Pushing the Performance of Electro-Mechanical Thin Films Paul - - PowerPoint PPT Presentation

STI-PM

1

Paul Muralt EPFL Swiss Federal Institute of Technology of Lausanne, Electroceramic Thin Film Laboratory 1015 Lausanne, Switzerland. IEEE UFFC Distinguished Lecturer

Pushing the Performance of Electro-Mechanical Thin Films

Ljubljana, 27 September 2016

STI-PM

Downscaling

2 Feynman: “There is plenty of room at the bottom”

Ultra-high density solid state logic and memory IC’s

22 nm

World wide web *) by Photolithography In spite of wavelength limitations

*

1dim precision, in pocket and wrist watches

𝜇

STI-PM

3

MEMS research MEMS products

!

1995 2001 2010

Downsizing of RF filters for mobile phones

(courtesy R. Ruby, AVAGO)

Smart phones

There is more than microelectronics!

Example of thin film bulk acoustic wave filters: Preceding materials research, process technology, and component innovation

More than Moore!

STI-PM

Towards industry 4.0, robot society, etc.

4

Myriads of micro sensors, micro actuators, micro transducers are needed …. Electro mechanical transducer voltage strain stress charge = Micro-Electro-Mechanical System (MEMS) + Downscaling

STI-PM

Important Role of Piezoelectrics

5

Piezo- electric material voltage strain stress charge = Micro-Electro-Mechanical Systems + Downscaling

Pb2+ Ti4+ Zr4+ O2-

Pb(Zr,Ti)O3 (PZT) Strongest piezoelectric ceramics Electromechanics of ionic crystals Major competing principle based on electrostatics forces vacuum/air gap

STI-PM

Important Role of Piezoelectrics

6

voltage strain stress charge = Micro-Electro-Mechanical Systems + Downscaling

• 0.05

0.05 0.1 0.15 7000 7200 7400 7600 7800 8000 8200 8400 8600 G Conductance B Susceptance

Admittance [S] Frequency [MHz]

30 µm square Q=460 k

2=5.9%

P E(t) ΔD(t)

Trapped wave within thin film AlN wurtzite

3 2

Very low loss as compar ed to real coil

STI-PM

Outline

• What is piezoelectricity?
• How to put materials into microsystems?
• Which are the achieved properties?
• What are the issues in performance limitation?

1. Nucleation of PZT thin films 2. Piezoelectric cracking 3. Heating by unipolar cycling 4. Imprinting, Aging

7

STI-PM

ks1 ks2

• +

+

• Charge separation

upon pulling ks1<ks2

d=charge/force=

" # 𝑙%& '& − 𝑙%& '&

• q

+q

d=strain/E=

" # 𝑙%& '& − 𝑙%& '&

Basics of piezoelectrics: polar chain example

𝐷 𝑛 𝑂 𝑛 = 𝐷/𝑂 𝐷 𝑛 𝐷𝑊 𝑛 / 𝑛 = 𝑛/𝑊 c Brothers Pierre and Jacques Curie Discovered the direct effect in 1880 at Tourmaline known as pyroelectric Direct: Converse:

The effect works in both directions:

STI-PM

Consequences and Symmetry requirements

Absence of inversion center, no polar axis (like in quartz)

The gravitation center of positive ions moves differently than the one

• f negative ions.

Piezoelectricity is strongest in ionic crystals (large q), and proportional to the dielectric constant (strain per voltage, charge per stress). The linear relation between second order stress tensor and vector of electric field 𝐸 = 𝑒𝑈 3 requires absence of inversion center among the symmetry elements, since stress T is invariant, and displacement field D changes sign upon inversion.

STI-PM

• +

+

• Basics of piezoelectrics: polar structures
• +
• +
• +

+

• +
• +
• Polar surfaces

Example AlN: N-polar surface Al-polar surface Transition to 3d structures: Coupling with transverse and shear deformations

STI-PM

Frequently encountered situation with one polar axes (4 and 6 fold rotation axis, random in plane, poled ceramics)

PP

S3=d33E3 S1=d31E3

P

P S5=d15E1 Longitudinal (33) – Transverse (31) Shear (15, 24)

Shear deformation in the plane of electric field and polarization

 E / /  P

E

 E ⊥  P

Strain, stress in reduced index notation: ii=i, 12=6, 13=5, 23=4

STI-PM

Concept of Coupling Factor K2

12

K2=

Created mechanical energy Electrical input energy in capacitor Ratio of stored electrical and mechanical energy in piezoelectric material or device structure (MEMS)

K2=

Created electrical energy in the capacitor Mechanical input energy ( Converse) (Direct) Bulk materials: Champions: PZT ceramics and related single crystals (PMN-PT) > 50 % LiNbO3 single crystal SAW structures (Surface acoustic wave) 5-7 % Quartz single crystal < 1 %

STI-PM

Ferroelectric materials as piezoelectric ceramics

T > Tc E = 0 E

Ferroelectric domains T

c: critical or Curie temperature

Most performing ceramics: Pb(Zr,Ti)O3 52/48: 14 possible domain orientations Thin films can be made textured. All grains equivalent. Larger piezoelectric effect?

