Spatial Atomic Layer Deposition: ASpatial Atomic Layer Deposition: A Path to High-Quality Films on

C ti S b t tContinuous Substrates

David H. Levy, Roger S. Kerr, Shelby F. Nelson, Lee W. Tutt, and Mitchell Burberry

Eastman Kodak CompanyRochester, NY

Agenda

Atomic Layer Deposition (ALD) as a processSpatial ALDSpatial ALD• Approach• Performance

Devices and patterning using Spatial ALD• Working demonstrations of film quality• Effective film patterning with ALD

Atomic Layer Depositions (ALD)ALD: process where a substrate is exposed to reactive gases one by one

Film growth occurs

s re

peat

ed

Film growth occurs layer by layerPrecursor (I)

Cyc

le is

High quality

ConformalPrecursor (II) Conformal

Low temperature

© Eastman Kodak Company 3

Atomic Layer Deposition Uses

Barrier layers• Very conformal and dense coatingVery conformal and dense coating• Prevent moisture and oxygen transmission• Thin layers (100–200 Å) are effective

Thin, high-performance dielectrics• New generation silicon chips: 25 A layers with low electrical

leakageg

Many other applications• Coating of high aspect ratio structures• Transparent conductors• Oxide and other binary/ternary semiconductors

Spatial Atomic Layer Deposition (S-ALD)

ChamberALD

SubstrateExposureALD

Inert (II)(I)Time

Exposure

SpatialALD S-ALD Head

Q

SubstrateExposure(Point Q)

Spatial Process

ALD S ALD HeadTime

p• Steady-state gas flows• Can be “open air”• Suitable for large or continuous substrates

Isolating the Reactive Gases

Gas confinement is keyThere is a variety of proposed y p psystems for gas confinement

Inert (II)(I)Gas

regions

Source andexhaust slots

The ALD Coating Head

PP

Small GapL G

Large gap to substrate

Large Gap

• Low pressure gradients• Gas will mix across many channels

Small gap → Substrate floats (gas bearing)Small gap → Substrate floats (gas bearing)• High pressure to drive from source to exhaust: Good Isolation• Excellent control of substrate position• Very small “chamber”• Very small chamber

Equipment Design

Current work is on aCurrent work is on a laboratory scale unit• 2" wide coating width• Used with discrete 2.5" square

substrates

Process demonstrationsProcess demonstrations• Gas isolation• ALD film growth and saturation• Open air operation →

extendability to long substrates

8

Isolation of Precursor GasesHow good is the gas separation?• Measure by using a “tag” gas (NH3) in the metal channels

P

• Look for crossover of this gas to the oxygen channels

“metal”h t

ppm level NH3 detector

Results

exhaust “oxy-”exhaust

Pure NH3

pp 3

• Stationary operation: No detectable mixing• At our current maximum velocity (0.26 m/s): ~23 ppm mixing• Gas phase reaction minimal factor

© Eastman Kodak Company 9

• Gas phase reaction → minimal factor

Saturation Behavior for TMA/Water

Relative residence times hard wired by h d d ihead designHowever• Constant flow: accurate

1.2

1.4 200 °C

• Constant flow: accurate control over chemistry levels

• Very sharp chemistry changes 0 6

0.8

1.0

h/C

ycle

(Å)

PTMA(mTor)

PWater(mTor)

g

Clear saturation behavior

Å0.2

0.4

0.6

Gro

wth r) r)

300 170

30 170

• Saturation near 1.2 Å/cycle for TMA/Water

0.00 500 1000 1500 2000 2500

Residence Time (msec)

300 17

30 17

© Eastman Kodak Company 10

Equipment Development (underway)

Objectives• Increased coating width• Web handling

d

Phase A• Coating head migration to 6” width

6 in

ch ri

gid • Coating head migration to 6 width

• DEMO: Ability to construct wider heads

• DEMO: Uniform delivery of gas6s

DEMO: Uniform delivery of gas

Phase B

Sho

rt P

ass

Web

• Short pass 6” web unit• DEMO: Ability to handle free

standing webs

S

Throughput

Empirical model can be constructed for a given reaction systemreaction system

1.200

1.400

GPC

(A)

