in situ studies of ald processes & reaction mechanisms...in situ studies of ald processes &...

In situ Studies of ALD Processes & Reaction Mechanisms

Erwin Kessels

[email protected]/pmp

This tutorial presentation will give …(1) an overview of methods for in situ studies of ALD processes &

reaction mechanisms; and(2) some insight into these processes and mechanisms

Don’t expect:• A comprehensive overview• Techniques explained in large detail

Do expect:• Focus on what can be learned from the methods• Their pros and cons articulated & practical comments• An overview based mainly from own experience

Department of Applied Physics – Erwin Kessels

For more information & feedbacksee blog:

www.AtomicLimits.com

Atomic layer deposition (ALD)


A B A Bt In situ studies:

• Quartz crystal microbalance• Spectroscopic ellipsometry• Mass spectrometry• Gas phase infrared spect.• Surface infrared spect.• Optical emission spect.• X-ray photoelectron spect.• X-ray diffraction• Sum-frequency generation• Adsorption calorimetry• Scanning tunneling micros.• …

In situ studies of ALD processes


• Quartz crystal microbalance• Spectroscopic ellipsometry• Mass spectrometry• Gas phase infrared spect.• Surface infrared spect.• Optical emission spect.

• X-ray photoelectron spect.• X-ray diffraction• Sum-frequency generation• Adsorption calorimetry• Scanning tunneling micros.• …

Discussed today:

Monitoring (linear) film growth


0 50 100 150 200 250 3000

10

20

30

40(a)

Al2O3

Ta2O5

TiO2

Film

thick

ness

(nm

)

Elam et al., Rev. Sci. Instr. 73, 2981 (2002).Langereis et al., J. Phys. D: Appl. Phys. 42, 073001 (2009).

EllipsometryQCM

Al2O3

Al2O3: 1.1 ÅTa2O5: 0.8 ÅTiO2: 0.04 Å

Material Growth per cycle

Al2O3 1.2 Å (100 °C)

Ta2O5 0.80 Å (225 °C)

TiO2 0.45 Å (200 °C)

AB cycles

ALD saturation curves


(b)

(a)

0 20 40 60 80 1000.00

0.05

0.10

0.15

0.20

Gro

wth

per C

ycle

(nm

/cyc

le)

Dose time (ms)0 1 2 3 4 5

Purge time (s)

CVD

0 20 40 60 80

Subsaturation CVD

H2O dose (ms)0 1 2 3

Purge time (s)

Al(CH3)3 precursor Purge H2O reactant Purge

ALD of Al2O3 from Al(CH3)3 and H2O (200 ⁰C)

20 ms – 2 s – 40 ms – 1 sAl(CH3)3 – purge – H2O – purge

Vary one parameter while keeping other constant:

Quartz crystal microbalance (QCM)


• Cheap device and relatively easy-to-implement on many reactors• Directly measures mass gain/loss in quantitative way• Very helpful for process development• Very sensitive to variations in pressure, gas flows and temperature

Measures mass variation of a quartz crystal resonator from its frequency change

Precursor Reactant

Pump

QCM sensor = quartz crystal in resonator housing

QCM 1

QCM 2

Elam et al., Rev. Sci. Instr. 73, 2981 (2002).Rocklein and George, Anal. Chem. 75, 4975 (2003).

Quartz crystal microbalance (QCM)

Department of Applied Physics – Erwin Kessels www.inficon.com

• Cheap device and relatively easy-to-implement on many reactors• Directly measures mass gain/loss in quantitative way• Very helpful for process development• Very sensitive to variations in pressure, gas flows and temperature

Measures mass variation of a quartz crystal resonator from its frequency change

QCM – Monitoring mass gain (Al2O3)

Department of Applied Physics – Erwin Kessels Elam et al., Rev. Sci. Instr. 73, 2981 (2002).Wind et al., J. Phys. Chem. A 114, 1281 (2010).

