Chapter 4 Pulsed Laser Deposition - Basic Mechanisms and Diagnostics 4.1 Introduction Pulsed laser deposition (PLD) is an extremely versatile technique for growing thin films of a wide variety of materials. The materials it may be applied to have profoundly different properties and include metals, insulators, semiconductors and superconductors" ®, The charm of the technique is in its simplicity. A pulsed laser beam (usually from an excimer laser) is directed into a vacuum chamber and is incident on the material to be deposited. Each laser pulse ablates a few hundred Angstroms of the target surface, causing a bright plume of material to be ejected normal to the surface of the target. The plume impinges on a substrate, which may be heated, positioned a few centimetres away from the target. Depositing the film is a relatively fast process - for most materials, it typically takes only 30 minutes to deposit a film approximately 1 pm thick. Perhaps the most important advantage of PLD over all other deposition techniques is its ability to transfer the composition of the target material directly to the deposited film, This stoichiometric transfer is a characteristic feature of the PLD process, and makes it ideal for growing thin films of photorefractives, which are usually complicated multicomponent oxides. Under the right experimental conditions, PLD also offers the potential for epitaxial thin film growth. This is vital for the deposition of photorefractive crystals since, in order for the deposited film to exhibit photorefractive properties, it must be a single crystal as well as of the correct stoichiometry. Furthermore, the growth of single crystal films does not require a single crystal target, and the target may be fabricated from the raw material in its relatively inexpensive powder form. Clearly then, PLD has a number of key features that make it ideally suited to the growth of photorefractive thin film optical waveguides. BBChapter 4: PLD - Basic Mechanisms and Diagnostics 4.2 Historical Review The initial proof of concept for using a laser as a tool for thin film deposition was carried out by Smith and Turner in 1965 Si 6) ysing a ruby laser. Thin films of a variety of materials, including semiconductors, dielectrics and organometallic compounds were deposited with some degree of success. As new types of lasers became available the experiments were repeated using CO, and Nd:YAG lasers. However, limitations due to the wavelength (absorption depth and photon energy), and power density of the lasers resulted in limited material selection, and produced films of inferior quality to those grown by more conventional techniques. The first attempt to grow a ferroelectric film was carried out in 1969 by Schwartz and Toutellotte working on BaTiO, ), but no systematic research was carried out to optimise the growth conditions. During the 1970's the Q switch was developed which allowed a reduction in laser pulse width and peak powers of > 10° W to be obtained. These improvements enabled the observation of congruent evaporation (see section 4.3.1), which is the key to the transfer of the target composition to the film. In the late 1970's reliable excimer lasers became available which provided both high energy density and ultraviolet wavelengths, allowing the disassociation of even very stable chemical bonds. By the 1980's PLD was used to deposit high quality epitaxial semiconductor films, comparable to those grown by the more conventional process of molecular beam epitaxy" *), Considerable progress was also made in the growth of oxides and ceramics, which exhibited improved crystallinity compared with e-beam deposited films"), In 1987 Dijkkamp deposited the first thin film of a high temperature superconductor using the PLD technique"? *?, where all other processes had failed to reproduce the complicated stoichiometries of these ceramics. This provided the necessary incentive for many other research workers and led to an explosive growth in the number of publications in the field, with the successful deposition of over 100 different inorganic and organic materials being reported in the following few years. In 1989 single crystal thin films of the ferroelectric BaTiO, were epitaxially grown onto LiF substrates, using the PLD technique™* ®, The extension of PLD to ferroelectrics was a natural 14Chapter 4: PLD - Basic Mechanisms and Diagnostics progression from the work on superconductors because many of the photorefractive ferroelectrics are structurally related to high T,, superconductors. However no attempt ‘was made to characterise the optical and waveguide properties of the films. The work in this thesis started in October 89 and was prompted by this extremely high interest in PLD, and its aim was to develop the PLD technique to epitaxially grow thin film, crystalline, optical waveguides, with primary emphasis on photorefractive materials. 