This chapter describes the process of sputter deposition and the experimental arrangement used to obtain the highly magnetostrictive FeSiBC films used in this study. It also describes the methods used to study the films magnetically and to characterise them, along with the a section on thermal heat treatments employed.

The sputtering process consists of the bombardment of the target material by fast moving, heavy, inert gas ions from a plasma. The bombarding ions cause atoms to be ejected from the target material by momentum transfer between the colliding ions and the target atoms. The process is schematically shown in Figure 3.1a, where a number of processes are shown to occur when the ions collide with the target material. Some of the bombarding ions are reflected back and are neutralised, but may still be sufficiently energetic to reach the substrate were the film is being deposited. This can be a source of substrate bombardment (back scattering) which can effect the resulting properties of the film. The majority of the colliding ions tend to induce sputtering by ejecting atoms of the target material by momentum transfer. This is a secondary collision process, as shown schematically in Figure 3.1a. The ejected atoms will have random directions but, as discussed in Chapter 4, the sputtering process can induce texture in the resulting films due to the sputtering conditions. Secondary electrons which are emitted either join the oscillating plasma gas, which cause the continuous ionisation of the gas to sustain the incident ions needed for sputtering, or they liberate their energy in the form of heat on colliding with the substrate or other parts of the chamber. The sputtered target atoms which are deposited at the substrate form the resulting thin film.

The basic processes occurring at the surface of the substrate are shown in Figure 3.1b. The mobility of the incident atoms arriving at the substrate is highly dependent upon the sputtering parameters (pressure and power), the temperature of the substrate, the distance between target and substrate, and the surface

Figure 3.1: An illustrative diagram showing the sputtering process at the target (a), and film formation at the substrate (b).

of the substrate itself. Any surface defects or texturing of the substrates can effect the mobility of the incident atoms and will act as a barrier, and this can be reflected in the resulting properties of the film. The incident atoms with sufficient mobility will diffuse to join other incident atoms to form islands which will continue to grow until they coalesce to form a continuous film (see Wagendristel et al (1994) for details). Other species of particles besides the target atoms can also bombarded the substrate, which can influence the growth of the film. The neutral reflected atoms and gas particles are particularly difficult to control, as they cannot be manipulated using electric or magnetic fields. These particles can have sufficient energy on arrival at the substrate such that they sputter the film or become incorporated into the film itself. The bombardment of these particles can be controlled by working at pressures where the mean free path of the particles is small compared to the target-substrate separation, or by working at lower sputtering powers. In addition to these neutral particles, the substrate can also suffer bombardment from reflected gas ions. Control of these particles can be achieved by applying a bias to the substrate; this is known as bias sputtering. This effect has been extensively utilised in the study of magnetic films (Chapter 4). All particles impinging on the substrate, including electrons, will liberate energy in the form of heat, increasing the temperature of the substrate which effects the mobility of the target atoms and hence the depositing film.

In this study all magnetic films were grown by Radio Frequency (RF) magnetron sputtering using a Nordiko NM2000 system. The system is schematically shown in Figure 3.2. The sputtering system was configured to operate in the sputter-up mode, where the substrates were mounted 6cm directly above the target electrode. The sputter-up configuration had the advantage of not suffering from the problem of eroded target fragments falling on the substrate electrode and thus affecting the depositing film. Both the substrate and target electrodes were 15cm in diameter and were water-cooled. The Nordiko NM2000 consisted of three target electrodes (referred to as 1,2,3) which were mounted on a rotatable carousel. This provided the facility to sputter from three different target materials separately, and therefore allowed the deposition of multi-layered films. A grounded stainless steel shielding plate ensured that only the target below the substrate electrode was being sputtered from. In this study all films were grown using target electrode 1 unless otherwise stated. The importance of this is discussed in Section 5.8.1. The main chamber is attached to a mechanical hoist in order that the vacuum chamber could be lifted clear of the base of the machine to provide access for the mounting of targets and general maintenance. Two Viton seals were used to provide a vacuum seal for the main chamber as shown Figure 3.2. The rotatable shutters were positioned to provide shielding for either the target or substrate from the plasma. This allowed the target or substrate to be sputter-cleaned prior to the deposition of the film. A stainless steel clamping ring was used to secure the target to the copper electrode which ensured good electrical and thermal contact. To prevent sputtering from the clamping ring itself, an over-sized earthing shield was mounted directly above to prevent the plasma from coming into contact with the ring. The earthing shield also prevented the sputtered material from being deposited onto the insulator which isolated the target electrode from ground. Substrates were mounted

