Aug 20, 2004 - HETEROSTRUCTURES ON SILICON by. Dong Min Kim. A dissertation submitted in partial fulfillment of the requirements for the degree of.
EPITAXIAL PIEZOELECTRIC THICK FILM HETEROSTRUCTURES ON SILICON
by Dong Min Kim
A dissertation submitted in partial fulfillment of the requirements for the degree of
Doctor of Philosophy (Metallurgical Engineering)
at the
UNIVERSITY OF WISCONSIN-MADISON 2004
i EPITAXIAL PIEZOELECTRIC THICK FILM HETEROSTRUCTURES ON SILICON Dong Min Kim Under the supervision of Professor Chang Beom Eom At the University of Wisconsin-Madison
The significantly higher dielectric permittivity, piezoelectric coefficients and electromechanical coupling coefficients of single crystal relaxor ferroelectrics make them very attractive for medical ultrasound transducers and microelectromechanical systems (MEMS) applications.
The potential impact of thin-film relaxor ferroelectrics in
integrated actuators and sensor on silicon has stimulated research on the growth and characterization of epitaxial piezoelectric thin films. We have fabricated heterostructures by (1) synthesizing optimally-oriented, epitaxial thin films of Pb(Mg1/3Nb2/3)O3-PbTiO3 (PMN-PT) on miscut (001) Si wafers with epitaxial (001) SrTiO3 template layers, where the single crystal form is known to have the giant piezoelectric response, and (2) nanostructuring to reduce the constraint imposed by the underlying silicon substrate. Up to now, the longitudinal piezoelectric coefficient (d33) values of PMN and PMN-PT thin films range from 50 to 200 pC/N have been reported, which are far inferior to the properties of bulk single crystals value (d33 ~ 2000 pC/N). These might be attributed to substrate constraints, pyrochlore phases and other effects. Here, we have realized the giant d33 values by fabricating epitaxial PMN-PT thick films on silicon. When the PMNPT film was subdivided into ~1 µm2 capacitors by focused ion beam processing, a 4 µm thick film shows a low-field d33 of 800 pm/V that increases to over 1200 pm/V under bias, which is the highest d33 value ever realized on silicon substrates. These high piezo-
ii reponse PMN-PT epitaxial heterostructures can be used for multilayered MEMS devices which function with low driving voltage, high frequency ultrasound transducer arrays for medical imaging, and capacitors for charge and energy storage. Since these PMN-PT films are epitaxially integrated with the silicon, they can make use of the well-developed fabrication process for patterning and micromachining of this large-area, cost-effective substrate. We believe the technology of heteroepitaxial growth and photolithographic processing to produce single crystal thick film ferroelectrics on single crystal metallic oxide electrodes is a significant step in improving medical ultrasound transducers and MEMS devices applications.
iii
Acknowledgements
I would like to express my deepest gratitude to my academic advisor, Professor Chang-Beom Eom, for his continual guidance and endless patience during my study and research at Duke University and the University of Wisconsin-Madison. Without his encouragement and support, I could not have finished this work. I would also like to express sincere appreciation to my advisory committee, Professor John Perepezko, Professor Eric Hellstrom, Professor Mark Rzchowski and Professor Hongrui Jiang, for providing valuable suggestions and advice. I would also like to thank the many talented collaborators, including V. Nagarajan, Jun Ouyang, Professor Ramamoorthy Ramesh, the late James Lettieri, V. Vaithyanathan, Professor Darrel G. Shlom, W. Tian, Professor X.Q. Pan, Dr. Stephen K. Streiffer, Professor S. Trolier-McKinstry and Dr. John Fournelle. I had lots of fun working with my lab colleagues. I wish to express my deepest appreciation to all of the members of the Oxide Thin Films laboratory from Duke university to University of Wisconsin – Tapan, Minku, Sangdon, Rajesh, Dan, Quing, Jinseo, Jung-Hoon, Kyoun-Jin, Rasmi, Land, Jonathan, Scott, Xiowei, undergraduate student Peter, Jolene, Jeremy and Scott. Especially, I would like to thank Rasmi and Land for their help and comments about my thesis. I would like to remember the good times with our department members – Joonsig, Sung-tae, Dalhyun, ByoungNam, Hyunseog, Sangil, Jiwhan and Hyunmin. I would like to remember my friends Jinwhan and Changhyun who, together, made our dreams of studying abroad come true and my friends in Korea - Younghoon,
iv Kwangsoo, Jungho, JongHun, Bosun. I would like to express my cordial appreciation to many people who gave me encouragement, Prof. ByoungRyul Min, Prof. Jinnam Rhim, Dr. Whagi Kim and Young-wook Seo. I would especially like to thank my wife, Sung Kyu, for her full support and patience throughout my research. Without her endless support, I could not have finished my work. I can not forget the encouragement and help from my parents-in-law, brothers, sisters, and cousins, brothers-in-law, uncles, aunts and grandmothers. Finally, I do not know how I can express my deepest gratitude to my mother’s never ending sacrifice for me. I want to dedicate this thesis to my late my father. He will be happy in heaven with his son’s successful research. God bless to all… August 20, 2004 Dong Min Kim
v
Table of Contents Abstracts…………………………………………………………………………………...i Acknowledgements………………………………………………………………………iii Table of Contents………………………………………………………………………….v List of Figures………………………………………………………………………….…ix List of Tables……………………………………………………………………............xiv
Chapter 1 Introductions……………………………………………………………………………..1 1.1 Motivations…..…..…………………………………………………..........................1 1.2 Statements of Research……….……………………………………………………...6 References…………………………………………………………………………………9
Chapter 2 Characteristics of Piezoelectric Materials…………..…………………...………..…..12 2.1 Dielectric Permitivity………..……………………………………………………...12 2.2 Dielectric Loss………………..…………………………………………………….15 2.3 Electromechanical Coupling Coefficient…………………………………………...15 2.4 Elastic Compliance and Stiffness Properties………………………………….……17 2.5 Piezoelectric Effect…………………………………………………………………17 2.6 Pyroelectric Materials………………………………………………………………18 2.7 Order-Disorder and Displacive Ferroelectrics……………………………………...19 2.8 Type of Distortions…………………………………………………………………21
vi 2.9 Perovskite Structure………………………………………………………………...22 2.10 Ferroelectric Materials……………………………………………………………...23 2.11 Relaxor Ferroelectric Materials…………………………………………………….27 2.12 Pb(B1,B2)O3-PT Solid Solution System……………….…………………………....31 References………………………………………………….…………………………….37
Chapter 3 Synthesis and Characterization of Piezoelectric Thin Films.…………………..……39 3.1 Growth Method for Piezoelectric Materials…….……………………………….…39 3.1.1
Sputtering……………..…………….……………...............…………….…40
3.1.2
Pulsed Laser Deposition.………………………………………….………...43
3.2 Bottom Electrode ……………..……………………………………………….…...45 3.3 Fabrication of Bottom Electrode and Piezoelectric Thin Films……………………46 3.4 X-ray Diffraction…………………………………………………………………...48 3.5 Atomic Force Microscopy………………………………………………………….51 3.6 Electric and Piezoelectric Properties Characterization……………………….…….52 References………………………………………………………………………………..54
Chapter 4 Giant Piezoelectric Response in Epitaxial Pb(Mg1/3Nb2/3)O3-PbTiO3 on Silicon …..56 4.1 Introduction…………………………………………………………………...........56 4.2 Experimental Method...…………………………………………………………….57 4.3 Results and Discussion……………………………………………………………..57
4.3.1
vii Miscut and thickness effect on the formation of PMN-PT perovskite and pyrochlore structure…….………………………………………………….57
4.3.2
Phase Purity and Microstructure…………………………………………...61
4.3.3
Microstructure of Pyrochlore Phase by TEM……………………………...65
4.3.4
Ferroelectric and Piezoelectric Properties………………………………....68
References………………………………………………………………………………..72
Chapter 5 Piezoelectric response in epitaxial 0.67[Pb(Mg1/3Nb2/3)O3]-0.33[PbTiO3] thick films on (001) SrTiO3 substrates……………………………………………………………..73 5.1 Introduction…………………………………………………………………………73 5.2 Experimental Method……………………………………………………………….74 5.3 Results and Discussion……………………………………………………………..75 References……………………………………………………………………………….82
Chapter 6 Thickness dependence of structural and piezoelectric properties of epitaxial Pb(Zr0.52Ti0.48)O3 Films on Si and SrTiO3 substrates ……………………………….84 6.1 Introduction…………………………………………………………………………84 6.2 Experimental Method……………………………………………………………….85 6.3 Structure Characterization………………………………………………………….86 6.4 Ferroelectric properties……………………………………………………………..90 6.5 Piezoelectric Properties……………………………………………………………..93
viii 6.6 Conclusions…………………………………………………………………………96 References………………………………………………………………………………..97
Chapter 7 Conclusions……………………………………………………………………………..98
ix
List of Figures 2.1 Microscopic origins of the electronic polarization…………………………………13 2.2 Definition and physical meaning of dielectric permittivity………………………...14 2.3 Schematic drawing of the order-disorder and displacive ferroelectric……………..20 2.4 Schematic of fundamental type of structural phase transition from a cetrosymmetric prototype…………….….………………………………………...21 2.5 Perovskite structure and phase transition from Cubic to tetragonal structure……...22 2.6 Illustration of the change in a ferroelectric material, which transforms from a paraelectric into ferroelectric tetragonal phase with temperature…………23 2.7 Distortion of BaTiO3 unit cell in its polymorph forms………………………….…24 2.8 Ferroelectric Polarization-Electric field hysteresis data for 4µm PMN-PT film grown on (100) SrTiO3 ……..…. …………………………………………………26 2.9 Relationship among ferroelectric, pyroelectric and piezoelectric………………….26 2.10 Dielectric dispersion in Lead Magnesium Niobate (PMN) as a function of temperature and frequency…………………………………………………………28 2.11 Dielectric hysteresis in PMN as a function of temperature………………………..28 2.12 Structure of relaxor ferroelectric Pb(Mg1/3Nb2/3)O3…………………………….…30 2.13 B-site cations ordered structure and disordered structure………………………….30 2.14 Phase diagram of Pb(Mg1/3Nb2/3)O3-PbTiO3 (PMN-PT) system near rhombohedral-tetragonal morphotropic phase boundary………………………......31 2.15 d33 as a function of crystal composition and orientation…………………………..33 2.16 Polarization and strain as a function of E-field (bipolar) curves for PZN crystals oriented along (a), (b) and (c), (d)………………. ……….35
x 2.17 Strain as a function of E-field (unipolar) curves for PZN crystals oriented along (a) and (b)………………………………...…………………..36 2.18 Domain configuration for poled rhombohedral crystals…………………...36 3.1 Thin film heterostructures and deposition methods………………………………...40 3.2 Physical sputtering process…………………………………………………………41 3.3 Sputtering system geometry………………………………………………………..42 3.4 Photograph of the planar magnetron sputtering chamber…………………………..43 3.5 Schematic drawing of PLD deposition chamber with RHEED…………………….44 3.6 The orthorhombic SrRuO3 structure……………………………………………….46 3.7 Schematic illustration of the geometry and angular relationship for four-circle x-ray diffractometer……..…………………………………………………………48 3.8
Schematic geometry of rocking curve measurement. S= x-ray source, D=Detector ………………………………………………………………………...49
3.9 Schematic geometry of off-axis azimuthal phi-scan in four-circle x-ray diffractometer………………………………………………………………………50 3.10 Four-circle x-ray diffractometer a) HR-XRD with point detector b) GADDS with area detector………………………………………………….…..50 3.11 AFM images of a) BHF treated and annealed (001) SrTiO3. b) sputter deposited (600Å) SrRuO3 / (001)SrTiO3…….. ………………………………………………51 3.12 Schematic drawing of the capacitor measurement…………………………………52 3.13 Piezoelectric Force Microscopy……………………………………………………53 4.1 2-D area detector diffraction images of (a) PMN-PT films on exact (001) SrTiO3 which shows a high content of textured and polycrystalline
xi pyrochlore phases (b) PMN-PT films on 4° miscut (001) SrTiO3 which shows nearly pure perovskite phases ……………………………………………...59 4.2 A schematic of surface of a miscut substrate. Adatoms move to kink site because it is more stable………...…………………...............................………….59 4.3 (a) 4 µm thick PMN-PT films on 4° miscut (001) SrTiO3 which shows nearly pure perovskite phases. (b) 6 µm thick PMN-PT films on 4° miscut (001) SrTiO3 which shows a high content of pyrochlore phases and polycrystalline phases……………………………………………………..…..60 4.4
(a) X-ray θ-2θ diffraction spectra of epitaxial PMN-PT (3.5 µm thick) grown on a SrRuO3 thin film grown on a SrTiO3 buffered (001) silicon substrate. (b) X-ray θ-2θ diffraction spectra of epitaxial PMN-PT (3.5 µm thick) grown on a SrRuO3 thin film on a bulk SrTiO3 substrate (c) φ-scan of the 202 PMN-PT reflection for the PMN-PT on Si. The FWHM of the 002 PMN-PT peak is 0.3° in 2θ and 0.18° in ω (rocking curve)……………………………………………………………….……61
4.5 (a) Rocking curve of (002) PMN-PT on (001) Si. (b) φ-scan of (101) PMN-PT on (001) Si. (c) Rocking curve of (002) PMN-PT on (001) SrTiO3. (d) φ-scan of (101) PMN-PT on (001) SrTiO3…………………………………….62 4.6 A comparison of the in-plane and out-of-plane lattice parameters of the PMN-PT films grown on SrTiO3 and SrTiO3/Si, illustrating the different stress states experienced by the films on the two substrates. As a reference, the pseudocubic lattice parameter of PMN-PT of a similar composition is also given…………………………………………………………….…………..63
xii 4.7
(a) Bright-field cross-sectional TEM image of 1 µm thick PMN-PT / SrRuO3 thin film grown on a SrTiO3 buffered (001) Si substrate, (b) SAED image of SrTiO3 along the [100] zone axis, (c) SAED image of SrRuO3 along the [110] zone axis, and (d) SAED of PMN-PT along the [100] zone axis …………………………..64