s-OH

s-OAl(CH3)x s-AlOH

s-AlCH3

Mass gain/loss can be monitored per half-cycle

Spectroscopic ellipsometry (SE)


• Directly measures thickness, very helpful for (fast) process development • Yields also insight into many other material properties (optical/electrical) • Optical modelling can be challenging for some layers/materials• Rather expensive and requires special ports for optical access

Measures change of polarization of light upon reflection (multiple wavelengths)

Precursor Reactant

Pump

Broadband light source

+polarizers

Polarizers +

spectrograph +

detector

Window +

Protection valve

Window +

Protection valve

Langereis et al., J. Phys. D: Appl. Phys. 42, 073001 (2009).See blog post about ellipsometry & ALD at www.AtomicLimits.com

Spectroscopic ellipsometry (SE)


Measures change of polarization of light upon reflection (multiple wavelengths)

www.cambridgenanotechald.comSee blog post about ellipsometry & ALD at www.AtomicLimits.com

• Directly measures thickness, very helpful for (fast) process development • Yields also insight into many other material properties (optical/electrical) • Optical modelling can be challenging for some layers/materials• Rather expensive and requires special ports for optical access

Spectroscopic ellipsometry – Saturation (TiN)


0 50 100 150 200 250 3000

4

8

12

16

20

Film

thick

ness

(nm

)

Number of cycles

15 s

60 s

15 s

30 s

0 20 40 60 80 100 120 1400.00

0.02

0.04

0.06

0.08

0.10

Gro

wth

rate

(nm

/cyc

le)

Plasma exposure time (s)

single deposition run separate depositions

Langereis et al., J. Phys. D: Appl. Phys. 42, 073001 (2009).

TiN

Monitor film thickness while changing precursor/reactant dosing timeprovides a fast method to determine saturation curves

Spectroscopic ellipsometry – Nucleation (Pt)

0 100 200 300 400 500 60002468

101214

Th

ickne

ss (n

m)

Number of cyclesDepartment of Applied Physics – Erwin Kessels

100 nm

ALD of Pt from MeCpPtMe3 and O2 on “foreign” Al2O3 substrate (300 ⁰C)

Mackus et al., Chem. Mater. 25, 1905 (2013).

Nucleation delay on

Al2O3

Spectroscopic ellipsometry – Resistivity (Pt)

Department of Applied Physics – Erwin Kessels Leick et al., J. Phys. D 49, 115504 (2016).

Imaginary part of dielectric function ε

Drudeterm

(resistivity)

Lorentz terms

1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.00

20

40

60

80

100

Photon energy (eV)

53 nm

10 nm

5 nm

3.5 nmDie

lect

ric fu

nctio

n ε 2

ALD of Al2O3 [Case study]


s-OH*+ Al(CH3)3 s-OAl(CH3)2 + CH4

s-AlOH + CH4s-AlCH3* + H2O

Prototypical ALD process

Precursor: Al(CH3)3

Reactant: H2O

Temperature: 25-400 ⁰C

A - 1st Half Cycle

B - 2nd Half Cycle

Simplified reaction scheme:

Mass spectrometry ― Reaction products (Al2O3)


Gas phase reaction products

0 25 50 75 10010-11

10-10

10-9

10-8

H2O

Mas

s sp

ectro

met

ry s

igna

l (A)

Time (s)

CH4

Al(

CH

3)3

Al(

CH

3)3

Al(

CH

3)3

Al(

CH

3)3

H2O

H2O

H2O

H2O

0 25 50 75 10010-11

10-10

10-9

10-8

H2O

Mas

s sp

ectro

met

ry s

igna

l (A)

Time (s)

CH4

0 25 50 75 10010-11

10-10

10-9

10-8

H2O

Mas

s sp

ectro

met

ry s

igna

l (A)

Time (s)

CH4

Al(

CH

3)3

Al(

CH

3)3

Al(

CH

3)3

Al(

CH

3)3

H2O

H2O

H2O

H2O

Al(CH3)3 dosing: CH4

H2O dosing: CH4

Vandalon and Kessels, J. Vac. Sci. Technol. A 35, 05C313 (2017).