Figure 3.2: A schematic representation of the Nordiko NM2000 sputtering system used to sputter deposit the amorphous thin films.

on a removable copper substrate platter which interlocked firmly into the substrate electrode with a copper gasket which insured good thermal and electrical contact.

The power from the RF amplifier could be directed either to the target or the substrate electrode by means of a relay switch through a matching circuit. This allowed the impedance of the power supply and electrode to be matched, so the required power could be delivered efficiently. Directing the power to the substrate electrode allowed the substrates to be sputter etched if needed before the deposition of the film. The sputtering gas used was zero-grade high purity argon (99.99%); the gas flow was controlled by a needle valve and by partially opening the high vacuum baffle valve.

The high vacuum pumping system consisted of a diffusion pump (Balzers DIF 200) which was backed by a mechanical rotary pump (Balzers DUO 030A). A liquid nitrogen trap was situated between the high vacuum baffle valve and the diffusion pump so as to prevent oil or water vapour from entering the chamber. The chamber could be isolated from the pumping system by the high vacuum baffle value which allowed the chamber to be vented to atmospheric pressure without the requirement of shutting the pumping system down. The baffle valve also allowed the chamber to be gradually exposed to the high vacuum pumping system. This was important since the chamber was only evacuated to a pressure of 10^-2 mTorr using the mechanical rotary pump, whereas the high vacuum pumping system was in the low 10^-8 Torr region. Any sudden opening of the high vacuum valve would have resulted in oil vapour passing into the chamber, and disrupting the flow of the diffusion pump. Pneumatic valves were used to isolate or open the rotary pump to either the chamber or the diffusion pump. The chamber pressure was monitored by a Pirani gauge (A2) in the range 10²-10^-4 mTorr and by a cold cathode Penning gauge (B1)

in the range ~10^-5-10^-8 Torr. The foreline (backing) pressure for the diffusion pump was monitored using a second Pirani gauge (A1).

All removable items (earthing shields, shutters, etc.) within the chamber were shot-blasted using a fine glass bead prior to the growth of any FeSiBC films. This was to remove any foreign material which had previously been grown in the chamber. The sputtering process deposited material on all surfaces within the chamber, and therefore it could also sputter material back off these surfaces thereby contaminating the films (secondary sputtering). All items were de-greased after shot-blasting and handled with clean-room gloves. The chamber was periodically baked out at 70⁰C to reduce out-gassing, and hence improve the vacuum. The contaminants within the system were monitored using a residual gas analyser (Balzers QMG64 RGA).

Use of a magnetron source increased the growth rate by ensuring that a high density of electrons exists near the surface of the target. The field from the magnetron forms a race track from where the majority of the material is sputtered. The downside to this is that the target erodes at a much higher rate in these regions, making very little use of the majority of the target. The growth rate is dependent mainly upon the pressure and power at which the films are deposited. The growth rate is also affected by substrate and target separation, but this was always fixed at 6cm. Increasing the sputtering power increases the growth rate and therefore erosion of the target, but it also has the effect of raising the temperature of the target and substrate. Increasing the pressure will also increase the sputtering rate at low pressures, but this will level off and even decrease at higher pressures because the various particles are slowed down by inelastic collisions. The magnetic properties of the films can be highly sensitive to pressure, power and the temperature at which they are deposited.

