Transmission Electron Microscopyfor

metrology and characterization of semiconductor devices

Bert Freitag, Laurens Kwakman, Ivan Lazic and Frank de Jong

FEI / ThermoFisher Scientific, Achtseweg Noord 5, 5651 GG Eindhoven, The Netherlands

03/05/2017

2

Trends in Semiconductors Industry

Device scaling (“More Moore”), todays technology challenges:

Complex, marginal processes …. Critical dimensions in 3D, < 10 nm

+45 Elements(Potential)

New materials & chemistryNew 3D device structures

Trends in Semiconductors Industry

TEM Microscopy for two use cases: analytics & metrology

Metrology : CD’s, pitch, LER, OL,…

Cu BEOL

Analytics & Defect analysis(EDX, EELS, Diffraction)

b

b c

Co Ti Si

Silicide short LER: LER: FIN CD’s

SiGe SiGe CD’s

ONO CD’sStrain

Ni2Si

NiSi

Si

Ti SiN Si

Metal Barrier

1 2

3

Trends in Semiconductors Industry

TEM demand explodes to meet 3D transistor and litho challenges

• Faster TEM data and more TEM data:

• TEM transition: from Lab to Fab

TEM microscopy needs to be fast, automated and easy to use

(and sensitive, accurate & reproducible for metrology)

20

00

20

01

20

02

20

03

20

04

20

05

20

06

20

07

20

08

20

09

20

10

20

11

20

12

20

13

20

14

20

15

20

16

20

17

20

18

TEM

sam

ple

vo

lum

e

TEM sample volume trend

metrology

materials science

Information source: Intel, SPIE 2016

4

A new STEM metrology workflow

TEM microscopy requires a workflow with different tools to• Prepare thin TEM lamellas from full wafers• Extract TEM lamellas from wafer, return wafers in manufacturing line• Measure TEM lamellas in TEM microscope

FEI’s TEM workflow developments in EU projectsA ‘’TEM metrology box’’ with FEI technology inside

03/05/2017

6

Made inThe EU

What Newly developed Technologies are inside?

• Phase 1: new METRIOS TEM platform– Automated microscope alignements, active drift control

– Automated image acquisition

– Automated STEM and TEM metrology

• Phase 2: extended capabilities– Software Embedded spherical aberration Corrector

– Software Embedded EDX acquisition

• Phase 3: improved capabilities– New EDX detector

– Automated EDX metrology

– Improved accuracy: calibration and distortion corrections

7

(STEM) Probe Cs correctors, ease-of-use!

Corrector Technology is not new, but the important innovation is that

Abberation correction has become a simple push button operation !

• operators can use it instantaneously, automatically; it works and provides consistent, excellent results

• Example: 14 nm device using OptiSTEM

Starting Point: C1A1 correction at ‘low’ mag:

focus + stigmatism

1st order

8

(STEM) Probe Cs correctors

Corrector Technology is not new, but the important innovation is that

Abberation correction has become a simple push button operation !

• operators can use it instantaneously, automatically; it works and provides consistent, excellent results

• Example: 14 nm device using OptiSTEM

Zoom in on silicon after 1st order: Once per day, A2B2 correction:

Axial coma + 3fold stigmatism

2nd order

9

The (demonstrated) benefits of automation

Manual STEM metrology

• High Resolution ( ~ 1.1 A)

• Accurate (~ 1 %)

• CD info in 3D • Image based,

No modeling

• Slow• No Statistical data• Poor precision:

(3% 3 s)

• Fast (~ 10 x faster)

25 nm

Automated STEM metrology

0

20

40

60

80

100

120

Automated Manual

Tim

e p

er

Sam

ple

(m

in)

Sample Prep

Imaging & Metrology

Automation

10

The (demonstrated) benefits of automation

Manual STEM metrology

• High Resolution ( ~ 1.1 A)

• Accurate [~ 1 %]

• CD info in 3D • Image based,

No modeling

• Slow• No Statistical data• Poor precision:

(3% 3 s)

• Fast (~ 10 x faster)• Statistical Data (10 x more)