4.8 TEM image of 4 µm PMN-PT films on (001) Si. There are some voids in PMN-PT films which are believed to pyrochlore phases………………………….65 4.9 TEM image of 4 µm PMN-PT films on (001) Si. which shows grain boundaries ……………………………………………………...………………….67 4.10 TEM image of 4 µm PMN-PT films on (001) Si. There grain boundaries are believed to pyrochlore phases ………………………………………………….….68 4.11 Piezoelectric coefficient measurement with a) continuous PMN-PT film b) cut-island film milled by FIB………………….………………………………..69 4.12 (a) Polarization vs. electric field 3.5 µm thick PMN-PT films for both continuous and nanostructured capacitor on STO/Si. (b) Polarization vs. electric field for 3.5 µm thick PMN-PT film for a continuous capacitor on STO. (c) d33 vs. electric field for 3.5 µm thick PMN-PT film for continuous and separated capacitors on SrTiO3/Si. (d) d33 vs. electric field for 3.5 µm thick PMN-PT film for continuous and separated capacitors on SrTiO3…………..70 5.1 X-ray diffraction theta-twotheta scans of PMN-PT films of 1, 4.5 and 7.5 µm thickness grown on 4o miscut (001) SrTiO3 substrates …………………...75 5.2 Cross-sectional bright field TEM image of a 9.0 µm thick PMN-PT film near the surface. Pyrochlore impurity phases are arrowed………………..……….76 5.3 Polarization vs. electric field hysteresis loops for 1.0, 4.5, and 9.0 µm
xiii thick films………………………………………………………………………….77 5.4 Temperature dependence of permittivity and dielectric loss for the (a) 1.0 µm and (b) 4.5 µm thick film…………………………………………………………..78 5.5 d33,f data as a function of dc bias field for the 4.5 µm thick PMN-PT film………...79 5.6 e31,f as a function of poling field and time for the 4.5 µm thick film………………80 6.1 X-ray diffraction θ-2θ scan of 0.8 and 3.8 µm thick PZT films on (001) Si Substrates. The inset shows the azimuthal φ-scan of PZT 101 reflection which shows the PZT films on Si grow with cube-on-cube epitaxy………………87 6.2 In-plane and out-of -plane lattice parameters vs. films thickness of PZT films on (001) SrTiO3 and (001) Si substrates…………………………..…………89 6.3 The composition of PZT films analysed by WDS……………………...…………..90 6.4 Polarization hysteresis loop as increasing film thickness for the PZT films on (001) SrTiO3………………..……………...…………………………...92 6.5 Thickness dependent polarization of PZT films on (001) SrTiO3 and (001) Si substrates…………...……………………………….…………………...92 6.6 Typical longitudinal piezoelectric coefficient of 3.8µm thick PZT films on (001) SrTiO3 and (001) Si substrates……………………………..……..94 6.7 Thickness dependence of longitudinal piezoelectric coefficient (d33) of PZT films on (001) SrTiO3 and (001) Si substrates……………………………………94 6.8 Thickness dependent piezoelectric properties of PZT on (001) Si……………..….95 6.9 SEM image of the 4um PZT on Si ……………..……………………..……………95
xiv List of Tables` 2.1 Electromechanical coupling coefficients of typical materials……………………...16 2.2 Typical dielectric constants, piezoelectric coefficients and electromechanical coefficients
for ceramic PZT and relaxor PMN-PT……………………………...27
2.3 Comparison of normal and relaxor ferroelectric materials…………………………29 2.4 Dielectric and piezoelectric properties as a function of crystallographic orientation for rhombohedral PZN and PZN-8%PT…………………………..…...34
1
Chapter 1. Introduction 1.1. Motivations The piezoelectric effect is a physical property that exists in various materials. The name is made up of two parts; piezo, which is derived from the Greek word for pressure, and electric from electricity. It can be described as pressure induced be electricity or electricity induced by pressure [1]. The application of a mechanical force or stress results in the development of a surface charge in the material. This is known as the direct piezoelectric effect. Conversely, the application of a charge to the same material will result in a change in the dimensions or strain. This is known as the converse piezoelectric effect. This phenomenon was discovered in 1880 by the brothers Jacques and Pierre Curie [2]. They measured the surface charges appearing on the specially prepared crystals such as quartz, Rochelle salt, tourmaline, topaz and cane sugars which were subjected to mechanical stress. The converse piezoelectric effect, strain deformation by applied electric field, was mathematically deduced from fundamental thermodynamics by Lippman in 1881 and confirmed by the Curie brothers. A new series of ferroelectric materials were discovered during World War П, by Thurnauer and Deaderick at the American Lava Corporation. They observed a very high dielectric constant (~1100) in BaTiO3 ceramic and studied its ferroelectric properties. The value was enormously higher than rutile TiO2 which has a dielectric constant of about 100 and was known to be the highest value at that time. Soon after, lead zirconate titanate solid solution (PZT) was discovered in 1952. It was theoretically [3,4] calculated and experimentally observed that the piezoelectric properties at the morphotropic phase boundary (MPB) of PZT (Zr/Ti : 52/48) ceramic are much higher (~220 pm/V) than any
2 know single crystals of BaTiO3 [2]. Hence the PZT ceramics have been the dominant piezoelectric material over single crystals for the past 50 years in industries for the ultrasound transducers and actuator applications. A large number of efforts have been done to grow PZT single crystals to get higher piezoelectric properties for better transducer performance, but single crystals of a usable size could not be grown due to its peritectic phase diagram. Another series of ferroelectrics discovered by Smoleskii and Agranovkaya [5,6], display a very broad dielectric maxima versus temperature curve with strong frequency relaxation dispersion in the vicinity of and below the maximum dielectric temperature. These kinds of ferroelectric materials are called relaxor ferroelectrics or simply “relaxors”. Compared with the ferroelectric PZT, the single crystal relaxor ferroelectrics have higher dielectric constants, much more elevated electrostrictive and piezoelectric coefficients and also extremely high electromechanical coupling coefficients. Furthermore, large crystals of relaxor ferroelectrics are relatively easy to grow in comparison with the PZT. In 1982 Kuwata et al. [7] discovered enormously high electromechanical coupling coefficients as high as 0.9 in the single crystal solid solution of lead zinc niobate and lead titanate, that is, relaxor ferroelectric Pb(Zn1/3Nb2/3)O3PbTiO3 (PZN-PT). The discovery of PZN-PT relaxor did not receive much attention at that time. In 1990 Shrout et al. [8] demonstrated the dielectric behavior of a Pb(Mg1/3Nb2/3)O3-PbTiO3 (PMN-PT) single crystals. Relaxor ferroelectrics are in the family of complex lead-based perovskites and its general formula is Pb(B1,B2)O3, where B1 is a low-valence cation, such as Mg+2, Ni+2, Fe+3, Sc+3 or Zn+2, and B2 is a high valence cation, such as Nb+5, Ta+5, or W+6 [9,10].
The lead magnesium niobate
3 Pb(Mg1/3Nb2/3)O3 (PMN) exhibits a diffuse phase transition at Tc ~10°C and strong frequency dispersion at the dielectric maximum which are typical behavior of relaxor ferroelectrics. There was a report of the MPB in the solid solution of PMN with PbTiO3, which exhibits normal ferroelectric and have Curie temperature at 490°C [11,12]. It was reported that the MPB separated the rhombohedral and tetragonal phases of (1-x)PMNxPT and it occurs at approximately x = 0.33 [8,13,14]. Both normal ferroelectric (1x)PbZrO3-xPbTiO3 (PZT) and relaxor type (1-x)Pb(Mg1/3Nb2/3)O3-xPbTiO3 (PMN-PT), compositions near the MPB exhibit anomalously high dielectric constant and piezoelectric coefficients which makes them candidates for a wide range of applications such as ultrasonic transducers, resonators and actuators. Park et al [15] have extensively studied the ferroelectric and piezoelectric properties of many Pb-based relaxor ferroelectrics and their solid solutions. They have reported that relaxor ferroelectrics exhibit a strong anomaly with higher piezoelectric properties whenever they are mixed with lead titanate. It was experimentally observed that ultrahigh longitudinal piezoelectric coefficients (d33) > 2500 pC/N and higher strains are achieved along directions as large as 1.7% in 0.91PZN-0.09PT bulk single crystals along the directions, even though is the polar axis of these compositions [16]. The higher piezoelectric characteristics along the non-polar direction of relaxors have drawn great attention and have been investigated thoroughly in the theoretical and practical aspects. Typical characteristics of relaxor ferroelectrics are ultrahigh dielectric constant and temperature and frequency dependent dielectric properties. A giant dielectric constant results in large electrostriction and electro-optical
4 effects. Hence, relaxor ferroelectrics can be used not only for capacitors but also actuator and transducers. The scientific and technological developments of single crystals of piezoelectric materials, which have enormously higher ferroelectric and piezoelectric properties than conventional bulk ceramics, have attracted many research concerns of piezoelectric device applications. As the size of the electronic device is reduced considerably, it is required to reduce the size of each and every component of the device. The use of single crystals in small dimension ( Tc
Displacive
Order-disorder Figure 2.3. Schematic drawing of the order-disorder and displacive ferroelectric [7].
21 2.8. Type of Distortions
Structural phase transitions can be categorized by their neutrality of sitting charges and dependence of applied electric field. A schematic representation of the various basic structural transitions and definitions is shown in figure 2.4 [6].
T > TC
T < TC
T < TC
Ferrodistortive Applied Field Antidistortive Pyroelectric
Ferroelectric
Antipolar
Antiferroelectric
Charged atoms or groups Figure 2.4.
Uncharged atoms or groups
Schematic of fundamental type of structural phase transition from a cetrosymmetric prototype [6].
22 2.9. Perovskite Structure
In general, ferroelectric have the simple perovskite structure ABO3, where large A-cations occupy the corner positions of the unit cells, smaller B-cations occupy the body-center position and oxygen anions are situated at the face-centered positions which consist of BO6 octahedra surrounded by A-cations as shown in Figure 2.5. The valence of A is in the range of +1 to +3 and that of B is in the range of +3 to +6 [8]. The single crystal ferroelectrics experience phase transition from paraelectric cubic structure to ferroelectric tetragonal structure with decreasing temperature. Therefore, cubic structure at high temperature does not have piezoelectric properties. As the temperature is lowered, the unit cells start to change their shape from cubic and the body-centered B-cations shift away from their central position which change the symmetry of the unit cell. At first, the B-cations shift toward the face center which gives tetragonal structure. At the much lower temperature B-cation shifts along the diagonal direction which gives rhombohedral structure. Due to the shift of B-cation and face-centered oxygen, the direction of Bcation gives polarization direction.
c
a a a
a
Pb O Mg, Nb or Ti
a
Figure 2.5. Perovskite structure and phase transition from Cubic to tetragonal structure.
23 2.10. Ferroelectric Materials
Ferroelectrics are a sub-class of piezoelectric materials. All the ferroelectric materials have piezoelectric characteristics, but the reverse is not true. In other words, not all the piezoelectric materials are ferroelectric; for example ZnO and Al2O3 have the piezoelectric effect but they do not have signature of ferroelectric properties. Most ferroelectric materials undergo phase changes from high temperature paraelectric phase to low temperature ferroelectric phase as shown in Figure 2.6. The cubic paraelectric phase is piezoelectric or nonpiezoelectric or rarely polar [6]. In case of low symmetry tetragonal phases, the centers of the positive and negative charges do not coincide at zero electric field. This gives electric dipoles and the crystal has spontaneous polarization. Ferroelectric materials have at least two equilibrium states of the spontaneous polarization [9]. When the spontaneous polarization can be reversed by applied electric
Dielectric permitivity
field, it is called ferroelectric.
Ferroelectric teragonal phase
Paraelectric cubic phase
Spontaneous Polarization
TC Temperature (°C)
Figure 2.6. Illustration of the change in a ferroelectric material, which transforms from a paraelectric into ferroelectric tetragonal phase with temperature.
24 The temperature of the phase change is called Curie temperature, TC. At the cubic paraelectric phase above the Curie temperature, the dielectric permitivity follows CurieWeiss law
ε=
C T − T0
----------------------------------------- (2-13)
where C is the Curie constant and T0 is the Curie-Weiss temperature, which is less than or equal to the Curie temperature. One example of ferroelectrics, BaTiO3 undergoes several phase transitions to successive ferroelectrics, which is shown in Figure 2.7.
Monoclinic Cubic
Tetragonal
=Orthorhombic
Rhombohedral α
a
c
a
a a
β
a
a
a Tc=130°C
c
0°C
α
a α a
a -90°C
Figure 2.7. Distortion of BaTiO3 unit cell in its polymorph forms [10].
Even though there are several phase transitions, only the first transition temperature from paraelectric to ferroelectric phase is called the Curie temperature. Above 130°C, the BaTiO3 phase is cubic paraelectric. At 130°C, the perovskite unit cell elongates along the face-centered direction, which leads to extreme dielectric permittivity along with changes in elastic properties, thermal properties and the dimensions of the unit
25 cell. Upon further cooling, the structure undergoes another phase change from tetragonal to orthorhombic at 0°C. At even lower temperatures, the orthorhombic structure changes to rhombohedral structure which has diagonal directions for polarization. The spontaneous polarization develops along the following directions as phase changes.
