Figure 5.1 shows a schematic diagram of the four-point-probe measurement set-up; it is shown that the current is applied through probes 1 and 4 and the voltages measured between probes 2 and 3.

Figure 5.1: Schematic diagram of a four-point-probe measurement set-up for a ITO film of arbitrary shape and dimensions

In addition to corrections needed for varying film thickness, accurate measurement using the four-point-probe method must take into account the shape of the test sample. This has its own correction factor, Cf, which depends on the relative size and shape of the test sample with respect to the probe spacing and the previous equation is then modified to:

Therefore, the TLM method, although slightly more time consuming in post-deposition sample preparation, was chosen to safeguard the integrity of the results which is vital for any comparative analysis work. The theory and the basic set-up of this technique have already been discussed in Chapter 4 therefore only the modifications needed to apply it for these purposes will now be described.

Following ITO deposition on highly resistive S.I. GaAs substrates, the samples were etched with TLM patterns to form ITO mesas. Further photolithography was used to deposit metal pads on the ITO mesas. The purpose of the ITO mesa is to confine the current between two metal pads when a voltage is applied to them. A schematic diagram of this set-up is shown in Figure 5.2:

Figure 5.2: Schematic diagram of a typical ITO film sample prepared for electrical measurements using the TLM method

The TLM method has various advantages over the four-point-probe method:

• sample shapes and sizes are always identical as they are produced using the same TLM mask set hence no correction factors have to be taken into account
• many accurate measurements can be made from a relatively small sample (a 1 cm2 sample contains 20 TLM mesas allowing as many independent measurements)
• metal pads ensure that the current is uniformly spread out minimising the risk of either heating or burning the ITO underneath, which would increase the resistance or render it unusable
• repeatable results were obtained with acceptable standard deviation (below �5%) for samples from the same deposition run.
• the existing automated TLM measurement system could be used for swiftness and accuracy.

Table 5.1 shows the sheet resistance and the associated standard deviation of the same ITO film measured at four different parts using both the four-point-probe method and the TLM method:

Therefore, all subsequent electrical analysis of ITO films were carried out using the TLM method.

5.1.2 Optical Properties of the Deposited ITO Films

During all ITO depositions, the film transparency was monitored by placing a microscope cover slide on the Nordiko 1500 substrate table. The transmittance measurements were carried out in a double beam spectrometer; details of these measurements are described in Chapter 4.

Figure 5.3 shows the measured transmittance of a typical ITO film over the wavelength range 200 to 2000 nm. It is seen that above 90% transmittance is achieved for the 500 to 2000 nm wavelength range corresponding to 2.48eV to 0.62eV energy range which covers the majority of the III-V compound semiconductors such as AlAs, AlGaAs, GaAs, InP and InGaAs as well as Si and Ge. These results are a marked improvement over those obtained by Zhang et al who reported a transmittance of 80% over the 400 to 1300 nm wavelength range for ITO films grown by d.c. magentron sputtering [139].

Figure 5.3: Graph of measured transmittance vs. wavelength for a typical ITO film grown by reactive r.f. sputter deposition. Film thickness was approximately 1,500�.

Figure 5.3 shows that the transmittance sharply decreases below a threshold wavelength; this is due to fundamental absorption of ITO for which the widely reported value of the bandgap is approximately 3.75eV [7,13]. The measured transmittance data can be used to obtain the bandgap of the ITO film - in particular, the dependence of the absorption co-efficient, a, in the absorption edge on the wavelength can be used to calculate this bandgap. The transmittance, Tr, as a function of the wavelength, l, for a film of thickness d is given by [14]:

For k'2 << nITO2, the transmittance is mainly dependent on the exponential term of (eqn. 5.3) and the constant A can be approximated to 1; hence taking this into account and combining (eqn. 5.3) and (eqn. 5.5), Tr can be simplified to:

Now, (eqn. 5.7) can be used to plot a2 as a function of energy, hn, and extrapolated to a2 = 0 yielding the bandgap of the ITO film as shown in Figure 5.4.

Figure 5.4: Graph of a2 vs. energy for a typical ITO film grown by reactive r.f. sputter deposition. The extrapolated bandgap for this ITO film is (3.7�0.03) eV

In this study, the value of the determined bandgap for a typical ITO film deposited by reactive r.f. sputtering is (3.7�0.03) eV. This is in good agreement with the values reported in the literature within the bounds of experimental error. However, it should be noted that most of the data in the literature were measured using ellipsometric measurements - a technique which is more accurate but more cumbersome than the one used here.

5.1.3 Effect of Total Pressure

In a reactive r.f. sputtering environment, a minimum gas pressure, or a threshold pressure - PT - is required in order to excite the plasma. In our case this was 1e-3 torr. As most depositions were carried out below this pressure, it was therefore necessary to raise the chamber pressure momentarily to PT in order to excite the plasma and then reduce to the required pressure. The plasma was self-sustaining throughout the entire deposition pressure range.

