verticale

Synthesis and characterization of ferroelectric lead zirconate titanate Pb(Zr,Ti)O3 thin films

Sintesi e deposizione, attraverso l’utilizzo della tecnica di spin coating, di film sottili di Titanato Zirconato di Piombo. Questo materiale possiede tra le più alte proprietà piezoelettriche e ferroelettriche ed è largamente utilizzato nell’industria elettronica per produrre sensori, come quelli presenti nei dispositivi airbag o negli accelerometri, attuatori, capacitori, trasduttori e si sta via via allargando il suo utilizzo all’interno dei dispositivi MEMS grazie ai vantaggi che offre come alta sensibilità, efficienza e basso consumo energetico. Le linee guida che sono state seguite per sviluppare un metodo efficiente per la produzione di questo tipo di film sottili sono state la semplificazione dei processi produttivi, partendo dalla sintesi della soluzione sol-gel fino al trattamento termico, e l’utilizzo di precursori chimici meno tossici e pericolosi del comunemente usato 2-Me.

Scarica il PDF Scarica il PDF
Aggiungi ai preferiti Aggiungi ai preferiti


Articoli tecnico scientifici o articoli contenenti case history
Tesi di Laurea, Politecnico di Milano, Anno Accademico 2012-2013

Pubblicato
da Alessia De Giosa
VerticaleSegui aziendaSegui




Settori: 

Parole chiave: 


Estratto del testo
POLITECNICO DI MILANO School of Industrial Engineering Master of Science in Materials Engineering and Nanotechnology Synthesis and characterization of ferroelectric Lead Zirconate Titanate Pb(Zr,Ti)O3 thin films. Supervisor: Ing. Luca Magagnin Master degree thesis of: Co-supervisor: Prof. Walter Navarrini Lorenzo Miglioli Matricola n°: 766272 Academic year: 2012/2013 2 Abstract This thesis deals with the processing and characterization of Lead Zirconate Titanate (PZT) sol-gel films for potential applications in microelectronic industry as actuators, capacitors, transducers, sensors and for their implementation into MEMS devices. The deposition method chosen was the spin-coating. The aim was to reduce the number of steps in the film processing by simplifying the production procedures and to use the less hazardous chemicals as possible to make the PZT thin sol-gel films attractive for the industrial applications. Several trials of non 2-Methoxyethanol based solution deposition routes have been attempted, finding an acetate-based hybrid sol-gel route to provide films of good quality with decreased processing times. First of all, a new route to synthesize stable and homogeneous sol-gel PZT (52/48) precursors solutions has been developed. These solutions have been spin coated on three different substrates: Au/TiW/SiO2/Si, Pt/TiO2/SiO2/Si and ITO coated glass. For each kind of substrate, a proper thermal treatment to obtain dense, uniform and cracks free PZT films that possess a fully perovskite crystallographic structure has been defined. The films morphology has been analyzed using SEM and optical microscope and the crystallographic structure has been studied using grazing incident X-ray diffraction. The influence of substrate texture and properties and of the performed heat treatment on the films crystallization mechanism has been discussed. On each kind of substrate, dense, uniform and crack free PZT films with a fully perovskite crystallographic structure and preferentially oriented towards the (110) plane direction, have been successfully deposited. Some deviations from this behavior occurred and these have been discussed. The electrical features of these films have been analyzed calculating the C-V and the ε-V curves, from which it has been confirmed the ferroelectric behavior of these films. The values of the dielectric constant obtained were 484 for PZT films deposited on gold electrodes, 655 for films deposited on ITO coated glass and 770 for PZT films deposited on platinum. From the C-V curve, an approximate value of the coercitive field (Ec) around 10-15 kV/cm has been also extrapolated, suggesting that these films behave like soft ferroelectrics. 3 Abstract Lo scopo di questa tesi è la sintesi e la deposizione, attraverso l''utilizzo della tecnica di spin coating, di film sottili di Titanato Zirconato di Piombo, meglio conosciuto con la semplice sigla PZT. Questo materiale possiede tra le più alte proprietà piezoelettriche e ferroelettriche ed è largamente utilizzato nell''industria elettronica per produrre sensori, come quelli presenti nei dispositivi airbag o negli accelerometri, attuatori, capacitori, trasduttori e si sta via via allargando il suo utilizzo all''interno dei dispositivi MEMS grazie ai vantaggi che offre come alta sensibilità, efficienza e basso consumo energetico. Le linee guida che sono state seguite per sviluppare un metodo efficiente per la produzione di questo tipo di film sottili sono state la semplificazione dei processi produttivi, partendo dalla sintesi della soluzione sol-gel fino al trattamento termico, e l''utilizzo di precursori chimici meno tossici e pericolosi del comunemente usato 2-Me. Prima di tutto, dopo numerose prove, un metodo per sintetizzare soluzioni sol-gel di PZT omogenee e stabili su lunghi periodi è stato definito. Usando Piombo Acetato tri-idrato, Zirconio propossido e Titanio isopropossido come precursori dei metalli, due tipologie di soluzioni sono state preparate in questo lavoro: la prima utilizza come solvente acqua e acido acetico, la seconda isopropanolo e acido acetico. In entrambe le soluzioni l''utilizzo di AcOH permette di inibire le reazioni di idrolisi e condensazione tra i metallo organici, poichè il gruppo acetato sostituendosi ai gruppi propossido e isopropossido riduce la velocità della reazioni di policondensazione, prevenendo in questo modo la gelificazione della soluzione. I sol gel preparati sono stati depositati utilizzando la tecnica di spin coating su tre substrati differenti: Au/TiW/SiO2/Si, Pt/TiO2/SiO2/Si e ITO su vetro. Per ogni tipo di substrato è stato messo a punto un diverso trattamento termico in modo tale da ottenere film densi, uniformi e senza alcun tipo di cricche. Questo è stato suddiviso in tre step: il primo a 150°C per 5 minuti, il secondo a 370°C per 5 minuti e il terzo a 650°C. La durata del processo di cristallizzazione è stata ottimizzata a seconda del tipo di substrato utilizzato in modo tale da ottenere una struttura cristallina a perovskite in tutto il film e di evitare la formazione di cricche e l''interdiffusione di elementi tra il substrato e il deposito. La morfologia dei film depositati è stata studiata attreverso l''uso del microscopio ottico e della microscopia elettronica (SEM). L''influenza del tipo di substrato e del tempo di ricottura sul meccanismo e sulla tipologia di cristallizzazione dei film è stato analizzato e discusso attraverso l''analisi di diffrazione a raggi-X. Per tutti i tipi di substrato usati, film sottili di PZT con buona morfologia e con la giusta struttuta cristallina sono stati ottenuti con successo. I film depositati hanno mostrato durante il processo di cristallizzazione una crescita di strutture a rosetta identificate come porzioni di materiale cristallizzato la cui estensione cresce aumentando il tempo di ricottura. I film depositati hanno evidenziato la formazione di una struttura policristallina con una forte crescita preferenziale del piano (110) per tutti i substrati usati e le condizioni per cui sono state ottenute diverse orientazioni preferenziali sono state analizzate e discusse. Matrici di capacitori sono state ottenute da questi film, depositando su di essi elettrodi d''oro attraverso la tecnica di spattering, in modo tale da ottenere una struttura metallo/PZT/metallo che si presta a misurazioni elettriche. La caratterizzazione 4 elettrica dei film è stata effettuata misurando le curve C-V e ε-V di questi capacitori. Queste hanno mostrato, per tutti i dispositivi analizzati, una forma a farfalla tipica dei materiali ferroelettrici, confermando cosi l''effettiva ferroelettricità dei film depositati. Le costanti dielettriche per ogni tipologia di substrato sono state ricavate dalle misure di capacità e i valori ottenuti sono: 484 per film di PZT depostati su elettrodi d''oro, 655 per film depositati su ITO e 770 per film depositati su platino. Dalle curve C-V è oltre possibile ricavare valori approssimati del campo coercitivo (Ec) del corrispondente ciclo di isteresi. Questo è stato calcolato intorno ai 10-15 kV/cm. 5 Contents
Abstract ............................................................................................................................................................ 2 Abstract..............................................................................................................................................................3 List of figures .................................................................................................................................................... 7 Chapter 1 : Introduction ................................................................................................................................... 9 1.1 Research goals: ..................................................................................................................................... 12 Chapter 2: Background ................................................................................................................................... 13 2.1 Historical background: .......................................................................................................................... 13 2.2 Polarity of crystal classes: ..................................................................................................................... 14 2.3 Ferroelectricity: .................................................................................................................................... 16 2.4 Piezoelectric materials: ........................................................................................................................ 20 2.5 Piezoelectric actuating and sensing ...................................................................................................... 22 Chapter 3 : Lead Zirconate Titanate ............................................................................................................... 25 3.1 Crystal structure and phase formation ................................................................................................. 25 3.2 Thin films deposition methods: ............................................................................................................ 30 3.2.1 Physical vapor deposition: ............................................................................................................. 31 3.2.2 Chemical vapor deposition: ........................................................................................................... 32 3.3.3 Chemical Solution deposition: ....................................................................................................... 33 3.3 The Sol gel route ................................................................................................................................... 34 Chapter 4: Experimental procedures .............................................................................................................. 37 4.1 Precursor solution synthesis ................................................................................................................. 37 4.1.1 State of the art: ............................................................................................................................. 37 4.1.2 2-Me based solution ...................................................................................................................... 38 4.1.3 Carboxylic acid based solution ....................................................................................................... 38 4.1.4 PZT precursor solution synthesis ................................................................................................... 41 4.2 Substrate influence on crystallization mechanism ............................................................................... 46 4.2.1 Crystallization mechanism: ............................................................................................................ 46 4.2.2 Choice of the substrate.................................................................................................................. 47 4.3 Sol-gel deposition of PZT thin film ........................................................................................................ 50 4.3.1 Substrate preparation and spin coating: ....................................................................................... 50 4.4 Thermal treatment : ............................................................................................................................ 53 4.4.1 Thermal analysis: ........................................................................................................................... 53 4.4.2 Selecting an heating sequence: ..................................................................................................... 54 4.4.3 Stresses evolution during heating :................................................................................................ 59 6 4.5 Top electrodes deposition: ................................................................................................................... 61 Chapter 5: Results and discussion .................................................................................................................. 62 5.1 PZT films on Au/TiW/SiO2/Si substrate ................................................................................................. 63 5.1.1 X-ray diffraction analysis ............................................................................................................... 67 5.2 PZT films on ITO coated glass ............................................................................................................... 72 5.2.1 X-ray diffraction analysis ............................................................................................................... 76 5.3 PZT film on Pt/TiO2/SiO2/Si substrate ................................................................................................... 78 5.3.1 X-ray diffraction analysis ............................................................................................................... 81 5.4 Electrical characterization .................................................................................................................... 85 5.4.1 PZT films on Au/TiW/SiO2/Si substrate .......................................................................................... 86 5.4.2 PZT films on ITO coated glass substrate ........................................................................................ 87 5.4.3 PZT films on Pt/TiO2/SiO2/Si substrate........................................................................................... 89 Conclusions and further works: ...................................................................................................................... 91 References: .................................................................................................................................................... 94

7 List of figures Figure1: The direct (left) and converse (right) piezoelectric effect ''''''''''''''''''''''..14 Figure2: Scheme of crystal classes in function of pyroelectric, piezoelectric and ferroelectric properties '15 Figure3: Tetragonally distorted perovskite structure of PZT ''''''''''''''''''''''''''..16 Figure4: Domain formation upon cooling, arrows show polarization direction ''''''''''''''''17 Figure5: Hysteresis loop of ferroelectrics '''''''''''''''''''''''''''''''''''''.18 Figure6: a) Cubic Perovskite structure of non-polarized PZT. (b) Tetrahedral structure of polarized PZT
W (c) Wurtzite structure of AlN. In yellow Aluminium and in grey Nitrogen ''''.''''''''''21 Figure7: Direct and converse piezoelectric effect ''''''''''''''''''''''''''''''''..22 Figure8: Piezoelectric cantilever stack '''''''''''''''''''''''''''''''''''''''23 Figure9: a) PZT perovskite structure, b) PZT pyrochlore structure '''''''''''''''''''''''25 Figure10: Phase diagram of PbZrO3-PbTiO3 solid solution ..'''''''''''''''''''''''''''26 Figure11: (a) PZT Cubic structure above Tc (b) Movement of Ti or Zr ions below Tc ''''''''''''..28 Figure12: Piezoelectric coefficients of Pb(Zr,Ti)O3 ceramics as a function of composition close to the
m morphotropic phase boundary ''''''''''''''''''''''''''''''''''''...28 Figure13: Drawings of (100), (110) and (111) plains '''''''''''''''''''''''''''''''.29 Figure14: Overview of the sol gel process '''''''''''''''''''''''''''''''''''''.35 Figure 15: Firing schedule example for PZT sol gel ''''''''''''''''''''''''''''''''.36 Figure16: Flow chart of Yi and Sayer recipe ''''''''''''''''''''''''''''''''''''.40 Figure17: Hydrolysis and condensation reaction for substituted metal alkoxides ''''''''''''''..42 Figure18: PZT precursor solution synthesis flow chart ''''''''''''''''''''''''''''''.45 Figure19: Stage of spin coating :a) fluid deposition b) spin up; c) fluid spin off; d) solvent evaporation ''51 Figure20: Effect of spinning striation a) and particles contamination b) after film crystallization '''''52 Figure21: Thermal treatment flow chart ''''''''''''''''''''''''''''''''''''''58 Figure22: Optical microscope image of cracked PZT mono layer on a) Au/TiW/Si b) ITO coated glass '.60 Figure23: Micro cracks in PZT layer on Au/TiW/Si ''''''''''''''''''''''''''''''''..60 Figure24: a) PZT capacitors structure b) electric scheme of two capacitor in series. ''''''''''''..61 Figure25: Optical microscope image of gold electrodes matrix patterned onto PZT films ''''''''''61 Figure26: Optical microscope image of PZT films on Au electrode derived from 0,5M isopropanol based o
u solution a) double layer b) single layer '''''''''''''''''''''''''.'''''''.63 Figure27: Double PZT layer derived from isopropanol based solution '''''''''''''''''''''65 Figure28: Micro-cracks into 15 minutes annealed PZT film derived from isopropanol based solution '''65 Figure29: Star shape cracks onto PZT film derived from 1M water base solution. a) optical microscope
image b) SEM image '''''''''''''''''''''''''''''''''''''''''''65 Figure30: Island of different morphology into PZT film derived from 0,5M water based precursor
solution and annealed 5 minutes ''''''''''''''''''''''''''''''''''''.66 8 Figure31: XRD of PZT (Zr,Ti) 52/48 film derived from 1M water based solution and annealed at 650°C for 5 minutes '''''''''''''''.''''''''''''''''''''''''''..67 Figure32: XRD of PZT (52/48) double layer film derived from isopropanol based precursors solution
annealed 650°C for 5 minutes '''''''''''''''''''''''''''''''''''''..68 Figure33: XRD of 15 minutes annealed PZT layer derived from isopropanol based precursors solution '..69 Figure34: XRD of PZT single layer derived from isopropanol based precursors solution and annealed
at 650°C for 30 minutes '''''''''''''''''''''''''''''''''''''''''.69 Figure35: XRD of PZT monolayer obtained using 0,5M water based solution and 5 minutes annealing ''70 Figure36: XRD of PZT monolayer obtained from 0,5M water based solution and annealed 15 minutes at 650°C ''''''''''''''''''''''''''''''''''''''''''.'71 Figure37: XRD of PZT double layer obtained from 0,5M water based solution and annealed 5 minutes at 650°C ''''''''''''''''''''''''''''''''''''''''''''71 Figure38: Optical microscope transmission mode image of PZT films on ITO coated glass substrate annealed at 650°C for a) 5min b) 10min c) 15min ''''''''''''''''''''..73 Figure39: SEM image of a) transversal section of single PZT film b) top view of 10 minutes annealed PZT double layer '''''''''''''''''''''''''''''''''''''''..75 Figure40: Optical microscope image of PZT double layer a) fully crystallized b) after 10 minutes annealing at 650°C ''''''''..''''''''''''''''''''''''''''''.75 Figure41: XRD of PZT films deposited on ITO coated glass substrate annealed at 650°C for a) 5 min b) 10 min ''''''''''''''''''''''''''''''''''''''''''.76 Figure42: XRD of PZT film deposited on ITO coated glass substrate and annealed 15 min at 650°C ''''76 Figure43: XRD of PZT double layer films on ITO coated glass substrate annealed for 15 min at
650°C a) film derived from isopropanol based solution b) film derived from water based solution ''''''''''''''''''''''''''''''''''''''''''''''..77 Figure44: Optical microscope images of double layer PZT films deposited on Pt/TiO2/Si substrate
and annealed for 15 min at 650°C a) film derived from 0.5M isopropanol based solutions b) film derived from 0.5M water based solution '''''''''''''''''''''''''''.78 Figure45: Optical profilometer analysis of PZT single layers deposited on Pt/TiO2/Si substrate
a) film derived from isopropanol based solution b) film derived from water based solution ''.79 Figure46: Optical microscope image of PZT single layer deposited on Pt/TiO2/Si substrate and
annealed at 650°C for 30 minutes ''''''''''''''''''''''''''''''''..''..80 Figure47: XRD of PZT films deposited on Pt/TiO2/Si substrate derived from 0,5M isopropanol
based PZT precursors solution and annealed at 650°C a) 5 min b) 15 min c) 30 min ''''''..81 Figure48: XRD of PZT films derived from 0.5M water based PZT precursors solution deposited
on Pt/TiO2/Si and annealed at 650°C for a) 5min b) 15 min c) 30 min'''''''''''''''...83 Figure49: a) C-V and b) ε-V curves for 1µm PZT film deposited on Au/TiW/Si substrate annealed 5 minutes ''''''''''''''''''''''''''''''''''''''''''''86 Figure50: a) C-V and b) ε -V curves of Au/PZT/ITO capacitor fabricated using isopropanol based
PZT precursors solution '''''''''''''''''''''''''''''''''''''''''..87 Figure51: a) C-V and b) ε -V curves of Au/PZT/ITO capacitor fabricated using water based PZT precursors solution '''''''''''''''''''''''''''''''''''''''''''...88 Figure52: a) C-V and b) ε -V curves of Au/PZT/Pt capacitor fabricated using isopropanol based
PZT precursors solution ''''''''''''''''''''''''''''''''''''''''''89 9 Chapter 1 : Introduction
The objective of the modern materials science is to tailor materials, starting with their chemical composition, constituent phases, and microstructure, in order to obtain a desired set of properties suitable for given applications. Moreover, as the size of electronic devices is getting smaller, and the operational speed required by devices is getting faster, the electronic components industries are encountering many challenges [5]. The need for integrated circuits with ever increasing functions and capacity are driving today's semiconductor technology . This, combined with an ever increasing degree of miniaturization, requires properties from materials which border their physical limits. A class of material that shows very interesting electrical properties and whose use is gaining great attention in the electronic industry, is the class of ferroelectric ceramics. Ferroelectric materials are a group of crystalline dielectrics having in a certain temperature range spontaneous polarization, and whose orientation can be reversed by an electric field and varies significantly under the influence of external factors. The electrical properties of ferroelectrics are in many aspects similar to the magnetic properties of ferromagnetics. Nowadays, several hundred ferroelectrics are known. Barium titanate, Rochelle salt (which has given its name to an entire group of crystals), triglycine sulfate, potassium di-hydrogen phosphate, lead titanate (PT), titanate zirconate of lead (PZT), are among the most widely used and thoroughly investigated ferroelectrics. Ferroelectric materials display unique dielectric, pyroelectric, piezoelectric and electro-optic properties that are utilized for a variety of applications such as: capacitors, dielectric resonators, sensors, transducers, actuators, ferroelectric non-volatile memories, dielectric memories, optical waveguides, displays, micro'' electro - mechanical systems (MEMS), miniaturized mechanical and electro-mechanical elements, made using techniques of micro-fabrication. Furthermore, ferroelectric ceramics possess very interesting electrical features, and they can be considered for some applications much better than for example the silicon dioxide, that nowadays it''s the most used material in the semiconductor industry. The use of ferroelectric material would sometimes be preferred to create high performance devices and lots of materials whose bulk properties are well known, are finding new applications in integrated circuits in the form of thin films. Unfortunately the use of these materials is still difficult due to thermal processing requirements which are incompatible with other components of the integrated circuits. A particular characteristic of these materials, on which this thesis is focused, is the piezoelectricity. The piezoelectricity is the ability of some material, called piezoelectrics, of producing a mechanical output when subjected to an electrical input, and vice versa of creating free charges in response of a mechanical 10 stimulus. Piezoelectric materials have been used since the very beginning of the XX century. They were used in ultrasonic devices or in phonograph. They are now present in many simple macroscopic devices either sensors or actuators. With the rise of Micro-Electro-Mechanical-Systems (MEMS ) a new focus is on piezoelectricity. By definition the best sensors are designed to exchange minimal amount of energy with the measurand since this fact ensures minimal perturbation in measurements. For this perspective is evident the interest in miniaturize sensors and using piezoelectrics. The great advantage of the piezoelectric phenomenon is that it does not exhibit failure modes associated with charge storage and it is reversible and linear (properties which electrostatic capacitors do not have)[7] . On the other hand the importance in miniaturize actuators is related to the less invasive size of small devices, in fact the smaller is the devices size, the less is the energy used for the actuation operations. This fact suggests to use materials with high energy density as piezoelectrics. Moreover, one can expect larger force output from piezoelectric MEMS actuators than that available from electrostatic actuators now being employed in MEMS. Furthermore, micromachined piezoelectric actuators are capable of sub-microsecond response times and can operate at low bias voltages ( 5-15 V ) while consuming only microwatts of power [3]. Piezoelectric materials show also high acoustic quality. One can summarize their advantages as follows: - Strong forces, or alternatively large excursions in bending structures;
- Low voltage because of high dielectric constant;
- High efficiency in energy conversion
- High-speed and high-frequency
- High acoustic qualities - Linear behavior In spite of these promising properties, piezoelectric materials have some disadvantages. Because of the small ultimate strain (usually less than 10% ) and high stresses (several MPa ) of piezoelectrics materials, their use is only suitable for applications with large forces but small displacements. Moreover, the most popular piezoelectric material (PZT ) contains lead which is toxic. Researchers are now trying to develop some new lead-free piezoelectric materials as the Barium Titanate (BTO) or the Polyvinylidene fluoride (PVDF ) which is a synthetic polymer that exhibits low piezoelectricity. Until now, such researchers are far from reaching results comparable to that of PZT and the process temperature used in lead containing material is still lower than that of lead free material as BTO . Finally, the problem of integration of small amount of piezoelectric materials (as thin films or as bulk) in MEMS is the major challenge researchers are dealing with. The complications are given both by the technological feasibility to deposit thin layers of piezoelectric materials on silicon substrate and by the reduction of the technologically caused constraints, like thickness deviation and deformation by residual stress. There are essentially three approaches to realize devices: the first one is based on the deposition of piezoelectric thin films on silicon substrates with appropriate insulating and conducting layers followed by surface or silicon bulk micromachining to realize the micro-machined transducers (additive approach); the second one is the direct bulk micromachining of single crystal or polycrystalline piezoelectrics and 11 piezoceramics which are thereafter appropriately electroded to realize micro-machined transducers (subtractive approach) and the third one is the based on the integration of micro-machined structures in silicon via bonding techniques onto bulk piezoelectric substrates (integrative approach) [7]. The most used approach is the additive one, based on the deposition of thin oxide film. The best route to create this kind of piezoelectric film is the sol-gel method. The sol''gel method has been widely studied as new route for the preparation of ceramics, due to its advantages such as: the mixing of reactants on a molecular level, a better control of the stoichiometry, higher-purity raw materials, easy formation of ultra- fine and crystallization powders with controlled morphology and grain size. It should display considerable stability in solution to guarantee the reproducibility of the materials preparation and it has to be easy to be purified to provide sufficient chemical quality of the final products. Generally, physical methods have shown considerable success in the growth of high quality ultrathin oxide films but these methods are expensive and unfavorable when large substrate surfaces need to be deposited. Chemical methods appear to be more suitable for easier fabrication of larger areas, simpler composition control and better thin film homogeneity as well as for lowering the synthesis temperature. Unfortunately, some problems, such as microstructural instability, could appear as a result of high temperature or long annealing time when chemical methods are used for ultrathin film preparation. In other words, it was found that oxide thin films with thickness below the critical value break up and form nanoislands after high temperature processing. This type of micro structural instability might have a negative influence on ferroelectric properties and needs to be avoided in the preparation of ultrathin films. One of the major advantages attributed to the use of chemical methods is the ability to produce fine active powders with high surface areas. Ceramics derived from powders of this type are typically characterized by lower sintering temperatures and fine microstructures. Furthermore, the ability to process this class of materials at low temperatures has allowed for the use of less expensive materials in thick film capacitor applications, and for sol''gel derived films, integration directly onto semiconductors without deleterious interactions. The development of sol''gel technology has, at very early step, put forward a request on the development of precursor compounds that are easily transformed into chemically reactive forms or hydrated oxides on hydrolysis [6]. Therefore, the ferroelectric ceramics thin films and their ferroelectric, dielectric and piezoelectric properties are important to be reviewed by providing an insight into different processes which may affect the behavior of ferroelectric devices and specific applications. Among the available piezoelectric materials, lead zirconate titanate Pb (ZrxTi(1-x))O3, also called simply PZT as already mentioned, is the most popular due to its superior dielectric constant, piezoelectric constants, and thermal stability. PZT finds broad application as capacitor, transducer, sensor and actuator. The growing need of miniaturization of such devices and their implementation onto a silicon chip leads to the development of micro-electromechanical systems (MEMS). To exploit the good PZT properties in a wide range of piezoelectric actuator applications in MEMS devices thick films are needed from some microns to 12 several 10''s micron. This kind of thick films are not easy to be obtained via sol gel technique because of film cracking above a critical thickness of few microns. Many attempts were undertaken to increase the single layer thickness and so to reduce the quantity of processing steps. When the thickness of a single layer increases, it simultaneously increases the internal stress into the film which easily leads to cracking. It''s thus fundamental to define a procedure to deposit dense, thick and free from cracks PZT films that has to be as simple as possible and guarantee good electrical properties. 1.1 Research goals: The purpose of this study is to develop a reproducible method to synthesize Lead Zirconate Titanate (PZT) thin films for ferroelectric and piezoelectric applications using sol gel technique. The role of this type of material in the electronic devices and MEMS industry is gathering great attention for the production of high density memories, actuators, transducers and sensors thanks to the ability to integrate this type of films inside the silicon chip, increasing in this way the miniaturization of the whole device with all the resulting benefits, as low power consumption and high frequency response. Starting from PZT sol gel precursor solution synthesis, different routes will be studied to find which is the one that permit to obtain the best electrical features from the ferroelectric film. The ease of solution preparation, as the reduction of processing steps and the using of less toxic agents as far as possible, together with the solution stability over long period of time, will be taken as main points to choose the best PZT precursor solution synthesis route. Different types of substrates will be used to study the crystallization behavior of PZT thin film on different materials and to analyze the influence of the bottom electrode type and texture, on the PZT crystals growth and on its final electrical properties. A right thermal treatment, that is necessary to crystallize the film and it has to avoid any type of cracks formation , will be developed for each type of substrate used, and a right annealing time and temperature will be chosen to obtain a complete PZT perovskite phase formation. The influence of the substrate electrode on the annealing procedures will be analyzed and discussed, and a reproducible method to obtain the best features from the different type of substrates and sol-gel solutions used, will be established. The final films morphology, crystallographic phase formation and ferroelectric characteristics will be analyzed using optical microscope, SEM, grazing incident X-ray diffraction and measuring the C-V and ε-V curves respectively.
13 Chapter 2: Background
2.1 Historical background: The discovery of piezoelectricity by J. Curie and P. Curie in 1880 followed the experiments on the pyroelectricity. The pyroelectricity is the ability of some materials to exhibit a spontaneous electric dipole moment when their temperature changes over certain values. These studies led to piezoelectricity, or the generation of charges by application of mechanical stress. The ferroelectricity, that it''s the ability of polarization reversal at reversed electric field, was discovered much later because the resulting net polarization of such crystals is very small and it was difficult to detect it. The formation of domains in ferroelectrics, or the formation of differently oriented polarization within virgin single crystals, leads to a lack of any net polarization and to very small pyroelectric and piezoelectric response. Only in 1920 Valasek discovered that the polarization of Rochelle salt (NaKC4H4O6 x 4H2O) could be reversed by application of an external electric field. Ferroelectricity, which name refers to certain magnetic analogies, though it is somewhat misleading as it has no connection with iron at all, has also been called ''''Seignette electricity'''', as Seignette or Rochelle Salt was the first material found to show ferroelectric properties such as a spontaneous polarization on cooling below the Curie point, ferroelectric domains and a ferroelectric hysteresis loop. For many years, Rochelle salt was the only crystal known to have this ferroelectric property. A huge leap in the research on ferroelectric materials came in the 1950''s, leading to the widespread use of barium titanate (BaTiO3) based ceramics in capacitor applications and piezoelectric transducer devices. Shortly after the investigation of lead zirconate titanate Pb(ZrxTi1-x)O3 ceramics was started, Jaffe found the highest permittivity in the composition PbZr0.53Ti0.47O3 and subsequently patented it and named it shortly PZT. The highest permittivity found at this composition is a consequence of the fact that near that composition the PZT crystal structure is across the rhombohedral-tetragonal limit of the phase diagram, called morphotropic phase boundary( MPB). The instability of the crystallographic phase near this composition is the reason of the high permittivity values. After the discovery of PZT, lots of research has been done to synthesize thin films of this materials and to incorporate these films into the silicon chip. To do that, the best route turned out to be the sol-gel deposition and further works have been focused on the synthesis and deposition of PZT film using this technique. Among all, very important contributes were given by Yi and Sayer and Budd et al. works. 14 2.2 Polarity of crystal classes: Depending on the crystal geometry there are 7 major crystal systems: cubic, hexagonal, orthorhombic, monoclinic, tetragonal, triclinic and trigonal. These crystal groups can be subdivided into point groups (crystal classes) according to their symmetry with respect to a point. In total, there are 32 such classes and 11 of them posses a centre of symmetry. If a uniform stress is applied to a centrosymmetrical crystal the resulting small movement of charge is symmetrically distributed around the centre of symmetry and so the relative displacement is fully compensated. The remaining 21 non-centrosymmetric classes, all except one, exhibit polarity when subject to stress because the charge displacement is not compensated. The polarity is linear with stress and the reversal of the stimulus results in a reversal of the response. This effect is termed piezoeffect and the materials that possess this property are defined as piezoelectric material. The piezoelectric effect was discovered by Pierre and Jacques Curie in the late 1800 and occurs, as explained above, in materials with dipole moments resulting from non-centrosymmetric crystal structures. The direct piezoelectric effect occurs when a charge is generated due to a change in the dipole movement caused by the application of a mechanical stress to the crystal. Coupled with some electrode it actually works as a transducer which produces charge (voltage) at the electrode upon the application of stress, exploiting the direct piezoelectric effect. The converse piezoelectric effect occurs when a strain is generated on the crystal by the application of an electric field. Quartz crystals in bulk resonators are examples of natural piezoelectric crystals. In Fig. 1 the direct and converse piezoelectric effect are depicted. Of the 20 piezoelectric crystal classes only 10 have a unique polar axis. Crystals belonging to these classes are called polar because they possess a spontaneous polarization or electric moment per unit volume. The spontaneous polarization in such crystals is temperature dependent and its existence can be detected by observing the flow of charge to and from the surface on change of temperature. This is the pyroelectric Figure 1:The direct (left) and converse (right) piezoelectric effect [3] 15 effect and the 10 polar classes that have a unique polar axis are often referred to as the pyroelectric classes. Finally, the spontaneous polarization vector can be reversed, thus it''s possible to reverse the barycenter of positive and negative charges. The materials that possess this property are called ferroelectrics. .

