gennadi bersuker

19
Advanced Materials Research Center, AMRC, International SEMATECH Manufacturing Initiative, and ISMI are servicemarks of SEMATECH, Inc. SEMATECH, the SEMATECH logo, Advanced Technology Development Facility, ATDF, and the ATDF logo are registered servicemarks of SEMATECH, Inc. All other servicemarks and trademarks are the property of their respective owners. Reliability assessment for new materials: Generation and activation of electrical defects in high-k gate stacks Gennadi Bersuker

Upload: rogan-rich

Post on 02-Jan-2016

29 views

Category:

Documents


0 download

DESCRIPTION

Reliability assessment for new materials: Generation and activation of electrical defects in high-k gate stacks. Gennadi Bersuker. Dielectric degradation: multilayer gate stack. - Defect location : in high-k or IL? Defect origin : intrinsic or process-related? - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Gennadi Bersuker

Advanced Materials Research Center, AMRC, International SEMATECH Manufacturing Initiative, and ISMI are servicemarks of SEMATECH, Inc. SEMATECH, the SEMATECH logo, Advanced Technology Development Facility, ATDF, and the ATDF logo are registered servicemarks of SEMATECH, Inc. All other servicemarks and trademarks are the property of their respective owners.

Reliability assessment for new materials: Generation and activation of electrical defects in high-k gate stacks

Reliability assessment for new materials: Generation and activation of electrical defects in high-k gate stacks

Gennadi Bersuker

Page 2: Gennadi Bersuker

Dielectric degradation: multilayer gate stack

- Defect location: in high-k or IL?

- Defect origin: intrinsic or

process-related?

- Defect generation mechanism:

stress condition-dependent or

‘universal’?

Page 3: Gennadi Bersuker

• Defects in interfacial SiO2

Process-related High-k-induced: O-vacancies & Hf impurities

• Defects in high-k As-grown: O-vacanciesStress generated – at high stress biasesPolarons

• CharacterizationCombining electrical and physical

techniques, and modeling

Page 4: Gennadi Bersuker

SILC evolution for monitoring breakdown

100 101 102 103 10410-8

10-7

10-6

TiN/3nm HfO2/2.1 nm IL

9.4 9.6 9.8 10.0 10.210

20

30

40

50

60

70

80

90

100

Gat

e C

urre

nt [n

A]

Time [103 sec]

Gat

e C

urre

nt [A

]

Time [sec]

65 nA

500 nASBD

nFET 1x10-8 cm2

CVS 4.6 V

TiN/ 3nm HfO2/2.1nm SiO2

CVS 4.6V

0.0 0.5 1.0 1.5 2.010-15

10-13

10-11

10-9

10-7

10-5

10-3

10-1

101

500 A

1 mA

10 mA

I g[A

]Gate Voltage [V]

TiN/3nm HfO2/ 2.1 nm IL

nMOSFET 1x10-8 cm2

65 nA

100nA; 500 nA1 A; 100 A

0.0 0.5 1.0 1.5 2.010-15

10-13

10-11

10-9

10-7

10-5

10-3

10-1

101

0.0 0.5 1.0 1.5 2.010-15

10-13

10-11

10-9

10-7

10-5

10-3

10-1

101

500 A

1 mA

10 mA

I g[A

]Gate Voltage [V]

TiN/3nm HfO2/ 2.1 nm IL

nMOSFET 1x10-8 cm2

65 nA

100nA; 500 nA1 A; 100 A

Stress-induced leakage current reflects on the formation of percolation path

G.B., IRPS 2007

Page 5: Gennadi Bersuker

Probing SiO2 traps

1.E-01

1.E+00

1.E+01

1.E+02

1.E+00 1.E+01 1.E+02 1.E+03 1.E+04

Stress time (sec)

DN

it /

Nit

0 @

CP

frq

. 1

KH

z

7nm/1nm;4.6V5nm/1nm;4.2V3nm/1nm;4.1V

TiN/HfO2/IL/p-Si

nMOSFET W/L = 1 / 0.2 m

A: 2x10-8 cm2

tH-K / tIL ; Vstress

t1.0

0.5

0.9

1.E-01

1.E+00

1.E+01

1.E+02

1.E+03

1.E+00 1.E+01 1.E+02 1.E+03 1.E+04

Stress time (sec)

DJ

g/J

g0

@ V

g =

2V

7nm / 1nm; 4.6V5 nm / 1nm 4.2V3nm / 1nm; 4.1V

0.4

1.0

t0.9

TiN/HfO2/IL/p-Si

nMOSFET W/L = 1 / 0.2 m

A: 2x10-8 cm2

tH-K / tIL ; Vstress

SILC Charge Pumping

Since CP probes IL, similar CP and SILC growth rates for each dielectric stack points to the same contributing defects in IL

