Intros

You might also like

Download as doc, pdf, or txt
Download as doc, pdf, or txt
You are on page 1of 27

Chapter 1

Structural Phase Transitions: An Introductory Note

1.1 Introduction
Phase transitions are exhibited by a wide variety of systems from simple metals and alloys to complex inorganic and organic materials. Phase transition is an area of great academic interest and has proved to be of technological importance in material science [1-6]. Various aspects such as critical phenomena, soft modes, mechanisms and changes in properties at the phase transition have been discussed in several reviews and books [7-12]. Systems undergoing transitions are routinely discovered and study of their physicochemical properties lead to the development of new materials.

1.2 Definition
Any phase is characterized by thermodynamic properties like volume, pressure, temperature and energy. An isolated phase is stable only when its free energy is a minimum for the specified thermodynamic condition. The system is said to be in a metastable phase if it exhibits several local minima along with the global minimum and settles in one of the local minima [9]. As the temperature, pressure or any other variable like electric field or magnetic field acting on a system is varied, the free energy of the system changes continuously. Whenever such variations of free energy are associated with changes in structure of the phase (i.e. either electronic or spin configuration), a phase transformation is said to occur [10].

1.3 Thermodynamic Considerations


Changes in quantities like entropy and volume which represent discontinuities in the first derivative of Gibbs free energy follow the Clausius Clayperon equation [10]. These phase changes are referred to as first order transitions and are generally brought about by variation in temperature or pressure. In a second order transition, there is a discontinuity in the second derivative of the free energy such as heat capacity and compressibility. These include transitions where the heat capacity tends towards infinity at the transition temperature and are referred to as -transitions. Every phase transition is associated with a change in symmetry as well as in order. The concept of order parameter was introduced by Landau [13]. The average value of vanishes above Tc (critical temperature) in a second

order transition. Ubbelohde [14] has classified phase transitions broadly into two groups, continuous and discontinuous. In the former, there is no discontinuity in enthalpy at Tc and the crystal structure changes continuously from one polymorph to another.

1.4 Kinetic Considerations


Phase transition in many solids occur through the process of nucleation and propagation, each of these processes being associated with specific activation energy. The formation of the product phase in the matrix of the parent phase is called nucleation. This requires higher activation energy than the propagation step [15,16]. The theory of nucleation has been invoked to satisfactorily to explain the kinetics of phase transition. The transformation in TiO2 from anatase to the rutile form is an example of a nucleation-growth mechanism [16]. The empirical relation developed by Avrami [17] is an example of a kinetic expression to measure the overall rate of transformation.

1.5 Structural Considerations


While the thermodynamic treatment of phase transition is very fundamental and useful, it does not provide a structural insight into the microscopic changes accompanying a transition. Thus, an essential part of the study of a phase transition in solids [18,19] involves a detailed understanding of crystal chemistry in terms of atomic arrangements and bonding. A new phase obtained after a transition may be related to the parent phase in more than one way. The transition may be accompanied by a change in the primary coordination and/or secondary coordination and/or there could be major changes in the electronic structure and/or bond type. Hence, a detailed study of the structures of the parent and transformed phases, especially in terms of their orientational relation, becomes important to understand the transition mechanism. Buerger [20,21] classified phase transitions on the basis of structural changes involving primary or higher coordination as follows: (i) Transformation involving first coordination (a) Reconstructive (sluggish) (b) Dilatational (rapid) (ii) Transformations involving second or higher coordination 4

(a) Reconstructive (sluggish) (b) Displacive (rapid) Reconstructive transitions involve a major reorganization of the crystal structure, in which many bonds are broken and new bonds formed. An example is the reversible transition from graphite to diamond involving a complete change in crystal structure. Reconstructive transitions usually have high activation energies due to bond breaking and making and therefore proceed slowly. Transformations involving second coordination are reconstructive only if the mechanism involves breaking and forming of bonds of first coordination like in the different polymorphs of silica [10]. In dilatational transitions the extent of bond breaking is much less than for reconstructive type with no intermediate state of high energy and hence transition rates are rapid. Buerger proposed a simple mechanism for the transformation of CsCl or NH4Cl from the CsCl structure to the NaCl structure [22] as shown in figure 1.1.

Figure1.1: Dilatational mechanism for the transformation from CsCl structure to NaCl structure Displacive phase transitions involve the distortion of bonds and the structural changes that occur are usually small. Hence, these transitions take place with zero or very small activation energies, and a symmetry relationship usually exists between the parent and the transformed phase. Figure 1.2 illustrates the distinction between reconstructive and displacive phase transitions [10]. In order to convert structure A into any of the other structures B, C and D, bond breaking is essential and hence the transition is reconstructive. On the other hand, interconversions between structures B, C and D involve only small

rotational movements and no bond breaking. These transitions are hence displacive in nature.