The total pressure effects the deposition process in a number ways:

• greater the pressure, greater the number of particles in the plasma
• increased number of particles increases the probability of sputtering from the target at a given r.f. power and therefore will increase the deposition rate
• however, greater number of particles increases the probability of scattering of sputtered particles during transit from the target to the substrate and will have an adverse effect on the deposition rate beyond a given pressure.
• the total pressure at a given r.f. power also effects the induced bias on the electrodes which in turn influences the deposition rate.

These findings about the dependence of the deposition rate on the chamber pressure are in good agreement with experimental observations of Sreenivas et al [6]. Therefore, it is important to take this parameter into account for optimizing ITO deposition. In this investigation, the total chamber pressure was kept at a constant 5 mtorr.

5.1.4 Effect of Oxygen Partial Pressure, PO2

There is a strong dependence of ITO film properties on the oxygen partial pressure which demands close control for reproducible preparation of such films. For r.f. sputtered ITO films, it is well known that while the transparency of the ITO films is directly proportional to the oxygen content of the plasma in the sputter chamber, the sheet resistance is inversely proportional to it [7,8,140]. Similar findings have also been reported for these films grown by other techniques such as ion-beam sputtering [11] and spray pyrolysis [141].

Figure 5.5 shows the results of ITO films deposited under various oxygen partial pressures at a r.f. power of 150 W. It is evident that the sheet resistance is directly proportional to the exponential of the oxygen partial pressure. Whereas the transmittance of these films were above 90% in the 500 to 2000 nm wavelength range, further reductions in the oxygen content severely degraded the transparency to below 60%. Total absence of O2 resulted in dark brown deposits with maximum transmittance of 50%, for very thin films of 500�, which diminished for thicker films.

Figure 5.5: The sheet resistance, Rsh, and the resistivity, r, of ITO vs. the oxygen partial pressure, PO2, in the r.f. reactive sputtering plasma; the corresponding O2 flow is also shown for reference

The high conductivity of the ITO films has been attributed to both substitutional tin and oxygen vacancies, created either during film growth or post-deposition annealing resulting in a material represented as In2-xSnxO3-2x [3]. Increasing the oxygen partial pressure above a value which yields a near stoichiometric film composition would thus result in an accumulation of excess oxygen, mainly at the grain boundaries acting as trapping centres for free carriers [7]. This means that the barrier scattering becomes the dominant process since the barrier height increases strongly. At the same time the density of free carriers is reduced as they are partly localised at the trapping sites [12]. This results in a decrease in the carrier mobility, and therefore a corresponding fall in the film conductivity is observed. Buchanan et al [8] have also suggested that the decrease in the conductivity in increased PO2 is primarily due to a reduction in the carrier concentration caused by the occupation of oxygen vacancies rather than due to tin doping. A rapid increase in the ITO resistivity has been consistently observed by Kellet et al on films grown on both GaAs [142] and Si [143]; they have related this observation to ITO's wide oxygen stoichiometry and its degeneracy (resulting in high conductivity) when the PO2 (and hence the oxygen content) is reduced.

With regard to the dependency of the transmission of ITO films on the PO2, Fan et al [5] have reported complimentary findings to those obtained in this study. They observed that the ITO film transmittance increased rapidly with increasing oxygen partial pressure exceeding 80% at a PO2 of 3e-3 then saturating around 90% for further increase. In addition to increasing the transmission of the films, there is also a shift of the intrinsic absorption edge to longer wavelengths (i.e. lower photon energies) due to an increase in PO2. Simple calculations show that this shift can be as much as 10%. Study of r.f. sputtered ITO films by Ohhata et al relate this phenomena to a change in the carrier concentration [144].

5.1.5 Effect of r.f. Power

The r.f. power has a direct influence on the deposition rate and the induced voltage as well as generating heat due to the electron bombardment intrinsic to this technique. After experimenting with various r.f. power for ITO deposition we have seen its marked influence on the electrical properties of the deposited films.

Figure 5.6: Sheet Resistance, Rsh, of ITO vs. r.f. sputtering power

During each deposition, at various powers, the plasma was 'tuned' to maximise the forward power and reduce the reflected power to zero. Our results from depositions using a oxygen partial pressure of 12e-3 show that the sheet resistances of the ITO films are inversely proportional to the exponential of the r.f. power used as shown in Figure 5.6. Films grown at 100W showed typical Rsh of about 70 kW/�, at 150W this was 4kW/� while at 200W the Rsh was 40W/�. These results are in close agreement with those reported by Sreenivas et al [6] and Mansingh et al [140], who used very similar deposition techniques and conditions to those in this study.

It should be noted that these values were obtained without any post-deposition annealing; after annealing in forming gas, Rsh < 10 W/� was achieved. On the other hand, very high r.f. powers of 500 W during deposition have been reported to generate sufficient intrinsic heating thus eliminating the need for further treatment usually carried out to obtained good quality ITO films [145]. In this study, however, films deposited at 200 W were difficult to etch; hence the r.f. power of 150 W was used for all subsequent work. Also, from a device application point of view, the use of excessive r.f. power is not suitable.