Page 6: Gennadi Bersuker

Effect of stress voltage on reliability assessments

0

2

4

6

8

10

12

14

16

18

0 1 2 3 4

Stress periods (sec)

Nit

grow

th r

ate

Rit

X10

10/c

m2

0

0.5

1

1.5

2

2.5

3

3.5

4

SIL

C g

row

th r

ate

RS

ILC0.67nm

0.84nm

1.0nm

Vg=0.6V

Vg=1.0V

1st 300s 2nd 300s 3rd 300s

Nit probing distance from Si:

SILC voltage:

0

2

4

6

8

10

12

14

16

18

0 1 2 3 4

Stress periods (sec)

Nit

grow

th r

ate

Rit

X10

10/c

m2

0

0.5

1

1.5

2

2.5

3

3.5

4

SIL

C g

row

th r

ate

RS

ILC0.67nm

0.84nm

1.0nm

Vg=0.6V

Vg=1.0V

1st 300s 2nd 300s 3rd 300s

Nit probing distance from Si:

SILC voltage:

1.1nm SiO2/ 3nm HfO2

Vstress = 2.4 V

0 100 200 300 400 500 600 7000

2

4

DNit

/Nit0

Stress time (s)

4KHz

1MHz

1.1nm SiO2/3nm HfO2

Vstress = 4.1 V

Low voltage High voltage

• Low voltage: activation of precursor defects in IL• High voltage: defect generation in IL

Page 7: Gennadi Bersuker

High-k–induced O vacancies in SiO2 IL: EELS

Higher O deficiency higher density of precursor defects (Si-Si) converted by stress into electron traps Si-

HfO2

SiO2

Si

Solid – as-depositedDashed – after 1000C anneal

100 110 120

5000

10000

15000

20000

25000

30000

35000

40000

Y A

xis

Title

Energy-loss [eV]

Si

Si/SiO2

SiO2

SiO2/HfO2

Si L2,3-edge EELS

G.B., JAP 2006

K. van Benthem, Pennycook

Page 8: Gennadi Bersuker

Metal/high-k-induced O defects in SiO2: ESR

Metal/high-k process significantly enhances E’

center density in interfacial SiO2 layer

J. Ryan et al., APL 2007

3nm HfO2/1nm SiO2/TiN+ 1000C PDA

3nm HfO2/1nm SiO2

3nm HfO2/1nm SiO2 +1000C PDA

Page 9: Gennadi Bersuker

3461 3463 3465 3467 3469

Magnetic Field (Gauss)

ES

R A

mp

litu

de

(Arb

. Un

its)

g = 2.0025

g = 2.0035

g = 2.0005

a

b

SiO2 (20Å) + HfO2 (30Å)/TiN + 1000ºC/10s

12

SiO2 (10Å) +HfO2 (30Å)/TiN + 1000ºC/10s

Metal/high-k-induced O defects in SiO2: ESR

High-k-induced (process-related) generation of E’ centers is much more effective in thinner SiO2 layers

J. Ryan et al.

Page 10: Gennadi Bersuker

Fast interface trap generation: DCIV

High-k devices show strong initial increase of both trapped charges and interface traps

SiO2

Neugroschel, IEDM 2006

0

1000

2000

3000

4000

5000

-0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0

VG(V)

DI B

(pA

)

3 nm HfO2

EOT @ 1.2 nm

125 oC, VG = - 1.8 V

Eo @ 3.4 MV/cm

t = 10 s 5 s 2 s 1 s 0 s

[email protected] V

High-k

DDit

DVt

DCIV measurements

Page 11: Gennadi Bersuker

3430 3440 3450 3460 3470 3480

Magnetic Field (G)

SDR

Inte

nsity

(Arb

. Uni

ts)

g = 2.0060 ± 0.0003

g = 2.0033 ± 0.0003

Possible E' Center

0

3

6

9

12

3360 3375 3390 3405 3420 3435Magnetic Field (G)

SDR

Ampl

itude

(Arb

Sca

le)

g = 1.9998

Magnetic Field (G) Magnetic Field (G)