The magnetic thin films studied throughout this thesis were sputter deposited from an amorphous METGLAS^® 2605SC ribbon material of composition Fe₈₁Si_3.5B_13.5C₂. The material was supplied by the Allied Signal Corporation and came in the form of an amorphous melt-spun ribbon 18cm in width. The target used for sputtering consisted of two circular 15cm diameter targets which were carefully cut from the 18cm width melt-spun ribbon. The matt side of each target was lightly cleaned with iso-propanol/acetone and immediately dried in a stream of dry nitrogen gas. The two targets were then

Figure 3.3: A diagram of the target electrode illustrating the mounting of the targets.

assembled on target electrode one unless otherwise stated, matt side up as shown in Figure 3.3. A 1mm thick copper and FeCo backing plate was also placed beneath the two METGLAS^® 2605SC targets. The copper plate was used as a spacer and to protect the electrode and the FeCo disc from the plasma in the instances where the METGLAS^® 2605SC targets eroded though. Two METGLAS^® 2605SC targets were used to ensure that any small pinholes in the top target were covered by the target below, thus preventing any sputtering of the copper backing plate. The purpose of the FeCo plate was to increase the sputtering rate, since it increased the density of electrons near the target. The targets were clamped lightly to accommodate for thermal expansion and prevent shattering because of brittleness of the target after being heated during the sputtering process. The top target was changed after approximately 10 hours of growth at 75 watts or when the target shattered or eroded through.

The high quality of the deposited films was maintained by ensuring that the substrates were clean and grease free. This prevented the films from peeling away from the substrate and having imperfections such as pin holes. The glass-based substrates were ultrasonically de-greased, rinsed in acetone/isopropanol and dried in a stream of dry nitrogen gas. Silicon and GaAs substrates are extremely clean on purchase and were only cleaned by a stream of dry nitrogen gas to remove any surface dust particles. Kapton^® substrates were cleaned in a similar manner to the glass-based substrates. Initially 1cm² squared substrates were used. These were glued onto glass slides using a high temperature vacuum compatible glue. The substrates were always glued in the same positions for consistency. The glass slides were then clamped to the copper substrate holder as shown in Figure 3.4a. Further into the study, films were grown on substrates of dimensions up to 7.6cm by 2.6cm. In this case, substrates were mounted on the substrate holder using a picture frame design such that they were only held under their own weight. This eliminated any clamping forces upon the substrate. All substrates were handled with non-magnetic tweezers wherever possible, and using clean-room gloves. The acetone and isopropanol used in the preparation of the substrates were of the analar grade.

Figure 3.4: A diagram representing the two main types of holders used during the sputter deposition of the films in this study.

The growth of the films commenced once a low base pressure (B1) was established within the chamber in the low 10^-7 Torr region. The high vacuum baffle valve was reduced to 30%, and a continuous flow of argon was allowed into the chamber via an inlet needle valve. The required pressure (A2) was obtained by careful adjustment of the needle valve which controlled the flow of argon into the chamber precisely. For consistency the baffle valve was always set to 30%, to ensure the same flow-rate of gas through the chamber during each growth. The continuous flow of argon into the chamber ensured that any gas contaminants produced by out-gassing of the chamber were removed. The plasma was ignited by setting the RF power to 20 W and then tuning the matching circuit so that the substrate or target impedance matched that of the power supply (50 Ohms). This was indicated by the reflected power; the smaller the reflected power, the better the match. The argon pressure was momentarily increased until the plasma has ignited using a second argon inlet valve. Once ignited the reflected power was readjusted to zero and the power was increased in steps of 1W to the required value.

During the initial stages of the study, the substrates were pre-sputtered to clean their surface prior to the deposition of the film. This procedure was later abandoned for reasons discussed in Chapter 4. The METGLAS^® 2605SC ribbon targets were always pre-sputtered for the following reasons:

[1] It ensured that any contamination or oxidation of the surface of the ribbon targets which may have resulted during the fabrication process was removed, and therefore did not get incorporated into the depositing film.