4.3 Basic Principles and Physical Effects The theoretical description of PLD is highly complex and draws on many disciplines of science, and for this reason the calculations required to model it are highly complex and can be performed only numerically. An overview of PLD models has been given by Bauerle®"="*”), Even the most complicated models are still unsatisfactory because of the great number of parameters involved (some of which are unknown), and none of the models to date explains all the observations. For these reasons, we will discuss only simple models here, in order to qualitatively illustrate the three main underlying, physical stages in PLD. These are: the laser-target interaction, the plume-laser interaction and the film growth. A basic understanding of these three stages is important, since the processes that occur within them affect the quality and composition of the films grown. 4.3.1 Laser-Target Interaction We shall consider a simple physical description for the laser-target interaction in which material is removed from the target by several processes, as illustrated in Fig. 4.1. The laser radiation is absorbed by the surface of the target where the photons’ energy causes surface heating. The increase in the surface temperature, which can easily exceed the melting point of ceramics (~ 150°C), depends on the optical penetration depth of the material, the thermal conductivity of the target and the rate at which the photons are supplied®"*"*®, As the laser pulse continues to irradiate the target the molten surface layer is evaporated. Congruent evaporation, in which all the components of the target material are effectively liberated simultaneously, is achieved because the laser energy is supplied to a small volume (107 m) in a short time (typically 20 ns). This ensures that the ejected material has the same stoichiometry as the target material. 15Chapter 4: PLD - Basic Mechanisms and Diagnostics Target Time Incident _laser_beam ~20ns Absorption Thermal conduction Surface melting Vaporisation Tonisation Plasma expansion Liquid expulsion Self - regulating Fig. 4.1 Schematic illustration of the processes involved in the laser target interaction. 76,Chapter 4: PLD - Basic Mechanisms and Diagnostics Many studies have shown that there is a threshold energy density for the removal of target material by ablation®“* %), ‘This threshold is approximately 0.7 J/cm? for BaTiO, at 308 nm" ®) and decreases with increasing absorption coefficient, which is in tun wavelength and material dependent. The typical shape of the graph of deposition rate versus laser flux is shown in Fig. 4.2. Above the ablation threshold the graph shows that material removal varies linearly with laser fluence until a plateau is reached, which for BaTiO, is typically around 2 J/em?. The ablation rate in the linear region of the graph is also dependent on wavelength, material and the duration and shape of the laser pulse. The ablation threshold can be used to explain the observation of two distinct components to the laser plume‘V*'**"®), The ablated plume of stoichiometric material is highly cone shaped, and is characterised by a forward peaked distribution with a cos"® dependence, where 9 Epp (the strong binding case) a layer-by-layer mechanism is dominant. The type of mechanism obviously depends on the choice of substrate and film. Both mechanisms rely on an initial nucleation process, which takes place when the deposited thickness is less than a molecular layer. In this process, islands of deposited film of approximately 100 A diameter form nucleation sites. The number of islands reaches a saturation density and subsequent deposition leads to the growth of the nuclei until they coalesce. ‘The basic difference between the mechanisms is where the nucleation takes place. In the layer-by-layer mechanism, the nucleation takes place on substrate lattice sites, and the film grows taking on the lattice structure and orientation over the entire layer. However, in the case of the island growth mechanism, most of the nucleation takes place on substrate defects, and islands of deposited film grow out from these nuclei with differing orientation from island to island. When the epitaxial growth is by the strong, bonding mechanism obviously the role of the substrate is more important and the lattice mismatch between the film and substrate, chemical bonding type and crystal symmetry type must be taken into consideration. A certain amount of mismatch between the crystal lattice constants can be accommodated by uniform elastic strain, and other crystal defects such as dislocations and slip. The amount of mismatch that can be 80Chapter 4: PLD - Basic Mechanisms and Diagnostics tolerated depends on many parameters but in particular, the film thickness. For thicknesses of a few um, misfits of <5% are commonly acceptable Tom=rrt¥y®, If the bonding is weak and epitaxy is by the island growth mechanism then the strict requirements on the substrate are relaxed and epitaxial growth with misfits > 13% are possible ans 9), It has been observed that most of the multicomponent oxides follow a complex mixture of the two growth mechanisms®"“” ®), Epitaxial films of these compounds are often characterised by a mosaic microstructure, indicative of their formation from separate islands, but with a significant degree of preferred orientation which implies a layer-by- layer mechanism. 