Figure 2: Scheme of crystal classes in function of pyroelectric, piezoelectric and ferroelectric properties [8] 16 2.3 Ferroelectricity: Ferroelectric materials are generally defined as materials that possess oriented spontaneous polarization in the absence of external electric field, and by the ability of the spontaneous polarization vector to be switched between different orientations by an electric field [3]. The spontaneous polarization Ps is generated from non-centrosymmetric arrangement of ions in unit cell, which produces a permanent electric dipole moment and the appearance of charges on the surface of the element. The unit cell of a typical ferroelectric material has a perovskite structure (ABO3) as illustrated in Figure 3 where A atom, B atom, and oxygen occupy the corner site, body-centered site, and face-centered site, respectively. When an external field (E) is applied, the polarization (P) direction switches as the TiO6 octahedra deforms to accommodate the electrical stress imparted by the external field. In the polarization state, the oxygen octahedra is regular and the symmetry is cube. As a rule, ferroelectrics are not uniformly polarized and a single crystal can be divided into spatial regions having different directions of polarization. Adjacent unit cells are inclined to polarize in the same direction and form a region called a ferroelectric domain. Domains are region of space characterized by having the same alignment and the same absolute value of the spontaneous polarization. These domains are separated by boundaries which are referred to as domain walls. A certain amount of energy is associated with the domain walls. In non-ideal crystals that contain surfaces and defects some depolarization along with the polarization can occur. The depolarization energy arises because of the accumulation of charges on impurities or close to the surface (deviation from the ideal crystal) where Ps decreases to zero. Figure 3: Tetragonally distorted perovskite structure of PZT [3] 17 A decreasing polarization Ps close to a surface can act as a source of the depolarization field. The depolarization fields which appear on cooling are usually sufficient to prevent any net polarization. The final domain configuration is determined by minimizing an appropriate free energy equation including the term of depolarization energy and domain wall energy. The equilibrium domain structure of a ferroelectric is determined by the balance between the decrease in the energy of electrostatic interaction of domains when the crystal is broken down into domains and the increase in energy arising from the formation of new domain walls having excess energy. In equilibrium, when all depolarizing fields are compensated, the minimum energy in the absence of defects would correspond to a single domain configuration. However, such an equilibrium state is rarely achieved in a virgin crystal in absence of electric fields and thus, multi- domains form. The number of different domains and the relative orientation of the spontaneous polarization in the domains, depends on many factors including the crystal symmetry, internal stresses, the electrical conductivity, the defect structure, the magnitude of spontaneous polarization and elastic and dielectric compliances, as well as the crystal preparation and sample geometry. Depending on the particle size of the polycrystalline ceramic material, the individual crystallites contain only a few domains, bounded by domain walls. In the absence of external factors the total electric dipole moment of a specimen is practically equal to zero because the polarization of the domains self equilibrates. If an electric field or a mechanical stress strong enough is applied, shifting occurs and the polarity of whole regions can be reversed as a result of domain reforming. These processes and the irreversible displacement of domain walls are some of the reasons for the familiar phenomenon of ferroelectric hysteresis in fact for any ferroelectric material a plot of the polarization versus the applied field will give an hysteresis loop, showing the non linearity between the applied field and the polarization of the material. The field required to cause this reorientation of the domains is called the coercive field (Ec). The remnant polarization (PR), or charge remaining on the material surfaces after polarization, is a measure of the ferroelectric performance of the material. A typical hysteresis loop is shown in Fig.5. At low fields well below the coercive field Ec and at high fields above the coercive field Ec a ferroelectric behaves like an ordinary dielectric but at the coercive field Ec polarization reversal occurs, giving a large dielectric non-linearity. The area within a loop is a measure of the energy required to twice reverse the Figure 4: Domain formation upon cooling, the arrows show polarization direction [2] 18 polarization. At zero electric field the electric displacement within a single domain can have two values corresponding to the opposite orientations of the spontaneous polarization. In a multi-domain crystal the average electric displacement at zero field can have any value between these two extremes. In principle, the spontaneous polarization is equal to the saturation value of the electric displacement extrapolated to zero field. The remanent polarization Pr (the displacement at zero field) may be different from spontaneous polarization Ps if reverse nucleation occurs before the applied field reverses. This can happen in the presence of internal (or external) stresses or if the free charges below the surfaces cannot reach their equilibrium distribution during each half-cycle of the loop. Additionally, if the material is constraint the saturated polarization may not be achieved. The defects such as impurities affect the dielectric properties and the switching behavior in fact in any crystalline lattice they generally cause deformation of the surrounding volume and modification of the local fields and in general the presence of defects tends to increase the coercive field.

Most ferroelectric materials undergo a structural phase transition from higher temperature paraelectric phase into a lower temperature ferroelectric phase, and the temperature where this transition takes place is called the Curie point, Tc. Figure 5: Hysteresis loop of ferroelectrics [6] 19 When the temperature is close to the Curie point, thermodynamic properties (such as spontaneous polarization, specific volume, and entropy) show anomalies, and the structure of the crystal changes. For example, the dielectric constant in most ferroelectric crystals has a very large value near their Curie point, which is usually referred to as the ''dielectric anomaly '. Above the Curie point the dielectric permittivity falls off with temperature according to the Curie-Weiss law:
where C is the Curie-Weiss constant and T0 is the Curie-Weiss temperature. T0 may be different from the Curie point Tc. In the case of a first-order phase transition, T0 < Tc, while for the second''order phase transition T0 = Tc. Representative ferroelectric materials include: barium titanate BaTiO3, lead zirconate titanate Pb(Ti,Zr)O3 (PZT), lead titanate (PbTiO3), lead lanthanum zirconate titanate (PLZT), Pb(ZnNb)O3 and ferroelectrics like lead magnesium niobate (PMN) which have been developed and utilized for a variety of applications. Generally, these materials have a high dielectric constant and are characterized by large piezoelectric, pyroelectric and electro-optic effects as well as nonlinear optical effects. On the basis of these unique properties, ferroelectric materials have been applied to produce active elements of various devices. Generally, there is a considerable interest in ferroelectric crystals as transducer materials for their spontaneous polarization and the strong sensitivity that is attributed to their higher electromechanical coupling than piezoelectric crystal, such as quartz. While ferroelectric materials have a higher electromechanical coupling, they are not as stable as the single crystal piezoelectric materials. Both the poly and the single crystals are made of brittle materials, which limit their practical size for high frequency ultrasonic applications where thin films on wafer are required.


20 2.4 Piezoelectric materials: Piezoelectric materials, as mentioned above, are a class of materials that produce a mechanical output when subjected to an electrical input and vice versa. Space charges are created on the surface of the material when a certain amount of stress is applied. This happened because of the deformation of the crystal structure and the ions displacement, that creates a dipole moment into the crystal. The inverse effect is the displacement of ions into the crystal structure and consequently the deformation of the material ,due to the application of an electric field. Piezoelectric materials exist spontaneously in nature and examples are quartz, Rochelle salt and Tourmaline. The first application of these materials was during the First World War when quartz was used as ultrasonic transducer to fabricate sonars. [8] An important step in the development of these materials was during the Second World War with the synthesis of ferroelectric oxides as barium titanate BaTiO3 and lead zirconate titanate Pb(Zr, Ti)O3 in ceramic form. These two last materials possess piezoelectric properties one hundred of times higher than the natural one as quartz and moreover it''s possible to tailor the fabrication parameters, as composition or material''s doping, to increase specific properties and to adapt the ceramic to specific applications. More recently the processing of these materials, and in particularly of Pb(Zr, Ti)O3, in the form of thin films has been developed, due to the necessity of incorporating these materials in microelectronic devices and in micro electro-mechanical systems (MEMS). Since 1976, PZT thin films have been developed using both vacuum and chemical solution techniques and these films have been widely exploited for current and potential use in such applications as nonvolatile memories (such as DRAMs), decoupling capacitors, sensors (acoustic emission sensors, vibration monitors, chemical microsensors), piezoelectric micromotors, and MEMS devices [2,4]. With the integration of PZT into such microelectronics and MEMS devices a more complete understanding of the thin film behavior, including nucleation and grain morphology and their effects on the properties is needed. Further work has to be done to obtain a processing methods to achieve structures with properties comparable to and exceeding those of the bulk material as well as possessing viability for integration into microelectronic fabrication schemes [1]. Nowadays, the most commonly used piezoelectric materials are ceramics and among these, three materials are principally used in microelectronics: Aluminium Nitride (AlN), Zinc oxide (ZnO) and the lead zirconate titanate (PZT). The AlN is a type III semiconductor with a large gap (6ev) [8] and is very used in optics and in acoustics to create elastic-waves resonators, in which its piezoelectric properties are exploited. AlN has a Wurtzite crystal structure, Fig.6c. Due to its structure the polar axis of this material cannot be oriented by the application of electric field. Therefore, development of any deposition process for these films has to result in films with a well-oriented polarization axis. Although its piezoelectric coefficients are nine/ten times smaller than the PZT , it has been used in many applications, such as bulk acoustic wave resonators. In any 21 case, it is usually used in sensing devices because as actuator it does not assure the required electromechanical coefficients. Zinc oxide, ZnO, is a piezoelectric materials which also has a Wurtzite structure and it has been used in many applications instead of AlN. The lead zirconate titanate is a perovskite-type (ABO3) ferroelectric ceramic that possess excellent dielectric and piezoelectric properties and it is the most used material for actuation thanks to its high piezoelectric coefficients, especially when it''s synthesized in the morphotropic phase. This work will be focused on this material and its properties will be presented in details in the next chapter. Recently lots of interest is focused on the development of piezoelectric materials without the use of lead, that is a well known toxic agent and its use should be reduced to a minimum. Unfortunately the materials that show the best piezoelectric coefficient are lead-based materials as PZT, but a family of new materials based on BaTiO3 is being studied and so far it''s showing good result even if still not comparable with PZT. Other materials as Lithium Niobate and Lithium Tantalate are gathering interest for their important piezoelectric coupling coefficients. The very last researches in piezoelectricity regard new materials with high piezoelectric coefficients. Such materials are called relaxors. Relaxors are very promising for the development of more efficient piezoelectric materials because they assure piezoelectric coefficient three order higher than the classical PZT. At this class belong the lanthalum modified lead zirconate titanate (PLZT), the lead magnesium niobate-lead titanate(PMN-PT) and others [6]. The usefulness of these new piezoelectric materials is still an open issue. Their integration on silicon substrate is not as understood as that of PZT and it present lot stability issue. Moreover, many authors are trying to really understand the origin of such giant piezoelectric coefficients but they have not yet arrived to a shared solution. Figure 6: (a) Cubic Perovskite structure of non-polarized PZT. (b) Tetrahedral structure of polarized PZT (c) Wurtzite structure of AlN. In yellow Aluminium and in grey Nitrogen. [7] 22 2.5 Piezoelectric actuating and sensing
Piezoelectric transduction is an alternative to electrostatic transduction for MEMS resonators and offers the potential advantage of increased electromechanical coupling strength, inherently linear performance, and an elimination of the need for small electrode gaps potentially leading to larger power handling. As already mentioned , the direct piezoelectric effect occurs when a charge is generated due to a change in the dipole movement caused by the application of a mechanical stress to the crystal. The converse piezoelectric effect occurs when a strain is generated on the crystal by the application of an electric field. The direct and the converse piezoelectric effect are illustrated in Figure 7. The equations of state relating the electric and elastic variables for the piezoelectric element are S1= s E 11T1+s E 12T 2+sE 13T3 +d31E3 (2.1) for strain generated in the x direction by an applied stress and electric field in the z direction and D3=d31(T1+T2) + d33T3 + ε3E3 (2.2) for the electric charge generated by an applied stress and electric field. In (2.1) and (2.2) S is the strain, T is the stress, s E is the elastic compliance, d is the piezoelectric constant, E is the electric field intensity, D is the electric charge generated and ε is the material permittivity. To obtain a complete description for all of the directions a matrix can be constructed using the equations above. Figure 7: Direct and converse piezoelectric effect [2] 23 Therefore from (2.1), in the absence of an applied stress, an applied electric field in the z direction results in a strain in the x direction via the piezoelectric coupling coefficient d31. Likewise, from (2.2) in the absence of an electric field, an applied stress in the x direction results in a charge in the z direction via the piezoelectric coupling coefficient d31. The efficiency of this conversion of mechanical energy to charge or charge to mechanical energy is given by the piezoelectric coupling coefficient If an additional material layer is used to move the piezoelectric material off of the neutral axis of a beam, the induced strain in the piezoelectric material when an electric field is applied generates a moment that causes the beam to bend. Figure 8: Piezoelectric cantilever stack [2]
If the input electric field is an RF signal that matches the natural resonant frequency of the beam, the resultant strain from the converse piezoelectric effect can induce resonance of the beam and vice versa. From (2.1), the maximum strain that can be induced in the piezoelectric element by an applied field is given by (2.4)
where V is the applied voltage and tp is the thickness of the piezoelectric layer. 24 The maximum force that can be generated when the piezoelectric element is held in the zero strain condition is given by Fmax= d31EbV (2.5) where E is the Young''s modulus of the piezoelectric element, b is the width of the element, and V is the applied voltage. In almost all applications, the piezoelectric element is mounted or attached to a substrate. The inclusion of a substrate can cause a reaction force to the piezoelectric-induced strain that reduces the actual amount of induced strain from the maximum that can be achieved in (2.4). Moreover, also when the piezoelectric element is attached to a surface, like in the case of a cantilever beam, the equations of the bending moment and of the strain on the top of the surface can be extrapolated and these depend not only on the voltage applied to the element but also on the thickness and stiffness of both beam and the piezoelectric layer. If the oxide beam is too stiff, the actuator will not induce any displacement. On the other hand, if the beam is too soft, it will only stretch under the drive electrode. For these reason a complete mechanical analysis of the whole system is necessary to increase the performance of the device.


