POLING AND SWITCHING OF PZT CERAMICS. field and grain size effects. Talal M. Kamel

Transcription

1 POLING AND SWITCHING OF PZT CERAMICS field and grain size effects Talal M. Kamel

2

3 POLING AND SWITCHING OF PZT CERAMICS field and grain size effects PROEFSCHRIFT ter verkrijging van de graad van doctor aan de Technische Universiteit Eindhoven, op gezag van de Rector Magnificus, prof.dr.ir. C.J. van Duijn, voor een commissie aangewezen door het College voor Promoties in het openbaar te verdedigen op woensdag 5 september 27 om 16. uur door TALAL MOHAMMAD KAMEL geboren te Caïro, Egypte

4 Dit proefschrift is goedgekeurd door de promotor: prof.dr. G. de With T. M. Kamel A catalogue record is available from the Eindhoven University of Technology Library ISBN: Copyright 27 T. M. Kamel Front cover page: A humorous resemblance between combined hysteresis loop characteristics of soft and hard PZT with... a dancing penguin! Back cover page: The principle of switching in PZT perovskite. The off-centred titanium/zirconium ions move up and down following the direction of the electric field. Cover Design: T. M. Kamel An electronic copy of this thesis is available at the web-site of the Eindhoven University Library in PDF format ( Printed by PrintPartners IpsKamp, Enschede, The Netherlands.

5

6

7 TO NAOOM NOUR AND JOURINE

8

9 C ONTENTS 1 INTRODUCTION Abstract Dielectric Materials Polar Materials Piezoelectricity Pyroelectricity Ferroelectricity Applications of Polar Materials Perovskites PZT Soft and Hard PZT Ferroelectric Domains Poling of Ferroelectrics Domain Switching Aim of the Thesis Thesis Outline References MATERIALS AND METHODS Abstract Soft and Hard PZT Ferroelectric Hysteresis Loop Poling Pyroelectric Current DC Conductivity Dielectric and Piezoelectric Parameters References POLING OF SOFT PZT Part I: Field Effect Abstract I.1. Introduction I.2. Theory I.2.1. Dielectric Response I.2.2. Polarization Switching I.3. Experimental I.4. Results I.4.1. Microstructure I.4.2. Hysteresis Loop I.4.3. Polarization Current I.4.4. Pyroelectric Current I.4.5. Dielectric Properties I.5. Discussion I.5.1. Polarization Current Curves I.5.2. Low Field Regime I.5.1. High Field Regime Conclusions References Part II: Grain Size Effect Abstract II.1. Introduction II.2. Experimental II.3. Results II.4. Discussion II.4.1. Limits of the Grain Size Effect II.4.2. Intrinsic and Extrinsic Contributions Conclusions References i

10 4 SWITCHING OF SOFT PZT Part I: Field Effect Abstract I.1. Introduction I.2. Experimental I.3. Results I.4. Discussion I.4.1. Physical Model Conclusions References Part II: Grain Size Effect Abstract II.1. Introduction II.2. Experimental II.3. Results II.4. Discussion II.4.1. Double Switching II.4.2. Pyroelectric Coefficient Conclusions References PYROELECTRICITY VERSUS CONDUCTIVITY IN SOFT PZT Abstract Introduction Doping Dependent Resistivity Resistance Degradation Experimental Results Discussion Field Effect Grain Size Effect Conclusions References POLING OF HARD PZT Abstract Introduction Literature Review Experimental Material Microstructure Hysteresis Loop Poling Dielectric, Piezoelectric and Pyroelectric Properties Results Microstructure Hysteresis Loops after Ageing Poling Effect after Ageing Hysteresis Loop Depinching Poling Effect after Deageing Discussion Model of Ageing Phenomena Model of Thermal Deageing Interpretation of the Pyroelectric Curves Dielectric Loss Factor Data Conclusions References ii

11 7 EPILOGUE Abstract What Was Planned What Was Achieved Future Research References SUMMARY APPENDIX A SAMPLING AND HOMOGENEITY APPENDIX B EQUATIONS OF PIEZOELECTRIC PARAMETERS ACKNOWLEDGEMENTS ABOUT THE AUTHOR iii

12

13 CHAPTER 1 INTRODUCTION ABSTRACT In this chapter a brief essay is given on the fundamental concepts and terminologies used throughout the thesis. A logical sequence is followed to reach eventually the ultimate aim of the current study. The chapter reviews the piezoelectricity, pyroelectricity, ferroelectricity, perovskites, soft and hard PZT, domain switching and poling.

14 Chapter (1) Introduction 1.1. DIELECTRIC MATERIALS Dielectric Materials can be classified into two classes 1 : non-polar (neutral) dielectrics and polar (dipole) dielectrics. If all positive and negative charges of a molecule are replaced by one positive and one negative charge mutually equal in absolute magnitude and located at the centres of gravity of the separate positive and negative charge distribution, these summary charges may either coincide or not coincide in space. In the first case we have a nonpolar molecule and the matter composed of such molecules is also called nonpolar. In the second case, the molecule has an effective and permanent electric dipole moment different from zero even in the absence of an external field, and the materials containing such kind of dipoles are called polar POLAR MATERIALS According to the symmetry operations, crystals are commonly classified into seven crystal classes 2 : triclinic, monoclinic, orthorhombic, tetragonal, trigonal, hexagonal, and cubic. These systems can again be subdivided into point groups according to their symmetry with respect to a point. All crystals can be subdivided into 32 point groups. Of the 32 point groups, 11 classes possess a centre of symmetry. The latter, of course, possess no polar properties. If, for example, a uniform stress is applied to such a centrosymmetric crystal, the resulting small movement of charge is symmetrically distributed about the centre of symmetry in a manner, which brings about a full compensation of relative displacements. But the application of an electric field does produce a strain. This property is termed as electrostriction and occurs naturally in all substances, crystalline or otherwise Piezoelectricity Of the remaining 21 non-centrosymmetric crystal classes, all except one point group, possess the so-called piezoelectric effect. One point group, although lacking a center of symmetry, is not piezoelectric because of other combined symmetry elements 3. Piezoelectricity is the ability of a material to exhibit a spontaneous polarization when subjected to a mechanical stress, Figure 1.1A. The material also exhibits the converse effect by undergoing a mechanical deformation under the application of an electric field, Figure 1.1B. The direct and converse piezoelectric effects can be expressed in tensor notation as: P = σ (direct effect ) (1.1) i d ijk ij kij jk s = d E (converse effect) (1.2) k 2

15 Introduction where P i is the polarization generated along the i-axis in response to the applied stress σ jk, and d kij is the piezoelectric coefficient. For the converse effect, s ij is the strain generated in a particular orientation of the crystal on the application of electric field E k along the k-axis 4,5. Piezoelectricity derives its name from the Greek word piezein meaning to press or squeeze. F Polar Axis + _ + F _ + _ + _ + (A) + + _ (B) _ Figure 1.1: 1: (A) Direct piezoelectric effect, (B) Converse piezoelectric effect Pyroelectricity Of the 2 piezoelectric crystal classes 1 are characterized by a unique polar axis. Crystals belonging to these classes are called pyroelectrics. Pyroelectricity, one of the least-known properties of solid materials 6, is the manifestation of the temperature dependences of the spontaneous polarization of certain solids which may be either single crystals or polycrystalline aggregates 7. The Greek word pyro denotes heat or high temperature. 3

16 Chapter (1) Figure 1.2: If a pyroelectric crystal with an intrinsic dipole moment (top) is mounted into a circuit with electrodes attached on each surface (middle), an increase in temperature T prompts the spontaneous polarization P S to decrease as the orientation of the dipole moments, on average, diminishes. A current flows to compensate for the change in bound charge that accumulates on the crystal edges. Adopted from ref. [6]. A simple structural model of a pyroelectric substance may be developed as follows 7. In order to be pyroelectric, each structural unit of the substance (unit cell or coordination polyhedron) must have an electric dipole. This dipole is produced because the centers of positive and negative charge do not coincide 1,8. If the dipoles throughout the sample are aligned in such a way that self cancellation among them does not occur, the material will exhibit an electrical polarization called the spontaneous polarization. The word spontaneous means that the polarization has a non-zero value in the absence of an external electric field. If the temperature of the pyroelectric material remains constant for a sufficiently long time, surface charges will accumulate on the substance which will mask the internal spontaneous polarization. An increase or decrease in temperature T may change the strength of the dipoles. The surface charges will then redistribute themselves in order to compensate the new internal spontaneous polarization. This redistribution may be observed, for example, by connecting an ammeter between conductive electrodes placed on appropriate surfaces of the substance, Figure 1.2. The observed flow of electric charges P S is the pyroelectric effect 7 : P S = p T (1.3) where p is the pyroelectric coefficient. 4

17 Introduction Ferroelectricity Ferroelectrics or Seignette-electrics are a subgroup of the pyroelectrics 9. Their outstanding property is the reversibility of the permanent polarization by an electric field 1. This process is known as switching, and is accompanied by hysteresis loop, Figure 1.3. In many ways these materials are electrical analogues of ferromagnetics, in which the magnetization is reversed by a magnetic field. In some practical ways ferroelectrics differ from ferromagnetics, and in their fundamental mechanism they are totally different. The name ferroelectrics arose from the analogy with ferromagnetics. But there is no iron in ferroelectrics as the prefix ferro- might seem to imply. In the past, these materials are called Seignette-electrics, after P. de la Seignette, who first prepared Rochelle salt, the first crystal known to be ferroelectric. Figure 1.3: Hysteresis loop and domain switching. Three prominent properties of ferroelectrics are: reversible polarization (e.g. P E) hysteresis, anomalous properties (e.g. ε r and tan δ T) and non-linearities (e.g. ε r E) 1. Most ferroelectrics cease to be ferroelectric above a temperature T C known as the Curie transition temperature. The permittivity ε r rises sharply to a very high peak value at the temperature T C. The permittivity values of most ferroelectrics are high even at room temperature. The ferroelectric phase transition can be of a first or second order. In both cases the transition may be regarded as a transition from one branch (P S = ) 5

18 Chapter (1) into another branch (P S ) of the same free energy function. The dielectric constant follows a Curie-Weiss law in the unpolarized phase: ε = C r (1.4) T T o where T o and C are the Curie-Weiss temperature and Curie constant, respectively, and ε r is the dielectric constant. The parameter T o equals to the Curie transition temperature T C only for the case of a continuous transition. The phase of the ferroelectric above T C is often termed as paraelectric. Polar crystals whose sublattices characterized by an equal and opposite polarization orientation are called antiferroelectrics. There are also crystals that exhibit at a given temperature antiferroelectric properties along one axis and ferroelectric properties along another axis. They are called ferrielectrics 2. A schematic representation of the dielectric family tree can be demonstrated as shown in Figure 1.4. Table 1.1 summarizes the history of the major discoveries in the field of polar materials. Dielectrics Polar Crystals 32 point group Piezoelectrics 2 point group Pyroelectrics 1 point group Ferroelectrics Non-Polar Amorphous Non-Piezoelectrics 12 point group Non-Pyroelectrics Non-Ferroelectrics Antiferroelectrics Ferrielectrics T C Paraelectric Figure 1.4: Dielectrics family tree Applications of Polar Materials The applications for ferroelectric ceramics are manifold and covering all areas of our workplaces, homes, automobiles and computers. Similar to most materials, the successful application of these piezoelectric, pyroelectric, ferroelectric, and electrostrictive ceramics and films are highly dependent on the relative ease with which they can be adapted to useful and reliable devices. Table 1.2 summarizes some applications of ferroelectrics, pyroelectrics and piezoelectrics. 6

19 Introduction Table 1.1: Notable events in the history of polar materials (modified after adoption from ref. [3]). Timeline Event Timeline Event 1824 Pyroelectricity discovered in Rochelle 1961 PMN relaxor materials reported salt 188 Piezoelectricity discovered in Rochelle salt, quartz, and other minerals 1964 Oxygen/atmosphere sintering for FEs developed 1912 Ferroelectricity first proposed as 1964 FE semiconductor (PTC) devices developed property of solids 1921 Ferroelectricity discovered in Rochelle salt 1967 Optical and E/O properties of hot-pressed FE ceramics reported 1935 Ferroelectricity discovered in KH2PO Terms ferroic and ferroelasticity introduced 1941 BaTiO 3 high-k (>12) capacitors developed 1969 Optical transparency achieved in hot-pressed PLZT ceramics 1944 Ferroelectricity discovered in ABO3-type perovskite BaTiO PLZT compositional phase diagram established, Pat. No BaTiO3 reported as useful piezo transducer, Pat. No Useful E/O properties reported for PLZT, Pat. No Phenomenological theory of BaTiO 3 introduced 1973 Oxygen/atmosphere sintering of PLZT to full transparency 1949 LiNbO 3 and LiTaO 3 reported as FE 1977 FE thin films developed 1951 Concept of antiferroelectricity introduced 1978 Engineered (connectivity designed) FE composites developed 1952 PZT reported as FE solid-solution system, phase diagram established 198 Electrostrictive relaxor PMN devices developed, Pat. No PbNb 2 O 6 reported as FE 1981 Sol gel techniques developed for the preparation of FE films 1954 PZT reported as useful piezo transducer, Pat. No Photostrictive effects reported in PZT and PLZT 1955 PTC effect in BaTiO3 reported 1991 Moonie piezo flextensional devices developed, Pat. No Chemical coprecipitation of FE materials introduced 1992 RAINBOW piezo bending actuators developed, Pat. No Alkali niobates reported as FE 1993 Integration of FE films to silicon technology, Pat. No BaTiO 3 barrier layer capacitors developed 1997 Relaxor single-crystal materials developed for piezo transducers 1959 PZT 5A and 5H MPB-type piezo compositions, Pat. No New modified PZT phase diagram containing monoclinic ferroelectric phase F M reported Lattice dynamics theory for FE materials, soft modes introduced 21 Ferroelectric Bi-Cuprate glasses prepared 12 Table 1.2: Piezoelectric, pyroelectric, and ferroelectric ceramics applications 3,13,14,15. Generators Motors Hydrophones and microphones Actuators (micro and macro) Phonograph cartridges Loud speakers, tweeters Gas igniters flash bulbs Camera shutters, autofocusing buzzers Accelerometers Ink jet printers Power supplies Fish finders Photoflash actuators Micropositioners Piezoelectric pens Valve controllers Impact fuzes Pumps Composites Video head positioners Detection system in machinery Nebulizers Ultrasonic motors Motor/Generator Combination devices Piezoelectric tans Relays Sonar Ranging transducers Resonant Devices Non-destructive testing Medical ultrasound Ultrasonic cleaners Fish finders Ultrasonic welders Fiters Filters (IF, SAWs) Piezo transformers Transformers Delay lines Pyroelectric Applications Echo sounding Radios, TVs and remote control Intruder alarms (pyroelectric detectors) Fire alarms Ferroelectric Switching Pollution monitoring and gas analysis Laser detectors Non-volatile FE memory NVFRAM Pyroelectric thermal imaging High permittivity DRAM Pyroelectric vidicon 7

20 Chapter (1) 1.3. PEROVSKITES Perovskite (calcium titanium oxide, CaTiO 3 ) is a relatively rare mineral occurring in orthorhombic crystals. Perovskite was discovered in the Ural mountains of Russia by Gustav Rose in 1839 and named for Russian mineralogist, L. A. Perovski ( ). Perovskite is also the name of a more general group of crystals which take the same structure 16. The perovskite structure is fairly simple, with a general formula ABO 3, Figure 1.5. The prototype structure is cubic, with a large cation A at the cube corners, a smaller cation B at the body centres, and oxygen O at the face centres. The structure can also be regarded as a set of BO 6 octahedra arranged in a simple cubic pattern and linked together by shared oxygen atoms, with the A atoms occupying the spaces between. The first ferroelectric perovskite to be discovered was 2,17 BaTiO 3. A B O 2 Figure 1.5: The perovskite structure PZT Lead Zirconate Titanate Pb(Zr,Ti)O 3 has a perovskite structure. The Ti and Zr atoms occupying, at random, the centres of the unit cell (B-sites) with O atoms located at the centre of the unit cell faces while Pb atoms occupying the corners of the unit cell (A-sites). The phase diagram of PZT was originally established by Jaffe et al. 17 as shown in Figure 1.6A. At high temperature PZT has the cubic perovskite structure. As the temperature is lowered, PZT undergoes a paraelectric-toferroelectric phase transition and the cubic unit cell is distorted due to the off-centre shift of the Zr/Ti ions with respect to the oxygen octahedra. The direction of the shifts is, however, different depending on the composition. For Ti-rich compositions PZT has tetragonal symmetry. For Zr-rich compositions, it becomes a rhombohedral. The most remarkable properties are those of compositions of almost equal amount of Zr and Ti, around the steep boundary dividing the rhombohedral and tetragonal regions, known as the morphotropic phase boundary MPB, where the piezoelectric coefficients and dielectric properties show an anomalous maximum 17. 8

21 Introduction Temperature ( o C) P C F R(HT) T C Line Morphotropic Phase Boundary (MPB) F T! Jaffe et al. " Noheda et al. A O F R(LT) PbZrO 3 PbTiO 3 (mol. %) PbTiO 3 Figure 1.6A: Phase diagram of the PbZrO 3 -PbTiO 3 (PZT) system. The various regions in the phase diagram correspond to: a cubic paraelectric phase (P C ); a tetragonal ferroelectric phase (F T ); two rhombohedral ferroelectric phases F R (LT) and F R (HT); an orthorhombic antiferroelectric phase A O ; and a tetragonal antiferroelectric phase (A T ) 17. Figure 1.6B: A modified PZT phase diagram according to Noheda et al 18. Region assigned by F M indicates ferroelectric the monoclinic phase 18. Noheda et al. 18 have discovered a monoclinic phase in Pb (Zr 1-x /T x )O 3 with.45 x.52 and associated with that a new phase diagram has been reported 18. The monoclinic region forms a narrow triangle in between the tetragonal and rhombohedral phases, Figure 1.6B, with a vertical rhombohedral monoclinic boundary and a slightly inclined tetragonal monoclinic boundary Soft and Hard PZT Although the maximum piezoelectric effect was found in the composition of pure PZT at the MPB of Zr/Ti = 52/48, in practice, PZT ceramics are often modified to meet certain requirements for various applications. Basically, three types of additives have been employed in the compositional modification of PZT 5,17,19,2. These typical additives and their effects on the piezoelectric properties of PZT are summarized in Table 1.3. The first additives used in compositional modification of PZT are the isovalent ones, i.e. Pb 2+, Ti 4+ and Zr 4+ are replaced partially by other cations with the same chemical valence and similar ionic radii as those of the replaced ions. The major effect of such isovalent additives is lowering of the Curie temperature, giving enhanced permittivity. The second additives are donor type substitutions, known as soft dopants. Ions for soft doping include La 3+, Nd 3+ and other rare earth ions, such as Sb 3+, Bi 3+, Nb 5+, Sb 5+, W 6+ etc. The most outstanding changes caused by soft doping are increases in piezoelectric coupling coefficient, relative permittivity and 1 3 increases in bulk electrical resistivity. In general, when the ions with larger ionic radii, such as La 3+, Nd 3+, etc., occupy the A-sites to replace Pb 2+ ions, extra positive charges are introduced into 9

22 Chapter (1) the lattice due to the fact that the valence of the doping ions is higher than that of Pb 2+ ions. A Pb vacancy is created in the lattice to maintain electroneutrality. When ions with smaller ionic radii, such as Nb 5+, Ta 5+, etc., enter into the perovskite lattice, they occupy the B-sites to replace Zr 4+ or Ti 4+ ions. Since the doping ions have a higher valence than (+4), extra positive charges enter the lattice and again Pb vacancies have to be created to ensure electroneutrality. These Pb vacancies make the transfer of atoms easier than in a perfect lattice; thus domain motions can be caused by a smaller electrical field (or a mechanical stress). The third additives are acceptor type substitutions, or known as hard dopants. Hard doping ions in PZT include K + and Na +, which occupy the A-sites and Fe 2+, Fe 3+, Co 2+, Co 3+, Mn 2+, Mn 3+, Ni 2+, Mg 2+, Al 3+, Ga 3+, In 3+, Cr 3+ and Sc 3+, etc., which occupy the B-sites in the perovskite structure. Unlike soft doping, hard doping increases the hardness of PZT properties, i.e. lowers the dielectric constant, lowers the dielectric loss tan δ, increases the coercive field, and results in high mechanical quality factor Q m, a slightly lower k p and a lower bulk resistivity. When hard doping ions with a lower positive valence replace metal ions with a higher positive valence, oxygen vacancies are created in the lattice, on account of the requirements of electroneutrality. For example, when two Pb 2+ ions are replaced by two K + ions, or when two Zr 4+ or Ti 4+ ions are replaced by two Fe 3+ ions, one oxygen vacancy is created. Thus two hard doping ions may cause the creation of one oxygen vacancy which makes a defect dipole with the acceptor dopant ion. Such defect dipoles inhibit the domain wall motion. The inhibition of domain motion reduces the dielectric loss and therefore Q m is enhanced. Table 1.3: Typical additives to PZT and their major effects on piezoelectric properties. Ionic radii, in nm, are given in parentheses. Doping Additives Major Effects Isovalent additives Ba 2+ (.134) or Sr 2+ (.112) for Pb 2+ (.132) Sn 4+ (.71) for Zr 4+ (.68) or Ti 4+ (.79) Lower T C Higher ε r Soft dopants La 3+ (.122), Nd 3+ (.115), Sb 3+ (.9), Bi 3+ (.114), Th 4+ (.11) for Pb 2+ (.132) Nb 5+ (.69), Ta 5+ (.68), Sb 5+ (.63), W 6+ (.65) for Ti 4+ (.68) or Zr 4+ (.79) Higher permittivity Higher k p Much lower Q m higher resistivity Hard dopants K + (.133) or Na + (.94) for Pb 2+ (.132) Fe 3+ (.67), Al 3+ (.57), Sc 3+ (.83), In 3+ (.92), Cr 3+ (.64) for Ti 4+ (.68) or Zr 4+ (.79) Lower ε r Lower tan δ Lower k p Much higher Q m 1

23 Introduction 1.5. FERROELECTRIC DOMAINS The electric dipoles in a ferroelectric crystal (or a grain in a ferroelectric film or ceramic) are usually not uniformly aligned throughout the material along the same direction when the crystal is cooled through the ferroelectric phase-transition temperature 21. Directions along which the polarization will develop depend on the electrical and mechanical boundary conditions imposed on the sample. The regions of the crystal with uniformly oriented electric dipoles are called ferroelectric domains. The boundary between two domains is called a domain wall. The walls that separate domains with oppositely oriented polarization are called 18 o walls and those that separate regions with mutually perpendicular polarization are called 9 o walls, Figure 1.7. Cubic (PE) phase T>T C E d P S 18 o P S Ferroelectric Domain Walls P S P S 9 o P S Tetragonal (FE) phase T<T C P S Ferroelastic Domain Walls Figure 1.7: The formation of 18 o and 9 o ferroelectric domain walls in a tetragonal perovskite ferroelectric, after Damjanovic 21. The ferroelectric domains form to minimize the electrostatic energy of the depolarizing fields and the elastic energy associated with the mechanical constraints to which the ferroelectric material is subjected to as it is cooled through the paraelectric-ferroelectric phase transition 2,22. Onset of spontaneous polarization at the transition temperature leads to the formation of surface charges. This surface charge produces an electric field, called the depolarizing field E d, which is oriented oppositely to P S (Figure 1.7). The electrostatic energy associated with the depolarizing field may be minimized if the ferroelectric splits into domains with oppositely oriented polarization, Figure 1.7. Splitting of a ferroelectric crystal into domains may also occur due to the influence of mechanical stresses, also shown in Figure

24 Chapter (1) Poling of Ferroelectrics If the direction of the spontaneous polarization through the material is random or distributed in such way as to lead to zero net macroscopic polarization, the piezoelectric effects of individual domains will cancel out and such materials will not exhibit piezoelectric or pyroelectric effect, which requires a non-zero spontaneous polarization 23. Polycrystalline ferroelectric materials (ceramics) may be brought into a polar state by applying a strong electric field (1 1 kv/cm), usually at elevated temperatures. This process, called poling, can reorient domains within individual grains along those directions that are permissible by the crystal symmetry and that lie as close as possible to the direction of the field. A poled polycrystalline ferroelectric exhibits piezoelectric and pyroelectric properties, even if many domain walls are still present. After the removal of the poling field, a ferroelectric material possesses macroscopic polarization, called spontaneous polarization P S Domain Switching The mechanism of polarization switching has been studied in detail for many bulk and thin-film ferroelectrics 2,22. Yet, there is no universal mechanism which would be valid for polarization reversal in all ferroelectrics. The polarization reversal takes place by the growth of existing antiparallel domains, by domain-wall motion, and by nucleation and growth of new antiparallel domains 23. Experimentally 24,25,26, polarization switching is usually studied by applying a constant voltage to a polarized sample of opposite polarity to the voltage. Figure 1.8 shows a typical applied constant voltage and switching current, i dp/ dt, for a ferroelectric. The field, applied antiparallel to the polarization, switches polarization from state P S to +P S. E i RC Peak t.1i max i max t Figure 1.8: Switching current i s versus time t s under constant electric field 24,25,26. 12

25 Introduction The total current consists of two parts. The first spike is due to the fast linear response of the dielectric and the bell-shaped curve is the current due to polarization switching. The total area under the curve is equal to t i ( t) dt = ε ε AE + 2AP (1.5) o r S where i is the switching current density and A is the area of electrodes. Once the ferroelectric is switched the same pulse may be applied again, this time parallel to polarization to obtain only the transient from the fast response of the dielectric. The area under the fast transient spike is ε oε rae and from this experiment and eq. 1.5 the remanent polarization may be calculated. In some practical cases, the first term can be neglected 22. The kinetics of polarization reversal may be described by measuring the switching time t S for different amplitudes E of electric field pulse. Since the current decreases exponentially it is difficult to measure the total switching time, and t S is usually taken as the value where current falls to.1 i 22 max. The switching time can be expressed as t = t (1.6) E s e α/ where α is called the activation field. At very high fields this relation changes to a n power law of the form ts E where n depends on the material. The maximum switching current may be described by i = i (1.7) E s e α / Constants α, n, t and i are temperature dependent, with switching time decreasing as the Curie point is approached AIM OF THE THESIS The polarization is the true measure of the degree of ferroelectricity 17. Generally, ceramic materials with a high value of remanent polarization show usable piezoelectric effects 17. Poling conditions required to achieve the optimal piezoelectric characteristics widely vary for different materials 17. Accordingly, the poling of piezoceramics after manufacturing is an important process. The major objective in all poling stages is to induce the maximum degree of domain alignment by using the lowest electric field, the closest temperature to room temperature and in the shortest possible duration. There is not yet a solid model or theory that governs the poling mechanism for a wide spectrum of piezoelectric materials. 13