1.200

1.400

GPC

(A)

Cyc

le

Slot spacing gth

0.000

0.200

0.400

0.600

0.800

1.000

G

GPC

GPC-Calc

0.000

0.200

0.400

0.600

0.800

1.000

G

GPC

GPC-Calc

Gro

wth

per

C p g

Rea

ctor

Len

RequiredThickness

Currently have good data on Al O and ZnO

0.000 1.000 2.000 3.000Residence (sec)

0.000 1.000 2.000 3.000Residence (sec)

Residence TimeModel

Web Speed

Currently have good data on Al2O3 and ZnO system

12

Throughput for Al2O3 or ZnO

30

00A

(m)

The material system tt

30

00A

(m)

230 mTorr

15

20

25

Leng

th fo

r 100 matters

Slot spacing is a weaker dependence

15

20

25

Leng

th fo

r 100

Al2O3Water

5 9 Torr

0

5

10

Zone

weaker dependenceLonger spacing• Longer residence →ZnO0

5

10

Zone

L 5.9 TorrWater

00 5 10

Speed (m/min)

• Longer residence → more deposition per cycle

• Easier head assembly

ZnO00 5 10

Speed (m/min)

Gro

wth

Per

Cyc

le • Easier head assemblyTo date: Small to no effects when not in complete saturation

Example:200 Å Film5 m/min web speed

Residence Time saturation → 1.6 m zone

S-ALD ZnO Thin-Film Transistors (TFT)

TFTs: the drive element for flat displays• Laptop screen: a-Si TFTs with mobility ~1 cm2/V-sLaptop screen: a Si TFTs with mobility 1 cm /V s

To drive an OLED• Higher mobility is needed to handle the pixel current

E t d

• Higher stability is needed to continuously supply the pixel

ZnO is a promising alternative

250 Å ZnOBy the SpatialALD Process

EvaporatedAl Contacts

Vd

i

Glass ITO Gate Layer on Glass

1100 Å Al2O3

ALD Process

Vg

© Eastman Kodak Company 14

(commercially obtained)Side View, Schematic

Typical Device Performance

1.E-03

m2) DF141-6-D6C-ID-[VA=0]

DF141 4 D1E ID [VA 0]1E 02 (A)

Al2O3 Dielectric Leakage

1.E-07

1.E-06

1.E-05

1.E-04

eaka

ge (A

/cm DF141-4-D1E-ID-[VA=0]

DF141-2_D2C-ID-[VA=0]

1E-06

1E-05

1E-04

1E-03

1E-02

Dra

in C

urre

nt

150 °C200 °C

W/L = 600/50 mmTox= 1100 ÅITO GateShadow mask Al contacts

1.E-10

1.E-09

1.E-08

0 2 4 6 8 10Applied Field (MV/cm)

Le

1E-11

1E-10

1E-09

1E-08

1E-07 250 °C

High on/off ratio >108

IgApplied Field (MV/cm)

1E-12

1E-11

-10 0 10 20 30Gate Voltage (V)

Additional Characteristics…St bilit C bl t SiHigh on/off ratio 10

Low gate leakage <2.5 × 10-8 A/cm2

Mobility: ~15 cm2/V-s

Stability: Comparable to a-Si.2.3 MHz ring oscillator circuits: Fast (J. Sun, et al., IEEE Electron Device Lett.))

Mapping Electrical Characteristics

Deposited film on Si( h thi k t ) Map of Linear Differential Mobility(shows thickness steps)

Thinner dielectric and

Measurement region: Central area

Thinner dielectric and semiconductor

© Eastman Kodak Company 16

TFTs with Shadow-masked Al Contacts

Linear Mobility Vth12 3 ± 0 6 cm2/V-s 4 68 ± 0 04 V12.3 ± 0.6 cm /V-s 4.68 ± 0.04 V

240 devices

© Eastman Kodak Company 17

Bias Stability

Initial observations• Stability depends on Gate Bias (not current flow)Stability depends on Gate Bias (not current flow)• Mobility shows little change