4.4 Experimental Technique ‘The pulsed laser deposition work in this thesis was carried out at several laboratories - the Rutherford Appleton Lab, UK., the European Ultraviolet Laser Facility, Crete and at Southampton University. For this reason the development of the experimental arrangement was very much an iterative process, with modifications made to the apparatus each time the experiment was set up. Much of the experimental technique section will describe the various modifications and improvements made during the course of the work, but unless otherwise stated, the discussions that follow apply generally to all the experiments carried out. ‘The basic experimental arrangement was developed from the set-up used by Dijkkamp for the deposition of high T, superconductors" §? and the work of Davis on the growth of BaTiO,©"*"™, The basic experimental arrangement for the PLD work in this thesis is shown schematically in Fig. 4.3, and a photograph of the pulsed laser deposition of BijzGeOzq is shown in Fig. 4.4. An excimer laser beam was focused using ‘a 40 cm lens through a quartz window, into a vacuum chamber capable of a base pressure of 10° mbar. The particular specifications of the lasers used at the various laboratories are given in Table 4.1. The laser beam was incident on the target material at an angle of 45° and caused a plume of highly energetic material to be ejected normal to the target surface. (The angle at which the plume was ejected was observed to be 81Chapter 4: PLD - Basic Mechanisms and Diagnostics Excimer laser Fig. 4.4 Photograph taken during the pulsed laser deposition of BiyGeOpo, 82Chapter 4: PLD - Basic Mechanisms and Diagnostics Table 4.1 —— Laser Energy per | Pulse Width | Repetition ] Wavelength | Laser Pulse Rate (am) (mJ) (as) (Hz) Rutherford 308 ~100 20 20 Appleton Laboratory Foundation for 248 ~200 20 20 Research and Technology - Hellas Southampton 248 ~200 20 40 [ vsivesiy independent of the angle of incidence of the laser beam.) The ejected material impinged on a substrate, held parallel to the target plane and a few centimetres from it. The substrate could be heated, as will be described in section 4.4.2. Prior to mounting the substrates on the heater, they were wiped with acetone and methanol. Ultrasonic cleaning in ‘micro’ surfactant was also performed but this did not significantly improve the quality of the films deposited, and appeared to etch specific substrates such as LiF and MgO. 4.4.1 Target Preparation ‘The targets used for PLD were mainly pressed powder pellets but also off-cuts of bulk crystals, crystal rods and bulk glasses. In order to make the pressed powder pellets, a stoichiometric powder was mixed with a few drops of distilled water and pressed, using a pressure of 10* kg/em?, into disks of 13 mm diameter by 2 mm depth. The pellets were then sintered in a tube furnace with a flow of oxygen. The oxygen was provided to avoid problems of oxygen loss at elevated temperatures. The sintering temperature 3Chapter 4: PLD - Basic Mechanisms and Diagnostics was chosen to be typically ~ 100°C below the melting point of the various crystals, so that the powder did not melt completely but could coalesce. The furnace temperature was slowly ramped up at a rate of 400°C/hour to the sintering temperature, which was then maintained for 2 hours before cooling at 200°C/hour. Once sintered the pellets were fairly robust and did not crumble on handling or ablation. The targets fabricated were of a single homogeneous crystalline phase in order to avoid production of microstructure on the target surface which can lead to particulate deposition (section 43.1). The various target materials were moved relative to the laser beam during deposition to avoid deflection of the laser beam. If the laser is stationary on the target surface a groove is cut as the laser beam ablates material. This groove can be up to ~ 1 mm deep after only ~ 10,000 pulses and is cut into the target surface at the angle of incidence of the laser beam (in this case 45°). Since the plume is ejected normal to the target surface, this groove causes the plume to be deflected