26 Chapter (1) In this thesis we are studying the ferroelectric and piezoelectric properties and aftereffects in soft and hard PZT near the MPB. The properties can be influenced by various parameters either extrinsic such as poling or intrinsic such as grain size. Thus the ultimate aim of the study is to gain deep insight on mechanism of the poling processes in order to reach and understand the optimum functional behavior of these materials THESIS OUTLINE Poling of PZT materials is the main focus of the present thesis. The contents of the thesis are outlined as follows. In Chapter 2, 2 a description of the main setups used in the experimental work will be set out. Chapter 3 comes in two parts; in Part I the phenomenological behavior of the polarization of soft PZT (PXE52) will be studied as a function of different factors, considering the applied electric field is the main variable. In Part II, the grain size effect on the polarization phenomenology will be investigated. Chapter 4 comes also in two parts; in Part I the switching of soft PZT (PXE52) will be studied as a function of the applied field. In Part II, the grain size effect on switching process will be discussed. In Chapter 5, 5 the pyroelectric current and conduction current interplay will be studied. The resistivity as a function of temperature will be examined as well. In Chapter 6 the poling of hard PZT (PXE43) will be extensively investigated. Different techniques for rejuvenation will be presented. Lastly, the main conclusions of the thesis results will be elaborated in Chapter 7. 7 Suggested studies for future research will be also addressed. 14

27 Introduction REFERENCES 1 B. Tareev, Physics of Dielectric Materials, Moscow, Mir Publisher, M. E. Lines and A. M. Glass, Principles and Applications of Ferroelectrics and Related Materials Clarendon Press, Oxford, G. H. Haertling, J. Am. Ceram. Soc., 82, 797, J. F. Nye, Physical Properties of Crystals, Clarendon Press, Oxford, N. Setter, Piezoelectric Materials in Devices, N. Setter, Ceramics Laboratory, EPFL, S. B. Lang, Phys. Today, 58, 31, S. B. Lang, Sourcebook of Pyroelectricity, Gordon and Breach Science Pub., London, A.K. Jonscher, Dielectric Relaxation in Solids, Chelsea Dielectric Press, London, W. Kænzig, Ferroelectrics and Antiferroelectrics, Academic Press, New York and London, J. C. Burfoot, Ferroelectrics, Introduction to the Physical Principles, D. Van Nostrand Co. Ltd., London, B. Noheda, D. E. Cox, G. Shirane, J. A. Gonzalo, L. E. Cross and S-E. Park, Appl. Phys. Lett., 74, 259, A. A. Bahgat and T. M. Kamel, Phys. Rev., B63, 1211, S. L. Swartz, IEEE Trans., Elec. Insulat., 25, 935, R. W. Whatmore, Rep. Prog. Phys. 49, 1335, O. Auciello, J. F. Scott and R. Ramesh, Phys. Today, 51, 22, L. G. Tejuca, Properties and applications of perovskite-type oxides. New York, Dekker, B. Jaffe, W. R. Cooke and H. Jaffe, Piezoelectric Ceramics, Academic Press, New York, B. Noheda, D.E. Cox, G. Shirane, R. Guo, B. Jones, L. E. Cross., Phys. Rev., B63, 1413, A. J. Moulson and J. M. Herbert, Electroceramics, Chapman & Hall, London, Y. Xu, Ferroelectric Materials and Their Applications, North Holland, D. Damjanovic, The Science of Hysteresis, Volume 3, I. Mayergoyz and G. Bertotti (Eds.), Elsevier, J. C. Burfoot and G. W. Taylor, Polar Dielectrics and their Applications, Macmillan, London, D. Damjanovic, Rep. Prog. Phys., 61, 1267, W. J. Merz, Phys. Rev., 95, 69, E. Fatuzzo and W. J. Merz, Phys. Rev., 116, 61, E. Fatuzzo and W. J. Merz, Ferroelectricity, North-Holland, Amsterdam, FURTHER READING R. von Hippel, Dielectric Materials and Applications, John Wiley & Sons, Inc., New York, J. Moulson and J. M. Herbert, Electroceramics; Materials, Properties, Applications, John Wiley & Sons LTD, London, 23. S. Zheludev, Physics of Crystalline Dielectrics, Plenum, New York, T. Mitsui et al, An Introduction to the Physics of Ferroelectrics, New York: Gordon and Breach, J. M. Herbert, Ceramic Dielectrics and Capacitors, Gordon and Breach, New York, U. Böttger, Polar Oxides, Edited by R. Waser, U. Böttger and S. Tiedke, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, 25. K. Uchino, Ferroelectric Device, Dekker, New York, 2. A. Strukov and A. P. Levanyuk, Ferroelectric Phenomena in Crystals, Springer, T. Ikeda, Fundamentals of Piezoelectricity, Oxford University Press, J. van Randeraat and R. E. Setterington, Piezoelectric Ceramics, Mullard Ltd, London, A. F. Devonshire, Phil. Mag., 4, 14, A. F. Devonshire, Phil. Mag., 42, 165, A. F. Devonshire, Adv. Phys., 3, 85, A. S. Bhala, R. Guo and R. Roy, Mat. Res. Innovat., 4, 3, 2. R. von Hippel, Rev. Mod. Phys., 22, 221, 195. W. G. Cady, Piezoelectricity, Dover Publications, INC. New York, F. S. Galasso, Structure, Properties and Preparation of Perovskite-Type Structure, Pergamon Press, A. Safari, R. K. Panda and V. F. Janas, Key Eng. Mater., 35, 122, J. Ravez, C.R. Acad. Sci. Paris Serie IIc, Chimie/Chemistry, 3, 267, 2. T. Lee and I. A. Aksay, Cryst. Growth Des., 1, 41,

28

29 CHAPTER 2 MATERIALS AND METHODS ABSTRACT In this chapter, detailed descriptions of the main setups used in the experimental work of this thesis are outlined.

30 Chapter (2) Materials and Methods 2.1. SOFT AND HARD PZT Two main types of PZT ceramic have been used in the present thesis. A modified donor doped soft PZT (PXE52) (Pb x Sr 1-x Zr.415 Ti.585 O 3 ) and Fe doped hard PZT (PXE43) (Pb y Sr 1-y Zr.532 Ti.468 O 3 ) were provided by Morgan Electro Ceramics BV, Eindhoven, The Netherlands. For studying the grain size effect on the poling process, the following procedure has been followed for both materials. The powder was pressed into bars of at least 15 mm in diameter and sintered at different fixed temperatures ( o C), with an interval of 25 o C, in a tube furnace with flowing oxygen. The following sintering profile was applied for all bars. The bar was heated from room temperature to the desired temperature in 8 hours and held at this temperature for 4 hours and then cooled down to room temperature in 8 hours. The sintered bars were diced into discs of 1 mm in diameter. Next, the discs were subjected to clean firing at 7 o C for 48 hours, cleaned in methanol in an ultrasonic bath for two minutes, washed in water and soap and then dried in air. Finally, the discs were electroded with Ni by sputtering for electrical measurements. Before starting the experimental work, a set of preliminary experiments was carried out to verify the material s properties reproducibility as a consequence of the material s chemical homogeneity. The description of these experiments is summarized in Appendix A FERROELECTRIC HYSTERESIS LOOP The most important characteristic of ferroelectrics is their polarization reversal (switching) with an electric field. One consequence of this switching in ferroelectric materials is the occurrence of the ferroelectric hysteresis loop. The hysteresis loop can be observed experimentally by using different techniques. We used the Sawyer- Tower 1 setup. A home-made computer controlled Sawyer Tower circuit, as schematically represented in Figure 2.1, with an applied frequency range from.1 to 2 Hz and an electric field amplitude of 15 kv/cm was used. Both P and E were obtained by measuring the voltage drop V Y and V X over a reference capacitor and the sample, respectively. For P we use C r VY P = (2.1) A 18

31 Materials and Methods where C r is the capacitance of the reference capacitor (1 µf), V Y is the Y axis voltage which is corresponding to the polarization of the sample, A is the electroded area. For E we use VX E = (2.2) d V X is the X axis voltage which is corresponding to the applied electric field, and d is the sample thickness. Signal Amplifier PC Interface Card Input ± 1V 3x Output V x = ± 3V C x V x V ref A/D X!V x /1 Y! V ref C ref GPIB Figure 2.1: A computer controlled Sawyer-Tower circuit. For observing the hysteresis loop for hard piezomaterials, a higher electric field was required. Hysteresis loops of hard PZT were observed using a computercontrolled virtual ground method (Radiant Technologies Inc., RT6 HVA-z 2V amplifier and RT6 HVS-z high voltage input test system) using the fast mode (.4 ms) at different temperature steps (RT 15 o C). The maximum applied electric field was 5 kv/cm. Silicone oil (Wacker Silicone Fluid, AP 15) was used as an insulating medium POLING The polarization current measurements were carried out using a homemade setup schematically represented in Figure 2.2. The electric field was applied using a Keithley built-in voltage source with maximum field of 3 kv/cm and a response time of 5 ms (up to 1 V) and 8 ms (up to 1 V). Upon applying the electric field, the polarization current is recorded, using a Keithley 6517 Electrometer, as a function of time at constant temperature. The polarization was then calculated by integrating the poling current density J pol over the whole measuring time t according to t pol P = J t (2.3) pol d 19

32 Chapter (2) % $ #! " + Voltage Current Time Temperature ) ' & + - ( * Figure 2.2: The setup used for the poling experiment.! the sample holder, " the sample, # the thermocouple, $ electrical shielding, % sample leads, & Keithley electrometer, ' connection to Keithley, ( Parameters measured by Keithley, ) grounding, * thermocouple connection PYROELECTRIC CURRENT The pyroelectric measurements have been performed using the direct method 2. The poling setup was modified for measuring the pyroelectric current after poling as illustrated in Figure 2.3. % $ & #! " + Current (na) Temperature ( o C) + - ( ) ' * Figure 2.3: The setup used for the pyroelectric current experiment. The numbers refers to the components as indicated in Figure 2.2. Using this method a poled sample is situated in an oven heated at a constant heating rate dt/dt of.58 o C/s. The sample s two electrodes were short-circuited via Keithley 6517 electrometer to measure the pyroelectric electric current. The pyroelectric coefficient p was calculated in terms of the pyroelectric current I pyro by 2

33 Materials and Methods I 1 pyro dt p = (2.4) A dt where A is the electrode area. The spontaneous polarization P S was calculated by integrating the pyroelectric coefficient P = p dt (2.5) S 2.5. DC CONDUCTIVITY The dc conductivity as a function of temperature under various electric fields has been carried out using the same setup as used for the poling experiments, see Chapter DIELECTRIC AND PIEZOELECTRIC PARAMETERS The dielectric parameters (ε r and tan δ) of poled samples were measured using an impedance analyzer (HP4294A Precision Impedance Analyzer) at 1 khz in the temperature range from room temperature up to 4 o C. Among several techniques for investigation of the piezoelectric response, the resonant technique is one of the most efficient ones; therefore it is commonly used to determine material coefficients of piezoelectric ceramics 3,4. When excited at the resonance frequency f r, the piezoelectric disc will resonate at a greater amplitude. For the anti resonant frequency f a, the impedance of the piezoelectric disc is at a maximum and the oscillation amplitude is at a minimum 5. By measuring the resonance and anti resonance frequencies f r and f a, along with dielectric parameters, the electromechanical coupling factors k p, k 33, k 31 and piezoelectric charge constants d 33 and d 31 can be calculated using the equations as listed in Appendix B. 21

34 Chapter (2) REFERENCES 1 C. B. Sawyer and C. H. Tower, Phys. Rev., 35, 269, S. B. Lang, Source Book of Pyroelectricity Gordon and Breach Science Publishers, London, B. Jaffe, W. R. Cook and H. Jaffe, Piezoelectric Ceramics, Academic Press New York, A. J. Moulson and J. M. Herbert, Electroceramics, Chapman & Hall, London, T. L. Jordan and Z. Ounaies, Piezoelectric Ceramics Characterization, National Aeronautics and Space Administration (NASA), ICASE Report No ,

35 CHAPTER 3 POLING OF SOFT PZT Part I: Field Effect ABSTRACT The properties of piezoelectric ceramic materials are strongly dependent on the degree of polarization as set by the poling process. In this chapter, a soft piezoceramic PZT material was polarized using different poling conditions. The hysteresis loop, the polarization current, and pyroelectric current measurements were used to evaluate the polarization state of the material. The hysteresis loop was monitored using a homemade computer controlled Sawyer-Tower circuit. The polarization current was recorded during the poling process at different applied electric fields, poling time and temperature. The pyroelectric coefficient and the polarization were calculated from the pyroelectric current. The polarization calculated from these data was in excellent agreement with the polarization as calculated from the poling current. The relative permittivity and loss factor were measured as a function of temperature after different poling conditions. The effects of the various poling conditions on the dielectric and ferroelectric properties of the soft PZT are discussed. It is shown that, contrary to common practice, poling at a field slightly larger than the coercive field is adequate to reach full polarization at room temperature.

36 Chapter (3) Part I Poling of Soft PZT Field Effect! 3I.1. INTRODUCTION Over decades, great attention has been paid to (PbTi x Zr 1-x O 3 ) solid solutions for their extremely strong piezoelectric effect 1-4. The coupling factor and relative permittivity are the highest near the morphotropic phase boundary 1,2,4 making these materials extensively used in actuator devices and microelectromechanical systems (MEMS) 1-3. Poling conditions required to achieve the optimal piezoelectric characteristics widely vary for different materials 1. Generally, the major objective in all poling stages is to induce the maximum degree of domain alignment by using the lowest electric field, at a temperature as close as possible to room temperature (RT) and in the shortest possible time. Although the poling of piezoceramics after manufacturing is a crucial process and the effect of poling conditions on the piezoelectric properties has been studied by many authors 5-9, there is not yet a solid model or theory that governs the poling mechanism for a wide spectrum of piezoelectric materials. In this chapter, we give preliminary results for studying the effect of the poling conditions on the polarization state as well as on the dielectric and ferroelectric properties. The polarization state has been evaluated by different techniques via which an attempt has been made to explain the poling process. 3I.2. THEORY 3I.2.1. Dielectric Response The most obvious physical reason for time-dependent dielectric response is the inevitable inertia of polarization processes 1. By contrast to a dielectric material system the response of free space is instantaneous and therefore the induced charge (ε o E) arising from the response of free space follows the field instantaneously. Thus the charges induced at the sample electrodes will be given by the sum of an instantaneous free space contribution and the delayed material polarization () t E P () t D ε + (3I.1) = o Here D(t) denotes dielectric displacement and it represents the total charge density induced at the electrodes. Assuming that a homogeneous electric field E is applied to a dielectric material, the current density J(t) through the surface of the material is! This part has been published as: T. M. Kamel, F. X. N. M. Kools and G. de With, J. Eur. Ceram. Soc., 27, 2471,

37 Poling of Soft PZT; field effect the sum of the displacement current dd/dt and the conduction current E/ρ and can be written as: () dd t J () t = + d t E ρ (3I.2) where ρ is the dc electrical resistivity. Substituting eq. 3I.1 in eq. 3I.2 and taking into account that E is constant during poling, leads to: () t dp J () t = + d t E ρ (3I.3) The polarization current dp/dt arises from the tendency of the polarizing species in the material to respond in a delayed manner to the exciting field and must go to zero at infinitely long time. In this sense this current characterizes the most important property of the dielectric system. On the other hand, the conduction current E/ρ is constant with time and arises from a continuous movement of free charges across the dielectric material from one electrode to another and this current does not change in any way the centre of gravity of the charge distribution in the system. 3I.2.2. Polarization Switching Grains in ferroelectric ceramics and polycrystalline films contain always multiple domains. Each domain has its own polarization direction. If the polarization directions through the material are random or distributed in such a way that it leads to a zero net polarization, the pyroelectric and piezoelectric effects of individual domains will cancel and such material is neither pyroelectric nor piezoelectric. Polycrystalline ferroelectric materials can be brought into a polar state by applying an adequate electric field. This process, which is referred to as poling, can reorient domains within individual grains in the direction of the field. A poled polycrystalline ferroelectric exhibits pyroelectric and piezoelectric properties, even if many domain walls are still present. By definition, poling is the dipole alignment by the electric field. The polarization after the removal of the field (at zero field) is called remanent polarization, P r. The field necessary to bring the polarization to zero is called the coercive field, E 11 C. The spontaneous polarization P S can be defined as the surface charge density 16,31 or the dielectric displacement of polar material when 2 ε >> r 1, or the dipole moment per unit volume 2,21. It should be noted that the coercive field E C that is determined from the intercept of the hysteresis loop with the field axis is not an absolute threshold field 12. If a low electric field is applied opposite to the polarization over a long time, the polarization will eventually switch to the opposite direction 11,12. 25

38 Chapter (3) Part I The mechanism of polarization switching has been studied in detail for many bulk and thin-film ferroelectrics However, the issue is not yet well generalized and there is no universal mechanism valid for polarization reversal in all ferroelectrics. The switching in ferroelectrics takes place by nucleation of domains, characterized by the nucleation time t n, the time necessary to form all nuclei and domain wall motion, characterized by the domain wall motion time t d, the time necessary for one domain to move through the sample. As proposed by Merz the total switching time can then be approximated by d ts tn + td = tn + (3I.4) µ E where d is the distance that the wall travels, and µ is the mobility of the domain wall. In Merz s model it was assumed that the nucleation of new domains is governed by a statistical law, in which at low fields, the probability of forming new domains p n depends exponentially on the applied field in the following way: p n = p / E oe α and, since t n ~ 1/p n, t n = t (3I.5) / E o e α Then, the maximum switching current may be described by J = J where J = ( J ) (3I.6) max E= E max e α / Parameters α, p o, t o, and J are temperature dependent, with switching time decreasing as the Curie point is approached 11,13-15, The parameter α is called the activation field, and can be considered as the threshold of the field needed to initiate nucleation 15. In the low-field range, the experimental curve can be described by eq. 3I.5 only while in the high field eq. 3I.4 is required. Sincet s = tn + td, one can conclude that the switching time t s is determined by the slower of the two mechanisms (nucleation or domain wall motion). At low fields the rate of nucleation is low so that the switching is primarily governed by the nucleation (t n >> t d ) leading to an exponential law for the switching time. On the other hand, we assume that at high fields the rate of nucleation is extremely large so that the switching time is primarily determined by the velocity of the domain walls (t d >> t n ). 26

39 Poling of Soft PZT; field effect The field applied antiparallel to the polarization switches the polarization from state P to +P. In his experiment, Merz showed that the total polarization current curve consists of two parts 12,13. The first part is due to the fast linear response of the dielectric and the second part is the current due to polarization switching. The total area under the curve is equal to t s () t dt E 2P J = ε o + (3I.7) Although Merz s theory was developed for single crystals, it has been applied to polycrystalline materials as well, e.g. [18]. 3I.3. EXPERIMENTAL Non-poled polycrystalline ceramic samples of soft PZT (PXE52, donor doped PbZr.415 Ti.585 O 3 ) were obtained from Morgan Electro Ceramics BV, Eindhoven, The Netherlands. The dimensions of the samples under study were 5x5x.2 mm 3. The first step in examining the poling effect was to reveal the domain structure. The microstructure of poled as well as non-poled samples was investigated using scanning electron microscopy (JEOL, JSM 84A, Japan). First, the sample was molded in conductive resin, ground and successively polished. Then the sample surface was etched using 95% water, 5 % HCl, 5 6 drops of HF as an etching agent 19. Next, the sample surfaces were coated with Gold by sputtering. The cross-section normal to the polarization direction was investigated. The measurements of polarization can be directly made by reversing the spontaneous polarization with the applied electric field. A computer controlled Sawyer Tower circuit was used to examine the ferroelectric hysteresis loop. The remanent polarization P r and coercive field E C were measured as a function of temperature. The polarization current measurements were carried out using a Keithley 6517 Electrometer on initially non-poled samples, provided with Ni-sputtered electrodes covering the complete surface, at different applied electric fields and temperatures. The electric field was applied using the Keithley built-in voltage source and varied from 2.5 kv/cm to 25 kv/cm with a response time of 5 ms (1 V) and 8 ms (1 V). For high temperature measurements, a set-up in a Linn Elektronika oven was used. The temperature was measured using a type-k thermocouple. The poling current was continuously recorded while the sample temperature was kept at a fixed value. The polarization was then calculated by integrating the experimental poling current density J pol over the whole measuring time t 1. P t = J pol d t (3I.8) 27

40 Chapter (3) Part I After the poling process, the sample was short-circuited. The pyroelectric measurements were carried out using the direct method 2. Using this method a sample poled at least a day in advance was situated in an oven heated at a constant heating rate dt/dt of.58 o C/s. The two electrodes were shortcircuited via Keithley 6517 electrometer to measure the pyroelectric electric current. The pyroelectric coefficient p was calculated in terms of the pyroelectric current I pyro by I A 1 pyro dt p = (3I.9) dt where A is the electrode area. The spontaneous polarization P S was calculated by integrating the pyroelectric coefficient P = pdt (3I.1) S The dielectric parameters ( ε r, tan δ) of poled and non-poled samples were measured using RCL Bridge (4284A Precision LCR Meter) at 1 khz in the temperature range from room temperature up to 25 o C. 3I.4. RESULTS 3I.4.1. Microstructure Figures 3.1A-1B show the SEM micrographs of a nonpoled and poled sample, respectively. The alignment of the domains is clearly revealed after the application of the applied electric field (15 kv/cm). The average grain size is 4.5 µm as determined by the mean linear intercept, counting about 1 grains. A B 5 µm 5 µm Figure 3I.13.1: SEM micrograph of (A) a nonpoled and (B) a poled sample at an applied electric field of 15 kv/cm. 28

41 Poling of Soft PZT; field effect 3I.4.2. Hysteresis Loop The hysteresis loop was observed at 1 Hz to get the highest polarization. At room temperature the hysteresis loop showed that the maximum polarization P max is 42.5 µc/cm 2, the remanent polarization P r is 35.7 µc/cm 2 and the coercive field E C is 7.5 kv/cm. Both P r and E C were measured as a function of temperature (Figure 3I.2) and are typically decreasing with temperature, rapidly approaching (nearly) zero at 17 o C. This temperature is referred to as the Curie temperature T C. Worthwhile to mention is that P r still survives even above T C which means that the material shows relaxor-like behavior 11, Also the hysteresis loop was traced with temperature (Figure 3I.3). As the temperature increases the loop collapses and becomes slimmer with vanishing P r and E C but still surviving even above T C. P r (µc/cm 2 ) Coercive Field, E 2 C Remanent Polarization, P r Temperature ( o C) 1 8 E C (kv/cm) 6 4 Figure 3I.2: Remanent polarization (P r ) and coercive field (E C ) as a function of temperature. The applied frequency is 1 Hz. Polarization (µc/cm 2 ) o C 75 o C 15 o C o C Electric Field (kv/cm) 125 o C 18 o C Figure 3I.3: Hysteresis loop as a function of temperature. The polarization reversal with small E C and P r is still persistent after T C = 17 C. The applied frequency is 5 Hz. 29

42 Chapter (3) Part I 3I.4.3. Polarization Current The polarization current as a function of time at different low fields (E < E C ) (2.5 6 kv/cm) was measured at room temperature (Figure 3I.4). The polarization current is initially high and then a shows fast decay with time passing a hump at a certain time for certain field. The hump is moving towards shorter time as the applied field is increasing until it completely diffuses within the initial high pulse when the coercive field is approached. No steady state current is observed in this range of electric field meaning that the material is still being polarized. Polarization Current (na/cm 2 ) Low Field Regime (E < E C ) Electric fields (kv/cm): Time (s) REGION (I) Maxwell-Wagner Polarization REGION (II) Polarization Switching Figure 3I.4: Polarization current as a function of time at different applied poling fields (E < E C ). For each curve a different non-poled sample was taken. Worthwhile to mention is that at very low field no humps were observed in the time interval used. Apparently, it needed a longer time than 3x1 3 s. It was, however, found that by increasing the sample temperature, the hump appeared at reasonable time and moved towards shorter time as the temperature increases at a rate dependant on the field applied (Figures 3I.5 and 6). Polarization Current (na/cm 2 ) 1, 1, Poling Field: 2.5 RT o 6 o 8 o 9 o C Time (s) Figures 3I.5: Polarization current as a function of time at 2.5 kv/cm at different temperatures. Polarization Current (na/cm 2 ) Poling Field: 3.25 RT o 55 o 7 o C Time (s) Figures 3I.6: Polarization current as a function of time at 3.25 kv/cm at different temperatures. 3

43 Poling of Soft PZT; field effect For higher fields (E E C ), the shape of poling current curves looks very different. As the initial current is inversely proportional to the applied electric field (contrary to a low field case), a steady state current, directly proportional to the applied field, appeared after a relatively long time (Figure 3I.7). Polarization Current (na/cm 2 ) High Field Regime (E > E C ) Tail of REGION (II) Polarization Switching REGION (III) Steady State Conduction Current 1 Electric Fields RT Time (s) Figure 3I.7: Polarization current as a function of time at different applied poling fields (E >E C ). For each curve a different non-poled sample was taken. 3I.4.4. Pyroelectric Current The pyroelectric coefficient p was calculated using eq. 3I.9 (Figure 3I.8) and shows a very sharp peak at 168 o C which is associated with the ferroelectric transition. The other curve represents the temperature dependence of spontaneous polarization as calculated from eq. 3I.1. p (µc/cm 2. o C) Poling field: 6. kv/cm Heating rate: 3.5 o C/min. Pyroelectric Coeffiecient, p Polarization, P Temperature ( o C) Polarization (µc/cm 2. o C) Figure 3I.8: The pyroelectric coefficient and the calculated polarization for a prepoled sample. 31

44 Chapter (3) Part I The pyroelectric characteristic was employed to assess the polarization state of samples previously poled at different conditions of electric field and time. Samples poled at RT and 8 o C were subjected to heating at a constant rate and the pyrocurrent was monitored for both (Figure 3I.9). It is clear that the sample poled at RT (no hump in the poling current curve) has very low pyroelectric activity while the sample poled at 8 o C showed a significantly higher pyroelectric activity. Pyroelectric coefficient (µc/cm 2. o C) kv/cm, 8 o C 2.5 kv/cm, RT 1E Temperature ( o C) Figure 3I.9: Pyroelectric coefficient of the two samples poled at 2.5 kv/cm at room temperature and 8 C, respectively. The same procedure was repeated for different samples previously poled at the same electric field (4 kv/cm) at RT for 1 s and 25 s (Figure 3I.1). For 1 s poling the current didn t hump since the time was too short. For 25 s the hump was observed as expected. The pyrocurrent measured for the non-humping poling curve showed a much lower pyroelectric coefficient than the humping curve (Figure 3I.11). It should be noted that a further increase of the temperature during poling leads to rapid depolarization due to the pyroelectric effect, as will be elaborated elsewhere 32. As a final confirmation of the significance of the appearance of the hump in the poling curve as finger-print of the full polarization the following test has been carried out. Two samples were poled at 3.75 and 4.75 kv/cm for 1 3 s at RT. They both showed a humped poling curve (Figure 3I.12). The pyrocurrent was measured for both and it was found that for both samples the pyroelectric coefficient peaks were almost identical (Figure 3I.13). Figure 3I.14 shows a significant increase of residual polarization by increasing the poling field. The remanent polarization as observed by hysteresis loop vs. temperature measurements shows a higher magnitude than that of the calculated polarization from pyroelectricity. This difference can be attributed to the backswitching of some reversible domains after removing the field

