Ferroelectricity in Doped Hafnium Oxide

Ferroelectricity in Doped Hafnium Oxide

U. Schroeder1, E. Yurchuk1, J. Müller2, D. Martin1, T. Schenk1, C. Richter1 C. Adelmann3, S. Kalinin5, U. Boettger6, A. Kersch7, and T. Mikolajick1,4 ISAF, State College May 15th, 2014 2 1

3

5

6

7

1

Outline

1. Motivation: Ferroelectricity in HfO2 2. Stabilization of the Ferroelectric HfO2 Phase 3. Ferroelectric Switching Behavior 4. Device Application: 1T FeFET

5. Summary

2

U. Schroeder et al. ISAF 05/2014

Outline


5. Summary

3


Motivation: Ferroelectric HfO2 Ferroelectrics enable

130nm FRAM

fast low power non-volatile memories

e.g. FRAM: -

current scaling limit: 130 nm due to material properties

TI & Ramtron

 new material necessary A lot of industry experience

CMOS

integrating HfO2 / ZrO2:

-

CMOS compatible

-

scalability below 50nm

-

ALD process available

-

ferroelectric properties

(IEDM 2011 / VLSI 2012 / IEDM 2013)

4


chipworks

sub 30 nm

DRAM

Outline

1. Motivation: Ferroelectricity in HfO2 2. Stabilization of the Ferroelectric HfO2 Phase 3. Ferroelectric Switching Behavior 4. Device Application:1T FeFET

5. Summary

5


Capacitor Route Route

Layer depositi on

Anneal + Pt dots

Electrode Deposition

Silicon

Wet Etch

HfO2 deposition

Platinum dots S. Mueller

Simple capacitor processing

7


ALD Process: doped HfO2 Other precursors used for dopant supercycles: • tetrakis(ethylmethylamino)hafnium (TEMAHf) • hafnium tetrachloride (HfCl4) • silicon tetrachloride (SiCl4)

• tetrakis(dimethylamino)silane (4DMAS) • tris(dimethylamino)silane (3DMAS) • tris(isopropylcyclopentadienyl)gadolinium (Gd(iPrCp)3)

• tris(methylcyclopentadienyl)-yttrium (Y(MeCp)3) • strontium di-tert-butylcyclopentadienyl (Sr(tBu3Cp)2) • and trimethylaluminium (TMA) + water, ozon or O2-plasma 8


Pt

Pt

Ti

Ti

TiN

TiN HfO2

SiO2

TiN Native SiO2 Si- wafer

Electric Displacement [C/cm2]

Effect of Si -Doping 60 40

Capacitance [F/cm2]

Si-substrate

Electric Field [MV/cm]

Para  FE

AFE



20

Increase of Si content

0

concentration

-20 -40 -60

4.5

→ Change of electrical SiO2 0 mol %

4.4 mol %

5.6 mol %

6.6 mol %

8.5 mol %

-3

-3

-3

-3

-3

0

3

0

3

0

3

0

3

0

properties :

3

4.0

Effect was confirmed by

3.5

polarisation and

3.0

capacitance -voltage

2.5 2.0 1.5

measurements

0 mo %

SiO2

9 nm Si:HfO after 800oC8,5 Anneal 5,6 mol % 2 6,6 mol % mol %

4,4 mol % -3

0

3

-3

0

3

-3

0

3

-3


9

Pt TiN Si:HfO2 TiN


0

3

-3

0

3

Pr~∫C(V)dV E. Yurchuk et al., Thin Solid Films 2012 A. Toriumi at al. APL 86, 2006

HfO2 phase stabilization

Anneal + Doping 2

Amorphous HfO2

High-symmetry / high-k phases

+ Doping

Anneal 1

Non-centrosym. / FE phase

Non-centrosym. / ‚AFE‘ phase

Low-symmetry / lower-k phase

Cubic Fm3m or Tetragonal P42/nmc

Orthorhombic Pbc21 Monoclinic P21/c 1 Tetragonal/Cubic 13

Tetragonal*

14 Si

39 Y

40 Zr

14 Si

38 Sr

13 Al

64 Gd

13 Al


depending on dopant

S. Müller AVS-ALD conference 2012

HfO2 phase

C. Richter BALD 2014

Ferroelectricity observed when orthorhombic phase dominant 14


Polarization forms

Crystallographic lattices monoclinic

orthorhombic

tetragonal C. Richter BALD 2014 15

mono

orth


orth

HfO2-Phase: TEM Y: HfO2 Grain size ~30nm

TEM

Y: HfO2 Clear assignment ongoing: - no monoclinic grains found - difference cubic/tetrag./orthorh. not detectable within TEM resolution - XRD + TEM: phase different to monoclinic phase  orthorhombic