Quadrupole mass spectrometry (QMS)


• Easy-to-implement on all types of reactors (with differential pumping)• Wide range of species can be detected (but heavy masses difficult)• All reaction products measured (not only from substrate)• QMS cracks molecules into fragments complicating data interpretation

Precursor Reactant

Pump

Mass filter

Detector

Valvewith

pinhole

Pump (p < 10-5 Torr)

Ionization of gas extracted from the reactor & mass filtering of the ions

Quadrupole mass spectrometry (QMS)


• Easy-to-implement on all types of reactors (with differential pumping)• Wide range of species can be detected (but heavy masses difficult)• All reaction products measured (not only from substrate)• QMS cracks molecules into fragments complicating data interpretation

Ionization of gas extracted from the reactor & mass filtering of the ions

Mass spectrometry ― Reaction products (Al2O3)


0 25 50 75 10010-11

10-10

10-9

10-8

H2O

Mas

s sp

ectro

met

ry s

igna

l (A)

Time (s)

CH4

Al(

CH

3)3

Al(

CH

3)3

Al(

CH

3)3

Al(

CH

3)3

H2O

H2O

H2O

H2O

0 25 50 75 10010-11

10-10

10-9

10-8

H2O

Mas

s sp

ectro

met

ry s

igna

l (A)

Time (s)

CH4

0 25 50 75 10010-11

10-10

10-9

10-8

H2O

Mas

s sp

ectro

met

ry s

igna

l (A)

Time (s)

CH4

Al(

CH

3)3

Al(

CH

3)3

Al(

CH

3)3

Al(

CH

3)3

H2O

H2O

H2O

H2O

CH4 probed at 16 (CH4+); 15 (CH3

+), etc.

H2O probed at 18 (H2O+); 17 (OH+); 16 (O+)


Mass spectrometry probes mass/charge ratios

Gas-phase infrared spectroscopy (FTIR)


• Calibration is quite straightforward to yield absolute densities• High sensitivity for certain species but not all species can be detected• All reaction products measured (not only from substrate)• Confinement of reaction products might be necessary for sufficient S/N ratio

Absorption of infrared light (from FTIR interferometer) by rovibrational transitions

Precursor Reactant

Pump

IR light source +

Interferometer IR detector

IR window +

Protection valve

IR window +

Protection valve

Gas-phase infrared spectroscopy (FTIR)


• Calibration is quite straightforward to yield absolute densities• High sensitivity for certain species but not all species can be detected• All reaction products measured (not only from substrate)• Confinement of reaction products might be necessary for sufficient S/N ratio

Absorption of infrared light (from FTIR interferometer) by rovibrational transitions

Department of Applied Physics – Erwin Kessels Vandalon and Kessels, J. Vac. Sci. Technol. A 35, 05C313 (2017).

Gas-phase FTIR ― Reaction products (Al2O3)

QMS and gas-phase FTIR installed in exhaust


Precursor Reactant

Pump

Mass Spectrometer

(p < 10-5 Torr)

Infrared gas cell

(p up to 1 atm)

IR light source +

Interferometer

IR detector

Can quite easily be implemented in industrial (spatial) ALD equipment

Gas-phase FTIR in exhaust of spatial ALD setup


Reaction products: O3, combustion products and CH4

1000 1500 2000 2500 3000 3500 4000

H2OCH4

COAbso

rban

ce

Wavenumber (cm-1)

CH4

CO2

O3H2O

Spectrum of O2 plasma reactant step during plasma-assisted spatial ALD of Al2O3

Mione et al., to be published (2018).

