

• The evaporation rate is estimated using enthalpy of organic material.

• DSMC-Neutrals’ results agree with experimental data in OVPD with

carrier gas.

• In nozzle geometry dependence, the DSMC-Neutrals’ results agree

with experimental data.

Summary

Dust Simulation


Behavior of dust in the chamber

Dust size is micro meter

Thermophoresis

high temperature

low temperature

Dust (micro-particle) simulation

The dust sometimes has a bad influence on the wafer. The behavior of dusts

is computed by DSMC-Neutrals. It is difficult to simulate behavior of dust

because cross section of the dust is much large and time step becomes too

small for conventional DSMC method.

DSMC-Neutrals can simulate dust behavior with collision weight factor, where

the collision weight factor is the number of molecules in a molecule-dust collision

event or a time step.

The DSMC method with the collision weight factor considers drag and

thermophoretic forces.


1. Molecule–particle collision model

collision factor


1. Validation of collision model

As the validation of molecule – particle collision model, thermophoretic velocity

of micro size particle in an Ar gas is computed. Furthermore, the results compares

with Waldmann’s thermophoretic velocity which is theoretical solution.

The validation model is one dimension model. The time step is less than one

millionth of the one for conventional DSMC method. The computation time

becomes to be 1/50,000 with considering the number of sample particles per cell.


particle speed dependence on momentum

ratio, where ratio of momentum of particle

and momentum transfer in a collision event.

particle speed dependent on accommodation

factor

1. Results (particle in Ar gas)

Below figures show particle speed versus distance from particle generating point.

The particle initial speed is very small. The results is different from Waldmann’s

results because of slip wall around the wall. The DSMC-Neutrals’ thermophoretic

speed agrees with Waldmann’s results.


2. Thermophoresis phenomena

speed 8.5 m/s

The behavior of 220 nm PSL particle in Air is computed. DSMC-Neutrals’ results

are compared with the experimental data of Kim et. al.. The two models with

temperature gradient 10 K/mm and without are simulated. This phenomena is

used as thermophoretic protection.

J.H.Kim and H.Fissan and C.Asbach and Se-Jin Yook and Y.H.Pui and K.L.Orvek, J. Vac. Sci. Technol.

B, Vol.24. No.3, (2006), pp.1178


2. Results (density and velocity)

Particle density Particle velocity

The some particles can not arrive at upper wall when temperature gradient exists

because of thermophoretic force. This result is the same as the experimental

results of Kim et. al..


Inlet : Ar

diameter ： 5 mm

gas flow rate ： 280 slm1 m

1 mParticles (dusts) are located

around bottom of the chamber

3. Dust simulation

• The micro meter particles are set on bottom of the chamber. The initial dust

density is uniform.

• Ar gas inflows from the nozzle whose diameter is 5mm. Gas flow rate is 280 [slm].

• As the initial condition, Ar gas does not exist in the chamber.

• Wall temperature is 300 K.

• Particle - Particle, Ar atom - Particle, Ar atom - Ar atom collisions are considered.

• Chamber size is 1x1x1 [m3].


• Density gradient is very large around inlet. Therefore, shock wave appears in

the computation results.

• Particles are blown up by Ar gas flow.

3. Results (Dust simulation)


• It is important to investigate the particle behavior in the chamber.

• Conventional DSMC method can not be applied to the dust

simulation because time step size becomes much smaller.

• The computation speed with molecule – particle model becomes

50,000 times than conventional DSMC method.

• DSMC-Neutrals can simulate the micro size particle behavior.

Summary

Documents

OLED Simulation using DSMC method · • DSMC-Neutrals is composed of WF-Geom, WF-Geom2D, DSMC-Neutrals-GUI including solver, WF-View. Furthermore, the output file is available for