45 Poling of Soft PZT; field effect Polarization Current (na/cm 2 ) 1, 1, kv/cm, RT, 1 s. 4. kv/cm, RT, 25 s Time (s) Figure 3I.1: Poling current of two samples poled at 4. kv/cm at room temperature for different poling time. Polarization Current (µa/cm 2 ) E-3 1E kv/cm, RT, 1 3 s kv/cm, RT, 1 3 s Time (s) Figure 3I.12: Poling current of two samples poled at 3.75 and 4.75 kv/cm at room temperature for 1 3 s. Pyroelectric coefficient (µc/cm 2. o C) kv/cm, RT, 1 s. 4. kv/cm, RT, 25 s. 1E Temperature ( o C) Figure 3I.11: Pyroelectric coefficient of the two samples poled as shown in Figure 3I.1. Pyroelectric coefficient (µc/cm 2. o C ) kv/cm, RT, 1 3 s kv/cm, RT, 1 3 s Temperature ( o C) Figure 3I.13: Pyroelectric coefficient of the two samples poled as shown in Figure 3I.12. Spontaneous Polarization, P S (µc/cm 2 ) Poling field (kv/cm) Temperature ( o C) Figure 3I.14: Spontaneous polarization as a function of temperature (after poling with different electric fields) as calculated from the pyroelectric coefficient. 33

46 Chapter (3) Part I 3I.4.5. Dielectric Properties Figure 3I.15 shows the temperature dependence of the relative permittivity at 1 khz for different samples poled at different poling fields including a non-poled sample. The relative permittivity typically drastically increases as the temperature increases to peak at 175 o C and thereafter decreases again. In the low temperature region (in the ferroelectric phase) the poling field (until 7.5 kv/cm) shows significant enhancement of the relative permittivity and no further enhancement is noticed for poling fields higher than 7.5 kv/cm. As the poling field enhances the relative permittivity near room temperature it suppresses the peak height at the Curie transition and has no influence in the paraelectric phase. Figure 16 shows the dielectric loss factor as a function of temperature. The poling field considerably reduces the loss factor till 7.5 kv/cm while no further effect was noticed at higher fields. As a common behavior in polycrystalline ferroelectric material the loss factor peaks at a temperature lower than the Curie transition temperature 21. For convenience, Table 1 summarizes the physical properties for PXE52. Dielectric Constant, ε r khz kv/cm 4.25 kv/cm 7.5 kv/cm 15. kv/cm Curie Transition tan δ khz kv/cm 4.25 kv/cm 7.5 kv/cm 15. kv/cm 25. kv/cm Temperature ( o C) Figure 3I.15: Relative permittivity as a function of temperature at 1 khz after poling with different poling electric fields Temperature ( o C) Figure 3I.16: Dielectric loss factor as a function of temperature at 1 khz after poling with different poling electric fields. Table 3I.1: PXE52 properties. Density (g/cm 3 ) 7.45 Grain size (µm) 4.5 Curie temperature ( o C) 168 (I pyro T) ; 17 (P r T) ; 175 (ε r T) Resistivity, ρ (Ω.cm) 3.5 x 1 12 Relative permittivity (1 khz/rt) 226 (nonpoled) ; 516 (after 25 kv/cm) tan δ (1 khz/rt).36 (nonpoled) ;.18 (after 25 kv/cm) Maxwell-Wagner time constant τ=rc, (s) ~ 2.6 Remanent polarization P r, (µc/cm 2 ) 35.7 Maximum polarization P max, (µc/cm 2 ) 42.5 Coercive field, E C (kv/cm) 7.47 Pyroelectric coefficient p at T C (µc/m 2 o C) 4.8 x

47 Poling of Soft PZT; field effect 3I.5. DISCUSSION 3I.5.1. Polarization Current Curves As has been shown in the hysteresis loop measurements, the coercive field is about 7.5 kv/cm. The polarization current curves (Figure 3I.4) revealed that the threshold field needed to view the hump within 5 s at room temperature is ~ (½)E C. Keep in mind that the coercive field is not an absolute quantity 12 since by subjecting a nonpoled sample to a low field for long time interval the polarization will eventually switch. In other words, since for a virgin sample (not poled) there is no net polarization, on average the positive polarization +(½)P and the negative polarization (½)P cancel and the field necessary to switch one to the other, or at least to make a net (non-zero) dipole moment, is approximately (½)E C, as is experimentally observed. The polarization of PZT is provided by 18 o and/or 9 o (for tetragonal structure) and 71 o and/or 19 o (for rhombohedral structure). All types of switching can occur, however, the distinction between the different types cannot be made without further information. The polarization current curves (Figure 3I.4 and Figure 3I.7) for low and high fields respectively can be divided into three regions: Region I: Maxwell-Wagner polarization 25-29, Region II: polarization switching and Region III: steady-state conduction current 1 We assume that the low field curves (Figure 3I.4) show only regions (I) and (II) since, as we said earlier, at low fields a steady-state current was not reached. On the other hand, the high field curves (Figure 3I.7) show only a tail of region II and region III. In the following, we discuss in some detail the mechanism occurring in each region at low and high poling electric fields. 3I.5.2. Low field regime Applying a poling dc field to a ferroelectric ceramic leads to so-called Maxwell- Wagner polarization indicated by an initial current which is decaying by time constant τ = R B C GB, where, according to Waser 25-28, R B is the grain bulk resistance and C GB is the grain boundary capacitance. The current decays as J MW = J /τ oe t The Maxwell Wagner (M W) (or space charge polarization as it is referred to in some texts 28 ) is limited to short time only (Region I). The space charge polarization is attributed to the pile up and depletion of the mobile charge carriers that arise at the interface between two different media 25-28,3, leading to a limited transport until the charge carriers are stopped at a potential barrier, possibly a grain boundary 2. The 35

48 Chapter (3) Part I resultant current due to this polarization decays until the charging of the barrier capacitor is completed [28]. For that reason M W or space charge polarization is limited to short time only. The polarization current curves in Figure 3I.4 show initial current decays with a time constant (τ ~2.6 s) for all fields. Maxwell-Wagner polarization is followed by polarization switching (region II), characterized by a limited window of applied fields ((½)E C E C ) for which the polarization current shows a hump due to the switching of the antiparallel dipoles ( P +P) 13,14. As stated by Merz 13,14 and many others 11,12,15,16, the switching process shows that, the higher the applied field, the faster the appearance of the hump, implying a faster dipole switch and thus a larger peak current J max. As shown in Figures 3I.5 and 6, at higher temperature the switching time becomes shorter and thus the switching current increases significantly 13. For extremely low field or moderate field applied for shorter time, the hump is not observed, Figures 3I.5 and 1. That means that switching is not complete. In other words the polarization didn t reach its full magnitude, as confirmed by the pyroelectric current curves, Figures 3I.9 and 11. The total switching polarization is given by eq. 3I.7. Due to its extremely fast response and negligible magnitude compared to the second term, the first term is negligible 12,14,15,18. For non-poled material there is no initial polarization and the total area under the poling current curve gives P instead of 2P. Consequently, eq. 3I.7 becomes: J () t dt P = (3I.11) Since the current decreases exponentially, the total switching time t sw is usually taken as the value where current falls to.1 J max 12. Since switching occurs at any field given enough time, the polarization is independent of the field and the charge density is constant after the switching time 15. Therefore eq. 3I.11 becomes tsw J () t dt = P = Constant (3I.12) sw This can be verified by examining Figures 3I.11 and 12. When different fields are applied on two samples, long enough for the current to reach.1 J max, the two samples show equal pyroelectric activity. Table 3I.2 shows the switching polarization for different poling fields and shows that the charge density is indeed constant. Worthwhile to mention is that, it is believed that there is no ionic polarization in polar dielectrics, since the permanent dipoles already exist in the material. Polar 36

49 Poling of Soft PZT; field effect materials reveal a tendency towards dipolar or rotational polarization 33. Dipolar polarization is characterized by a slow behavior 33 as can be observed from the polarization current curves. Adding to that, the ionic polarization requires a time as short 33 as ~1-12 s and therefore is impossible to be detected by our experiment. This implies that this process is already beyond the time scale discussed in our experiment. Table 3I.2: Polarization results according to Merz s theory. Applied Poling Field (kv/cm) J max na/cm 2 Switching time t s, s Switching polarization P sw, µc/cm 2 according to eq. 3I I.5.3. High Field Regime At high field the domain growth is extremely rapid 13. By means of our techniques it is not possible to record the maximum polarization current (the hump) and the corresponding switching time. For that reason the higher the applied field, the lower the initial current. In other words, at high field we see only the decaying tail of the hump at the very beginning of the experimental curve. If we integrate the experimental J t curve to calculate the polarization, the charge will be lower than that for the low field case since the integration range is incomplete. In reality this is not the case. Calculating the spontaneous polarization from pyroelectricity we found the higher the applied field, the higher the polarization (Figure 3I.14), as expected. Contrary to the low field regime, the current at high electric field shows a steady-state at very long time. This steady-state increases with increasing applied field in such a way that the ratio E/J gives a constant value. The steady state current is ascribed to electrical conduction of the material 1 and the constant ratio E/J represents the electrical resistivity ρ (Table 3I.1). In a recent paper Lupascu et al. 34 showed also the time dependence of the polarization process when reversing the direction of the electric field, but only at high fields. The fast response is attributed to domain switching only. Our results corroborate theirs since in our E < E C polarization results, only the tail of M W process can be identified. For the E > E C polarization current decreases with increasing the electric field and refers to the domain switching only. The long time scale tail is attributed to the steady state conduction current. 37

50 Chapter (3) Part I CONCLUSIONS Poling is a crucial step in the manufacturing process for ferroelectric materials and in this paper the poling state has been determined phenomenologically by different techniques. It appears that the pyroelectric activity, regardless its own usage, can be used as a good tool to measure the degree of polarization. The polarization current in the low field regime can be described with the switching polarization theory and according to this theory, for optimal poling, an electric field slightly higher than E C field should be applied at least for the switching time of the domains at room temperature. Contrary to common practice, good dielectric and ferroelectric properties can be reached at room temperature at a field slightly larger than the coercive field. Finally, the material showed high stable electrical insulation and no resistance degradation has been observed. 38

51 Poling of Soft PZT; field effect REFERENCES B. Jaffe, W. R. Cook, and H. Jaffe, Piezoelectric Ceramics, Academic Press New York, A. J. Moulson and J. M. Herbert, Electroceramics, Chapman & Hall, London, N. Setter, Piezoelectric Materials in Devices, Ceramics Laboratory, EPFL, 23. J. Ravez, C.R. Acad. Sci. Paris Serie IIc, Chimie/Chemistry, 3, 267, 2. M. Sayer, B. A. Judd, K. El Assal and E. Prasad, J. Can. Ceram. Soc.,, 5, L. Kholkin, D. V. Taylor, and N. Setter, August, ISAF XI p.p H.-W. Wang, S.-Y. Cheng and C.-M. Wang, IEEE/CHMT Japan IEMT Symposium, 263, L. J. Raibagkar and S. B. Bajaj, Solid State Ionics, 18, 15, S.-B. Kim, D.-Y. Kim, J.-J. Kim and S.-H. Cho, J. Am. Ceram. Soc., , 199. A. K. Jonscher, Dielectric Relaxation in Solids, Chelsea Dielectrics, London, D. Damjanovic, Rep. Prog. Phys , J. C. Burfoot and G. W. Taylor, Polar Dielectrics and their Applications, University of Calif. Press Berkeley and Los Angeles, W. J. Merz, Phys. Rev., 95, 69, E. Fatuzzo and W. J. Merz, Phys. Rev., 116, 61, C. F. Pulvari and W. Keubler, J. App. Phys., , U. Böttger, Polar Oxides, Edited by R. Waser, U. Böttger and S. Tiedke, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, 25, and references therein. M. E. Drougard, J. Appl. Phys., 31, 352, 196. P. H. White and B. R. Withey, Brit. J. Appl. Phys., 2, 1487, M. Hammer and M. J. Hoffmann, J. Am. Ceram. Soc., 81, 3277, S. B. Lang, Source Book of Pyroelectricity, Gordon and Breach Science Publishers, London, M. E. Lines and A. M. Glass, Principles and Applications of Ferroelectrics and Related Materials, Clarendon Press Oxford, A. Safari, R. K. Panda, and V. F. Janas, Key Eng. Mater., 35, 122, C. Elissalde and J. Ravez, J. Mater. Chem., 11, 1957, 21. W. Y. Pan, T. R. Shrout, and L. E. Cross, J. Mater. Sci. Lett., 8, 771, M. Vollman and R. Waser, J. Am. Ceram. Soc., 77, 235, R. Waser and R. Hagenbeck, Acta Mater., 48, 797, 2. M. Vollman and R. Waser, J. Electroceramics, 1, 51, T. H. Hölbling, N. Söylemezoğlu and R. Waser, J. Electroceramics, 9, 87, 22. E. Bouyssou, P. Leduc, G. Guégan and R. Jérisian, J. Phys., 1, 317, 25. N. E. Hill, W. E. Vaughan, A. H. Price and M. Davies, Dielectric Properties and Molecular Behaviour, Van Nostrand Reinhold London, B. A. Strukov and A. P. Levanyuk, Ferroelectric Phenomena in Crystals, Springer-Verlag Berlin Heidelberg, Chapter 4 of this thesis. B. Tareev, Physics of Dielectric Materials Mir Publishers Moscow, D. C. Lupascu, S. Fedosov, C. Verdier, J. Rödel, H. von Seggern, J. Appl. Phys., 95, 1386,

52

53 POLING OF SOFT PZT Part II: Grain Size Effect ABSTRACT The properties of piezoelectric ceramic materials are strongly dependent on the degree of polarization as set by the poling process. As the polarization depends on the domain wall mobility, which in turn depends on the domain size and grain size, this part focuses on the effect of the grain size on the poling process and dielectric and piezoelectric properties. Polarization and pyroelectric, dielectric and piezoelectric properties of modified donor doped ferroelectric Pb(Zr,Ti)O 3 ceramics having different grain sizes have been studied. The dielectric and piezoelectric parameters greatly improved at room temperature with increasing grain size. The Curie temperature (16 o C) is found to shift slightly (by about 7 o C) towards higher temperatures as the grain size increases (2-1 µm). It is concluded that, an optimum choice of a combination of poling field and grain size leads to optimum physical properties. A limited dependence of the dielectric properties on the grain size in the nonpoled state exists. The dominant contribution to the dielectric properties in the different conditions of the material is discussed.

54 Chapter (3) Part II Poling of Soft PZT Grain Size Effect! 3II.1. INTRODUCTION Investigations of the effect of grain size on the phase transition in ferroelectrics date back to the 195s. Kniekamp and Heywang 2 showed that the ferroelectricity in (BaTiO 3 ) BT single crystals disappears for sufficiently small crystal size. In polycrystalline BT the permittivity increases with decreasing grain size 3. Coincidently with Kniekamp and Heywang, Känzig and co workers 4,5 investigated the effect of grain size on T C for polycrystalline ferroelectrics. With decreasing grain size in the range of 1 5 µm for sintered samples with high density, it was found that: (i) there is a decrease in the dielectric constant height, (ii) the dielectric constant peak becomes broader, (iii) the Curie transition temperature T C rises. However, with the availability of more advanced chemical preparation techniques (sol gel, hydrosol, aerosol, microemulsion, etc.), it is now established that the decrease in the grain size causes: (i) a monotonic reduction in the transition temperature T C contrary to early observation (ii) the dielectric constant peak becomes wider (ii) the ferroelectric phase transition becomes increasingly diffuse, (iv) the crystallographic distortion of the unit cell, which characterizes the ferroelectric phase, decreases and tends towards higher symmetry 6. Uchino et al. 8 made a detailed study on the variation of the tetragonal distortion (c/a) with particle size in BaTiO 3 and have identified the T C as that temperature at which (c/a) 1. They estimated the critical size of the crystal below which the ferroelectricity cannot be observed as 12 nm. However, in a more recent study Saad et al. 9 showed that the ferroelectricity in BaTiO 3 can be observed in unconstrained materials in sample as small as 7 nm. Recently, the study of grain size effects (on the nanometer scale) in ferroelectric systems has become very important because of their potential applications, especially as nonvolatile memory elements. The miniaturization of ferroelectric electroceramic components is accompanied by the necessity of reducing of the grain size. The domain size, domain configuration and domain wall mobility all change with a reducing grain size 7. There have been many earlier attempts to investigate the influence of grain size on the piezoelectric properties in PZT ceramics 7. In general, grain size studies were limited! This part has been accepted for publication as: T. M. Kamel and G. de With, J. Eur. Ceram. Soc.,

55 Poling of Soft PZT; grain size effect in a range of 1. 1 µm in bulk ceramics. The variations in properties with grain size, as summarized in Table 3II.1, showed little consistency. Reasons for the inconsistencies are believed to be related to processing and the control of grain boundary resistivities that lead to space charge accumulation, which masks the size effect response in the ferroelectric grains 7. Table 3II.1: Grain Size Effects on PZT Systems (adopted from Randall et al. 7 and references therein) Reference Year Grain size µm Dielectric constant ε r at 25 o C Piezoelectric property at 25 o C Transition parameters Model Comments Haertling Increases Geseman Webster and Weston Increases As E C increases, P r decreases As k p decreases, P r decreases As k p and Q m decrease, P r decreases Internal stress model Hot pressed ceramics Clamping of domain motion Okazaki and Nagata Ouchi et al Martirena and Burfoot Keizer and coworkers Decreases k ij decreases As d ij and k p decrease, E C increases Decreases d ij decreases As T C increases, ε max decreases As T C increases, ε max decreases As T C increases, ε max decreases As T C increases, ε max decreases Space charge model Gaussian distribution Internal stress model Porosity effects are also important Lanthanum doped PZT Hot pressed niobium doped PZT Propose hydrostatic effect; lanthanumdoped PZT Pisarski Rossetti As T C increases, ε max decreases 2D T C increases 2D or 3D stress model Width of MPB region change Devonshire theory; thin film PbTiO 3 Yamamoto As E C increases, P r and k p Decrease As T C decreases, ε max decreases PbO grain boundary Processing dependence One of the intrinsic microstructural characteristics in ferroelectric PZT ceramics are the internal stresses that develop because of the incompatible strains occurring during the paraelectric ferroelectric transition 1. In addition, the thermal expansion anisotropy of tetragonal crystals may also contribute to the stress development. According to the internal stress model proposed by Buessem et al. 3 the large internal stress developed during cooling of PZT is released by the formation of a polydomain structure. A polydomain structure is possible if the grain is large enough to contain multiple domains. Consequently the internal stress along the grain boundary would be substantially relieved in large PZT grains by the possible formation of multidomain structure. For small PZT grains, however, the formation of polydomain structure may 43

56 Chapter (3) Part II not occur. Consequently, the internal stress increases during the cooling. Such a stress will not be relieved by the formation of 9 o domains and leads to a decrease in T 3 C. In Part I of this chapter, the optimum poling conditions of soft ferroelectric ceramic PZT 1 were introduced. The approach was to employ some preconditions of poling followed by evaluation of the degree of polarization by different methods. Using the switching current and pyrocurrent data, insight to the mechanism of polarization in soft ferroelectrics was gained. The internal stresses can be relieved by increasing the grain size 3,11 as a result of polydomain formation, resulting in easy domain switching (i.e. easier poling). Therefore, the present study focuses on the dependence of the polarization and consequently the dielectric and piezoelectric characteristics on the grain size variation. 3II.2. EXPERIMENTAL Non poled soft PZT (PXE52, modified donor doped PbZr.415 Ti.585 O 3 ) ceramic powder coded PXE 52 were provided by Morgan Electro Ceramics B.V. Eindhoven, the Netherlands. The powder was pressed into bars of at least 15 mm in diameter and sintered at different temperatures ( o C) in a tube furnace with flowing oxygen. The following sintering profile was applied for all bars. The bar was heated from room temperature to the desired temperature in 8 hours and held at this temperature for 4 hours and then cooled down to room temperature in 8 hours. The sintered bars were diced into discs of 1 mm in diameter. Next, the discs were subjected to clean firing at 7 o C for 48 hours, cleaned in methanol in an ultrasonic bath for two minutes, washed in water and soap and then dried in air. Finally, the discs were electroded with Ni by sputtering for electrical measurements. The microstructure of the as prepared samples was investigated using polarized light optical microscopy. First, the sample was molded in a resin before grinding and successive polishing. Thereafter the sample surface was etched as described previously 1. The grain size d gs was measured using linear intercept method, counting about 1 grains. Hysteresis loops were observed using a computer controlled virtual ground method (Radiant Technologies Inc., RT6 HVA z 2 V amplifier and RT6 HVS z high voltage input test system) using the fast mode (.4 ms) at room temperature. The maximum applied electric field was 2 kv/cm. The polarization current measurements were carried out on non poled samples at applied electric fields of 3.75, 7., and 25 kv/cm at room temperature. The electric field was applied using the Keithley built in voltage source. The pyroelectric measurement was carried out using the direct method 12. Experimental details for pyroelectricity measurements are described in detail in ref. [1]. 44

57 Poling of Soft PZT; grain size effect The resonance method is commonly used to determine material coefficients of piezoelectric ceramics 13,19,2. When excited at the resonance frequency f r, the piezoelectric disc will resonate at a greater amplitude. For the anti resonant frequency f a, the impedance of the piezoelectric disc is at a maximum and the oscillation amplitude is at a minimum 14. By measuring the resonance and anti resonance frequencies f r and f a, along with dielectric parameters, the electromechanical coupling factors k p, k 33, k 31 and piezoelectric charge constants d 33 and d 31 can be calculated using the equations analyzed in Appendix B. The dielectric and piezoelectric parameters of poled samples were measured using a HP4294A Precision Impedance Analyzer at 1 khz in the temperature range from room temperature up to T C. 45

58 Chapter (3) Part II 3II.3. RESULTS The microstructure of polished and etched surface of nonpoled samples was analyzed using optical microscopy. Figure 3II.1 shows the optical micrographs for samples sintered at different temperatures ( oc). As shown in Figure 3II.2 and summarized in Table 3II.2, the grain size and density increased as a function of the sintering temperature. The average grain size ranges from ~2 µm at 115 oc to ~1 µm at 13 o C. " $ % & ' ( # Figure 3II.1: 3II.1: Optical micrographs of samples (" #) sintered at different temperatures ( oc), see Figure 3II.2. 46

59 Poling of Soft PZT; grain size effect Grain Size (µm) GS 8. Density Sintering Temperature ( o C) Density (mg/cm 3 ) Figure 3II.2: Grain size and density as a function of sintering temperature for PXE 52. The line serves as a guide to the eye. Table 3II.2: Sintering temperature and the corresponding resultant density and mean linear intercept (Sample standard deviation in brackets) of PXE 52. Sample Name Sintering Temperature ( o C) Grain Size (µm) Density (ρ) g/cm 3 " (±.4) 7.29 $ (±.2) 7.72 % (±.4) 7.82 & (±.9) 7.82 ' (±1.2) 7.93 ( (±1.1) 7.93 # (±2.2) 7.98 Figure 3II.3 depicts the hysteresis loops for the different grain sizes. At first glance, no major change is observed in the gross shape of the loops. The rectangular ratio of the hysteresis loops, defined as P r /P max where P r is the remanent polarization and P max is the maximum polarization 15, did not significantly change. However, the coercive field slightly decreased with increasing grain size, as shown in Figure 3II.4. For the smallest grain (~2 µm), E C = 8. kv/cm and for the largest grain (~1 µm) E C = 7.5 kv/cm. 47

60 Chapter (3) Part II Polarization (µc/cm 2 ) Electric Field (kv/cm) Figure 3II.3: 3: Hysteresis loops as a function of grain size at room temperature and maximum field of 25 kv/cm. 48 E C (kv/cm) Grain Size, d gs (µm) Figure 3II.4: The variation of coercive field E C with grain size. As we discussed in the preceding part 1, the poling field enhances the spontaneous polarization and dielectric properties. Figure 3II.5 summarizes the main results obtained previously 1 for the poling effect on P S, ε r and tan δ. It is clearly shown that when an electric field slightly higher than the coercive field (7.5 kv/cm) is applied at room temperature, a maximum enhancement in properties results. Following the same procedure as applied before 1, Figure 3II.6 shows the polarization current as a function of time at a fixed field of 3.75 kv/cm for the different grain sizes.