[2] It allowed sufficient time for an equilibrium of sputtering particles to be established. At the start of sputtering, the target and surrounding chamber are cold and as the target heats up, the sputtering properties of the target will also change.

[3] Contamination of the films from secondary sputtering from the surrounding chamber environment is removed, since the chamber is coated with a thin layer of METGLAS^® during the pre-sputter.

[4] It allowed an adequate length of time for any substantial out-gassing of the chamber and target to occur and an equilibrium state to be reached. This also allows the pumping system to reach an equilibrium state, which will ensure a constant flow of argon gas through the chamber and hence constant pressure.

[5] Finally it permitted the whole sputtering system to reach thermal equilibrium before the deposition of the film began.

There were two steps involved in the pre-sputter; the target was initially sputtered at 200 W for 15 minutes to remove any surface oxides and speed up any out-gassing processes (points [1] & [4]), the power was then reduced down to 75 W and sputtered for a further for 5 minutes to allow time for the target and surrounding environment to reach thermal equilibrium [points [2],[3] & [5]). At this stage the substrate shutter which is shielding the substrate from the plasma is rotated to commence the deposition of the film. The pre-sputter times were reduced to 10 and 5 minutes respectively for further growths of the same target. The time of 15 minutes at 200 W was only used for the first growth of each new target.

Upon completion of the deposition, the substrate is shielded from the plasma using the shutter, and the power is decreased to zero in steps of one watt. This procedure was adopted to allow the target to cool gradually, and prevent the target from shattering, increasing the life of the target. The whole system was then allowed to cool for 5 minutes with the flow of argon still on. The continuous flow of argon reduced the possibilities of any contamination from sticking to the cooler film surface from any out-gassing processes. The flow of argon was then turned off and the chamber isolated by closure of the HV baffle valve. The chamber was then vented to atmospheric pressure using dry nitrogen gas which reduces contamination and the substrates removed.

The deposition rate was determined for each series of films grown at the same growth parameters, or whenever the growth parameters were changed. The deposition rate, for a given set of growth parameters, was determined by masking a standard glass slide with a 1mm strip of Kapton^®. The Kapton^® strip was tightly clamped along the length of the slide onto the copper platter using the clamping arrangement shown in Figure 3.4a. A film was then deposited for a fixed period of time and, upon removal of the Kapton^® strip, a clean step existed to the surface of the un-coated substrate. The profile of this step was determined using a stylus measurement. The DEKTAK profilometer used had a vertical resolution of 1nm. The deposition rate was then determined in nm/minute. It was found that the deposition rate was linear (Fig. 3.5) with time, and provided a simple means of depositing a known thickness. No pre-sputtering of the glass slide was undertaken for the calibration films, since this would have eroded the substrate and therefore affected the determination of the thickness. Thickness measurements were done on the central region of the glass slides, since initially samples were grown on 1cm² samples which were mounted to the central region of the glass slides. Measurements performed along the length of the slide revealed that a thickness profile existed along the slide. Figure 3.6 shows there is a 5% decrease in the thickness, as one moves away from the central region of the slide. The film thickness was also monitored on films which were patterned using photo-lithography techniques.

Figure 3.5: Calibration of deposition rate.

Figure 3.6: Thickness profiles across the length of two FeSiBC films. Films deposited at an argon pressure of 4mTorr at 75W.

The Inductive Magnetometer shown in Figure 3.7 was used to provide bulk magnetic hysteresis loops (referred to as MH loops). The magnetometer is based on a similar design to that which is described by Squire et al (1988). It is an induction method which is dependent on Faraday’s law of electromagnetic induction; this states that the voltage V induced in the search coils is equal to the rate of change of flux linking the coil.