25 Chapter 3 : Lead Zirconate Titanate
3.1 Crystal structure and phase formation One of the intensively studied ferroelectric materials system is the solid solution of lead zirconate (PbZrO3) and lead titanate (PbTiO3). This solid solution ceramic is abbreviated as PZT and normally written as Pb(ZrxTi1-x)O3 [3]. Although the properties and structure of bulk lead zirconate titanite have been widely studied and well established, there is much interest and consequently, many researches into thin film PZT. PZT is a perovskite type (ABO3) ferroelectric material well known for excellent dielectric and pyroelectric properties and for having the highest values of piezoelectric coefficients. The basic crystal structure of PZT is of the perovskite type (ABO3) illustrated in figure 9a , but often the non- ferroelectric pyrochlore phase is obtained after crystallization, figure 9b. Pyrochlore is a ternary compound with an A2B2O7 stoichiometry, where A is a larger and trivalent cation and B is a smaller and tetravalent cation. The pyrochlore structure is related to the fluorite structure. a) b) Figure 9: a) PZT perovskite structure, b) PZT pyrochlore structure [2] Pyrochlore phase arises due to lead deficiency in the deposit and un-equilibrated oxygen transport and it is considered to be transient or metastable. During the crystallization process, oxygen and Pb react readily to form PbO. A shortage of Pb will keep the oxygen gain greater than the oxygen loss allowing the metastable pyrochlore to stabilize. By adding excess lead (10-20%), the volatility of PbO can be compensated but too much lead will cause the oxygen loss to be greater than the gain, transforming the metastable pyrochlore in a fine pyrochlore matrix. The latter case is the most common, although it is 26 possible to achieve complete pyrochlore transformation to perovskite through an approximate balance between oxygen loss and gain. The presence of a pyrochlore phase affects strongly the properties of the material changing its micro structure and in particular it decreases the electrical and piezo-electrical performances. For all these reasons it''s important to find a processing method that eliminates, or strongly reduces, the presence of pyrochlore phase and this it''s done well balancing the oxygen gain and loss, tailoring the lead quantity. The sub-solidus phase diagram for PbZrO3-PbTiO3 is shown in Figure 10. The PbZrO3-PbTiO3 system exhibits multiple ferroelectric, anti-ferroelectric, and paraelectric phases, along with various symmetries, including tetragonal, rhombohedral, orthorhombic and cubic. Figure 10: Phase diagram in PbZrO3-PbTiO3 solid solution [3] Ferroelectric properties of PZT depend on the Zr/Ti ratio below the Curie temperature (Tc). The Curie temperature varies from 230 °C to 490°C depending upon the Zr/Ti ratio. Above the Curie temperature of PZT, paraelectric cubic phase exists and the Zr/Ti atoms are located at the exact center of the unit cell, regardless of the composition and so it exhibits zero polarization (figure 11a). Below the Curie temperature the centers of positive and negative charges don''t compensate due to the Zr or Ti ions shift and for this reason the PZT possesses a finite polarization Fig. 11b. This final polarization can be modified by an external field. 27 a) b) Figure 11: (a) PZT Cubic structure above Tc (b) Movement of Ti or Zr ions below Tc [8]

Deviations from the ideal cubic perovskite structure start to occur when PZT is cooled below the Curie point. When the PZT is cooled below the Tc, it moves from a cubic to a tetragonal or rhombohedral phase, depending on the Zr/Ti ratio, that are still very close to the cubic phase Fig. 11a. For an high Ti concentration (Ti > 48%), below the Curie point, the ferroelectric phase is tetragonal. In this case, the Zr/Ti atom displaces off center and a distortion of the TiO6 octahedra occurs along one of the six <00l> directions. These six directions yield six possible domain configurations separated by either 90° or 180 ° domain walls. The direction of the Zr/Ti displacement is the tetragonal c-axis in the resulting unit cell. Due to this distortion, as mentioned before, a dipole is created since the centers of positive and negative charge no longer coincide and this atomic displacement is the source of the spontaneous polarization (figure 10b). For higher Zr concentrations (Zr > 52%), PZT is rhombohedral below the Curie temperature. The rhombohedral distortion occurs when the Zr or Ti atom moves towards the face-center of the oxygen octahedra (the cubic <111> direction). In this case there are eight equivalent directions in which the Zr/Ti atom can move. The most widely studied composition of PZT occurs at the boundary between the tetragonal and rhombohedral phases. This is known as the morphotropic phase boundary (MPB) and its composition is approximately 52 mol% of PbZrO3 and 48 mol% of PbTiO3 at room temperature. This composition bounds the abrupt structural change between the rhombohedral and tetragonal phases and exhibits greatly enhanced dielectric and piezoelectric properties. The values of these characteristic coefficients show a peak at the MPB, explaining why these compositions are technologically so interesting. At the MPB composition, the free energy of the rhombohedral and tetragonal phases are equal and it is probable that an electric field may easily cause switch between tetragonal and rhombohedral domain states. There are thus effectively 14 available directions (8 from the rhombohedral and 6 from the tetragonal ) along which polarization may be reoriented, leading to a large effective remanent polarization 28 for compositions near the MPB. The large remanent polarization at MPB in randomly oriented PZT also helps to increase the peak in the piezoelectric coefficients at MPB. Good-quality single crystals of PZT with near MPB compositions are not available due to the phase separation, therefore comparable measurements could not be made on single-domain single crystals. Calculations using the thermodynamic phenomenological theory have shown, however, that a peak in the piezoelectric coefficients and dielectric permittivity should be expected at MPB in mono-domain single crystals. Figure 12: Piezoelectric coefficients of Pb(Zr,Ti)O3 ceramics as a function of composition close to the morphotropic phase boundary [3]
When in the perovskite phase, the PZT can be orientated in three different major directions: <100>, <110> and <111>. It''s necessary, moreover obtaining a perovskite phase that have good ferroelectric and piezoelectric properties, obtain the PZT preferentially oriented in one of the above directions. In fact the electrical properties of the PZT depend on the crystal orientation and so accordingly to the final use of the device and on the properties required, it''s necessary to grow PZT crystals oriented in the direction that enhances those required properties. PZT orientation is described starting from the plane that is parallel to the surface of the substrate. Miller indices (h,k,l) are used to indicate the different crystallographic planes. When the plane is parallel to a certain direction, the value of the Miller index for that direction is 0. The <100>, <110> and <111> planes are illustrated in figure 13. The a1, a2 et a3 axes are defined referring to the cubic perovskite phase. The a3 axis correspond to the c axis when the PZT is in the tetragonal phase. 29 Figure 13: Drawings of (100), (110) and (111) plains [8] When the PZT is grown on a non epitaxial substrate, a polycrystalline phase with different amount of the above planes is obtained. This means that the deposit is formed by mono-crystalline grains orientated toward different crystallographic planes. When a epitaxial substrate like Platinum is used, it''s possible to grow crystals epitaxially, obtaining in this way films with crystallographic planes preferentially oriented toward a mono- direction. The mechanism of crystallization is controlled by the nucleation of piezoelectric grains. It means that the energy required for the nucleation of a grain must be higher than the energy required for its growth. This balance of energy defines the way it grows, and therefore also the preferential orientation (texture) the lattice is going to assume. The (111) is usually stiffer than the (100) direction and it has lower piezoelectric coupling coefficients. Moreover, the (100)-oriented PZT shows a weaker non-linearity. This is probably due to the different role that the lattice boundaries play for the two configurations. Recent studies show that the domain-wall contributions to the piezoelectric properties for the (100) PZT are much less significant than for the (111). The growth of (100) or (111) plane depends essentially on kinematics and thermodynamics of grains growth. The thermodynamics of the phenomenon promotes the growth of PZT (100), because it is the more stable phase. On the other hand the kinetics tends to minimize the interfacial energy which depends on the substrate properties and composition. For this reason, if a substrate which lattice constants match the (111) PZT phase, the growing layer will tend to be oriented according to that direction. It is clear now that the under- layers play a fundamental role in the definition of the final PZT -silicon stack [7].
30 3.2 Thin films deposition methods: Although many applications of electrical ceramic materials are found in the form of bulk or thick films, an increasing need for thin film ceramics to meet the needs of new devices has been apparent in recent years [3]. The process of depositing PZT as a thin film involves many common practices adopted from bulk and thin film production techniques. Commonly, precursors are used in either in gaseous (CVD), solid (PVD), or liquid (solution deposition) form to react or combine to produce a thin film on a substrate. The main approaches for thin film PZT processing can be categorized into the following categories 1- Physical vapor deposition (PVD) 2- Chemical solution deposition (CSD) PVD techniques require medium vacuum, usually between 10 -2 and 10-6 torr, to increase the mean free path of the atoms ejected from the target in order to obtain a sufficient flux of ions capable of depositing onto a substrate. Sputtering, evaporation and laser ablation are examples of well-established PVD techniques. Among the advantages of the PVD techniques there are high purity, cleanliness, potential compatibility with semiconductor integrated circuit processing and epitaxial/single crystal film growth is possible. However, these are certain drawbacks as slow deposition rates, and difficult stoichiometry control in multi- component systems where evaporation or sputtering rates differ considerably, moreover high temperature post deposition annealing is often required and high capital equipment acquisition and maintenance costs are necessary. Ion-beam assisted evaporation and sputtering are potentially capable of increasing film uniformity and increasing deposition rates. In general, laser ablation, a technique which is similar in concept to flash evaporation, and is based on the creation of jet of atoms from the target material using an ultra high-energy laser beam seems advantageous for congruent transfer of target material on the final substrate, however, the control of the final composition and morphology is not that easy due to the different reaction rate of the elements to be deposited. The CVD techniques are usually characterized by higher deposition rates, good stoichiometry control, large area pinhole free films and lower initial equipment costs. However, the limited availability, stability, and toxicity of the precursors for the ferroelectric compositions has posed a serious, technical challenge. Another method that has been quite extensively employed for producing ferroelectric thin films is chemical solution deposition. Chemical solution deposition has the advantages of precise composition and thickness control in multicomponent system, as lead zirconate titanate, ease of process integration with standard semiconductor manufacturing, process simplicity, and much cheaper equipment costs [3].
31 3.2.1 Physical vapor deposition: By using a solid source and heat, plasma, ion-beam, or laser to vaporize the solid target, a thin film can be deposited onto a substrate. Although this technique has been used often in the production of metal films, a growing number of processes are being modified and developed to deposit ceramics, especially stoichiometric multicomponent films like PZT. Physical vapor deposition (PVD) techniques used in ceramic processing include ion-beam, magnetron, and rf-diode sputtering; thermal, electron beam, and flash evaporation; and pulsed laser ablation. Sputtering is an ablative plasma assisted method operated in high vacuum. It utilizes a inert gas ion bombardment, very often Ar+ ions. The Ar+ ions are generated by cold plasma. The sputtering relies on the transfer of physical momentum and kinetic energy from the incident particle (ions) to the surface target atoms. The ion bombardment leads to structural changes in the near surface atoms and/or clusters of atoms by breaking the bonds and dislodging the surface atoms. Due to a considerable kinetic energy of some eV the sputtered atoms and ions move away from the target and condense on a substrate, located opposite to the target at small distances. Plasma discharge can be realized by DC or RF electrical field. The advantage of RF sputter sources is the possible deposition of non-conducting and semiconducting materials. The deposition rate is reduced significantly and cannot be compensated by higher electrical power supply. DC power supplies are able to provide more plasma energy. [1] Plasma sputtering deposition (PSD) is practical for both commercial and research environments, but often problems of thickness and compositional uniformity from using bulk targets alter the process. Also, extreme sensitivity to deposition conditions and geometry result from the nature of the physical configuration. Depending on the process conditions a high amount of defects like grain boundaries, pores and dislocations, vacancies, impurities and significant residual stress can be generated in the deposited thin films. As is the case for general PZT processing, the formation of the perovskite phase is greatly influenced by the incorporation of Pb. With PSD, the type of system can influence the transport of Pb to the surface depending on the gas pressure. Another PVD method, ion-beam sputtering deposition (IBSD) uses multiple ion beams directed a single elemental target, for example Pb, Zr, and Ti for PZT, to counteract preferential sputtering of multi- component oxides. Thus, control of the film stoichiometry stands out as this is an often a drawback for many vapor-based deposition processes. Although IBSD has lower deposition rates than PSD, some advantages include lower pressure during deposition and the production of smoother films. This technique is on the other hand quite difficult to control and quite expensive. Thermal evaporation is also a vacuum-based PVD deposition technique in which a metal is heated to generate a metal vapor that can condense onto a substrate surface inside a vacuum chamber. For fabrication of high quality thin films a collision-less flight of particles has to take place from the vapor 32 source to the substrate, so a vacuum of better than 10-5 Pa is required. The temperature needed is a function of vapor pressure of the material [2]. Primarily still a research tool, pulsed laser ablation deposition (PLAD) is good for rapid exploration of novel chemistries and has been shown to produce single phase PZT at temperatures from 600-700°C. It is most widely used for its ability to deposit a variety of ceramic films [12]. Exposing targets to an high intensity laser beam creates a plume of material from the target, providing an evaporation source and transferring the target atoms to a heated substrate where these atoms can be incorporated in the substrate forming a thin film with a composition similar to the target. This method is good for multicomponent system like PZT due to the accurate transfer of the target stoichiometry to the film in the case of oxides. Complexity arises in the spatial and time dynamic evolution of the ablated plume, and there are problems of conformal and large-scale uniform coverage. For PZT, the presence of Pb may cause contamination problems during fabrication [11]. It is a very quick and expensive method due to equipment like laser and geometry-dependent chambers, and hence limited to research laboratories and not suitable for industry [1].
3.2.2 Chemical vapor deposition: Chemical vapor deposition (CVD) processes differ from PVD in the fact that the deposition is due to chemical reactions of the precursor gases that react on the surface of the substrate. Typically, this is a thermally driven process because using heat energy to activate the reactions. In the case of depositing PZT, a specific kind of CVD employed is metalorganic CVD (MOCVD) distinguished by the generally high volatility of the precursors permitting lower temperature processing [12]. Unlike PVD processes, stoichiometry in MOCVD is not such an issue due to the ability to precisely control the flows of the gaseous precursors and so the amount of precursor reacted. This make it a suitable candidate for the preparation of PZT and other ferroelectric oxides films [12]. In many cases of CVD processing, the modification of using glow-discharge plasmas enables low temperature deposition of films. Some drawbacks include complex metal precursor selection, (typically metal alkoxides) and lower deposition temperatures. Further complexities arise from the precursor delivery technique. A proposed option called liquid-source injection produces a viable, but expensive, alternative technique for multicomponent systems. Even with the decrease in processing temperature, the high costs override as a major drawback to use such methods. 33 3.2.3 Chemical Solution deposition: CSD processes, mainly including metallo-organic deposition (MOD) and sol-gel process, have been the subjects of considerable research interest relative to the deposition of thin ceramic films like PZT. Fukushima et al. in 1984 published their work on metallo-organic deposition (MOD) processing of thin films of PZT and right after that, Budd et al [10] published their work on the sol-gel processing of thin films of PZT. These pioneers works demonstrated that it is possible to obtain the desirable properties of bulk perovskite materials in thin-film form. The early work by these investigators and others led to a rapid expansion of research in this area [2]. MOD and sol-gel process differ in the chemistry in the solution. They afford the potential to control both chemical composition and microstructure to the levels required in electronic materials processing applications. In addition, experimental evidence indicates that temperature-time requirements for the subsequent crystallization/densification processing can often be minimized due to the high degree of chemical homogeneity in the polymeric ceramic precursor systems. The general principle of CSD involved in the solution deposition of ceramic films is to prepare a homogeneous solution of the necessary precursors dissolved in an appropriate solvent, and later to spread the solution onto a substrate using spin coating or dip coating. The homogeneity of solution is one of the most important parameter in assuring a high quality thin film, because it prevents the formation of nano- island or nano-clusters. Other important parameters that will define the solution precursor properties are the precursor size, structure, shape and rate of reaction. The starting reagents are dissolved in a common solvent. The selection of the starting reagents, depending on the solution routes, is dictated by solubility, accessibility and reactivity considerations of solution precursor species desired. Similar consideration have to be done for the choice of the solvent. For depositing the solutions onto substrates, the basic process involves substrate dipping or spin coating followed by some baking or pyrolysis, conducted at relatively low temperature (100-300°C), and crystallization, carried out at high temperature (550-700°C for PZT) in either a standard conventional furnace or better in a rapid thermal annealer (RTA). The use of the latter increases the final films quality and reduces the possibility of film cracking, as discussed hereafter. Among the processing methods available to produce thin PZT films, solution deposition stands out as the most adaptable and versatile as it can be used in the production of thin and thick films, tailoring such parameters as chemistry and thickness fairly simply and covering large surfaces easily and with a low cost. The deposition is divided into two parts, the solution synthesis and the film deposition. Prepared solutions can be easily made in-house or commercially purchased. More details in the preparation of the precursor solution are in the next chapter. 34
3.3 The Sol gel route
A colloid is a suspension in which the dispersed phase is so small that gravitational forces are negligible and the interaction are dominated by short range forces, such as Van der Waals and surface charges. The inertia of the suspended phase is so small that it exhibits Brownian motion [39]. A sol-gel is a colloidal solution of metalorganic [M(OR)n] or metalorganic-oxygen- metalorganic [(OR)n-1M-O- M(OR)n-1] molecules which are linked into a -O-M-O network. The solution is formed by controlled hydrolysis and condensation reactions of metal akoxides. M(OR)n + H2O '' OH - M(OR)n-1 + ROH (hydrolysis)
M(OR)n-1 - OH + OR - M(OR)n-1 '' M(OR)n-1 -O - M(OR)n-1 + ROH (Condensation) M(OR)n-1 - OH + HO - M(OR)n-1 '' M(OR)n-1 -O - M(OR)n-1 + H2O In the equations above, R represents an alkyl radical, such as methyl (CH3), ethyl (CH2CH3), propyl (CH2CH2 CH3) or others .The hydrolysis process consists of replacing one OR group of the alkoxide with an hydroxyl ion (OH), releasing an alcohol molecule in the process as second product. Condensation is the reaction between two partially hydrolyzed molecules [(OR)n-1M-OH] or one partially hydrolyzed and one alkoxide [M(OR)n] to form a M-O-M link with a (OR)n-1 group linked to each metal atom. By definition, condensation produces as reaction product either an alcohol or a water molecule, depending on the initial compounds. If the hydrolysis and condensation reactions of the metal alkoxides go unchecked or are not correctly equilibrated, macro-molecules containing many M-O-M segments will form and depending on the reaction speed, precipitates or extend polymeric networks throughout the solution will eventually form. As reported in literature, when macro-molecules or polymeric networks form in a liquid solution, its viscosity increases in relation to the amount of these aggregates until it reaches the gelation point, that is the point in which the liquid solution become a gel. Thus, if the hydrolysis and condensation reactions in a sol-gel are not well controlled, the solution will become a highly viscous or solid gel. The overall process is called gelation. Precipitation of a single reagent can also occurs when the reaction rates of the reagents in a complex solution have different speed and so the condensation reaction for a single element occurs faster, leading to differential precipitation. These reactions can take from less than one second to weeks or month depending on temperature, on pH, on the reaction speed, on the initial concentration of the reagents and on the evaporation tendency of the solvent. Moreover the speed of the reaction can be altered by other reactants which inhibit the gelation process and prevent the solution precipitation. The use of elements, 35 called chelating agents, which slow down the hydrolysis / condensation reaction will be discussed later in the experimental section. The hydrolysis and condensation reactions are reversible. If these reactions reach the equilibrium, so the speed of the two is the same, the solution will never gel in a sealed container. Such solution is called a stable sol-gel. By varying the amount of water/alcohol, it is possible to control the speed of each reaction and shift the equilibrium in both directions. If one were to remove alcohol from a stable sol-gel, this would decrease the speed of the reverse reactions, and would shift the equilibrium toward hydrolysis and condensation. If enough alcohol is removed from the solution, it will eventually gel. This is actually desired for the fabrication of thin films. If the chemical reactions are controlled properly, the solution will become a structured liquid containing a matrix of metalorganic chains. When the sol-gel is spread out over a surface (either by dip coating or spinning), the solvent evaporates, the equilibrium of the reaction is shifted towards hydrolysis and condensation and the solution gels, creating a branched polymeric network into the thin film, that is rigid enough for handling. Controlling the evaporation rate of the solvent is possible to produce also Aerogels, that is a suspension of particles in a gas and it forms when the solvent is evaporated in supercritical conditions, and Xerogels, that is a solid with high porosity that forms from a gel when the solvent is extracted slowly with unhindered shrinkage. If the drying of the solvent is followed by heat treatment, it is possible to create dense films or dense ceramic. These procedures are summarized in Fig.14. Figure 14: Overview of the sol gel process. [39] The thermal treatment for the production of PZT films, once that the solution has been spinned on the wafer, can be divided into three steps. The first step is the drying of the film. This is usually accomplished on a hot plate in air, at temperatures ranging from 100°C to 200°C depending on the composition of the film. During this stage, the excess solvent is eliminated and the film shrinks and densifies as it loses liquid. After drying, the film is much denser, and any remaining liquid is held in by capillary action. 36 After drying, the as-deposited film is an amorphous gel of metalorganic compounds. In order to eliminate the organic components and the residual liquid, the film is pyrolysed, which involves heating in the range of approximately 300-500°C to burn the organic compounds off. This step is often referred to in the literature as the "firing" stage because in this stage the organic component is carbonized and oxidized. After this stage the film is amorphous but with the PZT composition. It is quite flexible, mechanically tough, but capable of being patterned by photolithography with an appropriate etch. After the firing of the film, the amorphous metalorganic gel should often be converted, as in the case of PZT, to a crystalline layer of inorganic metal oxide compounds to exploit the material properties. This step is often referred to as annealing and is a thermal treatment performed at high temperature (550°- 700°C for PZT) that guarantees the elimination of all the organic traces and the crystallization of the film. The temperatures used to crystallized thin films are much lower than the sintering temperatures of the same bulk material made from ceramic powders and this is a great benefit when planning an industrial procedure for oxides film production. This final step is usually performed using a rapid thermal annealer (RTA), which produces films with better properties and reduces cracking, but can be done also in a conventional furnace. Figure 15: Firing schedule example for PZT sol gel.
37 Chapter 4: Experimental procedures
4.1 Precursor solution synthesis
4.1.1 State of the art: During years lots of different routes to synthesize PZT precursor solutions have been studied. Changing the precursors type for PZT synthesis inevitably will impact the materials properties and, at the same time, each route requires significant processing changes as crystallization temperature and film thickness. Typical precursors to prepare a Lead Zirconate Titanate sol-gel are metals alkoxides with the general formula M(OR)n where M is a metal atom and R is an organic ligand. Commonly as metal precursor Lead acetate tri- hydrated, Zirconium n-propoxide and Titanium n-propoxide are used. These compounds posses a great tendency towards hydrolysis and condensation and are strongly moisture sensitive thus, during preparation, preventative measure must be adopted. What change among the preparation routes, is mainly the type of solvent used, which will impact strongly on the homogeneity of the final solution because it changes the hydrolysis/ condensation tendency of the metal-alkoxide precursors. The solvent actually acts, beside as a solvent, as a chelating agent, forming complexes with the metal precursors and modifying their reaction pathway. The choice of the solvent is also dictated by solubility consideration, toxicity, viscosity, surface tension and boiling point. Alcohol is a common choice of the solvent, since the alkoxides tend to react with all hydroxy compounds with the replacement of alkoxy group. These reaction can be written as: M(OR)n + x R''OH '' M(OR)n-x(OR'')x + xROH
Two main routes for the synthesis of PZT precursors solution have been developed during years: the 2-Methoxyethanol (2-Me) based solution and the carboxylic acid, acetic acid or acetylacetone, route. Other recipes that combine the advantages of the two routes ( the high quality of 2-Me as solvent and the benefit of the chelating effect of carboxylic acids on metals alkoxide) are present in the literature and are based on a mix between the two route [25,26,27,28]. 38 4.1.2 2-Me based solution The first route, that has became the most widely used for the production of PZT films, was first developed by Gurkovich and Blum and developed by Budd et all. [10] in 1985. It''s based on the use of 2- Methoxyethanol (2-Me) both as a solvent and as a modifying agent, and Lead acetate tri-hydrated, Zirconium iso-propoxide and Titanium iso-propoxide as metals precursors. The lead acetate is first dissolved in 2-Me and, after refluxing and de-hydrating of the solution , Zirconium and Titanium iso-propoxide are added sequentially. The solution is finally mixed and refluxed to assure high homogeneity. Modification of this recipe, as the addition of Ethylene glycol or Glycerol to improve the mechanical properties of the final film or the use of Formamide as drying control agent, has been developed during years. The use of 2-Me assures high homogeneity of the precursors solution since, bridging with the metal alkoxides, 2-Me slows down the hydrolysis/condensation reactions , stabilizing the solution and almost eliminating the aging effects. The metal alkoxides, as reported in [3], become metal methoxy-ethoxide which are more stable due to the inductive effects and the steric hindrance of the methoxy-ethoxide ligands. The use of 2-Me as solvent also enhances the final crystallization of the film. This has been attributed to the smaller colloidal molecules of the 2-Me sol which are a direct result of the chelating effects [5]. Due to all these advantages and to the simplicity of the preparation, this route has become the most widely used. The major drawback of the 2-Me is its high toxicity and for this reason during years lots of research has been done to develop a precursor synthesis route based on the less hazardous elements. 4.1.3 Carboxylic acid based solution