61 Poling of Soft PZT; grain size effect P s (µc/cm 2 ) E C ε r P S tan δ Poling Field (kv/cm) Dielectric Constant Figure 3II.5: Summary of poling effect on the polarization and dielectric properties as presented in Part I of this Chapter. Polarization Current (na/cm 2 ) Poling Field 3.75 kv/cm tan δ d gs (µm): Time (s) Figure 3II.6: The poling current as function of time for different grain sizes under applied poling field of 3.75 kv/cm at room temperature. The polarization current profile shows the typical switching peak which is moving towards shorter time as the grain size increases. It is interesting to observe that for grain size 4.5 µm although the applied poling field is quite low (.5E C ), the switching peak is observed. This behavior indicates the possibility of complete switching and consequently maximum polarization even at low field if the grain size is large enough. Figure 3II.7 illustrates the decrease of the switching time t sw with the grain size. As can be expected, t sw becomes probably constant at larger grain sizes than those used in the present work. For the smallest grain size the switching peak was not observed and this can be ascribed to the unrelieved internal stresses developed in fine grains which suppresses the orientation process Either the switching may occur at much longer time scale than 49

62 Chapter (3) Part II used in our experiments or the switching process never occurs due to the indomitable counteracting internal stress. Switching Time, t sw (s) Grain Size, d gs (µm) Figure 3II.7: Decreasing switching time t sw with increasing grain size. By integrating the polarization current over the poling time, the polarization is obtained and plotted in Figure 3II.8A. The increasing grain size greatly enhances the polarization from 7.5 µc/m 2 for d gs = 2 µm to 4 µc/cm 2 for d gs = ~1 µm, as illustrated in Figure 3II.8B. Polarization (µc/cm 2 ) 4 A Poling Field: 3.75 kv/cm Time (s) d gs (µm): (2.) (3.45) (4.5) (5.7) (7.) (8.6) (9.78) Maximum Polarization (µc/cm 2 ) Poling Field: 3.75 kv/cm Grain Size (µm) Figure 3II.8: (A) Polarization as calculated form curves in Figure 3II.6; (B) Maximum polarization as a function of grain size. Figures 3II.9A, B, and C show the pyroelectric coefficient as a function of temperature for the different grain sizes after different poling fields of 3.75, 7.5, 25 kv/cm, respectively. The pyroelectric activity is remarkably enhanced by increasing the grain size as a result of the increase of the spontaneous polarization. Furthermore, as a typical grain size effect, the inset figures show the shift of the Curie temperature towards higher temperature by increasing the grain size. 5

63 Poling of Soft PZT; grain size effect Pyroelectric Coefficient (µc/cm 2. o C) Pyroelectric Coefficient (µc/cm 2. o C) Pyroelectric Coefficient (µc/cm 2. o C) T C ( o C) d gs (µm) Poling Eield: 3.75 kv/cm d gs (µm) Temperature ( o C) Poling Eield: 7.5 kv/cm d gs (µm) d gs (µm) T C ( o C) T C ( o C) Temperature ( o C) d gs (µm) Poling Field: 25 kv/cm d gs (µm) Temperature ( o C) Figure 3II.9: Pyroelectric coefficient after different poling conditions. As we mentioned previously, the poling field enhances the dielectric properties. Figures 3II.1, 11, 12 and 13 show that increasing the grain size enhances the dielectric and piezoelectric parameters as a consequence of the improvement made in the 51

64 Chapter (3) Part II polarization. The dielectric constant ε r and dielectric dissipation factor tan δ for poled materials greatly improve with respect to the nonpoled case. The piezo properties are also enhanced by increasing the grain size. The ultimate enhancement is a function of the poling field. Relative Permittivity 4 38 A Grain Size, d gs (µm) tan δ B.22 Poling Field Nonpoled kv/cm 7.5 kv/cm 25 kv/cm Grain Size, d gs (µm) Figure 3II.1: Room temperature relative permittivity ε r (A) and dielectric dissipation factor tan δ (B) as a function of grain size at different poling conditions. Planar coupling factor, k p.7 A µm 7. µm µm 8.6 µm 9.78 µm Poled at 3.75 kv/cm Temperature ( o C) Poled at 25 kv/cm 3.45 µm 4.5 µm 7. µm 8.6 µm 9.78 µm Temperature ( o C) C.8 B Poled at 7.5 kv/cm 2. µm 3.45 µm 4.5 µm 5.7 µm 7. µm 8.6 µm 9.78 µm Temperature ( o C).8 D Poled at: 3.75 kv/cm 7.5 kv/cm 25 kv/cm Grain Size, d gs (µm) Figure 3II.11: Planar coupling factor k p as a function of temperature at different grain sizes after different condition of poling, (A) E = 3.75 kv/cm (B ) E = 7.5 kv/cm, (C) E = 25 kv/cm. (D) k p as a function of grain size at room temperature. 52

65 Poling of Soft PZT; grain size effect d 33 (pc/n) µm 5.7 µm 7. µm 8.6 µm 9.78 µm Temperature ( o C) µm 4.5 µm 7. µm 8.6 µm 9.78 µm A Poled at 3.75 kv/cm C µm 3.45 µm 4.5 µm 5.7 µm 7. µm 8.6 µm 9.78 µm Temperature ( o C) 9 D 8 B Poled at 7.5 kv/cm Poled at: kv/cm kv/cm Poled at 25 kv/cm kv/cm Temperature ( o C) Grain Size, d gs (µm) Figure 3II.12: Piezoelectric charge constant d 33 as a function of temperature at different grain size after different condition of poling (A) E = 3.75 kv/cm (B ) E = 7.5 kv/cm, (C) E = 25 kv/cm. (D) ) d 33 as a function of grain size at room temperature. -d 31 (pc/n) Poled at 3.75 kv/cm Temperature ( o C) Poled at 25 kv/cm A 4.5 µm 5.7 µm 7. µm 8.6 µm 9.78 µm C 3.45 µm 4.5 µm 7. µm 8.6 µm 9.78 µm Poled at 7.5 kv/cm 2. µm 3.45 µm 4.5 µm 5.7 µm 7. µm 8.6 µm 9.78 µm Temperature ( o C) B D Poled at: 3.75 kv/cm 7.5 kv/cm 25 kv/cm Temperature ( o C) Grain Size,d gs (µm) Figure 3II.13: Piezoelectric charge constant d 31 as a function of temperature at different grain sizes after different condition of poling (A) E = 3.75 kv/cm (B ) E = 7.5 kv/cm, (C) E = 25 kv/cm. (D) ) d 31 as a function of grain size at room temperature. 53

66 Chapter (3) Part II It is worth to point out that the dielectric properties (ε r and tan δ) for nonpoled samples did not considerably change with changing grain size. However, the post poling properties are noticeably grain size/poling field dependent. This behavior raises the question of what are the effective contributions to the dielectric parameters before and after poling. This question will be discussed in the next section. 3II.4. DISCUSSION 3II.4.1. Limits of the Grain Size Effect From the results of the dielectric and piezoelectric properties (Figures 3II.1 13) it can be concluded that by optimum combination of poling conditions and grain size the dielectric properties are significantly enhanced. However, the curves in Figure 3II.1 13 show that the effect of increasing the grain size saturates at a certain grain size. This behavior can be explained as follows 16. During the transformation from the paraelectric phase to the ferroelectric phase, the crystal splits into multiple domains. This splitting is necessary to lower the free energy by reducing the electrostatic energy of the spontaneous polarization charges. However, domain splitting cannot be continued without limit because a definite energy is required for the formation of domain walls. The splitting of a crystal into domains stops when the energy gained by reduction of the electrostatic energy becomes equal to the energy lost in the formation of domain walls 16. 3II.4.2. Intrinsic and Extrinsic Contribution There are several polarization mechanisms contributing to the dielectric response 17,18 ; (i) electronic polarization: the relative displacement of the negatively charged electron shell with respect to the positively charged core; (ii) ionic polarization: as observed in ionic crystals and describes the displacement of the positive and negative sublattices under an applied electric field; (iii) orientation polarization: the alignment of permanent dipoles via rotational movement; (iv) space charge polarization (Maxwell Wagner polarization): polarization due to spatial inhomogeneities of charge carrier densities 2 ; (v) domain wall motion: plays a decisive role in ferroelectric materials and contributes significantly to the overall dielectric response. The total polarization of dielectric material results from all the contributions mentioned above. In bulk ferroelectric ceramics, the room temperature dielectric and piezoelectric properties result from a combination of the intrinsic lattice related response and extrinsic responses originating from domain wall motion, phase boundary, space charge, etc. 18,21,22,23. ε = ε + ε (3II.1) r int ext The extrinsic contributions can be separated in two types: reversible and irreversible. Both types are presented schematically in Figure 3II.14. Under a small (subswitching) 54

67 Poling of Soft PZT; grain size effect field reversible movement of the wall is regarded as a small displacement either by vibration or bending around a local minimum. When the field becomes high enough to switch the domain, an irreversible contribution is expected 18. Small displacements of all types of domain walls will affect the polarization of the material whereas the movement of non 18 o walls, in addition to the polarization change, directly contribute to the piezoelectric effect 24,25,27. Displacement of domain walls also contributes to the dielectric and mechanical losses of ferroelectric materials and, particularly near the phase transition temperature, may dominate other loss mechanisms 24. Figure 3II.14: Schematic of the movement of a domain wall in the lattice potential. By considering the ε r E curve (Figure 3II.15), the sharp peak at the coercive field is due to high domain wall density and motion (extrinsic contribution). At large fields, domain wall contributions are reduced due to a decrease in the number of domain walls. Therefore, the dielectric constant measured at large applied voltage values will approach the intrinsic dielectric constant 28,29. Based on the domain wall motion model 3,31, it was proposed that the losses are due to domain wall motions, and the wall contribution is proportional to the total domain wall area per unit volume. During poling many domain walls are eliminated and do no longer contribute to the loss, so that the loss factor would be expected to reduce. Polarization P ε r Relative Permittivity Electric Field Figure 3II.15: P E hysteresis loop and the derivative (dp/de) ε r as a function of an AC electric field. 55

68 Chapter (3) Part II Cao and Randall 32 indicated that the domain switching process involves transgranular cooperation of domains. Clearly, if this is indeed the general case, the quality of grain boundaries, presence of intergranular phases and porosity will affect domain wall switching. This may explain, in our case, why the smaller grain materials show lower polarization. From the Landau Devonshire phenomenological theory, it has been proposed that for PZT at compositions near the morphotropic phase boundary, the domain wall contribution accounts for more than half of the room temperature dielectric and piezoelectric responses 22. There is still an extensive debate on how to separate the intrinsic and extrinsic portions from the apparent measured quantities 24,25,26. Herbiet et al. 22 argued that the extrinsic contribution is the main source of dielectric loss, because the domain wall motion can induce mechanical friction. They derived that the extrinsic contribution to the dielectric property can reach 7% of the total at room temperature. A comparable extrinsic contribution percentage was observed in PZT system at room temperature 33,35. Experimentally, there are two methods to separate the extrinsic and intrinsic contributions to dielectric and piezoelectric properties in ceramics 33 : One is based on the frequency dispersion characteristics of the dielectric constant and the other is based on freezing out domain wall oscillations at very low temperature so that the dielectric constant converges towards value calculated from the phenomenological theory (which does not account for the domain contribution) 23. Zhang et al. 33 assumed that the intrinsic polarization change will be accompanied by a change in the unit cell volume while the polarization change induced by domain wall motion will not cause volumetric changes. As a consequence, the domain wall motion has no contribution to the hydrostatic response of a material. Accordingly, the change in the hydrostatic piezoelectric coefficient d h will not be related to the response from domain walls. With this assumption, they calculated the temperature dependence of the dielectric and piezoelectric constants for pure and soft PZT. They attributed most of the increase of these properties to the extrinsic responses. They mentioned that, according to the phenomenological theory 7,33, the piezoelectric response in a single domain ferroelectric can be viewed as polarization biased electrostriction. Hence the intrinsic piezoelectric and dielectric response are related through the electrostrictive coefficient Q ij and remanent polarization P r. d h 2ε 33ε oq hpr = where (3II.2A) d h =d d 31 (3II.2B) 56

69 Poling of Soft PZT; grain size effect represents the intrinsic hydrostatic piezoelectric coefficient, ε 33 the intrinsic dielectric permittivity in the direction of the polarization and ε o the permittivity of free space. Randall et al. 7 argued that eq. 3II.2A is not valid for the PZT system, because, if it were true, Q h should be practically constant, which is not the case. They added that the invalidity of eq. 3II.2a in the polycrystalline system is the result of large extrinsic contribution to ε 33 and accordingly d h is not purely intrinsic in nature. However, following the criteria of Zhang et al. 33, the hydrostatic piezoelectric coefficient d h was calculated according to eq. 3II.2B for all grain sizes and for the different poling conditions, the results of which are shown in Figure 3II.16. It is clearly seen that d h is almost constant at and above room temperature up to about 8 o C and increases at still higher temperature. Adding to that, the constant value of d h with temperature is increasing with increasing poling field. The constancy of d h according to Zhang et al. 33 is indicating the constant intrinsic contribution over the temperature interval mentioned above. This behavior could mean that eq. 3II.2A may be valid over that temperature range and cannot be applied above 8 o C due to the increasing extrinsic contribution. d h (pc/n) µm A 2. µm B µm 3.45 µm µm 4.5 µm µm 5.7 µm 9.78 µm µm 8.6 µm µm Poled at 3.75 kv/cm Poled at 7.5 kv/cm Temperature ( o C) Temperature ( o C) 3 C Poled at: D µm 4.5 µm 7. µm 8.6 µm 9.78 µm Poled at 25 kv/cm Temperature ( o C) kv/cm 7.5 kv/cm 25 kv/cm Grain Size, d gs (µm) Figure 3II.16: Hydrostatic piezoelectric charge constant d h as a function of temperature at different grain size after different condition of poling (A) E = 3.75 kv/cm (B ) E = 7.5 kv/cm, (C) E = 25 kv/cm. (D) ) d h as a function of grain size at room temperature. Since the magnitude of the elastic/dielectric properties is dependent on both intrinsic and extrinsic properties, it is expected that both intrinsic and extrinsic contributions are influenced by grain size variations in the PZT ceramics 7. Therefore, d h is plotted as a function of grain size in Figure 3II.16D. Contrary to what observed by 57

70 Chapter (3) Part II Randall et al. 7, d h is found to decrease with increasing grain size, oppositely to the behavior of d 33 and d 31 (the latter which is entirely extrinsic). This behavior can be attributed to suppression of the intrinsic dielectric contribution as a result of the growing extrinsic contribution due to increasing domain wall number in larger grain sizes. Damjanovic 24 and co-workers 37 have shown that in compressively stressed crystals or crystals biased by antiparallel field, the intrinsic piezoelectric properties are enhanced. This is equivalent to the case of small grains were depolarizing fields and compressive stresses could not be completely compensated by domain walls. Thus, intrinsic effects could be enhanced in small grains by internal stresses and depolarizing field, while extrinsic effects would be smaller because the domain wall density is smaller. In large grains the internal stresses and depolarizing fields are compensated by domain walls structure, leasing to smaller intrinsic but larger extrinsic contributions. Hence we have two oppositely changing effects: a decreasing intrinsic contribution with increasing grain size (represented by d h ) and increasing extrinsic contribution with increasing grain size (represented by d 33 and d 31 ). This opposite effects of the intrinsic and extrinsic contributions may lead to a constant behavior of the dielectric parameters as a function of grain size in the nonpoled state. 58

71 Poling of Soft PZT; grain size effect CONCLUSIONS It was previously 1 found that an electric field slightly higher than the coercive field can efficiently polarize soft PZT PXE52 to maximum polarization leading to largely enhanced dielectric and piezoelectric properties. In the present work, it was found that, increasing the grain size maximizes the dielectric and piezoelectric properties as compared to properties of grain size often used in practice. However, we noticed a limit for the grain size effect. Thus, an optimum choice of a combination of the poling field and the grain size leads to maximum enhanced properties. It was also found that the extrinsic dielectric contribution increases with increasing grain size due to increasing domain wall numbers. 59

72 Chapter (3) Part II REFERENCES 1 T. M. Kamel, F. X. N. M. Kools and G. de With, J. Eur. Ceram. Soc., 27, 2471, H. Kniekamp and W. Heywang, Z. Angew Phys., 6, 385, W. R. Buessem, L. E. Cross and A. K. Goswami, J. Am. Ceramic Soc., 49, 33, M. Anliker, H. R. Brugger and W. Känzig, Helv. Phys. Acta, 27, 99, W. Känzig, Phys. Rev. 98, 549, S. Chattopadhyay, P. Ayyub, V. R. Palkar and M. Multani, Phys. Rev., B52, 13177, C. A. Randall, N. Kim, J. P. Kucera,W. Cao and T. R. Shrout, J. Am. Ceram. Soc., 81, 677, 1998, and references therein. 8 K. Uchino, E. Sadanaga and T. Hirose, J. Am. Ceram. Soc., 72, 1555, M. M. Saad et al., IEEE Trans. UFFC, 53, 228, S. B. Kim, D. Y. Kim, J. J. Kim and S. H Cho, J. Am. Ceram. Soc., 73, 161, M. E. Lines and A. M. Glass, Principles and Applications of Ferroelectrics and Related Materials, Clarendon, Oxford, S. B. Lang, Source Book of Pyroelectricity, Gordon and Breach Science Publishers, London, T. Tsurumi, H. Kakemoto and S. Wada, IEEE, 375, T. L. Jordan and Z. Ounaies, Piezoelectric Ceramics Characterization, National Aeronautics and Space Administration (NASA), ICASE Report No , L. Mitoseriu et al., Jpn. J. Appl. Phys, 35, 521, I. S. Zheludev, Physics of the Crystalline Dielectrics; Crystallography and Spontaneous Polarization, Vol. 1, Plenum Press, New York, B. Tareev, Physics of Dielectric Materials, Mir Publishers, Moscow, U. Böttger, In Polar Oxides, ed. R. Waser, Bottger and S. Tiedke, Wiley VCH Verlag GmbH & Co. KGaA, Weinheim, B. Jaffe, W. R. Cook and H. Jaffe, Piezoelectric Ceramics, Academic Press New York, A. J. Moulson and J. M. Herbert, Electroceramics, Chapman & Hall, London, W. A. Schulze and K. Ogino, Ferroelectrics, 87, 361, Herbert, U. Robels, H. Dederichs and G. Arlt, Ferroelectrics, 98, 17, T. M. Shaw, S. Trolier McKinstry, P. C. McIntyre, Annu. Rev. Mater. Sci., 3, 263, D. Damjanovic, Rep. Prog. Phys., 61, 1267, D. Damjanovic, J. Am. Ceram. Soc., 88, 2663, D. Damjanovic and M. Demartin, J. Phys., 9, 4943, D. J. Kim, J. P. Maria and A. I. Kingon, S. K. Streiffer, J. Appl. Phys., 93, 5568, S. Hiboux, P. Muralt and T. Maeder, J. Mater. Res., 14, 437, S. H. Kim, J. S. Yang, C. Y. Koo, J. H. Yeom, E. Yoon, C. S. Hwang, J. S. Park, S. G. Kang, D. J. Kim and J. Ha, Jpn. J. Appl. Phys., 42, 5952, G. Arlt, H. Dederichs and R. Herbiet, Ferroelectrics, 74, 37, R. Herbiet, H. Tenbrock and G. Arlt, Ferroelectrics, 75, 319, W. Cao and C. A. Randall, J. Phys. Chem. Solids, , Q. M. Zhang, H. Wang, N. Kim and L. E. Cross, J. Appl. Phys., 75, 454, M. J. Haun, Ph.D. thesis, The Pennsylvania State University, X. L. Zhang, Z.X. Chen, L. E. Cross and W. A. Schulze. J. Mater. Sci., 18, 968, A. F. Devonshire, Phil. Mag., 42, 165, M. Budimir, D. Damjanovic and N. Setter, Phys. Rev., B72, 6417, 25. 6

73 CHAPTER 4 SWITCHING IN SOFT PZT Part I: Field Effect ABSTRACT In this chapter, switching current measurements have been carried out on a soft PZT (PXE52, donor doped, modified PbZr.415 Ti.585 O 3 ). The experiments showed a single switching current peak during the application of electric field to a nonpoled (virgin) sample. However, an unusual doublepeak for the switching current was observed upon reversing the electric field polarity. The pyrocurrent for a forward and reverse poled sample showed a related behavior. A single pyro-peak is observed in the forward poled case and broadened peak was observed in case of a reverse poled sample. This behavior is attributed to non-18 o domain switching during the reverse poled case as a result of residual stresses developed during the forward poling. In Part I, the switching experiment as a function of the applied field will be studied.

74 Chapter (4) Part I Switching in Soft PZT Field Effect! 4I.1. INTRODUCTION The first quantitative experiments to determine the electric field and time dependence of domain reorientation were reported by Merz 1 for BaTiO 3 (BT) single crystals. In Merz s experiments, a sample was subjected to square pulse to align all dipoles in one direction followed by a second square pulse of opposite polarity. A single switching current peak as a function of time was obtained. In the preceding chapter we discussed the optimum poling conditions for a soft ferroelectric ceramic PZT 2. The approach was to employ some preconditions of poling followed by evaluation of the degree of polarization by different methods. Pyrocurrent measurements were found to be a very useful tool to assess the polarization state. The measurements of the switching current enabled us to calculate the polarization, which was found to be consistent with the spontaneous polarization as calculated from the pyroelectric current. Thus, using both switching current and pyrocurrent, insight to the mechanism of polarization in soft ferroelectrics can be gained. In an extension to our previous work on a soft PZT ceramic, following an approach similar to Merz s, we observed that the switching current showed two peaks as a function of time instead of the commonly observed single one. In the present paper a possible explanation for the appearance of the double peaked switching current is introduced. The explanation is based on the phenomenon of field-induced internal stresses 3-6. Merz 1 and his co-worker 7 proposed that the field applied antiparallel to the polarization direction switches the polarization from -P to +P showing a single switching current peak. The area under the curve equals: t sw J sw dt = 2 Psw = Constant (4I.1) where J sw and t sw are the switching current density and switching time, respectively. The last step in eq. 4I.1 can be made since switching occurs at any field given enough time and the polarization is independent of the applied field and constant after sufficient switching time. Several authors studied field-induced phase transformations in ferroelectrics. Fan and Kim 8 showed that, when ferroelectric domains in a polycrystalline ceramic are subjected to a static electric field, the polar axes of (both the tetragonal and rhombohedral phases) tend to be aligned with the direction of the! This part has been accepted for publication as: T. M. Kamel and G. de With, J. Appl. Phys.,

75 Switching in Soft PZT; field effect applied field. Using X-ray diffraction they showed that, when the tetragonal and rhombohedral phases are co-existing together in the ferroelectric ceramic, upon applying an electric field the relative intensities of the rhombohedral reflections decreased after poling, while those of the tetragonal reflections increased. When the perovskite was mainly in the rhombohedral phase, the poling process increased the intensity of the rhombohedral peak, but there was no phase transformation to the tetragonal phase. Similarly, the phase transformation did not occur when the specimen was composed mainly of the tetragonal phase but in this case extensive 9 o domain switching occurred during the poling process 8. Liu et al. 3 made in situ X-ray diffraction experiments to observe non-18 o domain switching and a phase transition at different electric fields for nonpoled and poled PZT specimens. They showed that, upon application of an electric field of E C to a nonpoled sample, the intensity of the tetragonal T(2) peak increases while that of T(2) decreases, indicating 9 o domain switching of the tetragonal phase from the (2) to the (2) orientation. Moreover, the intensity of the rhombohedral R(2) peaks decreases upon applying an electric field, indicating a phase transition from rhombohedral to tetragonal, R(2) to T(2). In the case of a polarized sample, they found that upon application of an electric field of +E C, 9 o domain switching occurs, indicated by a change from T(2) to T(2) and a phase transition, resulting in change from R(2) to T(2), is induced but only to a very minor extent. While applying an electric field of -.8E C induces 9 o domain switching as indicated by a change from T(2) to T(2) and a phase transition resulting in a change from T(2) to R(2) is induced to a large extent. They attributed those field-induced phase transformations to the residual stresses developed during poling. 4I.2. EXPERIMENTAL Non-poled polycrystalline ceramic samples of soft PZT (PXE52, donor doped, modified proprietary composition with the overall formula PbZr.415 Ti.585 O 3 ) were obtained from Morgan Electro Ceramics BV, Eindhoven, The Netherlands. The dimensions of the samples under study were 5x5x.2 mm 3. The microstructure of the samples was investigated after grinding, successive polishing and etching. The switching current measurements were carried out using a Keithley 6517 Electrometer on initially nonpoled samples using a ramping electric field of.5 kv/cm.s. The electric field was applied using the Keithley built-in voltage source with a response time of 5 ms at.1 kv and 8 ms at 1 kv and varied until a maximum field of 25 kv/cm was reached. The pyroelectric measurements were carried out using the direct method 9. Experimental detail for the pyroelectricity experiments is described in detail in ref. [2]. 63

76 Chapter (4) Part I 4I.3. RESULTS The microstructure of the nonpoled ceramic PZT used is shown in Figure 4I.1. The density was 7.45 g/cm3 while the average grain size as measured using linear intercept method for 2 grains, was 4.5 µm. Attempts to visualize the changed domain structure after switching failed due to the influence of the surface on the domain structure: the surface domain structure appeared to be largely fixated. It is well-known that the domain structure in the bulk of the specimens may behave rather different from that near the surface19. In view of the expected complexity, no attempts were made to reveal the crystallographic nature of the specimens by XRD. Figure 4I.1 4I.1: SEM micrograph of a non-poled sample. In Merz s experiments on BT single crystals, a sample is subjected to square pulse to align all dipoles in one direction followed by a second square pulse of opposite polarity and a single switching current peak as a function of time was obtained1,7,1-15. In our experiments we followed a slightly different approach. An. initially nonpoled sample is subjected to an electric field rate E f of 5 x 1-2 kv/cm.s in the forward direction, showing a single switching peak at 4.9 kv/cm (Figure 4I.2A). The corresponding polarization at the end of the process (5 s) is 45 µc/cm2.. Upon applying the same field rate in the reverse direction E f, a double switching peak is observed at Er1 = 6. kv/cm and Er2 = 6.6 kv/cm (Figure 4I.2B). The corresponding maximum polarization is Pmax = 85 µc/cm2 implying that Merz s thesis, eq. 4I.1, is satisfactorily verified (Figure 4I.3). 64

77 Switching in Soft PZT; field effect Switching Current (µa/cm 2 ) A B Field ramp rate: +/-.5 kv cm -1 s -1 Switching Current (µa/cm 2 ) Forward E.F. (kv/cm) Reverse E.F. (kv/cm) Figure 4I.2: (A): Switching current peak during positive poling, (B): Double switching current peak during subsequent negative poling. Polarization (µc/cm 2 ) 9 8 A B Electric Field (kv/cm) Polarization (µc/cm 2 ) Figure 4I.3: The polarization as calculated from the pyro-current for forward and reverse field. The double peak switching current was confirmed by recording the pyroelectric signal for forward and reverse poled samples. Figure 4I.4A shows the pyroelectric coefficient after a forward poling. A single pyro-peak is observed at the Curie transition temperature (168 o C). Figure 4I.4B shows the pyro-signal after reverse poling. Consistently with the switching curves a broad peak appeared at the Curie temperature, indicative of a combination of two peaks close to each other. Both the switching and the pyroelectric current double peak are continuously observed confirming this behavior. 65

78 Chapter (4) Part I Pyroelectric Coeffieceint ( µc/cm 2. o C) 1 1 A 1 1 B Temperature ( o C) Figure 4I.4: Pyroelectric coefficient as a function of temperature for a forward (positive) and reverse (negative) poled sample. Figure 4I.5 shows a subsequent poling-switching process. In this process, a nonpoled sample is subjected to a ramping poling field reaching a maximum of 2.7 kv/cm. In this process a single peak is observed. Upon reversing the field on the opposite direction a double peak is observed. Upon subsequent field reversal (forward and reverse) the double peak switching is permanently observed. Switching Current (µa) ramp rate: 1 V/s maximum field: 2.7 kv/cm 1 st -E (2) 2 nd -E (4) Electric Field (kv/cm) st +E (1) 2 nd +E (3) Figure 4I.5: Successive switching processes showing a single peak in the first poling for a virgin sample and permanently appearing double peak upon subsequent poling. The development of the poling was studied by gradually increasing the maximum forward field. Figure 4I.6 shows that, as long as during the poling process the ramping is stopped before major polarization takes place at approximately 5 kv/cm, only a single reverse switching peak is observed. Surprisingly the reverse peak for the very low maximum forward field of 2.3 kv/cm occurs at a field of about

79 Switching in Soft PZT; field effect kv/cm, which is lower than the forward poling peak. We have no explanation so far for this effect. With increasing maximum forward field, a double reverse peak develops as soon as the major forward polarization has taken place. The splitting of the reverse peak is illustrated in Figure 4I.7 and increases from initially.28 kv/cm to about.61 kv/cm, remaining largely constant for larger fields. E max = kv/cm E max = kv/cm Switching Current (a.u.) E max = kv/cm E max = 6.29 kv/cm E max = 6. kv/cm E max = 5.67 kv/cm E max = 5.27 kv/cm E max = 5.15 kv/cm E max = 4.69 kv/cm E max = 4.48 kv/cm E max = 3.91 kv/cm Negative Reverse E.F. (kv/cm) E max = 2.33 kv/cm Positive Forward E.F. (kv/cm) Figure 4I.6: Developing the double peak gradually with increasing the maximum forward field. Peaks Position on Reverse Poling (kv/cm) E=.28 kv/cm E=.61 kv/cm 5. Original Peak Position Developed Peak Position Forward E.F. (kv/cm) Figure 4I.7: Reversed double peak splitting E as a function of the maximum forward field. 67

80 Chapter (4) Part I It has been shown for PZT-5 2 that a sudden poling instead of gradual poling results in a quite different behavior for switching. The effect is ascribed to the collective motion of all domain walls during a sudden pulse as compared with a more sequential change in domain structure for gradual poling. For our material a maximum field of 15 kv/cm was suddenly applied and the switching monitored. Peak splitting was again observed and found to be same as for the gradual poling, Figure 4I.8, although the ratio of the two peaks is different. Switching Current (µa/cm 2 ) Electric Field (kv/cm) Figure 4I.8: Double peak switching after sudden forward poling with 15 kv/cm. 4I.4. DISCUSSION It is well known that during the phase transition of PZT from the paraelectric (cubic) to the ferroelectric (tetragonal) phase, spontaneous polarization occurs. This tetragonal structure is in a stable (equilibrium) state, in which both 9 o and 18 o domain walls are formed to reduce the effects of the elastic energy and depolarizing electric fields, respectively. However, some residual stress will exist. Other stable (equilibrium) states exist and they can be realized by reversing the direction of polarization (18 o domain switching) and changing the polarization direction to the transverse directions (9 o domain switching). When an external force (mechanical stress or electric field) is applied, the ferroelectric material may move from one equilibrium state to another, due to domain switching. This switch depends on the magnitude and direction of the load. An electric field can cause both 18 o and 9 o domain switching, depending on the direction of the electric field, but a mechanical stress can only cause 9 o domain switching 5. Several criteria have been advanced to explain this domain switching. A brief review is given in [5]. Two criteria are relevant at present. The first criterion, proposed by Liu et al. 3 and not reviewed in [5], suggests a threshold in electric field. These authors propose that in non-poled samples residual stresses are low and uniformly distributed throughout the material. This assumption possibly leads to a threshold electric field for domain switching similar in all grains. Once such a threshold is reached, 18 o and non-18 o domain switching may occur. However, in a polarized sample due to the complicated and inhomogeneous residual stress 68

