Veeco Instruments - TurboDisc Operations Reactor Design ......Superposition technique (L.Kadinski et al., Jounal of Crystal Growth 261, 2004., 175-181) Objective: Find the optimal

TurboDisc Operations

Veeco Instruments - TurboDisc Operations

Reactor Design Optimization Based on Reactor Design Optimization Based on 3D CFD Modeling of Nitrides Deposition in 3D CFD Modeling of Nitrides Deposition in MOCVD Vertical Rotating Disc ReactorsMOCVD Vertical Rotating Disc Reactors

B. Mitrovic, A. Gurary, and L. KadinskiVeeco TurboDisc Operations

394 Elizabeth Avenue, Somerset, NJ 08873, USA

22 June 2005Barga, Italy


IntroductionMOCVD (Metal-Organic Chemical Vapor Deposition) Vertical Rotating Disc Reactors (RDR) are widely used for the large-scale production of GaN-based semiconductor devices such as blue and green light-emitting diodes (LED), ultra violet LED, solid-state lasers, heterojunction bipolar transistors.

In RDRs rotation of the wafer carrier results in an effective averaging of the deposition rate distribution and this is a key mechanism providing growth of epitaxial layers with highly uniform properties.

Inlet

Outlet Outlet

Heated Susceptor

Rotation


Introduction

Alkyls (TMG, TMIn..)

Hydrides (NH3)

Shroud flow (N2,H2)

The necessity to utilize a number of different precursors for nitride deposition, many of which actively react with each other in the gas phase, presents significant challenges for the reactor development, particularly the reactant injection elements.

Proper design of such components is practically impossible without detailed flow modeling based on CFD that addresses optimization of both reactor components and process parameters and is based on an ability to predict GaN/InGaN growth rate and uniformity under different process conditions.


Geometry and Process Parameters in theVeeco MOCVD TurboDisc Reactors

Inlet Gases (Reactants)

Alkyls (Ga(CH3)3, In(CH3)3..)

Hydrides (NH3)

Shroud flow (N2,H2)

Outlet Outlet

Heated Substrate

ω

Water-cooled walls

ts = 500 – 1100oC

ω = 0 – 1500 rpm

Total Flow Rate:Q = 5 – 250 slm

Reactor Pressure:P = 50 – 1000 Torr

Diameterd = 75 – 450 mm

Spindle

tw = 25 – 75oC

Solid Products:GaN, InGaN, AlGaN …



Surface chemical processes during GaN growth

Substrate

NH3MMGa GaN2 H2 CH4

Rate-limiting processes• Low temperatures (400-600 ºC): group-III adsorption site

blocking by methyl radicals• Intermediate temperatures (600-1050 ºC): transport of

group-III species to the growth surface• High temperatures (above 1050 ºC): gallium desorption


Flow Dynamics in Rotating Disc MOCVD Reactors

Rotating Disc MOCVD Reactors involve complex flow dynamics driven by interactions between buoyancy forces, wafer carrier rotation and forced convection.

There is considerable interest in understanding and controlling reactor gas flow dynamics as the flow regimes in the process reactor are largely responsible for compositional and thickness non-uniformity.

Gas flows smoothly over the substrate without any recirculations above the wafer carrier

Vortex forms near the reactor wall close to the upper disc surface

Thermal recirculation – density difference between the disk and the incoming gas stream overcomes the stabilizing influence of the viscous forces


Flow Patterns in the Rotating Disc Reactors

Computer Generated Flow Patterns in Rotating Disc System

Computer Generated Flow Patterns Computer Generated Flow Patterns in Rotating Disc Systemin Rotating Disc System

Smoke Flow Patternsin Rotating Disc System

Smoke Flow PatternsSmoke Flow Patternsin Rotating Disc Systemin Rotating Disc System

Data Courtesy of Sandia National LaboratoriesData Courtesy of Data Courtesy of SandiaSandia National LaboratoriesNational Laboratories


CFD Modeling at Veeco TurboDisc Operations

Reactor and process development guided by CFD

Feasibility, evaluation of design concepts

Equipment design optimization

Process optimization

Customer support

Commercial CFD solver FLUENT by Fluent Inc.flow dynamics, heat and mass transport of precursors and reaction products, chemical reactions.


Solid Works Example – D125GaN Reactor

model mesh

Solid Works (Computer Aided Design Software) models are used for the grid generation by direct import to CFD code


Screen provides flow uniformity

All geometrical details are resolved

Alkyls Hydrides

Grid ~ 750,000 control volumes


CFD – DOE Optimization of the Injector Plate

A

1/8” 1/8”

B

1.5”

A

1/8”1/8” 1/8”

B

1.5”

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035

Radius (m)

Gro

wth

Rat

e (a

.u.)