Lenahan, IRW 2006

Hf defects in IL: spin dependent recombination

Fast transient defect generation might be

associated with Hf atoms in interfacial SiO2 layer

SiO2HfO2/SiO2

Page 12: Gennadi Bersuker

Fast degradation: Hf in SiO2 IL

S. Rashkeev, INFOS 2005Hf can diffuse through voids in SiO2

SiO2Si Amorphous layers

SiO2 HfO2Si

Hf

“Regular” structure

G.B., JAP 2006

Page 13: Gennadi Bersuker

Long-term instability: defects in SiO2

10

100

1000

1.E+00 1.E+01 1.E+02 1.E+03 1.E+04 1.E+05 1.E+06 1.E+07 1.E+08 1.E+09

Stress Time (s)

VTH

(m

V)

75 oC, VG = - 1.6 V

n @ 0.19

3.3 nm SiO2

75 oC, VG = - 4 V

HF-last + 3 nm HfO2

Filled: DVTH

Open: adjusted DVTH n @ 0.13

n @ 0.2

103

102

101

100 102 104 106 108 1.E+10

1.E+11

1000 10000 100000

Stress Time (s)D D

IT (1

/cm

2-e

V)

HF-last + 3 nm HfO2

28 oC, - 1.6 V

n @ 0.16

SiO2High-k DDIT - DDIT(1s)High-k

1011

1010

103 104 105

Similar degradation rates in high-k stack and control SiO2 same mechanism

Neugroschel, IEDM 2006

Threshold voltage Interface states

Page 14: Gennadi Bersuker

Defect generation in high-k film

Low Vg:mostly reversible

High Vg:w/ continues degradation

020406080

100120140160180200

-0.5 0.5 1.5 2.5 3.5 4.5 5.5Stress time x1000 (sec)

after discharge

020406080

100120140160180200

-0.5 0.5 1.5 2.5 3.5 4.5 5.5Stress time x1000 (sec)

after discharge20

60

100

140

180

0.5 1.5 2.5 3.5 4.5 5.5

Stress time x1000 (sec)

after discharge

DVt (

mV

)

1.1nm SiO2/ 3 nm HfO2

Vstress= 2.4 V

• Low stress voltage: reversible filling of pre-existing traps• High voltage: trap generation

0.1

1.0

101 102 103

Stress time (sec)D

Vt

(V)

after detrappingbefore detrapping

0.1

1.0

101 102 103

Stress time (sec)D

Vt

(V)

after detrappingbefore detrapping

1.1nm SiO2/3nm HfO2

Vstress = 5 V

Page 15: Gennadi Bersuker

Defect generation in high-k: pulse measurements

102 103

100

DVth

(m

V)

Stress Time (sec)

Breakdown

VSTRESS

= 3.5 V

T = 25 oCPulse time = 300 ns

0.0 0.5 1.0 1.5 2.0

Dra

in C

urr

en

t

Gate Voltage [V]

102 103

100

DVth

(m

V)

Stress Time (sec)

Breakdown

VSTRESS

= 3.5 V

T = 25 oCPulse time = 300 ns

0.0 0.5 1.0 1.5 2.0

Dra

in C

urr

en

t

Gate Voltage [V]

1.1nm SiO2/3nm HfO2

High-k

GateGate

0 500 1000 1500 20000.0

0.1

0.2

0.3

0.4

Def

ect g

ener

atio

n ra

te (m

V/s

)

Stress Time (s)

Defect generation at as-grown defect precursors

Page 16: Gennadi Bersuker

Trapping in amorphous high-k

J. Gavartin, ECS 2006

Injected electron can trap via self-localization (polaron formation) No defects needed to charge high-k film

Page 17: Gennadi Bersuker

Summary• Interfacial SiO2 layer:

- Low bias stress: trap generation at as-processed precursor defects (O vacancies/Hf atoms) induced by high-k dielectric

- High bias stress: new “conventional” defects • High-k film:

- Low bias stress: instability due to reversible electron trapping on as-processed defects (O-vacancies) or polaron formation(?)

- High bias stress: defect generation at as-processed precursors: Defect nature? Mechanism?

Page 18: Gennadi Bersuker

Specifics of metal electrode/high-k dielectric gate stacks

– Multi-layer dielectric stacksInterfacial SiO2, high-k dielectric, metal/high-k

interface– Ultra-short characteristic times

Transient charging/discharging (relaxation) effects– High density of pre-existing defects

O vacancies, under-coordinated metal and Si atoms

Question applicability of SiO2 test methodologies

Page 19: Gennadi Bersuker

New Materials Reliability Issues

• Reversible parameter instability – sensitive to measurement times; can be partially addressed by design

• Stress-dependent degradation mechanisms - test close to use conditions

• Strong process-dependent characteristics – reliability assessment requires extensive set of gate stacks of variety of compositions/processing