81 Switching in Soft PZT; field effect distribution 24 the threshold for switching may differ significantly from grain to grain 3. Therefore, different types of domain wall switch at different applied electric field levels. The developing of internal stresses due to an electric field is an observation that dates from two decades ago 16. The second criterion, proposed by Sun and Achuthan 5 and the one that showed the best agreement with experiment, is an internal-energy-density criterion. They assumed that, when an external electric field is applied in the direction of the spontaneous polarization, minor or no switching is expected. On the other hand, when an electric field is applied in a direction opposite to the polarization, switching to the 18 o equilibrium state is enabled. Thus only an electric field in the direction opposite to the direction of spontaneous polarization can produce 18 o domain switching. In the case of an electric field applied in the direction perpendicular to the spontaneous polarization, deformation is due to shear strain. Such a deformation can only result in 9 o domain switching. The internal-energy-density criterion as set by Sun and Achuthan 5 is based on the assumption that the 18 o and 9 o domain switching are two independent modes of switching and that they take place when the respective internal-energy density reaches a critical value. It is assumed that the 18 o domain switching can only be realized by a reverse electric field and the internalenergy density due to this reverse electric field is considered to be the driving force for the 18 o domain switching. Similarly, the 9 o domain switching can only be produced by a compressive stress in the poling direction or an electric field perpendicular to the poling direction and the internal-energy density due to these two loads is considered the driving force for the 9 o domain switching. However, a reversed electric field enables 18 o domain switching and acts in the same time as a shear stress force for 9 o domain switching as long as the threshold energy for this switching is reached 3. If we assume that the difference between the two reverse peak positions is due to internal stress, an estimate for this stress can be made using the model as described in [5]. Using the values for the field as observed in our experiments, we estimated the stress to be about 3 MPa. This value seems to be not unreasonable, in view of the limited amount of information on internal stress that is available 5. Using the hypotheses briefly discussed above, we can speculate on a physical model to describe our results. 4I.4.1. Physical Model It is widely accepted that 9 o domains are highly mobile in the donor doped (soft) PZT ceramics as a consequence of the increase of lead vacancies in the crystal lattice as charge compensation 21 as well as reducing the concentration of the oxygen vacancies 22 which are thought to clamp the domain wall motion 23. Based on this assumption, we may expect that 9 o domains can be switched relatively easily upon applying an electric field. In fact, theoretical calculations 28 of PbTiO 3 support the existence of a relatively low barrier of 9 o domain wall switching. However, a large 69

82 Chapter (4) Part I internal stress is developed due to the dimensional deformation associated with 9 o domain 24. It has been shown in this work that the double peak switching is only occurring once the forward single switching peak is developed, Figure 4I.6. The separation between the two reversed peaks is also function of the forward applied field, Figure 4I.7. The splitting is the same for a ramped field and a suddenly applied field. This confirms the relatively easy switching for the soft PZT, contrary to the behavior of the PZT-5 2. Based on the information mentioned above, we may consider the following model. In the virgin state, when all domains are randomly oriented, the poling electric field induces mainly 18 o and 9 o domains. Possibly, the 9 o domains can be switched relatively easily. However, a large inhomogeneous internal stress is developed due to the dimensional deformation caused by 9 o domain switching 24. Upon reversing the applied field in the opposite direction (E to -E), the reverse peak is gradually splitting depending on how far the forward peak is developed. This splitting may indicate the amount of 9 o domain switching and consequently the amount of internal stress introduced. As the forward applied field increases, more 9 o domains switch introducing additional stresses. Upon a subsequent field reversal and due to the already developed stresses instead of direct 18 o switching, the switching occurs favorably via two successive 9 o domain rotations 25,24,26 as the large residual stress and electric field can make 9 o domain switching energetically easier than direct 18 o domain switching 24. It is assumed in this explanation that the maximum polarization in both directions of the field is the same. It is worthwhile to mention that this behavior is occurring on subsequent field reversal as a result of the permanent residual stress, Figure 4I.5. The reverse switching (both steps) is found to be slightly delayed with respect to the forward peak position due the fact that the threshold electric field needed for 9 o is always higher due to the low mobility of the 9 o domain walls which must be reflected in the poling process as a function of time 4. It has been shown that the second step of 9 o domain switching should take place in a short interval of electric field, estimated 24 as.4 kv/cm. We found in our case that the observed two switching peaks on reverse poling are separated in.6 kv/cm in a good agreement with the results of Achuthan and Sun 24. Broadening of the pyro-peak for the reversely poled sample, Figure 4b, enhances the idea of two successive thermal depolarization transitions for the 9 o domains. It is commonly believed that a negative electric field induces 18 o domain switching (e.g. [27]). However, we have shown here that electrical behavior is consistent with the microstructural observations made by Liu et al. 3 indicating non-18 o or 9 o domain switching. This would also be an explanation for the same behavior observed by other authors 6,18. A study of the effect of the grain size on this 9 o domain switching as seen by our switching current experiment will be elaborated in Part II. 7

83 Switching in Soft PZT; field effect CONCLUSIONS Modified donor doped PZT shows a single polarization current peak during poling. During reverse poling a double peak in the switching current is observed. Similar behavior is also observed in the pyroelectric current curve. This double peak in the switching curve is attributed to the residual stresses developed during forward poling that transforms the switching mode from a single threshold coercive field to double threshold coercive field. It is concluded that the residual stress makes switching via two successive rotations energetically easier than direct 18 o domain switching. 71

84 Chapter (4) Part I REFERENCES W. J. Merz, Phys. Rev., 95, 69, T. M. Kamel, F. X. N. M. Kools and G. de With, J. Eur. Ceram. Soc., 27, 2471, 27. M. Liu, K. J. Hsia and M. R. Sardela, J. Am. Ceram. Soc., 88, 21, 25. M. H. Lente and J. A. Eiras, J. Appl. Phys., 89, 593, 21. C.-T. Sun and A. Achuthan, J. Am. Ceram. Soc., 87, 395, 24. U. Belegundu and K. Uchino, J. Electroceramics, 6, 19, 21. E. Fatuzzo and W. J. Merz, Phys. Rev., 116, 61, H. Fan and H. E. Kim, J. Appl. Phys., 91, 317, 22. S. B. Lang, Source Book of Pyroelectricity, Gordon and Breach Science, Publishers, London, C. F. Pulvari and W. Keubler, J. Appl. Phys., 29, 1315, A. Levstik, M. Kosec, V. Bobnar, C. Filipič and J. Holc, Jpn. J. Appl. Phys., 36, 2744, K. Dimmler, M. Parris, D. Butler, S. Eaton, B. Pouligny, J. F. Scott and Y. Ishibashi, J. Appl. Phys., 61, 5467, V. Shur, E. Rumyantsev and S. Makarov, J. Appl. Phys., 84, 445, S. Ikeda, T. Fukada and Y. Wada, J. Appl. Phys., 64, 226, V. Shur, E. L. Rumyantsev, S.D. Makarov, A. L. Subbotin and V. V. Volegov, Int. Ferroelectrics, 1, 223, H.-T. Chung, B.-C. Shin and H.-G. Kim, J. Am. Ceram. Soc., 72, 327, 1989, and references therein. S. Li, A. S. Bhalla, R. E. Newnham, L. E. Cross and C. Huang, J. Mater. Sci., 29, 129, V. Gopalan and T. E. Mitchell, J. Appl. Phys., 83, 941, G. Arlt, J. Mater. Sci., 25, 2655, 199. F.-X. Li, D.-N. Fang and A.-K. Soh, Smart Mater. Struct., 13, 668, 24. P. Gerthsen, K. H. Hardtl, N. A. Schmidt, J. Appl. Phys., 51, 1131, 198. T. Tsurumi, Y. Kumano, N. Ohashi, T. Takenaka and O. Fukunaga, Jpn. J. Appl. Phys., 36, 597, K. Carl and K. Härdtl, Ferroelectrics, 17, 473, Achuthan and C.-T. Sun, Proceedings of SPIE-The International Society for Optical Engineering, , 379, 24. N. Bar-Chaim, M. Brunstein, J. Grunberg and A. Seidman, J. Appl. Phys., 45, 2398, S. Li, A. S. Bhalla, R. E. Newnham, L. E. Cross and C. Huang, J. Mater. Sci., 29, 129, E. Fatuzzo and W. J. Merz, Ferroelectricity, North-Holland, Amsterdam, B. Meyer and D. Vanderbilt, Phys. Rev. B65, 14111,

85 SWITCHING IN SOFT PZT Part II: Grain Size Effect ABSTRACT In Part I of this chapter, we reported on the appearance of a double peak in the switching current during the reverse poling. In this part, the switching current measurements have been carried out on a soft PZT as a function of grain size. While in small grains only a small single switching peak is observed, large grains, however, showed double peak switching, as commonly observed in this material. The pyroelectric coefficient curves show a consistent trend with the switching curves. This behavior is attributed to non-18 o domain switching during the reversed poling case as a result of residual stresses developed during forward poling.

86 Chapter (4) Part II Switching in Soft PZT Grain Size Effect! 4II.1. INTRODUCTION In Part I of this chapter 1 we discussed the appearance of a double peak in the switching current during the reverse poling. The observation was attributed to the residual stresses developed during forward poling. Due to the residual stresses, the switching mode transforms from a low single peak coercive field to high double peak coercive fields as a result of two 9 o rotations instead of a direct 18 o switching. The internal stresses developed during the forward poling makes the switching via two successive rotations energetically more favorable than one direct 18 o 2. In the present part, we study the effect of grain size of the appearance of double peak switching. It is well known that, on cooling from the paraelectric phase to the ferroelectric phase, a large mechanical stress may be generated by the anisotropic spontaneous strain, and these stresses in turn affect the domain configuration and domain dynamics within the grain 3. In small grains the polarization may be completely clamped by this effect preventing domain reversal under an applied field. At intermediate grain sizes domain reversal becomes possible but the phase transition is broadened over a large temperature range and the peak dielectric constant is suppressed because of the inhomogeneous distribution of stresses and electric fields. Eventually, in large grain sizes bulk ferroelectric behavior becomes dominant. 4II.2. EXPERIMENTAL Non-poled polycrystalline ceramic samples of soft PZT (PXE52, modified and donor doped PbZr.415 Ti.585 O 3 ) with different grain sizes were obtained from Morgan Electro Ceramics BV, Eindhoven, The Netherlands. The microstructure of the non-poled samples was investigated using optical microscopy after grinding, successive polishing and etching. The grain size and density as a function of sintering temperature are given in Table 3II.1. The switching current measurements were carried out using a Keithley 6517 Electrometer on initially non-poled samples at ramping electric field (.5 kv cm -1 s -1 ). The electric field was applied using the Keithley built-in voltage source and varied until a maximum field of 25 kv/cm with a response time of 5 ms (.1 kv) and 8 ms (1 kv).! This part has been submitted for publication as: T. M. Kamel and G. de With, J. Appl. Phys.,

87 Switching in Soft PZT; grain size effect Systematic pyroelectric measurements were carried out for all grain sizes using the direct method as described in ref. [4]. 4II II.3. RESULTS Initially nonpoled samples were subjected to a positive ramping electric field of E f = +5 x 1-2 kv cm -1 s -1 ) and show a single switching peak at 6.2 kv/cm for the smallest grain (2 µm). This peak moves towards smaller fields as the grain size increases to reach 4. kv/cm for the largest grain size (1 µm), Figure 4II.1A. When the electric field is applied in the reverse direction (E r ) with the same ramping rate, a single switching peak is observed at 6.9 kv/cm. However, there is an increasing tendency for broadening or doubling of the switching curve as the grain size increases, Figure 4II.1B. The double peak switching current was confirmed by recording the pyroelectric signal of each state (forward and reverse poled samples). Figure 4II.2A shows the pyroelectric coefficient after a forward poling. A single pyro-peak is observed at the Curie transition temperature (168 o C). The Curie temperature also slightly increases with increasing grain size. Figure 4II.2B shows the pyroelectric signal after reversing the field polarity. Consistent with the switching curves, a broad peak appeared at the Curie temperature. Both the switching and pyroelectric current double peaks are consistently and repeatedly observed confirming the existence of such behavior. A B Poling Rate: +1 V/s Maximum Field: 25 kv/cm Polarization Current (na/cm 2 ) (7) (6) (5) (4) (3) (2) (1) Electric Field (kv/cm) Polarization Current (na/cm 2 ) (log scale) (7) (6) (5) (3) (2) Poling Rate: -1 V/s Maximum Field: 25 kv/cm (1) Electric Field (kv/cm) Figure 4II.1: The polarization current curves during (A) positive poling and (B) and negative poling (switching) for different grain sizes. 75

88 Chapter (4) Part II Pyroelectric Coefficient (µc/cm 2. o C) Poling Field: 25 kv/cm Grain size (mµ): Temperature ( o C) A Pyroelectric Coefficient (µc/cm 2. o C) After Switching Max Field: 25 kv/cm Grain size (µm) Temperature ( o C) Figure 4II.2: The pyroelectric coefficient as a function of temperature for different grain sizes; (A) after poling and (B) after switching. B 4II.4. DISCUSSION 4II.4.1. Double Switching The experimental results described above are attributed to polarization changes. The question is whether the phase coexistence 9 or the field-induced phase transformation phenomenon 7,8 or non-18 o domain switching 5,6 is responsible for the observation of double-peak current during the polarization switching. In lead zirconate titanate ceramics, the low temperature tetragonal and rhombohedral ferroelectric phases coexist for compositions near the morphotropic phase boundary (MPB). Several authors studied the field-induced transformation between those two phases 7,9. X-ray diffraction measurements conducted 7 on polycrystalline lead zirconate niobate-lead zirconate titanate (PZN-PZT) indicated that phase transitions between rhombohedral and tetragonal phases and non-18 o domain switching (9 o domain switching for tetragonal phases and 71 o or 19 o domain switching for 76

89 Switching in Soft PZT; grain size effect rhombohedral phases) are induced by the poling process. Liu et al. 8 showed that the internal stress developed during poling can be responsible for splitting the threshold electric field needed for different types of domain switching 8. Belegundu and Uchino 9 ascribed the appearance of double peak switching current to the coexistence of both rhombohedral and tetragonal phases. A model using a statistical distribution of the phases by Cao and Cross 1,11 can describe the coexistence of the phases assuming that it results from the thermal fluctuation quenching during cooling from the paraelectric to the ferroelectric phase. In this model the width of the phase coexistence region ( x) is related to the ceramic grain size d gs and obeys x according to Cao and Cross or 3 d gs x 3/ 2 d gs according to Soares et al. 12. According to this criterion, there is a smaller probability of coexistence of the tetragonal and rhombohedral phases with increasing the grain size increases. However, it was noted, in our case, that there is an increasing tendency of switching curve doubling with increasing the grain size. This means that the double switching peaks cannot be attributed to switching of different phases. One of the intrinsic microstructural effects in ferroelectric PZT ceramics is the internal stress developed due to the incompatible strains arising at the Curie temperature 13. In addition, the thermal expansion anisotropy of tetragonal crystals may be also contributing to the stress development. The large internal stress developed during cooling of PZT is released by the formation of a polydomain structure which is known to be always present if the grain size is large enough to contain multiple domains. Consequently the internal stress along the grain boundary is substantially relieved in large PZT grains by the formation of multidomain structure. For small PZT grains, however, the formation of a polydomain structure within the grain may not occur. Consequently, the internal stress might be substantial during the cooling from the paraelectric to the ferroelectric phase transition. Buessem, Cross and Goswami 14 showed that the internal stress in fine-grained BT ceramic must be greater than that in coarsegrained ceramics. The strong coupling between the grain boundaries and domain walls makes domain reorientation more difficult and severely constrains its motion. This coupling affects the extrinsic contribution to the polarization in two ways: (i) the value of remanent polarization becomes smaller, because of the reduction of the achievable domain alignment, and (ii) the domain wall mobility decreases, which reduces the effective dielectric and piezoelectric coefficients. Generally speaking, the 18 o domains reorient easier than do the 9 o domains, because when 9 o domain switches into a new direction, they will be accompanied with a dimensional change 15. A poled ceramic has a smaller number of domain walls than a nonpoled ceramic and since a polydomain configuration in a ceramic grain can relieve 77

90 Chapter (4) Part II the internal stress along the boundary, a poled ceramic must have a larger internal stress than is a nonpoled ceramic. In nonpoled ceramics, residual stresses are lower and relatively uniformly distributed throughout the material 8. Therefore, the intrinsic threshold energy for domain switching is similar in all grains. Once such a threshold energy level is reached upon applying an electric field, non-18 o domain switching occurs. In a polarized sample, however, the residual stress level differs significantly from grain to grain and so does the threshold energy for switching. Therefore, non-18 o domain switching and phase transition in different grains/domains may be induced at different applied electric field levels. Accordingly, in small grains, the internal microstresses are high in the nonpoled state and hinder domain motion and hence reaching full polarization. Thus, upon poling higher stress is introduced and lower polarization develop. This condition makes the switching of non-18 o tremendously restricted. For larger grains, the stresses become lower and full polarization can be reached by smaller electric fields, since the coercive field is smaller for larger grains 16. However, internal stresses can also develop because of poling. The stresses developed may facilitate the switching to take place by two rotations via two 9 o domain rotations rather than a direct 18 o. The movement of 9 o domains becomes easier as the grain size increases. We showed previously 17 that the switching via two successive 9 o rotations can be energetically favorable than a direct 18 o switching. 4II.4.2. Pyroelectric Coefficient In the previous part 1, the pyroelectric coefficient curves showed, qualitatively, a consistent behavior with the switching current characteristics. In this part a qualitative and quantitative investigation of the pyroelectric effect, as a switching-related property, is given. The pyroelectric coefficient curves after switching have a maximum distinctly lower than that of their poled-only counterparts. However, calculating the area under peak (value of the polarization charge) of the corresponding curves in each case yields the same magnitude. In other words, although the pyroelectric coefficient maxima in the poled and switched states are not equal, the total charge released during the ferroelectric phase transition is the same for a given grain size. Table 4II.1 summarizes the total polarization charge after poling and after switching for different grain sizes. This consistency gives good evidence that the broadening of the phase transition peak is the result of a two-step depolarization process (i.e. two 9 o rotations) rather than one 18 o rotation, as the total charge must be conserved in each case. 78

91 Switching in Soft PZT; grain size effect Table 4II.1: Depolarization charge after poling and after switching for different grain sizes of PXE52 as calculated from area-under peak of the pyroelectric coefficient as a function of temperature. Temperature range used for the calculation is o C. Sample Name Grain Size (µm) Pyroelectric coefficient maximum (at T C ) After poling (µc/cm 2. o C) Pyroelectric coefficient maximum (at T C ) After switching (µc/cm 2. o C) Charge released after poling (µc/cm 2 ) Charge released after switching (µc/cm 2 ) " # $ % & '

92 Chapter (4) Part II CONCLUSIONS Grain size variation was found to have a large effect on the double peak switching in soft PZT. Small grains showed a small single switching peak due to hindrance of the 9 o domain motion. Large grains, however, showed double peak switching, the commonly observed process in this material, due to the higher mobility of the 9 o domains that transforms the switching from a single 18 o switching to two 9 o domain wall rotations. The pyroelectric coefficient curves showed a behavior consistent with these switching curves. While the pyroelectric charge after poling is released via one discharge, the pyroelectric charge after switching is released via two discharges steps. This indicates that the thermal discharging follows the same mechanism. 8

93 Switching in Soft PZT; grain size effect REFERENCES 1 T. M. Kamel and G. de With, J. Appl. Phys., (accepted) C.-T. Sun and A. Achuthan, J. Am. Ceram. Soc., 87, 395, M.E. Lines and A.M. Glass, Principles and Applications of Ferroelectrics and Related Materials, Clarendon, Press, Oxford, T. M. Kamel, F. X. N. M. Kools and G. de With, J. Eur. Ceram. Soc., 27, 2471, S. Li, A. S. Bhalla, R. E. Newnham, L. E. Cross and C. Huang, J. Mater. Sci., 29, 129, M. H. Lente and J. A. Eiras, J. Appl. Phys., 89, 593, H. Q. Fan and H.-E. Kim, J. Appl. Phys., 91, 317, M. Liu, K. J. Hsia and M. R. Sardela, J. Am. Ceram. Soc., 88, 21, U. Belegundu and K. Uchino, J. Electroceram., 6, 19, W. Cao and L.E. Cross, J. Appl. Phys., 73, 325, W. Cao and L.E. Cross, Phys. Rev. B47, 4825, M. R. Soares, A. M. R. Senos and P. Q. Mantas, J. Eur. Ceram. Soc., 19, 1865, S.-B. Kim, D.-Y. Kim, J.-J. Kim and S.-H Cho, J. Am. Ceram. Soc., 73, 161, W. R. Buessem, L. E. Cross and A. K. Goswami, J. Am. Ceram. Soc., 49, 33, S. Cai, Y. Xu, C. E. Millar, L. Pedersen and O. T. Sorensen, Electronic Components and Technology Conference, Proceedings 45 th, 84, T. M. Kamel and G. de With, J. Eur. Ceram. Soc., (accepted)

94

95 CHAPTER 5 PYROELECTRICITY VERSUS CONDUCTIVITY IN SOFT PZT CERAMICS ABSTRACT High insulation resistance is desirable for many technological applications of PZT piezoceramics. The electrical conduction of modified soft PZT ceramics has been studied as a function of temperature at different dc electric fields and grain sizes. As ferroelectrics are highly polarizable materials, poling, depolarization and electric conduction contribute to the total current. These contributions are discussed and it was found that the displacement current hides the conduction current near room temperature. The electrical resistivity of the soft PZT used has a typical, relatively high value of 3.6 x 1 12 Ω.cm near room temperature due to compensating effect. The resistivity above the Curie temperature was two-orders of magnitude lower than the room temperature value because the thermal polarization no longer compensates the intrinsic conductivity. The resistivity decreases with increasing grain size probably due to the increased Pb vacancy concentration resulting as a consequence of a higher sintering temperature. However, no resistance degradation has been observed in all cases. The values of activation energies suggest that the conduction carriers at high temperature are mainly ionic due to oxygen vacancies motion.

96 Chapter (5) Pyroelectricity versus Conductivity in Soft PZT Ceramics! 5.1. INTRODUCTION Doping Dependent Resistivity Resistance degradation is one of the issues encountered during the functioning of piezoceramics. In ceramics defects, which can serve as donor or acceptor, contribute to the extrinsic electrical conduction. The major defects in PZT ceramics are lead and oxygen vacancies caused by the volatility of the lead oxide during sintering 1,2,3. Gerson 4 explained the conduction mechanism in PZT ceramics. The p-type conduction is due to the Pb vacancies and the charge carriers (holes) are generated by the ionization of Pb vacancies 4. Alternatively, doping is another mechanism for introducing defects. When lower valence ions enter into the lattice, the charge deficit is compensated by oxygen vacancies 5,6,7 which are responsible for the resistance degradation 8,11. When higher valence ions enter into the lattice, excess charge is compensated by vacancies in the perovskite structure. This process leads eventually to a reduction in the charge carrier concentration and accordingly the resistivity is enormously increased 4,5,6,7,9, Resistance Degradation The insulation resistance degradation is characterized by the increase of the leakage current under simultaneous influence of increased temperature and an applied electric field 11. The majority of theories assume that the relatively high mobility of oxygen vacancies is essential to the degradation process. Oxygen vacancies are positively charged with respect to the regular lattice and, thus, under a dc bias they experience an electromigration towards the cathode. In front of the cathode, the oxygen vacancies pile up accompanied with a depletion of oxygen vacancies at the anode 11. The accumulation of oxygen vacancies at the cathode leads to a significant increase in electron concentration injected from the cathode, whereas in the oxygen vacancies depleted region, near the anode, the hole concentration increases 11. Based on this model, one expects resistance degradation to be more pronounced in systems that have a higher oxygen vacancy concentration 11. Therefore improved-resistance degradation behavior is attributed to a lower oxygen vacancy concentration (e.g. undoped or donor doped PZT). Donor doping is generally reported to lead to stabilization against degradation 11.! This chapter has been submitted for publication as: T. M. Kamel and G. de With, J. Mater. Res.,

97 Pyroelectricity versus Conductivity in Soft PZT Pyroelectric Influence In ferroelectrics, apart from the localized permanent dipoles, free charges (e.g. holes or oxygen vacancies) can exist in the bulk. As the temperature rises, the dipoles tend to disorder gradually due to the increasing thermal motion. Thus, ferroelectric materials exhibit a temperature dependent polarization (i.e. pyroelectric current) which disappears at Curie temperature T C. The space charges trapped at different depths are gradually set free, leading to a thermally activated electrical conduction current which overlaps with the pyroelectric current. Therefore, it is important to separate the different thermally stimulated currents of a ferroelectric material during heating under an electric bias. In this chapter, we shall focus on the dc conductivity of soft (modified donor doped) PZT ceramics under different applied fields. The influence of pyroelectricity on the conduction behavior will be investigated as well. As there is no clear evidence of an effect of the grain size on the dc conductivity of PZT ceramics according to the literature, this effect will be also investigated EXPERIMENTAL Non-poled polycrystalline ceramic samples of soft PZT (PXE52, modified and donor doped Pb(Zr.415 Ti.585 )O 3 ) were obtained from Morgan Electro Ceramics BV, Eindhoven, the Netherlands. The current measurements were carried out using a Keithley 6517 Electrometer on initially non-poled samples, with carefully cleaned surfaces. Ni-electrodes were sputtered to cover the sample s surfaces. The electric field was applied using the Keithley built-in voltage source and varied from.75 to 1 kv/cm. The current is continuously recorded while the sample is heated at a constant rate of 5 o C/min from room temperature (RT) to 3 o C. The temperature was measured using a type-k thermocouple. Previous results 12 for the pyroelectric current and temperature dependence of the coercive field are presently reused. For studying the grain size effect on the resistivity, the same composition was used and the grain size was varied in the same way as described previously 12,13. 85

98 Chapter (5) 5.3. RESULTS Ferroelectrics are highly polarizable materials. Thus, there is a considerable charging current upon applying a dc voltage to the specimen. To get the true values of leakage current, one usually has to collect readings from the current-time curve after reaching a steady state 14. As we have shown previously 12,15, the polarization current must go to zero at sufficiently long time, while the direct conduction current reaches a steady state 16. In the present study, the nonpoled sample is subjected to a dc voltage for a relatively long time to reach a steady state upon which heating starts. The recorded current under different dc voltages is plotted as a function of temperature in Figure 5.1. We can classify the curves, according to the applied field, to two groups; low fields and high fields. At low electric fields (E < E C = ~ 7.5 kv/cm), the current shows a complicated behavior (Figure 5.1A, B, C). As the temperature increases the current increases until it shows a maximum and thereafter it decreases. As the temperature increases further, a negative current develops and a minimum appears. Upon still further heating the current starts to increase continuously to the positive side. Figure 5.2 shows the current profile for different grain sizes. Increasing the grain size has a noticeable effect on the current behavior: the distance between the positive and negative current peaks increases as the grain size increases. Total Current (na/cm 2 ) E) 1 kv/cm D) 7.5 kv/cm C) 2.5 kv/cm B) 1.75 kv/cm A) 1.25 kv/cm H i g h F i e l d L o w F i e l d Temperature ( o C) Figure 5.1: The current as a function of temperature at different dc electric fields for nonpoled samples. Grain size used is 4.5 µm. 86