Tetragonal
: ! 6 equivalent directions
Orthorhombic : ! 12 equivalent directions Rhombohedral: ! 8 equivalent directions
As mentioned above, the most typical characteristic of ferroelectric materials is polarization reversibility by an applied electric field. In the presence of an electric field, most of the dipoles cluster together and form a set depending on their easy direction, which is called a domain. These domains are separated by a thin wall, called a domain wall. Instead of single dipoles, conventionally domain walls are described. Their movement has a significant contribution to ferroelectric properties. The domain wall switching in ferroelectrics results in the ferroelectric hysteresis loops as shown in Figure 2.8. As the field is increased, the polarization of domains starts to switch in the direction of the applied field. This switching continues to saturate all polarization of domains. After the saturation, if the electric field starts to decrease, some of the domains recover their original state. However, even though the electric field reduces to zero, some of polarization does not recover to their original state. This is called remnant polarization (Pr). A further decrease of the electric field into negative directions results in a new alignment of the polarization. The necessary field to bring the polarization to zero value
26 is the coercive field, (Ec). Ultimately, the polarization will saturate again. Then the filed is reduced to zero, which completes the hysteresis loop.
Polarization (µC/cm²)
Ps Pr
-Ec
+Ec
Electric field (kV/cm) Figure 2.8. Ferroelectric Polarization-Electric field hysteresis data for 4µm PMN-PT film grown on (100) SrTiO3. . All ferroelectric materials are pyroelectric but only some pyroelectric materials, those in which an external field may switch the polarization, are ferroelectric. Moreover, all pyroelectric materials are piezoelectric, but only some piezoelectric materials, those whose symmetry belongs to polar groups, are pyroelectric. The relationships between these materials are shown in Figure 2.9. The typical physical properties of Piezoelectric materials are shown in Table 2.2.
ferroelectric
pyroelectric
piezoelectric
Figure 2.9. Relationship among ferroelectric, pyroelectric and piezoelectric.
27 Table 2.2. Typical dielectric constants, piezoelectric coefficients and electromechanical coefficients for ceramic PZT and relaxor PMN-PT Pb(Zr0.53Ti0.48)O3*
.67(PbMg1/3Nb2/3O3)-.33PbTiO3#
Polycrystalline
Single Crystal
386°C
150°C
Κ3T (free)
730
4000
Κ3S
400
d33 (pC/N)
223
d31 (pC/N)
-93.5
k33
0.67
Tc
1500
0.90 ~ 0.94
* Jaffe et al. [5], # Park et al. [11].
2.11. Relaxor Ferroelectric Materials
Relaxor ferroelectrics were discovered by Smolenskii et al [12] in 1958. The general formula of relaxors can be expressed as Pb(B1,B2)O3, where B1 is a low valence cation, such as Mg+2,Zr+2, Ni+2,Fe+3, or Sc+3, and B2 is a high valence cation, such as Nb+5, Ta+5, or W+6 [13]. These materials exhibit distinct relaxor properties from normal ferroelectric materials. In the dielectric response shown in Figure 2.10, the weak field dielectric permittivity exhibit diffuse phase transition near ferroelectric Curie temperature and the frequency dispersion of the dielectric permittivity below the temperature of the dielectric permitivity maximum Tm, where the temperature of maximum increases with frequency [14]. The associated maximum in the loss tangent (tan δ) is also typical of a relaxor response. Another characteristic of relaxor is shown in Figure 2.11. The
28 polarization vs. hysteresis loop decays more gradually to zero without abruptly loosing the spontaneous polarization at Curie temperature. The most surprising behavior is that there is no evidence of optical anisotropy or x-ray splitting at very low temperatures which would be characteristic of a macro-volume change to a polar phase. The major property differences which distinguish relaxor versus normal type ferroelectrics are summarized in Table 2.3.
Figure 2.10. Dielectric dispersion in Lead Magnesium Niobate (PMN) as a function of temperature and frequency [15].
Figure 2.11. Dielectric hysteresis in PMN as a function of temperature [15].
29 Table 2.3. Comparison of normal and relaxor ferroelectric materials [16] Property
Normal Ferroelectric
Relaxor Ferroelectric
Dielectric temperature Dependence
Sharp 1st or 2nd order transition at Curie point Tc
Broad diffuse phase transition at Curie maxima
Dielectric frequency Dependence
Weak Frequency Dependence
Strong frequency dependence
Dielectric Behavior in Paraelectric range ( T > Tc )
Follows Curie - Weiss law 1/K = C/(T-TC)
Remanent polarization (PR)
Strong PR
Scattering of light
Strong anisotropy
Diffraction of X-Rays
Line splitting due to deformation from paraelectric to ferroelectric phase
Follows Curie - Weiss square law 1/K = 1/Kmax + (T-Tmax)2/2Kmax δ2 Weak PR Very weak anisotropy to light (psuedo-cubic) No X-Ray line splitting (pseudo-cubic structure)
The most widely accepted model for understanding the relaxor ferroelectric phenomena is the nano-scale heterogeneity postulated by Smolenskii et al. [17] The reason for relaxor behavior is that the body-centered ions, Mg and Nb in Pb(Mg1/3Nb2/3)O3, do not order, so there are concentration fluctuations. This mixture of the ordered and disordered phases leads to a diffuse phase transition due to intimate mixture of ferroelectric and paraelectric regions. Figure 2.12 and 2.13 show the typical relaxor structure of PMN with the order and disorder structure of the B-site cations. The compositional ordering in relaxor is on the nanoscale (10 ~100nm). Ordered complex perovskites with long coherency ( ≥ 100 nm) show normal ferroelectric or antiferroelectric behavior rather than relaxor behavior [16]. In conclusion, it is expected
30 that sharp normal ferroelectric transition in the ordered structure and diffuse relaxor transition at disordered structure.
Pb Pb
Pb Pb
Pb Pb
Pb Pb
Mg
Nb
B-site cations
Figure 2.12. Structure of relaxor ferroelectric Pb(Mg1/3Nb2/3)O3.
Pb
Pb
Pb
Pb
Pb
Pb
Pb
Pb
Pb Pb
Pb
Pb
Pb
Pb
B-site cation disordered structure Mg
Pb Pb
B-site cation ordered structure Nb
B-site cations
Figure 2.13. B-site cations ordered structure and disordered structure.
31 2.12. Pb(B1,B2)O3-PT solid solution system
As we discussed above, the complex perovskite lead magnesium niobate (PMN) exhibits a diffuse phase transition near the curie temperature ~ 10°C and strong frequency dispersion in the dielectric permitivity which are typical behavior of a relaxor ferroelectric [18]. In solid solution with the simple perovskite PbTiO3 (PT) there is a morphotropic phase boundary (MPB), where the PT is a normal ferroelectric with large spontaneous polarization, small coercive field and large Curie temperature at 490°C [19]. The MPB separate the rhombohedral and tetragonal phases occurs at approximately 0.33 mol PbTiO3 Figure 2.14 shows the typical phase diagram of PMN-PT solid solutions.
250
Temperature (°C)
PMN-PT
Cubic
150
50
Rhombohedral (pseudocubic)
Tetragonal
- 50 0.0
0.1
0.2
0.3
0.4
0.5
Mole of PbTiO 3
Figure 2.14.
Phase diagram of Pb(Mg1/3Nb2/3)O3-PbTiO3 (PMN-PT) system near rhombohedral-tetragonal morphotropic phase boundary [20].
32 Usually, normal ferroelectric Pb(ZrxTi1-x)O3 (PZT) and relaxor ferroelectric Pb(Mg1/3Nb2/3)O3-PbTiO3
(PMN-PT)
and
Pb(Zn1/3Nb2/3)O3-PbTiO3
(PZN-PT)
compositions near the MPB exhibit anomalously high dielectric and piezoelectric properties, which make them good candidates for a wide range of MEMS device applications. According to Kuwata et al. [21], a large piezoelectric coefficient (d33 ~ 1600 pC/N) was found for PZN-PT with MPB compositions (9.5% PT). It should be noted that all the rhombohedral crystals oriented along their pseudocubic direction exhibit large piezoelectric coefficients, where the polarization direction is diagonal . As shown in Figure 2.15, d33 increases gradually with increasing amount of PbTiO3 in the oriented rhombohedral side until MPB compositions. The maximum d33 value was
~2500 pC/N for the oriented rhombohedral (0.92)PZN-(0.08)PT crystal. On the contrary, d33 value amorously decreases at MPB to the lower value ~500 pC/N for tetragonal composition. For the oriented (0.92)PZN-(0.08)PT crystal, the piezoelectric coefficient exhibits much lower than oriented crystal as shown in figure 2.15.
33
3000
2000 1500 Rhombohedral (pseudocubic)
1000
MPB
d33 (pC/N)
2500
Tetragonal
500
0
5
10
Pb(Zn1/3Nb2/3)O3
15
20 PbTiO3
Figure 2.15. d33 as a function of crystal composition and orientation [22].
Table 2.4 shows dielectric and piezoelectric properties for two rhombohedral crystals of PZN and (0.92)PZN-(0.08)PT with respect to the crystallographic orientation. For the oriented crystals, low dielectric loss values (< 1%) and high electromechanical coupling coefficient values ( > 90% ) have achieved, which are superior properties for transducer applications and high performance MEMS device applications. These oriented crystals exhibit ultrahigh longitudinal piezoelectric coefficients of ~1100pC/N for PZN and up to ~2500 pC/N for (0.92)PZN-(0.08)PT. However, oriented rhombohedral crystal have only ~38% of the electromechanical coefficient and ~80 pC/N of piezoelectric coefficients irrespective of the composition.
34 Table 2.4
Dielectric and piezoelectric properties as a function of crystallographic orientation for rhombohedral PZN and PZN-8%PT [11]
Crystal orientation
Composition Mode Coupling
111
PZN
K33
0.38
111
PZN-8%PT
K33
001
PZN
001
PZN-8%PT
S33E (10-2m2/N)
Dielectric constant
Loss
7.4
900
0.012
83
0.39
7.4
1000
0.012
84
K33
0.85
48
3600
0.008
1100
K33
0.94
130
5000
0.010
2500
d33 (pC/N)
The inferior piezoelectric properties of oriented rhombohedral crystals are related to domain instability. Since is the polar direction in the rhombohedral crystal, it is expected that complete poling will give a single domain and hysteresis-free piezoelectric strain behavior for oriented PZN crystals. However, as shown in Figure 2.16 and 2.17, there is remarkable hysteresis with enormously high strain ~ 0.1% which explains domain reorientation under bias. Complete single domain configuration through 109° and 71° domain reorientation may cause elastic energy in the crystals due to depoling after removing the field. On the contrary, oriented rhombohedral crystals were known to be stable. As shown in Figure 2.18, each domain of rhombohedral −
−
crystals have one of four possible polar directions < 111 >, < 111 >, < 111 >, and −−
< 111 >. Therefore, it is expected abrupt domain switching at 3kV/cm through 71° domain
switching. In Figure 2.16, poled PZN crystals exhibit hysteresis-free strain behavior, which is a consequence of domain stability. These result that piezoelectric properties depend of crystallographic dependence of piezoelectric is very useful for transducer applications.
35
Figure 2.16. Polarization and strain as a function of E-field (bipolar) curves for PZN crystals oriented along (a), (b) and (c), (d).
36
Figure 2.17 Strain as a function of E-field (unipolar) curves for PZN crystals oriented along (a) and (b).
Possible polarization directions of domains
E
E
Net polarization Figure 2.18. Domain configuration for poled rhombohedral crystals.
37
References [1]
Niels. Bohr, Philos. Mag. 26, 1 (1913).
[2]
HyperPhyscis, http://hyperphysics.phy-astr.gsu.edu/hbase/hframe.html.
[3]
D. Damjanovic, Rep. Prog. Phys. 61, 1267 (1998).
[4]
G.S. Kino, Acoustic Waves: Devices, Imaging, and Analog Signal Processing. (Prentice Hall, Englewood Cliffs, 1987).
[5]
Bernard Jaffe, William R. Jr. Cook, and Hans Jaffe, Piezoelectric Ceramics. (Academic Press, London, 1971).
[6]
M.E. Lines and A.M. Class, Principles and Applications of Ferroelectrics and Related Materials. (Clarendon Press, Oxford, 1997).
[7]
S. K. Streiffer, presented at the Panamerican Advanced Study Institute, Rosario, Argentina, 2002 (unpublished).
[8]
R.R. Das, University of Puerto Rico, 2003.
[9]
F. Jona and G. Shiraine, Ferroelectric Crystals. (Dover Publications, Inc., New York, 1993).
[10]
H.F. Kay and P Vousden, Philos. Mag., 1019 (1949).
[11]
S. E. Park and T. R. Shrout, IEEE Trans. Ultrason. Ferroelectr. Freq. Control 44, 1140 (1997).
[12]
G.A. Smolenskii and V.A. Isupov, Soviet Physics - Tech. Phys., 1380 (1958).
[13]
L. E. Cross, Ferroelectrics 151, 305 (1994).
[14]
Z. Kighelman, D. Damjanovic, A. Seifert, L. Sagalowicz, and N. Setter, Appl. Phys. Lett. 73, 2281 (1998).
[15]
L. E. Cross, Ferroelectrics 76, 241 (1987).
[16]
C. A. Randall, A. S. Bhalla, T. R. Shrout, and L. E. Cross, J. Mater. Res. 5, 829 (1990).
[17]
G.A. Smolenskii and V.A. Isupov, Soviet Physics - Solid State 1, 1429 (1959).
38 [18]
G.A. Smolenskii, V.A. Isupov, A.I. Agranovskaya, and S.N. Popov, Soviet Physics - Solid State 1 2, 2584 (1961).