STI-PM

Strong piezoelectric effects are found in inorganic crystals close to morphotropic phase transitions

14

Pb2+ Ti4+ Zr4+ O2-

Perovskites: usually cubic phase at high temperatures. Phase transition to lower symmetry required. In PZT=Pb(Zr,Ti)O3, transition at 250 to 490 °C, depending on Zr/Ti ratio. At 52/48 = MPB transition, T

c at around 360 °C.

P // [001] P // [111] 90 ° domains in {100} film Sol-gel thin film, piezo AFM a/c domain pattern

STI-PM

+ Ferroelectricity, adding highly polarizable ions (e.g.,

Pb) and polarization switching on the nano scale

15

Ferroelectric domains provide amplification on the nanoscale: Charge and strain amplification by domain effects However:

• Non linearity
• Increased Losses
• Back switching – Poling issues
• Fatigue (bipolar operation)
• Irreversibility – history dependence
• 200
• 150
• 100
• 50

50 100 150 200 200 400 600 800 1000 1200 1400 1600

1st cycle 2nd cycle 3rd cycle

Bias field to top electrode (kV/cm)

er

0.02 0.04 0.06 0.08 0.10 0.12 0.14

tan d

• 600
• 400
• 200

200 400 600

• 40
• 20

20 40

Polarization (µC/cm

2)

Electric field to top electrode (kV/cm)

P P CV-curves PZT thin film PZT PZT

εr

tan(δ )

Sol-Gel

• N. Chidambaram

Transverse stress loops

T

1 = −e31, f ⋅E3

500 MPa ! @ 300 kV/cm 1 µm

A.Mazzalai

Recently: e31,f > 20 C/m2

STI-PM

Thin film situation on elastic layer/substrate

Q3 A T1 T2 T3 T3 S1 T1 T3 T3 S1 T2 S2 S2 Piezoelectric bulk sample Piezoelectric thin film on substrate

• P. Muralt, Integrated Ferroelectrics 1997

R z x x hn tp

S1(z ) = z R

Mixed boundary conditions: d33 à d33,f : only thickness changes e31à e31,f : T3 does not change (=0). Bending controlled by e31,f

STI-PM

Piezoelectric MEMS structures and devices

17

Piezoelectric micromachined ultrasonic transducers (pMUT) in motion, 2-dim array, and as ultrasound probe in emission-receive experiment

STI-PM

More motion

18

1995 2006 Towards ink-jet

STI-PM

Flexural devices (a) 0 1 2 3 4 5 6 7 8 9 10 Contour modes (b) Thickn modes (d) Log(frequency(Hz)) Lamb wave modes (c) b c A d c S a

Immense Frequency range of piezoelectric thin films

Auto-focus lens for mob.phone Inkjet printing head RF filters for mob.phone Loudspeaker and Microphone for mob.phones Energy harves ting Bio-MEMS: particle filters / transport Limited by Qf = 5’000 GHz

STI-PM

Materials “Road Map”

20

Category BULK Thin Film

Category Bulk Thin film High Q, Low K2 High stability Quartz single crystal AlN/SiO2 multilayer High Q, medium K2, medium stability (in T) LiNbO3 single crystal AlN, AlScN AlScN, high Sc content Lower to low Q, Large K2 Less stable (in T, and properties) PZT ceramics PZT thin film AlScN thin film, high Sc ….........ev. lead free substitute

Time and frequency control RF filters Energy harvesting Acoustics Low frequency devices

STI-PM

21

Progress in PZT and related films

( till 2008: see Muralt, J.Am.Cer.Soc. May 2008)

e31, f = d31 s11

E + s12 E

σ1 = −e31, f E3 D3 = e31, f (x1 + x2)