(3.1)

Here, N is the number of coil turns linking the flux, A is the cross-sectional area of the search coil, and B is the flux density which is defined as

(3.2)

From equations 3.1 and 3.2 one finds that the magnetisation, M, of the sample is proportional to the integral of the induced voltage

(3.3)

and this is the basis of the magnetometer described here.

The magnetometer consists of a solenoid one metre in length, in which two identical search coils were positioned 50cm apart along the central axis of the solenoid. The search coils were wired in series opposition through a balance circuit. This ensured that when no sample was present within the search coil S₁, there was no net signal from the applied field, but the system was still sensitive to changes in

Figure 3.7: Schematic diagram of inductive magnetometer used to obtain bulk magnetic hysteresis loops.

the magnetisation. The arrangement allowed the magnetisation of the samples to be measured as a function of magnetic field up to 28kA/m. The signal from the balance unit was integrated digitally before being recorded and stored on the computer. The applied field was software controlled similarly to that of the MOKE magnetometer described in Chapter 2. The digital field ramps were user-adjustable, and a number of user-adjustable parameters were available, as with the MOKE system (see Chapter 2). The magnetic field generated by the KEPCO power amplifier was monitored by measuring the voltage across a standard non-inductive resistor which was in series with the solenoid.

The dimensions of the search coil S₁ allowed sample dimensions of 1´1´0.15cm to be measured in two orthogonal directions. It was not possible to investigate fully any in-plane magnetic anisotropy using this system. This was overcome by the use of the MOKE magnetometer and domain imaging system. The inductive magnetometer was able to provide magnetic information from sample volumes as small as 1´10^-11 m^-3. This corresponds to thin films approximately 100nm in thickness. Thicknesses lower than this were found to be more difficult to characterise, but this was overcome by using the MOKE magnetometer which was able to measure sample volumes as small as 1´10^-16 m^-3 without any difficulty.

The integration unit which was used to integrate the induced voltage was prone to drift linearly with time. This produced hysteresis loops which failed to close and therefore it was mandatory to implement a linear drift routine as described for the MOKE magnetometer in Chapter 2. This was possible because the drift was found to be linear with time [Squire et al (1988)].

It was important that the search coils were carefully balanced so there was no net signal from the applied field with no sample present in the search coil. Otherwise the hysteresis loops would give the impression that the samples could not be saturated. This is shown in Figure 3.8 where a loop was obtained from a sample where the search coils were unbalanced and in the situation where the search coils were correctly balanced.

Figure 3.8: MH loops obtained from a FeSiBC film with (a) search coils unbalanced and (b) search coils correctly balanced.

The heat treatments used to anneal the samples in this study were performed at a temperature of 390⁰C for 60 minutes in a low vacuum of the order of 10^-2 Torr. This temperature was found to be sufficient to allow the as-deposited stresses in the films to be relieved, but sufficiently low that crystallisation of the films did not occur. The crystallisation temperature for METGLAS^® 2605SC is 480⁰C [Allied (1995)], and X-ray diffraction analysis of the films revealed no signs of crystallisation after annealing.

The annealing process relieves the internal inhomogeneous stresses which can occur during the deposition of the films. This arises mainly as a consequence of the difference in the thermal expansion coefficients between the depositing film and substrate and the dynamics of the sputtering process itself. In highly magnetostrictive materials such as the FeSiBC investigated in this thesis, this can severely affect the magnetic properties by generally increasing the coercive and anisotropy fields. Random stresses can also give rise to complicated domain structures which can make the interpretation of the magnetisation processes difficult. Annealing of the samples has the effect of removing any magnetic anisotropy which may have been induced during the growth process, for instance by stray magnetic fields. In the case of magnetron sputtering as used in this study, it has been shown in Chapter 5 that under certain sputtering conditions the stray field from the magnetron source induces a unique radial anisotropy.

The annealing process was also employed to induce and control the magnetic anisotropy by applying a suitable magnetic field or mechanical stress to the film as discussed in Chapter 5.