Figure 1.2: Transformation from structure A to any other structure requires the breaking of first coordination bonds. Transformation among B, C and D are only distortions. Structural transitions are also classified into ferrodistortive (representing either no change in the number of formula units in the unit cell) or antiferrodistortive (change in the number of formula units in the unit cell) [7]. Ferrodistortive transitions can however be displacive or of order-disorder type . Such transitions are exhibited only by ferroelectric materials. Antiferrodistortive displacive transitions are shown by both ferroelectric and antiferroelectric materials.

1.5.1 Order-disorder transitions It is known based on third law of thermodynamics that only at zero Kelvin perfect order in solids is realized and the extent of order or disorder in a system thus becomes relevant at any other temperature. These transitions are generally classified as (i) positional disorder (ii) orientational disorder (iii) disorder of electronic (or nuclear) spin states. The configurational entropy due to disorder is given by S =Rln (II/I) where I and II are the total number of configurations in the product and parent phases respectively. An example of positional disorder is AgI, which undergoes a transition from a hexagonal to a cubic phase, resulting in a high ionic conductivity due to the randomization of Ag+ ions in the higher symmetry system. 1.5.2 Martensitic transitions Although this type of transformation was originally discovered in steel, they were considered to provide the mechanism for transitions in a variety of inorganic solids. A martensitic [23] transition is a structural change caused by atomic displacements corresponding to a homogeneous deformation which gives rise to an invariant strain plane. At this plane the parent and product phases are related by a precise orientational relationship [23].

1.6 Material Properties and Phase transition


The occurrence of a phase transition in a solid alters several material properties. Measurement of a particular property across the phase transition using sophisticated techniques provides insights into the nature of phase transition. Indeed, such changes in properties result in the development of new materials of technological interest. The following sections discuss various types of phase transitions and their correlation with material properties such as ferroelectricity and ferroelasticity.

1.6.1 Phase transitions in ferroics 7

A ferroic may be defined as a material possessing two or more orientation states or domains, which can be switched from one to another by the application of appropriate forces [24,25]. In a ferroelectric, the orientation state of spontaneous electric polarization can be altered by the application of an electric field and in a ferromagnet, the orientation state of magnetization in domains can be switched by the application of a magnetic field while in a ferroelastic, the direction of spontaneous strain in a domain can be switched by the application of a mechanical stress. Examples for these materials are BaTiO3 (ferroelectric), CrO2 (ferromagnetic) and CaAl2Si2O8 (ferroelastic). The electric polarization, magnetic polarization and elastic strain are the properties for which directionality changes in the above examples. Ferroics are classified into primary and secondary ferroics depending on the nature of the switchable property. Table 1.1 lists the different types of ferroic effects [7]. In primary ferroics the switchability involves the property (for example electric polarization) while in secondary ferroics, the switchability occurs on the derivative of the property (for example dielectric susceptibility). At hig temperature, ferroelectric materials transform to the paraelectric state, ferromagnetic to the paramagnetic state and ferroelastic to the paraelastic state. The transitions are characterized through order parameters [9] which are specific properties parameterized in such a way that the resulting quantity is unity for the ferroic state, below Tc and zero in the nonferroic phase beyond Tc. Whenever transitions are governed by the expected variations of these order parameters they are called proper ferroics or else the materials are termed as improper ferroics. Polarization, magnetization and strain are the proper order parameters for the ferroelectric, ferromagnetic and ferroelastic transitions respectively. Newnham and Cross have proposed a hexagonal representation (figure 1.3) of proper and improper primary ferroics [25]. The order parameter for proper ferroics appears on the diagonals of the hexagon, while the sides of the hexagon represent improper ferroics.

Table 1.1 Primary and secondary ferroics Ferroic property Primary ferroics Ferroelectric Ferromagnetic Ferroelastic Secondary ferroics Ferrobielectric Ferrobimagnetic Ferrobielastic Ferroelasoelectric Ferromagnetoelastic Ferromagnetoelectric Dielectric susceptibility Magnetic susceptibility Elastic compliance Piezoelectric coefficients Piezomagnetic coefficients Magnetoelectric coefficients Electric field Magnetic field Mechanical stress Electric field and mechanical stress Magnetic field and mechanical stress Magnetic field and electric field SrTiO3 NiO -quartz NH4Cl FeCO3 Cr2O3 Spontaneous polarization Spontaneous magnetization Spontaneous strain Electric field Magnetic field Mechanical stress BaTiO3 CrO2 CaAl2Si2O8 Switching field Example

Figure 1.3: Illustration of several types of order parameters in proper and improper ferroics. 1.6.2 Symmetry aspect in ferroics Ferroic phase transitions in crystals usually involve a change of the space group symmetry. It enables the classification of ferroic crystals in terms of their macroscopic physical properties such as polarization and strain. According to the Newnhams principle, any macroscopic physical property of a crystal must include all the symmetry elements of its point group [25]. Figure 1.4 illustrates the symmetry classification of phase transitions in crystals based on the work of several groups [26]. The term distortive is used for both displacive and orderdisorder phase transitions [27]. This can be divided into two categories: isomorphous and nonisomorphous In case of isomorphous phase transition, there is no change in the spacegroup symmetry of the crystal. A well studied example is the transition occurring in Ce at a pressure of 7.7kbar (0.77GPa), involving a volume decrease of about 14% leaving the space-group symmetry unchanged as Fm3m. The change of space group symmetry at a nonisomorphous phase transition can be either nonferroic or ferroic. In a nonferroic phase, transition a change in the translational symmetry is seen without change in the point-group symmetry [28]. The order-disorder phase transition occurring in the alloy Cu3Au where the symmetry changes from Fm3m to Pm3m is a classic example.