[19]
M. de Keijser, G.J.M. Dormans, J.F.M. Cillessen, and D.M. de Leeuw, Appl. Phys. Lett. 58, 2536 (1991).
[20]
T. R. Shrout, Z. P. Chang, N. C. Kim, and S. Markgraf, Ferroelectr. Lett. Sect. 12, 63 (1990).
[21]
J. Kuwata, K. Uchino, and S. Nomura, Jpn. J. Appl. Phys. Part 1 - Regul. Pap. Short Notes Rev. Pap. 21, 1298 (1982).
[22]
S. E. Park and T. R. Shrout, J. Appl. Phys. 82, 1804 (1997).
39
Chapter 3. Synthesis and Characterization of Piezoelectric Thin Films 3.1. Growth Method for Piezoelectric Materials Several methods have been used to fabricate PMN-PT and PZT thin films including metal-organic chemical vapor deposition (MOCVD), pulsed laser deposition (PLD), and sputtering. Each growth method has its own advantages in the thin film fabrication. Several deposition methods are summarized in the next sections. Among the various growth techniques, magnetron sputtering is well suited for the fabrication of PMN-PT films because of its reproducibility, controllability and ease of extending results obtained empirically on a small, research-scale sputter chamber to a highly reliable production process. Hence we selected magnetron sputtering to grow epitaxial PMN-PT thick films. Several deposition methods are summarized in the next sections. To grow epitaxial PMN-PT films, we have used 5 mol% PbO excess to stoichiometric PMN-PT target. Even though off-axis sputtering gives broader area homogeneity, we decided to use on-axis sputtering because of its fast deposition rate to grow thick films. However, we have used off-axis geometry to fabricate SrRuO3 bottom electrodes which are shown in Figure 3.1. The important factors to grow high quality films are temperature, total pressure, Ar : O2 gas ratio and sample positions. We have decided carefully the optimum conditions experimentally.
40
Piezoelectric
On-axis sputtering
Piezoelectric
1000 Å SrRuO3
Off-axis sputtering
1000 Å SrRuO3
100 Å SrTiO3
MBE growth (Penn. State Univ.)
(001) Si
(001) SrTiO3
Figure 3.1. Thin film heterostructures and deposition methods.
3.1.1. Sputtering Sputtering is one of the most well developed and versatile thin film deposition techniques. It was developed by Grove (1852) and Plücker (1858) who first reported vaporization and film formation of metal films by sputtering [1]. The sputtering process is essentially a kinetic process which involves momentum exchange between the energized Ar+ ion and sputter target atoms, rather than a thermal process. When a solid surface is bombarded with accelerated ions, the surface atoms of the solid are scattered backward due to collisions between the surface atoms and the accelerated ions as shown in Figure 3.2. Typically, the target is connected to a negative voltage supply, which can be DC or RF. A gas is introduced which provides a medium for glow discharge. When the glow discharge is started, Ar+ ions hit the target surface and kick the neutral atoms out by momentum transfer. The sputtered atoms delivered to the surface of the substrates. Figure 3.3 shows a comparison of on-axis and off-axis geometries. Under normal on-axis geometric sputtering, severe back sputtering of the substrate by O2- ions causes changes of stoichiometric composition of the thin film [2]. The back-sputtering can be overcome to some degree by using sputtering at high pressure which will scatter the O2-
41 ions before arrival at the substrate surface and by using a non-stoichiometric target which will compensate the selective back-sputtering effect.
Sputtered Species
Incident Ion
Target Surface Target Ion
Figure 3.2. Physical sputtering process.
Conventionally, metals were fabricated with on-axis sputtering in the range of 3 10 mTorr. For the growth of perovskite oxide thin films, the oxides are not completely formed at this low pressure, where the energetic species emitted from the target do not suffer enough scattering with the background gas and hence they bombard the surface of the growing thin film without reducing their high energy. Thus, on-axis sputtering for epitaxial oxide thin film fabrication has been carried out in the range of high pressures 400 ~ 600 mTorr. Even though on-axis geometry provides higher deposition rate, its uniformity is low. If the target-to-substrate distance is increased, the thickness uniformity increases, but deposition rate decreases. The back-sputtering problem can be overcome by 90° off-axis geometry sputtering, in which thin films have almost the same composition as the target over large areas. This off-axis geometry gives excellent composition and thickness uniformity with
42 better step coverage and an extremely smooth surface. One drawback of off-axis geometry is the low deposition rate.
Sputter Gun
Target
Substrate Heater Block
On-axis Sputtering Geometry
Target
Substrate Sputter gun
H eater Block
O ff-axis Sputtering G eom etry
Figure 3.3. Sputtering system geometry.
43 Here we adopted on-axis geometry sputtering because we would like to form films over a micron thickness . Figure 3.4 shows a of sputtering system which were used to grow SrRuO3 bottom electrode and ferroelectric films, where SrRuO3 were grown by off-axis geometry and ferroelectric films were grown by on-axis geometry.
Figure 3.4. Photograph of the planar magnetron sputtering chamber.
3.1.2. Pulsed Lased Deposition Pulsed-laser-deposition (PLD) is a very suitable physical vapor deposition technique which is based on the evaporation of materials by a focused beam of an excimer laser such as KrF, ArF, XeCl from a stoichiometric target [3]. PLD provides several advantages over other thin-film deposition techniques [4]. The schematic diagram of PLD technique is shown in Figure 3.5. One of the important advantage is that a wide variety of materials from binary to multicompoment oxides and nitrides can be
44 successfully grown on metal and oxide substrates using PLD technique. Artificial superlattices can be easily grown by using multiple targets without breaking the vacuum. The high deposition rate and the simplicity and flexibility of PLD are also noteworthy. The disadvantage is the small deposition area on the substrate and particulate formation during ablation, which can be overcome by rastering the target and the substrate during deposition of thin films [5].
Figure 3.5. Schematic drawing of PLD deposition chamber with RHEED [3].
High pressure reflection high energy electron diffraction (RHEED) has been employed for PLD to monitor in-situ epitaxial growth of the thin films during the film growth. Electrons generated by an electron gun are typically 10 ~ 30 KeV with low divergence [6], which are detected on a phosphorescent screen as a reciprocal diffraction pattern. The electron gun and the screen are positioned remotely from the substrate. Therefore the RHEED system does not interfere with the growth process and is not exposed directly to the atomic sources and surface of the heater. Furthermore, the
45 incident electron beam strikes the sample at a grazing angle of ~ 1° which makes RHEED a surface-sensitive technique [7].
3.2. Bottom Electrode For high quality electronic heterostructures or microelectromechanical system (MEMS) applications, we selected SrRuO3 as the bottom electrode. SrRuO3 has a GdFeO3-type pseudocubic perovskite structure with a bulk lattice parameter of 3.93Å. The lattice parameters of the orthorhombic structure are a=5.53Å, b=5.57Å and c=7.85Å. as shown Figure 3.6 [8,9]. The lattice mismatch with PMN-PT is fairly small (2.3%), which allows us to grow high quality epitaxial heterostructures. SrRuO3 is stable at high temperature up to 1200 K in severe conditions. The resistivity of SrRuO3 film is ~340 µΩ·cm at room temperature and essentially isotropic. Additionally, SrRuO3 can be used as a bottom electrode for oxide thin films because of its good metallic behavior. According to the characterization of SrRuO3 films by XRD and TEM [10-12], they exhibit pure {110} texture normal to the substrate. From these considerations, we selected SrRuO3 as an ideal electrode material for epitaxial ferroelectric heterostructures.
46 c (7.8446 Å)
[001]o
b (5.5304 Å)
a
[110]o
(5.5670 Å)
O(2)
O(1)
Sr
Ru
Figure 3.6. The orthorhombic SrRuO3 structure ( Eom et al.).
3.3. Fabrication of Bottom Electrode and Piezoelectric Thin Films The thin films were deposited using Radio-Frequency (RF) magnetron sputtering. The sputter gun was mounted in a vacuum chamber equipped with turbo-molecular pump which is backed up with mechanical pump. The base pressure was 2×10-6 Torr without bake-out. The SrRuO3 thin films were deposited from a 2 in. diameter stoichiometric, composite target using the 90° off-axis sputtering technique on (001) SrTiO3 and (100Å) SrTiO3 / (001) Si substrates. During the deposition, the operating pressure was held at 200 mTorr and the gas input ratio of Ar vs. O2 was 60% vs. 40%. The substrate heater block temperature was in the range of 600 ~ 680 °C. After deposition, the samples were cooled to room temperature in an oxygen pressure of 300 Torr. Film thicknesses
47 were in the range of 1000Å ~ 2000Å. With these growth conditions, we can easily grow (110) oriented SrRuO3 thin films on (001) SrTiO3. Piezoelectric PZT and PMN-PT epitaxial films were deposited on a SrRuO3 bottom electrode with SrTiO3 and Si substrates. In the case of SrTiO3, we used several different substrate orientations, including (100), (111) and (110) to understand the orientation dependence of
the
piezoelectric characteristics. We used 0° and 4° miscut (001) SrTiO3 and (001) Si substrate to understand the miscut dependent structural and electrical properties. The substrates were heated by fixing them to a stainless steel block-style hot stage using silver paint which increases heat conductivity. The temperatures given refer to the readings taken from a K-type thermocouple embedded in the heater block. We used sintered targets with Pb(Zr0.52Ti0.48)O3 and 0.67PMN-0.33PT and excess PbO due to its high volatility. The quantities of excess PbO were changed from 0% to 15%. Two inch diameter sputtered targets were used to grow thin films on various substrates. Thickness dependant structural and piezoelectric properties of the PZT and PMN-PT films were studied by growing the films for various time intervals at similar deposition environments. Because of its thickness variation, we used two-gun on-axis geometry together with onegun geometry. The distance between the two guns is 3 ~ 4 inch. The target-to-substrate distance was 1 ~ 1.5 inch, with an operating pressure of 400 mTorr. The substrate temperatures were 600 ~ 800°C. The post deposition cooling method was the same as in the SrRuO3 deposition.
48
3.4. X-ray diffraction The structural orientation and lattice parameters of the crystalline materials can be studied using x-ray diffraction and it provides a non-destructive probe of the samples. The detailed structural analysis of thin films and single crystals can be done by a fourcircle x-ray diffractometer which gives lattice parameters, grain size, presence of second phases, and domain information. Figure 3.7 shows the geometry of the four-circle diffractometer.
Figure 3.7.
Schematic illustration of the geometry and angular relationship for fourcircle x-ray diffractometer (courtesy of Schlom [6]).
49 The crystalline quality of thin films can be determined by measuring full width at half maximum (FWHM) of rocking curve as shown in Figure 3.8. The width of the rocking curve is a direct measurement of non ideality of the crystal quality. As each misaligned subgrain of a typical mosaic crystal continuously comes into Bragg reflection the intensity of the diffracted spot changes as the crystal is rocked [13].
D
S
Sample
Figure 3.8. Schematic geometry of rocking curve measurement. S= x-ray source, D=Detector. Normal θ-2θ scans can identify phases and crystallographic orientations of the film’s out-of-plane. Furthermore, θ-2θ scans give the in-plane and out-of-plane lattice parameters, where in-plane lattice parameter can be determined by azimuthal off-axis reflections. The in-plane film textures of films can be determined by off-axis azimuthal φ-scans of off-axis peaks. With well aligned samples and all of the off-axis peaks of the
phi scan, in-plane epitaxial relations are established. The full width at half maximum (FWHM) of peaks in phi-scan also gives information of the twist misalignment of epitaxial grain in the plane. The schematic geometry of off-axis phi-scan is shown in Figure 3.9.
50 Detector
2θ
θ
χ
q
Crystallograph ic Plane s
In cident x-rays
Figure 3.9
ϕ
χ 2000 pm/V) [6]. Non-piezoelectric pyrochlore phases dominate at the larger film thicknesses of most interest for piezoelectric applications, and significantly reduce the piezoelectric responses [7]. We have fabricated epitaxial PMN-PT thick films on both (001) Si and (001) SrTiO3 to study lattice and thermal expansion mismatch effects on piezoelectric properties. We also have studied the reasons for inferior electric and piezoelectric properties of the thin films than bulk single crystals by reducing the substrate constraint with FIB milling nano-structuring.
57
4.2. Experimental method The first critical aspect of our approach is the use of (001) Si wafers with an epitaxial SrTiO3 template layer as the substrate. The SrTiO3 layer is deposited by molecular beam epitaxy (MBE) using a process described elsewhere [8], followed the deposition of a conducting SrRuO3 bottom electrode. Finally, the 0.67Pb(Mg1/3Nb2/3)O30.33PbTiO3 films are deposited by on-axis RF-magnetron sputtering, in a thickness range of 1 to 4 µm. During film deposition, the substrate temperature is maintained at 670oC with argon and oxygen pressures of 240 mTorr, and 160 mTorr. Chemical measurements show that all the films maintained the target stoichiometry [7]. SrRuO3 top electrodes were deposited by pulsed-laser deposition (PLD) at lower temperature through shadow masks. To relieve the effects of substrate-induced constraint on the piezo-response, the films were patterned by focused ion beam (FIB) milling down to the bottom electrode [9], thus yielding capacitors with lateral dimensions in the 0.5-3 µm range.