Surface infrared spectroscopy (FTIR)


• Direct measurement of surface groups created, removed or incorporated• Probes only surface groups which are changing every (half-)cycle• Poor S/N ratio for some species – long integration times required• Requires dedicated reactor with optical access and IR-transparent substrate

Precursor Reactant

Pump

IR light source +

Interferometer IR detector

IR window +

Protection valve

IR window +

Protection valve

Absorption of infrared light by vibrational transitions by (surface) groups

Chabal et al., Surf. Sci. Rep. 8, 211 (1988).

Various configurations infrared spectroscopy


Gas phase species Surface species - wafer

Surface species – particles(enlarged surface area by particles)

Surface species – ATR element(multiple reflections at surface)

Chabal et al., Surf. Sci. Rep. 8, 211 (1988).

Surface FTIR – Surface groups (Al2O3)


4000 3500 3000 2500 2000 1500 1000

Wavenumber (cm-1)4000 3500 3000 2500 2000 1500 1000

Abso

rban

ce

Wavenumber (cm-1)

5×10-5

Thermal ALD

Al(CH3)3

-CH3

-CH3

-CH3

-CH3-OH

-OH

H2O or O2 plasma

Thermal ALD

Plasma ALDPlasma ALD

150 °C 150 °C

Langereis et al., ECS Transactions 16, 247 (2008).

-CH3 and -OH are surface groups for both thermal and plasma ALDDifferential spectra: show changes per half cycle

Plasma-enhanced ALD of Al2O3 [Case study]


s-OH*+ Al(CH3)3

s-AlOH + CO + CO2 + H2Os-AlCH3* + 3O

Precursor: Al(CH3)3

Reactant: O2 plasma

Temperature: 25-400 ⁰C

s-OAl(CH3)2 + CH4

A - 1st Half Cycle

B - 2nd Half Cycle

Simplified reaction scheme:


Ar H2 N2 O2

Plasma radiation – feed gas dependent

Profijt et al., J. Vac. Sci. Technol. A 29, 050801 (2011).

Optical emission spectroscopy (OES)


• Ideally suited for process monitoring of plasma-based processes• Extremely easy to implement & cheap• Yields only information about excited species – not ground state species • Typically yields very indirect and qualitative information

Mackus et al., J. Vac. Sci. Technol. A 28, 77 (2010).

Measures (visible) radiation from excited species decaying to lower levels

Precursor Reactant

Pump

Optical fiber

SpectrographImaging lenses

Spectrograph

Window+

Protection valve

Option 1 Option 2

Optical emission spectroscopy (OES)

Department of Applied Physics – Erwin Kessels Mackus et al., J. Vac. Sci. Technol. A 28, 77 (2010).

• Ideally suited for process monitoring of plasma-based processes• Extremely easy to implement & cheap• Yields only information about excited species – not ground state species• Typically yields very indirect and qualitative information

Measures (visible) radiation from excited species decaying to lower levels

Optical emission spectroscopy – Plasma (Al2O3)


0400800

12001600 Pure O2 plasma O2 O

300 400 500 600 700 8000

400800

12001600 ODuring ALD

HβOH

CO

Hα

Emiss

ion

inte

nsity

(a.u

.)

Wavelength (nm)

Plasma half-cycle

AlOH* + CO2 + H2OAlCH3* + 4O

CO2 + H2O + e COex + Hex + ... + e

Plasma is “disturbed” by reaction products

Heil et al., Appl. Phys. Lett. 89, 131505 (2006).Knoops et al., Appl. Phys. Lett. 107, 014102 (2015).

0 s 0.4 s 0.8 s 1.2 s 1.6 s 2.0 st:

Atomic layer deposition (ALD)


A B A Bt Discussed next:

ALD merits:• Conformality• Uniformity• Growth control

Advanced methods:• Sum-frequency generation• Adsorption calorimetry

Conformality – Reaction- vs. diffusion-limited


s0 << 1

s0 → 1

Knoops et al., J. Electrochem. Soc. 157, G241 (2010).Elam et al., Chem. Mater. 15, 3507 (2003).