10

DISTORTIVE PHASE TRANSITIONS

ISOMORPHOUS
7.7kbar

NONISOMORPHOUS

Ce: Fm3m

Fm3m NONFERROIC FERROIC

Cu3Au: Fm3m Pm3m NONFERROELASTIC TGS: 2/m 2 SiO2: 622 32 FERROELASTIC BiVO4: 4/m 2/m BaTiO3, Mn3O4

Figure 1.4: A classification of phase transitions based on considerations of symmetry descent from prototype symmetry Ferroic phase transitions involve a change of the point-group symmetry with or without a change of the translational symmetry. If there is a change in the point-group symmetry but no change of the crystal system, the phase transition is nonferroelastic [29]. The 2/m(C2h) to 2(C2) transition (figure 1.4) in triglycine sulfate (TGS) is such an example. In general, it is seen that a change of crystal system is a necessary and a sufficient condition for a ferroelastic phase transition [30]. The crystal structure of BiVO4 [26] is an example of a purely ferroelastic transition while BaTiO3 [26] displays both ferroelastic and ferroelectric transitions. 1.7 Major Techniques Employed to Characterize Phase Transitions Structural phase transitions can be analyzed using several characterization techniques such as diffraction, dielectric measurements, spectroscopy, thermal analysis and microscopy. A brief description of these techniques is outlined.

11

1.7.1 Diffraction techniques The two important thermodynamic variables in the study of phase transitions are temperature and pressure. Any study of a phase transition would therefore involve measurement of properties as a function of temperature or pressure. Variable temperature X-ray diffraction studies of crystalline substances are useful in the study of phase transitions, thermal expansion and thermal vibrational amplitudes of atoms in solids. For temperatures as low as 80K liquid nitrogen cryosystem is used. High temperature studies are carried out using a graphite heating filament fitted with a thermocouple. High pressure apparatus employ a diamond anvil cell for the study of phase transitions. All these apparatus are commercially available. 1.7.2 Single crystal X-ray diffraction One of the best means of obtaining an accurate and detailed structural analysis of a crystalline solid is by single crystal X-ray diffraction technique. If a crystal is to be satisfactory for collecting X-ray diffraction data, two main requirements must be met. It must possess a uniform internal structure and must be of proper size and shape. This means that the crystal should not be twinned or composed of microscopic subcrystals. It should not be grossly fractured, bent, or otherwise physically distorted. It need not, however, have particularly uniform or well-formed external faces [31]. The morphology of the crystal can be screened rapidly and conveniently using a polarizing microscope. If rotated about an axis, normal to the polarizing material, the crystals should either appear uniformly dark in all positions or be bright and extinguish, i.e., appear uniformly dark, once every 90. Crystals which are made up of two or more fragments with different orientations will often reveal themselves by displaying both dark and light regions at one time. The ultimate evidence, however, of the internal structure of a crystal is furnished by the diffraction pattern itself. The reflections that appear should be single spots, without tails or streaks connecting them and should be uniquely indexable. The size of the crystal is normally of the order of 0.2mm3 for structures with light atoms whereas even smaller size crystals are suitable for diffraction if the compound has one or more heavy atoms. The structure determination from the single crystal diffraction involves

12

the measurement of intensity data from the reciprocal lattice, reduction and scaling of the data using well established data reduction procedures. Any structure determination has to be done in two stages because of the well known phase problem in crystallography. The determination of the phases of each reflection is performed using standard packages such as SHELXS[97] [32] and SIR97 [33], which allow the use of either direct methods or the Patterson methods. The starting model thus derived can be refined using the standard packages that are available for example on a program suite like WinGX [34] as described in subsequent chapters. 1.7.3 Powder X-ray diffraction It is not always possible to obtain good quality crystals of suitable size for single crystal structure determination by X-ray diffraction. In fact traditionally, phase transitions have been studied mainly via powder X-ray diffraction techniques. With the advent of the Rietveld method [35], the refinement of the structures has received an enormous impetus. A brief description of the Rietveld method followed by a discussion on the abinitio approach is given below. 1.7.3.1 The Rietveld method In the Rietveld method, the least-squares refinements are carried until the best fit is obtained between the observed powder diffraction pattern taken as a whole and the calculated pattern [36]. A powder diffraction pattern of a crystalline material may be thought of as a collection of individual reflection profiles, each of which has a peak height, a peak position, a breadth, tails which decay gradually with distance from the peak position, and an integrated area which is proportional to the Bragg intensity, Ik, where k stands for the Miller indices, h,k,l. Ik is proportional to the square of the absolute value of the structure factor. The diffraction pattern is recorded in digitized form i.e. as a numerical intensity value, y i, i representing the step value in 2 [36]. Depending on the method, the increments may be either in scattering angle 2, or some energy parameter such as velocity (for time of flight neutron data) or wavelength (for X-ray data collected with an energy dispersive detector) and an incident beam of white X-radiation. Typical step sizes range from 0.01 to 0.050 in