InnerOuterOptimized

Wafer

Inner Injector Outer Injector

Simplified model – streamlines and growth rate

Superposition technique (L.Kadinski et al., Jounal of Crystal Growth 261, 2004., 175-181)

Objective:Objective:

Find the optimal sizes and positions of the alkyl zones (parameters A and B) that provide the best growth rate deposition uniformity on the wafer in a wide range of process conditions

Inner alkyl injector

Outer alkyl injector


CFD – DOE Optimization of the Injector Plate

0.12

5

0.16

25 0.2

0.23

75

0.27

5

0.31

25

0.35

0.38

75

0.42

5

0.46

25

0.5

0

0.075

0.15

0.225

0.30.375

0

1

2

3

4

5

6

Unifo

rmity

(%)

A

B

Surface Plot of A vs B Constants:

5-6

4-5

3-4

2-3

1-2

0-1

0

1

2

3

4

5

6

0 0.1 0.2 0.3 0.4 0.5 0.6

A (in)

Uni

form

ity (%

)

B = 0 inchB = 0.25 inchB = 0.375 inch

An optimal geometrical position of the alkyls zones is found for different process conditions which correspond to GaNbased LED process development


Detailed numerical grid of P75 VEECO Reactor and of Flow Flange

Based on the optimized injector plate a new modification of P75 TurboDisc Reactor has been designed

Grid: 500,000 control volumes


Flow Visualization in the P75 Reactor


Flow Optimization in the ReactorVelocity matched conditions

between alkyl and hydride zones Velocity profiles

Temperature profiles

Momentum matched conditions between alkyl and hydride zones

Recirculation areas


Flow Optimization in the Reactor

Impulse matched conditions provide flow with no recirculation and the best growth rate uniformity in P75 reactor

0

0.5

1

1.5

2

2.5

3

0 0.1 0.2 0.3 0.4 0.5 0.6

Alkyl velocity (m/s)

Uni

form

ity (%

)InGaN process conditionsGaN process conditions

Velocity matched

Reynolds number matched

Impulse matched


Experimental Verification - Effect of Pressure

0

1

2

3

4

5

6

7

8

9

0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08

Distance (m)

Gro

wth

Rat

e ( µ

m/h

)

Inner, ExperimentInner, ModelOuter, ExperimentOuter, Model

waferInner Injector Outer InjectorOuter Injector

200 torr

500 torr

For superposition, a set of independent runs (a number of runs is equal to the number of alkyls zones) are conducted wherein each run, the alkyls flow into the reactor only in a single zone. All of the other zones contain push gas. Each run gives an individual growth rate/composition “response” for the zones

0

2

4

6

8

10

12

0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08

Distance (m)

Gro

wth

Rat

e ( µ

m/h

)Inner, ExperimentInner, ModelOuter, ExperimentOuter, Model

waferInner Injector Outer InjectorOuter Injector

Excellent qualitative and solid quantitative agreement between the modeling results and experiments is observed


CFD Model for high capacity multi-wafer reactors

Temperature ProfilesGrid Streamlines

Temperature profile

Velocity profile


Comparison with the Experimental Data

Good qualitative and quantitative agreement between the modeling results and experiments is observed

0.0

5.0

10.0

15.0

20.0

25.0

30.0

35.0

40.0

45.0

0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16

Radius (m)

Thic

knes

sInner-1_EXPInner-2_EXPMiddle_EXPOuter_EXPInner-1_ModelInner-2_ModelMiddle_ModelOuter_Model

GaNGaN Growth ConditionsGrowth ConditionsP = 200 P = 200 TorrTorr

ωω = 1500 rpm= 1500 rpm

ttss = 1050 = 1050 ooC

QQN2N2 = 15 = 15 slmslm

QQH2H2 = 80 = 80 slmslm

QQNH3NH3 = 40 = 40 slmC slm


Model Verification

Developed CFD model helped us to identify the potential problems for the flow flange design of the reactor

Deposition on the cold plate (black area on the photograph) corresponds to the area where model predicts the recirculation pattern (white area on the modeling figure)


Flow Flange Optimization Based on CFD Modeling

Old Flow Flange Redesigned Flow Flange


Flow Stability Mapping Approach in “P –ω“ (Pressure –Rotation Rate) Diagram

0

100

200

300

400

500

600

0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 1600

Rotation Rate (ω ), rpm

Pres

sure

(P),T

orr

Stable Flow

Unstable Flow

ReRe

constGr=

ω

Re

constGrm =

“Pressure-Rotation” diagram flow regime mapping approach is based on extensive 2D/3D numerical modeling for the entire domain of possible process parameters.