Lugienbuhl et al 1996 Muralt et al, 1996 Kanno et al 1997 Shepard et al, 1998 Dubois et al. 1999 Yoshimura et al, 2001 Seifert et al. 2001 Zhang et al. 2002 Ledermann et. al. 2003 Zhou et al, 2003 Kanno et al. 2004 Nino et al. 2004 Maria et al. 2005 Bassiri et al. 2005, Yokohama et al. 2006 Tyholdt et al. 2007 Krume et al. 2007 Calame et al. 2007

Baek et al, Science 2011 Pb(Mg,Nb)O3-PTO

3 1 Sol-gel sputter

STI-PM

AlScN: Comparison of e31,f values

Industrial project with EVATEC

About d33

STI-PM

How to put materials into microsystems

23

STI-PM

Chemical solution deposition (sol-gel): Automatic CSD tool

Rotating covered chuck and media arm

Wafer on chuck with media arm

High volume piezoelectric thin film production process for microsystems (2010-2013)

Piezo-MEMS workshop, Lausanne, 2011

wafer

Spin coat

s

• l
• g

e l s

• l

u t i

• n

Hot plate (350oC)

STI-PM Piezo-MEMS workshop, Lausanne, 2011

600 to 700 °C

STI-PM

In-situ Sputter deposition of PZT thin films

cluster PZT Module at EPFL

STI-PM

Sputter deposition of AlN

STI-PM

28

Integration: PZT in silicon MEMS

SiO2 With or without Pt

Buried oxide

4 µm PZT

cavity substrate 10 µm Si

Silicon Micromachining: Controlled properties like resonance frequency of laminated plate SOI substrate

Stable electrode with passivated Ti- TiO2 adhesion layer, or equiv.

Resist residues

Dense PZT film for lateral force development, Optimal properties = Controlled texture / Composition / grain size

• B. Belgacem, F. Calame, P. Muralt, 2006

SOI: Silicon on Insulator Si/SiO2/Si

STI-PM

• 1. Nucleation issue in PZT growth

29

Nanostructural tuning of the texture of PZT pervoskite thin films grown by RF sputtering for piezoelectric MEMS

Andrea Mazzalai, Martin Kratzer, Cosmin Sandu, Ramin Matloub and Paul Muralt

STI-PM

Nucleation controlled growth of Pb(Zr,Ti)O3

Consequences: Strong influence of substrate:

• Most easy is epitaxy on perovskite single crystal substrate
• For flexible, stress compensated MEMS integration, another

concept is needed: seed layer, such as PbTiO3 {100}, LaNiO3 (100) , etc. on Pt

• Or TiO2 island structure on Pt(111) !

Application: industry wants to save one chamber. Ti Pt PZT PZT

Magnetron sputter chambers of a 200 mm cluster

STI-PM

• 4.4eV/unit cell to nucleate the perovskite phase *
• 1.1eV/unit cell to continue the growth *
• PbO is highly volatile
• Parasitic phases are more tolerant to stoichiometric

deviations

• Process is nucleation-controlled
• T>550°C required
• PbO excess in the target required
• Film quality is defined by microstructure

Thermodynamics of PZT deposition

*)

STI-PM

Pb-Ti-O seed layers on Pt(111)

32

2 to 3 nm TiO2/Pt(111) Good for PZT(111)

PbTiO3{100}/Pt(111) Good for PZT {100}

• P. Muralt et al, JAP 1998
• S. Hiboux, P. Muralt, JECS 2004
• P. Muralt, JAP 2006

Pt(111) Seed layer PZT (???)

• ΔG for lead

titanate larger than for lead zirconate (100) (111) (100) (110) (111) ???

STI-PM

20 22 24 26 28 30 32 34 36 38 40 42

No seed 5 Å 10 Å

Pt(111) PZT(111) Pyrochlore(222)

Log(Intensity) [arb. units]

2Q [degrees]

PZT(100)

X-Ray diffraction, sputter deposited PZT from compound target

STI-PM

• A. Mazzalai - Nanostructural tuning of sputtered PZT for piezoMEMS

Difficult to analyze with standard techniques (SEM, XPS) Conductive Tip Atomic Force Microscopy (CTAFM)

• nm-scale lateral resolution
• high contrast between TiO2 and bare

Pt

TiO2 tends to accumulate into the grain boundaries If TiO2 is too extended, then PZT grows (111)

STI-PM

35

Binding energy of PbO (or Pb) TiO2 island TiO2 plane film Gas phase Chemisorbed PbO Diffusion flux PbO adatom density Considerable desorption PbO excess: Surface PbO terminated: (100)

PbO adatoms on Pt: large diffusion paths

STI-PM

Patterned TiO2 Platinum TiO2

Growth of (100)

• rientation

indicates large Pb flux to TiO2 seeds Site controlled + size controlled !