13

2 for fixed wavelength X-ray data on a conventional powder X-ray diffractometer. With the advent of synchrotron radiation, it is now possible to record the pattern at step sizes of ~ 0.0050 in 2. The number of steps in the powder diffraction pattern is usually in thousands (e.g. Number of data points for a scan range of 3 to 80 in steps of 0.02 is 3850). The quantity minimized in the least-squares refinement is the residual, Sy : Sy = wi (yi-yci)2 Where wi = 1/yi yi = observed intensity at the ith step yci = calculated intensity at the ith step Typically, many Bragg reflections contribute to the intensity, yi; observed at any arbitrarily chosen point i in the pattern. The calculated intensities yci are determined from the k F 2 values calculated from the structural model by summing up the calculated contributions from neighboring (i.e. within a specified range) Bragg reflection and the background. yci = s LK |FK|2 (2i - 2K)PKA+ybi where s is the scale factor, K represents the Miller indices, h,k,l for a Bragg reflection, Lk contains the Lorentz polarization and multiplicity factors, is the reflection profile function. Pk is the preferred orientation function, A is an absorption factor, Fk is the structure factor for the Kth Bragg reflection, ybi is the background intensity at the ith step. In a number of available computer programs for the Rietveld method, the ratio of the intensities for the two X-ray K wavelengths (if used) is included in the calculation of k , F 2 so that only a single scale factor is required. The model parameters that may be refined include not only the atomic parameters like positional, thermal and site-occupancy but also parameters for the background, lattice, instrumental geometrical - optical features, specimen aberrations (e.g. specimen displacement and transparency), an amorphous component and specimen reflection. Although it is in general a much less severe problem

14

with powders than with single crystals, extinction can be quite important in some powder specimens. Multiple phases may be refined simultaneously and comparative analysis of the separate overall scale factors for the phases is probably the most reliable method for doing quantitative phase analysis. The Rietveld method is a powerful tool, but it is limited by the same drawback that affects powder methods in general. There is loss of information that arises from the compression of the three-dimensional diffraction pattern into a single dimension. It is also important to mention that the Rietveld method, though an excellent technique for refining structures, requires a good starting model if it is to converge successfully. Standard packages such as GSAS [37], FULLPROF [38], DBWS [39] and RIETAN[40] perform Rietveld refinements for single as well as multiple phase in a routinely and user-friendly manner. 1.7.3.2 The abinitio approach There has been a great deal of interest concerning the determination of unknown structures from powder diffraction data during the last decade and there have been several reviews on the subject [41-44]. The process may be broken into several steps. An essential prerequisite for crystal structure determination is that the lattice parameters and the space group are known. Determination of the lattice parameters from the powder diffraction pattern requires accurate determination of the peak positions (accurate d- spacing data), which can normally be achieved using a peak-search process, provided all systematic errors have been eliminated. Although in some cases the lattice parameters can be determined from first principles, it is usually necessary to use an autoindexing program such as ITO, TREOR or DICVOL [45-47]. In general these programs generate several possible sets of lattice parameters that are consistent with a set of measured peak positions; a variety of figures of merit can be used to rank the proposed sets of lattice parameters. The space group is assigned by identifying the conditions for systematic absences in the indexed powder diffraction data. From the unit cell parameters and the selected space group, the positions of the reflections in the diffraction pattern can be calculated. The diffraction intensity associated with each reflection can then be determined by applying a whole-profile fitting technique similar to that used in a Rietveld refinement but with the intensities of the reflections rather than the structural parameters. This procedure, known as intensity