This route uses less toxic agents compared to the 2-Me solution and it was first initiated by Yi and Sayer at Queen''s University in 1988 [11,12,13] and then developed by many researchers. This route is based on the use of carboxylic acids, as acetic acid or acetylacetone, as chelating agents and water or alcohol as solvent. It is often referred to as ''chelate' process because the use of carboxylic acids, such as acetic acid or acetylacetone, impacts heavily on the molecular modifications of the precursors with chelating ligands. Typically, the carboxylic acid groups coordinate to the metal species in a bidentate fashion for Zirconium and in a monodentate fashion for Titanium as reported in [11] and [13] , and frequently it acts as bridging ligands, linking metals together to form oligomers and small polymers. Chelation of the metal alkoxides by these carboxylate groups, results in the formation of precursors that possess reduced sensitivity toward hydrolysis in respect to with the unmodified chemicals, as reported in many literatures. Reducing the tendency of metal alkoxides towards hydrolysis is of fundamental relevance, since when titanium iso-propoxide and zirconium propoxide are added into a solution of water or alcohol, precipitates often form due to their high hydrolysis and condensation tendency. In order to 39 prepare a useable sol or gel, titanium iso-propoxide and zirconium propoxide must be chemically modified to change their hydrolysis and condensation behavior. The gelation behavior of titanium and zirconium alkoxides can be modified by acetic acid, and this has been attributed to the formation of chelates or to the nucleophilic substitution of a propoxide group with an acetate group that is less sensitive towards hydrolysis [11]. While chelate processes are simple and rapid, the chemistry involved in solution preparation is quite complex due to the number of reactions that occur. Key reactions were found to be chelation, esterification, hydrolysis and condensation. The complexity of the reactions results in a diminished ability to control the precursor structure compared to true sol-gel approaches, and thus the gain in process simplicity thus comes at a cost [2]. Another drawback of chelate processes is that continued reactivity in the precursor solution after synthesis, can result in a change in precursor characteristic over time (weeks to months) and thereby a degradation in film properties [13]. This occurs because constituent groups such as acetate, even if less susceptible to hydrolysis than alkoxy groups, may still be attacked by water, resulting in a change in the molecular structure, and other reactions result in continued oligomerization of the chemical species, eventually causing precipitation [2]. Yi and Sayer in their work chose as carboxylic acid the acetic acid. The basis of this sol-gel rests with the property of acetic acid to slow the hydrolysis and condensation reactions of transition metal alkoxides by forming more stable metal alkoxo-acetylates. The original solution is prepared by dissolving lead acetate trihydrate (Pb(CH3CO2)2 · 3H2O) in acetic acid (CH3COOH). This is followed by de-hydratation of the solution since water may cause non uniform gelation of titanium isopropoxide and zirconium propoxide. The zirconium propoxide Zr(OCH2CH2CH3)4 (70 wt% in propanol), and the titanium isopropoxide [Ti ((CH3)2CHO)4] are then sequentially added to the acetic acid / lead acetate mixture to form the sol-gel after than the solution was cooled to room temperature. In this type of sol-gel, the order of metal alkoxides addition has a fundamental importance because of the different hydrolysis / condensation reaction speeds of these compounds [11,12,13]. The higher reaction speed of the titanium iso-propoxide leads to the formation of large titanium alkoxide chains which create local in homogeneities, or in some cases, the formation of precipitates in the solution when it is the first metal alkoxide added to the acetic acid / lead acetate mixture. To prevent this phenomenon, the zirconium propoxide is usually added first and allowed to react with the acetic acid / lead acetate mixture to form Zirconium propoxide di-acetate before the addition of the titanium iso-propoxide. Water is added to the solution to provide excess reagent to the hydrolysis reaction and to control the viscosity [5]. Exothermic reactions between the acetic acid and the metal alkoxides warm the solution during its preparation [11]. As it cools down, the lead acetate re-crystallizes in the solution due to the common ion effect with the excess acetic acid [5]. Lactic acid added to the sol-gel before cooling prevents the 40 appearance of these crystals by reacting with, thus effectively removing, the excess acetic acid by a process known as esterification. This assure high solution stability over a long period. In order to improve the drying behavior of the film and to avoid cracking Ethylene Glycol and Glycerol are finally added [11,12,13]. A flow chart of the whole process is represented in Fig. 16. Figure 16: Flow chart of Yi and Sayer recipe [11] A single crack free layer of about 0.5 µm thickness can be prepared with this solution by spin coating. Thinner films were prepared by adding some combination of water and propanol to dilute the solution. It has been demonstrated that lowering the concentration of the solution helps to produce crack free films. The addition of propanol also lowers the surface tension of the solution and can improve the wettability for some substrates. Thicker films were built up by multilayer coating.
41 4.1.4 PZT precursor solution synthesis In this work the choice of PZT precursors and solvent has been done considering first of all the toxicity of the elements used and their cost, and after all the possibility to extend the process to the industry. Moreover, the simplest synthesis route, therefore the process that requires less steps, has been considered to be the most convenient. Following these considerations the using of 2-Me as solvent has been discarded due to its high toxicity, thus the choice turned to water or 1-Propanol. As metal precursors the commons Lead acetate trihydrate (Pb(CH3CO2)2 · 3H2O), Zirconium propoxide Zr(OCH2CH2CH3)4 (70wt% in 1-propanol) and Titanium iso-propoxide (Ti[OCH(CH3)2]4) by Sigma Aldrich have been chosen since these elements are the most used in the literature and their behavior is well known. These chemicals are mixed with a molar ratio of 1.1/0.52/0.48, to obtain a final deposit that match the target stoichiometry of Pb1.1 Zr0.52 Ti0.48 that is the ratio of Zr/Ti at the morphotropic phase boundary between the tetragonal and rhombohedral phase, where the PZT is expected to show the higher electrical properties as mentioned before. The 10% excess of Lead in the formulation has been chosen to compensate the Lead evaporation during the heat treatment. The lead loss, that results in a lack of lead in the final deposit, favors the growth of the non ferroelectric pyrochlore phase and so this loss must be prevented tuning the initial formulation. Moreover Pb excess is generally believed to favor perovskite formation [8,30] in PZT films during annealing and most of the authors used an enriched lead formulation to deposit PZT films. Even though, due to their high tendency toward hydrolysis, is suggested to use these chemicals in a glove box or in a dry nitrogen atmosphere, for simplicity the synthesis process has been performed in the normal laboratory atmosphere, paying great attention to the absence of water or moisture in all the flasks and instruments used during preparation. Different solutions have been prepared following both Yi and Sayer recipe [9,11,12,13,14] and 1-Propanol based methods [1,17,18,20,21]. In both those recipes , acetic acid (CH3COOH) has been chosen, among the possible carboxylic acids, as chelating agent. When acetic acid is present in solution with the metal alkoxides, this chelating reagent modifies the metal alkoxides into a mixed alkoxide acetate derivative substituting the propoxide groups for Zirconium and the isopropoxy groups for Titanium, and these will form 1-propanol and isopropanol in solution. Yi and Sayer proposed that, due to steric hindrance, it''s impossible for all the four surrounding alkoxides groups around each metals to be replaced by these chelating ligands but acetic acid would substitute for only two propoxy groups of a zirconium propoxide molecules and for one isopropoxy group of a titanium isopropoxy molecule [11]. The maximum replacement is therefore two. The as formed metal alkoxides acetate possess a lower tendency towards hydrolysis compared to the previous metal alkoxides 42 because these latter are expected to have fewer functional groups, since the acetate group has low tendency toward hydrolysis and is quite stable. As proposed by Q. Zhang et al. in [23], after groups substitution, molecules would undergo through hydrolysis and condensation reactions even though with a slower rate compared to the unmodified metal alkoxides. Due to the way in which the acetate group substitute the alkoxides before hydrolysis and condensation occurs, it''s more likely that, due to condensation, chain-like molecules would form instead of spherical particles, as proposed in [23] and illustrated in Fig.17. Here M is the metal atom (Zr or Ti), R is the alkyl group (propoxide or isopropoxide) and R'' is tha acetate group CCH3. Moreover, as already discussed, the acetate group posses lower tendency toward hydrolysis compared to the alkoxide group and this fact prevents the solution from excessive condensation and subsequent gelation. The solution, anyway, undergoes with time to a certain degree of polymerization but this can be controlled quenching the solution with acetic acid or excess solvent, preventing in this way excessive condensation. The effect of adding excess solvent is of reducing the collision rate of the molecules, increasing the mean free path between collisions and also to further substitute the metal ligands, as in the case of water, and form ionic species. When water is added into a solution of acetic acid, titanium isopropoxide and zirconium propoxide, several reactions will occur. Since water is a stronger nucleophile, the isopropoxy groups of the titanium isopropoxide acetate and the propoxy groups of the zirconium propoxide acetate are removed by water to form isopropanol and 1-Propanol. The acetate groups of the titanium isopropoxide acetate and the zirconium propoxide acetate are also removed by water to form Figure 17: Hydrolysis and condensation reaction for substituted metal alkoxides [23] 43 hydrated acetate anions AcO - since the Ti-OAc and the Zr-OAc bonds are highly polar or ionic in nature and will decompose in water. Therefore, hydrated titanium hydroxide and zirconium hydroxide cations are formed in the solution. Depending on the pH, so on the amount of acetic acid, the hydrated titanium hydroxide and zirconium hydroxide cations may further react with water or with hydrated protons. The charge density of the polymeric particles also depends on the pH of the solution. This is due to the following equilibrium since the surface of the polymeric particles are covered with hydroxy groups and hydrated water: Under acidic conditions, the equilibrium is shifted to the right and the hydrated titanium hydroxide and zirconium hydroxide cations are further positively charged. The repulsive interaction between the positively charged cations prevents them from approaching to each other and the positive charge of the cations reduces the lability of the bonded water, making the condensation reaction more difficult. Therefore, the condensation reaction of the cations is retarded and the stability of the solution on storage is high.[13] Following this observation is possible to assume that using acetic acid as chelating agent in a mixture of metal alkoxide molecules, it''s possible to obtain polymeric chains of the metal precursors in an acetic acid/ propanol or water solution and thus, what is called a stable sol gel. The amount of acetic acid in solution has a strong impact on the final material properties and particles size, since it is responsible for the metal alkoxides groups substitution, pH control and thus solution stability over time. During this work, solutions containing different amount of acetic acid were prepared. The [AcOH] / [Ti+Zr] molar ratio has been changed from 2, that is the minimum amount of acetic acid necessary to substitute for two propoxy group of Zr and one iso-propoxy group of Ti accordingly with Yi and Sayer works [9,11], to a maximum of 15. When a very little acetic acid is present, [AcOH]/[Ti+Zr] = 2, the metals alkoxides precipitate over time creating a milky dense solution. This is probably due to the shortage of modifier, caused also by acetic acid evaporation during mixing and storage , that leads to hydrolysis and condensation of the metal alkoxides during time. The presence of moisture from the atmosphere and from the not completely un-hydrolized Pb acetate are also causes of sudden condensation and precipitation during solution preparation. On the other hand, when a large amount of acetic acid is added during the alkoxides mixing the solution still goes towards precipitation. In Q. Zhang et al. in [23] work is suggested that a [AcOH]/[Ti+Zr] molar ratio of 5 is enough to avoid the solution gelation and adding more acetic acid won''t influence further the solution evolution and the particle growth over time. This is in agreement with the conclusions of Yi and Sayer works [9,11,12,13]. Following these 44 considerations a [AcOH]/[Ti+Zr] molar ratio of 5 has been chosen as mixing ratio. Two different routes has been adopted to obtain a mixture of the metal alkoxides. The first one approximately follows the process described by Yi and Sayer and illustrated in Fig.17. The lead acetate tri-hydrated was mixed with acetic acid with a [AcOH]/[Pb] molar ratio of 5 and heated up at 110°C under vigorous stirring. The solution was thus de-hydrate for at least 30 min at constant temperature. After that, the solution was cooled to room temperature and the right amount of Zirconium propoxide was added. As already mentioned the PZT target stoichiometry is Pb1.1 (Zr0.52 Ti0.48). If the solution is not cooled down before adding the Zirconium propoxide, this will undergo gelation after few minutes. After at least ten minutes mixing, an adequate amount of Titanium isopropoxide was added and the solution was mixed for 30 min before the addition of water. Heat was evolved after that Titanium precursor was added and that is due to the ligands exchange process between metals alkoxides and acetic acid. A golden yellow homogeneous solution of PZT metals precursors was thus obtained. When the solution came back to room temperature de-ionized water was added with a molar ratio [H2O]/[Zr+Ti] of 15 to dilute the solution and to make it more stable. To prevent metals precipitation over time Lactic Acid was added following Yi and Sayer considerations and to increase the mechanical film properties and the drying behavior during thermal treatment, Ethylene Glycol was added with a [PZT]/[E.G] molar ratio of 1. The solution was then stirred for one night and the PZT molar concentration was finally adjusted to 0,5 M adding de-ionized water. After pressure filtering using a 0,45 µm filter membrane, the solution was stored in a sealed container and it showed stability over long periods. A second route to obtain a sol gel solution of the PZT precursors is based on the use of iso-propanol as solvent and acetic acid both as solvent and modifier to control the hydrolysis rate [1,18,21]. Lead acetate trihydrate was mixed with isopropanol with a ratio of 10g:1ml and the solution was refluxed at 110°C for one hour to remove water. The solution was then naturally cooled to 80°C and acetic acid was added with a [AcOH]/[Pb] molar ratio of 4. After mixing, Zirconium propoxide was put into the solution and that was refluxed at 80°C for one hour under vigorous stirring. Finally Titanium Isopropoxide was combined with the previous solution and it was refluxed for other two other hours at 80°C to assure complete mixing. An adequate amount of acetic acid is then added to the PZT precursor solution to obtain a final [AcOH]/[Ti] molar ratio of 25. Isopropanol was then added into the previous solution until it was diluited to 0,5M. Ethylene glycol was finally combined with the precursors solution with a [PZT]/[E.G] molar ratio of 0.68 to start the hydrolysis reactions. The right amount of Ethylene Glycol was chosen following the considerations done in previous works *17+. It''s important that the E.G in solution is enough to start the hydrolysis reactions but its concentration has not to be too high , to prevent excessive condensation of the metals precursors that lead to precipitation and phase separation. After almost six hours mixing, the solution was filtered using a 0,45 µm filter membrane and then it was stored in sealed glass container to prevent solvent evaporation. Also this solution showed stability over long periods. A flow chart of the PZT precursor solution preparation process for the isopropanol based method is illustrated in Fig.18. 45 Figure 18: PZT precursor solution synthesis flow chart
46 4.2 Substrate influence on crystallization mechanism
The choice of the substrate for PZT thin film deposition has fundamental relevance on the morphological and electrical properties of the final deposit, as the presence of cracks, the density of the film and the ferroelectric and piezoelectric features . Contrary to the bulk PZT fabricated using sintering process, which crystals don''t possess a preferential orientation and are randomly oriented, for PZT thin films the substrate strongly influences the growth of PZT grains and their final crystalline planes direction. The dielectric, ferroelectric and piezoelectric properties are strongly influenced by the preferential crystallographic plane orientation and on the density of the film, so the relevance of the substrate on the final properties of the deposit is obviously fundamental. 4.2.1 Crystallization mechanism: The PZT films are built on a substrate and during annealing these films transform from an amorphous state to a crystalline one. This process starts from the substrate and it''s principally governed by the nucleation of crystalline grains. This means that the energy needed for the grain''s nucleation is bigger than the energy required for their growth. The nucleation is governed by many factors as the texture of the substrate, its residual stresses and composition. All these factors can lead to crystals growth with different preferential orientations. The grain growth is governed by the competition between the kinetics, that promotes the (111) or (110) planes growth when the substrate properties favor it, and the thermodynamic, that promotes the (100) planes growth, since it is the most thermodynamically stable. For this reason if the substrate is oriented (111), as in the case of Pt, the PZT will tend to be oriented according to the same direction because the kinetics tends to minimize the interfacial energy between the substrate and the deposit, which depends on the substrate properties and composition. On the other hand the thermodynamic favors the (100) plane growth and so the final PZT orientation will depend on which of the two mechanism prevails. The grains growth process depends also on the number of nucleation sites; if these sites are numerous, the kinetic will govern the process, if there are just few sites the thermodynamic will be predominant [8]. Moreover has been noted that when the kinetic governs the growing process, as in the case of (111) plane growth, the grains dimension is smaller compared to the grains of the (100) PZT. This fact demonstrates as the kinetic governed growth is more rapid and is favored by a large amount of nucleation sites, while the thermodynamic controlled growth is slower and favored by a small amount of nucleation sites, that also permits the growing of larger grains. The firing temperature, that influences the stability of the pyrochlore phase, is another parameter that has a certain relevance on the grain growth mechanism because it makes the phase transformation easier or not. In the first case the kinetic will be 47 favored, in the second the process will be slower, favoring the thermodynamic. Other factors influence the growth mechanism, for example the substrate properties before sol gel spinning and the annealing procedure. Using and rapid thermal annealer the crystallization process is in fact very quick, and the kinetic of grain growth will be predominant in respect to the thermodynamic. This will promote the formation of the (111) planes instead of the (100). 4.2.2 Choice of the substrate During years different types of substrate have been used to fabricate PZT films. Some of these are summarized in the table below with their behavior: Substrate materials Stannic Oxide Medium resistance Indium tin oxide (ITO) Medium resistance Ti/ Platinum Good and reproducible Aluminum Temperature limited Silver High adhesion Stainless steel Good at low T° Ni, Co, Ti, W Oxidized Nickel based alloy Acceptable Ruthenium Oxide High conductivity Yttrium Copper Barium Oxide Lattice matched W,Ti,Pt silicides Some success As noted earlier, the definition of a good electrode material is not trivial. It has to survive processing in oxygen without degradation or the development of unacceptable interface layers and the thermal expansion difference between the electrode and the film must be lower as possible. The metallization has two roles to play with respect to the film: it must nucleate the required phase and provide adequate conductivity for device purposes [29]. Many metals such as Ni, Co, Ti and W oxidize rapidly under oxygen processing. Aluminum is often acceptable, although an interfacial film can nucleate a non-piezoelectric phase of PZT and care has to be taken to maintain the processing temperature below that of the Al-Si eutectic temperature which is around 480°C [29]. Platinum electrodes with a 0,01 µm thick titanium adhesion layer are most often reported as having the highest reliability . In both cases, the quality of the metallization is of great significance and low density, porous deposits can lead to irreproducible crystallization conditions and interfacial barriers. This often makes simple evaporation suspect and the deposition of metallic layer by sputtering onto heated substrates has been found to be the most reliable. Stannic oxide and indium tin oxide (ITO) are transparent electronic conductors and remain excellent prototyping electrodes for piezoelectric films. Such substrates are available commercially and provide good results at low frequency [29]. The reason for this is two-fold. The material remains conducting in an oxygen 48 atmosphere up to temperatures of 400°C, and it tends to promote crystallization of the required perovskite phase. On ITO coated glass the films usually growth with a (100) preferential orientation and with a less pronounced pyrochlore phase. The main disadvantage of ITO is that its n-type conductivity, associated with oxygen vacancies, reduces when the film is fired in oxygen. This can lead to interfacial diffusion and a limitation of the high frequency response [29]. For ferroelectric memory applications it''s the relatively high sheet resistance of the film which limits the upper frequency response of the device. This is less important at the frequencies normally employed in piezoelectric devices. The most used substrate is Platinum deposited on a SiO2/Si wafer. This material is chemically inert and possess a very low diffusion tendency. Moreover it can form a metallic-ferroelectric interface that promotes the formation of a perovskite phase and doesn''t reduce the electric properties of the film. Platinum is usually oriented towards the (111) direction and this can promote the nucleation of PZT with the (100) or (111) planes depending on parameters as the deposition temperature of Platinum and its residual stresses as reported in [8]. To enhance the adhesion of Platinum, often Ti is used as adhesive layer between the Silicon dioxide and the Platinum. As reported in [8], at the PZT annealing temperature, the diffusion of Titanium into the Pt and PZT layers is probable, but this can be beneficial for PZT grains growth. In any case, to reduce the Ti diffusion, the creation of an oxide layer on the top of Ti, or the direct use of TiO2 has been proposed. Certain conditions promote the nucleation and growth of grains with the (111) preferential plane direction, even if the most thermodynamically stable orientation is the (100), and one of the key points for this, is the substrate texture. Three main conditions can promote the (111) plane nucleation: the diffusion of Ti towards the surface of the substrate; the formation of an intermetallic phase of Pt-Ti or Pb-Pt; and the Pt residual stresses. It''s also possible to use a seeding layer on the top of the substrate to promote the nucleation of only one plane. Such seeding layer has the aim to promote the kinematics which assures the growing of the preferred orientation of PZT layer. The presence of Ti or TiO2 layers (even very fine ) will assure the growing of (111)-oriented PZT while PbTiO3 (PTO) layer, will promote the growing of (100) planes [7]. The PTO layer can be obtained both via sputtering and sol-gel deposition. Residual stresses in the Pt layer also have a great influence on the final PZT orientation, as mentioned before. When the Pt layer is subjected to a thermal pre-treatment, that consists in an annealing cycle at 500 -600°C before spin coating, residual stresses are created inside the layer. It has been demonstrated by Quin et al. [31] that the residual tension will promote the growth of the (111) PZT and a residual compression will favor the (100) planes growth. When the substrate is annealed before spin coating in a conventional furnace at 500°C the (111) plane growth is preferred. When the Pt layer is annealed in a rapid thermal annealer (RTA) at 500°C the PZT (100) will preferentially grow. In both cases a residual tension will be present in the Pt layer and it will be stronger in the RTA annealed sample. The residual tension will decrease the mismatch, and consequently the interfacial energy, between the (111) Pt and the (111) PZT and so the latter plane will easily grow. 49 When the residual stresses are more important, as in the case of RTA annealing, the lattice mismatch will be still large and the thermodynamically more stable (100) plane will preferentially grow because of the high interfacial energy between the (111) Pt and the (111) PZT plane. The use of a barrier layer is very important to avoid the diffusion of the substrate elements towards the deposited film. PZT films are usually annealed at temperatures higher than 550°C, temperature at which the perovskite phase begins to form, and at this temperature the risk of Lead diffusion into the Silicon and of Titanium diffusion towards the surface is very high. For this reason, usually, a silicon dioxide layer is thermally fabricated on the top of the Si wafer to prevent elements diffusion. Films may be prepared either by successively coating a substrate with repeated layers deposition with 0.2- 0.3 µm each in thickness to build up a final layer thicknesses of 1-2 µm , or by using single coats of up to 1 µm. The successful application of thicker coats depends on the nature and quality of the substrate, and on the effectiveness of additives designed to control the firing process. Thermal expansion mismatch with metallic substrates such as platinum tends to induce cracking, while conducting oxides such as indium tin oxide (ITO) have stronger adhesion. In this work PZT films of 0.2-0.3 µm thickness each have been deposited on three different substrates: ITO coated glass, Pt/Ti02/SiO2/Si and Au/TiW/SiO2/Si. For each substrate the morphology and electric properties of the final film have been studied. The electrical resistivity of ITO has been measured after and before the annealing step. As reported also in literature, the resistivity of ITO increases after thermal treatment due to oxygen diffusion. Just to have an idea of the change of resistivity, using a digital voltmeter an initial resistance of 55/60Ω and a final one of 270/300Ω has been measured. This obviously impacts on the electrical features of the final device.