18


Polarization vs. Piezoelectricity FE

Dielectrics Piezoelectrics

‚AFE‘

Piezoelectric (for FE) or Electrostrictive

Pyroelectrics

(for ‘AFE’) behavior visible

Ferroelectrics

T. Boescke et al., APL 99, 102903 (2011) by Laser Interferometer 19


Piezo Force Microscopy

Ref.: http://en.wikipedia.org/wiki/ Piezoresponse_Force_Microscopy

- Local distribution

- Phase: Polarization direction detectable

D. Martin @ Oak Ridge Nat. Labs

20


Piezo Force Measurements

3 nm nm

180°

a.u.

+4.2 V

2

0°

-4.2 V

1 0

-180°

Topography

Piezo responce polarization value visible

Phase  two polarization direction

- Most HfO2 grains switchable - Rough edges likely caused by domains overlapping the scan area

23


D. Martin et al., Adv. Mat 2014 U. Schroeder et al., IWDTF 2013/ JJAP 2014

Piezo Force Measurements

3 nm nm

180°

a.u.

2

0°

1 0

Topography

-180°

Piezo responce polarization value visible

Phase  two polarization direction

- Most HfO2 grains switchable -  field induced phase change - - similar effect known in literature*

24


* X.Tan et al. Phys. Rev. Lett 105, 2010 AFE  FE

Phase change barrier height (K. Rabe) First-principles calculations (squares) and LandauDevonshire model (solid line) S. E. Reyes-Lilli et al. arXiv:1403.3878 (2014)

•

Energy profile between tetragonal and orthorhombic phases

 ZrO2 barrier height: 35meV/f.u.  Ec~1.2MV/cm  field induced phase change possible

•

HfZrO: polar orthorhombic structure is favored over the nonpolar tetragonal phase by ~23 meV/f.u.

25


Polarization reversal barrier height

•

Polarization by displacement of four O ions in the orthorhombic unit cell

•

Maximum saturation polarization: 53 µC/cm2

•

Correlation between coercive field and barrier height Clima et al., Appl. Phys. Lett. 104, 092906 (2014)

26


Different HfO2 dopants paraelectric

‚antiferroelectric‘

( )

ferroelectric

P

paraelectric

0 0

E

•

Ferroelectricity visible for dopands with different crystal radius

•

‚Antiferroelectricity‘ only for dopands with radius smaller than HfO2

•

Dopant range larger for higher crystal radius Schroeder et al., JSS 2012/JJAP 2014

28


Different HfO2 dopants paraelectric

‚antiferroelectric‘

( )

ferroelectric

P

paraelectric

0 0

Si, Al

E

(FE/AFE)

monoclinic  orthorhombic  tetragonal  cubic ? field induced phase change

Y, Gd, La, Sr (FE)

monoclinic  orthorhombic  cubic Schroeder et al., JSS 2012/JJAP 2014

29


Different HfO2 dopants

Maximum polarisation typically at about 3-6 mol% dopant concentration Schroeder et al., JJAP 2014 accepted 30


Different HfO2 dopants

Si

Al

Y Gd

Sr

Crystal Radius (pm)

Coercive field seems to increase for larger ionic radius of dopant atom

Schroeder et al., JSS 2012

31


Outline

1. Motivation: Ferroelectricity in HfO2 2. Stabilization of the Ferroelectric Phase 3. Ferroelectric Switching Behavior 4. Device Application: 1T/1C FRAM – 1T FeFET

5. Summary

32


Switching Behavior: Speed

Pt TiN Si:HfO2 TiN Si-substrate

S. Mueller et al., IEEE TDRM 2012 J. Müller et al. APL (2011) U. Schroeder et al., JSS 2012

2Pr (µC/cm2)