15

extraction, can be performed using either a least-square method (Pawley 1981) [48] or an iterative approach (Le Bail et all.1988) [49]. Even if integrated intensities are not used in subsequent steps, this procedure is still necessary to establish the appropriate profile parameters for whole-profile applications. Those reflections that are too close to one another to be considered independently (i.e. strong overlapping reflections) were earlier ignored or assigned arbitrary (often equal) contributions to the total intensity of the overlapping set. However, this approach was clearly unsatisfactory, and more sophisticated approaches for determining reliable relative intensities from overlapping peaks have been developed [41]. These pattern decomposition techniques are incorporated in a number of programs including ALLHKL, WPPF [50], GSAS [37], FULLPROF [38], LSQPROF [51] and EXTRA [52]. As most approaches for structure solution from powder diffraction data depend heavily on extracting reliable intensity information, pattern decomposition constitutes an important step dictating the overall success of these approaches. Methods are currently being developed to allow the relative intensities of such overlapping peaks to be determined accurately and include the applications of relations between structure factors derived from direct methods and the Patterson function (DOREES) [53]. This is an iterative procedure involving the calculation of a squared Patterson map and the subsequent backtransformation giving a new set of structure factors for the overlapping reflections. Another program (FIPS) [54] is a method based on entropy maximization of a Patterson function and a Bayesian fitting procedure. The result of the above procedure is a pseudo single crystal dataset (i.e. a list of hkl and Ihkl) which can be subjected to (a) adaptations of singlecrystal techniques, (b) direct-space methods that exploit prior chemical knowledge and (c) hybrids of the two to arrive at the satisfactory structural model. The final and often most time-consuming step in the structure determination maze is the completion of the structure (eg. finding any missing atoms by Fourier analysis, resolving disorder problems, etc) and the refinement of the structural parameters using the Rietveld method. Only when the refinement has been brought to a successful conclusion can the structure proposed from the structure determination step be considered to be confirmed. Throughout the whole procedure, chemical information and intuition play an important role in guiding the user through the maze [41].

16

1.7.4 Neutron diffraction Thermal neutrons with a velocity of about 4000ms-1 associated with a wavelength of~1.0 are used in neutron diffraction experiments. Variable temperature neutron powder diffraction experiments are used to study phase transitions in many materials. Whereas Xrays are scattered primarily by the electrons in atoms, neutrons are scattered mainly by the atomic nucleus. Since the neutron-scattering amplitude does not show a smooth dependence on the atomic number of the atoms, neutron diffraction is particularly useful in locating light atoms in crystals. Additional scattering of neutrons can arise owing to the magnetic moment of neutrons. In the absence of an external field, the magnetic moments of atoms in a paramagnetic crystal are arranged at random, so that the magnetic scattering of neutrons by such a crystal is also random. It only contributes diffuse background to the diffraction pattern. In magnetically ordered materials however, the magnetic moments are regularly aligned. Neutron diffraction provides an experimental means whereby the magnetic structures can be determined. In addition to the two scattering effects that are elastic, neutrons can also undergo inelastic scattering by crystals. This involves energy exchange between the lattice and neutrons. Inelastic neutron scattering by crystals is used to study quantized vibrational modes and dynamics in solids [55]. Since neutron beams are much weaker in intensity than X-rays, neutron diffraction requires large single crystals. However, in recent years, powder neutron diffraction analysis has also been used to obtain structural information [56]. The Rietveld method described in earlier sections is also used for analysis of the powder neutron data [35]. The profile analysis is particularly suitable for neutron data since the profiles could be accurately described by a Gaussian function. In case of structures consisting of atoms which are near neighbors in the periodic table, it is recommended to use neutron diffraction rather than X-ray diffraction techniques. In general, neutron diffraction is employed as a surrogate technique along with either single crystal or the powder X-ray diffraction techniques. 1.7.5 Electron diffraction Electron beams have small wavelengths than X-rays and carry charge. Hence, electron diffraction is a valuable technique for structural studies of solids. These are the small

17

wavelength of electron beams and charge carried by them. Smaller wavelength leads to smaller Bragg angles in electron diffraction. The radius of the Ewald sphere, 1/, is therefore much larger for electron diffraction than X-ray diffraction. This makes it possible to record extensive sections of the reciprocal lattice with a small stationary crystal. Due to the charge, interaction of electrons with atoms is about 103 times stronger than that of Xrays and this makes it possible to record electron diffraction patterns almost instantaneously. Electron diffraction patterns are readily obtained with commercial electron microscopes. It is possible to investigate defect ordering, the Bravais lattice type, superstructures and fine particle sample by electron diffraction. However, it has the disadvantage of having secondary diffraction effects that severely limit its application as a stand alone structure determination technique. The requirement of very high vacuum also becomes a serious rate limiting step especially for the study of phase transition. 1.7.6 Dielectric measurements The dielectric properties of a material are governed by its response to an applied electric field at the electronic, atomic, molecular and macroscopic levels. Application of a potential difference across a dielectric leads to polarization of charge within the material although long range motion of ions and electrons cannot occur. The polarization disappears when the voltage is removed. Polarization and dielectric loss in materials are phenomena of interest, and thus are generally studied as a function of frequency. Piezoelectrics, pyroelectrics, ferroelectrics, paraelectrics, ferroelastics all fall under a broad class of dielectrics since most of the materials exhibiting these properties undergo transitions with changes in polarization on application of an electric field. Ferroelectric materials retain a large, residual polarization of charge after the electric field has been removed. One of the important characteristics of a ferroelectric is that the dielectric constant obeys the Curie-Weiss law, e = r 1 = 3Tc/T-Tc where e is the susceptibility of the material, r is the relative dielectric constant of the material and Tc is the critical or Curie temperature. A number of displacive, order-disorder and hydrogen bonded ferroelectric sulfates have been evaluated for their pyroelectric behavior. For example, ferroelectricity was observed in A3H(SO4)2 (A=K,NH4,Rb) through