Conditions• Typically stress time = 10,000 s• Bias applied Vg = 10 V (for gate dielectric thickness = 50 nm)• Relatively high field (2 × 106 V/cm)Relatively high field (2 × 10 V/cm)

For W/L = 500/50• Linear: Vd = 0.25 V, drain current ~50 μA• Saturation: Vd = 10 V, drain current ~0.9 mA

© Eastman Kodak Company 18

Passivated TFTPassivation with alumina in Spatial ALD system

200 °C processThickness = 50 nm

1.0E-06

1.0E-05

1.0E-04

t (A

)

Al

Al2O3

Chromium Gate

ZnOAl

1.0E-09

1.0E-08

1.0E-07

Dra

in C

urre

n

t = 0 s

t = 10000 s

Glass

1.0E-11

1.0E-10

-10 -5 0 5 10Vg (V)

0 80.91.0

ent Low

0.00.10.20.30.40.50.60.70.8

1 10 100 1000 10000 100000

Nor

mal

ized

Cur

re

Low movement of threshold or

current

© Eastman Kodak Company 19

Stress Duration (s)

Patterning and R2R ALD

Typical Semiconductor ProcessingPhotolithographic processPhotolithographic processLayers applied, then patterned with photoresist + etching

Large-area processing: A new landscape

Material Process Printed Patterning

Direct Print Functional Materials

Physical Vapor Deposition

Chemical Vapor Deposition

ALD

?Selective Area

Selective Area Deposition

Reaction inhibition• If precursors cannot react

with the substrate, the film does not grow

Advantages• Thin inhibitor layers

I hibit b i t d• Inhibitors can be printed• After ALD, film is ready for

next layer

Characterization of Growth Inhibition

ess)

½ Sample N I hibit

½ Sample O.D

. (~t

hick

ne

(inhibition)

Inhibitor spun on sample

No Inhibitor InhibitorALD Cycles

During ALD growth, sample removed periodically• No inhibitor: Normal surface growth

I hibit id R d d/ li i t d th• Inhibitor side: Reduced/eliminated growth

Growth of ZnO characterized with 355 nm optical density

PMMA: a Good Inhibitor

PMMA solutions spun on borosilicate glass• 950 000 MW (Microchem 950 PMMA A4)

38 A PMMA

• 950,000 MW (Microchem 950-PMMA-A4)• Thicknesses by ellipsometry on silicon controls (3000 rpm)

– 9 Å (0.025% solution)18 Å (0 05% l ti )

1

1.2

1.418 A PMMA

9 A PMMABare

@ 3

55 n

m

– 18 Å (0.05% solution)– 38 Å (0.1% solution)

Inhibition results

0.4

0.6

0.8

O.D

. @

Inhibition results• Strong inhibition even for

9 Å film40 Å it bl f t

0

0.2

0 500 1000 1500

ALD Cycles

• 40 Å suitable for most applications

• Thinness: Quick inhibitor l f i li ALD Cyclesremoval for inline process

TFT Structure Completely by Selective Area ALD

1000 Å Doped ZnO 1100 Å Al2O3 300 Å Intrinsic ZnO 1000 Å Doped ZnO

StSt St StStampStamp Stamp Stamp

PDMS Based StampingPDMS Based Stamping

Result: Working transistor with mobility ~3 cm2/V-s 1.0E-06

1.0E-05

1.0E-04

1.0E-03

Drain Current (Vd=10V)Drain Current (Vd=20V)Gate Leakage (Vd=10V)Gate Leakage (Vd=20V)

mobility ~3 cm /V-sAll layers by ALDAll patterning by selective area depositionTransparent too! 1.0E-12

1.0E-11

1.0E-10

1.0E-09

1.0E-08

1.0E-07

© Eastman Kodak Company 24

Transparent, too! -10 0 10 20 30

Conclusions

Spatial ALD Approach• Open air performance demonstrated on rigid

substrates

• Scaleup and flexible work underway• Scaleup and flexible work underway

A li iApplications• High performance semiconductor /

dielectricsdielectrics

• Accessible patterning

© Eastman Kodak Company 25