by up to 20°. This deflection is disadvantageous because it causes the film to be deposited from the wings of the plume rather than ftom its highly forward directed ablated component. This leads to a much lower deposition rate and the films may be non-stoichiometric. ‘The targets were moved relative to the laser beam using several techniques. Fig. 4.54 shows how the target pellets were mounted so that they rotated eccentrically, thus ensuring maximum ablation of the surface area, and also minimising the formation of microstructure on the target due to the laser beam. Fig. 4.5b illustrates how crystal rods. and stacks of pellets were mounted so that the laser beam cut a spiral pattern around the target edge. Both 4.5a and 4.5b maintain the position of the ejected plume relative to the substrate. In 4.5e, the laser beam is stepped or scanned across the target surface, and consequently in this arrangement the position of the plume is moved relative to the substrate. This actually proved to be advantageous since it served to improve the thickness uniformity of the deposited film in the direction of the scanning, as illustrated in Fig. 4.6a and 4.6b and observed in section 5.5. No detrimental affects to the stoichiometry of the deposited films were seen, for the small distances (<1 em) which the laser was scanned over the target. As a precaution against contaminants, a flag 84Chapter 4: PLD - Basic Mechanisms and Diagnostics Target motion (a) O Excimer —_—e beam (b) a ——— (c) Fig. 4.5 Several schemes to ensure ablation of maximum target surface area: a) pressed powder pellets were rotated eccentrically to produce a spiral pattern, b) crystal rods and stacks of pellets ‘were rotated and translated to produce a helical ablation pattern, c) bulk crystals and glasses were held stationary while the laser was stepped across the target surface. 85Chapter 4: PLD - Basic Mechanisms and Diagnostics {a) Target rotated (b) Laser stepped vertically Fig. 4.6 Contour plots of the variation in thickness of the deposited film at the substrate. 9) laser stationary relative to substrate and b) laser moved vertically relative to substrate. could be positioned in front of the substrate prior to deposition. This allowed the target to be preablated before the deposition of the film, in order to remove any target surface impurities. 4.4.2 Substrate Heating The successful deposition of crystalline films by epitaxial growth requires substrate heating to ensure that the deposited material has sufficient energy to relocate and reorientate on the substrate crystal lattice, as discussed in section 4.3.3. Several methods of heating the various substrates were used, the simplest of which was the filament heater. In this design a wire filament was wound around a stainless steel block held underneath a stainless steel plate. The filament used in this case consisted of an inconel metal current-carrying core, capable of withstanding high temperatures (=1100°C) and currents <10 A, insulated by a ceramic powder and contained in an inconel sheath. The temperature of the heater plate was varied by altering the currentChapter 4: PLD - Basic Mechanisms and Diagnostics supplied to the filament, and using this system, temperatures of up to 750°C could be reached in vacuo. In general the heater needed to be as compact as possible to minimise thermal radiation to the rest of the vacuum chamber. A high power (250 W) light bulb was also tried in place of the filament, but the system was found to be too thermally massive and heated the whole chamber to over 250°C. The substrate was attached to the heater plate by means Of small clips, and the temperature of the hot plate was monitored using a thermocouple. The heater plate had to be as flat and free from defects as possible to ensure uniform thermal contact between the substrate and heater plate. Using a simple analysis, the temperature of the front face of the substrate was calculated, for the two extreme cases of perfect and negligible thermal contact. Since the films were grown under vacuum, heat transfer by convection could be neglected, and only conduction and radiation need be considered. A schematic representation of the heater is illustrated in Fig. 4.7. Considering the case of poor thermal contact (when Chamber Ve e o Substrate Heater | _Heat__ — oe! transfer transfer —- |, & & _—_ 4 a Bh Fig. 4.7 Schematic representation of the heat transfer between the heater plate, substrate and vacuum chamber. 