18

dielectric measurements [57,58]. A conspicuous increase in the Curie temperature and enhancement in the spontaneous polarization by deuterium substitution was found in (NH4)H3(SO4)2 [12]. 1.7.7 Spectroscopic techniques A number of spectroscopic methods are available for the analysis of structure and dynamics during phase transition. Most of these techniques yield valuable information regarding the nature of phase transition. Magnetic transitions in solids can be studied by Mossbauer spectroscopy and magnetic resonance spectroscopy (NMR, ESR and NQR). Inelastic neutron scattering provides valuable information on phonons and magnons in crystals.The slow vibrational motion of H atom in RBHSO4 governed by the reoriented motion of the sulfate ion through the riding model mechanism due to the isotope effect is studied by this technique [59]. NMR spectroscopy has been employed to study phase transitions of solids containing the appropriate nuclei as in the case of NaCN and NaHS [9]. Studies of hindered rotations of CH3 or NH4+ groups and phase transitions in hydrogenbonded ferroelectrics like KH2PO4 are other important applications of NMR spectroscopy. ESR spectra of solids undergoing transitions have been reported in the literature [9]. NQR spectroscopy has been employed to study phase transition of halides, nitrates and nitrites containing nuclei with quadrupole moments [9]. Raman spectroscopy is specially used for investigating soft modes [9]. The changes in electronic structure can be followed by X-ray and ultraviolet photoelectric spectroscopy. Optical spectroscopy is used for studying phase transformations, particularly with respect to the movement of boundaries, growth of nuclei and changes of grain size [9]. Spectroscopic techniques like mass spectrometry, atomic absorption spectrophotometry and electron microprobe analysis are used to confirm the purity of the material which is crucial to study phase transitions [9]. 1.7.8 Thermal analysis Thermal measurements have been widely used to identify and characterize transitions. Heat capacity measurements provide precise enthalpy changes and indicate thermodynamic order. The two main thermal analysis techniques are thermogravimetric analysis (TGA) which automatically records the change in weight of a sample as a function of either

19

temperature or time and differential thermal analysis (DTA) which measures the difference in temperature, T, between a sample and an inert reference material as a function of temperature. A technique that is closely related to DTA is differential scanning calorimetry (DSC), where, the equipment is designed to allow a quantitative measure of the enthalpy changes that occur in a sample as a function of either temperature or time. The H values obtained here are more reliable. A fourth type of thermal analysis technique is dilatometry, in which the change in linear dimension of a sample as a function of temperature can be recorded. Thermal changes can be followed on cooling as well as on heating. It is evident that if a particular process, on heating, is endothermic, then the reverse process on cooling must be exothermic. This enables us to study reversibility of any phase transition. Another phenomenon is the thermal hysteresis that occurs when the exotherm may be displaced to occur at lower temperatures than the corresponding endotherm. Hence hysteresis is more in transitions involving breaking of strong bonds.

1.7.9 Electron microscopy Electron microscopy is an extremely versatile technique capable of providing structural information over a wide range of magnification. It gives useful information on dislocations and structural aspects. High resolution lattice imaging of transforming solids is an area of great potential. The main mode of operation makes use the fact that when a sample is placed in the microscope and bombarded with high energy electron, X-rays characteristic of the elements present in the sample are generated. By scanning either the wavelength (wavelength dispersive,WD) or the energy(energy dispersive, ED) of the emitted X-rays it is possible to identify the elements present. By a suitable calibration a quantitative elemental analysis may be made for elements heavier than sodium. The scanning electron microscopy (SEM) allows for preliminary verification of the purity of the phase along with transmission electron microscopy (TEM). Other techniques include use of electron probe microanalysis (EPMA), electron microscopy with microanalysis (EMMA) and analytical

20

electron microscopy (AEM) to gain insights into phase transitions A recent advance is the development of the scanning transmission electron microscope (STEM). This combines the scanning feature of the SEM with the intrinsically higher resolution obtainable with TEM. SEM is invaluable for surveying materials under high magnification and providing information on particle sizes and shapes. Using thin foils with TEM (transmission electron microscopy), crystal defects such as dislocations, stacking faults, anti phase boundaries may be seen directly. With HREM (high resolution electron microscopy), it is now possible to it to view details on an atomic scale. Variations in local structure such as site occupancies and vacancies can be observed directly; this is especially useful in studying compositional phase transitions by doping.