• S. Buehlmann, P. Muralt, S. Von Allmen, Site controlled nucleation and growth
• f small ferroelectric PZT single crystals, Appl.Phys.Lett. 84, 2614-16 (2004).

No TiO2: (111)

STI-PM

Sputter Deposition from a single ceramic target

Deposition Temperature: 600 °C EPFL – Oerlikon collaboration

STI-PM

EDX analysis of sputtered PZT thin film

Regions 1 to 4: Zr/(Ti+Zr) variation between 0.57 and 0.59 In-situ growth: diffusion along thickness direction is very limited, to a couple of lattice constants only, as cation diffusion in the perovskite is absent below 800 °C.

STI-PM

Example of sputtered PZT thin film

(Oerlikon-EPFL collaboration)

t = 1.31μm; EDX: Pb1.01(Zr0.54Ti0.46)0.99O3 Very compact microstructure. Normal grain size (200nm). Stoichiometry is very close to the ideal values. Values of e31,f are very good. Very high is the maximum achieved stress: 565MPa The non-linearity of the piezoelectric effect is

• bserved: increases at high electric fields

(above 150kV/cm)

1.3 µm 1.7 µm

STI-PM

• 2. Performance issue: break down and

film cracking

40

Cantilever bending à In-plane film stress by piezoelectric effect

Mazzalai et. al., JMEMS 2015

• A. Mazzalai, D. Balma, N. Chidambaram

STI-PM

Cracks evidencing in stress loops

Sol-gel film Critical stress is about 350 MPa. Including residual stress (+ 150 MPa), around 500 MPa Amazing agreement with estimated toughness as derived from ceramics data.

STI-PM

“Self cracking”

• Cracking due to piezoelectric

stress

• Electrical break down

PZT{100} film (111) grain

STI-PM

Stress resistance - Toughness

σc = K1C / πc

V

tp

σ1

( p ) = −e31, f E3 = −e31, f

V t p

As e31,f is negative, the piezoelectric stress is positive (tensile) at E>Ec in ferroelectrics Critical stress for crack propagation: c: relevant length of crack, K1c: Toughness coefficient à c < tp:

σ1

( p )

σc ≥ K1C / πt p

STI-PM

How large is the toughness coefficient?

• K. Mehta, A.V. Virkar, JACS (1990):

PZT 54/46: K1c = 0.8……..1.8 MPa*m1/2 S.L. dos Santos, D.C. Lupasco, J. Rödel, (JACS 2000)

K1c = 0.8……..1.2 MPa*m1/2

450MPa < σc < 670MPa

With 1 µm film thickness

STI-PM

E-field ---- e31,f ---- Stress

45

K1c=0.975 MPam1/2

σ r = K1c πt p

Break down field can go as t-1/2 also because of standard dielectric breakdown with percolation path, however, if it decreases when e31,f goes up, then it is mechanical in the origin.

Critical stress: Theory and experiment

For less than 1 µm: other origin than cracking, or microstructure is different

Calculation for crack propagation model

STI-PM

Lessons learned

• Piezoelectric stress of 600 MPa was achieved
• The higher the piezoelectric coefficient, the lower the

level for self-destruction (Ecrit)

46

σcrit = K1c πt = −e31, f Ecrit .

Is limited to maximally 600 MPa

STI-PM

• 3. Aging of interdigitated transducers with Lead

Zirconate Titanate thin films

• R. Nigon, T. Raeder, N. Chidambaram

STI-PM

Motivations for IDE on PZT

48

§ Pb(ZrxTi1-x)O3 or PZT is a ferroelectric material with advantageous piezoelectric properties. § Interdigitated electrodes (IDE) have some advantages over the parallel plate capacitor (PPE) design, such as a higher voltage response and no requirement for a bottom electrode stable at high temperature in

• xidizing conditions.

PPE configuration IDE configuration

STI-PM

Vibration energy harvesting with IDE’s

49

Distribution of ferroelastic domains on silicon more suited with interdigitated electrode (a-domains are good) Voltage response and figure of merit higher with IDE

Chidambaram et al, TUFFC 2013

STI-PM

50

§ An MgO buffer layer is evaporated onto a Si substrate with 2 µm wet

• xide.