4.3. Results and discussion 4.3.1. Miscut and thickness effect on the formation of PMN-PT perovskite and pyrochlore phases As discussed previously, many researchers have been interested in the growth of high-quality single crystal epitaxial PMN-PT films for electromechanical device applications due to their excellent piezoelectric properties. However, growth of pure perovskite epitaxial PMN-PT films is known to be difficult due to the appearance of stable pyrochlore phases which are cubic pyrochlore Pb3Nb4O13 and rhombohedral pyrochlore Pb2Nb2O7 [10]. The pyrochlore phases are severely detrimental to the piezoelectric properties. According to Bu et al. [11], highly pure epitaxial PMN-PT thin
58 films can be fabricated up to 0.5 µm thick by deposition onto miscut (001) SrTiO3 single crystal substrates. Films grown on high miscut (>4°) SrTiO3 substrates showed the pure perovskite phase. In contrast, films grown on exact (001) SrTiO3 substrates contained a higher content of pyrochlore phases with lead deficiency. In order to grow much thicker films over a few µm with pure perovskite phase, we employed the same approach. Figure 4.1, shows the 2-D area detector images of x-ray diffractometer for a ~4 µm thick PMNPT films on exact (001) SrTiO3 and miscut (4°) SrTiO3. It is clear that the low miscut substrate gives higher contents of pyrochlore phases. The thick PMN-PT films on exact (001) SrTiO3 substrates exhibit highly textured pyrochlore phases and polycrystalline powder pattern. However, PMN-PT films on miscut (4°) SrTiO3 show nearly pure epitaxial perovskite phases with strong 00l reflections. This is also pronounced with the PMN-PT films grown on silicon substrates. We believe the formation of pyrochlore phases on low miscut substrate is due to lead volatility and longer terrace length. In the case of a high miscut substrate, the terrace length is shorter than exact (001) SrTiO3 substrates which are shown in Figure 4.2. When atoms arrive on the surface of the substrate, each atom moves to a kink site to lower its energy. At the initial stage of the deposition, lead has a much higher probability to evaporate before the atoms reach the kink sites due to its high volatility, and leaddeficient pyrochlore phase can be formed. Once pyrochlore phases are nucleated, it is easier to grow pyrochlore phases because they are more stable than perovskite phases. The contents of the pyrochlore phases will increase as thickness increases. Therefore, the shorter terrace length of higher miscut substrate will suppress the formation of pyrochlore phases.
59
(001) PMN-PT
Pyrochlore
(002) PMN-PT
(a) (101) PMN-PT
(001) PMN-PT
(002) PMN-PT
(b)
(111) PMN-PT
Figure 4.1. 2-D area detector diffraction images of (a) PMN-PT films on exact (001) SrTiO3 which shows a high content of textured and polycrystalline pyrochlore phases (b) PMN-PT films on 4° miscut (001) SrTiO3 which shows nearly pure perovskite phases.
Kink
Ledge
Terrace
Figure 4.2. A schematic of surface of a miscut substrate. Adatoms move to kink site because it is more stable. According to reports, it is not easy to grow thick films over a micron due to the pyrochlore phase formation during the thin film growth. We have successfully fabricated epitaxial 4µm PMN-PT thick films on (001) SrTiO3 and (001) Si. We have fabricated 6
60 µm think PMN-PT films on (001) SrTiO3 to see the thickness effect on the formation of pyrochlore phase. Figure 4.3 shows the x-ray area detector image of 4 µm and 6 µm PMN-PT thick films. It shows that 6 µm thick PMN-PT films have higher contents of pyrochlore phases. The 4 µm PMN-PT films on 4° miscut SrTiO3 show small contents of pyrochlore phases. It is clear that the contents of pyrochlore phases increase as thickness increases.
(001) PMN-PT
(002) PMN-PT
(001) PMN-PT
(a)
Pyrochlore
(002) PMN-PT
(b) (101) PMN-PT
(111) PMN-PT
Figure 4.3. (a) 4 µm thick PMN-PT films on 4° miscut (001) SrTiO3 which shows nearly pure perovskite phases. (b) 6 µm thick PMN-PT films on 4° miscut (001) SrTiO3 which shows a high content of pyrochlore phases and polycrystalline phases.
61
4.3.2. Phase Purity and Microstructure The precise lattice parameter, crystal quality and epitaxial arrangements were studied using a four circle x-ray diffractometer with high resolution x-ray diffraction (HR-HRD) with a four-bounce monochromator. The θ-2θ scans in Figure 4.4 (a) and (b) show the strong (00l) peaks from perovskite PMN-PT phase in 3.5 µm thick films grown on 4° miscut (001) Si and (001) SrTiO3 substrates. We observed the formation of pyrochlore phases above a film thickness of 4µm. The full width at half maximum (FWHM) of the rocking curve for the PMN-PT (002) reflection are 0.26° for the PMNPT on (001) SrTiO3 and 0.29o for the PMN-PT on (001) Si. The FWHM of the off-axis φ scan of PMN-PT (101) reflections are 0.4° for PMN-PT on SrTiO3 and 0.7° for the PMNPT on (001) Si. This confirms the high crystalline quality of the films as shown in Figure 4.5. As expected, azimuthal φ-scans, show in-plane epitaxy with a cube-on-cube epitaxial
100
220 SrRuO3
(b) 002 SrTiO3
c = 4.032 ± 0.001Å a = 4.000 ± 0.003Å
002 PMN-PT
001 SrTiO3
Intensity (cps)
102
104
001 PMN-PT
106
(a) 220 SrRuO3
104
c = 3.998 ± 0.002Å a = 4.027 ± 0.002Å
002 PMN-PT
Intensity (cps)
106
001 PMN-PT
relationship, [100] PMN-PT//[110][001] SrRuO3//[100] SrTiO3 /[110] Si.
102
100 20
30
40
2θ (degrees)
50
20
30
40
50
2θ (degrees)
Figure 4.4. (a) X-ray θ-2θ diffraction spectra of epitaxial PMN-PT (3.5 µm thick) grown on a SrRuO3 thin film grown on a SrTiO3 buffered (001) silicon substrate. (b) X-ray θ-2θ diffraction spectra of epitaxial PMN-PT (3.5 µm thick) grown on a SrRuO3 thin film on a bulk SrTiO3 substrate.
Intensity (a.u.)
FWHM=0.29°
(a)
Intensity (a.u.)
62
Intensity (a.u.) 0
90
180
270
φ (degrees)
360
θ (degrees)
(d)
FWHM=0.4°
Intensity (a.u.)
θ (degrees)
(b)
(c)
21.5 22.0 22.5 23.0 23.5
21.5 22.0 22.5 23.0 23.5
FWHM=0.7°
FWHM=0.26°
0
90
180
270
φ (degrees)
360
Figure 4.5. (a) Rocking curve of 002 PMN-PT on (001) Si. (b) φ-scan of 101 PMN-PT on (001) Si. (c) Rocking curve of 002 PMN-PT on (001) SrTiO3. (d) φscan of 101 PMN-PT on (001) SrTiO3. Figure 4.6 compares the out-of-plane and in-plane lattice parameters of the 3.5
µm thick films on both Si and SrTiO3 substrates. We find that the film on Si is under tension due to the thermal expansion mismatch of PMN-PT with Si, with an in-plane lattice parameter of 4.027Å and an out-of-plane lattice parameter of 3.998Å, compared to the pseudocubic bulk lattice parameter of 4.02Å. On the other hand, the films on bulk SrTiO3 show the opposite behavior. The x-ray diffraction results in Figure 4.3 (a) and (b) indicate a clear peak shift towards lower angles (or bigger lattice parameters) for the film on bulk SrTiO3 compared to Si, with out-of-plane lattice parameter of 4.032 ± 0.001 Å and in-plane lattice parameter of 4.000 ± 0.003 Å. The influence of this remnant stress is further discussed when the ferroelectric and piezoelectric properties are described.
63
Lattice parameter ( Å)
4.04 4.03 4.02
Bulk PMN-PT
4.01 4.00
In-plane Out -of-plane
3.99
Si
SrTiO 3 Substrates
Figure 4.6. A comparison of the in-plane and out-of-plane lattice parameters of the PMN-PT films grown on SrTiO3 and SrTiO3/Si, illustrating the different stress states experienced by the films on the two substrates. As a reference, the pseudocubic lattice parameter of PMN-PT of a similar composition is also given. Transmission electron microscopy (TEM) studies were performed to obtain direct proof of epitaxial growth of the PMN-PT on Si. Figure 4.7 (a) is a low magnification bright-field TEM image of a 1µm thick PMN-PT/SrRuO3/SrTiO3/Si heterostructures. Figures 4.7 (b), (c), and (d) are the selected-area electron diffraction (SAED) patterns taken from the SrTiO3 as well as the underlying Si substrate, SrRuO3, and PMN-PT layers, respectively. They were identified as the superimposition of the [010] zone axis diffraction pattern of SrTiO3 and the [110] zone axis diffraction pattern of Si, the superimposition of the [001] zone axis and [110] zone axis diffraction patterns of SrRuO3, and the [010] zone axis diffraction pattern of PMN-PT. The epitaxial growth of PMN-PT is evident. The TEM observations confirm the high crystalline quality of the PMN-PT, which is consistent with the x-ray diffraction.
64
PMN-PT
(a)
SrRuO3 SrTiO3 Si
200 nm
(b)
(c)
(d)
002
002
[010]SrTiO3 +[110]Si [001]+[110]SrRuO3
[010]PMN-PT
Figure 4.7 (a) Bright-field cross-sectional TEM image of 1 µm thick PMN-PT / SrRuO3 thin film grown on a SrTiO3 buffered (001) Si substrate, (b) SAED image of SrTiO3 along the [100] zone axis, (c) SAED image of SrRuO3 along the [110] zone axis, and (d) SAED of PMN-PT along the [100] zone axis
65
4.3.3. Microstructure of Pyrochlore Phase by TEM We analyzed the microstructure the 4 µm thick PMN-PT films by Transmission Electron Microscopy to analyze the pyrochlore structure. The low magnification image of the layered structures in Figure 4.8 shows some porous morphology. These regions inside PMN-PT films are sensitive to Ar+ ion milling and will become amorphous or even void by sputtering after the TEM sample preparation, which indicates the existence of a second phase in the PMN-PT films and are believed to be pyrochlore phases.
Voids PMN-PT
SrRuO3 SrTiO3 Si
Figure 4.8. TEM image of 4 µm PMN-PT films on (001) Si. There are some voids in PMN-PT films which are believed to pyrochlore phases.
We also found a columnar structure with grain boundary regions. Figure 4.9 shows grain boundaries in the PMN-PT films. HR-TEM image in Figure 4.10 shows microstructures of secondary phase. According to EDS spectra, the second phase S1 is non-stoichiometric compared with S2 phase which is PMN-PT stoichiometric
66 compositions. The Plan view TEM image in Figure 4.10 (b) also shows clearly the different microstructure of secondary phases. The exact mechanism of the formation of pyrochlore phases is not yet clear. Composition instability will increase the possibility of nucleation of pyrochlore and the lamellar structure of the growing films will be separated by the grain boundary which is believed as the pyrochlore phase. Finally, the surface roughness of growing films will promote the nucleation of pyrochlore phases.
Grain Boundary
PMN-PT
(a)
300 nm PMN-PT
(b) Figure 4.9.
PMN-PT
(c) TEM image of 4 µm PMN-PT films on (001) Si which shows grain boundaries.
67
(a)
(b)
(PMN-PT)pero
s1
s2 (PMN-PT)pyro
4 nm
Figure 4.10. TEM image of 4 µm PMN-PT films on (001) Si. There grain boundaries are believed to pyrochlore phases.
4.3.4. Ferroelectric and Piezoelectric Properties The films were cut by focused ion beam (FIB) milling and their properties were compared to continuous films as shown in Figure 4.11.
Figure 4.12 show the
piezoelectric and ferroelectric measurements of the 3.5 micron thick films on both Si and SrTiO3. The polarization hysteresis loops were measured using a Radiant Technologies RT 6000 tester and an Aixacct TF2000 analyzer. Figure 4.12 (a) plots the P-E loop measured for the film on Si while Figure 4.12 (b) is a plot of the P-E hysteresis loop for a film on SrTiO3. We observe that the P-E loops for continuous films (Pr from 5 to 8 µC/cm2), is strongly tilted and is not saturated. This is a consequence of the tensile strain
imposed by the Si substrate, as evident from the x-ray data, and is consistent with previous reports of low remnant polarizations in random and oriented PMN-PT films on Si [12]. In contrast, films on SrTiO3 show much squarer behavior with remnant polarizations of ~ 22µC/cm2 (again consistent with the effect of compressive stresses). In
68 direct measurements of the properties of PMN-PT films on LaNiO3/Si, it has been shown that when a biaxial tensile stress is applied via flexure of the substrate, the hysteresis loop rotated clockwise, resulting in lower remnant polarizations. Compressive stress resulted in a counterclockwise rotation, increasing the measured remnant polarization [13]. The changes are often large enough to suggest that it may be possible to induce the tetragonal phase (with the polarization in the plane) in films under large tensile stresses. Interestingly, when the film on Si is laterally subdivided by FIB as shown in Figure 4.12 (a), the hysteresis loop recovers to a shape comparable to that of the epitaxial film on SrTiO3 (Pr ~ 25-30 µC/cm2).
This is a consequence of the removal of the biaxial
constraint on the film which alters the electromechanical boundary conditions and hence the ferroelectric behavior.