1.8 Foreword
The subsequent chapters report the detailed phase transition analysis mainly using X-ray diffraction techniques. The materials investigated belong to the family of sulfates and complex oxides. The following preamble is an account of the existing literature related to phase transitions in these materials. 1.8.1 Sulfates The alkali and alkaline earth metal mixed sulfates have been investigated for exhibiting properties such as ferroelectricity, pyroelectricity, ferroelasticity and proton ion conduction. These properties have been characterized by dielectric, optical, spectroscopic and X-ray diffraction measurements. The compounds which have been subjected to these studies are A3H(SO4)2 (A= K, Cs, Rb, NH4) [60-70], AHSO4 (A=NH4, K, Cs, Rb), A4LiH3(SO4)4 (A= NH4, K, Rb, Cs) and AB(SO4)3 (A=Rb,K,NH4,Tl; B= Ca, Cd, Mg, Mn, Zn), ABSO4(A=Li, K, Na, Cs; B=NH4, Li, Rb) [71-74]. The family of ABSO4 (A=Li, K, Na, Cs; B=NH4, Li, Rb)[74-77] and AHSO4 (A=NH4, K, Cs, Rb)[78-80] have been shown to exhibit ferroic properties and possess extensive hydrogen bonding network. For example, the crystal structure of LiNH4SO4 has been studied in detail and the ferroelastic to ferroelectric phase transitions characterized by modulated differential scanning calorimeter, capacitance and ac thermal measurements [81,82]. 21

1.8.1.1 Rubidium hydrogen sulfate The ferroelectric transition in NH4HSO4 was observed by dielectric measurements and attributed to the character of the N-HO bond [83] and hence it was surprising to discover that RbHSO4 [79], which is isomorphous with NH4HSO4 at room temperature, is also ferroelectric below -15C. Dielectric and thermal measurements confirmed this phase transition [84]. Preliminary X-ray diffraction reports showed that the room temperature paraelectric phase crystallized in a pseudo orthorhombic system (space group B21/a), while the ferroelectric phase crystallized in a noncentrosymmetric (space group Pc or P1). The room temperature crystal structure of RbHSO4 was determined by X-ray and neutron diffraction [85]. There was evidence for disorder of one of the sulphate groups and the hydrogen atoms were ordered in the paraelectric phase. In another report [86] the disordered structure of RbHSO4 at room temperature was redetermined by assuming that one of the two independent sulfate ions is disordered

. Figure 1.5. Pressure-temperature phase diagram of RbHSO4. A high pressure phase of this compound has also been observed by dielectric constant measurements and the crystal structure at 1GPa has been analyzed [87]. The pressure temperature phase diagram of RbHSO4 is shown in figure 1.5.

22

1.8.1.2 Langbeinites Langbeinites are a class of sulfates having the general formula (A+1)2(B2+)2(SO4)3, where A+1=K,Rb,NH4, Tl and B2+=Zn,Co,Cd,Mn or Mg [88]. The structural phase transitions in these compounds were first characterized by Dvorak [89], using Landau theory. According to Dvoraks theory, the langbeinite type crystal could transform from the parent cubic phase (space group P213) to any one of the four space groups P2 1, P1, P212121 and R3. Langbeinites are classified into three categories based on their successive phase transition sequence as shown below where P and F represent paraelectric and ferroelectric or ferroelastic phases respectively [90]: Type ICubic (P213) (P) Monoclinic (P21) (F)* Triclinic (P1) (F)* Orthorhombic (P212121) (F) * This type has one or two intermediate phases followed by the lowest phase. Eg. Tl2Cd2(SO4)3, Rb2Cd2(SO4)3, K2Zn2(SO4)3, K2Co2(SO4)3 [91] Type II Cubic (P213) (P) Orthorhombic (P212121) (F) Eg. K2Mn2(SO4)3, (NH4)2Cd2(SO4)3 Type III There are no phase transitions eg. K2Ni2(SO4)3 Several Type II structures have been analyzed by single crystal X-ray diffraction and the mechanism of phase transition has been attributed to the simultaneous translation and rotation of the SO4 group and the subsequent rearrangement in the B2+ and A+ cation positions [88,92]. It is of interest to note that so far there is only one detailed single crystal X-ray diffraction study followed by a subsequent phase transition analysis in literature for type I langbeinite Tl2Cd2(SO4)3 [ 93]. Rb2Cd2(SO4)3 also belongs to the typeI langbeinite as observed from dielectric studies[94]. This compound undergoes successive ferroelectric phase transition at -144C and -170C [94]. Preliminary studies suggest the presence of super lattice reflections in both these phases [95]. The lattice parameters were measured as a function of temperature and the phase transitions were confirmed by piezoelectric resonance, EPR and dielectric measurements [95-97]. The phase transition pathway indicated in Tl2Cd2(SO4)3 is not supported by any of these measurements.

23

1.8.1.3 Tetra rubidium lithium tri hydrogen tetra sulfate (Rb4LiH3(SO4)4) The crystals of the common formula A4LiH3(SO4)4 were found to be tetragonal at room temperature by X-ray diffraction and crystal morphology studies [98]. Pyroelectric, thermal and elastic studies of Rb4LiH3(SO4)4 show the existence of a phase transition at 137K [99].