50 4.3 Sol-gel deposition of PZT thin film
For thin film fabrication, dip coating and spin coating are typically used to coat sol-gel onto substrates. In this thesis, PZT thin films were fabricated exclusively using the spin coating process. Spin coating has been used extensively in the semiconductor industry because of the numerous advantages that it offers as low cost, great control of the precursor chemistry, possibility to cover large surfaces and good thickness on morphology control. This method also assures the highest ferroelectric and piezoelectric properties of the final thin layer. In this work the sol gel precursor solution has been spinned on three different substrates: ITO coated glass, Au/TiW/Si and Pt/TiO2/Si. Depending on the substrate, the solutions showed different adhesion and different behaviors when those were spinned on the wafer. This latter depends on the viscosity and surface tension of the solutions. Moreover the substrate influence on the final crystallization of the film has been discussed in the previous section. After spinning all the films were subjected to the same thermal treatment that is subdivided in three steps: drying, firing and annealing. 4.3.1 Substrate preparation and spin coating: To guarantee good morphology of the final film, first of all the substrates must be perfectly cleaned prior spin coating. This step is crucial since all the defects of the substrate will be transferred on the film creating in-homogeneities and cracks. The Pt and Au coated silicon wafers have been cleaned in an ultra sounds bath using acetone to eliminate all the organic contaminants and then quickly dried under strong Nitrogen flow. ITO coated glass substrates have been cleaned using alternatively two different solutions. The first was a Piranha solution and the substrates have been immersed into it for 15 minutes to assure complete cleaning and removing of dirt, and finally they were washed with de-ionized water and dried using Nitrogen flow. The Piranha solution is a mixture of sulfuric acid (H2SO4) and hydrogen peroxide (H2O2). In our case the volume ratio of thesechemicals was set at 7:3. Since the mixture is a strong oxidizing agent, it will remove most organic matter, and it will also hydroxylate most surfaces (add OH groups), making them highly hydrophilic. This can be useful to improve the adhesion of the water-based precursors solution onto the substrate and can have some relevance on the final films morphology. The second solution used was a mixture of Ammonia, Hydrogen Peroxide and de-ionized water, with a volume ratio of 5:5:1. Also this solution is a strong oxidizing agent and it will remove all the organic content onto the substrate surface. The substrates were immersed into this solution for 30 min at 80°C to assure complete cleaning and then, they were washed with de-ionized water and dried under strong nitrogen 51 flow. Finally, the absence of rifts and scratches on the surface of all the substrates has been checked before the spin coating step, to assure a good final film morphology. The sol-gel spin coating step can be divided into 4 main stages [2]. In the first stage the sol is dispensed onto the substrate Fig 19a. In the second stage the substrate is accelerated to its final desired rotational speed, Fig. 19b. This stage is usually characterized by aggressive fluid expulsion from the wafer surface by the rotational motion. Eventually, the fluid is thin enough to be completely co-rotating with the wafer and any evidence of fluid thickness differences has gone. Ultimately, the wafer reaches its desired speed and the fluid is thin enough that the viscous shear drag exactly balances the rotational accelerations. In the stage three, Fig.19c, the substrate is spinning at a constant rate and fluid viscous forces dominate fluid thinning behavior. This stage is characterized by gradual fluid thinning. Depending on the surface tension, viscosity, rotation rate, there may be a small bead of coating thickness difference around the rim of the final wafer. If the fluid thickness was initially uniform across the wafer, then the fluid thickness profile at any following time will also be uniform, leading to a uniform final coating (under ideal circumstances). In the fourth stage, Fig.19d, the substrate is spinning at a constant rate and solvent evaporation dominates the coating thinning behavior. Figure 19: Stage of spin coating :a) fluid deposition b) spin up; c) fluid spin off; d) solvent evaporation [5] As the prior stage advances, the fluid thickness reaches a point where the viscosity effects yield only rather minor net fluid flow. At this point, the evaporation of any volatile solvent species will become the dominant process occurring in the coating. At this point the coating effectively gels. As these solvents are removed, the viscosity of the remaining solution will rise [2]. The hydrolysis is allowed to occur since the film has been deposited onto the substrate to create a dense film . Moisture in the atmosphere is used to supply water for the reaction. The reaction should proceed at significant rate implying that gelation will occur in a short period of time. The final coating thickness depends on viscosity, type of substrate, surface tension, spin rate and solvent evaporation rate. 52 In this work the spin coating stage was divided in two steps. Initially the solution was spun for 8 seconds at 500 rpm with an acceleration of 100 rpm/s, then the spin speed was increased to 3000 rpm with an acceleration of 200 rpm/s and the sample was maintained at this speed for 40 sec. In this way during the first step the solution wets all the substrate slowly, reducing the possibility of striations and increasing the film homogeneity and then, at higher speed, the liquid film becomes thinner and denser along with time and solvent evaporation. When the spinning cycle finished the sample was immediately moved onto an hotplate to start the thermal treatment. With this procedure single films of 0,2-0,3 µm thickness have been successfully deposited onto all the substrate used. To create thicker films, multi-layers have been deposited sequentially after thermal cycle. Some defects that can occur during spin coating include comets, striations and wafer edge effects. Comets occur when relatively large solid particles impede the normal flow of the solution on the spinning wafer. The presence of comets can be reduced or eliminated by working in cleaner environments and by filtering solutions as part of the dispense process. Striations are radialy oriented lines of thickness variation in the as-coated film, Fig. 20a. Their occurrence is thought to arise because of evaporation driven surface tension effects. The early evaporation of light solvents can cause an enrichment of water and/or other less volatile species in the surface layer. If the surface tension of this layer is larger than the starting solution, then instability exists where the higher surface tension actually draws material in at regular intervals and the spaces in-between are more able to evaporate, and surface relief develops. Other defects, like concentrical circles, can form due to solution contamination with particle during spinning or drying steps Fig20b. The formation of defects during spinning implies the presence of the same kind of irregularities after the heat treatment. This phenomenon decreases the ferroelectric and piezoelectric features of PZT because defects form areas with tampered composition and make impossible the surface area control when other structures, like electrodes, have to be built on it. The figure below shows the effect of striation and punctual defects on film morphology after thermal treatment. a) b) Figure 20: Effect of spinning striation a) and particles contamination b) after film crystallization. 53 4.4 Thermal treatment :
Even though the sol-gel method seems to be a simple process, it does possess a number of variables and constraints allowing exploitation of its unique features. These mainly are related to the volume of organic content in the solution, that may influence the mechanical film behavior and the formation of cracks during high temperature treatment. The thermal treatment is the fundamental step to transform the PZT precursors solution film into the final crystalline material. There are essentially four steps: - Deposition of the solution containing the precursors and jellification on the substrate - Drying of the layer for evaporating the solvents - Calcination of the layer for eliminating the remaining solvents and the organic components. After this step the material reaches an amorphous phase - Crystallization for obtaining the Perovskite phase and densification of the material As can be imagined, significant weight loss and shrinkage occur in the film during thermal stages. Hence, a careful control on these parameters and on processing conditions is required and their influence on the final structure must be understood to minimize the issues. 4.4.1 Thermal analysis: In order to justify a heating sequence for the PZT films, it is necessary to know the effect of heating the sol gel precursors solution at different temperatures. TGA studies from the literature [5,8,33] confirm that the maximum weight loss occurs at 100-200°C and this is justified with the solvent and most volatile species evaporation. At temperature between 200-300 °C the evaporation of the residual solvent hold in the sample continue and there are the first traces of CO2, that prove the beginning of the carbon chains burning into the film. This release will reflect the bonding of the organic species to the inorganic components of the gel. For example, acetic acid, that in solution acts both as chelating ligand and solvent, is evolved from the coating in different temperature regions. The acetic acid that is vaporized in the temperature region from 70°C to 120°C is free in the solution . As water is lost , more acetate groups are bonded to the titanium and zirconium ions. The bonded acetate groups are driven off around 175°C. This thing justifies the continuous evaporation of acetic acid until high temperatures. A weight loss peak is present at 285°C and this is attributed to the Lead acetate dissociation 54 and production of PbO. In fact, when the gel films are heated to T=285 °C, lead acetate melts and starts to decompose to the more stable lead carbonate. The amount of acetic acid driven off increases drastically, carbon dioxide is evolved, and network movement takes place. At temperature of 300-450 °C the loss of weight continues, even if with a slower rate. In this range of temperature there is still evolution of acetic acid, that comes probably from lead acetate decomposition, CO, CO2, and PbO. Upon heating in this range of temperature the organic residuals are burned off, forming an amorphous solid. This step is referred to as firing stage. This solid is transformed into an intermediate pyrochlore phase at pyrolysis temperature between 300 and 470°C. The firing has a great relevance on the subsequent crystallization step and the temperature at which this occurs strongly influence the crystallization mechanism as reported later on. At temperature of 450-700°C the weight loss is minimal and there are traces of evaporated PbO, CO2 and CO [33]. The residual organic content in the film is completely burned off during this stage and the film transforms from an amorphous solid to a crystalline perovskite structure. This step is referred to as annealing. The amorphous film transformation begins at different temperature, depending on the type of sol gel used, and it''s observed in the TGA as an exothermic peak that indicates phase transformation . Using low carbon content solution it''s possible to obtain crystallization at low temperature as 550°C. Usually it is necessary to select an appropriate temperature to obtain the largest amount of perovskite phase and to eliminate the non ferroelectric pyrochlore phase. The temperature usually used varies from 550 °C to 700 °C. It is also important to balance the lead loss during high temperature heating, since the Pb content decreases monotonically with increasing annealing temperature and the lead loss may obstruct the formation of the perovskite phase. 4.4.2 Selecting an heating sequence: The thermal analysis provides guides for choosing the different temperatures to process the PZT films. Other considerations such as the film thickness and the specific type of sol-gel need to be considered before selecting a final thermal processing sequence. After spinning the PZT sol onto the substrate, as already discussed, a thermal treatment is required to promote gelation, pyrolysis and crystallization of the film. The thermal processing commonly used for 2-Me based sol-gels consists of quickly heating the film to 400°C by placing the sample onto a hotplate and then annealing it at the desired temperature in a RTA. Acetic acid based PZT films require a more carefully chosen heating sequence in order to maximize the crystallization and the electrical properties. In addition to the high carbon content, cracking of the film is also a problem when using this type of sol-gel [5]. The heat processing in this work has been divided into three steps. During the first step the film is dried in air on a hot plate, at temperature between 100-200 °C. At this range of temperature the most volatile 55 element will evaporate (acetic acid, water and alcohols in our case). Solvent evaporation produce shrinking and densification of the film as it loses liquid. During this process bubbles and in-homogeneities can appear on the film surface when the solvent is not free to evaporate from the film. Moreover, during the drying process, some solvent will flow across the polymeric network present in the film, to compensate for the different solvent evaporation rate through the film. This latter causes a gradient of pressure inside the matrix of the sol gel that can lead to the breaking of the rigid network and defects formation. For this reason an appropriate temperature and time must be chosen to improve the quality of the dried film. After drying, the film is much denser, and any remaining liquid is held in by capillary action. In this work the drying stage has been performed in air on a hot plate at 150 °C for 5 minutes for all the samples. This choice has been done comparing the data from literature about the mass loss rate at different temperature for the kind of solutions used in this work. The data in literature have been gathered using thermo gravimetric analysis (TGA). This process assures the complete evaporation of the solvent without producing any type of defects. After drying, the as-deposited film is an amorphous gel of metallorganic compounds. After drying, the sample is immediately moved on another hot plate at temperature between 300°C and 450°C. This step is referred to as firing or calcinations. At these temperatures the residual solvent is eliminated and lead acetate dissociation occurs, producing free acetic acid. The organic fraction in the film is burned off and the gel transforms into an amorphous solid. The firing temperature and time must be chosen in order to assure the organic component burning and the amorphous solid phase formation but to prevent excessive PbO evaporation. Firing at lower temperatures for prolonged periods results in only partial removal of the residual carbon, and contributes to the evaporation of lead oxide. In order to maximize the removal of the carbon and to minimize the lead loss, a rapid increase of the temperature from 150°C to the crystallization temperature would be desirable. However, increasing the temperature directly from 150°C to 650°C is not beneficial since stresses in the film due to the sudden volume change can lead to cracking. For this reason an intermediate firing step is crucial to improve the film features and an excess of lead is added in all the solution to compensate the lead loss. The amorphous solid formed during heating creates an intermediate pyrochlore phase that will evolve in a perovskite structure during high temperature annealing. As reported in the M.Cueff work [8], the stability of the intermediate pyrochlore phase is fundamental for the final properties of the crystalline film, and this depends directly on the calcination temperature. When the temperature of calcinations is high, also the oxygen concentration in the pyrochlore phase increases and this latter thing promote the stabilization of the pyrochlore phase. When lower firing temperatures are used, as 350°C, the pyrochlore phase obtained is metastable and it will transform easily in perovskite. The fact that the phase transformation is made easier by low calcinations temperature, increases the crystallization kinetics. As discussed in the previous section, when the kinetics dominates the growth mechanism, the (111) plane orientation is preferentially obtained on Pt substrates. On the other hand, when the calcinations temperature is high as 420°C, the more stable 56 pyrochlore phase created slows down the crystallization kinetic. Under these conditions the thermodynamic dominates the further crystallization and the (100) phase growth is favored. Many other authors noted that varying the calcination temperature from 200 °C to 500 °C the final PZT orientation varies from the (111) to the (100). In any case, it''s difficult to establish a defined temperature at which the growth mechanism change, but it remains a good method to qualitatively control the final grains orientation. The calcinations temperature, the annealing method, the substrate type and its thermal pre-treatment, that induce residual stresses into it and promote the formation of intermetallic species that favor the (111) plane growth, are so far the most important parameters to control the grains growth mechanism and final crystallographic planes orientation. The table below reports some results from the literature and shows how the calcinations temperature and the annealing method actually influence the plane orientation. Reference Calcination temperature Crystallization Temperature and annealing method Final orientation Alkoy [34] 400 °C 600°C, normal furnace 650°C, RTA (100)
(111) Aoki [35] 650°C
480°C 650°C, normal furnace (100)
(111) Brooks [36] 450°C
350°C 600°C, RTA (100)
(111) Gong [37] 450°C
200°C 650°C, RTA (100)
(111) Kim [38] 350°C
330°C 650°C, RTA (100)
(111) Calcination and annealing temperature influence on grain growth. In this work, 350°C has been chosen as firing temperature and this step was performed on a pre-heated hot plate, moving the sample onto it immediately after the first drying step. The use of low temperature as 350°C has been done to decrease the possibility of cracking due to thermal mismatch between the substrate and the amorphous film and too rapid densification . A calcinations time of 5 minute assures the burning of the residual solvent and of the organic species present into the film. After calcinations follows the third and last step of the thermal treatment . This is performed at high temperature using a normal furnace or a rapid thermal annealer (RTA), and this step assures the crystallization of the amorphous deposit. The annealing temperatures used in the literature vary from 550°C to 750°C and the heating ramp should be fast enough to prevent the excessive lead loss from the film. Lack of lead during thermal processing is a factor inhibiting crystallization of the PZT into the perovskite phase. The choice of a proper annealing temperature and time must be done to guarantee the maximum percent of perovskite phase formation into the film. This depends on many factor as the type of PZT precursors solution used, substrate and firing temperature. Usually high temperatures guarantee 57 complete perovskite phase formation, but the Pb content decreases monotonically with increasing annealing temperature due to PbO evaporation, and similarly to the Pb content, also the oxygen content decreases with increasing annealing temperature. For this reason a right balance between the lead loss and the necessity of obtaining a total perovskite phase must be found and that it''s done tuning the annealing temperature and time for the different substrates and solutions used. The PZT crystallization temperature for thin films is in general much lower than that one used in bulk PZT processing, and it depends strongly on the carbon content into the gel film and on the evolution degree of the rigid structure formed by poly- condensation of the metal alkoxides during the first drying processes. The presence of an ordered rigid structure into the film, before thermal treatment, helps the PZT crystallization and contributes to the enhanced electric properties. For these reasons, it has been possible to crystallize thin films at low temperature as 550°C. Usually PZT film are annealed between 650 - 700°C, which assures the formation of the perovskite phase. In the M.Dueff work [8] the crystallization temperature is not considered very relevant for the final films properties, since it has to be just high enough to transform the amorphous solid film into a crystalline perovskite structure. Besides the crystallization temperature, the method used to anneal the film has a big relevance on the grains growth mechanism. The two annealing methods, adopted during years, imply the using of a normal furnace, or the crystallization in a rapid thermal annealer (RTA). Using a normal furnace the kinetic of crystallization is slow and for this reason the most thermodynamically stable phase, in our case the (100) for PZT, will tend to form. On the other hand, using a RTA, the increase of temperature in the furnace is very quick and the kinetic will be dominant in the growing mechanism, promoting the formation of the phase that will be kinetically favored, as the (111) for PZT on Pt substrate. Another time, it can be shown how the PZT crystallization mechanism depends on the balance between the kinetic and the thermodynamic of grains growth. In this work the crystallization temperature has been set at 650°C and the annealing time has been optimized for each one of the different substrates used to obtain a complete perovskite phase formation and to avoid any presence of non-ferroelectric pyrochlore phase. Moreover, when choosing a proper annealing time, it must be taken into account the substrate''s elements diffusion into the film during annealing, that can lead to a change in film stoichiometry and can also induce cracking. In particular when Au/TiW/Si and Pt/TiO2/Si substrate are used, the titanium diffusion towards the surface, especially for Au electrodes, lead to inter-metallic Titanite formation and can influence the PZT grains growth mechanism as discussed previously. Titanium diffusion into the film and Titanite formation, are also causes of micro- cracking during the annealing step. On the other hand, when using a ITO coated glass as substrate, the elements diffusion into the growing film is not significant and the problem of element''s diffusion induced cracking is almost completely avoided. The annealing step has been performed for all the samples using a conventional furnace under normal atmosphere. After the firing step, the samples are carefully removed from the hot plate and immediately 58 inserted into the furnace, which has been previously heated-up at 650°C and cleaned to avoid particles contamination in the film. Summarizing, the thermal treatment that has been chosen and performed on all the samples is illustrated in the flow chart below, Fig.21. The annealing time, as reported previously, has been optimized for each substrate, and after some tests it has been set at 5 minutes for the Au/TiW/Si substrate and 15 minutes for the ITO and for the Pt/TiO2/SiO2/Si substrate. Figure 21: Thermal treatment flow chart
59 4.4.3 Stresses evolution during heating : Gel films are converted into ceramic films by thermal treatment, and this latter is divided into three steps as discussed in the previous section. Drying and firing causes densification, pyrolysis and the release of volatile compounds from the film. These changes in volume combined with a mismatch between the thermal expansion coefficient of PZT and substrate, produce internal stresses in the film, which may lead to cracking. Stress that generates during firing is the origin of the cracks in gel-derived films, and can also provides substrate bending. Properties of the films like Curie point and crystallographic plane orientation could also be affected by stress into the film. Therefore, scientific understanding is quite important on stress evolution in gel-derived films. When fired, gel films experience: heating-up, isothermal heating (annealing), cooling-down stages. In the heating-up and isothermal heating stages, gel films undergo changes in structure and chemistry, which results in evolution of stress. The stress thus generated at any occasion during heating-up or isothermal heating stage is called intrinsic stress. In the heating-up and isothermal heating stages, gel films are densified with several 10% in volume reduction due to capillary pressure on solvent vaporization and polycondensation reaction between PZT metals precursors. Densification leads to in-plane tensile stress because the films are constrained on the substrates. Thermal stress could also be generated in the heating up stage when the films and substrates have difference in thermal expansion coefficient. In the range of temperature from 200°C t o 300°C, gel films crack readily . As the volatile components vaporize, the films must shrink . However, because it is constrained by the substrate and by internal parts of the network, the surface cannot shrink freely and internal tensile stresses will be created. If these cannot be relaxed the film will crack when the internal tensile stress achieves a critical value. For this reason fresh films are generally dried at lower temperature to reduce the rate of solvent evaporation and consequentially to allow the stress release , and only after drying, treated at higher temperature. This reduces the thermal shock at which the film is exposed. The film densification during the heating up stage is the main source of stresses. In fact, because the difference in thermal expansion coefficient 'α and the change in temperature 'T , would be in the order of 10 -7/10-5 K-1and 102 K respectively, and the resulting strain, 'T 'α, would be 10 -5 10-3 in order, which is much smaller than those occurring via densification. [32] Therefore, stress due to densification would be dominating in the heating-up (and isothermal heating) stage. In the cooling down stage, on the other hand, no major changes are expected in structure and chemistry. Therefore, it would be thermal stress that is expected to be generated in the cooling stage. The intrinsic stress at the end of the isothermal heating stage and the thermal stress generating in the cooling stage are accumulated in films, giving rise to residual stress, which can be measured at room temperature after firing. In our case Pt, Au and ITO acting as bottom electrodes, have a larger thermal coefficient of expansion than the PZT. This mismatch leads to stresses build up during the subsequent heat treatments. Other stresses can arise from the contamination of the film with other particle like dust. These obviously 60 create defects and irregularities that lead to punctual stress generation and cracking. When the substrate, as in the case of Platinum, is annealed before PZT deposition to improve its (111) orientation, further stress is collected on the electrode. On application of PZT, the stress prevalent in the annealed substrate decreases as the film is initially stretched and then shrinks following the heat treatment for the PZT layer. The pictures below, taken with an optical microscope, illustrate cracked PZT layers on Au substrate,Fig.22a, and on ITO coated glass, Fig.22b. a) b) Figure 22: Optical microscope image of cracked PZT mono layer on a) Au/TiW/Si b) ITO coated glass
The figures above illustrate macro cracks over all the film. These are very long and interconnected cracks that arise in few seconds when the sample is put on the hotplate at 350°C for the firing steps, and they can be seen by naked eye. They are caused by sudden film densification and thermal mismatch between the substrate and the PZT layer as discussed previously. It''s possible to prevent the formation of these cracks depositing thinner film or adding into the sol gel solution some plasticizer as glycerol. During the annealing step micro cracks can arise, Fig.23, and these are caused either by thermal mismatch, stresses induced by phase transformation, and substrate''s elements diffusion into the film. Figure 23: Micro cracks in PZT layer on Au/TiW/Si 61 4.5 Top electrodes deposition: After PZT film deposition, top electrodes must be patterned onto the film to check the electrical properties of the PZT ferroelectric ceramic film deposited. Pattering top electrodes with the use of an appropriate shadow mask, a matrix of capacitors is obtained and this structure is suitable for electrical measurement, Fig.24a. More exactly, a matrix of circular gold electrodes has been deposited using Argon ions cold sputtering technique. With the appropriate time of exposure, a gold layer of 80 nm thickness has been deposited on the top of the mask. The diameter of each electrode is around 500 µm, so the total electrode area is around 0,785 mm 2. To create the pattern onto the film, a shadow mask has been fabricated making a (9x9) matrix of holes of the desired dimension, into brass foil of 0,45 mm thickness. The mask is placed on the sample that has to be patterned and removed at the end of the sputtering process to create the desired circular gold electrodes, Fig.25. a) b) Figure 24: a) PZT capacitors structure b) electric scheme of two capacitor in series. The electrical characterization will be performed connecting in series two capacitors and checking the final capacitance and polarization of the system. The total capacitance will be equal to that one of a capacitor with a double dielectric thickness, considering that the area of all the electrodes is the same. A scheme of the final system that has been tested is reported in Fig.24b. Figure 25: Optical microscope image of gold electrodes matrix patterned onto PZT films 62 Chapter 5: Results and discussion
The final PZT ceramic films morphology, crystal structure and electrical features have been analyzed and discussed after thermal treatment for all the different combinations of solutions and substrates used. The film morphology has been checked using optical microscope and scanning electron microscope (SEM) and the detection of the crystallographic phase of the films, has been performed using grazing incident X-ray diffraction to increase the signal from the films and eliminate the substrate interference. The final film morphology and crystal structure show great differences depending on the annealing time chosen, on the thickness of the film, the substrate used and the type of PZT precursors solution utilized. These factors will obviously impact on the electrical characteristics of the final devices. For all the tree types of substrates used, it has been possible to deposit uniform, dense and crack free films that showed a fully crystallization into a perovskite phase. The annealing times have been optimized for the different samples, to obtain a fully crystallization of the films but to reduce all the problems that arise when long annealing time are chosen, as elements inter diffusion and film cracking. The electrical features of the films that possess good morphology, uniformity and a right XRD pattern have been checked measuring the C-V curves and extrapolating the dielectric constant and the coercitive field values. As will be discussed, all the C-V curves derived from the films deposited on Au, ITO and Pt have a butterfly shape and this confirmed that the deposited films shown a ferroelectric behavior. In fact, the non linearity between the applied field and the capacitance of the dielectric is a manifestation of the presence of remanent polarization and coercitive field of the ferroelectric film. Due to large differences in crystallization behavior and films morphology shown by the various types of substrate used, this chapter will be divided into three sections, and in each one of these the results for a single type of substrate will be presented and discussed. Finally the electrical characterization of those samples that possess the best qualities will be presented and discussed and the differences between the films deposited on the various substrates analyzed.
63 5.1 PZT films on Au/TiW/SiO2/Si substrate Gold seems to be a good candidate as PZT thin film electrode material, thanks to its high conductivity, its compatibility with the thermal treatment at which it has to be subjected, reduced reactivity with other elements and lattice constants matching with the PZT, that favors the film crystallization. Moreover, either the water or the isopropanol based solutions showed good wetting behavior and for both it has been possible to completely cover the substrate and create an homogeneous film using spin coating technique, even though the isopropanol based solution shows, during the drying step, better adhesion with the substrate and less pronounced shrinking. In our case, the major drawback with the use of this kind substrate, has been the high Ti diffusion from the TiW adhesion layer under Au, to the surface and the PZT top layer. Titanium diffusion into the growing film induces stresses and leads to micro-cracks formation when the annealing time exceeds approximately 15 minutes. Furthermore, Ti diffusion alters the stoichiometry of the growing film and creates an intermetallic layer between the Au electrode and the PZT layer, that strongly impacts on the electrical features of the device. For these reasons it has been impossible to use high annealing times and this fact made the crystallization step, to obtain a complete perovskite phase, more difficult. Using the 0,5M isopropanol based solution, and choosing short annealing times as 5 minutes, it has been possible to create single and double PZT homogeneous layers on gold electrodes Fig.26. a) b) Figure 26: Optical microscope image of PZT films on Au electrode derived from 0,5M isopropanol based solution a) double layer b) single layer. 64 Either the single and the double layered films were cracks free, even if some defects were still present due to film contamination with atmosphere''s particles during the spinning and thermal process, or derived from possible irregularities, scratches and defects already present on the substrate before deposition. Fig.27 shows a scanning electron microscope image of a double layer film of PZT derived from isopropanol based precursors solution. The annealing time was set at 5 minutes and in this way both film macro and micro-cracking were avoided, as can be seen. The film shows a good density and an homogeneous grains growth. The film is completely crystallized and, as will be showed by the XRD analysis, the PZT film has a complete perovskite structure with no remanent pyrochlore phase. Figure 27: Double PZT layer derived from isopropanol based solution The increase of the annealing time leads to micro-cracks formation, when this exceeds approximately 15 minutes, and high Titanium diffusion into the growing film, Fig.28. Micro cracks are caused either by thermal expansion coefficient mismatch between the substrate and the PZT film , densification and shrinking of the growing layer, stresses induced by phase transformation inside the film and stresses due to Titanium diffusion. Micro-cracks differ from macro-cracks in the extension of the cracked area and because the formers are not interconnected but isolated. The effects of long annealing time are illustrated in Fig.28, that shows PZT film deposited on Au substrate and annealed for 15 minutes. It''s possible to see micro-cracks on all the film surface and the effects of the diffusion of Titanium, that appears as darker areas inside the film. Due to those reasons, the annealing time when using this kind of substrate, can''t exceed 15 minutes to avoid micro cracking, and this impacts strongly on the PZT crystallization process, since this time could be too short to complete the transformation of the non ferroelectric pyrochlore phase into the perovskite one, as discussed later on. 65 Figure 28: Micro-cracks into 15 minutes annealed PZT film derived from isopropanol based solution. PZT thin films derived from water based solution show different morphology and crystallization behavior respect to the isopropanol based sol gel. First of all, it hasn''t been possible to obtain homogeneous films using this solution and, depending on solution molarity and on the annealing time, film cracking or formation of areas with different morphology always occurred. This phenomenon is probably caused by a poorer wetting behavior of the solution on the substrate, that is derived from its different surface tension and viscosity, compared with the isopropanol based PZT precursor sol-gel. When 1M solution has been used, the film crystallized in an homogeneous way and the film surface appeared dense and uniform, even if star shape cracks were always present, even if short annealing time as 5 minutes had been used, Fig.29. The high molarity of the solution implies the formation of a denser film on the substrate after spin coating, and during the drying process these films showed higher rigidity compared with the 0,5M solution derived films. a) b) Figure 29: Star shape cracks onto PZT film derived from 1M water base solution. a) optical microscope image b) SEM image 66 This type of cracking never happened using isopropanol based solution, until high film thickness, approximately 0.4 µm, was reached. The higher molarity presupposes also an higher particles concentration and so an higher number of nucleation sites for crystallization. For this reason the constraints and stresses at which the grains are subjected during the annealing step are strong due to grains growth competition. When 0,5M water based solution has been utilized, the film uniformity increased and only few cracks appeared. Single and double PZT layers, derived from 0,5M water based solution, were fabricated. They appeared dense and uniform, but unfortunately island of different morphology were always present, Fig.30. The film surface appears smooth and so the presence of these islands couldn''t be attributed to areas of different heights, but the reasons of the formation of these agglomerate could be attributed either to different crystallization behavior of some areas and presence of inhomogeneities in solution, as condensed or not completely dissolved particles. Figure 30: Island of different morphology into PZT film derived from 0,5M water based precursor solution and annealed 5 minutes. Similarly to the isopropanol-based solution derived films, also for the PZT films derived from water based precursor solution, film cracking has been avoided and the titanium diffusion limited, choosing short annealing times. When longer times had been used, also for this kind of films, cracking and stoichiometry alteration of the PZT layer occurred, due to high titanium diffusion. It can be concluded that PZT thin films with good morphology, dense and crack free can be deposited on Au/TiW/Si substrate, using both the 0,5M isopropanol and water based PZT precursors solution and choosing short annealing time as 5-10 minutes. The overall film morphology was in any case better for the films derived from isopropanol based solution , since into the PZT layers derived from 0.5M water based solution, island of different morphology were always present. For both the solutions used, when the annealing time exceeded approximately 15 minutes or higher molarity solutions were used, macro or micro-cracking occurred overall the film and high titanium diffusion took place. 67 5.1.1 X-ray diffraction analysis
The crystallographic structure of all the films deposited on Au electrode, has been studied using grazing incident x-ray diffraction to increase the signal from the thin film and to reduce the interference of the substrate. As discussed previously, the annealing time has a great relevance on the final PZT crystallographic structure, and the development of a complete perovskite phase into the film is strongly dependent on the annealing time and temperature. The latter has been set at 650°C for all the samples and the annealing time has been changed from 5 to 30 minutes. The kinetics of crystallization for the films derived from the two different solutions is not equal, and it depends on the solution molarity as well as on the substrate used and on the annealing temperature. The films derived from the 1M water based solution crystallized almost completely with only five minutes annealing into a perovskite phase with a preferential (110) orientation, Fig.31. As already mentioned the (110) phase is the most thermodynamically stable and its peak is placed around 2θ = 31.1°. The peak placed at 2θ =21.8° in Fig.33 indicates the (100) plane and through this peak is possible to calculate the lattice constant of the rhombohedral PZT phase that in our case is 4.04 '. This value is in agreement with those in literature. The peak at 38.3° the (111) plane and the peak at 44.5° is the (200) plane. All these peaks are characteristic of the rhombohedral PZT, and the angles at which they are indicated are in perfect agreement with the literature. The peak at 2θ = 50° is characteristic of the (201) or (102) plane of the tetragonal phase and it doesn''t appear in the rhombohedral PZT. An explication of the presence of both the peaks of the tetragonal and rhombohedral phase, is that the crystal structure, in our case, is a mix between the tetragonal and rhombohedral PZT phase, because the stoichiometry of the precursors solutions were always around the morphotropic phase boundary between these two phases, that is obtained with a Zr/Ti molar ratio of 52/48. Other peaks that are characteristic of the tetragonal phase, as the (001) peak at 21.6° ,can''t be seen, probably because it superimposes to the (100) peak at 21.8°. Figure 31: XRD of PZT (Zr,Ti) 52/48 film derived from 1M water based solution and annealed at 650°C for 5 min. 68
The peak at 55° is characteristic of the (121) plane and the peak at 64,8° of the (222) plane. These last two peaks are present both for the tetragonal and the rhombohedral PZT phase, even if peak splitting occurs for the tetragonal PZT. All these values are in agreement with the literature and so it can be concluded that PZT films with the right crystallographic structure have been obtained. The pyrochlore phase, that is normally identified with the peaks at 2θ = 29.6° and 2θ = 34.4°, is almost completely disappeared in PZT films derived from 1M water based solution and annealed 5 minutes, and the transformation into a perovskite phase is complete, as can be seen in Fig.31. The peaks position slightly shift changing the Zr/Ti ratio, the film thickness and the type of substrate, and so the angles at which the various peaks are identified will slightly change during this work. The position at which the peaks are collocated is related with the lattice constants of the PZT phase, and theirs values are strongly affected by the Zr/Ti ratio and by the residual stresses present into the film. Figure 32 illustrates the x-ray diffraction results for a double PZT layer obtained using the isopropanol based solution. The annealing time was set at 5 min for both the layers and the annealing temperature at 650°C. Also in this case the films fully crystallized in a perovskite structure with a preferential (100) orientation and the typical peaks of the PZT perovskite phase can be seen. For this kind of film it''s possible to see the splitting of 2θ = 44.8° peak into two new peaks, that identify the (200) and (002) planes. This phenomenon is characteristic of the PZT tetragonal structure. Other typical peaks as the 2θ =21.6° of the (001) plane of the tetragonal structure, can''t be seen, since it superimposes with the peak of the (100) plane at 21.8°. Another sign of the formation of a mix between the tetragonal and the rhombohedral structure is the splitting of the peak at 2θ = 64.8° , characteristic of the (222) plane ,into two peaks that indicate the (022) and (220) planes of the tetragonal PZT phase. Also in this case, the pyrochlore phase, identified with the two peaks at 2θ = 29.6° and 2θ = 34.4°, is almost disappeared, suggesting the complete transformation into a perovskite phase. Figure 32: XRD of PZT (52/48) double layer film derived from isopropanol based precursors solution, annealed 650°C for 5 min 69 The low intensity peak at 2θ =27.4° is not characteristic of the PZT crystal structure, but it probably indicates the formation of Titanite and it relates with the Titanium diffusion from the substrate. The intensity of this peak can be taken as a good indicator of the amount of titanium diffused from the substrate into the growing film during annealing, and it can be seen how its intensity increases with increasing annealing time. The presence of Titanium on the substrate surface, and the formation of an intermetallic phase of Pt-Ti or Pb-Pt are factors that promote the growing of the (111) plane during PZT crystallization. Figure 33 illustrate the XRD of a single layer PZT film derived from isopropanol based solution and annealed for 15 minutes. It''s possible to see the growth of the titanite peak at 2θ =27.4° that demonstrates how the titanium diffusion increases, increasing annealing times. Moreover, Fig.33 illustrates how the peak representative of the (111) plane at 2θ = 38.6° tends to grow more than the (100) peak at 21.6°, and this is probably due to the influence of the titanium diffusion. Furthermore, Fig. 34 shows the same kind of film annealed for 30 minutes. It''s easy to see the further growth of the peak at 2θ =27.4° that depicts an high titanium diffusion and intermetallic titanite formation . The high titanium diffusion favors the growth of the (111) plane preferential orientation as expected, and this can be seen in Fig.34 where the peak at 2θ = 38.3° becomes predominant, indicating a preferential (111) plane orientation. Figure 33: XRD of 15 minutes annealed PZT layer derived from isopropanol based precursors solution Figure 34: XRD of PZT single layer derived from isopropanol based precursors solution and annealed at 650°C for 30 min 70 Unfortunately the high titanium diffusion into the growing PZT film is not beneficial for its final electrical properties and, in addition, it induces film micro-cracking. Moreover when long annealing times are used the lead loss increases with time, and this loss promotes the formation of the non-ferroelectric pyrochlore phase, as illustrated in Fig. 34 and discussed earlier, that can be detected at 2θ = 29.6° and 2θ = 34.4°. For all these reasons, when PZT layers derived from the isopropanol based solution are deposited, the optimum annealing time to obtain a complete perovskite formation and to avoid an excessive titanium diffusion and film cracking, can''t exceed 10 minutes. PZT films derived from the 0,5M water based solution showed a different crystallization kinetics compared to those ones obtained from the isopropanol and the 1M water based PZT precursors solutions. When short annealing times were performed on PZT mono layer films obtained using the 0,5M water based solution, PZT crystals grew with a preferential orientation along the (222) plane, peak at 2θ=64.8°, as illustrated in Fig.35. With 5 minutes annealing the pyrochlore phase peaks at 2θ = 29,6° and 2θ = 34.4° are still present, indicating that the crystallization towards a perovskite structure is not complete. Also the (100), the (110) and the (111) plane peaks were present at 2θ = 21.8°, 31.1°, 38.3° respectively, but they were not well defined, suggesting that the film crystallization was not complete and so the grains were not well defined. Figure 35: XRD of PZT monolayer obtained using 0,5M water based solution and 5 minutes annealing Increasing the annealing time to 15 minutes, all the characteristic peaks became better defined and the PZT crystal were still strongly preferentially orientated towards the (222) plane direction, Fig.36. Titanium diffusion took place and it is identified with the growing of the 2θ =27.4° peak. Either the (100), (110) and (111) planes peaks were present with a major growth of the (111) plane direction compared with the 5 minutes annealed sample and this could be probably due to higher titanium diffusion, that as already discussed favors the growth of the (111) plane. These plane peaks were definitely better defined, suggesting an higher film crystallization, even if a low pyrochlore phase peak could be seen at 2θ =29.6°, Fig.36. 71 PZT double layers derived from 0,5M water based solution showed a different XRD pattern, Fig.37. The film crystallizes with a preferential (110) orientation and not with a (222), as showed by the mono-layer. The (222) peak at 2θ=64.8° was still present but with a lower intensity. This fact suggests that the substrate texture could be responsible for the un-expected crystals growth towards the (222) orientation of the PZT monolayer. The others representative peaks of the PZT perovskite phase and low percentage of the pyrochlore phase were still present, indicating a not completed transformation into the perovskite structure. Figure 37: XRD of PZT double layer obtained from 0,5M water based solution and annealed 5 minutes at 650°C In summary, single and double PZT layers with a complete perovskite structure and preferentially oriented towards the (110) plane direction have been obtained on Au/TiW/Si substrate. Either for the isopropanol and for the water based solutions a fully perovskite structure has been obtained even if, when 0.5M water Figure 36: XRD of PZT monolayer obtained from 0,5M water based solution and annealed 15 minutes at 650°C. 72 based solution has been used, the mono layered films unexpectedly crystallized with a preferential (222) plane direction. Probably the substrate has a certain influence on this crystallization pathway, because when multilayers of the same solution has been deposited, they crystallized along the most thermodynamic stable (110) plane direction. Using short annealing times as 5 minutes a complete transformation into a perovskite structure has been reached for both the type of sol gel used, but not for the 0.5M water based solution. In the XRD pattern of the film derived from this latter solution a certain amount of pyrochlore phase was in fact always present. Increasing the annealing times leads to a better crystallization but also to an higher titanium diffusion from the substrate that can influence the crystal growth, favoring the orientation towards the (111) plane direction, as shown in Fig.34. However, long annealing times and high titanium diffusion induce film cracking and film stoichiometry alteration and so to produce good quality and performing devices, short annealing times must be used when depositing PZT film on this kind of substrates.
5.2 PZT films on ITO coated glass ITO coated glass is not one of the most used substrate to deposit PZT thin films, and not so many researches have been done on the using of this material as PZT film substrate. One the major drawbacks with the using of this material, is the low conductivity associated with the ITO electrode, that is at least one order of magnitude lower than that one of gold or platinum. Just to have an idea of the conductivity values of the various substrates, the electrode resistivity has been measured using a multi-meter and it''s around 3Ω for platinum electrode and around 55Ω for ITO before thermal treatment. Moreover, the ITO resistivity further increases when it''s subjected to high temperature thermal treatment and it becomes one order of magnitude higher. The measured value of resistivity in fact increases approximately to 300Ω. So, the final conductivity of the ITO electrode after thermal treatment is two order of magnitude lower than that one of the most commonly used platinum layer. This reduced conductivity obviously impacts on the electrical response of the final device built on this material, and in particular the high frequencies response is the more influenced and limited feature, due to the raising of ITO resistivity. For ferroelectric memory applications it''s the relatively high sheet resistance of the film which limits the upper frequency response of the device. This phenomenon is less important at the frequencies normally employed in piezoelectric devices [29]. Apart from its high resistivity, the ITO coated glass shows very good compatibility with both the PZT precursors solutions used, and the morphology of the films deposited on that material is normally very good and there aren''t appreciable differences between films derived from the water and the isopropanol based precursors solutions. Moreover, PZT thin films show very good adhesion on ITO coated 73 glass and so it has been possible to deposit single and multiple dense, uniform and transparent films of 200-250 nm thickness each. With both the type of solutions used film cracking never occurred unless very long annealing times had been used or very thick films had been deposited. With approximately one hour annealing film micro-cracking occurred but for shorter annealing times the films crystallized completely without any cracks formation. The crystallization kinetics is characterized by the nucleation and growth of typical rosette structures in an amorphous matrix, and the number and dimension of these agglomerates keep growing until all the film is completely crystallized. Using optical microscope in transmission mode it''s possible to detect the nucleation and growth of these rosette structures, Fig.38. PZT films on ITO coated glass are transparent and for this reason using the microscope in the normal mode it''s not possible to well appreciate the kinetics of film crystallization. a) b) c) Figure 38: Optical microscope transmission mode image of PZT films on ITO coated glass substrate annealed at 650°C for a) 5min b) 10min c) 15min 74 Fig.38 shows optical microscope images, obtained in transmission mode, of PZT films on ITO coated glass substrate with different annealing time. It''s possible to see in this figure how the nucleation and growth of the rosette structures took place during crystallization. XRD analysis have confirmed that these agglomerates are areas in which the crystallization into a perovskite structure with a (110) preferential orientation has been completed, and these crystallized structures grow inside an amorphous PZT matrix. With 15 minutes annealing, the film fully crystallizes into a PZT perovskite phase and the growth of the rosette structures terminates because of grains growth competition, since these agglomerates are constrained by the presence of a fully crystallized matrix in which they cannot expand anymore. The single film thickness is around 250-300 nm as can be seen from the SEM image below of the section of a single PZT layer deposited on ITO, Fig.39a. The presence and the growth of rosette structures during films crystallization has been confirmed by SEM analysis, Fig.39b. a) b) Figure 39: SEM image of a) transversal section of single PZT film b) top view of 10 minutes annealed PZT double layer. The same crystallization behavior is valid for multilayer films, Fig.40. As discussed previously, the rosette structures growth stops when the film is fully crystallized and so, when depositing multilayer films, if the first layer is not fully transformed into a perovskite phase, the rosette structures growth can start again when the second layer is deposited and the annealing step begin. The agglomerate growth goes ahead until the layer is fully crystallized and in this way a complete film crystallization can be obtained using two annealing steps. 75 a) b) Figure 40: Optical microscope image of PZT double layer a) fully crystallized b) after 10 minutes annealing at 650°C
In summary, PZT thin films with a Zr/Ti molar ratio of (52/48) deposited on ITO coated glass show very high adhesion with the substrate and they are cracks free, unless very long annealing times has been reached. These films are dense, uniform and transparent and this features open to the possibility of building PZT transparent capacitors on ITO coated glass that can be used for example in photovoltaic application. The films crystallize with a typical rosette structure and this has been confirmed either by SEM and optical microscope analysis. The dimension of these agglomerates keep increasing until the growing competition between them begin and the film is fully transformed from an amorphous pyrochlore phase into a ferroelectric perovskite structure. The average diameter of these rosette structures when the film is fully crystallized is around 10µm. Both the two type of solution synthesized have shown good compatibility with this kind of substrate and the kinetics of crystallization is similar. The thickness of a single layer is around 250-300 nm as confirmed by SEM analysis, Fig.39a, and thicker films had been produced depositing multilayers.
76 5.2.1 X-ray diffraction analysis As discussed in the previous paragraph, when PZT films crystallize on ITO coated glass, they create typical rosette structures that nucleate when the annealing step begins and they keep growing on until all the film is fully crystallized and there is competition between the growth of these agglomerates. XRD analysis confirmed that these structures are areas in which the PZT film is crystallized into a perovskite phase with a strong (110) preferential orientation, and these areas expand into an amorphous PZT matrix. Figure 41: XRD of PZT films deposited on ITO coated glass substrate annealed at 650°C for a) 5 min b) 10 min. Figure 42: XRD of PZT film deposited on ITO coated glass substrate and annealed 15 min at 650°C. The percentage of the film transformed into a perovskite phase rises increasing the annealing time, as showed in Fig.41 and Fig.42, and this transformation goes in parallel with the growing of the rosette structures, Fig.38. With 15 minutes annealing the films fully crystallize into a perovskite phase and there is not remanent pyrochlore phase, Fig.42. This is valid when using either water or isopropanol based precursors solution. All the films always crystallize with a polycrystalline structure but with a strong (110) preferential orientation independently from the annealing time and from the kind of precursors solution used. On the other hand, the percentage of pyrochlore phase into the film decreases increasing the 77 annealing time, until it disappears when 15 minutes annealing have been reached. This is demonstrated by the vanishing of the peaks that identify the pyrochlore phase at 2θ = 29.6° and 2θ = 34.4° when the annealing time is changed from 5, Fig.41, to 15 minutes, Fig.42. The peak that identifies the (110) plane direction is at 2θ = 31.1°, as for PZT film on gold electrode, and the other typical peaks of the PZT perovskite phase at 2θ = 21.8°, 38.3°, 44.6°, 55° and 64.8° are present with lower intensity, indicating a polycrystalline structure. Also for this kind of films, as for those ones on gold electrodes, there is a coexistence between the tetragonal and rhombohedral PZT structures because of the stoichiometry of the PZT precursors solutions, that is close to the morphotropic phase boundary, even if for PZT films deposited on ITO coated glass, peaks splitting is less pronounced and this could probably indicates an higher presence of rhombohedral phase. The same behavior is shown by the double layer films deposited on ITO coated, Fig.43. With 15 minutes annealing the PZT films fully transform into a perovskite phase with a strong (110) preferential orientation and no remanent pyrochlore phase was detected. These results are in agreement with the SEM and optical microscope observations. Also for the double layer films, a polycrystalline structure forms during crystallization and this structure is a mix between the rhombohedral and tetragonal PZT phase. Peaks splitting, that could indicate the presence of a tetragonal phase, occurs at 2θ = 44.8° for PZT films derived from isopropanol based solutions, Fig.43a, and this could be due to the presence of the (200) and (002) planes of the tetragonal PZT. For films derived from water based solution, peaks splitting occurs at 2θ = 50°, Fig.43b, and this could identify the (102) and (201) planes presence. All the other peaks are characteristic of the rhombohedral PZT phase. a) b) Figure 43: XRD of PZT double layer films on ITO coated glass substrate annealed for 15 min at 650°C a) film derived from isopropanol based solution b) film derived from water based solution In summary, it can be conclude that PZT films derived from both the precursors solutions used, have shown good crystallization behavior on ITO coated glass substrates and the XRD peaks were always very well defined and the values in agreement with the literature. Either single and double layers films crystallized completely into a perovskite phase with a strong (110) preferential orientation with 15 minutes annealing 78 and no remanent pyrochlore phase was detected, even increasing the annealing times. Using shorter annealing time, five and ten minutes in our case, the films didn''t crystallized completely and a certain amount of pyrochlore phase was always detected and the other peaks worse defined. All the films deposited showed good morphology and they were smooth, transparent and free from cracks and defects. The average roughness measured by AFM was around 10 nm and the presence of rosette structures formation was detected. No problems of elements diffusion from the substrate came across using 650°C as annealing temperature and all the peaks reported by the XRD belong only to the PZT crystallographic phase. So it can be concluded that there is no formation of intermetallic phases or inter-diffusion between the PZT film and the electrode. For all these reasons, ITO coated glass seems to be a suitable substrate to obtain well defined and crystallized PZT thin films with a strong (110) preferential orientation and with a good morphology. 5.3 PZT film on Pt/TiO2/SiO2/Si substrate
Platinum is the most used material as bottom electrode when depositing PZT thin films, thanks to its high stability, low reactivity, high conductivity and lattice constant matching with the PZT crystallographic structure. Both the water and the isopropanol based 0,5M precursors solutions have demonstrated good wetting behavior on platinum and it has been possible to obtain uniform films of PZT gel using spin coating technique. These films showed very good adhesion on platinum, and the final deposits demonstrated good morphology after thermal treatment. All the films deposited were free either from macro and micro cracks, even when long annealing times have been adopted. Only some delamination problems occurred with films derived from water based precursors solution when long annealing times have been used. a) b) Figure 44: Optical microscope images of double layer PZT films deposited on Pt/TiO2/Si substrate and annealed for 15 min at 650°C a) film derived from 0.5M isopropanol based solutions b) film derived from 0.5M water based solution 79 Thanks to the using of TiO2 as platinum adhesion layer, it has been possible to reduce the titanium diffusion towards the PZT film, and this opens to the possibility of performing long annealing times during thermal treatment to obtain a complete films crystallization. As it will be demonstrated by XRD analysis, there was no evidence of Ti diffusion towards the PZT film or Ti-Pt and Ti-Pb intermetallic phases formation, even when long annealing times as 30 minutes has been performed. Also the cracking problems related with the use of long annealing times and the consequent high titanium diffusion, had been eliminated using TiO2 as adhesion layer. Moreover, the platinum layer shows low reaction tendency with the PZT film and it''s stable at the temperature used in the annealing steps. Figure 44 shows optical microscope images of PZT double layers films deposited on platinum and annealed for 15 minutes. Both the samples show good density and homogeneity and the films are free from cracks and have only few punctual defects. The average films roughness measured with AFM is around 15 nm, and the thickness of a single layer measured with optical profilometer is around 250nm, as illustrated in Fig.45. Figure 45: Optical profilometer analysis of PZT single layers deposited on Pt/TiO2/Si substrate a) film derived from isopropanol based solution b) film derived from water based solution The behavior of the two different solutions on Pt electrode is similar, as for ITO coated glass samples. PZT films possess good morphology even when long annealing times as 30 minutes were performed, thanks to the beneficial effects of the reduced diffusion tendency of TiO2, and the high stability of the Pt bottom layer. Figure 46 shows an optical microscope image of a PZT single layer deposited on Pt substrate and annealed for 30 minutes. As can be seen there is no evidences of macro or micro-cracking and the films morphology is definitely better than that one of PZT films deposited on Au/TiW/Si substrates and annealed for the same time. In the latter the titanium diffusion was high and the bottom layer texture strongly changed during the annealing step. This phenomenon leads to film cracking, intermetallic phases formation 80 and alteration of the PZT layer composition. Fortunately all these issues have been solved using TiO2 as adhesion layer. Figure 46: Optical microscope image of PZT single layer deposited on Pt/TiO2/Si substrate and annealed at 650°C for 30 min. It can be concluded that Pt/TiO2/Si is a good kind of substrate for PZT thin films deposition and it matches all the requirements needed to assure good final films morphology. With both the solutions used the film appearance and morphology were good, independently from the annealing time chosen. As X-ray diffraction analysis will confirm, with less than 15 minutes annealing a complete film crystallization into a perovskite phase can be obtained and so dense, uniform and cracks free PZT thin films with the required crystallographic structure can be deposited on this kind of substrates. Only few delamination problems occurred with films derived from 0.5M water based solution when multilayers had been deposited and long annealing time performed. Since long annealing times are not required to obtain a complete film crystallization, as it will be discussed later, these problems have been solved reducing the crystallization period to a maximum of 25 minutes that are anyway enough to fully transform the PZT films into a perovskite structure.
81 5.3.1 X-ray diffraction analysis PZT films derived from both the precursors solutions have been deposited on Pt/TiO2/Si substrates and annealed for different times to check their crystallization behavior. As discussed in the previous paragraph, using TiO2 as Pt adhesion layer strongly reduces the Ti diffusion towards the surface and leaves the platinum electrode chemically unchanged after thermal treatment. Consequently, all the problems related with high Ti diffusion can be avoided and long annealing times can be performed on PZT films deposited on this kind of substrate. The absence of high titanium diffusion and intermetallic titanite formation, can be demonstrated by the omission of the 2θ =27.4° peak in the XRD pattern. This peak has been associated with the presence of titanite and, as it has been discussed for PZT films deposited on gold electrodes, it is a good indicator of the amount of Ti diffused from the substrate towards the growing film. When depositing on Au/TiW/Si substrates, the 2θ =27.4° peak was already present after only 10 minutes annealing and its intensity increased raising the annealing time, while when using Pt/TiO2/Si substrate, that peak never appeared even if long annealing times had been used. This fact is believed to be a consequence of the low diffusion tendency of titanium into the TiO2 adhesion layer. Fig.47 illustrates the XRD results for PZT films derived from isopropanol based precursors solution and annealed for different times, from 5 to 30 minutes. It can be seen how, with only 5 minutes annealing the PZT film crystallize almost completely into a perovskite phase and only a low peak of the pyrochlore is present at 2θ =29.6°, Fig.47a. a) b) c) Figure 47: XRD of PZT films deposited on Pt/TiO2/Si substrate derived from 0,5M isopropanol based PZT precursors solution and annealed at 650°C a) 5 min b) 15 min c) 30 min 82 Increasing the annealing time, and maintaining always the annealing temperature at 650°C , the intensity of the pyrochlore peak decreases, until it completely disappears when 30 minutes annealing have been reached, Fig.47c. Both the films derived from water and isopropanol based precursors solutions crystallized with a strong (110) preferential orientation, 2θ =31.1°, as in the case of films deposited on ITO coated glass. The different preferential orientation, in particular the (111), obtained for PZT films deposited on gold electrode was due to titanium diffusion or to Pb-Ti, Au-Ti intermetallic phases formation, and this could explain why for PZT film deposited on Pt electrodes, in which the Ti diffusion has been reduced, the films crystallize always along the most thermodynamic stable (110) direction, as in the case of ITO. The growth of the (111) PZT preferential orientation, should be also promoted by the presence of the (111) oriented platinum bottom electrode, that should favor the kinetics of crystallization promoting the PZT crystals growth along the (111) direction, but evidently its influence is weaker than that one of the titanium diffusion, and during crystallization is the thermodynamics that governs the growing process. For this reason PZT films oriented towards the (110) plane direction have been always obtained, independently from the annealing time chosen and from the type of solution used. These films show as usual a polycrystalline structure and in this case the (110) preferential orientation is weaker than that one of PZT films deposited on ITO electrodes. The (100) peak is collocated at 2θ =21.9° for PZT films derived from isopropanol based solution and annealed for 5 and 30 minutes, while for films annealed 15 minutes there are two peaks at that angle: one at 2θ =21.8° and the other at 2θ =22°, Fig.49b. This is probably due to the coexistence of the rhombohedral and tetragonal structures. The peak at 2θ =21.9° is associated with the (100) plane of the rhombohedral PZT structure and it defines a lattice constant a = 4.04', while the two peaks at 2θ =21.8° and 2θ =22° probably refer to the (100) and (001) planes of the tetragonal PZT phase and they define the two lattice constants a= 4.036' and c = 4.07' with tetragonality (c/a) = 1.008. The peak at 2θ =38.3° represents for all the samples in Fig.49 the (111) plane, while the low peak at 2θ =40.11° comes from the (111) plane of Pt and is due to substrate signal noise. The peak at 2θ = 44.8° refers to the (200) plane and for the sample annealed 15 minutes peaks splitting occurs at this angle, indicating another time the coexistence between the tetragonal and rhombohedral PZT phases. The other typical PZT peaks at 2θ =50° , 2θ =55° and 2θ = 64.8° are present and also at these angles peak splitting occurs at 2θ = 55°, for the samples annealed 5 and 30 minutes, Fig.47a-c, and at 2θ = 64.8° for the sample annealed 5 minutes. This results are in agreement with the XRD analysis of the PZT films deposited on the other substrates, that have always shown coexistence between peaks that come from the tetragonal and the rhombohedral PZT phases. In summary, films derived from 0.5M isopropanol based solutions crystallize quickly and with only 5 minutes annealing an almost complete perovskite phase can be obtained. Increasing the annealing time leads to a fully transformation into a perovskite phase and to a better definition of the characteristics peaks. These peaks identify the formation of a polycrystalline structure but with a strong (110) preferential orientation in which the tetragonal and rhombohedral PZT phases coexist because of the film stoichiometry, that is close to the PZT morphotropic phase boundary. 83 Films derived from 0.5M water based PZT precursors solution, show a different crystallization kinetics and they take more time to fully transform into a perovskite phase. XRD results for this kind of films are illustrated in Fig.48. As can be seen, a certain amount of pyrochlore phase is present into these films until the annealing time doesn''t reach 30 minutes, Fig.48c. The pyrochlore phase peaks are located at 2θ = 29.6° and 2θ = 34.2°. These films take more time to fully transform into a perovskite phase and this fact could be due to the presence of a much disordered structure inside the gel film before annealing. As reported in the literature, the presence of an ordered structure inside the PZT films before the beginning of thermal treatment is one of the parameters that much influences the crystallization kinetics. Anyway, with 25-30 minutes annealing, the film fully crystallized into a perovskite phase and peaks splitting is less pronounced if compared to the XRD pattern of the films derived from isopropanol based precursors solution. Also in this case all the peaks belong to the PZT crystallographic structure except from the 2θ =40.11° peak that identifies the (111) plane of the platinum bottom electrode. Even though the crystallization kinetic is slower, also these films crystallize with a strong (110) preferential orientation. The peak that identifies this plane is at 2θ = 21.9° and it defines a PZT lattice constant a = 4.04 '. This value is in agreement with the literature. Peak splitting occurs only for the 5 minutes annealed sample and this occurs at 2θ = 50° and 2θ = 55°, Fig.48a, suggesting the presence of the (201)/(102) and (211)/(112) planes respectively, that belong to the tetragonal PZT structure. a) b) c) Figure 48: XRD of PZT films derived from 0.5M water based PZT precursors solution deposited on Pt/TiO2/Si and annealed at 650°C for a) 5min b) 15 min c) 30 min 84 In summary, PZT films deposited on Pt electrode crystallize with a (110) preferential orientation independently from the annealing time and from the type of solution used. On the other hand, the time required to obtain a complete film transformation into a perovskite phase changes, depending on the type of precursors solution used. PZT films derived from isopropanol based precursors solution fully crystallize with only 15 minutes annealing at 650°C, while films derived from water based solution take at least 25 minutes to fully transform into a perovskite structure with no remanent pyrochlore phase. This could be due to a different extent of ordered structures inside the film before the beginning of the thermal treatment. The films deposited are always polycrystalline, with a major presence of crystals orientated towards the (110) direction, and the XRD results show, for both the type of samples, a coexistence between the rhombohedral and the tetragonal PZT phases. Evidences of the presence of a tetragonal crystals structure are stronger for films derived from isopropanol based solution compared to those ones derived from water based solution, in which peaks splitting occurs only for the 5 minutes annealed sample.