50

1070

slm 140Q tot

Ct

Cto

w

os

=

=

=

ReRe constn =ω


“P-Q-ω “ (Pressure-Flow-Rotation) Flow Stability Diagram

50 ; 1070 ; 2 ; 4 3

2

2

2 CtCtQQ

QQ o

wo

sNH

H

N

H ====GaN process conditions:

1

100

200

300

400

500

600

700

800

900

1000

1100

1200

1300

1400

150040 slm

80 slm

125 slm

170 slm215 slm

250 slm 250 slm

250 slm230 slm215 slm200 slm185 slm170 slm155 slm140 slm125 slm110 slm95 slm80 slm65 slm

0.00

100.00

200.00

300.00

400.00

500.00

600.00

P (torr)

ω (rpm)

Q (slm)

40 slm50 slm65 slm80 slm95 slm110 slm125 slm140 slm155 slm170 slm185 slm200 slm215 slm230 slm250 slm


Flow Stability Map and Relative Growth Rate

GaN process conditions:

0

50

100

150

200

250

300

350

400

450

500

0 200 400 600 800 1000 1200 1400 1600

Rotation Rate (ω), rpm

Pres

sure

(P),

Torr

0

0.2

0.4

0.6

0.8

1

1.2

Rel

ativ

e G

row

th R

ate

Flow Stability BoundaryRelative Growth Rate

50 ; 1070 ; 2 ; 4 ; 140Q 3

2

2

2tot CtCt

QQ

QQ

slm ow

os

NH

H

N

H =====

~Re~1 ~ PRateGrowth ⋅ωδ ω

For the case of an infinitely large disc y proportional to the boundary layer th

under mass transfer limited growth conditions the growth rate is inverselickness and primarily depends on rotation rate and operating pressure.

Rapid decrease of the growth rate

High growth rate conditions


Criteria for the onset of buoyancy-induced flow

0

100

200

300

400

500

600

700

800

0 200 400 600 800 1000 1200 1400 1600


Pres

sure

(P),

Torr

ts = 500 oCts = 750 oCts = 1100 oC

5.13ReRe

=ω

Gr

50 ; 2 ; 4 ; 140Q 3

2

2

2tot Ctt

QQ

QQ

slm oow

NH

H

N

H =====

By decreasing the temperature gradient between the wafer carrier and the inlet, we also

reduce the tendency for natural convection (lower Gr), and hence, higher operating

pressures can be utilized for the same rotation rate in the buoyancy induced region.

Higher inlet temperature has two effects on flow stability: (1) decreases the temperature gradient

between the wafer carrier and the inlet which suppresses buoyancy-induced recirculation; (2)

increases the through-flow velocity, which suppresses the rotation-induced recirculation.

1070 ; 2 ; 4 ; 140Q 3

2

2

2tot Ct

QQ

QQ

slm os

NH

H

N

H ====

0

100

200

300

400

500

600

700

800

900

1000

0 200 400 600 800 1000 1200 1400 1600


Pres

sure

(P),

Torr

tw = 25 oCtw = 50 oCtw = 100 oCtw = 200 oC

5.13ReRe

=ω

Gr

( ) 2

32

w

ws

tttHg

Grµ

ρ −⋅=

oo tvtt ∝= ; w

( ) 2

32

w

ws

tttHg

Grµ

ρ −⋅=


Conclusions

Based on modeling results and DOE optimization, an optimal geometrical position of the alkyls zones is found for different reactor sizes, which provides the best growth rate deposition uniformity on the wafer within a wide range of process conditions.A new modifications of TurboDisc reactors has been designed based on the optimized injector plate.Detailed 3D reactor modeling from direct CAD geometry import into CFD is used to find optimal process parameters for III-Nitrides materials growth. Excellent qualitative and solid quantitative agreement between the modeling results and experiments is observed.Modeling drastically reduced the process development time to a few runs and resulted in significant improvement of growth uniformity ( δ < 1%) and alkyl efficiency.


Conclusions

Quantitative flow stability maps have been developed based on extensive 2D and 3D flow modeling for the entire domain of possible process parameters. It has been shown that all typical flow regime regions that can be encountered in a rotating disc reactor can be presented in a single P-ω diagram, which also transparently captures the effects of all other process parameters.New dimensionless criteria have been proposed (based on Grashof, Reynolds and rotational Reynolds numbers) for defining a boundary between stable and unstable flow regimes. The obtained stability criteria have both fundamental and practical significance, and allow one to predict the process window that is free of both thermal and rotation induced recirculation, without performing additional numerical modeling or costly experiments.

Documents

Veeco Instruments - TurboDisc Operations Reactor Design ......Superposition technique (L.Kadinski et al., Jounal of Crystal Growth 261, 2004., 175-181) Objective: Find the optimal