87Chapter 4: PLD - Basic Mechanisms and Diagnostics the heater is in close proximity to, but not in contact with the substrate), then no thermal conduction takes place and heat transfer is by thermal radiation alone. ‘According to Stefan’s Law, the net radiated power per unit time (AQ/A‘) from the heater plate (at a temperature Ty), to the back face of the substrate (at a temperature 7;) is given by aa. : ae = ATaencapttnne-m an where A is the area of the substrate and o is Stefan’s constant. The term containing the emissivities of the heater plate (¢,), and the substrate (¢,) takes into account the repeated radiation and absorption of heat from each of the surfaces. The heat flow through the substrate from the back face (at temperature T;), to the front face (at temperature T;) is given by the equation of heat conduction 4.2) where k is the thermal conductivity and d is the thickness of the substrate. Assuming that the system is in thermal equilibrium, then the net radiation from the heater plate, given by Eq.(4.1), can be equated to the thermal conduction through the substrate, given by Eq.(4.2), and therefore we have (,-T) (4.3) Z 4.3) €¢,0(Tg-T;) = k- 1 1-i-e)(l-e) Similarly, the thermal conduction through the substrate can be equated to the net radiation loss from the front face of the substrate, 88Chapter 4: PLD - Basic Mechanisms and Diagnostics 1 -—1 _ - 4.4) ieee ™ Gy Here e; is the emissivity, and Tp is the temperature of the vacuum chamber. The temperature of the front face of the substrate (7;) can therefore be calculated by solving equations 4.3 and 4.4 simultaneously. For the case of perfect thermal contact then the heater plate and the back face of the substrate are considered to be in thermal equilibrium and in this case Jy = T, in the above equations. Fig. 4.8 is a graph of 7) versus T; for the two extreme cases of perfect and negligible thermal contact, and shows how much the front face of the substrate can vary compared to the temperature of the heater plate. The implication of this graph is that for a measured heater plate temperature of T>=550°C (the optimum growth temperature for Bij,GeO,,) the actual temperature at the front face of the ZrO substrate, T, can be up to 120°C lower. ‘Temperature of Front Face of Substrate 1000 b) 200! ° 200 400 600 800 1000 Temperature of Heater Plate Fig. 4.8 Graph of temperature of the front face of a ZrO, substrate (7) versus the temperature of the heater plate (7;), for the two extreme cases of a) perfect and b) negligible thermal contact. Values of the parameters used in the calculation were as follows: €;=€,=0.9 (steel), €2=0.67 =1 mm, k=10 Wm"K" (Zr0,) and 0=5.67x10% Wm?K*. 89Chapter 4: PLD - Basic Mechanisms and Diagnostics If the substrate heating is non-uniform the extremely temperature-dependent nature of the epitaxial growth can cause effects such as that shown in Fig. 4.9. Any temperature gradients are exaggerated because the substrates used were poor thermal conductors, and consequently the non-uniformities could not be evened out. Fig. 4.9 illustrates how a small bump on the heater plate has caused a ‘hot spot’ on the substrate and has resulted in an area of transparent film growth, while the rest of the film has grown at too low a temperature (as indicated by the darker areas). Fig. 4.9 Photograph of a Bij,GeOz on ZrO, film, highlighting the extremely temperature- dependent nature of epitaxial growth. The circular area of transparent growth is the result of a hot-spot on the heater plate. To avoid the problems of poor thermal contact and non-uniformity a second heating method was tried, in which a CO, laser was used to heat the substrate from behind. The profile of the CO, laser beam was Gaussian, which obviously posed a problem if the heating was to be uniform across the whole substrate. The CO, laser beam was reflected using a gold coated mirror positioned outside the chamber, consisting of four square sections which could be angled independently. Using this mirror the beam was divided into quarters and then overlapped to form a homogenised ‘top hat’ function, as shown in Fig. 4.10. A ZnSe window, which was transparent at the CO, laser wavelength ofChapter 4: PLD - Basic Mechanisms and Diagnostics Segmented (a) (b) Fig. 4.10 Illustration of the reflection technique for homogenising a Gaussian laser beam. a) mapping of beam contour through system; b) mapping of beam profile through system. (The calculated 3-D profile was generated by Dr S. W. James of Southampton University). a1Chapter 4: PLD - Basic Mechanisms and Diagnostics 10.6 um, was provided in a port of the vacuum chamber, to allow the homogenised CO, laser beam to be directed onto the substrate. The CO, laser used was a 15 W laser which obviously could deliver considerably less power than that supplied by the filament heater (250 W). At a temperature of 600°C, the power radiated from a typical substrate was calculated from Stefan’s law to be 45 W. The CO, laser could therefore easily provide excess energy to overcome this loss and heat the substrate as long as the thermal conduction to, and thermal radiation from the substrate mount was small. To