85 5.4 Electrical characterization
After the analysis of the morphology and of the crystallographic structure of the deposited films, a matrix of top gold electrodes has been patterned onto those samples that have shown good morphology and that are fully crystallized into a perovskite phase, following the procedures described in the previous chapter. In this way, a metal/PZT/metal structure has been obtained and this is suitable for electrical measurements that in this work have been performed analyzing the C-V curves. Characteristics as absence of cracks, good density, good uniformity and a fully crystallization into a perovskite phase, are fundamental to fabricate performing devices. Using a capacimeter, measurements of C-V or ε-V curves have been made by applying simultaneously on the measured sample a DC field which changes as a step-like function and an AC voltage of relatively high frequency of small amplitude. In this case the frequency has been set at 100kHz and the oscillation level at 100mV for all the measurements. The AC voltage is used to measure the capacitance which is then plotted as a function of the DC bias field, giving a C-V graph. The C-V curves have for ferroelectric material a non-linear behavior, due to the existence of a remanent polarization and a coercitive field of the ferroelectric that alter the linearity between the capacitance of the dielectric and the applied voltage. The C-V curve for ferroelectric materials has usually a butterfly shape, in with the maximum values of capacitance are obtained near the 0V DC bias . Initial rise in the permittivity values of capacitance near the 0 DC bias is probably due to increased movement of the domain walls which become ''free'' from defects and will lock them at zero DC field. This phenomenon is due to partial switching of some domains whose coercive field happens to be small enough and which can be switched by the DC and AC field combination. The re-orientation of domains increases the polarization of the ferroelectric and thus rises the overall capacity measured. The maximum in the C - V curve is supposed to appear in the vicinity of the coercive field for the P-E hysteresis when most of the domains switch and the material appears to be dielectrically very ''soft''. In fact, when the applied field is close to the value of the ferroelectric coercitive field (Ec), very small changes in the external field could induce wide changes in the polarization status of the ferroelectric because of domains switching. Knowing the dimension and the thickness of the measured capacitors is thus possible to extrapolate an approximate value of coercitive field and usually this value is in agreement with that one given by the P-E hysteresis loop. A difference between those values may be attributed to the space charge regions, which could develop due to DC biasing in the C''V measurements. At high DC fields, the permittivity decreases, re'ecting two processes: the decrease in the number of domains as they become more and more aligned with the field (ideally the sample becomes a single domain and only lattice contributions are present) and inhibition of the movement of residual domain walls by the DC field. In this work the top and bottom electrodes were not always made of the same material and this can affect the symmetry of the C-V curves creating a mismatch between the measures performed at positive and negative DC bias. 86 -20 -15 -10 -5 0 5 10 15 20 300 340 380 420 460 500 540 580 Volts [ V ] D ie le ct ri c co n s ta n t 5.4.1 PZT films on Au/TiW/SiO2/Si substrate In the case of PZT films deposited on Au/TiW/Si substrates, only the films derived from the isopropanol based precursors solution and annealed for 5 minutes possess all the required features to produce performing devices, as good density, uniformity, homogeneity and absence of cracks, and so the electrical analysis has been performed only on this kind of samples. Moreover, as demonstrated by the XRD analysis, for this kind of substrate the titanium diffusion is actually a problem, because it becomes very high when long annealing times are performed and the consequent formation of an intermetallic phase is demonstrated to be one of the first factors that depress the electrical features of PZT thin films. Using a capacimeter set with the parameter described before, the capacity of the Au/PZT/Au capacitors fabricated has been measured and the C-V curve has been plotted, Fig.49a. Moreover, knowing the area of the electrodes and the thickness of the deposited films is possible to find out the value of the dielectric constant and the ε-V curves, Fig.49b. In this case the electrodes area is 500*500 µm 2 and the film thickness is considered to be 1 µm because the capacity is calculated for two equals capacitors in series. Each capacitor is made by a PZT film formed by two layers of approximately 0.25 µm each. a) b) Figure 49: a) C-V and b) ε-V curves for 1µm PZT film deposited on Au/TiW/Si substrate annealed 5 min. These curves have a typical butterfly shape that has been already obtained in other works [40,41] and the samples seem to stand the applied voltage in a symmetric way. The presence of a shoulder near 0 V is due to the presence of domains that remain oriented towards the same direction after that the voltage has been reversed. Increasing the DC voltage the domains switch when the value of coercitive field is reached. The values of voltages obtained from the two maxima of the C''V curve divided by thickness of the film should be nearly equal to twice the coercive field (Ec) of the P-E curve. In this case the maximum value of capacitance is obtained around 2 Volts and for a capacitor of 1 µm thickness it corresponds to an electric -20 -15 -10 -5 0 5 10 15 20 600 700 800 900 1000 1100 1200 Volts [ V ] C a p a ci ty [ p F ] 87 field of 20kV/cm and so a coercitive field of 10 kV/cm . A difference in the value may be attributed to the space charge regions, which could develop because of DC biasing in the C''V measurements. From the value of the capacitance at 0V, it''s possible to extrapolate the dielectric constant of the deposited films that in this case have an average value of 484. This quite low value compared to those one find on the literature, that are around 900, could be probably due the presence of intermetallic Ti-Pb, Ti-Au phase between the bottom electrode and the PZT film, caused by the high tendency of titanium to diffuse from the Au adhesion layer towards the surface. The presence of intermetallic layer is a factor that strongly decreases the electrical features of the thin film deposited. 5.4.2 PZT films on ITO coated glass substrate
PZT films deposited on ITO coated glass have shown very good morphology, good adhesion and it has been possible to obtain a fully crystallization without problems of cracking or elements inter-diffusion between the different layers, either for films derived from water and isopropanol based PZT precursors solution. The XRD and optical microscope analysis have shown that the crystallization behavior of these films is very similar and that for both the types of solutions used the films crystallized with a strong (110) preferential orientation. The capacitance of the Au/PZT/ITO capacitors, fabricated pattering a matrix of top gold electrodes onto the films derived from both the two types of solutions, has been measured and the C-V curve has been plotted, Fig.50-51a. Moreover, knowing the area of the patterned electrodes and the thickness of the deposited films is possible to find out the value of the dielectric constant and the ε-V curves, Fig.50-51b. As in the other cases, the electrodes area is 500*500 µm 2 but for PZT films deposited on ITO the film thickness is considered to be 0,8 µm, considering that the capacity is calculated for two equals capacitors in series. Each capacitor is made by a PZT film formed by two layers of approximately 200 nm each, as demonstrated by SEM analysis, Fig.41. Figure 50: a) C-V and b) ε -V curves of Au/PZT/ITO capacitor fabricated using isopropanol based PZT precursors solution. -20 -15 -10 -5 0 5 10 15 20 900 1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 Volts [V ] C a p a ci ty [ p F ] -20 -15 -10 -5 0 5 10 15 20 400,0 450,0 500,0 550,0 600,0 650,0 700,0 750,0 Volt [ V ] D ie le ct ri c co n s ta n t 88 -20 -15 -10 -5 0 5 10 15 20 1300 1500 1700 1900 2100 2300 2500 Volts [ V ] C a p a ci ty [ p F ] a) b) Figure 51: a) C-V and b) ε -V curves of Au/PZT/ITO capacitor fabricated using water based PZT precursors solution. Also these films show ferroelectricity and the C-V curves have the typical butterfly shape of ferroelectric materials, Fig.50-51a. Both the C-V curves are not symmetric and this probably happens because the electrodes are made of different materials. The C-V curves were indeed symmetric for Au/PZT/Au capacitors. In particular the values obtained for positive DC bias are slightly lower than those ones obtained for negative DC bias, and the negative wing of the C-V curve is narrower than the positive for both the curves. This phenomenon is probably related with the ITO n-type conductivity associated with the oxygen vacancies that could be suitable sites for charges trapping. Furthermore, the ITO conductivity is widely lower than the conductivity of the gold electrodes. The value of the dielectric constant at 0V for the two samples taken from the ε-V curves is very similar, and in particular is 655 for PZT films derived from isopropanol based precursors solution and 643 for films derived from water based precursors solution. These are quite high values of ε if compared with those ones found in the literature for PZT thin films deposited on ITO electrodes [40,41]. The good morphology and crystallization behavior of the deposited films might play a fundamental role to obtain these high values. The maximum values of capacitance for PZT films derived from isopropanol based solutions, are measured around 2V either for positive and negative DC bias and this voltage corresponds to an electric field of 25 kV/cm if considering a 0,8 µm PZT dielectric layer. From this value is possible to predict the coercitive field of the P-E loop, that should be around 12 kV/cm. For PZT films derived from water based solution the maximum values of capacitance are around 4 V for positive DC bias and 2V for negative DC. The films thickness is still considered to be 0,8µm and so this voltage corresponds to a coercitive field (Ec) of 12 kV/cm for negative bias and 25 kV/cm for positive bias. This asymmetric behavior is attributed to the differences in the working potential between the top Au electrode and the bottom ITO layer. -20 -15 -10 -5 0 5 10 15 20 350 400 450 500 550 600 650 700 750 Volts [ V ] D ie le ct ri c co n s ta n t 89 5.4.3 PZT films on Pt/TiO2/SiO2/Si substrate
PZT films deposited on Pt electrodes have shown good density, uniformity, a fully crystallization into a perovskite structures with a preferential (110) orientation and absence of both macro and micro-cracks. Both the films derived from the two solutions have demonstrated good behavior on Pt electrodes even if PZT films derived from water based precursors solution have shown some delamination problems. Gold top electrodes have been patterned only onto those films derived from isopropanol based precursors solution, since they have better morphology and crystallographic structure. Another time, the capacity of the as made capacitors, have been tested using a capacimeter and the C-V and ε-V curves plotted, Fig.52. For this kind of capacitors, the thickness of the dielectric between the electrodes, has been considered to be 0,5 µm, since each dielectric film is made by two PZT layers of 0,25µm each, as demonstrated by optical profilometer analysis, Fig.45. When calculating the dielectric constant from the capacity values, the capacitor thickness is considered to be 1 µm, since the capacity is calculated for two equal capacitors in series. The electrodes area is as usual around 500*500 µm 2. a) b) Figure 52 : a) C-V and b) ε -V curves of Au/PZT/Pt capacitor fabricated using isopropanol based PZT precursors solution. Another time the C-V curve has a butterfly shape, confirming the ferroelectricity of these films. The curves are definitely more symmetric than those one of PZT films deposited on ITO, even if the positive wing is narrower then the negative and the positive values of capacitance are slightly higher than the negative. This fact is probably due to an higher electronic affinity between the Pt and Au electrodes, compared to that one of the ITO. The value of the dielectric constant at 0 V is approximately 770, that is an acceptable value for this kind of capacitor. Therefore, the PZT films deposited onto Pt electrodes possess the higher values of dielectric constant obtained in this work, as it was expected. The high values of capacity, if compared with those ones of PZT films deposited on Au electrodes, are probably due to the higher stability -20 -15 -10 -5 0 5 10 15 20 600 700 800 900 1000 1100 1200 1300 1400 1500 1600 1700 Volts [ V ] C a p a ci ty [ p F ] -20 -15 -10 -5 0 5 10 15 20 300 350 400 450 500 550 600 650 700 750 800 850 900 Volts [ V ] D ie le ct ri c co n s ta n t 90 of this substrate, in which the titanium diffusion from the TiO2 layer is very low and the Pt stability high. For this substrate, no intermetallic phases were present and this is beneficial for the electrical features of the deposited films. Another time the higher values of capacity are obtained around 2 V both for negative and positive bias. This voltage corresponds to an electric field of 20 kV/cm inside the ferroelectric and so a value of coercitive field (Ec) around 10 kV/cm. If confirmed, this value of coercitive field indicates that the deposited films behave as a '' soft ' ferroelectric.