99 Pyroelectricity versus Conductivity in Soft PZT Current (na/cm 2 ) (4) (3) (7) (6) (5) (2) Applied Field: 2.5 kv/cm 2 (1) Temperature ( o C) Figure 5.2: The current as a function of temperature for nonpoled samples for different grain sizes under low electric field. (1) 2. µm; (2) 3.45 µm; (3) 4.5 µm; (4) 5.7 µm; (5) 7. µm; (6) 8.6 µm; (7) 9.8 µm. At high fields (E E C ), the current curves have different shape (Figure 5.1D, E). Upon heating, a negative current occurs until it shows a diffuse minimum at ~215 o C followed by an increase in a similar manner as for the low field behavior. A similar behavior is observed for all grain sizes, Figure 5.3. Conduction Current (µa/cm 2 ) Applied Field: 7.5 kv/cm Temperature ( o C) Figure 5.3: The current as a function of temperature for nonpoled samples at different grain sizes at a field equal to the coercive field. (1) 2. µm; (2) 3.45 µm; (3) 4.5 µm; (4) 5.7 µm; (5) 7. µm; (6) 8.6 µm; (7) 9.8 µm

100 Chapter (5) 5.4. DISCUSSION Field effect It is well known that, increasing the temperature enhances the polarization switching process 17. Merz 17 showed that the switching time decreases as the temperature increase. The switching current is also increasing with increasing temperature 17. Similarly, we showed previously 12 that, at low fields, comparable to the low field range presently used, switching is not possible at RT. However, with increasing temperature switching commences and becomes more pronounced as the temperature increases. We showed also that the increase in the switching current with temperature was in agreement with Merz s observations 17. Coercive Field (kv/cm) 8 7 Grain size = 4.5 µm Temperature ( o C) Pyroelectric Current (na/cm 2 ) Figure 5.4: The temperature dependence of the pyroelectric current (!) and the coercive field ("). Based on the above information, the complexity of the current curves shown in Figure 5.1 can be resolved. By selecting a current curve (say the one at 2.5 kv/cm), and considering a typical pyrocurrent curve as shown in Figure 5.4 as well as the temperature dependency of the coercive field E C as shown in Figure 5.4, we may explain the current behavior as follows. The current curve can be divided into three zones as indicated in Figure 5.5: Total Current (na/cm 2 ) Zone 1 pol Applied Field: 2.5 kv/cm Zone 2 Zone 3 depol Temperature ( o C) Figure 5.5: A model current-temperature curve for a nonpoled sample. Zone 1 represents the thermal assisted polarization current (charging of the sample); pol is the total polarization charge. Zone 2 represents the thermal depolarization (pyroelectric) current; depol is accounted for the depolarization charge. Zone 3 shows the conductive part of the current at high temperature (well above the Curie temperature). 88

101 Pyroelectricity versus Conductivity in Soft PZT Zone 1: the positive current branch (at T >T C ); assigned to thermal assisted polarization. Zone 2: the negative current branch (at T >T C ); assigned to thermal depolarization (pyroelectric effect). Zone 3: the growing positive current (at T >>T C ); assigned to electrical conduction. Figure 5.4 shows that switching occurs at lower field by increasing the temperature due to the decrease in the coercive field E C. Accordingly, the positive current branch (Zone 1) can be attributed to the charging (poling) process. The charging current is thermally activated and goes to zero as expected for a charging process. As the material now is charged (poled), further heating leads to a thermal depolarization current (Zone 2) with a sign opposite to the polarization current. Accordingly, the current peak shown in the negative direction is associated with the ferroelectricparaelectric (FE) phase transition with a transition temperature T C rather different from the normal value as observed from the pyroelectric curve due to the applied field, Figure 5.4. It is well known that, at a temperature close to a FE phase transition, the polarization still can be induced by a bias field 18. Consequently, the actual FE phase transition (P = ) moves towards higher temperature due to the effect of the external bias field 18. This explanation as discussed above can be checked out by comparing the value of polarization and depolarization charges which must, theoretically, be equal. Table 5.1 shows the area under the current-temperature curves pol and depol representing the polarization and depolarization charge, respectively, at different applied fields (low field regime). Table 5.1: Polarization and depolarization values at different applied electric fields calculated from the areas under peaks (according to the definition as given in Figure 5.5). Applied Field kv/cm pol µc/cm 2 depol µc/cm It is obvious that the value of the depolarization charge is close to that of the polarization charge. The differences can be attributed to the counteracting contribution of the conduction current during the temperature increase, as will be shown later. Finally, the positive current in Zone (3) is purely conductive. The electrical conduction as a function of temperature will be discussed in detail later. Current curves at higher fields, Figure 5.1D, E (E > E C ) show a different shape. The current is almost negligible with a negative sign over the temperature range up to 2 o C. Around 215 o C the current starts to increase nonlinearly in the positive direction. It should be noted that at these fields (E > E C ) the sample can be 89

102 Chapter (5) instantaneously polarized in a time as short as 1 s implying that the polarization current already faded out a few seconds after turning on the voltage on. Consequently, the current shown in Figure 5.1 represents the thermal depolarization. Since it is counteracted by the increasing thermal/bias assisted conduction current, it shows a small value until it completely vanishes at 2 o C. The nonlinear positive current at higher temperature is due to electric conduction similar as for the low field case. Conduction Current (µa/m 2 ) Grain size = 4.5 µm At 28 o C Bias Field (MV/m) Figure 5.6: Conduction current as a function of bias field at 28 o C. Figure 5.6 shows the I-V characteristic at high temperature (28 o C). At this temperature the I-V characteristic is fairly linear (ohmic resistance) over the electric field range used and consequently the temperature dependence of resistivity ρ at high temperature, Figure 5.7, can be well fitted with an Arrhenius equation 9,19,2,24,27 (ignoring a 1/T factor 21 ): W / kt ρ = ρo e (5.1) where a ρ o is constant, k is Boltzmann s constant and W is the activation energy for conduction. Considering only the high temperature part, since at low temperature the displacement current due to the (de-)polarization process dominates the current, the activation energy W is calculated as 1.55 ev. This value is comparable to reported values by other authors on similar PZT systems 1,2,24,27,28,29. Figure 5.6 shows also that, the average resistivity at 28 o C is 2.2 x 1 1 Ω.cm which is almost 2 orders of magnitude lower than the value at RT (3.6 x 1 12 Ω.cm). The room temperature resistivity is comparable to values previously reported by other researchers 6,26,29. For convenience, Table 5.2 summarizes the material s electrical parameters. 9

103 Pyroelectricity versus Conductivity in Soft PZT ln (Resistivity ρ, Ω.m) T ( o C) Grain size = 4.5 µm Applied Field: 1.25 kv/cm /T ( o K -1 ) Figure 5.7: Arrhenius plot for the dc resistivity (high temperature region) at applied electric field: 1.25 kv/cm. Undoped PZT ceramics exhibit p-type conductivity due to the presence of unintentionally introduced acceptor species. In PZT, the dominant acceptor species are expected to be Pb vacancies, which are associated with PbO volatilization during sintering. These negatively charged acceptors require the introduction of positively charged species to maintain bulk charge neutrality. The charge neutrality condition can be given using Kröger-Vink notation 22 by 24,25 V V + [ h] (5.2A) Pb Pb 2[ V Pb ] = 2 [ VO ] + [ h] (5.2B) where [ ] denotes concentration and h, V Pb, V Pb, and VO denote electron holes, lead vacancies, doubly ionized lead vacancies, and oxygen vacancies, respectively. According to eq. 5.2A, the unintentional incorporation of acceptor species generates defects that can increase the conductivity. The charge neutrality condition (eq. 5.2A) can be modified through the addition of donor impurities to be where D D + e A, (5.3) ( B) ( A,B) D is a B-site donor ion and ( A,B) D B ( A, ) is an ionized ion. Donor doping can result in a dramatic decrease in the concentration of the mobile defects ( h andv O ) and consequently a decrease in the electrical resistivity 26. Ionic vacancies are much more abundant than electronic, i.e., V Pb V O >> p. However, the hole mobility is significantly larger than ion vacancy mobility, and it 91

104 Chapter (5) appears that the conductivity of PZT is dominated by the motion of holes rather than oxygen vacancies 29. The resistivity measurements may show different activation energies for both the ionic and electronic conductivity. At higher temperatures, increased ionic vacancy mobility may eventually allow the ionic conduction mechanism to become significant 29,3. Therefore, the value of the activation energy for conduction at high temperature for the present material suggests that the conduction mechanism is ionic. Table 5.2: PXE52 electrical properties. Density (g/cm 3 ) 7.45 Grain size (µm) 4.5 Curie temperature T C ( o C) RT Coercive field E C (kv/cm) RT Resistivity (Ω.cm) Resistivity (28 o C) (Ω.cm) Activation energy W (ev) Grain Size Effect Increasing the grain size has a remarkable effect on the current profile. As shown in Figure 5.2, the distance between the polarization and depolarization peak increases by increasing the grain size. As we discussed previously 23, increasing the grain size enhances and motivates the polarization as well as increases the Curie temperature. Consequently, we may expect an earlier appearance of the polarization peak and later appearance of the depolarization peak during heating for larger grain size. At high field, the current is almost zero until 18 o C. Above 18 o C the current increases nonlinearly with temperature and proportional to the increase in grain size for the various specimens. At high field, both the displacement current and conduction current are nearly equal in magnitude and opposite in sign and therefore cancel each other. As the displacement current ceases at the Curie temperature, the conduction current increases again in nonlinear fashion with further increase of the temperature. Table 5.3: Electrical resistivity as a function of grain size of PXE 52. The activation energies were determined according to eq. 5.1, using the same temperature range as given in Figure 5.7. Grain size (µm) RT Resistivity (Ω.cm) Resistivity at 22 o C (Ω.cm) Activation energy W at high temperature (ev)

105 Pyroelectricity versus Conductivity in Soft PZT According to the literature, there is no clear evidence of an effect of the grain size on the dc conductivity in bulk PZT ceramics. It was shown that, in our case, at high temperature the resistivity decreases nonlinearly as the grain size increases, Figure 5.8. Because the activation energy of conduction does not change significantly with grain size (averaged over all grain sizes yielding 1.35 ev), this probably implies that the conduction mechanism doesn t change with grain size. There can be an increase in charge carrier density which didn t change in nature ( VO and holes) 24,25. Donor doping suppresses the conduction process in PZT by introducing electrons 24,26 to compensate the positive charge carriers ( V and holes), while increasing the grain size by sintering at elevated temperature produces Pb vacancies due to lead evaporation. As a result, more oxygen vacancies and holes are created to maintain the charge neutrality. Hence, an increase in conduction is expected. O Electrical Resistivity Temperature: 22 o C Applied Field: 2.5 kv/cm 7.5 kv/cm Grain Size, d gs (µm) Figure 5.8: The electrical resistivity as a function of grain size at different electric fields (2.5 and 7.5 kv/cm) at 22 o C. At low temperature the conduction current is entirely masked by the pyroelectric current and it is not possible to resolve the pure conduction contribution. The same behavior was also observed by other workers 27,28. However, it is widely accepted that hole conduction is predominant at RT 8,24,26,29. No major change in the resistivity as a function of grain size was observed. This behavior supports the fact that the conduction at room temperature is dominated by electronic mechanism, since holes are expected to be trapped by lead vacancies 26. Consequently, the conductivity is expected to be low 26 and almost constant. As the temperature increases, VO defects become more mobile and significantly contribute to the conduction process 8,26,3 leading to a higher conductivity proportional with the grain size. 93

106 Chapter (5) CONCLUSIONS The electric conduction of modified donor doped (soft) PZT ceramics as a function of temperature and grain size has been investigated. Due to the highly pronounced pyroelectric effect of this material, it was not possible to recognize the conduction current as a function of temperature near room temperature, since it is entirely hidden by the pyroelectric current. However, the steady state resistivity was found to have a typical high value of 3.6 x 1 12 Ω.cm at room temperature. The high temperature resistivity was found to be two-orders of magnitude lower than the RT value. Increasing the grain size has no large effect on the resistivity at room temperature due to the trapping of the dominating charge carriers (holes) by lead vacancies. Nevertheless, it led to resistance reduction at high temperature, probably due to the increased Pb vacancy concentration resulting as a consequence of a higher sintering temperature. However, no resistance degradation has been observed in all grain sizes at all conditions. 94

107 Pyroelectricity versus Conductivity in Soft PZT REFERENCES 1 B. Jaffe, W. R. Cook and H. Jaffe, Piezoelectric Ceramics, Academic Press, New York, D. A. Northrop, J. Am. Ceram. Soc., 5, 441, R. L. Holman, J. Am. Ceram. Soc., 55, 192, R. Gerson, J. Appl. Phys., 31, 188, R. Gerson and H. Jaffe, J. Phys. Chem. Solids, 24, 979, M. Takahashi, Jpn. J. Appl. Phys., 1, 643, L. Wu, T.-S. Wu, C.-C. Wei and H.-C. Liu, J. Phys., 16, 2823, H. N. Al-Shareef and D. Dimos, J. Am. Ceram. Soc., 8, 3127, M. M. Nodliisky, S. D. Toshev and T. K. Vasileva, Phys. Stat. Sol., 54, K145, N. Setter, Piezoelectric Materials in Devices, Ceramics Laboratory, EPFL, R. Waser, T. Baiatu and K. Härdtl, J. Am. Ceram. Soc., 73, 1645, 199, and references therein. 12 T. M. Kamel, F. X. N. M. Kools and G. de With, J. Eur. Ceram. Soc., 27, 2471, Chapter (3) Part II of this thesis. 14 H. Hu and S. B. Krupanidhi, J. Mater. Res., 9, 1484, Chapter (3) Part I I of this thesis. 16 A. K. Jonscher, Dielectric Relaxation in Solids, Chelsea Dielectrics, London, W. J. Merz, Phys. Rev., 95, 69, M. E. Lines and A. M. Glass Principles and Applications of Ferroelectrics and Related Materials, Clarendon Press, Oxford, Z. Zhu, N. Zheng, G. Li and Q. Yin, J. Am. Ceram. Soc., 89, 717, J. J. Dih, D. R. Biswas and R. M. Fulrath, J. Mater., Sci., 16, L. L. Hench and J. K. West, Principles of Electronic Ceramics, Wiley, New York, p. 139, F.A. Kröger and H.J. Vink, Solid State Physics, Vol. 3, 3 p. 37, eds. by F. Seitz and D. Turnbull. Academic Press, New York, T. M. Kamel and G. de With, J. Eur. Ceram. Soc., (accepted) J. J. Dih and R.M. Fulrath, J. Am. Ceram. Soc., 61, 448, B. A. Boukamp, M. T. N. Pham, D. H. A. Blank, H. J. M. Bouwmeester, Solid State Ionics, 17, 239, D. Dimos, R. W. Schwartz and S. J. Lockwood, J. Am. Ceram. Soc., 77, 3, E. Furman, W. Gu and L. E. Cross, IEEE Int. Symp. Appl. Ferroelectr., 7 th, 588, A. P. Barranco, F. C. Pinar, O. P. Martinez, J. Mater. Sci. Lett., 2, 1439, N. J. Donnelly, T. R. Shrout and C. A. Randall, J. Am. Ceram. Soc., 9, 49, J. Lappalainen and V. Lantto, Phys. Scripta, T79, 22,

108

109 CHAPTER 6 POLING OF HARD PZT ABSTRACT Poling of hard ferroelectric ceramics is not an easy process. Pinning of domain walls due to defects prevents the material to be switched by moderate conditions of poling. In this chapter, we try to explore a way to facilitate the poling process by defeating the counteracting defects. We shall focus on the structural and electrical effects that enhance the domain mobility, as presented by hysteresis loops, that lead to depinched hysteresis loops and consequently to easier poling and eventually to better piezoelectric properties. Different poling conditions were applied to ceramic samples sintered at different temperatures before and after hysteresis depinching. Dielectric parameters were measured after each poling state. The results showed that the hard ceramic can be efficiently polarized only after domain depinning. We showed that the domains can be depinned by different methods.

110 Chapter (6) Poling of Hard PZT! 6.1. INTRODUCTION Domain walls in ferroelectrics are the boundaries that separate domains polarized in different directions 1. The presence of domain walls is a result of multidomain structure of many ferroelectric crystals. Multidomain structure of a polar crystal minimizes the electrostatic energy of the depolarizing field and the elastic energy associated with mechanical constraints to which the ferroelectric material is subjected to as it is cooled through the paraelectric to ferroelectric phase transition. In soft ferroelectric ceramics, the domain walls move irreversibly while in hard ceramics, the domain walls are clamped and perform mainly reversible movements. The hard materials are characterized by biased and/or pinched hysteresis loops. This behavior is attributed to the crystal lattice defects that act as pinning centres for the domain walls causing its nonuniform movements. As is well known, a piezoelectric ceramic gets in use by applying adequate poling conditions (electric field, poling time, and poling temperature) to a ferroelectric ceramic. Consequently, the piezoelectric properties are greatly dependent on the poling conditions 2-8. It has been verified that there is a direct dependence between the dielectric and piezoelectric properties and the degree of domain orientation that is set by poling conditions In polycrystalline perovskite ferroelectrics domain wall motion is the dominant switching mechanism of polarization reversal. The domain wall mobility in turn is determined by the interaction with structural defects. While a homogeneous distribution of point defects in the bulk of a crystal yields only a slightly reduced domain wall velocity, larger clusters may completely block the motion of certain domain walls 16. Since the domain walls in ferroelectrics are relatively thin they can be pinned by point defects or charged species 1. Poling of hard ferroelectric ceramics is not an easy process. Pinning of domain walls due to defects prevents the material to be switched by moderate conditions of poling 17. Unfortunately, literature on poling of hard ceramics is very scarce. However, several authors have discussed the defects and degradation in hard ceramics (e.g. PZT) which are believed to be responsible for the difficulty of poling in these materials. In this chapter, we try to explore a way to facilitate the poling process by defeating the counteracting defects. Acceptor dopants for Pb 2+ (Zr 4+,Ti 4+ )O 3 such as Fe 3+ ions are compensated by oxygen vacancies as such a combination of defects satisfies the requirement for local charge neutrality. During sintering the (effectively negative) dopant ion and (effectively positive) oxygen vacancy ( V O ) form a! This chapter has been submitted for publication as: T. M. Kamel and G. de With, J. Eur. Ceram. Soc.,

111 Poling of Hard PZT 3+ defect dipole ( ) FeZr, Ti V O. However, the oxygen vacancies are still movable below the Curie temperature (i.e. at room temperature) because the oxygen ion and oxygen vacancy are adjacent (only 2.8 Å apart) and hopping easily occurs 3. The orientation of defect-dipoles takes place via oxygen vacancy diffusion 3. This diffusion motion leads to domain wall immobility and eventually a constricted hysteresis loops, low dielectric constant, very low dielectric loss, low piezoelectric coupling factor and consequently acceptor doped PZT (or Hard PZT as commonly used) are very difficult to polarize 18. In the following, the models suggested for the role of defects in hard ferroelectric ceramics will be summarized Models of Domain Wall Stabilization; Literature Review One consequence of domain wall switching in ferroelectrics is the occurrence of ferroelectric hysteresis loops 19. In some cases the hysteresis loop is biased and/or pinched 19. This phenomenon is usually attributed to domain wall stabilization caused by ordering of charged defects 17. Hysteresis loop asymmetry or constriction was observed in many ferroelectrics 2. When a ferroelectric is doped with an acceptor dopant (a lower valence than the host atom, e.g. Fe 3+ or Mn 3+ on Ti 4+ or Zr 4+ sites in PZT), a positively charged oxygen vacancy V O is created, which neutralizes the effective negative charge introduced by the acceptor dopant 1. Lambeck and Jonker 21 showed that reorientation of the defect-dipole is expected to be responsible for domain stabilization. A perovskite lattice cell containing such defect-dipoles may remain paraelectric even if all neighboring cells are ferroelectric 17. In the paraelectric phase the oxygen vacancy may jump by thermal activation to the energetically equivalent neighboring oxygen sites of the lattice cell, thus changing the orientation of the electric moment of the defect-dipole. Upon cooling from the paraelectric phase to the ferroelectric phase a spontaneous polarization P S is established. The energy of the electric dipole then becomes different in the six orientations as shown in Figure 6.1. Acceptor doped PZT.. V Ȯ 3+ Fe (Zr,Ti) 3+ Figure 6.1: PZT unit cell containing ( ) Fe V dipolar defect. Zr, Ti O 99

112 Chapter (6) If the lattice has a polarization direction in the +Z-direction, Figure 6.1, three possible positions of the oxygen vacancy with three different energy levels are expected. If a transition takes place to lower energy site, the point defect may contribute to ageing of the ceramic 17. Accordingly, ageing is defined as the spontaneous change (increase or decrease) of a property with time under zero external stress and isothermal conditions 22. Many authors investigated the ageing behavior and several plausible models have been developed to interpret this phenomenon. A quantitative theory about the physics of ageing mechanism is, however, not yet put forward. Most of the models are of a qualitative nature. In the following we shall present the models that may cause the domain wall stabilization in bulk ferroelectric ceramics. Carl and Härdtl 23 classified the possible mechanism that may responsible for the occurrence of internal bias in ferroelectric ceramics in three basic effects 23 : i) Volume effect. Ceramics always contain defects, intentionally (i.e. doping) or unintentionally (i.e. impurities), in the form of foreign atoms or vacancies. These defects can occupy energetically preferred sites in the lattice and favor a certain direction for the spontaneous polarization. This favoring of a particular direction of spontaneous polarization reveals itself experimentally as an internal bias. ii) Domain effect. This effect is also due to defects which diffuse in the course of time into the domain walls and fix their position. The driving forces may be elastic (neutralization of internal stresses) or electric (compensation of electric charges, e.g. by valency changes of the foreign atoms). iii) Grain boundary effect. Several ferroelectric ceramics contain a secondary phase along the grain boundaries and/or on the triple points. The second phase always occurs at high dopant concentrations if the concentration of the dope exceeds the solubility limit. Second phases lead to surface charges at the grain boundaries, which stabilize the domain configurations once these are established. During ageing the individual crystallite is biased by an overall preferred direction of polarization, but the individual domain wall remains mobile. Robels and Arlt 24 showed that the decrease of the material properties in time corresponds to an increasing shift of the hysteresis loop along the E-axis. This shift is called internal bias field, E i. The internal field in acceptor doped ceramics can be explained quantitatively 17 by oriented dipole defects. Domain clamping by orientation of defect-dipoles according to Robels and Arlt 24 can be summarized as follows. During ageing of acceptor doped ceramics, an increasing shift of the hysteresis along the E-axis occurs. The build up of E i in time has been explained by a slow relaxation of the defect-dipoles. When a wall is displaced by l (Figure 6.2), the dipolar defects in this region do not reorient immediately. Thus the total energy of a domain wall which is displaced by the length l increases by () t n( t A l We W = (6.1) DW ) 1

113 Poling of Hard PZT where n(t) A l is the number of defects in the shifted region as a function of time and A is the area of the domain wall and W e is the energy associated with a defect oriented in unfavorable direction with respect to the spontaneous polarization. Therefore the contribution of domain walls to the dielectric and piezoelectric properties should be decreasing with increasing orientation of the defects 22,23. P S P S l Figure 6.2: Most of the defect dipoles (small arrows) are oriented parallel to P S (large arrow) within the domains. If the domain wall is displaced by l, the P S and the defect dipole will be misaligned (Robels and Arlt 24 ). In this chapter we shall focus on the structural and electrical effects that enhance the domain mobility, as presented by hysteresis loops, that lead to depinched hysteresis loops and consequently to easier poling and eventually to better piezoelectric performance. Throughout this chapter the ceramic densification and thermal treatment (quenching) will be considered as structural effects whilst poling electric field and temperature as electric effects EXPERIMENTAL Material Non-poled [(Pb Sr.6 ) (Fe.8 Zr x Ti 1-x-.8 ) O 3 ] ceramic powder, near the morphotropic phase boundary, coded PXE43, were provided by Morgan Electro Ceramics B.V. Eindhoven, the Netherlands. Addition of Sr was intended to lower the Curie transition temperature 18. The powder was pressed into bars of at least 15 mm in diameter and sintered at different temperatures ( o C) in a tube furnace with flowing oxygen. The following sintering profile was applied for all bars. The bar was heated from room temperature to the desired temperature in 8 hours and held at this temperature for 4 hours and then cooled down to room temperature in 8 hours. The sintered bars were diced into discs of 1 mm in diameter. Next, the discs were subjected to clean-firing at 7 o C for 48 hours and then cleaned in methanol in an ultrasonic bath for two minutes and then washed in water and soap and dried in air. Finally, the discs were electroded with Ni by sputtering for electrical measurements. 11

114 Chapter (6) Microstructure The microstructure of the as-prepared samples was investigated using optical as well as scanning electron microscopy (JEOL, JSM 84A, Japan). First, the sample was molded in conductive resin, ground and successively polished. Then the sample surface was etched using 95% water, 5% HCl, and five drops of HF as an etching agent 31. To avoid charging during SEM measurements, the sample surfaces were coated with gold by sputtering. The grain size d gs was measured as a function of sintering temperature, using mean linear intercept, counting about 1 grains Hysteresis Loop Hysteresis loops of the hard PZT were observed using a computer-controlled virtual ground method (Radiant Technologies Inc., RT6 HVA-z 2V amplifier and RT6 HVS-z high voltage input test system) using the fast mode (.4 ms) at different temperature steps (RT 15 o C). The maximum applied electric field was (5 kv/cm). Silicone oil (Wacker Silicone Fluid, AP 15) was used as an insulating medium in an oil bath. The loops were not completely closed due to the nonswitchable polarization 32, Poling As mentioned before, the poling conditions are divided into structural and electrical conditions. Structural: ceramic densification effect and thermal effect (quenching). Electrical: Poling electric field and temperature. As-prepared non-poled samples, sintered at different temperatures, were poled at different temperatures (7, 9, 1 o C). The poling electric field was kept constant over the whole poling process at 3 kv/cm. The poling procedures can be described as follows. The sample is heated up to the desired temperature in the silicone oil bath under application of the electric field. The field increases gradually at a constant rate 1 V/s until it reaches 3 kv/cm. Then the sample is held under these conditions for 1 min. During the last minute the sample is cooled down, while the field still applied, to room temperature whereafter the electric field is removed. Worthwhile to mention is that, unlike for the soft ferroelectric ceramic (PXE52) 3, we found out that the polarization current couldn t be monitored during the poling process due to the high leakage current which completely hid the polarization current Dielectric, Piezoelectric and Pyroelectric Properties The dielectric and piezoelectric parameters (ε r, tan δ, k p, k 33, k eff ) of poled and (ε r, tan δ) of non-poled samples were measured using (HP4294A Precision Impedance Analyzer) at 1 khz in the temperature range from room temperature up to 4 o C. The pyroelectric measurements were carried out using the direct method 33. The setup used was described elsewhere 3. 12