avoid further heat loss a thermocouple could not be used to measure the temperature as it would simply act as a heat sink. The temperature therefore had to be measured remotely using a pyrometer. (This instrument measures the amount of thermal radiation emitted from a specimen and using black body radiation theory, infers its temperature.) However, in practice using a pyrometer presented two main problems. Firstly the pyrometer calibration depends on accurately knowing the surface emissivity of the sample. Obviously, since the emissivity of the surface was not constant during deposition, the pyrometer could not be calibrated properly. Secondly, the area analyzed by the pyrometer was =25 cm? compared to the 1 cm? of the substrates, and the measured temperature was therefore an average over all the radiating surfaces within the pyrometer acceptance angle (such as the vacuum chamber walls). The temperature of the substrate was altered by adjusting the power output of the CO, laser. This was continuously variable from 1 W to 15 W by switching of the RF power supply with a variable mark-space ratio square-wave. The small thermal mass of the substrates meant that even small changes in the laser power produced large changes in temperature (typically ~50°C), and it was difficult to maintain a constant substrate temperature using CO, heating. (If it were not for the problems with using the pyrometer for temperature measurement a feed-back loop from the laser power supply would have been used to stabilise the substrate temperature). For the above reasons it was found that the best results were achieved using the filament heater, provided the heater plate was completely flat and defect free. 4.5 Characterisation Techniques ‘The films grown by pulsed laser deposition were analyzed using a variety of methods 92‘Chapter 4: PLD - Basic Mechanisms and Diagnostics to determine their stoichiometry, crystallinity, phase, orientation, uniformity and morphology. As the techniques used to analyze the-films are all well-established tools for material characterisation, they will only be outlined here. 4.5.1 Scanning Electron Microscopy The scanning electron microscope (SEM) is primarily used as an imaging tool to determine the surface and near-surface structure of a specimen on a sub-micron scale(S>~dhew 88) In our case, an SEM was used to examine the morphology and topography of the deposited thin films. The basic principle of operation is similar to that of an optical microscope, except that the SEM uses electrons rather than photons to perform the imaging, and electrostatic and magnetic deflection rather than lenses as the focusing elements. This allows improved resolution, magnification and depth of field compared to an optical microscope. The typical resolution of an SEM is ~2 nm (which is limited by the minimum size of the probe beam) whereas the typical resolution of an optical microscope is only ~500 nm. This improved resolution allows magnifications of 16 000X to be achieved in the SEM, as opposed to only 1000X for an optical microscope. Possibly the most striking difference between the SEM and optical microscope is the depth of field which can be attained. For typical values of aperture and working distance, and at a magnification of 1000X, the depth of field in an optical microscope is ~ 1 ym, whereas for an SEM it is =40 um. ‘ Since the electrons which facilitate the imaging are charged particles, insulating specimens can charge up, which produces distorted images. To prevent this the specimens are coated with a thin (20 nm) conducting layer which can be either gold or carbon. This means that the samples are useless for optical applications unless the coating can be removed without damaging the sample. For this reason only small pieces of the deposited films were analyzed, leaving the majority of the sample available for further analysis. The region of the sample into which the primary electrons penetrate is known as the interaction volume, as shown in Fig. 4.11. The interaction of the primary electrons with the sample produces several detectable signals which can be used for analysis purposes 93Chapter 4: PLD - Basic Mechanisms and Diagnostics Incident beam (primary electrons) Secondary electrons r Interaction volume Fig. 4.11 Interaction volume of the primary electrons from an SEM with the sample, illustrating some of the detectable signals emitted. and the regions of the sample from which these are emitted is also shown in Fig. 4.11. It is the secondary electrons which are primarily used to perform the imaging in an SEM, but backscattered electrons can also be used to obtain a subtlely different type of ‘image of the sample surface. 