The experimental apparatus used to anneal the samples is shown in Figure 3.9. A wire-wound observation furnace was used to heat the samples which were placed inside a Pyrex tube. The low vacuum of 10^-2 Torr within the tube was maintained by a mechanical rotary pump. The annealing process was performed under a vacuum so as to reduce any oxidation of the films and prevented the surface of the films from being contaminated by impurities in the air. An AC current source, controlled by a Eurotherm temperature controller, was used to maintain the temperature to within ±1⁰C. In the situation, where the samples were field annealed, the furnace was positioned between the pole pieces of a water-cooled electromagnet which provided a magnetic field of 0.3T. On completion of the appropriate heat treatment, the samples were allowed to cool gradually to room temperature in a vacuum. Prior to any heat treatments, all films were cleaned to ensure no grease or other impurities were present on their surface; these could have been absorbed into the film during the anneal. It was found that the samples which were deposited onto Kapton^® substrates were not suitable under any of the heat treatments. This was due to the large difference in the thermal expansion coefficients of the Kapton^® substrate (20ppm/⁰C) and the METGLAS^® 2605SC (4.9ppm/⁰C) film. The films on Kapton^® were found to be highly stressed after annealing and it was not possible to measure their magnetic properties.

Figure 3.9: Apparatus used for thermal treatments of samples. The electromagnet is only present when the samples are magnetically annealed (field annealed).

Figure 3.10: An illustrative diagram of an X-ray diffractometer.

Figure 3.11: X-ray diffraction patterns from (a) METGLAS^® 2605SC ribbon targets. (b) Glass substrate and a 500nm film deposited on a glass substrate.(4mT, 75W). (c) Lorentzian fit to the (110) peak from ribbon - FWHM: 2.9⁰. (d) Lorentzian fit to the (110) peak amorphous film - FWHM: 3⁰.

X-ray diffraction measurements were performed to ensure that the films deposited by RF magnetron sputtering were amorphous; the analysis was performed using MoK_a radiation. A number of samples were also analysed in Madrid using CuK_a radiation. The X-ray diffractometer (Philips) was an automated system which operated in the q-2q geometry and is shown schematically in Figure 3.10.

The sample, X-ray source and detector are all in the same plane, and collimating slits are positioned in the beam path to ensure a well defined focused beam. The sample holder and detector are mechanically coupled such that a rotation of the sample through an angle q would ensure that the detector would rotate through an angle 2q. This ensured that the incident and reflected angles always remained equal. The angular range and step size were software controlled and the recorded data was stored digitally.

The X-ray diffraction patterns obtained using this technique were analysed using the Bragg equation

(3.4)

where l is the wavelength of the radiation, n is an integer, d_hkl is the interplanar spacing, and q is the angle between the plane of atoms and the x-ray beam (see Elliot (1984) for details). The interplanar distance between consecutive planes of atoms is a function of the Miller indices (h,k,l) and the lattice parameter a_l. For cubic structures the interplanar spacing d_hkl can be obtained from the following expression

(3.5)

In the case of amorphous materials, there is a distribution of the interplanar spacing because of the random nature of the structure. This leads to much broader peaks in the X-ray diffraction patterns. The amorphous nature of the thin films was compared with the amorphous METGLAS^® 2605SC ribbon which was used as the target material. Figure 3.11a shows a typical diffraction pattern obtained from a METGLAS^® 2605SC target ribbon; the peaks have been labelled using the lattice parameter of 0.2866nm for a-Fe. Figure 3.11b shows a typical X-ray diffraction scan for an amorphous FeSiBC film which has been deposited onto glass. The X-ray scan for the glass substrate alone has also been included for comparison. Figures 3.11c,d, represent the (110) peak to which a Lorentzian fit has been applied to obtain full width half maxima (FWHM). The values obtained for the amorphous films (3⁰) compare well with that of the target material (2.9⁰) which indicates a similar amorphous phase.