115 Poling of Hard PZT 6.3. RESULTS Microstructure Figure 6.3(1 7) depict the micrographs of the nonpoled samples. The average grain size ranges from.9 µm for 12 oc to 1.7 µm for 13 oc. Table 6.1 shows the variation of the density and grain size with sintering temperature. It is observed that the ceramic showed high densification at a sintering temperature above 12 oc reaching very dense structure at 13 oc. However, there is not a significant variation in grain size because the grain growth is inhibited due to the acceptor doping18. During sintering, due to the incorporation of the acceptor dopant (Fe3+) in the lattice, the material becomes oxygen-deficient due to the production of oxygen vacancies as a requirement of electrical neutrality. The charged oxygen vacancies accumulate at the grain boundaries making a screening space charge inhibiting the grain growth 35,36. On the other hand, the samples sintered at lower temperature showed a porous structure, so that the grain size couldn t even be determined properly. # " 2.5 µm 2.5 µm $ % 2.5 µm 1 µm & ' 2.5 µm 2.5 µm ( 2.5 µm Figure : SEM micrographs of Fe doped PZT samples sintered at: " 115 oc, # 1175 oc, $ 12 oc, % 1225 oc, & 125 oc, ' 1275 oc, (13 oc. 13

116 Chapter (6) Table 6.1: Sintering temperature and the corresponding resultant densities. Sample Sintering Temperature Grain Size Density (ρ exp ) Name ( o C) (µm) g/cm 3 ρ exp /ρ th % " 115 ) # 1175 ) $ % & ' ( )) Due to the porous structure, the grain size couldn t be determined Hysteresis Loops for As-Prepared Samples Figures 6.4, 5, and 6 show the hysteresis loops of samples % ' at room temperature. A double hysteresis loop is a common characteristic of all samples. As is shown in the figures, at room temperature the loops are still constricted even at very high field (5 kv/cm). Double (pinched) hysteresis loops are attributed to a reversible movement of domain walls as they are electrically and/or elastically 44 pinned by defects. Polarization (µc/cm 2 ) C A B D Electric Field (kv/cm) Figure 6.4: Hysteresis loops at different electric fields of non-poled sample % at room temperature. Sintering temperature: 1225 o C. 14

117 Poling of Hard PZT Polarization (µc/cm 2 ) C 15 D A B Electric Field (kv/cm) Figure 6.5: Hysteresis loops at different electric fields of non-poled sample & at room temperature. Sintering temperature: 125 o C. Polarization (µc/cm 2 ) C D A B Electric Field (kv/cm) Figure 6.6: Hysteresis loops at different electric fields of non-poled sample ' at room temperature. Sintering temperature: 1275 o C Poling Effect on the Dielectric and Piezoelectric Properties for Aged As-Prepared Samples The dielectric constant ε r, dielectric dissipation factor tan δ and the planar coupling factor k p at room temperature as a function of the sintering temperature after poling at different temperatures are presented in Figures 6.7A 7C. The poling was done for initially as-prepared nonpoled samples. At first glance, it is obvious that the densification of the ceramic considerably enhances the dielectric and piezoelectric 15

118 Chapter (6) properties due to relatively enhanced domain wall mobility as a result of decreasing of the pores number. However, the poling has a small influence. The dielectric dissipation factor tan δ shows anomalous behavior compared to the soft counterpart (PXE52) 3 as it increases by increasing the poling which means that the domain wall still movable or rather more precisely, the full polarization is not yet set. Adding to that, the noticeable low value of tan δ is attributed to ageing 22 and domain wall stabilization 23 as a consequence of the association of the domain wall movement and dielectric loss 6,22. This observation is consistent with the constricted hysteresis loops shown in Figures Dielectric Constant A 7 Nonpoled 7 o C 6 9 o C 1 o C Sintering Temperature ( o C) tan δ B Nonpoled 7 o C 9 o C 1 o C Sintering Temperature ( o C) Planar Coupling Factor (k p ) C.3 Poling Temp. 7 o C.25 9 o C 1 o C Sintering Temperature ( o C) Figure 6.7: Dielectric constant (A), dielectric loss (B) and planar electromechanical coupling factor (C) after direct poling at different temperatures. To determine the Curie transition temperature, the dielectric constant was measured as a function of temperature at 1 khz, Figure 6.8. As can be seen in this figure, the sample shows a typical ferroelectric phase transition peak at a Curie transition temperature of 335 o C. The onset of the paraelectric phase state is accompanied with the disappearance of the electromechanical coupling factor k p. This rules out any other possible causes for the anomalous behavior of the hysteresis loops at room temperature, e.g. antiferroelectricity. Planar Copuling Factor (k p ) Temperature ( o C) Figure 6.8: Dielectric constant and the planar electromechanical coupling factor as a function of temperature for sample poled at 3 kv/cm (7 o C) after electric depinning at (5 kv/cm, 15 o C) Dielectric Constant 16

119 Poling of Hard PZT Hysteresis Loop Depinching It is widely known that there are only two ways for depinching the hysteresis loop 1. Electrically, by repeated cycling (hundred or even thousands of times) of the hysteresis loop, the pinching will disappear and the internal bias will be reduced 1,21-23,34,38,39. This process is normally done above the Curie temperature 21,22. This effect is called hysteresis relaxation 21,23. The relaxation of the internal bias obeys, according to Carl and Härdtl 23, an Arrhenius law: E i t τ () t = E () exp i (6.2) The relaxation time τ depends on the temperature and the maximum applied electric field (i.e. a thermally and electrically activated process). In some cases the depinching cannot be completely relaxed even after a large number of cycles 1. The pinning can be structurally removed by quenching the sample from well above the Curie temperature to room temperature 1,2. This effect is called thermal relaxation 21,22. However, we have found, in our case, that the loop can be gradually depinched by increasing the temperature only up to the poling temperature, which is one-third the Curie temperature, and with no need of repeated cycling. Figures show the hysteresis loops at 7, 8, 95 and 15 o C. As shown in these figures, the loop is gradually changed to normal shape by increasing the temperature, Figures , and the applied field, Figures It can be easily noticed that, there is a transition temperature (obviously 95 o C) at which (along with the highest applied field) the pinched loop can be relaxed due to the thermal activation. In other words, assuming a maximum applied electric field, the remanent polarization increases from zero to non-zero magnitude by increasing temperature and/or by the densification of the ceramic. This behavior is clearly shown in Figure Worthwhile to mention is that, the bias internal electric field, as calculated according to [17 and 23], is significantly decreasing with increasing density of the ceramic, Figure This behavior can be explained as follows. At low ceramic density, the pores are very generously distributed. Hence, the domain wall mobility may be hindered because of space charges that accumulate on the pores. Nevertheless, the route of hysteresis loop depinching followed above was found to be effective for depinching and lowering the internal bias in the more dense samples. It was found that, the internal bias can be greatly reduced after quenching from well above (5 o C) the T C (~ 33 o C) to room temperature. The hysteresis loops after quenching are shown in Figure 6.2. Figure 6.21 shows the internal bias after quenching as a function of the sintering temperature. The internal bias is slightly decreasing in a nearly linear manner by increasing the ceramic density. Due to the fast cooling ordering of the oxygen vacancies or their dipoles are hindered. Consequently, the ceramic after the quenching process is in a metastable state. 17

120 Chapter (6) Polarization (µc/cm 2 ) 2 15 (A) 7 o C 2 15 (B) 8 o C (C) 95 o C 15 (D) 15 o C (E) 115 o C Electric Field (kv/cm) Figure 6.9: Hysteresis loops at different temperatures of a non-poled sample " at the maximum applied electric field (5 kv/cm). Polarization (µc/cm 2 ) (A) RT (C) 8 o C 2 (D) 95 o C Electric Field (kv/cm) (B) 7 o C (E) 15 o C Figure 6.1: Hysteresis loops at different temperatures of a non-poled sample # at the maximum applied electric field (5 kv/cm). Polarization (µc/cm 2 ) (A) RT 2 (B) 7 o C (C) 8 o C 2 (D) 95 o C (E) 15 o C Electric Field (kv/cm) Polarization µc/cm (A) RT 3 15 (B) 7 o C (C) 8 o C 3 (D) 95 o C (E) 15 o C Electric Field (kv/cm) Figure 6.11: Hysteresis loops at different temperatures of a non-poled sample % at the maximum applied electric field (5 kv/cm). Figure 6.12: Hysteresis loops at different temperatures of a non-poled sample & at the maximum applied electric field (5 kv/cm). 18

121 Poling of Hard PZT Polarization (µc/cm 2 ) 4 3 (A) RT 4 3 (B) 7 o C (C) 95 o C 3 (D) 15 o C Electric Field (kv/cm) Polarization µc/cm (C) 95 o C 3 (D) 15 o C (A) 7 o C (B) 8 o C Electric Field (kv/cm) Figure : Hysteresis loops at different temperatures of a non-poled sample ' at the maximum applied electric field (5 kv/cm). Figure 6.14: Hysteresis loops at different temperatures of a non-poled sample ( at the maximum applied electric field (5 kv/cm). Polarization (µc/cm 2 ) (A) (C) 15 (D) (B) Electric Field (kv/cm) Polarization (µc/cm 2 ) (A) 15 (B) (C) 15 (D) Electric Field (kv/cm) Figure : Hysteresis loops at different electric fields of a non-poled sample % at 7 o C. Figure 6.16: Hysteresis loops at different electric fields of a non-poled sample % at 8 o C. 19

122 Chapter (6) Polarization µc/cm (A) 28 (B) (C) 28 (D) Electric Field (kv/cm) Polarization (µc/cm 2 ) (C) 3 (D) (A) (B) Electric Field (kv/cm) Figure 6.17: Hysteresis loops at different electric fields of a non-poled sample ' at 95 o C. Figure 6.18: Hysteresis loops at different electric fields of a non-poled sample ( at 15 o C Poling Effect on the Ferroelectric and Piezoelectric Properties after Deageing Hysteresis Loops Poling of hard ceramics is a difficult process while they are in an aged state. However, they can be poled into a stable state after deageing. After hysteresis depinching by our method the ceramic can be subjected to the normal poling procedure described above. Figure 6.22 shows the P E curves at room temperature after poling at 1 o C under applied field of 3 kv/cm and field cooled. It is seen that a strong asymmetric polarization hysteresis curve results, revealing a unipolar polarization orientation after poling. Clearly, field cooling results in irreversible changes in the defect distributions and domain stability unless the temperature is increased to enhance defect mobility. A similar behavior is observed by other workers 39,4. The large field shift is due to domain pinning by defects. The defects rearrange with the ferroelectric polarization at higher temperature, and freeze-in upon cooling resulting a stabilized polarization 4. Remanent Polarization (µc/cm 2 ) [1] [4] [5] [6] [7] Maximum Applied Field: 5 kv/cm As prepared nonpoled samples Density increases Hysteresis Loop Temperature ( o C) Figure 6.19: Hysteresis loops opening as a function of temperature for different densities. 11

123 Poling of Hard PZT Polarization (µc/cm 2 ) Sample [1] Sample [3] Sample [7] Sample [2] Electric Field (kv/cm) Sample [5] Figure 6.2: Hysteresis loops opening at room temperature after quenching (depinning) Dielectric and Piezoelectric Properties The dielectric constant, dielectric dissipation factor and the planar coupling factor at room temperature as a function of the sintering temperature after normal poling at different temperatures are presented in Figures 6.23A 23C. The poling was done after the electrical deageing. As shown in these figures, there is no significant influence of the poling temperature as well as the deageing itself on the dielectric properties. Poling after deageing showed a little enhancement on the electromechanical coupling factor. It is also worth to mention that, still the densification of the ceramics is the main ruler of the improvement of the piezoelectric properties. 111

124 Chapter (6) Internal Bias E i (kv/cm) A B As 15 o C After RT Sintering Temperature ( o C) Figure : Internal bias after electrical (A) and structural (B) depinching. Polarization (µc/cm 2 ) 25 2 Sample [2] Electric Field (kv/cm) Polarization (µc/cm 2 ) 25 2 Sample [3] Electric Field (kv/cm) Polarization (µc/cm 2 ) 25 2 Sample [5] Electric Field (kv/cm) Polarization (µc/cm 2 ) 25 2 Sample [7] Electric Field (kv/cm) Figure 6.22: Hysteresis loops at room temperature for samples high poled (3 kv/cm, 9 o C) after electrical depinching (5 kv/cm, 15 o C). However, the structural deageing was found to be very effective for improving the dielectric as well as the piezoelectric properties. As shown in Figures 6.24A 24C and summarized in Table 6.2, the dielectric constant ε r and tan δ are greatly enhanced after quenching. 112

125 Poling of Hard PZT Table 6.2: Dielectric and piezoelectric parameters at different conditions of poling. Poled after electrical Nonpoled Samples Aged nonpoled Directly poled deageing after quenching Poled after quenching ε r tan δ ε r tan δ k p ε r tan δ k p ε r tan δ ε r tan δ k p " # $ % & ' ( By poling the samples after deageing, the electromechanical coupling factor becomes significantly higher than that of the aged state. Generally, we note that the thermally quenched hard ceramic samples behave qualitatively the same as soft materials. That transition can be interpreted as disordering of pinning centres resulting in randomization of the energy profile for domain wall motion. This behavior can be attributed to the effect of oxygen vacancies disordering caused by thermal quenching. Owing to fast cooling, some diffusion processes are hindered, hence, the quenched ceramics will be in metastable state. Planar Coupling Factor, k p A.3 7 o C 9 o C.2 1 o C Dielectric Constant B 7 o C 9 o C 1 o C tan δ Sintering Temperature ( o C) Figure 6.23: Dielectric and piezoelectric parameters after normal poling (before deageing). Planar Coupling Factor, k p (A) Dielectric Constant.35 1 o C after Ageing Quenching (B) Aged Nonpoled Quenched Nonpoled Poled after Quenching tan δ Sintering Temperature ( o C) C 7 o C.4 9 o C 1 o C (C) Aged Nonpoled Quenched Nonpoled Poled after Quenching Figure 6.24: Dielectric and piezoelectric parameters after structural deageing compared with aged state. 113

126 Chapter (6) Thermally Stimulated Discharge Current (TSDC) Figure 6.25A shows the thermally stimulated discharging current (TSDC) curve for sample (. After electrical depinning, the discharge current curves showed a certain scatter with density and/or poling conditions. Therefore the curve in Figure 6.25A is taken as a representative curve for the other samples. At first glance one cannot attribute the TSDC to the pyroelectric activity since the total charge released by thermal stimulation is about 8 µc/cm 2. Since the frozen-in defect-dipoles can be released by thermal activation and consequently prevail itself as anomalous TSDC. As the dipolar defects orient to certain favorable directions which are not necessarily in the spontaneous polarization direction, the most favorite orientation or favorite position of the oxygen vacancy depends on the off-centre ferroelectric ions (Ti or Zr). The energetically favored site also changes by increasing the temperature which leads to release of the defects charges 17. After quenching (structural depinning), the defect dipoles are randomly distributed. However, they can be oriented by the externally applied electric field at high temperature. After ageing the defect dipoles are already oriented by the ferroelectric polarization and they have no depolarization effect. As a consequence, the TSDC curve, Figure 6.25B, has a totally different profile. It shows only one single peak at the typical Curie transition temperature (~ 315 o C). Worthwhile to mention is that the total charge released in both cases is almost constant (~ 8 µc/cm 2 ) as such confirming our postulate of the unification of direction of ferroelectric polarization and defect dipoles in poled samples after deageing. Similar behavior was observed previously 18, Pyroelectric Current (na/cm 2 ) 1 (A) T C Temperature ( o C) 175 (B) T C Temperature ( o C) Figure 6.25: Thermal stimulated discharge current (TSDC) for samples poled after (A) electrical depinning and (B) structural depinning. 114

127 Poling of Hard PZT 6.4. DISCUSSION Model of Ageing Phenomena Ageing phenomena are observed in all ferroelectrics. It usually manifests itself as an appearance of a double P E hysteresis loop in nonpoled samples, a decrease of the dielectric constant at a small ac electric field and piezoelectric constant at small mechanical stress. The theory of ferroelectricity cannot directly explain ageing behavior. It is natural to speculate that ferroelectric ageing is caused by a certain relaxation process. Generally, it is agreed that ageing can be phenomenologically attributed to a gradual stabilization of domain pattern by D 3+ V O defect dipoles. This argument is widely verified, theoretically 1,17,21-24 and experimentally 49,46. However, Dimos and co-workers have shown that, pinning can be of electronic origin due to locking of domains by electronic charge trapping centres. The central controversy is, still, whether the domain stabilization stems from a domain effect or a volume effect or whether it is even ionic or electronic. For the domain effect, it is believed that defects migrate to the domain walls during ageing and consequently pin the domain walls. For the volume effect, it is considered that the defect dipoles align along the spontaneous-polarization P S orientation over the whole domain during ageing and thus make domain switching difficult. In the domain effect, however, the existence of domain walls is a necessary condition for ageing. However, in the volume effect model, ageing is expected to exist even in a single-domain sample. On this assumption, Zhang and Ren 5 prepared a single domain ferroelectric crystal and showed that ageing cannot be explained by domain effect since there is no domain wall structure. They found out that the volume effect rather than domain-wallpinning is the governing mechanism for ferroelectric hysteresis loop pinning. As it is impossible to distinguish between different mechanisms which cause the stabilization from purely electrical measurements 28,29, we will also assume that the volume effect can explain our results as a general mechanism governing the single- and multidomain wall stabilization. The model of ageing phenomena as initially set by Carl and Härdlt 23 and adopted by many others can be applied to our case as follows. The model is based on the fact that the ferroelectric crystal contains V O vacancies and Fe 3+ acceptor impurity ions occupying (Ti 4+ or Zr 4+ ) sites in PZT perovskite structure. Now we consider the principle of symmetry of the statistical distribution 48 of V O around a defect ion Fe 3+. In the cubic paraelectric phase Figure 6.26A, the defect probabilities P v should be the same (P v1 = P v2 = P v3 = P v4 ) of finding an V O in the neighboring sites (1, 2, 3, 4) of the defect Fe 3+ because these sites are equivalent to the defect Fe 3+ at site. Thus the symmetry of the defect probability about Fe 3+ is cubic, following the cubic crystal symmetry of the paraelectric phase. 115

128 Chapter (6) C1 C2 O 2 5% Probable O 2 or Vö Vö Zr 4+ or Ti 4+ Fe 3+ [A]@T >T C P S =+ P D =* [B]@T >T C, t= P S =! P D =* [C]@T <T C, t= P S =! P D =" [D]@T <T C, t= P S =! P D =* [E]@T <T C, t= P S =! P D =" A 3 Figure 6.26: Symmetryconforming property of point defects. Here, a typical ABO 3 (PZT) perovskite structure is considered, which contains an O 2- vacancy and an acceptor impurity Fe 3+ occupying the (Zr 4+,Ti 4+ ) site. The probability P vi refers to the conditional probability of finding an O 2- vacancy and O 2- ion at the site i (i =1,2,3,4), respectively, when an impurity ion Fe 3+ occupies site. State (A): represents a nonpolar/centrosymmetric cubic paraelectric state (T >T C ). (B): Fresh (unaged) polar tetragonal ferroelectric state (T <T C ) (thick line represents the domain wall). (C): Polar tetragonal ferroelectric state after O 2- vacancy diffusion (i.e. ageing). (C1 C1): A stable multidomain structure of the tetragonal ferroelectric crystal after ageing (diffusion) in the ferroelectric state, in which defect symmetry follows crystal symmetry in every domain. (C2 C2) Unstable state after domain switching by electric field E. Double hysteresis loop (P-E curve) during reversible domain switching between (C1) and (C2). Stable states are those where the defect symmetry matches the crystal symmetry; unstable states are those where they do not match. (D) Poled configuration in the fresh state before ageing dominates, in which the defect symmetry matches the crystal symmetry. Even upon the inevitable diffusion motion of the oxygen vacancies (E), the poled configuration (i.e. the stable state) is still preserved. C E B D 116

129 Poling of Hard PZT In the ferroelectric phase Figure 6.26B, the polar tetragonal symmetry makes the sites (1, 2, 3, and 4) no longer equivalent with respect to the defect Fe 3+ at site (site 1 and 2 are equivalent but 3 and 4 are not). That equivalent sites have the same defect probabilityv O and non-equivalent sites have unequal defect probability; thus it follows that the defect probabilities P v should bepv 1 = Pv 2 Pv 3 Pv 4, following the polar tetragonal symmetry of the lattice when in equilibrium. Therefore, the defect configurations have cubic symmetry when in equilibrium which matches the cubic crystal symmetry. Consider a macroscopic crystal of equilibrium cubic paraelectric phase, Figure 6.26A, which has a cubic symmetry of point-defect distribution (defect symmetry in short) everywhere according to the symmetry-conforming property 48. Upon cooling, the crystal undergoes a paraelectric ferroelectric (PE FE) phase transition transforming into a multidomain tetragonal ferroelectric phase, Figure 6.26B. A spontaneous polarization P S will be established along the off-centered Fe 3+ ions due to the relative displacement of positive and negative ions. During this abrupt lowering of crystal symmetry, the defects, however, cannot migrate since the PE FE phase transition is diffusionless. As a result, the polar tetragonal ferroelectric phase inherits a cubic defect symmetry after the transition, as shown in Figure 6.26B. However, the unaged or fresh ferroelectric phase, Figure 6.26B, is not a stable state in reality, because the defect symmetry (cubic) does not match the crystal symmetry (tetragonal) 17. If defect/vacancy re-distribution occurs 17, the cubic defect symmetry will be changed into a polar tetragonal one, Figure 6.26C. Such a process involves a short range migration of ions/vacancies. Site 3 and site 4 are no longer equivalent because site 3 is closer to the ion on site. Thus the site closer to Fe 3+ ion should have a higher defect probability due to the Coulomb attractive force between the effectively negatively Fe 3+ and effectively positively V O vacancy ions. This process requires some time (ageing) to complete. This is the microscopic mechanism of ageing (domain wall pinning) in ferroelectrics. After ageing, the defect symmetry in each domain follows the polar tetragonal crystal symmetry. The non-centric distribution of charged defects Fe 3+ and V exhibits a defect-dipole moment P D following the spontaneous polarization P S direction 45 of the residing domain, as shown in Figure 6.26C1. Now every domain is in its stable state. When such stable domains are switched by an electric field, domain switching occurs abruptly (without diffusion) with spontaneous polarization following the field direction, Figure 6.26C2. However, the defect symmetry and defect-dipole moment cannot be rotated in such a diffusionless process. Therefore, the unswitchable defect symmetry and the associated defect-dipole moment provides a restoring force or a reverse internal field favoring to reverse the domain switching when the electric field is removed, Figure 6.26C1 C2. As a consequence, the original domain pattern is restored so that the defect symmetry and dipole moment follow those of the crystal O 117

130 Chapter (6) symmetry in every domain. Consequently a pinched hysteresis loop in the P-E relation is observed as well as a shift in the E-axis due to the existence of P D Model of Thermal Deageing As we discussed earlier, in the domain wall pinning model, at a temperature well above T C (in the paraelectric phase) the oxygen vacancies V O are randomly distributed and all the oxygen sites have the same probability to be occupied by an oxygen vacancy. Thus the oxygen vacancy has a cubic defect symmetry 49 following the cubic symmetry of the host lattice. Upon quenching and immediately after the PE FE phase transition, the oxygen vacancies freeze in position and do not migrate to their favorite positions (i.e. close to the Fe 3+ ions). The oxygen vacancies rather adopt the cubic symmetry as they did in the paraelectric phase because the transition is diffusionless and involves no exchange of the ions. In this state the defect has zero dipole moment. This image is analogous to the fresh state in Figure 6.26B. At this stage we have two directions to go. Either to let the material to age and follows the same scenario we mentioned before or we apply very strong poling conditions. In this way the domain will be oriented easily without the hindrance of the defects since the defect polarization is still zero, Figure 6.26D. Even after removing the field the defects have no influence so-far and remain in the cubic symmetry. Consequently, there is no restoring force to switch the oriented domains back. However, in time (by ageing) the oxygen vacancies will migrate by diffusion following the host crystal symmetry and the spontaneous polarization P S. However, such diffusion redistribution of defects will not affect the domain orientation because defect dipoles are following the ferroelectric polarization direction. This new domain stability can be sketched as in Figure 6.26E. Figure 6.22 also shows a good verification of this assumption. The anomalous hysteresis loops show a unipolar polarization orientation for poled samples after deageing. Deageing directly followed by poling results in irreversible changes in the defect distributions and domain stability 39,4. This simple explanation also enables us to understand the large enhancement in the piezoelectric coupling factor k p after the same conditions of poling right after quenching, since the defects have no chance to age and thereafter to dominate Interpretation of the Pyroelectric Curves Before Deageing In the aged state, the sample is still not fully polarized as can be recognized from the value of the k p. That means that the majority of the domains are still pinned. The oriented domains are in fact in an unstable state because the defect polarization P D does not follow the spontaneous polarization P S of oriented domains. That means that the defect dipole moment P D can memorize the original crystal domain patterns after ageing, and do not change during subsequent domain switching 49. However, under thermal activation defects can migrate by diffusion in order to match the crystal 118

131 Poling of Hard PZT symmetry. This motion of defects, primarily the oxygen vacancies V O, produces a current which is opposite to the intrinsic thermal depolarization current which has also just started to contribute. At this point we have two types of current of different nature and direction. The intrinsic pyrocurrent which is rising in the positive direction and the extrinsic diffusion current of the oxygen vacancies which rising in the negative direction and counteracting the pyrocurrent to peak at lower temperature (25 o C) than the typical Curie temperature (33 o C), Figure 6.25A. By approaching the Curie transition temperature, the pyrocurrent completely diminishes whereas the diffusion motion of the defects may continue in the negative direction. The reason for that is the material undergoes a FE PE phase transition which is relatively fast in nature while the defects move by diffusion. Thus, it takes a relatively long time to adapt with the new (cubic) crystal symmetry. Consequently its current vanishes at relatively higher temperature (4 o C) After Deageing; Quenching Effect The thermally stimulated discharge current (TSDC) curve, as shown in Figure 6.25B, of a poled sample after quenching can be explained as follows. After quenching, as we illustrated previously, strong poling for a sample in the fresh quenched state can easily and permanently orient the domain. Although the defect-dipoles still have a cubic symmetry, they have no influence on the oriented domain walls even after ageing by diffusion to follow the crystal symmetry and the long-range spontaneous polarization P S. By short circuiting the sample with an electrometer and applying a regular heating, we may expect two types of discharge current: The defect dipole discharge current and the intrinsic depolarization current. This indeed was experimentally observed and we found an extremely high and broad TSDC curve peaking at the typical Curie transition temperature (~ 33 o C). At this point we have two important observations. Firstly, the single peak in the TSDC profile enhances the idea that the defect dipoles are not misaligned with the intrinsic ferroelectric polarization and ageing is not going towards depolarization or backswitching effects again. This is also consistent with the normal hysteresis loop at room temperature after quenching. Secondly, the extremely high TSDC can be understood now as a combination of two discharging currents and accordingly the real spontaneous polarization P S can not be determined by this measurement Dielectric Loss Factor Data In ferroelectric materials, all properties are closely linked. On the one hand, the ferroelectric P E hysteresis as a defining property of ferroelectric materials can give valuable information on different physical processes that take place in ferroelectric materials, e.g. domain-wall pinning, defect ordering, and nature of defects 1,19. On the other hand, by having a mechanism explaining the double hysteresis loops, it can 119