5.1.6 Effect of Induced Voltage

Figure 5.7 shows that the induced d.c. voltage on the substrate electrode is directly proportional to the r.f. power at a given pressure. The d.c. bias was independent of the oxygen partial pressure.

Figure 5.7: Induced d.c. bias on substrate electrode vs. r.f. Sputtering Power at a total chamber pressure of 5 mtorr

It was also observed that the induced d.c. bias changed if the plasma was de-tuned i.e. if the reflected power was non-zero. As all depositions were carried out in a tuned plasma, any dependence of ITO electrical properties on the induced d.c. bias would be the same as those seen for their dependence on the r.f. power.

5.1.7 Effect of Deposition Rate

Figure 5.8 shows that the deposition rate itself is directly proportional to the r.f. sputtering power at a given chamber pressure and partial pressure of oxygen. Reduction in the oxygen flow rate or the oxygen partial pressure, PO2, in the reactive plasma caused a proportional decrease in the deposition rate at a given r.f. power.

Figure 5.8: ITO deposition rate vs. r.f. sputtering power at a PO2 of 14e-3.

Since the ITO deposition rate is directly proportional to the r.f. power for a given set of conditions, any dependence of the film properties on the rate will be the same as those observed for their dependence on the r.f. power for this system.

5.1.8 Effect of Target Pre-conditioning

Pre-conditioning is a term used for attaining stability in the sputter chamber prior to ITO deposition on the substrate. This includes stabilising the chamber pressure, the plasma, the r.f. forward and reflected power. Pre-conditioning has two distinct effects:

1. It cleans the target surface of any debris and deposits a fine layer of ITO on the chamber walls thus preventing further contamination due to natural degassing from the surface.

2. It stabilises the release or sputtering rates of species from the target and particularly so if the target is a composite as in the case of ITO targets.

Without pre-conditioning, results were very difficult to reproduce between consecutive sputter depositions with nominally identical conditions. Both Ar on its own as well as Ar and O2 mixtures were used as the plasma gas in turn without any pre-conditioning, but neither of these conditions produced satisfactory repeatability.

We have studied the effect of pre-conditioning under varied or similar conditions to those used during the actual deposition; these include using different r.f. power and gas composition to those used in the respective depositions. In general, best results were obtained when the pre- conditioning parameters were the same as the deposition parameters. In all the cases the pre- conditioning period was 30 minutes whereas the deposition time was dictated by the specific needs of any particular experiment.

5.1.9 Effect of Film Thickness

Our initial experiment with Ar as the only plasma gas during ITO deposition showed that the optical transmission was inversely proportional to the thickness and remained below 50%. However, for addition of O2 in the plasma gas above partial pressures of 7e-3, consistent transmissions above 90% were obtained in the 500 to 2000nm wavelength range regardless of the film thickness.

Two sets of experiments showed a noticeable trend on the effect of thickness on the Rsh and r. In each set, ITO was deposited on two otherwise identical samples of S.I. GaAs; all preparation and deposition conditions were identical for each of the two samples except the deposition time, which was varied, resulting in two different thicknesses. The results and conditions are summarised in Table 5.2.

These results show that for a given set of deposition conditions, the thicker film has better electrical properties. These results are consistent with the findings of Sreenivas et al [6]. Studies involving very thick ITO layers (> 10,000�) deposited on metal alloys by Just et al suggest that since the stress within the film is independent of its thickness. Hence a shear stress is developed between this film and the substrate ultimately resulting in the loss of adhesion between thick films and substrates. They also observe a decrease in this stress due to annealing and that the ITO film thickness does not have any effect on its refractive index [146].

5.1.10 Effect of Target to Substrate Distance

For a given set of deposition parameters for the sputtering technique, the energy distribution of sputtered neutrals reaching the substrate depends on the distance between the target and the substrate due to molecular collisions in the plasma. Depending on their initial energy, which is proportional to the r.f. power and hence the self-induced bias, these particles travel a 'thermalization distance' until its energy reduces to the thermal energy kT. Transport beyond this 'virtual source' point towards the substrate is by diffusion under the material concentration gradient.

Experimental work carried out on a very similar system to that used here by Kumar et al show that the deposition rate is inversely proportional to the exponential of the distance between the virtual source and the substrate [147]. Thus the substrate position can be located either above or below this virtual source point by increasing or decreasing the distance between the target and the substrate. However, during the course of this investigation, this separation was fixed at a distance of 7.5 cm, slightly above the virtual source which varies between 3 and 6.5 cm for r.f. power of 50 to 200 W respectively. Kumar et al reported that ITO films with improved transmittance and more uniform properties were obtained for substrates located below this virtual source.

5.1.11 Effect of Substrate Temperature During Deposition

In the Nordiko 1500 sputtering system used in this work, it is not possible to measure the in- situ chamber temperature. However, reports in the literature from groups using an identical system suggests substrate temperatures rise to approximately between 60�C to 70�C during deposition due to intrinsic electron bombardment.