Piezoelectric Electrode Substrate Continuous film
Cut island film
Figure 4.11. Piezoelectric coefficient measurement with a) continuous PMN-PT film b) cut-island film milled by FIB
69 40
4µm * 4µm Cut capacitor
(a)
Continuous film
20 0 -20 -40
PMN-PT on Si
-60 -80 -60 -40 -20
0
20
40
60
60
Polarization (µC/cm2)
Polarization (µC/cm2)
60
20 0 -20 -40
80
-80 -60 -40 -20
4µm * 4µm Cut capacitor Continuous film
1800
(c)
1200
600 0 -600 -1200 PMN-PT on Si 0
20
Field (kV/cm)
20
40
4µm * 4µm Cut capacitor Continuous film
60
80
40
60
80
(d)
600 0 -600 -1200
-1800 -80 -60 -40 -20
0
Field (kV/cm)
d33 (pm/V)
d33 (pm/V)
1200
PMN-PT on SrTiO3
-60
Field (kV/cm)
1800
(b)
40
PMN-PT on SrTiO3
-1800 -80 -60 -40 -20
0
20
40
60
80
Field (kV/cm)
Figure 4.12. (a) Polarization vs. electric field 3.5 µm thick PMN-PT films for both continuous and nanostructured capacitor on STO/Si. (b) Polarization vs. electric field for 3.5 µm thick PMN-PT film for a continuous capacitor on SrTiO3. (c) d33 vs. electric field for 3.5 µm thick PMN-PT film for continuous and separated capacitors on SrTiO3/Si. (d) d33 vs. electric field for 3.5 µm thick PMN-PT film for continuous and separated capacitors on SrTiO3. Further evidence of this is observed in the piezoelectric measurements. The details of the experimental procedure and quantitative measurements of the piezoelectric coefficients are presented elsewhere [14]. Figure 4.12 (c) shows the longitudinal(d33,f) piezoelectric coefficients for a continuous (clamped) film capacitor and a milled 4µm x 4 µm island for the film on Si measured by piezoresponse microscopy. For the continuous
film capacitor the maximum d33 is approximately 800 pm/V. When measured after milling, the d33,f increases to 1200 pm/V under a DC bias. This is much higher than values reported for PMN-PT films, and is consistent with the release of lateral constraints
70 on the film. Furthermore, the cut capacitors exhibit a stronger dependence on the applied field compared to the continuous capacitor, similar to previous results on soft PZT compositions [15]. For the film on SrTiO3, FIB milling increases the d33 from 400pm/V to 600 pm/V. This large difference in the piezoelectric responses between the islands on Si and SrTiO3 might be due either to a change in the degree of clamping imposed by the substrate, or by differences in the residual stress values.
71
References [1]
S. E. Park and T. R. Shrout, J. Appl. Phys. 82, 1804 (1997).
[2]
D. Lavric, R. A. Rao, Q. Gan, J. J. Krajewski, and C. B. Eom, Integr. Ferroelectr. 21, 499 (1998).
[3]
J. P. Maria, W. Hackenberger, and S. Trolier-McKinstry, J. Appl. Phys. 84, 5147 (1998).
[4]
V. Nagarajan, S. P. Alpay, C. S. Ganpule, B. K. Nagaraj, S. Aggarwal, E. D. Williams, A. L. Roytburd, and R. Ramesh, Appl. Phys. Lett. 77, 438 (2000).
[5]
J. H. Park, F. Xu, and S. Trolier-McKinstry, J. Appl. Phys. 89, 568 (2001).
[6]
S. E. Park and T. R. Shrout, IEEE Trans. Ultrason. Ferroelectr. Freq. Control 44, 1140 (1997). S. D. Bu, D. M. Kim, C. B. Eom, S. K. Streiffer, W. Tian, X. Q. Pan, Yoshimura T., and S. Trolier-McKinstry, submitted to Appl. Phys. Lett. (2002).
[7] [8]
G. Y. Yang, J. M. Finder, J. Wang, Z. L. Wang, Z. Yu, J. Ramdani, R. Droopad, K. W. Eisenbeiser, and R. Ramesh, J. Mater. Res. 17, 204 (2002).
[9]
V. Nagarajan, A. Roytburd, A. Stanishevsky, S. Prasertchoung, T. Zhao, L. Chen, J. Melngailis, O. Auciello, and R. Ramesh, Nature Materials 2, 43 (2003).
[10]
S. L. Swartz and T. R. Shrout, Mater. Res. Bull. 17, 1245 (1982).
[11]
S. D. Bu, M. K. Lee, C. B. Eom, W. Tian, X. Q. Pan, S. K. Streiffer, and J. J. Krajewski, Appl. Phys. Lett. 79, 3482 (2001).
[12]
L. F. Francis and D. A. Payne, J. Am. Ceram. Soc. 74, 3000 (1991).
[13]
Z Zhang, J-H. Park, and S. Trolier-McKinstry, MRS. Proc. Ferroelectric Thin Films VIII 596, 73 (2000).
[14]
C. S. Ganpule, A. Stanishevsky, S. Aggarwal, J. Melngailis, E. Williams, R. Ramesh, V. Joshi, and C. P. de Araujo, Appl. Phys. Lett. 75, 3874 (1999).
[15]
V. Nagarajan, A. Stanishevsky, L. Chen, T. Zhao, B. T. Liu, J. Melngailis, A. L. Roytburd, R. Ramesh, J. Finder, Z. Yu, R. Droopad, and K. Eisenbeiser, Appl. Phys. Lett. 81, 4215 (2002).
72
Chapter 5. Piezoelectric response in epitaxial 0.67[Pb(Mg1/3Nb2/3)O3]0.33[PbTiO3] thick films on (001) SrTiO3 substrates
5.1. Introduction Single crystal relaxor ferroelectrics, such as Pb(Mg1/3Nb2/3)O3-PbTiO3 (PMN-PT) and Pb(Zn1/3Nb2/3)O3-PbTiO3 (PZN-PT), yield significantly higher piezoelectric constants and electromechanical coupling coefficients than polycrystalline ferroelectrics [1,2]. A major challenge is to prepare these materials as thick epitaxial heterostructures and
integrate
them
into
high
frequency
ultrasound
transducers
and
microelectromechanical systems. This would allow these superior properties to be utilized with all the advantages associated with microelectronics technology. Epitaxial PMN-PT thin films have been grown on a number of substrates, including LaAlO3 [3], SrTiO3 [4,5], MgO [6], and buffered Si [7]. Growth of epitaxial PMN-PT films is difficult due to the relatively poor thermodynamic stability of the perovskite phase relative to the pyrochlore phase and due to compositional complexity [3-5]. However, epitaxial films reported to date show appreciably squarer hysteresis loops and larger values of the remnant polarization (~15 to 20 µC/cm2 for PMN-(3035%)PT) than polycrystalline films on Si. In terms of piezoelectric response, PMN-PT films near the morphotropic phase boundary have been reported to have transverse piezoelectric coefficients e31,f ~ –9.5 C/m2 and a low field longitudinal piezoelectric coefficient d33,f of 50 to 200 pC/N [6,8], which are still substantially smaller than the corresponding single-crystal values. The differences between thin film and bulk crystal properties have been attributed to strain, substrate constraint, phase purity, and
73 nonstoichiometry. As e31,f is the most useful electromechanical coefficient for film-based devices, it is particularly desirable to explore routes to improve this value. Specifically, we have chosen to synthesize thick (≥1 µm) films, in which strain and the substrate constraint are expected to be partially relaxed, and extrinsic contributions to the piezoelectric behavior to be increased.
5.2. Experimental method It was recently demonstrated that high quality epitaxial PMN-PT thin films (0.5 µm thick) can be grown on miscut (001) SrTiO3 substrate using 90o off-axis magnetron
sputtering [5]. The PMN-PT thin films on high miscut substrates (>4o) were almost pure perovskite phase with nearly stoichiometric composition [5]. However, one of the drawbacks of off-axis sputtering is a slow growth rate and therefore, it is not ideal for thick film growth. In this paper, we report the growth of high quality epitaxial thick films of PMN-PT with superior piezoelectric properties using on-axis rf magnetron sputtering. Films with thicknesses from 1 to 15 µm were grown on 4° miscut (001) SrTiO3 substrates and on single crystal SrRuO3 (SRO) bottom electrode layers on 4° miscut (001) SrTiO3 [9]. A sintered target with stoichiometry 0.67PMN-0.33PT and 1.4% excess MgO was used for PMN-PT deposition. The substrate temperature was held at 650oC and the sputtering atmosphere consisted of 240 mTorr Ar and 160 mTorr O2. The deposition rate was 0.5 µm/h. Film composition was determined by wavelength-dispersive x-ray fluorescence spectroscopy, which showed a slight excess of Mg (1.2 ± 0.9 %) with a Ti/(Mg+Nb+Ti) ratio of 0.33 ± 0.01 and a Pb/(Mg+Nb+Ti) ratio of 1.05 ± 0.01, very close to the target composition.
74
5.3. Results and Discussion Figure 5.1 shows the x-ray diffraction theta-twotheta scans for 4.5 and 7.5 µm films on bare SrTiO3. Strong (00l) peaks from the PMN-PT perovskite phase are clearly visible. Azimuthal φ scans demonstrate in-plane epitaxy of the film and the substrate, with a cube-on-cube epitaxial relationship. Spectra also show low intensity second phase peaks; their intensities increase as the film thickness increases. The second phase peaks are tentatively identified as pyrochlore phases [10]. The full width at half maximum (FWHM) of the rocking curve for the PMN-PT (002) reflection is 0.30o for the 4.5 µm
3
(001)SrTiO
(002)
9.0 µm
103 102
4.5 µm
101
1.0 µm
100
Kb
104
(400)Pb 2 Nb 2O 7
105
(222)Pb 2 Nb 2O 7
Intensity (a.u.)
106
(001) (001)
film.
20
30
40
50
Twotheta Figure 5.1. X-ray diffraction theta-twotheta scans of PMN-PT films of 1, 4.5 and 7.5 µm thickness grown on 4o miscut (001) SrTiO3 substrates. Figure 5.2 is a cross-sectional transmission electron microscopy image of the near-surface region of a 9 µm thick PMN-PT film, showing perovskite phase with a small volume fraction of second phase embedded in the perovskite, as indicated by arrows. The
75 surface roughness of the films, studied by cross-sectional scanning electron microscopy, degraded with thickness.
(PMN-PT)pyro
surface
(PMN-PT)pero
500 nm
Figure 5.2. Cross-sectional bright field TEM image of a 9.0 µm thick PMN-PT film near the surface. Pyrochlore impurity phases are arrowed. In order to make electrical and electromechanical measurements, Pt top electrodes were deposited on the film surface. Figure 5.3 shows the polarization versus electric field for 1.0, 4.5, and 9.0 µm films measured at room temperature. The films show wellsaturated hysteresis loops, which are shifted in the positive direction for electric field applied to the top electrode. This imprint is more pronounced in the 1 m film. The 4.5 µm thick film shows the highest remnant polarization [11] of ~32 C/cm2, which exceeds
the value, 25 C/cm2, measured for flux-grown (001)-oriented 0.70PMN-0.30PT single crystals [12]. This suggests that it is possible to induce some tetragonal material at high electric fields, leading to an increase the measured remanent polarization. The 1.0 and 9.0 µm thick films show remanent polarizations of ~16 µC/cm2 and ~29 C/cm2, respectively. The coercive field decreased with film thickness.
76
2
Polarization (µC/cm )
60 40 20 0
1.0 µm 4.5 µm 9.0 µm
-20 -40 -60 -100
-50
0
50
100
Electric field (kV/cm)
Figure 5.3. Polarization vs. electric field hysteresis loops for 1.0, 4.5, and 9.0 µm thick films. Figure 5.4 shows the temperature dependence of the zero-bias permittivity at various frequencies for the 1.0 and 4.5 µm films. The 1.0 µm thick film shows clear relaxor behavior; in contrast, the 4.5 and 9.0 µm thick films show reduced dispersion. The temperature of the maxima in the permittivity at 1000 Hz is 197, 177, and 167 oC for the 1.0, 4.5, 9.0 µm films, respectively , which are higher than that (160 oC) of bulk single crystals with crystals with the same composition [2]. There are several possible reasons as to why the films show elevated maxima temperatures. First, transition temperatures in ferroelectric films depend on the level of biaxial stress imposed [13], and this stress state could be a function of the film thickness, although this remains to be thoroughly investigated. Second, the peak temperature in PMN-PT is a strong function of the Ti content [14]. A higher-than-expected Ti content would be consistent both with a high permittivity maximum temperature and with the high remnant polarization. Unfortunately, no rhombohedral to tetragonal phase transition
77 in the permittivity data was observed for these films, so it was not possible to use the two transition temperatures to assign a Ti stoichiometry unambiguously. Third, it is known that the transition temperature in bulk and thin film Pb-based perovskite depends on the lead stoichiometry, such that lead deficiency can lead to higher transition temperatures [15].
4000
0.6
1000 Hz 3162 Hz 10000 Hz 31622 Hz
0.4
2000 0 200
0.2
300
400
500
Temperature (K)
0 600
Dielectric Constant
6000
0.8
(b) 15000
10000
0.6
1000 Hz 3162 Hz 10000 Hz 31622 Hz
0.4
5000
0
200
Tan δ
0.8
(a)
Tan δ
Dielectric Constant
8000
0.2
300
400
500
0
600
Temperature (K)
Figure 5.4. Temperature dependence of permittivity and dielectric loss for the (a) 1.0 µm and (b) 4.5 µm thick film. A variation in Pb or Ti stoichiometry through the film thickness could thus be present in these samples, and yield the observed thickness-dependent behavior. Further research is required to distinguish which, if any, of these mechanisms dominates. Piezoelectric coefficients were measured using double beam interferometry for d33,f and the wafer flexure technique for e31,f [16]. The double beam interferometry measurements were carried out as a function of both ac and dc bias level. For e31,f measurements, data were collected for the direct effect, without dc bias field. In this case films were poled at room temperature prior to measurement, and the maximum values were collected immediately after poling to minimize aging. The piezoelectric response was then monitored as a function of time after poling to track the aging behavior.
78 Figure 5.5 shows the double beam interferometry data for the 4.5 µm thick film. The 4.5 µm thick film showed the highest d33,f coefficient, 320±10 pm/V, which is comparable to previous reports measured by x-ray diffraction at low field [8]. The piezoelectric response first increases, then begins to drop as the magnitude of the dc bias is increased. Values for the 1 and 9 µm thick films are much lower, with maximum d33 values of 70±5 pm/V and 30±15 pm/V, respectively. The reduced piezoelectric response of the 9 µm thick film can be attributed to the increased amount of pyrochlore phase, while that in the 1 µm thick film may come about from a larger substrate constraint.
450
d33 (pm/V)
400
350
300
250
AC Voltage : 1.0 Vrms Frequency : 1KHz 0
2.0
4.0
6.0
8.0
10.0
DC bias voltage (V)
Figure 5.5. d33,f data as a function of dc bias field for the 4.5 µm thick PMN-PT film.