4.5.2 Energy Dispersive X-ray Analysis The interaction of primary electrons with a specimen in the SEM, also produces characteristic x-rays which can be used to perform chemical analysis of a specimen. Energy dispersive x-ray analysis (EDX) relies on the fact that each element produces a characteristic x-ray signature, and by taking an EDX x-ray spectrum of an unknown compound, the wavelength of each characteristic peak in the spectrum may be used to identify the elements present” ®, Measurement of the number of x-ray photons 94Chapter 4: PLD - Basic Mechanisms and Diagnostics detected at each wavelength also allows quantitative analysis of the chemical composition of the sample. In modern EDX machines the recording and processing of the x-ray spectra is computerised, which allows contour maps, line scans as well as localised points of the specimen to be analyzed. Unfortunately, in practise, quantitative analysis may be inaccurate for the following reasons: 1) The x-rays produced by the direct electron excitation can themselves excite atoms which then emit characteristic ‘softer’ x-rays of a slightly lower energy. This fluorescence produces additional anomalous downshifted peaks in the x-ray spectrum. 2) A significant fraction of the x-rays produced are absorbed in the sample and therefore not detected. The absorption and therefore the number of x-rays collected is dependent on the energy of the electron-beam, the wavelength of the x-ray studied and the local atomic weight of the sample. The complexity of the relationship between these factors may make fitting of models to the collected spectra inaccurate. 3) Elements lighter than sodium cannot be detected by standard instruments because the soft x-rays produced from these elements are absorbed by the detector window. 4) As illustrated in Fig. 4.11, the sampling volume for x-rays is significantly larger and deeper than for secondary electrons, and in practise it is difficult to make the sampling volume for x-rays less that 1 zm?. This means that the chemical analysis is an average over this volume, and EDX scans of the sample volume may yield different elements to those imaged by the SEM. In the particular case of the thin films grown by PLD their thickness was typically 1 um. Consequently, although the SEM imaged the surface and no more then 0.1 um below it, by virtue of having a larger sample volume the EDX analysis showed elements known to be present in the substrate material. EDX analysis was therefore deemed unreliable if elements were common to the substrate and film. '5) A common practical problem is caused by the topography of the specimen. X-rays propagate rectilinearly, and if the thickness of the specimen is significantly uneven then the x-rays from certain regions of the sample may have trajectories that fall outside the detector solid angle, giving ambiguous readings. 95Chapter 4: PLD - Basic Mechanisms and Diagnostics 4.5.3 Rutherford Backscattering Rutherford backscattering (RBS) has its origin in the initial experiments by Rutherford on the size of nuclei. In a typical RBS analysis of a sample, 2 MeV He ions are accelerated onto a 1 mm? area of the specimen, and are elastically scattered by the target atoms via the Coulombic repulsion between the target nuclei and the ions‘""”), By measuring the energy of the backscattered ions the various component atoms of the thin film can be inferred. The results are plotted as counts versus energy as shown in Fig. 4.12. For thin samples (<500 A) the peaks for the various types of atoms can be resolved (Fig. 4.12a). However if the film is greater than 500 A then the scattering which occurs from atoms beneath the surface becomes significant. In this case ions (a) a=] a > energy (b) energy Fig. 4.12 Typical RBS data characteristic of a) a thin film (<500 A) and b) a solid. 96‘Chapter 4: PLD - Basic Mechanisms and Diagnostics scattered from beneath the surface lose energy in travelling through the material, and will therefore have a lower energy than those scattered from the surface. The net result is a spectrum with a high energy edge corresponding to surface scattering, with a continuous plateau to the lower energy side (Fig. 4.12b). A determination of the film stoichiometry and thickness can be performed by fitting computer simulations to the experimental data. Generally, the height of the peaks or plateaus corresponds to the number of atoms of that type present, and the width of the peaks or plateaus represents the thickness of the film. The positions of the edges or peaks identifies the type of atoms present in the sample. If the sample being analyzed is an insulator, as was the case for all the thin films deposited by PLD, then the incident ions can cause the sample to become charged. This can lead to the incident ions being deflected by the previous ions trapped on the surface, rather than by the Coulombic repulsion of the nuclei, and can therefore lead to misleading results. To avoid charging of the sample, the film can be coated in a thin layer of carbon, but of course this layer will show up in the analysis. 