Anim

atio

n -s

0<<

1An

imat

ion

-s0

→ 1

Conformality test structures

Department of Applied Physics – Erwin KesselsYlilammi et al., J. Appl. Phys. 123, 205301 (2018).

Gao et al., J. Vac. Sci. Technol. A 33, 010601 (2015).

PillarHall™ LHAR structures

Fabricated by Si micromachining

= 500 nm

Conformality tests – sticking probability (Al2O3)

Department of Applied Physics – Erwin KesselsSee presentation Karsten Arts – AF2-Tuesday afternoon at 5 pm

Ylilammi et al., J. Appl. Phys. 123, 205301 (2018).

100 150 200 250 300 350

10-5

10-4

10-3

Sum-frequency generation

(Initi

al) S

ticki

ng p

roba

bilty

s0

Substrate temperature (°C)

PillarHall structures

Initial sticking probability s0

Sticking probability of H2O during H2O step

Sticking probability of H2O <10-4 H2O is not very reactive with –CH3

Good agreement with sum-frequency generation (SFG, see later)

Uniformity – O3 surface loss (ZnO)


0 20 40 600

5

10

15

20

25

ZnO

Film

Thi

ckne

ss (n

m)

Reactor position (cm)

150 cycles

5 s O3

10 s O3

20 s O3

Knoops et al., Chem. Mater. 23, 2381 (2011).

Zn(C2H5)2O3

flow

Surface loss/recombination of O3

Depends on surface termination

Uniformity – O3 surface loss (ZnO)


0 20 40 600

5

10

15

20

25

ZnO

Film

Thi

ckne

ss (n

m)

Reactor position (cm)

150 cycles

5 s O3

10 s O3

20 s O3

10 20 30

Norm

alize

d Q

MS

m/z

= 4

8

Time (s)

20 s

O3 exposure

5 s 10 s

Knoops et al., Chem. Mater. 23, 2381 (2011).

Zn(C2H5)2O3

flowQMS

C2H5-term. surfaceLow O3 loss

ZnO surfaceHigh O3 loss

Growth control - initial growth on foreign surfaces


Spectroscopic ellipsometryALD Al2O3 on SiO2 and Si(111):H surfaces

On foreign surfaces initially no “ideal” ALD film growth

Additional insight is necessary for

• Ultrathin films• Area-selective ALD• Etc.

Vandalon et al., to be published (2018).

Sum frequency generation (SFG)


• Highly sensitive & specific for surface groups (sub-surface species not probed)• Good time resolution, reaction kinetics can be followed in time• Can give absolute values of reaction cross-sections/sticking probabilities etc.• Very complex method requiring highly dedicated setup with laser-system

Nonlinear optical technique with 2 laser beams probing vibrational transitions

Precursor Reactant

Pump

Tunable IR laser beam(broadband, fs resolution)

800 nm laser beam(ps resolution)

Sum-frequency radiation(broadband, fs resolution)


Sum frequency generation (SFG)


• Highly sensitive & specific for surface groups (sub-surface species not probed)• Good time resolution, reaction kinetics can be followed in time• Can give absolute values of reaction cross-sections/sticking probabilities etc.• Very complex method requiring highly dedicated setup with laser-system

Nonlinear optical technique with 2 laser beams probing vibrational transitions


Sum frequency generation – Al2O3 on Si(111):H


Si-H stretch @ 2083 cm-1

Al(CH3)3 reacts with Si(111):H breaking the Si-H bonds

Frank et al., Appl. Phys. Lett. 82, 4758 (2003). Vandalon et al., to be published (2018).