91 Conclusions and further works:
A method to synthesize and deposit via solution deposition technique Lead Zirconate Titanate (PZT) thin films with a composition of Pb(Zr0.52Ti0.48)O3 has been developed in this work. This kind of material is widely used in the microelectronic industry to produce actuators, sensors, transducers and MEMS devices with high performances and so it''s fundamental to develop a production process that has to be the simplest as possible and that has to guarantee good films quality and high performances. Moreover the use of the less hazardous chemicals as possible is another important parameter that have to be taken into account when developing a production plan. First of all a new route to synthesize the PZT sol-gel precursors solutions has been developed and successively an appropriate thermal treatment to obtain fully crystallized films has been chosen. Using as metals precursors Lead acetate 3-hydrated, Zirconium propoxide and Titanium isopropoxide, a simple method to obtain an homogeneous and stable sol gel solutions has been developed. As solvent, acetic acid and water or acetic acid and isopropanol have been used, obtaining two different types of solutions. An improvement offered by this method is the use of solvents that are very less toxic than the commonly used 2-Methoxyethanol and the procedure to obtain the final sol-gel solution is definitely simpler. Following the procedures described in the thesis, homogeneous and stable sol gel solutions have been prepared, using both water and isopropanol as solvent. The acetic acid has in the chemistry of the prepared solutions both the role of solvent and chelating agent , that is fundamental to modify the metals alkoxides in solution, lowering their tendency towards hydrolysis and condensation and so preventing the solution from gelification and metals precipitation. The amount of acetic acid chosen was 5 time the moles of lead for the water based precursors solution and 25 times the moles of titanium for the isopropanol based precursors solution. With this composition the sol gels have shown high stability over long periods and the metals precipitation has been avoided. After the sol gels synthesis, the prepared solutions have been spin coated on three different kind of substrates: Au/TiW/SiO2/Si, Pt/TiO2/SiO2/Si and ITO coated glass. A proper thermal treatment has been performed on the deposited films and it includes a drying step at 150°C, a firing step at 370°C and an annealing step at 650°C. For the drying and firing steps the process time has been set at 5 minutes and the effects of different annealing times on the films crystallization have been studied. A proper thermal treatment must prevent any type of films cracking and elements diffusion from the substrate to the PZT layer, and it must assure a fully films crystallization into a ferroelectric perovskite phase. For PZT films deposited on gold electrodes the diffusion of elements, in particular of titanium, has been a great problem, since at the chosen annealing temperature it diffuses from the TiW adhesion layer towards the PZT film, forming Pb-Ti and Au-Ti intermetallic layers, that decrease the electrical features of 92 the films and induce micro-cracking. These problems appear after only 10 minutes annealing and so for this kind of substrate short annealing times have been adopted. For the other types of substrates used this problem has been avoided, since the ITO, the TiO2 and the platinum layers posses higher stability at the annealing temperature. Uniform, dense and crack free PZT films with a fully perovskite crystallographic structure have been obtained, tuning the annealing times, on all the three types of substrate used. More exactly, films with a good morphology and with a fully perovskite crystallographic structure have been obtained on gold electrodes using the 0,5M isopropanol based PZT precursors solution, and on ITO and platinum electrodes using both the two type of 0,5M solutions. On all the substrates, the PZT films crystallized into a polycrystalline structure but with a strong (110) preferential orientation, that is the most thermodynamically stable phase. Only for films deposited on gold electrodes, the crystals preferential orientation changed from the (110). When long annealing times have been performed, the PZT films derived from isopropanol based precursors solution, crystallized with a (111) preferential orientation and this fact is due to the formation of titanite and Ti-Pb or Ti-Au intermetallic phases, whose presence has been confirmed by XRD analysis. The formation of these structures on the bottom electrode surface, as discussed in other works [8] , favors the growth of crystals with a (111) preferential orientation, since the lattice parameters of these phases match with that one of the PZT (111) phase. The lattice parameters matching favors the kinetics of crystallization over the thermodynamics, and it leads to the growth of crystals with a (111) preferential orientation. Also for mono layer films derived from the 0,5M water based precursors solutions, the preferential orientation obtained on gold electrodes wasn''t the (110) but unexpectedly the (222), even when short annealing times were performed. Also in this case the substrate texture is supposed to have an important role on the films crystallization mechanism, since when a second layer is deposited, this one comes back to crystallize with a (110) preferential orientation. On the other hand, when films derived both from the water and the isopropanol based 0,5M precursors solution have been deposited on ITO coated glass or on Pt electrode, these films always fully crystallized with a strong (110) preferential orientation and no problems of films cracking or elements diffusion have been encountered. The thickness of a PZT mono layer was around 0,25 µm when deposited on Au or Pt electrodes, and around 0,2 µm when deposited on ITO coated glass, as demonstrated by SEM and optical profilometer analysis. Double layer films have been deposited on the three different substrates, using for each one those combinations of solution type and annealing time that guarantee the formation of fully crystallized films with good morphology. In particular, good quality deposits have been obtained on gold electrodes using the isopropanol based 0,5M precursors solution and choosing as annealing time 5 minutes to avoid titanium diffusion and films cracking, while on ITO and platinum electrodes, using both the two type of solutions and choosing as annealing time 15 minutes. Finally, top gold electrodes with an average area of 500*500 µm 2 have been patterned, using cold sputtering deposition, on the deposited double layer films, obtaining in this way a metal/PZT/metal structure that is suitable for electrical measurements. The 93 electrical characterization has been done analyzing the C-V and the ε-V curves of the capacitors fabricated. These curves have shown a non linear relation between the capacity and the applied voltage and a butterfly shape that is typical of ferroelectric materials. From these curves it can be concluded that all the films deposited behave as ferroelectric and so the scope of this work has been reached. The C-V curves were not always symmetric and this phenomenon is due to the using of different materials to build the top and bottom electrode. The highest asymmetry has been detected in the C-V curve of the Au/PZT/ITO capacitor, since the electrodes have very different working functions. Symmetric curves have been instead obtained for the Au/PZT/Au capacitor as expected. From the values of capacitance, the dielectric constants of these films can be extrapolated and the values found are 484 for PZT films deposited on Au/TiW/SiO2/Si, 655 for films deposited on ITO coated glass and 770 for films on Pt/TiO2/SiO2/Si substrate. The low dielectric constant of the films deposited on gold electrodes is attributed to the formation of intermetallic phases between the PZT layer and the substrate surface, that decrease the electrical features of the capacitors. The dielectric constant of films deposited on ITO is on the other hand quite high, if compared to the values found in literature that are around 300-400. The deposited PZT films have shown very good density, uniformity and adhesion on ITO substrates and these features play an essential role in the film electrical properties. Finally the dielectric constant of PZT films deposited on platinum electrodes has an acceptable value of ε=770, even if it''s a bit lower than those ones found in literature that are around 900. From the C-V curves is also possible to predict the values of the coercitive field (Ec) of the deposited ferroelectric films. Their values are around 10-15 kV/cm and only for those films derived from water based solution and deposited on ITO coated glass is 25kV/cm. These low values indicate that the films behave as ''soft' ferroelectrics. Further works should be done to confirm the values obtained from the electrical characterization and to calculate the hysteresis loop of these kind of films, from which it''s possible to obtain exact values of the remanent polarization (Pr) and of the coercitive field (Ec). Moreover the piezoelectric characteristic of these films have to be checked in order to have a complete understanding of the properties of these ferroelectric PZT thin films.
94 References:

[1] Julia Vivian Martinez, ''Fabrication, materials and characterization for Efficient MEMS power
generation', Thesis of Master of science in Materials science and Engineering, Washington State
University, School of Mechanical and Materials Engineering, August 2004.

[2+ Silvana Corkovic,' Piezoelectric Thick Films for Micro-electromechanical Systems Application', PhD
Thesis, School of Applied Sciences, Cranfield University, 2007.

[3] Taeyun Kim, ''Lead Zirconate Titanate (PZT) based thin film capacitors for embedded applications', PhD dissertation, Graduate Faculty of North Carolina State University, Department of Material Science and
Engineering, Raleigh, 2003.

[4] Brett Harold Piekarski, ''Lead Zirconate Titanate thin films for piezoelectrical actuation and sensing of
MEMS resonators ', PhD dissertation, Faculty of the Graduate School of the University of Maryland,
College Park, Department of Mechanical Engineering and Bioengineering Graduate Program, 2005.

[5] Brian Leclerc, ''Process Optimization for Sol-Gel PZT Films' , Master of Science Thesis, Department of Physics, Queen's University, Kingston, Ontario, Canada, March 1999.

[6] Irinela Chilibon, Jose N. Marat-Mendes, ''Ferroelectric ceramics by sol''gel methods and applications: a
review', Journal of Sol-Gel Science Technology , Springer Science Business Media, New York, 2012.

[7] Giacomo Gafforelli, ''Deposition of PZT on silicon substrate ', Ph.D. School of Materials Engineering: Material for Electronics Applications , Politecnico of Milano.

[8] Matthieu Cueff , ''Micro-actionneurs Piézoélectriques' , Thesis project, 'cole Doctorale Electronique,
Electrotechnique, Automatique et Traitement du Signal, Universitè de Grenoble, France, November 2011.

[9] Ling-Sheng Jang, Kuo-Ching Kuo, ''Fabrication and Characterization of PZT Thick Films for Sensing and
Actuation' , Sensors 7, pg. 493-507, 2007.

[10] K.D. Budd, S.K. Dey, and D.A. Payne, ''Sol-gel processing of PbTiO3, PbZrO3, PZT, and
PLZT thin films,' Proc. Ceramic Proc. 36, pg. 107-121, 1985.

[11] Guanghua Yi and Michael Sayer, ''An Acetic Acid/Water Based Sol-Gel PZT Process I: Modification of Zr
and Ti Alkoxides with Acetic Acid', Journal of Sol-Gel Science and Technology 6, pg. 65-74 1996.

[12] Guanghua Yi and Michael Sayer, ''An Acetic Acid/Water Based Sol-Gel PZT Process II: Formation of a
Water Based Solution', Journal of Sol-Gel Science and Technology 6, pg. 75-82 1996.

[13] Guanghua Yi, Zheng Wu and Michael Sayer, ''Preparation of Pb(Zr,Ti)O3 thin films by sol gel
processing: Electrical, optical, and electrooptic properties' , Journal Applied Physics, 64 (5), 1 September
1988.

[14] Yi-Chu Hsu, Chia-Che Wu, Cheng-Chun Lee, G.Z. Cao, I.Y. Shen, '' Demonstration and characterization of
PZT thin-film sensors and actuators for meso- and micro-structures', Sensors and Actuators A 116, pg.
369''377, Elsevier B.V, 2004.

[15] H.Brunckova, L. Medvecky, J.Briancin, K.Saksl, ''Influence of hydrolysis conditions of the acetate sol-gel
process on the stoichiometry of PZT powders '' , Ceramics International 30, pg. 453''460, Elsevier B.V,
2004. 95
[16] Yirong Lin, Clark Andrews and Henry A. Sodano , ''Enhanced piezoelectric properties of lead zirconate
titanate sol-gel derived ceramics using single crystal PbZr0.52Ti0.48O3 cubes', Journal of Applied Physics
108, 064108, American Institute of Physics, USA, 2010.

[17] Z. J. WANG, R. MAEDA, K. KIKUCHI, ''Development of phases and texture in sol-gel derived lead
zirconate titanate thin films prepared by three-step heat-treatment process', Journal of Materials Science
35, pg. 5915-5919, Kluwer Academic Publishers, 2000.

[18] Tien-I Chang, Jow-Lay Huang, Jen-Fu Lin, Bing-Huei Chen, Long Wu, Hong-Ping Lin and Horng-Hwa Lu,
''Effect of Drying Temperature on the Characteristics of the Lead Zirconium Titanate Powders Prepared by
Sol-Gel Process', Materials Transactions, Vol. 45, No. 11 (2004) pg. 3150 '' 3155, The Japan Institute of
Metals, 2004.

[19] S. Marino, M. Castriota, G. Strangi, E. Cazzanelli and N. Scaramuzza, ''Asymmetric nematic liquid crystal
cells containing lead zirconium titanate (PZT) films', Journal of Applied Physics 102, 013112, American
Institute of Physics, 2007.

[20] K. Stanly Jacob , N. Rani Panicker, I. Packia Selvam and V. Kumar, ''Sol-Gel Synthesis of Nanocrystalline
PZT Using a Novel System', Journal of Sol-Gel Science and Technology 28, pg. 289''295, Kluwer Academic
Publishers, 2003.

[21] Bing-Huei Chen, Long Wu, Ming-Cheng Chure, Yeong-Chin Chen, '' Fabrication of PZT by sol-.gel
method ', IEE E 978-1-4244-9821-5/10, 2010.

*22+ Cheung Mei Kuen, ''Preparation and Characterization of ferroelectric ceramic /ceramic composite film
for MEMS applications', Thesis project , Master of Philosophy in Physics, Department of Applied Physics,
Hong Kong Polytechnic University, Hong Kong, August 2003.

*23+Q.Zhang, Z.Huang and R.W.Whatmore, '' Studies of Lead Zirconate Titanate sol ageing. Part I: Factors
affecting particle growth', Journal of Sol-Gel Science and Technology 23, p.g. 135-144, Kluwer Academic
Publishers ,2002.

[24] Q.Zhang, Z.Huang and R.W.Whatmore, '' Studies of Lead Zirconate Titanate sol ageing. Part II: Particle
growth mechanisms and Kinetics ', Journal of Sol-Gel Science and Technology 24, p.g. 49-55 , Kluwer
Academic Publishers ,2002.

[25] Barbara Malic, Iztok Arcon, Alojz Kodre, and Marija Kosec, ''Homogeneity of Pb(Zr,Ti)O3 thin films by
chemical solution deposition: Extended x-ray absorption fine structure spectroscopy study of zirconium
local environment' , Journal of applied physics 100, 051612, American Institute of Physics, 2006.

[26] Weiguo Liu , Weiguang Zhu, ''Preparation and orientation control of Pb1.1( Zr0.3 Ti0.7)O3 thin by a modified sol''gel process', Materials Letters 46 ,p.g 239''243, Elsevier,November 2000.

[27] Reji Thomas, M. Shoichi , M. Toshiyuki, I.Tadashi, ''Preparation of ferroelectric Pb (Zr0.5,Ti0.5)O3 thin
films by sol''gel process: dielectric and ferroelectric properties' , Materials Letters 57, p.g 2007'' 2014,
Elsevier ,2003.

[28] L. Mengwei ,W. Jing ,W. Liding and Tianhong Cui, ''Deposition and characterization of Pb(Zr, Ti)O3 sol''
gel thin films for piezoelectric cantilever beams, Smart Materials Structure 16, p.g 93''99, doi:
10.1088/0964-1726/16/1/011, Institute of Physics publishing, 2007.

[29] M. Sayer, ''PIEZOELECTRIC THIN FILM DEVICES', Department of Physics, Queen's University 96 Kingston, Ontario, IEEE 1051-0117/91/0000-059, Ultrasonic symposium 1991.

*30+ M. Natali, D. Garoli, V. Rigato, F. Romanato, ''Sputtering Depostion of PZT Thin Films on ITO, Pt and Au
Electroded Substrates' , LNL Annual Report, Applied, General and Interdisciplinary Physics, Dipartimento di
Fisica dell'Università di Padova, Padova, Italy.

[31] Qin, H. Zhu, J. Jin, Z. & Wang Y., '' PZT thin films with preferred-orientation induced by external stress',
Thin Solid Films, 379, pg. 72-75, 2000.

[32] Sumio Sakka, ''HANDBOOK of SOL''GEL SCIENCE and TECHNOLOGY: Processing, Characterization and
Applications', VOLUME I: Sol-gel processing, volume editor: Hiromitsu Kozuka, Kansai University,Suita,
Osaka, Japan.

[33+ Guanghua Yi and Michael Sayer; ''SOL GEL PROCESSING OF THICK PZT FILMS' , Department of Physics,
Queen's University Kingston, Ontario, Canada.

[34] E. Alkoy ,S.Alkoy and Shiosaki, ''The effect of crystallographic orientation and solution aging on the
electrical properties of sol-gel derived Pb(Zr0.45 Ti0.55)O3 thin films, Ceramics International 33, pg. 1455-1462,
2007.

[35] K. Aoki Y. Fukuda, K. Numata and A. Nishimura, ''Dielectric properties of (111) and (100) lead-
zirconate-titanate films prepared by sol-gel technique', Japanese Journal of Applied Physics, JJAP 33, pg.
5155-5158, 1994.

*36+ K.Brooks,M.Reaney, R.Klissurka, Y.Huang, L.Bursill and N.Setter, ''Orientation of rapid thermally
annealed lead zirconate titanate thin films on (111) Pt substrates', Journal of Materials Research 9,
p.g 2540-2553, 1994.

*37+ Gong , Li, Chu, Li, '' Effect of Pyrolysis Temperature on Preferential Orientation and Electrical Properties
of Sol-gel Derived Lead Zirconate Titanate Films' , Journal of the European Ceramic Society 24, p.g 2977-
2982, 2005.

*38+ C. Kim, D. Yoon, J. Lee, C. Choi and W.Lee, ''Electrical characteristics of (100), (111), and randomly
aligned lead zirconate titanate thin films', Journal of Applied Physics 76, pg. 7478-7482, 1994.

*39+C. Jeffrey Brinker, George W.Scherer, '' Sol gel science: The Physics and chemistry of sol gel processing',
ISBN-13:978-0-12-134970-7, Academic Press Inc. ,1990.

[40] S.K. Pandey, A.R. James, R. Raman, S.N. Chatterjee, Anshu Goyal,Chandra Prakash, T.C. Goel,
''Structural, ferroelectric and optical properties of PZT thin films', Physica B 369 (2005) p.g 135''142,
Elsevier, 2005.

[41+ K. K. Uprety, L. E. Ocola, and O. Auciello, ''Growth and characterization of transparent Pb(Zi,Ti)O3
capacitor on glass substrate', Journal Applied Physics 102, 084107 , American Institute of Physics ,2007.





© Eiom - All rights Reserved     P.IVA 00850640186