25 nm

Automated STEM metrology

Automation

21 samples/wfr45 devices/sample5 CD’s / device~ 5000 data /10 hrs

11

The (demonstrated) benefits of automation

Manual STEM metrology

• High Resolution ( ~ 1.1 A)

• Accurate (~ 1 %)

• CD info in 3D • Image based,

No modeling

• Slow• No Statistical data• Poor precision:

(3% 3s)

• Fast (~ 10 x faster)• Statistical Data (10 x more)• Adequate precision

(1% 3s, 3x better)• Accurate ( ~ 0.4 %, 2x better)

25 nm

Automated STEM metrology

Automation

0

0.2

0.4

0.6

0.8

1

TopWidth MiddleWidth BottomWidth HeightSiOC HeightSi Pitch

STEM

Pre

cisi

on

3s

(nm

)

STEM Metrology Dynamic Precision

Manual Metrology

Automated Metrology

Metrios analytical STEM application examples

EDX based metrology of 3D VNAND memory03/05/2017

13

Difficult (diffraction) contrast in direct STEM images for ONO metrology

Plan view STEM imaging of ONO dielectricschematic of 3D VNAND memory

Vertical X-section showing ONO stack

EDX based metrology of 3D VNAND memory03/05/2017

14

EDX provides chemical contrast and is well adapted for ONO metrology

Plan view EDX map with line scan

SiO2SiW

O N OTiN AlxOy

EDX line scan

SiO2Si Si N O N O Al Ti W

Si = 12.2 nmNONO = 2.6/4.2/4.3/9.7 nmTiN = 2.5 nm, AlOx = 3.0 nm

Plasma doping of FINFET (damage and profile)03/05/2017

15

HR STEM image showing silicon atoms (dumbells)

D03

D03

Silic

on

FIN

SiO2 STI

a-Silicon layer

2nm Oxide layer

As precipitates

Silic

on

FIN

Silic

on

FIN

45 nm

voids

Plasma doping does not create significant damage in the FIN silicon lattice

Imec’s FinFet structure

Plasma doping of FINFET (damage and profile)03/05/2017

16

Top of Fin EDX line scan profiles

STEM- EDX chemical map:Silicon, Oxygen, Arsenic

EDX shows As doping accumulation at Si-SiO2 interfaces

SiO2a-Si Si FIN

SiO2 a-SiSi FIN

Sidewall of Fin EDX line scan profiles

EDX scan stepsize:~ 0.15 nm !

As Si SiO2

iDPCa new STEM imaging technique

Imperfection of (S)TEM imaging techniques

18

ABF-STEM

N ?

Ga ?

HAADF-STEM

Ga

N ?

GaN

GaN 𝟏𝟏𝟎(schematic)

New technique: iDPC-STEM

19

GaN 𝟏𝟏𝟎

• High tension: 300 keV• Opening angle: 21 mrad• Cs corrected

Ga

N

Titan FEI: iDPC-STEM of GaN 𝟏𝟏𝟎

iDPC How does it work? GaN

20

𝜃

Q1 Q2 Q3 Q4

𝑰𝑫𝑷𝑪 𝒓𝒑 =

𝜵 ∙ 𝑰𝒊𝑫𝑷𝑪 𝒓𝒑x

y

Q2

Q3

Q4

Q1

Beam

r

4 Segment Detector

-Ex = ∂V/∂x = Q1-Q3

DPCx = Q1–Q3

-Ey = ∂V/∂y = Q2-Q4

DPCy = Q2–Q4

I. Lazić, E.G.T. Bosch, S. Lazar, Ultramicroscopy 160 (2016) 265-280 𝑰𝒊𝑫𝑷𝑪 𝒓𝒑

Integration(electric field is conservative)

iDPC(represents φ)

𝜑 = 𝜎𝑉Patent:US 9,312,098 B2Lazić et al.

ℱ 𝐼𝑖𝐷𝑃𝐶 =1

2𝜋ℱ 𝜓𝑖𝑛

2 ∙ ℱ 𝜑

iDPC (object 𝜑)[1]

Why do(n’t) we see Nitrogen?