40 30

-0.6 V -0.8 V -1 V -1.2 V -1.8 V -3 V

RC Delay

20 10

Pulse width (s)

50 1E+11

pos. bias

1E+07

neg. bias

1E+03 1E-01

1E-05

0 1E-8

1E-6 1E-4 1E-2 Pulse Width (s)

1E+0

RC Delay

1E-09 0.5

1.5

2.5

Programming bias (V)

Enhanced domain switching with increasing E-field

Pos/neg differences caused by asymmetry in film stack 34


3.5

Switching Behavior: Wake-up

Pt TiN Si:HfO2 TiN

9.9 mol% SrO

Remanent Polarization (μA)

Si-substrate

Electric Field (MV/cm)

Initial current peaks merge during cycling De-aging of polarization hysteresis

Activation energy for wake-up: ~100meV

35


T. Schenk, ESSDERC 2013 and ECAPD 2014

Switching Behavior: Wake-up


Sense

Cycling

Sense

Time

20

1 kHz

100 kHz

1 kHz

-20 -40

2

0

PRrel (C/cm )

2

Polarisation (C/cm )

40

Voltage

2

PR (C/cm )

Si concentration: 4.4 mol%

initial 100 cycles 1000 cycles 10000 cycles 80000 cycles

-4 -3 -2 -1 0 1 2 Voltage (V)

3

4

5.6 mol% 30 Cycling 3.5 V @ 100 kHz

6.6 mol%

15 0 -15 PR+

-30 30

PR-

60%

15

90% PR 50% PR

0 -15 PRrel+

-30 -1

10

PRrel1

10

3

5

10 10 Number of cycles

7

10

E. Yurchuk, PhD thesis Schroeder et al, JJAP 2014

De-aging of polarization hysteresis Less relaxation with cycling  Retention improved

36


Switching Behavior: Fatique

Pt TiN Si:HfO2 TiN

-3 2 -1 0 1 2 3 voltage

fatigue signal (10kHz)

Rem. Polarization (μC/cm²)

±20

1E+6

2.5V

1E+7

1E+8

2.25V

1E+9

Si-substrate

2 -1 0 1 2 3

2V

time

hysteresis signal

20

10

10 - 0

0

0

-10

-3 -2 -1 0 1 2 3 -10 10-10

-20

Field [MV/cm] -20 20 -20-Electric 1E+9 9 1E+10 10 10 10 -Electric 3 -2 -1Field 0 (MV/cm) 1 2 3

1E+55 10 1E+66 10 1E+7 7 10 1E+8 8 10

Fatigue Cycles No Imprint

-

With cycling: more domains are switching at higher fields

 Fatique causes mentioned for SBT/PZT:

37

[µC/cm P

S. Mueller et al., IEEE TDMR 2012

-

-

Domain wall pinning by mobile charged defects

-

Growth of domain nuclei inhibited

-

Origin: oxygen vacancies or free injected charges U. Schroeder et al. ISAF 05/2014

2]

40

20 20 0

-4

40 -20 -40

-4 -2 0 2 4

4080 K 140 K 20200 K 00 260 K 320 K --20380 K 470 K --40


0 -

Polarization (µC/cm2)

Polarization (µC/cm2)

2040

-4 -2 0 2 44

Si content

Polarization: Temperature dependence


20 10

Si

0

-10 -20 80

270 K 450 K

170 260 350 Temperature (K)

440

T. Boescke at al., APL 2011 S. Mueller et al., IEEE TDMR 2012 U. Schroeder et al., JSS 2012

- Polarization stable up to 400K - Transformation to AFE phase reduces polarization > 400K - Transformation stronger for higher Si content 38


Outline

1. Motivation: Ferroelectricity in HfO2 2. Stabilization of the Ferroelectric Phase 3. Ferroelectric Switching Behavior 4. Device Application: 1T FeFET

5. Summary

41


Device Application: 1T FeFET Metal-Gate Ferroelectric

Performance advantages: • non-volatility • non-destructive readout • low power consumption • switching speed in ns-time range • low operation voltages