§ A PTO seed layer is deposited by sol-gel. § Gradient-free PZT is deposited by sol-gel. § The interdigitated electrodes are patterned by photolithography, sputtering and lift-off. Pt electrode ITO electrode

2Chidambaram et al.

• J. Micromech. Microeng. 25 045016 (2015)

Processing – Fabrication Route

STI-PM

51

§ FEM simulations with PZT at high permittivity (ɛr=1000) dielectric material. § For Si substrate the field is nearly zero below the fingers, nearly homogeneous between the fingers, and is of the form E=V/(a+Δa) where Δa is proportional to the film thickness. Horizontal E Vertical E PZT MgO SiO2 Pt Pt Pt b a

2016 MRS Spring meeting – May 24

Finite element simulation (FEM)

STI-PM

52

Samples cycled at V/a=200 kV/cm, PV loops as-measured Loops corrected for parasitic capacitance and field lowering

Decreasing gap

2016 MRS Spring meeting – May 24

PV Loop Corrections

• Elimination of parasitics
• Electric field correction

STI-PM

53

MPB PZT with IDE aged for two weeks

§ The ageing seems to be driven by the net polarization, since unpoled samples age very slowly or not at all. § The ageing rate is not dependent on the finger distance nor on the electrode material. § UV exposure increases the ageing significantly. § The ageing speed is composition dependent. § The ageing is much less for epitaxial films.

IDE Properties - Ageing

STI-PM

54

§ Charge compensation at uncompensated grain boundaries. § Similar to ageing by space charge accumulation near grain boundaries in ceramics4.

Ageing

4Genenko et al.

Physical Review B 80, 224109 (2009)

Possible Model for Ageing with IDE’s

Aging is faster with IDE’s because there are more grain boundaries perpendicular to the electric field.

STI-PM

• 4. Reliability – Life time issue

Accelerated test

Motivated by collaboration with ST-Microelectronics

55

STI-PM

Accelerated cycling test

150 kV/cm 5 µs 5 µs Unipolar cycling, square wave, 150 kV/cm, 50 % duty cycle, 100 kHz 500 ms

E

time

Measurement cycle (3 per decade) RC time constant is about 1 µs! The peak voltages are thus really applied!

V

• A. Mazzalai, D. Balma, N. Chidambaran, R. Matloub, P. Muralt, JMEMS 24 (2015)

STI-PM

Testing a sol-gel film (with gradient)

• 150
• 100
• 50

0.0 0.2 0.4 0.6 0.8 1.0 Cantilever Tip Displacement (µm) Electric Field (kV/cm)

• 30
• 24
• 18
• 12
• 6

DP(µC/cm

2)

1 µm film, {100}-textured Before fatiguing

STI-PM

109 cycles

10

• 1

10 10

1

10

2

10

3

10

4

10

5

10

6

10

7

10

8

10

9 10 10

0.0 0.2 0.4 0.6 0.8 1.0 1.2

• Max. Cantilever Tip Displacement (µm)

number of cycles

• 42
• 35
• 28
• 21
• 14
• 7

Maximum DP (µC/cm

2)

• 150
• 100
• 50

0.0 0.2 0.4 0.6 0.8 1.0 Cantilever Tip Displ. (µm) Electric Field (kV/cm)

• 60
• 48
• 36
• 24
• 12

DP (µC/cm

2)

• A. Mazzalai et al, J.MEMS 2015

leakage

• 150
• 100
• 50

0.0 0.2 0.4 0.6 0.8 1.0 Cantilever Tip Displ. (µm) Electric Field (kV/cm)

• 30
• 24
• 18
• 12
• 6

DP (µC/cm

2)

Wait 10 minutes

Degradation caused by massive

• heating. Problem are the PV

integral, and insufficient cooling

P-loop like at 300 °C

STI-PM

Conclusions

59

There would be much more to tell, as e.g. how to make logic circuits with PZT nano elements.. Anyhow, there is still a lot to study, discover, learn and improve in piezoelectric MEMS and NEMS

Response (pm/V)

Domain wall contributions

STI-PM

60

Thanks for your attention

STI-PM

61

• N. Chidambaram, A. Mazzalai, R. Matloub, C. Sandu, D. Balma
• J. Conde, F. Calame, B. Belgacem,
• S. Bühlmann, M. Cantoni,
• N. Ledermann, S. Hiboux, ….

Electron Microscopy Center of EPFL Micro Machining Center of EPFL European projects: MUSTWIN, MEMSPIE, VIBES, PiezoVolume Swiss projects: COST, CTI, NSF

Acknowledgements