21

Periodic

system of

elements(simulation)

ABF[2]

GaN 𝟏𝟏𝟐𝟎(experiment)

ADF (object 𝜑2)[2]

Ga31

N7

Ga31

N7

Ga31

N7

[1] I. Lazić, E.G.T. Bosch, S. Lazar, Ultramicroscopy 160 (2016) 265-280 [2] E.G.T Bosch, I. Lazić, Ultramicroscopy 156 (2015) 59–72

Examples on GaN[211]

zeoliteinterfaces

LiTi2O4Semicon devices

Ultimate lateral resolution Comparison iDPC with HAADF on GaN [211]

63pm 63pm

More information accessibleLight and heavy elements imaged at the same time

with sub Angstrom resolution

2nmZeolite structure: MFI ZSM-5 [010]

Low-dose imaging with iDPC: Zeolite

• Zeolite imaged at 300 keV

• Dose: 961 e-/Å2

• Sample damages at dose of 5000 e-/Å2

• Not possible to focus using ADF image

iDPC

Imaging of extremely sensitive materials possible

Interface termination : iDPC-STEM images of ZrO2/Ni interface

Confidential

25

Titan Themis 300 STEM at 300kV

Image size: 512x512 Frame time: 6 s

Detector: DF4/iDPC Converg. angle: 21 mrad

iDPC

High pass filter

ZrO

Imaging of atomic configuration of the interface and the termination of the substrate

HAADF

Sample: courtesy of Prof. Dr. Wayne Kaplan, Israel Institute of Technology Technion

iDPCADF

Extremely light element imaging Simultaneous ADF-STEM vs. iDPC-STEM

Li clearly visible within LiTi2O4 using iDPC-STEM

LiTi2O4LiTi2O4

t = 19 nm Df = -15 nm

t = 8 nm Df = -5 nm

t = 19 nm Df = -22.5 nm

Experiment Simulation

Observation from comparison with simulations:

• Li is visible if beam best focus is inside sample

• In top-right corner Li is visible: Sample is probably thicker than 20 nm

• Defocus change consistent with thickness change

iDPC

Extremely light element imaging iDPC-STEM: Experiment vs. simulation

Contrast enhancement on semicon devices : STEM – EDS

Confidential

28

STEM EELS in Talos 200kV

Acquisition: 17 min System mag.: 910 kx Current 250 pA

Drift corr.: On Map size: 406x384 Display: Raw counts, smoothed

28

HAADF89 – 210

mrad

DF434 – 89 mrad

DF217 – 32 mrad

BF0 – 14 mrad

Contrast enhancement on semicon devices STEM & iDPC @ low dose

Confidential

29

29

iDPCDF434 – 89 mrad

BF0 – 14 mrad

STEM at 200kV

Image size: 1024 x 1024

Detector: HAADF, DF4, DF2, BF

Frame time: 40 s

Magnification: 640 Mx

Current 5 pA

idpc at 200kV

Image size: 1024 x 1024

Detector: DF4, 1.1m Cl

Frame time: 20 s

Magnification: 640 Mx

Current 5 pA, Filter 0 30

Contrast enhancement at only

~1000pe/pixel electron dose

Conclusions

30

• The newly developed automated Metrios Metrology TEM microscope combines ease of use with excellent metrology capabilities.now adequatefor 10 nm device technology:

precision ~ 0.2 nm 3s, accuracy ~ 0.4%

• Automated and sensitive STEM-EDX allows for fast chemical characterization (dopants, stoichiometry,..) but also complementsSTEM metrology for low contrast materials ( e.g. ONO layer stacks)

• Statistically relevant (S)TEM data provide new solutions for process control of advanced devices: Line roughness analysis (LWR and LER), Hybrid (reference) metrology

for e.g. OCD or CD-SEM

• iDPC-STEM is direct and linear imaging electrostatic potential of the sample .

• It is capable of imaging light and heavy elements together.

• It is a low dose technique with superior contrast, which enables to handle sensitive materials in metrology applications

Thank you!