Semiconductor n+

- -+ -+ - - -

+

n+

p-Substrate

Idrain

„1“ low Vth

no polarization

„0“ high Vth

Vgate

44




+

+ +

- - -

Semiconductor n+ + + + n+ p-Substrate

Idrain

„0“ high Vth

Vgate

45




Semiconductor n+

-+ -+ -+ - - -

n+

p-Substrate

Idrain

„1“ low Vth

Vgate

46


Ferroelectric Field Effect 1T Transistor Device Application: FeFET Memory Window

28nm N-channel MFIS-FET 60 50

+4 V 100 ns

-6 V 100 ns

ID (A)

40 30 20

MW ~1.2 V

10 0

-1

0 VG (V)

1

• World‘s first 28nm FeFET • Memory window ~1.2V • >90% yield on a 300mm wafer for single transistor devices E. Yurchuk et al., IMW 2012 51


Endurance

Retention programmed -6V 100ns Memory Window(V)

0.5

0.0

1.0

0.5 1

Experimental data Extrapolation

~ 0.9V

0 0

100

0.0

200

-0.5

erased +4V 100ns 0

10

1

10

2

10

3

10

10 years

Temperature (°C)

10 days

~ 0.9V

[ V] @ t4e-5A VtV (Volts)

1.0

4

10

5

10

6

10

7

10

-0.5

8

10

time [s]

K. Khullar Master Thesis

E. Yurchuk et al., IMW 2012

 Memory window after 105 cycles: ~0.9V  Accumulation of asymmetric charge injection closes MW  Detrap pulse can recover memory window  Memory window after 10 years: ~0.9V

57


Endurance 1.2

Vt (Volts)

0.8 0.4

Program

Erase

~ 0.9V

0 -0.4 1

10

100

1000

10000 100000

Number of cycle K. Khullar Master Thesis

58

K. Yurchuk PhD Thesis

•

Cycling in capacitor limited by breakdown

•

Cycling in transistor limited by charge trapping

•

Optimized operation conditions can significantly improve cycling U. Schroeder et al. ISAF 05/2014

Outline


5. Summary

74


Comparison: HfO2 vs. SBT/PZT SrBiTa*

Si:HfO2

Thickn. [nm]

200

10

Pr [µC/cm²]

10-20

25

Emax [MV/cm]

0.06

2.3

Ec [MV/cm]

0.04

>1

k

300

30

0.6 (PZT)

0.1

Memory Window [V]

0.4

1.2

Endurance [cyl]

109

105

Retention

0.2V @10 yr

0.9V @10 yr

FeFET

Ea

75

Wake-up

[eV]


* Sakai et al., NVMTS 2012 I. Stolichnov et al. APL 2003

Summary Material: 

A ferroelectric phase in HfO2 thin films can be stabilized



Ferroelectric phase  most likely orthorhombic phase



Several stabilizing dopants have been identified



Field induced phase change likely



Simulation results available: barrier heights

Ferroelectric Devices:

76



1T/1C: fe-HfO2 adds the 3rd dimension to FRAM scaling



World‘s first 28nm FeFET device - unmatched retention/scalability



HfO2-based FeFET added to ITRS roadmap in 2014


Thank you for your attention This work was supported by funding of the Deutsche Forschungs Gemeinschaft (DFG) (Project: Inferox) and by the EFRE fund of the European Commission within the scope of technology development and in part by the Free State of Saxony (Project: Cool Memory, Heiko, Merlin)

77


and

Thanks to the FeFET – TEAM: 2 5

4

3 6

7

8

9

and many more:

E. Yurchuk1, J. Mueller2, S. Mueller1, S. Slesazeck1, T. Mikolajick1 T. Boescke4, D. Martin1, D. Zhou1, J. Sundqvist2, P. Polakowski2, T. Schenk1, U. Boettger5, D. Braeuhaus5, S. Starschich5, C. Adelmann6, M. Popovici6, T. Schloesser3, M. Trentzsch3 , M. Goldbach3, R.v. Bentum3, S. Knebel1, T. Olsen1, R. Hoffmann2, J. Paul2, R. Boschke3, A. Kumar7, T.M. Arruda7, S.V. Kalinin7, M. Alexe8, A. Morelli8, A.Kersch9, R. Maverick9

78


1