The experimental apparatus used to perform the MI measurements presented in Chapter 6 is shown in Figure 3.12. The MI measurements were performed at the Instituto de Magnetismo Aplicado in Madrid. The system was constructed and maintained at the time of the measurements, by J.P. Sinnecker [J.P. Sinnecker et al (1998)].

The basic principle of the measurement is that a low intensity alternating current is passed through the magnetic sample and the changes in its impedance are measured as a function of applied field (see Chapter 6).

The alternating current was provided by a Hewlett-Packard 38589A spectrum/network analyser, which could deliver a current in the range of 1-100mA up to driving frequencies of 150MHz. At the time of the measurements, frequencies up to 10MHz only were available. To ensure that a constant current was being delivered, the voltage, V_R across a non-inductive resistor was monitored; this resistor was in series with the sample. The voltages across both the resistor and the sample were measured by the spectrum/network analyser. The samples were orientated so that the magnetic field of the Helmholtz coils was in the plane of the sample and parallel to the current direction. The Helmholtz coils, in conjunction with the KEPCO BOP 50-4M current source, provided a maximum DC field of 10.5kA/m.

The measurement system was fully automated using a computer; this swept the magnetic field and recorded the voltage across the sample. At each measurement point, the voltage across the resistor was monitored and recorded, and any changes in the current were then automatically corrected.

The electrical contacts to the samples were made using high purity copper strands which were bonded to the surfaces of the films using silver paint. This was allowed to dry for 24 hours to ensure a good electrical contact. Coaxial cabling was used throughout the apparatus to screen out unwanted signals.

Prior to the performance of any MI measurements it was first established that the current amplitudes which were being used were not causing the films to heat up due to eddy current heating, which would effect the impedance measurements. The impedance of the sample was monitored over a period of 20 minutes at a fixed current. It was found that the current amplitudes were sufficiently small and there appeared to be no measurable changes in the impedance. To verify that no external artificial signals were present in the system, a high purity copper strand was substituted for a magnetic sample; no changes in the impedance were obtained. The system was able to measure impedance ratios as small as 0.1 %.

Figure 3.12: An illustration of the MI apparatus used to perform the MI analysis.

Figure 3.13: Schematic representation of the stages involved in photolithography.

Photolithography is a technique which is widely used in industry in the manufacture of integrated circuits or semiconductor components. The technique involves transferring a pattern from a photographic mask onto a film so that a pattern can be chemically etched out of the film.

The cleaned film is first coated with an organic photosensitive material known as photoresist. This is allowed to dry and then exposed to UV light through a photographic mask (Fig. 3.13a). The photoresist is developed using a suitable developing agent which removes regions of the photoresist which have de-polymerised due to the expose to the UV light (Fig. 3.13b). The remaining regions of the photoresist act as a barrier during the etching stages. After compilation of the etching process, the remaining photoresist is removed using a suitable solvent (Fig. 3.13c) leaving an etched film of the mask.

The films were carefully cleaned by boiling the samples in a solution of trichloroethane. This removed any oily residues and any foreign surface particles. For more stubborn dust particles the surfaces of the films were wiped using cotton buds soaked with trichloroethane. In all instances the films were dried in a stream of dry nitrogen gas. It was important that the surfaces of the films was dust/grease free in order that the photoresist adhered to the surface of the films so as to protect the film beneath. The films were baked at 100⁰C on a hot-plate for 1 minute to ensure no water vapour was present before finally being coated by a few drops of photoresist, with the excess being spun off. The photoresist was also allowed to dry for 1 minute on a hot-plate before being exposed to the UV light through the photographic mask. The resist was developed, exposing regions of the films which were to be etched away. The etchant used was a solution of nitric acid (3 parts of water to 1 part of nitric acid). The etch time was determined by using sacrificial film pieces. On completion of the etching process, the remaining resist was removed using a solvent to reveal the patterned film.