Figure 1.6: Plot of nc against temperature for Rb4LiH3(SO4)4 crystal Linear birefringence measurements (figure 1.6) indicate a linear dependence of the morphic birefringence n on the order parameter confirming that the transition in Rb4LiH3(SO4)4 is a proper ferroelastic phase transition [100]. The Curie temperature (Tc) for this compound and its dueterated analogue are determined to be at 122.44K and 108.43K respectively [100]. The crystal structure of an optically laevorotatory crystal was determined in the space group P41 [101]. Dielectric and DTA measurements show anomalies at 458K, 470K and 490K indicating structural phase transitions. The behavior of the dielectric constant [102] was found to be very similar to that of a superionic conductor. 1.8.1.4 Tetra potassium lithium tri hydrogen tetra sulfate (K4LiH3(SO4)4) Low temperature pyroelectric and dielectric behavior of K4LiH3(SO4)4 show an anomaly at 114K [99]. EPR studies show a change in the line intensity with temperature relating to the existence of a phase transition at 110K [103].

24

Figure 1.7. Raman spectra of K4LiH3(SO4)4 crystal at (1) 90K (2) 300K Polarization Z(XX)Y Further thermal expansion investigations do not show any anomalous changes suggesting that either no phase transition related with change of the lattice parameter occurs in K4LiH3(SO4)4 or the relative length induced by these changes is less than 10-6 [99]. Brillouin scattering technique used to study elastic properties in the temperature range 90 to 300K confirm the nonferroelastic [104] character of the phase transition in this compound. Raman investigations unambiguously show the existence of a phase transition (figure 1.7) near 120K and possible changes related to the dynamics of the proton on the central H-bond were conjectured [105]. However the microscopic mechanism of the phase transition governing could not be determined. 1.8.2 Oxides 1.8.2.1 Aurivillius phases The complex bismuth oxides with layered structures were discovered by Aurivillius more than 50 years ago [106]. These compounds can be described by the general formula [Bi2O2] [An-1BnO3n+3] (n=1,2,3,4) where the [Bi2O2]2+ layers are interleaved with n perovskite-type layers having the composition [An-1 Bn O3n+1] [107]. To date, an intensive search will yield more than 100 compounds in this group, which have been characterized

25

by dielectric, single crystal and powder X-ray diffraction, Raman spectroscopy, electron diffraction and conductivity measurements [108-112]. Representative members are Bi2WO6 (n=1), Bi3TiNbO9 (n=2), Bi4Ti3O12 (n=3), Bi5Ti3FeO15 (n=4) (figure 1.8) [113]. All the compounds in this family are tetragonal (I4/mmm) and centrosymmetric at high temperature but transform to different polar groups on cooling to room temperature. The crystal structure of Bi3TiNbO9 and Bi4Ti3O12 was determined by single crystal X-ray diffraction methods and it was found that both crystallized in the orthorhombic system with different space groups [113,114]. Further studies describe the modulated structures of both these compounds in greater detail and accuracy [115,116]. The structure and property of the higher members of the series (n=4 and above) has also been a subject of interest as they exhibit ferroelectric properties in thin films [117].

Figure 1.8: Idealized structures of : (a) Bi2WO6 (n=1); (b)Bi3TiNbO9 (n=2) (c) Bi4Ti3O12 (n =3) projected along a axis [108]

26

1.8.2.2 Crystal structures of ABi4Ti4O15 (A=Ba, Sr and Pb) and phase transitions at high temperature Ferroelectricity has been observed in thin films of the n=4 Aurivillius oxides [117]. Raman studies on these oxides show under damped soft modes in SrBi4Ti4O15, low frequency over damped modes in BaBi4Ti4O15 in the vicinity of the ferroelectric to paraelectric transition [118]. In PbBi4Ti4O15, the softening of the under damped modes was not observed clearly. The reported Tc for these compounds is 803K for BaBi4Ti4O15, 693K for SrBi4Ti4O15 and 843K for PbBi4Ti4O15. Recently a variable temperature neutron diffraction study was reported for SrBi4Ti4O15 [119]. It is observed that these oxides exhibit no spontaneous polarization along the c axis. Thin film studies on BaBi4Ti4O15 show that the macroscopic ferroelectric properties of these layered oxides depend on the crystalline orientation of the films [120]. 1.8.2.3 Structural studies on lanthanum doped n=2 Bi3TiNbO9 The lanthanum substituted bismuth titanates have shown a high fatigue free resistance which find application in ferroelectric random access memories (FRAM) [121]. It has been observed previously that substitution of the trivalent Bi ions with increasing concentrations of trivalent La ions in Bi2-xLaxTiNbO9 result in smaller orthorhombic distortion [122]. At the maximum substitution rate for Bi2-xLaxTiNbO9 corresponding to x=1, it was observed that the value of the dielectric constant is reduced drastically in the vicinity of the transition. These results have been correlated to the change of symmetry with increasing concentration of La. In the analogous Bi2SrTa2O9, where the A site is occupied by the divalent Sr ion, the reduced orthorhombic distortion observed is attributed to the high BVS at that site [123]. The marked preference for Bi to occupy a highly distorted coordination site is complicated by the presence of the stereochemically active lone pair [124].

27

You might also like