4.5.4 X-ray Diffraction Texture Camera In order to determine the extent to which the deposited films were crystalline an x-ray texture camera was employed®*** °), In this technique the crystal lattice acts as a grating to diffract the incident x-rays. Samples of size ranging from 100A thick to bulk crystals can be investigated. Fig. 4.13 shows a schematic representation of the x-ray texture camera. The incident x-ray beam is parallel to the central axis of the cylindrical film holder, and the sample can be rotated at an angle ¢ to the incident x-ray beam. The positions to which the x-rays are diffracted on the camera film allows the determination of the lattice constants, and the definition of the diffraction maxima gives information about the crystallinity of the sample. Amorphous material is usually manifest by a general darkening over the central region of the pattern. If the sample is polycrystalline then the whole Bragg cone is recorded, and when the photographic film is opened flat, this cone appears as a straight line. Preferred orientation in polycrystalline films is indicated by changes in intensity along the Bragg cone while for single crystal thin films, sharp diffraction maxima (spots) rather than lines are recorded. 97Chapter 4: PLD - Basic Mechanisms and Diagnostics Cylindrical film holder Fig. 4.13 Schematic representation of an x-ray texture camera. 4.5.5 X-ray Diffractometer ‘The x-ray diffractometer is the most commonly used technique for routine crystalline phase analysis®°®**, Conventionally, the diffractometer is arranged as illustrated in Fig. 4.14, with an x-ray beam incident onto a 2 mm? area of the sample at an angle 0, with the detector at an angle 20 to the incident beam. The detector rotates at twice the rotational speed of the sample so that the diffracted x-ray beam is always aligned on the detector aperture as each diffraction peak is detected. A monochromator is placed between the detector and the aperture so that only characteristic x-rays are recorded. ‘The output trace of beam intensity versus Bragg angle is therefore produced by scanning the detector and recording its response. The measurement of 26 for each diffraction peak and calculation of the lattice d spacings from Braggs’ law md = 2dsind (4.5) (where m is the diffraction order and } is the x-ray wavelength) enables the constituent phases to be identified. 98Chapter 4: PLD - Basic Mechanisms and Diagnostics — X-ray \ Monochromator \ So Tr ~ \ / SMM_L_____ Specimen Detector Fig. 4.14 Schematic representation of an x-ray diffractometer. 4.6 References Avery A. J. Ion Beam and X-ray Analysis of Thin Films One day course, Royal Military College of Science, Swindon, 14th Nov. (1991) Barr W. P., J. Phys. E, 2, 2 (1969) Bauerle D., Appl. Phys. A 48, 527 (1989) Bauerle D., Luk'yanchuk B., Schwab P., Wang X. Z. and Arenholz E. Laser Ablation of Electronic Materials European Materials Research Society Monographs 4, 39 North- Holland (1992) Chan C.L. and Mazumder J., J. Appl. Phys. 62, 4579 (1987) Cheung J. T., Niizawa G., Moyle J., Ong N. P., Paine B. M. and Vreeland Jr. T., J. Vac. Sci. Technol. Ad, 2086 (1986) Cheung J.T. and Sankur H., CRC Critical Reviews in Solid State and Material Sciences 15, 63 (1988) Cheung J.T. and Horwitz J. Pulsed Laser Deposition: History and Laser-Target Interactions MRS Bulletin XVIIQ), 30 (1992)Chapter 4: PLD - Basic Mechanisms and Diagnostics Davis G. M. and Gower M. C., Appl. Phys. Lett. 55, 112 (1989) Dijkkamp D., Venkatesan T., Wu X. D., Shaheen S. A., Jisrawi N., Min-Lee Y. H., McLean W. L. and Croft M., Appl. Phys. Lett. 51, 619 (1987) Foltyn S, R., Dye R. C., Ott K. C., Peterson E., Hubbard K. M., Hutchinson W., Meunchausen R. E., Estler R, C. and Wu X. D., Appl. Phys. Lett. 59, 594 (1991) Goodhew and Humphreys Electron Microscopy and Analysis Taylor and Francis (1988) Knox B. E, and Ban V. S., Mater. Res. Bull. 3, 337 (1968) Kukuey V. L., Sevostanov M, A. and Tomashpolsky Y. Y., Phys. Chem. Mech. Surface 3, 2315 (1985) van der Merwe J. H., CRC Critical Reviews in Solid State and Material Sciences 7, 209 (1978) Rogers K. D. lon Beam and X-ray Analysis of Thin Films One day course, Royal Military College of Science, Swindon, 14th Nov. (1991) Sankur H., Appl. Opt. 25, 1962 (1986) Schwarz H. and Tourtellotte H. A., J. Vac. Sci. Technol. 6, 373 (1969) Smith H. M, and Turner A. F., Appl. Opt. 4, 147 (1965) Tomashpolskii Y. Y. and Sevostianov M. A., Ferroeletrics 29, 87 (1980) Venkatesan T., Wu X. D., Inam A. and Watchman J. B., Appl. Phys. Lett. 52, 1193 (1988) Wang S. Z., Xiong G. C., He Y. M., Luo B., Su W. and Yao S. D., Appl. Phys. Lett! 59, 1509 (1991) 100