10 ms Al(CH3)3pulses

or translated into sticking probability s0 = (1.9±0.2)x10-3

Reaction cross-section σ= (3.1±0.3)x10-18 cm2

Initial growth of Al2O3 on SiO2 and on Si(111):H


Spectroscopic ellipsometryInitial growth:

1st cyle on Si(111):Hs0 = (1.9±0.2)x10-3

1st cycle on SiO2

s0 = (1.2±0.1)x10-3

Steady-state growth:

x-th cycle (x>>1)s0 = (3.9±0.4)x10-3

Vandalon et al., to be published (2018).

Area-selective ALD (see tutorial Parsons)


Differences in nucleation behavior (initial growth) are often exploited to achieve area-selective ALD

Fundamental insight (preferable with quantitative information) in initial growth is required

Linear growth on growth area

Nucleation delayon non-growth

area

Selective growth

Number of cycles

Thic

knes

s

Growth area Non-growth area

Growth area Non-growth area

Mackus, ALD2017 tutorial (2017).

Adsorption calorimetry


• Provides additional thermodynamic and mechanistic insight• Can be used to verify and benchmark (half-cycle) reactions – also from DFT• New to the field of ALD – needs follow up work• ….

Measures half-cycle reaction heats pyroelectrically using a LiTaO3 crystal disk

Precursor Reactant

Pump

Lownsbury et al., Chem. Mater. 29, 8566 (2017).

QCM

Calorimeter

Adsorption calorimetry


• Provides additional thermodynamic and mechanistic insight• Can be used to verify and benchmark (half-cycle) reactions – also from DFT• New to the field of ALD – needs follow up work• ….

Measures half-cycle reaction heats pyroelectrically using a LiTaO3 crystal disk

Lownsbury et al., Chem. Mater. 29, 8566 (2017).Photos courtesy of Alex Martinson (Argonne National Lab)

Adsorption calorimetry – Reaction heats (Al2O3)

Department of Applied Physics – Erwin Kessels Lownsbury et al., Chem. Mater. 29, 8566 (2017).

s-OH*+ Al(CH3)3 s-OAl(CH3)2 + CH4

s-AlOH + CH4s-AlCH3* + H2O

A - 1st Half Cycle:

B - 2nd Half Cycle:

ΔH = -343 kJ/mol

ΔH = -251 kJ/mol

First-principle calculations

Department of Applied Physics – Erwin Kessels Widjaja and Musgrave, Appl. Phys. Lett. 80, 3306 (2002).

s-OH*+ Al(CH3)3 s-AlOH + CH4s-AlCH3* + H2O

A - 1st Half Cycle B - 2nd Half Cycle

s-OAl(CH3)2 + CH4

So far calculated reaction heats have remained untested with respect to experiment

Concluding remarks


• Various analytical tools for in situ studies of ALD have been discussedQMS, gas phase FTIR, QCM, SE, surface FTIR, OESMany more exist. Combine tools if you can!

Concluding remarks



• Focus can be on• Film growth & properties• Reaction mechanisms• Process monitoring & control

Concluding remarks



• Focus can be on• Film growth & properties• Reaction mechanisms• Process monitoring & control

• Take it to the next level (quantitatively!)• Sticking probabilities• Reaction heats• Transient states• ….Combine experiments with theory/simulations!

Acknowledgements


Plasma & Materials Processing group

Co-workers:

Dr. Adrie MackusDr. Harm Knoops

Dr. Fred RoozeboomDr. Marcel Verheijen

Dr. Ageeth BolDr. Adriana Creatore

&Many PhD students

and postdocs

For more information & feedbacksee blog:

www.AtomicLimits.com

ALD Academy (www.ALDacademy.com)

Dr. Erwin KesselsDept. of Applied Physics

Eindhoven University of Technology

Dr. Gregory ParsonsDept. of Chemical and Biomolecular Engineering

North Carolina State University

Mission: educate students and professionals on the principles, applications and future advancements of ALD and related atomic-scale processes.

in situ studies of ald processes & reaction mechanisms...in situ studies of ald processes &...

Documents