132 Chapter (6) accordingly be easier using it to interpret the other dielectric, piezoelectric and pyroelectric properties Aged Samples Herbiet et al. 53 stated that the extrinsic contribution is the main source of the dielectric loss because the domain wall motion can induce mechanical friction. The extrinsic contribution can be as much as 7 % of the total loss at room temperature 53. Before poling the domain walls are pinned by defects leading to restricted movement of the domain walls (assuming the movement of the domain wall is the main contribution to the dielectric loss). Consequently, a low dielectric loss is expected 22,56. Poling of hard ferroelectric does not greatly reduce the domain wall number (see Figure 6.26C1 C2). This can be evidenced from the double hysteresis loop characteristic. However, there is a significant movement of domain walls due to poling. As a result of pinning, which acts as an elastic restoring force, the domain walls in time (i.e. ageing) return back. Accordingly, we can understand why the dielectric loss increases after poling contrary to the common practice. The same thought is also applicable for poled ceramics after electrical deageing. Because the material in this case is more polarized than before electrical deageing, the increase in tan δ is less, Table 6.3. Table 6.3: Dielectric loss (tan δ) status. Before Poling After Poling Aged #! Electrically deaged #! Thermally deaged! # Deaged Samples Now we consider the case after structural deageing (thermal quenching). After quenching, the material undergoes a sudden ferroelectric (tetragonal) to paraelectric (cubic) phase transition. However, the oxygen vacancies preserve a cubic symmetry 48-5 so that no defect-dipoles are developed. Consequently, there is no domain wall pinning and no acting restoring force. Based on this assumption, the domain wall can freely move and this implies a high dielectric loss. After poling and before ageing (i.e. before defect-dipoles develop) the material can be relatively easy poled leading to a reduction in the number of domain walls, which are now pinned by the global ferroelectric polarization. Consequently, after poling the material shows domain wall stability and lower dielectric loss. Moreover a remarkable increase is observed in the electromechanical coupling factor k p. This assumption described above is again based on the major extrinsic contribution of the domain wall motion to the dielectric properties. 12

133 Poling of Hard PZT CONCLUSIONS In summary, poling of hard ferroelectric ceramics cannot be accomplished before domain wall depinning. The restriction of domain wall mobility can be released either by applying a high temperature and high periodic electric field before normal poling procedure. Another more efficient way for domain wall depinning is quenching effect where after the material showed remarkably improved piezoelectric properties. Pyroelectric current, in despite of its useless indication to the degree of polarization in the present case can nevertheless be used as monitoring tool to indicate to the degree of domain pinning by defects. 121

134 Chapter (6) REFERENCES D. Damjanovic, The Science of Hysteresis, Volume 3, 3 I. Mayergoyz and G. Bertotti (Eds.), Elsevier, 25. K. Okazaki, Jpn. J. Appl. Phys., 32, 4241, W. L. Zhong, Y.G. Wang, S. B. Yue and P.L. Zhang, Solid State Comm., 9, 383, T. M. Kamel, F. X. N. M. Kools and G. de With, J. Eur. Ceram. Soc., 27, 2471, 26. I. Ueda and S. Ikegami, Jpn. J. Appl. Phys., 7, 236, S. Cai, Y. Xu, C. E. Millar, L. Pedersen and O. Sorensen, IEEE, 84, P. Bryant and T Palmer, J. Aust. Ceram. Soc., 36, 153, 2. H. Nagata and T. Takenaka, IEEE, 45, 21. M. H. Lente and J. A. Eiras, J. Appl. Phys., 89, 593, 21. A. Yamada, Y. Chung, M. Takahashi and T. Ogawa, Jpn. J. Appl. Phys., 35, 5232, T. Ogawa, K. Nakamura, Jpn. J. Appl. Phys., 37, 5241, T. Ogawa, Ferroelectrics, 24, 75, 2. T. Ogawa, Jpn. J. Appl. Phys., 39, 5538, 2. T. Ogawa, Jpn. J. Appl. Phys., 4, 563, 21. T. Ogawa, Ferroelectrics, 273, 371, 22. D. C. Lupascu, U. Rabe, Phys. Rev. Lett. 89, 18761, 22. G. Arlt and H. Neumann, Ferroelectrics, 87, 19, B. Jaffe, W. R. Cook and H. Jaffe, Piezoelectric Ceramics. Academic Press, New York, M.E. Lines, A.M. Glass, Principles and Applications of Ferroelectrics and Related Materials, Oxford, Clarendon, M. Morosov, Softening and Hardening Transitions in Ferroelectric Pb(Zr,Ti)O 3 Ceramics PhD Thesis, EPFL, Lausanne, Switzerland, 25. P. V. Lambeck and G. H. Jonker, J. Phys. Chem., 47, 453, W. A. Schulze and K. Ogino, Ferroelectrics, 87, 361, K. Carl and K. Härdtl, Ferroelectrics, 17, 473, U. Robels and G. Arlt, J. Appl. Phys., 73, 3454, U. Robels, J. H. Calderwood and G. Arlt, J. Appl. Phys., 77, 42, D. Dimos, W. L. Warren and M. B. Sinclair, B. A. Tuttle and R. W. Schwartz, J. Appl. Phys. 76, 435, W. L. Warren, D. Dimos, B. A. Tuttle, R. D. Nasby and G. E. Pike, Appl. Phys. Lett. 65, 118, D. Dimos, W. L. Warren and H. N. Al-Shareef, Thin film ferroelectric materials and devices, Edited by R. Ramesh, Kluwer Academic Publishers, p.p. 199, W. L. Warren, B. A. Tuttle and D. Dimos, Appl. Phys. Lett., 67, 1426, K. Uchino Ferroelectric Devices Marcel Dekker, 2. M. Hammer and M. J. Hoffmann, J. Am. Ceram. Soc., 81, 3277, A. K. Tagantsev, I. Stolichnov, E. L. Colla and N. Setter, J. Appl. Phys., 9, 1387, 21. S. B. Lang, Source Book of Pyroelectricity, Gordon and Breach Science Publishers, London, B. Li, Z. Zhu, G. Li, Q. Yin and A. Ding., Jpn J. Appl. Phys., 43, 1458, 24. O. Parkash, L. Panedy, M. K. Sharma and D. Kumar, J. Mat. Sci., 24, 455, L. Wu, C.-C. Wei, T.-S. Wu and H.-C. Liu, J. Phys., 16, 2813, P. Singh, D. Kumar and O. Parkash, J. Appl. Phys. 97, 7413, 25. B. Li, G. Li, Q. Yin, Z. Zhu, A. Ding and W. Cao, J. Phys., 38, 117, 25. Q. Tan, Z. Xu and D. Viehland, J. Mater. Res., 14, 465, M. Kohli, P. Muralt and N. Setter, Appl. Phys. Lett., 72, 3217, H. L. Blood, S. Levine and N. H. Roberts, J. Appl. Phys., 27, 66, J. W. Northrip, J. Appl. Phys., 31, 2293, 196. K. Okazaki, Jpn. J. Appl. Phys., 32, 4241, S. K. Pandey, O. P. Thakur, A. Kumar and C. Prakash, J. Appl. Phys., 1, 1414, 26. R. Lohkämper, H. Neumann and G. Arlt, J. Appl. Phys. 68, 422, 199. T.J. Yang, V. Gopalan, P.J. Swart and U. Mohideen, Phys. Rev. Lett,. 82, 416, C-.C Chou, C-.S Hou and T-.H. Yeh., J. Eur. Ceram. Soc., 25, 255, 25. X. Ren, Nat. Mat. 3, 91, 24. L. Zhang and X. Ren Phys. Rev B71, 17418,

135 Poling of Hard PZT L. Zhang and X. Ren, Phys. Rev. B73 73, 94121, 26. W. Liu, W. Chen, L. Yang, L. Zhang, Y. Wang, C. Zhou, S. Li and X. Ren, Appl. Phys. Lett., 89, 17298, 26. C. A. Randall, N. Kim, J.-P. Kucera, W. Cao and T. R. Shrout, J. Am. Ceram. Soc., 81, 667, R. Herbiet, U. Robels, H. Dederichs and G. Arlt, Ferroelectrics, 98, 17, K. Uchino, E. Sadanaga and T. Hirose, J. Am. Ceram. Soc., 72, 1555, J. F. Meng, R. S. Katiyar and G T. Zou, J. Phys. Chem. Solids, 59, 1161, S.-J. Yoon, A. Joshi and K. Uchino, J. Am. Ceram. Soc., 8, 135,

136

137 CHAPTER 7 EPILOGUE ABSTRACT In this chapter, an evaluation of the main achievements of this thesis research is set out.

138 Chapter (7) Epilogue 7.1. WHAT WAS PLANNED? Poling is one of the crucial intermediate stages of the manufacturing process of the piezomaterials. Based on this fact, the main object of the thesis was to understand the polarization of the PXE materials and consequently the appropriate poling conditions can be adjusted to yield the optimum dielectric and piezoelectric performance WHAT WAS ACHIEVED? In chapter 3, we showed that the poling state can be determined phenomenologically by different techniques. It appears that the pyroelectric activity can be used as a good tool to measure the degree of polarization. The polarization current results can be described by the switching polarization theory of Merz 1. It was found that, contrary to common practice, for soft materials, good dielectric and ferroelectric properties can be reached at room temperature at a poling field slightly larger than the coercive field. The pyroelectric activity of the PXE52 material makes it a good pyroelectric candidate for technological purposes. Its pyroelectric coefficient at room temperature is 7. µc/cm 2. o C and at the Curie transition temperature is.5 7. µc/cm 2. o C indicating a potential applicability of these materials compared to other pyroelectrics 2-5. It has been also shown that, increasing the grain size maximizes the dielectric and piezoelectric properties as compared to properties of grain size often used in practice. Thus, an optimum choice of a combination of the poling field and the grain size leads to maximum enhanced properties. As a consequence of the softness of the PXE52, a double switching peak has been observed during reversed poling instead of the commonly observed single one. As we discussed in Chapter 4, this double peak in the switching curve is attributed to the residual stresses developed during forward poling that transforms the switching mode from a single coercive field to double coercive field. The residual stress makes that switching via two successive 9 o rotations energetically easier than direct 18 o domain switching. The grain size variation was found to have a large and important role on the observation of double switching phenomenon. While small grains showed a small single switching due to hindrance of the 9 o domain motion, large grains, however, showed double switching. The pyroelectric coefficient curves showed a consistent trend with the switching curves. While pyroelectric charge after poling is released via one depolarization step, the pyroelectric charge after switching is released via two depolarization steps. This indicates that the thermal depolarization follows the same polarization mechanism. 126

139 Epilogue High insulation resistance is desirable for many technological applications of PZT piezoceramics. Accordingly, resistance degradation is one of the issues encountered during the functioning of piezoceramics. The electric conduction of PXE52 has been studied in Chapter 5. Due to the highly pronounced pyroelectric effect of this material, it was not possible to recognize the conduction current near room temperature, since it is entirely hidden by the pyroelectric current. However, the steady state resistivity was found to have a typical high value of 3.6 x 1 12 Ω.cm at room temperature. The high temperature resistivity was found to be two-orders of magnitude lower than the RT value. However, no resistance degradation has been observed in PXE52. In Chapter 6, the poling of hard PZT (PXE43) was studied. It was found that, poling of hard ferroelectric ceramics cannot be accomplished before domain wall depinning. The restriction of domain wall mobility can be released either by applying a high temperature and high periodic electric field before the normal poling procedure. In this study we explored another more efficient way for domain wall depinning by the thermal quenching. This method resulted in remarkably improved ferroelectric and piezoelectric properties FUTURE RESEARCH It would help a great deal if an extensive domain structure investigation will be made for PXE materials to see the real effect of poling on the microstructure. Transmission electron microscopy (TEM) and atomic force microscopy (AFM) are be the candidate techniques for the domain structure investigation. The field-induced-phase-transition and domain-type-switching are considered to be interesting subjects of research. The poling and switching behavior and the subsequent transformations should be studied by in situ investigating tools (e.g. in situ XRD or AFM). The potential application of pyroelectric behavior of PXE52 is a promising point of research for the scientific and the technological prospects. Further investigation of the quenching effect on the rejuvenation of PXE43 is an exciting subject for research in the light of industrial applications. 127

140 Chapter (7) REFERENCES W. J. Merz, Phys. Rev., 95, 69, T. Takenaka, A. S. Bhalla and L. E. Cross, J. Am. Ceram. Soc., 72, 116, S. Lee, K. Lee and G. Park, J. Korean Phys. Soc., 3, 261, K. V. R. Prasad and K. B. R. Varma, J. Phys., 24, 1858, K. B. R. Varma and A. K Raychaudhuri, J. Phys., 22, 89,

141 POLING AND SWITCHING OF PZT CERAMICS field and grain size effects The polarization is the true measure of the degree of ferroelectricity. Generally, ceramic materials with a high value of remanent polarization show usable piezoelectric effects. Accordingly, the poling of piezoceramics after manufacturing is an important process. In particular, lead zirconate titanate (PZT) is a well-known ferroelectric and piezoelectric material, therefore, poling and switching of PZT ceramics is the main focus of the present thesis. Depending on the type of doping, PZT can be electrically soft or hard. In this thesis we are studying the ferroelectric, pyroelectric and piezoelectric properties and aftereffects in soft and hard PZT. The properties can be influenced by various parameters either extrinsic such as applied electric field or intrinsic such as grain size. Thus the ultimate aim of the study is to gain insight on mechanism of the poling and switching processes in order to reach and understand the optimum functional behavior of these materials. We have shown that the poling state can determined phenomenologically by different techniques. It was found that, contrary to common practice, for soft materials, good dielectric and ferroelectric properties can be reached at room temperature at a poling field slightly larger than the coercive field. The pyroelectric activity of the soft PZT makes it a good pyroelectric candidate for technological purposes. It has been also shown that, increasing the grain size maximizes the dielectric and piezoelectric properties as compared to properties of grain size often used in practice. Thus, an optimum choice of a combination of the poling field and the grain size leads to maximum enhanced properties. Switching of pre-poled soft PZT has been extensively studied as a function of electric field and grain size. An unusual double switching profile was observed in soft PZT instead of the common single one. This double peak in the switching curve is attributed to the residual stresses developed during forward poling that transforms the switching mode from a single coercive field to double coercive field. The residual stress makes switching via two successive 9 o rotations are energetically easier than direct 18 o domain switching. Double switching was found to be a function of applied electric field and grains size. These observations enabled us to access a new channel for understanding the essence of the polarization mechanism of PZT ceramics. Due to the highly pronounced pyroelectric effect of this material, it was not possible to recognize the conduction current at higher temperatures. However, the steady state resistivity was found to have a typical high value at room temperature. Poling of hard PZT was extensively studied. It was found that, poling of hard ferroelectric ceramics cannot be accomplished before defeating the pinning centres that counteracting the switching process. In this study we presented different methods to overcome the pinning centres and consequently an efficient poling was enabled that eventually led to considerably improvised ferroelectric and piezoelectric properties. 129

142

143 APPENDIX A SAMPLING AND HOMOGENEITY A.1. INTRODUCTION During the sintering process of PZT lead oxide PbO evaporates and diffuses into the grain boundaries and triple points as an amorphous secondary phase 1. The PbO secondary phase leads to inhomogeneous microstructure, irreproducible and eventually degraded electrical properties 2. Accordingly, prior to the main experimental work in this thesis, a preliminary set of experiments was performed in order to ascertain the homogeneity and reproducibility of the material s properties to be eventually worked out. A.2. EXPERIMENTAL RESULTS A.2.1. Block of PXE52 For verifying the microstructural homogeneity and consequently the properties reproducibility, a large block of soft PZT (PXE52) was prepared. Due to the evaporation of PbO, it can be naturally expected that, specimens from the outer surface of the block more likely lose lead than specimen from inner core do. Based on that, the PXE52 block was diced into a large amount of specimens, Figure 1, which can be labeled according to their original positions in the block. Thereafter, samples from rather different positions from the block are selected for the microstructural and electrical investigations.

144 Appendix (A) 25 mm 45 mm 84 plates 74 mm 28 Strips L R P84 P83 P82... P4 P3 P2 P1 Example P1-15R Figure A.1: A block of PZT (PXE52) diced into 84 plates (P 1 84 ). Each plate is cut longitudinally into two bands (L and R). Each band is cut into 28 stripes. Each stripe has a label (e.g., P1-15R) referring to its position in the block. A.2.2. Thermal analysis (DTA and TGA) The grain boundary glassy secondary phase has a lower melting point than the surrounding ceramic phase. Although this phase has a low fraction volume, its presence can be detected by means of differential thermal analysis DTA and/or thermogravimetric analysis TGA. Figure 2 shows both DTA and TGA traces for a PXE52 sample. The DTA curve shows melting transition temperature at 85 o C and the TGA thermogram monitors a slight weight loss qualitatively indicating the lead oxide evaporation. T (a.u.) PXE52 PbO Melting ( 85 o C) -.5 Flowing Gas: Nitrogen 1 ml/min Temperature o C Figure A.2: DTA and TGA traces for a PXE52 sample Weight Loss (mg) 132

145 Sampling and Homogeneity A.2.3. Grain Size Distribution A grain size distribution analysis was made for samples from rather different locations in the block to see whether there is a change in the grain size for the different locations in the block. The samples were polished, etched and examined by optical microscopy. Figure A.3 shows the micrographs of the samples. Each sample was labeled referring to its original location in the block. Table A.1 summarizes the grain size distribution data for samples from different positions in the block. As can be clearly seen, there is no change in the grain size as a function of the specimen location. P6-14L P47-1L P44-14R P6-1R P84-1R P83-14R Figure A.3 A.3: Micrographs of samples selected from different locations of the block. The sample label refers to the sample location, as indicated in Figure A

146 Appendix (A) Table A.1: A Grain size distribution analysis. Number of grains analyzed is 1. Sample Grain size (mean linear intercept) (µm) Mean St. Dev. (µm) P6-14L P47-1L P44-14R P6-1R P84-1R P83-14R A.2.4. Electrical Properties Distribution Figure A.4 shows an extensive scan of the electrical properties (dielectric constant, dielectric loss, and electric resistivity) throughout the block. The properties were scanned for all plates focusing on stripes on the top and in the middle of the block. It was found that the electrical properties are varying within a limited margin. The dielectric constant, dielectric loss, and electric resistivity fluctuate between 27-3, , and x1-11 Ω.m, respectively. Dielectric Constant L top stripes middle stripes Distance from the front plate (mm) A R Resistivity (Ω.m) x top stripes middle stripes L tan δ top stripes middle stripes R L top stripes middle stripes R top stripes middle stripes Distance from the front plate (mm) top stripes middle stripes Distance from the front plate (mm) C Figure A.4.4: Scanning of (A) the dielectric constant, (B) the dielectric loss, and (C) the electrical resistivity through the block. B 134

147 Sampling and Homogeneity A.3. DISCUSSION AND CONCLUSIONS Differential thermal analysis data clearly monitored the melting of the (lead oxide) second phase at 85 o C. However, the weight loss due to lead evaporation was quite low. This is probably the reason why there is no effect of lead evaporation on the grain size distribution. We have found a fairly constant grain size throughout the block. The electrical properties showed also limited fluctuation and changed within the indicated range. To conclude, the PXE52 ceramic was found to be homogenous and its properties were reproducible within satisfactory limits regarding it a good material for the purpose of this thesis. 135

148 Appendix (A) REFERENCES 1 B. Jaffe, W. R. Cooke, and H. Jaffe, Piezoelectric Ceramics, Academic Press, New York, H. Fan, G.-T. Park, J.-J. Choi, J.-R. Ryu and H.-E. Kim, J. Mater, Res., 17, 18,

149 APPENDIX B PIEZOELECTRIC CALCULATIONS The Equations Used for Calculating the Piezoelectric Parameters: Measurements of resonant frequency f r and the antiresonant frequenc f a were used to evaluate the piezoelectric parameters, Chapter 2. An important parameter of a piezoelectric material is the effective electromechanical coupling factor k eff which is defined as 1 : mechanical energy converted to electrical energy k 2 eff = (B.1) input mechanical energy or electrical energy converted to mechanical energy k 2 eff = (B.2) input electrical energy The coupling factor k eff is related to the resonance and antiresonance frequencies by the following relation 3 : fa fr eff 2 fa k = (B.3) The thickness extensional electromechanical coupling factor k 33 is related to the resonance and antiresonance frequencies as 1,2 : k 2 33 π f = 2 f r a π fa f tan 2 fa r (B.4) The elastic compliance is the ration of a material s change in dimension (strain) in relation to an externally applied load (stress) 3. For a piezoelectric material the elastic compliance coefficients s 33 and s 11 can be calculated by the following relations

150 Appendix (B) 1 s 2 2 = 4ρ fa l D (B.5) 33 s s E 33 = (B.6) 1 k 1 s s D = 4ρ fr w E (B.7) 11 2 ( ) D E 11 s11 1 k31 = (B.8) where ρ is the density of the material, l and w are the piezo plate thickness and width, respectively. The k 31 is the length extensional coupling factor. The superscript D and E signify that the sample is under constant dielectric displacement and constant electric field, respectively. One of the fundamental constants of piezoelectric ceramics is the planar coupling coefficient k p which is defined for thin discs and can be calculated as 1,2 2 k p f a fr = f Jo, J1, σ (B.9) 2 1 kp fa where J o and J 1 are Bessel functions and σ is Poisson s ratio. The planar coupling factor can be approximated 3 : f f k (B.1) p 2 2 a r 2 fa The length extensional coupling factor k 31 can be calculated as k σ 2 = k p (B.11) 2 Possion s ratio can be determined using the following equation 2 v f r. ω = [ ( σ.3) ] (B.12) 2 π 1 σ 1 where v = is the velocity of compressional wave in a disc normal to the poling E ρ s11 direction which in the same time equals to 2f r. ω, where ω is the diameter of the disc. 138

151 Piezoelectric Calculations The piezoelectric charge constant d ij can be calculated using the following equations X E d33 k33 εo ε 33 s33 = (B.12) X E d31 k31 εo ε33 s11 = (B.12) X where ε 33 is the dielectric permittivity of the material under constant stress X and εo is the space dielectric permittivity. 139

152 Appendix (B) REFERENCES 1 A. J. Moulson, and J. M. Herbert, Electroceramics; Materials, Properties, Applications, John Wiley & Sons LTD, England, B. Jaffe, W. R. Cook, and H. Jaffe, Piezoelectric Ceramics, Academic Press New York, T. L. Jordan and Z. Ounaies, Piezoelectric Ceramics Characterization, National Aeronautics and Space Administration (NASA), ICASE Report No ,

153 ACKNOWLEDGEMENTS The story has not yet come to an end. However, lots of achievements have been accomplished and difficulties have been smoothed out. After four years in the Netherlands, I have just folded a very important chapter in the story. During my journey, since I started to think about my PhD until writing this text, I have met a number of people who have greatly influenced me. As I should mention first, Allah is always the great helper in all I do. I owe huge thanks filled with gratefulness to my mother for her love, sacrifice and suffering of being away from her for several years. I would like to take this opportunity to remember my father, the man whose memory brings always the inspiration of high example. I must admit that I was fortunate to meet my journey-companion at the very beginning of the journey, my wife Naoom. She was always the source of love, inspiration and support. She would have been a senior teacher in her home country if she had continued her own career. Instead, she was always there for me. I appreciate, from the bottom of my heart, her great sacrifice for propping up our life against all sorts of threat. Nour, my lovely daughter, I love you in spite of all kinds of troubles you did at home. Jourine, my little doll, who was not even born while writing the draft version of this text, a huge kiss for you! I was also fortunate to have a great mentor in the most important part in the journey, my promoter Prof. Dr. Bert de With, who has a tremendous influence on my scientific progress. I had the privilege to pursue my PhD research under his supervision. His magic professional critics were always to make things easier and clearer. I would like to thank him for giving me the opportunity to work within his group, for his endless help and support, professional guidance, encouragement and unlimited trust in me. At the end of this work I have not only finished a PhD study but most importantly, my personality became mature in all aspects. I m deeply thankful for his contribution in this change. I m exceedingly grateful to the R&D team at Morgan Electro Ceramics, Dr. Pim Groen, Klaas Prijs, and René Kragt for providing the materials I needed during my work and for their great technical support in the piezoelectric measurements. I would like also to thank Adnan for his helpful collaboration. I would like to express my sincere gratitude to Frans Kools, firstly for recommending me as a PhD student at SMG group. Secondly for the beneficial discussion we have had on the results described in Chapter

154 I m deeply thankful to Imanda, Huub, Anneke, Marco and Sacha for providing me all support I needed during my work. I would like also to express my heartfelt thanks to Niek for his good-natured attitude. I would like to thank Mw. Lutgart van Kollenburg from the Backoffice for her wholehearted support in my official paper work. Thankful appreciation is extended to the great team (coached by Jovita Moerel) from GTD at Eindhoven University of Technology for manufacturing the setup I have used in the experimental work. Sincere appreciation must be given to a number of people I have met at Groningen University, Dr. Beatriz Noheda, Mufti and Umet, first for their hospitality during my stay at Groningen. Secondly, for the great help they offered me to do the ac dielectric measurements in their laboratory at the Material Science Center (MSC) at Groningen University. I m very much thankful to Amir Abdallah for our never-forgettable chat times, my former office mates Sjurt, Rob, Dennis, Bart, and Olavio. My current office mate Willem Jan, thank you for the quiet atmosphere you always kept up. I m also indebted to a number of people who I will never forget: Francesca, Daniela, Przemyslaw, Baris, Okan, Riadh, Marshal, Ming, Di Wu and ChunLing Xu, Kangbo, Welfred, Catarina, Jos, Paul, Josè and the other members of SMG group. No words can express my gratefulness to my sister Rasha for the great effort she has been doing for my little family and me. I would like also to extend my gratefulness to my sister Wesaam as well as Ashraf, Fatima, Hamad, Muhammad, Mashael for their mental support. My mother-in-law will be always remembered. I greatly acknowledge M. Alshourabgy for triggering the first contact between TU/e and me. My dear friends Tamer Gamal, Iman, Manal, thank you all for your mental support. Omar and Wijdan, thank you for your brotherly help. My final words will be to express my willing to move on to complete my journey. I wish I would have the chance to visit the TU/e campus later in the future with my kids to show and tell them about where I spent the most valuable and significant time in my entire life. 142

155 About the Author TALAL MOHAMMAD KAMEL was born in Cairo, Egypt, in November 9, He finished the High School in He studied Physics at Al Azhar University, Cairo. By 1997, he obtained the Bachelor degree (B.Sc. Physics) with distinction. Since 1997 he worked as a Physics lab tutor, meanwhile he continued a postgraduate study in Solid State Physics until He started his Master study in 1998 in Experimental Solid State Physics. He accomplished his Master research in the field of ferroelectricity in Bi-cuprate glasses. In 21 he awarded the M.Sc degree. In late 21 he was promoted as a Lecturer Assistant and continued his research in the field of ferroelectrics. In May 23 he joined SMG group at the Eindhoven University of Technology to start his Ph.D. research in the field of electroceramics. The results of his research are described in this thesis. In July 27, he has accepted an invitation to continue his career as a piezoelectric researcher at IMEC-NL Eindhoven. 143