79 -30
e31 (C/cm2)
-25
+ 11 kV/cm - 11 kV/cm + 22 kV/cm - 22 kV/cm + 33 kV/cm - 33 kV/cm
-20 -15 -10 -5 0 0
5
10
15
20
25
Poling Time (minutes) Figure 5.6. e31,f as a function of poling field and time for the 4.5 µm thick film.
Figure 5.6 shows the variation of the e31,f coefficient of the 4.5 µm thick film as a function of the poling electric field and poling time The maximum e31,f coefficient is ~ – 27 C/m2. This is considerably above the values reported either in polycrystalline films, or in thinner epitaxial PMN-PT films [8], or in Pb(Zr,Ti)O3 films [17]. For the 1 and 9 µm thick films, the maximum e31,f values were –18 C/m2 and –13 C/m2, respectively, a trend that is consistent with that of the ferroelectric and longitudinal piezoelectric properties. The strong imprint in the films is evident in the dependence of the piezoelectric response and the aging rate on the poling direction. When poled into to the preferred polarization orientation, the e31,f coefficient is large and ages slowly, approximately – 0.7%/decade of time. However, when the polarization is reversed, it is difficult to stabilize a large remnant polarization, and e31,f is small, and ages quickly. Similar aging behavior has been reported by Kholkin[18] and Maria [8]. The superior piezoelectric properties indicate that the 4.5 µm thick film is a high crystalline quality film, which is probably under relatively smaller substrate constraints
80 than thin films and has relatively small amounts of pyrochlore phase. These films will allow us to grow high quality multilayer heterostructures of single crystal piezoelectric PMN-PT/SRO conductive oxides for microelectromechanical systems and ultrasound devices.
81
References [1]
S. E. Park and T. R. Shrout, J. Appl. Phys. 82, 1804 (1997).
[2]
S. E. Park and T. R. Shrout, IEEE Trans. Ultrason. Ferroelectr. Freq. Control 44, 1140 (1997).
[3]
J. P. Maria, W. Hackenberger, and S. Trolier-McKinstry, J. Appl. Phys. 84, 5147 (1998).
[4]
G. R. Bai, S. K. Streiffer, P. K. Baumann, O. Auciello, K. Ghosh, S. Stemmer, A. Munkholm, C. Thompson, R. A. Rao, and C. B. Eom, Appl. Phys. Lett. 76, 3106 (2000).
[5]
S. D. Bu, M. K. Lee, C. B. Eom, W. Tian, X. Q. Pan, S. K. Streiffer, and J. J. Krajewski, Appl. Phys. Lett. 79, 3482 (2001).
[6]
V. Nagarajan, S. P. Alpay, C. S. Ganpule, B. K. Nagaraj, S. Aggarwal, E. D. Williams, A. L. Roytburd, and R. Ramesh, Appl. Phys. Lett. 77, 438 (2000).
[7]
N. Wakiya, K. Shinozaki, and N. Mizutani, 384, 189 (2001).
[8]
J. P. Maria, The Pennsylvania State University, 1998.
[9]
C. B. Eom, R. B. Vandover, J. M. Phillips, D. J. Werder, J. H. Marshall, C. H. Chen, R. J. Cava, R. M. Fleming, and D. K. Fork, Appl. Phys. Lett. 63, 2570 (1993).
[10]
JCPDS 40-828 for Pb2Nb2O7 and JCPDS 25-443 for Pb3Nb4O13 X-Ray Powder Diffraction Data Card.
[11] Remanent polarization was obtained as the intersection between the hysteresis loop and the positive y-axis for applied field reduced to zero. This value of remanent polarization may not be entirely accurate due to the imprint in the hysteresis loop. The remanent polarization from the intersection between the hysteresis loop and the negative y-axis was ~40 C/cm2. . [12] S.E. Park and T.R. Shrout, ((unpublished data)). [13]
N. A. Pertsev, A. G. Zembilgotov, and A. K. Tagantsev, Phys. Rev. Lett. 80, 1988 (1998).
[14]
S. W. Choi, T. R. Shrout, S. J. Jang, and A. S. Bhalla, Ferroelectrics 100, 29 (1989).
[15]
K. Okazaki and K. Nagata, J. Am. Ceram. Soc. 56, 82 (1973).
82 [16]
J. F. Shepard, Jr., P. J. Moses, and S. Trolier-McKinstry, Sens. Actuators, A A 71, 133 (1998).
[17]
Paul Muralt, IEEE Trans. Ultrason. Ferroelectr. Freq. Control 47, 903 (2000).
[18]
A. Kholkin, E. Colla, K. Brooks, P. Muralt, M. Kohli, T. Maeder, D. Taylor, and N. Setter, Micro. Eng. 29, 261 (1995).
83
Chapter 6.
Thickness dependence of structural and piezoelectric properties of epitaxial Pb(Zr0.52Ti0.48)O3 Films on Si and SrTiO3 substrates
6.1. Introduction High strain piezoelectric materials have drawn much attention for the fabrication of micro-electromechanical systems (MEMS), such as microactuators, micromotors, pressure transducers, strain gauges, and high frequency ultrasound transducers [1-3]. Pb(ZrxTi1-x)O3 (PZT) ceramics with the composition (x= 0.45-0.5) close to morphotrophic phase boundary (MPB) are the most commonly used piezoelectric material for various sensors and actuators [4]. With the increasing demand of the micro and nano-mechanical devices and their integration into Si-technology it is desirable to fabricate thick PZT films with high piezoelectric properties. Depending on various device dimensions and frequency range it requires growing piezoelectric films with different thicknesses (1-100 µm). The key advantages of the thick films over bulk ceramics are the lower driving voltage with multilayer stacks and parallel wiring [5] and high frequency applications [5]. Due to superior dielectric and piezoelectric properties of ceramic PZT, the ceramic PZT have been dominated in industries during past 50 years. Single crystal PZT is expected to have enhanced piezoelectric properties over the bulk ceramic. However, growth of single crystal PZT is not easy due their peritectic characteristics. To grow single crystal PZT, we can use thin film deposition method. There have been many endeavors and some progress on fabrication and understanding of piezoelectric characteristics of the PZT thin films. In most of the cases, the observed piezoelectric coefficients of the thin films were much lower than the respective bulk PZT ceramics and
84 was explained by the substrate induced constraints [6-8]. Recently, Nagarajan et al. [9] observed a reduction of the substrate clamping effect on PZT (20/80) thin films by patterning small (1 µm×1 µm) discrete islands, which facilitates the motion of ferroelastic domain walls. They obtained a high value of piezoelectric coefficient (~250 pm/V). Xu et al. [10] have studied the extrinsic and intrinsic contribution to piezoelectric properties
of sol-gel derived polycrystalline PZT films with thickness range 0.25 to 10 µm. They have observed that the non-180o domain wall switching is negligible for the film thickness less than 2µm, and ferroelastic domain pinning reduced significantly with increasing film thickness, which enhanced the piezoelectric coefficient (~130 pm/V). Haun et al. [11] and Du et al. [12] have predicted theoretically the d33 value of rhombohedral PZT single crystals about 600 pm/V using first principle calculations. However, there are no experimental results in which the d33 value is close to the theoretical prediction. It is desirable to fabricate high quality epitaxial PZT thick films that are expected to have higher piezoelectric response for high performance electromechanical systems. In this letter, we report the fabrication of high quality epitaxial PZT thick films up to 4 µm on both (001) SrTiO3 and (001) Si substrates with higher piezo response. We have also studied the thickness-dependent structural and piezoelectric properties with close correlation to the substrate constraints and degree of tetragonality of the of PZT lattice.
6.2. Experimental method Epitaxial PZT (100) films with various thicknesses (0.4-4µm) were grown on (001) SrTiO3 and (001) Si substrates using on-axis radio-frequency (RF) magnetron sputtering.
85 The growth of epitaxial PZT films on various substrates allows us to investigate the influence of strain states on piezoelectric characteristics. The nominal composition of the sputtering target was PZT(Zr/Ti=52/48). Molecular-Beam-Epitaxy (MBE) was used to fabricate 100Å of SrTiO3 on the Si substrate as a template layer in order to grow epitaxial PZT films. The details of the MBE process parameters for SrTiO3 deposition are described elsewhere [13]. Prior to the PZT film deposition, the epitaxial SrRuO3 bottom electrode was deposited by a 90° off-axis RF magnetron sputtering [14]. During the PZT film deposition the substrate temperature was maintained at 600°C with an oxygen pressure of 400 mTorr, followed by cooling under an oxygen pressure of 300 Torr.
6.3. Structure Characterization Epitaxial arrangement and three-dimensional strain states of the PZT films as a function of thickness were determined using a four-circle x-ray diffractometer (XRD). Figure 6.1 (a) shows the typical θ-2θ scans of 0.8 µm and 3.8 µm thick films on (001)Si. Thinner films are purely (00l) oriented texture. However, the textured (101) reflection was observed for thick films above 2 µm, and pronounced for higher thicknesses. The 3.8 µm thick PZT films on Si show relatively higher content of (101) phase than the PZT films on SrTiO3. This difference may be due to the lattice mismatch between the film and the substrate, and the crystalline quality of the SrTiO3 template layer. The crystalline quality of the PZT films was determined from their rocking curve widths of the PZT 002 reflections. With increasing film thickness on both the substrates the rocking curve width increases. The measured full width at half maximum (FWHM) of the rocking curve, for the 3.8 µm thick PZT films on SrTiO3, and Si was ~0.57° and ~0.67°, respectively.
86 Figure 6.1 (b) shows the azimuthal φ-scan of PZT 101 reflection. It is clear that in-plane texture is cube-on-cube epitaxy without misoriented grains. Similar cube-on-cube epitaxy
103
002
SrTiO3 002
104
101
Intensity (a.u.)
105
SrTiO3 001
001
106
SrRuO3 220
was also observed in the case of PZT films on (001)SrTiO3 substrates.
3.8µm
102 0.8µm
101 100 20
25
30
35
40
45
50
2θ (degrees)
Intensity (a.u.)
200 150 100 50 0 0
50
100
150
200
250
300
350
φ (degrees) Figure 6.1. a) X-ray diffraction θ-2θ scan of 0.8 and 3.8 µm thick PZT films on (001) Si Substrates. b) The azimuthal φ-scan of PZT 101 reflection which shows the PZT films on Si grow with cube-on-cube epitaxy Figure 6.2 shows the variation of in-plane and out-of-plane lattice parameters of
87 PZT films on Si and SrTiO3 substrates as a function of film thicknesses. The outof-plane lattice parameters were determined by normal θ-2θ scans. The in-plane lattice parameters were determined by off-axis reflections. It was found that the out-of-plane lattice parameter decreased and in-plane lattice parameter increased with film thicknesses irrespective of the substrate. As shown in the figure the out-of-plane lattice parameter of PZT thin films on SrTiO3 substrates has higher value in comparison to the respective bulk. Since the thermal expansion coefficients of PZT and SrTiO3 are nearly the same (~11×106
/ºC), it could be possible that the thinner PZT films (< 0.5 µm) carry a large amount of
compressive strain due to the lattice mismatch between the films and substrate (~3.35%). At higher thickness, the influence of epitaxial strain could be minimal and the lattice structure have the tendency to change from tetragonal to pseudo-rhombohedral behavior with reduced tetragonality factor, which is clear from the converging nature of the inplane and out-of-plane lattice parameters. The nature of variation of in-plane and out-ofplane lattice parameters of PZT films on Si is similar to those of films on SrTiO3, however, the effect is more pronounced. It could be possible that the films on Si carry a large amount of tensile strain in the plane due to the mismatch of thermal expansion coefficients. The exact mechanism of the converging nature of the in-plane and out-ofplane lattice parameters of PZT at higher thicknesses is not clearly understood. Pertsev et al. [15] reported the effect of mechanical boundary conditions on phase diagrams of epitaxial ferroelectric thin films. According to their theory, the substrate constraint can affect the stability of phases and the thermodynamic potential that can change with film thicknesses. Therefore, it could be possible with increasing thicknesses of the PZT films the crystal structure follow the pseudo-rhombohedral behavior.
88
Lattice parameter (Å)
4.20 Bulk c-lattice
4.16
Out-of-plane (PZT/STO)
4.12 Out-of-plane (PZT/Si)
4.08
In-plane (PZT/Si) In-plane (PZT/STO)
4.04
Bulk a-lattice
4.00 0
1
2
3
4
Thickness (µm) Figure 6.2. In-plane and out-of -plane lattice parameters vs. films thickness of PZT films on (001) SrTiO3 and (001) Si substrates.
This was supported by the reduction of the tetragonality (c/a) of PZT films with increase in thicknesses. The PZT films on Si showed tetragonality smaller in comparison to films on SrTiO3 substrates. This result suggests that the PZT films on Si have a higher tendency towards rhombohedral structure, which is expected to show enhanced piezoreponse. The puzzle of rhombohedral-like behavior in tetragonal phase was investigated using compositional analysis. The target and the film compositions were determined by the wavelength dispersive spectrometer (WDS) to verify the film is in the tetragonal region. It was found that the Zr/Ti ratio of the target and film were 52/48 and 45/55, respectively, which are shown in Figure 6.3. Although the compositions are in the tetragonal phase region of the phase diagram, it is possible that the lattice distortion increased with typical deposition parameters, cooling process, and growth induced strain.
89 As a result the lattice structure followed pseudo-rhombohedral characteristics rather than large tetragonality at higher thicknesses. Target composition composition Target Zr:Ti=52:48 Zr :Ti = 52 : 48
Film composition composition Film Zr:Ti=45:55
Temperature (°C)
500
Cubic
400
300
200
Tetragonal
Rhombohedral MPB
100
0 0
10
20
PbZrO3
30
40
50
60
70
80
90 100
PbTiO3
Figure 6.3. The composition of PZT films analyzed by WDS [4].
6.4. Ferroelectric properties The ferroelectric behavior of the films was examined using ferroelectric tester (RT 66A, Radiant Technologies) in virtual ground mode. It was found that the PZT films on Si have a relatively low value of remnant polarization (Pr) and a higher value of coercive field (Ec) than the films on SrTiO3 substrates not shown here. We believe that the PZT films on Si might carry a relatively large amount of epitaxial and thermal strain and as a result exhibit a larger value of Ec. Figure 6.4 shows the thickness-dependent remnant polarization of PZT films on Si and SrTiO3 substrates. It is clear that the remnant
90 polarization was increased with the film thicknesses up to 2 µm and reduced for higher thicknesses (>2 µm), which is shown in Figure 6.5 for the PZT films on (001) SrTiO3. It is well known that the intrinsic and extrinsic properties of the materials can have contributions towards remnant polarization in ferroelectrics. As we increase the thickness of films the switching of domains (both 180o and non-180o) might increase with reduction of substrate constraints that leads to increased remnant polarization even the tetragonality of (00l) domains reduced with the thicknesses. At higher thicknesses (> 2 µm), the PZT films get softened which reduces the Pr and Ec value irrespective of the
substrate. Thick PZT films on SrTiO3 substrates exhibit relatively higher value of remnant polarization in comparison to films on Si. These results are consistent with the lower tetragonality of PZT films on Si substrates, and also indicative of significant
2
80 60 40nm 40 (@800kV/cm) 20 0 -20 -40 -60 -80 -2.4 -1.8 -1.2 -0.6 0.0 0.6 1.2 1.8 2.4
80 60 40 20 0 -20 -40 -60 -80 -4
Polarization( µC/cm )
Polarization( µC/cm 2)
extrinsic contributions to the polarization components.
2
Polarization( µC/cm )
Polarization(µC/cm2)
nm
-4
-2
0
2
4
Applied voltage(V)
6
8
-3
-2
-1
0
1
2
3
4
Applied voltage(V)
Applied voltage(V) 80 60 160 40 20 0 -20 -40 -60 -80 -8 -6
80 nm
80 60 700 nm 40 20 0 -20 -40 -60 -80 -32 -24 -16 -8
0
8
16 24 32
Applied voltage(V)
Figure 6.4. Polarization hysteresis loop as increasing film thickness for the PZT films on (001) SrTiO3.
Polarization, 2Pr (( µC/cm ²)
91 160 140
PZT/STO PZT/STO -Si
120 100 80 60 40 20 0 0
1
2
3
4
Thickness ( µm) Figure 6.5. Thickness dependent polarization of PZT films on (001)SrTiO3 and (001)Si substrates.
6.5. Piezoelectric properties In general, the longitudinal piezoelectric coefficient (d33) of PZT thin films is influenced by the composition, orientation, and presence of non 180o domains. By fabricating ideal epitaxial films on suitable substrates, it could be possible to modify the domain orientations of PZT, and also their piezo-response. Figure 6.6 shows the typical field dependent d33 characteristics of 4µm PZT on SrTiO3 and Si substrates. It is clear that the films on Si have much higher values of d33 (~330 pm/V) than those of films on SrTiO3 (~200 pm/V). This result can be correlated to the pseudo-rhombohedral characteristics of PZT, as observed from structural data. Figure 6.7 shows the piezoelectric coefficients of the PZT films on SrTiO3 and Si substrates as a function of film thicknesses. Figure 6.8 shows piezoelectric hysteresis of PZT on Si as thickness increases. The nature of the increment of d33 value with film thicknesses is similar for the PZT films on both the substrates, however, the films on Si
92 have significant enhancement of d33. The increased piezoelectric coefficient with film thickness could be due to the reduction of substrate constraints and softening of the material by structural modification from higher tetragonal to lower tetragonal symmetry. This behavior could be directly correlated to the microstructure of the films on both the substrates. From the surface morphology by SEM, microcracks were observed at the thickness above 2 µm of PZT films on Si substrates, as shown in Figure 6.9. There were no cracks found on PZT films on SrTiO3 substrates. A large number of cracks on thick (>2 µm) PZT films on Si substrates could be analogous to the PZT cut-capacitors or islands of various sizes. However, the aspect ratio of those small capacitors is much higher than the observed cracks on PZT films on Si. We have observed cracks on PZT films at a separation 60 µm. We believe that our continuous films still have substrateinduced constraint and that by pattering into small capacitors (1 µm × 1 µm) could further improve the d33 value. 600
d 33 (pm/V)
400 200 0 -200
PZT/STO-Si
-400 -600 -150 -100
PZT/STO
-50
0
50
100
150
Field (kV/cm) Figure 6.6. Typical longitudinal piezoelectric coefficient of 3.8µm thick PZT films on (001)SrTiO3 and (001)Si substrates.
93 600
PZT/STO -Si
d 33 (pm/V)
500
PZT/STO
d 33,bulk,tet
400 300
d 33,clamped,tet
200 100 0 0
1
2
3
4
Thickness ( µm) Figure 6.7. Thickness dependence of longitudinal piezoelectric coefficient (d33) of PZT films on (001) SrTiO3 and (001)Si substrates.
600
d 33 (pm/V)
400
3.8 µm
0.7 µm
200
0.16 µm
0 -200
0.08 µm
-400 -600 -1200 -800
-400
0
400
800 1200
Applied field (kV/cm) Figure 6.8. Thickness dependent piezoelectric properties of PZT on (001) Si.
94
Figure 6.9. SEM image of the 4um PZT on (001) Si.
The other contribution to the high d33 value could be due to the reduction of the tetragonality of PZT films on Si, which is the responsible factor for the softening of the films. As the PZT films get softened, depinning of non-180o domains is easier. This facilitates the movement of ferroelastic domain walls and easy switching of such domains with minimal substrate constraints could be the origin of such a different value of d33 of PZT on SrTiO3 and Si substrates. This result suggests that the thick epitaxial PZT films on Si with such a high value of piezoelectric coefficients will open a direction for the fabrication of high performance electromechanical systems for high frequency applications.
95 6.6. Conclusions
In summary, we have successfully fabricated epitaxial PZT films up to 4 µm thick on SrTiO3 and Si substrates. Irrespective of the substrates, it is observed that increasing the thickness of the PZT films leads to the reduction of tetragonality factor of (00l) domains. The piezoelectric coefficients are found to be higher for PZT films with higher thicknesses. The enhanced piezoelectric coefficient of PZT films on Si over that of SrTiO3 is explained in terms of observed microstructure, lower tetragonality, and easy depinning of non-180o domains.
96 References
[1]
K. Uchino, Piezoelectric Actuators and Ultrasonic Motors. (Kuwer Academic, Boston, 1996).
[2]
A. M. Flynn, L. S. Tavrow, S. F. Bart, R. A. Brooks, D. J. Ehrlich, K. R. Udayakumar, and L. E. Cross, J. Microelectromech. Syst. 1, 44 (1992).
[3]
D. M. Kim, S. D. Bu, C. B. Eom, J. Lettieri, T. Yosimura, S. Trolier-McKinstry, D. G. Schlom, V. Nagarajan, A. Stanishevsky, B. Liu, R. Ramesh, W. Tian, X. Q. Pan, Yoshimura T., and S. K. Streiffer, (2003).
[4]
Bernard Jaffe, William R. Jr. Cook, and Hans Jaffe, Piezoelectric Ceramics. (Academic Press, London, 1971).
[5]
R. L. Goldberg and S. W. Smith, IEEE Trans. Ultrason. Ferroelectr. Freq. Control 41, 761 (1994).
[6]
H. D. Chen, K. R. Udayakumar, L. E. Cross, J. J. Bernstein, and L. C. Niles, J. Appl. Phys. 77, 3349 (1995).
[7]
Y. Wang, C. Ganpule, B. T. Liu, H. Li, K. Mori, B. Hill, M. Wuttig, R. Ramesh, J. Finder, Z. Yu, R. Droopad, and K. Eisenbeiser, Appl. Phys. Lett. 80, 97 (2002).
[8]
S. Hiboux, P. Muralt, and T. Maeder, J. Mater. Res. 14, 4307 (1999).
[9]
V. Nagarajan, A. Roytburd, A. Stanishevsky, S. Prasertchoung, T. Zhao, L. Chen, J. Melngailis, O. Auciello, and R. Ramesh, Nature Materials 2, 43 (2003).
[10]
F. Xu, S. Trolier-McKinstry, W. Ren, Xu Baomin, Z. L. Xie, and K. J. Hemker, J. Appl. Phys. 89, 1336 (2001).
[11]
M. J. Haun, E. Furman, S. J. Jang, and L. E. Cross, Ferroelectrics 99, 63 (1989).
[12]
97 X. H. Du, J. H. Zheng, U. Belegundu, and K. Uchino, Appl. Phys. Lett. 72, 2421 (1998).
[13]
G. Y. Yang, J. M. Finder, J. Wang, Z. L. Wang, Z. Yu, J. Ramdani, R. Droopad, K. W. Eisenbeiser, and R. Ramesh, J. Mater. Res. 17, 204 (2002).
[14]
C. B. Eom, R. J. Cava, R. M. Fleming, J. M. Phillips, R. B. Vandover, J. H. Marshall, J. W. P. Hsu, J. J. Krajewski, and W. F. Peck, Science 258, 1766 (1992).
[15]
N. A. Pertsev, A. G. Zembilgotov, and A. K. Tagantsev, Phys. Rev. Lett. 80, 1988 (1998).
98
Chapter 7. Conclusions
We have first demonstrated the giant piezoelectric responses in piezoelectric epitaxial thick film heterostructures on silicon. The ferroelectric material systems we have studied in this work include PMN-PT relaxor ferroelectrics and PZT ferroelectrics. Both materials systems have been widely used for various electromechanical systems in bulk
single
crystals
form or
polycrystalline
ceramics
form.
However,
the
electromechanical coupling coefficients of thin films are quite low, severely limiting performance. A major challenge is to prepare these materials as "single crystal" epitaxial thin or thick films between electrodes, and integrate them on silicon so that these properties can be utilized in electromechanical systems with all the advantages of yield, uniformity, cost and performance associated with microelectronic technologies. Growth of epitaxial PMN-PT films with good electromechanical properties is known to be difficult due to the relatively poor thermodynamic stability of the perovskite phase relative to the pyrochlore phase and due to compositional complexity. We have used novel approach to growth high quality piezoelectric thick films. First, we have developed a high pressure on-axis sputtering process to deposit ferroelectric thick films with high growth rate and stoichiometric composition. Second, we have employed miscut Si substrates to stabilize perovskite phase and suppress the formation of pyrochlore phases. We believe this is attributed to the terrace length variation with miscut angle. This in turn affects the stabilization of volatile Pb and maintains stoichiometry on short terrace length on high miscut angle substrates. Third, we have used an epitaxial SrTiO3 template layer to grow epitaxial piezoelectric thick films directly on silicon. It is difficult
99 to grow epitaxial perovskite films on the Si substrate directly because of the large lattice mismatch and formation of SiO2 on the silicon substrate. We have successfully prepared epitaxial SrTiO3 buffer layers on silicon substrates by molecular beam epitaxy (MBE). The heterostructures will allow us to fabricate high frequency medical ultrasound transducers, and MEMS devices with all the advantages associated with Si-based microelectronic technology. We have also used novel approach to reduce the substrate clamping effect in piezoelectric thick films and obtain the high piezoelectric coefficient in thick films. Nano-structuring of PMN-PT films has been carried out using Focused-Ion Beam (FIB) milling to reduce the constraint imposed by the underlying silicon substrates. The continuous 4 µm thick films showed a d33 value of about 800 pm/V in a low-field regime. When the films are subdivided into ~4 µm x 4µm capacitors by FIB milling, the d33 value of the films increased to over 1200 pm/V under bias, which is clear proof of severe constraint effect of substrate. Although we have achieved the highest longitudinal piezoelectric coefficients (d33 = 1200 pC/N) in our epitaxial PMN-PT films on (001) Si substrates, they are still smaller than the corresponding bulk single-crystal values. The differences between thin film and bulk crystal properties have been attributed to strain, substrate constraint, phase purity, and non-stoichiometry and need to be further studied. We have only achieved the highest transverse piezoelectric coefficients (e31,f = 27 C/m2) in our epitaxial PMN-PT films on (001) SrTiO3 substrates. As e31,f and k31 are the most useful electromechanical coefficients for MEMS devices, it is particularly desirable to explore routes to improve their values in silicon substrates. The strain state in
100 the piezoelectric layers can alter the electromechanical properties via change of polarization direction. For ferroelectric heterostructures on silicon, there is a large tensile stress in the piezoelectric film due to the thermal expansion mismatch between the perovskite oxide and silicon, which can pull the polarization vector into the plane of the film. We need to deposit epitaxial PMN-PT thick films with controlled strain states in order to study the strain effect on electromechanical coupling coefficients, k31. We still have remaining challenges to grow much thicker epitaxial PMN-PT films on silicon up to ~ 100 µm thickness and to understand the pyrochlore formation mechanisms for application to high frequency ultrasound imaging. We have found the substrate miscut suppresses the formation of pyrochlore phases although the thickness is limited less than 4 microns.
We need to understand the nucleation mechanism of
pyrochlore phases to grow thicker films above 5 microns. Investigations of single crystal epitaxial films will lead to significant improvements in our understanding of the piezoresponse of these materials. These films will serve as model systems for investigating the electromechanical response with an ability to control microstructure and composition in ways that are not possible in bulk materials processed at high temperature under thermodynamic equilibrium. We believe that the technology of heteroepitaxial growth and photolithographic processing to produce single crystal thick film ferroelectrics on single crystal metallic oxide electrodes is a significant step in improving medical ultrasound transducers and electromechanical systems such as MEMS devices.
Copy right by Dong Min Kim 2004 All Rights Reserved
