Physical properties of TCOs
In order to be transparent in the visible range, TCOs need to have a band gap that is larger than the energy of photons in the visible range (~3 eV). On the other hand, to be simultaneously conducting, the material needs to have a sufficient charge carrier concentration. The conductivity σ of a material is given by σ = n e µ , (2.1).
with n the charge carrier concentration, e the elementary charge and µ the mobility of the charge carriers. To achieve a large charge carrier concentration in a wide band gap semiconductor, shallow donor states need to be introduced below the conduction band .
When doping semiconductors, additional energy levels are intro-duced within the band gap close to the valence or conduction band in case of p-type or n-type semiconductors, respectively. These additional states can introduce charge carriers in the valence or conduction band. This causes a shift of the Fermi level EF that in the case of an n-type semiconductor is given by nC EC −EF = kBT ln nD , (2.2).
with EC the energy level of the conduction band, nC the effective density of states in the conduction band and nD the density of donor defects [13, p. 37]. Equation 2.2 shows that the Fermi level increases with increasing n-type doping. This is the case for non-degenerate semiconductors. If the doping is increased so far, that the dopant concentration is higher than the effective density of states in the conduction band, the Fermi level will lie inside the conduction band. In this case the semiconductor is called degenerate [13, p. 151]. In the case of a degenerate semiconductor, the position of the Fermi level is given by ħ2 2 2m ∗ EF−EC = h (3π2ne)3 , (2.3).
The fact that the Fermi level lies within the conduction band means that the bottom states of the conduction band are filled. The optical band gap measured by absorption techniques will therefore be shifted towards higher energies. This effect is called the Burstein-Moss effect and is shown in Figure 2.1 [14, 15]. To calculate the shift in the optical band gap, Equation 2.3 needs to be multiplied by (1 +me∗/m h∗), with mh∗ the hole effective mass, in order to take the curvature of the bands into account . The Burstein-Moss shift is then given by me∗ mh∗ ΔEBM = h (3π2ne)3 + . (2.4).
Applications of TCOs
The most important applications of TCOs are architectural window ap-plications, and as transparent electrodes in flat panel displays, touch-screens and photovoltaics [7, p. 3, 24, p. 768]. Other applications include transparent electronics and films for electrochromic devices [7, p. 5], as well as antistatic films [24, pp. 758-759] and electromagnetic shield-ing [24, p. 754]. In the following, a short description of the different applications will be given.
For architectural windows, the principal interest in TCOs lies in their low emissivity in the infrared. The emissivity ε is related to the reflect-ance R and the transmittance T by ε = 1−T −R . As TCOs are reflect-ive above the plasma wavelength, their emissivity is low in the infrared. This allows to lower the transfer of heat from the inside to the outside of a building via infrared radiation, while the solar radiation in the near-infrared can still pass through the window . An ideal material for this application would therefore be fully transparent at wavelengths between 400 and 3000 nm, and fully reflective at wavelengths above 3000 nm . Especially important is a low emissivity in the wavelength range between 8000 and 13000 nm in order to avoid the radiative cooling effect due to the transparency window of the atmosphere in that range . Low emissivity coatings are therefore also useful to prevent frost on car wind-screens . The most common material used for these applications is FTO due to its good chemical and mechanical stability . Addition-ally, it is also possible to use the TCO film as a transparent heater by applying a voltage across the film, in order to defrost car or airplane windscreens [24, p. 753].
In order to be used as a transparent electrode, a material needs to be transparent in a certain wavelength range that depends on the ap-plication of the device . For use in flat panel displays, the material needs to be transparent in the range that is visible to the human eye (400-800 nm) . For use in photovoltaics, the necessary transparency range is given by the solar spectrum (300-2500 nm)  and the band gap of the absorbing material.
In flat panel displays, two transparent electrodes are needed with a layer of liquid crystals in between. Applying a voltage to the device allows to change the orientation of the liquid crystals . For flat panel displays, the most commonly used TCO material is ITO due to its low resistivity [6, p. 24]. However, because of the high cost of In, materials with less or no In are also interesting for use as transparent electrodes. Recently, amorph-ous TCOs such as indium zinc oxide (IZO) have emerged , which offer a better uniformity over large areas [33, p. 157]. These materials are therefore also used in flat panel displays [7, p. 4].
On mobile devices, such as phones and tablets, flat panel displays are usually equipped with a touchscreen. There are generally two types of touchscreens [24, pp. 768-772]. The first type is the resistive type, where one transparent electrode is deposited on a flexible substrate. When this flexible layer is deformed by touching it, a contact with the second transparent electrode is formed. By applying a voltage gradient on each electrode, the position of the contact can be detected. The other type of touchscreen is the capacitive type. In this type, two grids of transpar-ent electrodes, that are perpendicular to each other and separated by a dielectric spacer, form capacitors. When the device is touched at a certain location, the local electric fields change, and the resulting change in capacitance can be detected. For such touchscreens, ITO films are usually used.
In solar cells, TCO films can be used as back and front contacts. ITO alternatives such as AZO or FTO are often used [6, p. 24]. For example, sputtered Al-doped ZnO layers are used as both back and front contacts in solar cells based on amorphous Si. The surface of these layers can be modified by chemical etching to improve the light trapping . Trans-parent electrodes based on doped zinc oxide are also used in thin film solar cells based on CuInGaSe2 (CIGS)  or CdS/CdTe .
Another application where transparent electrodes are needed, are electrochromic devices . Electrochromic devices experience a change of their optical properties when a voltage is applied. Such electrochromic devices can therefore be used in applications such as “smart windows” which allow to control the onset of reflectivity . This allows to switch the window from a mode, where near-infrared radiation from the sun is reflected, to keep the inside of the building cool, to a mode, where the onset of reflectivity is shifted to longer wavelengths, in order to keep heat from escaping through the window. Electrochromic devices are based on the transport of ions, such as H+ or Li+ through an ionic conductor from an ion storage film towards the electrochromic film. A typical elec-trochromic material is WO3. As ionic conductors, both organic and inorganic materials can be used. Finally, transparent electrodes are used on both sides of the device in order to apply a voltage. When a voltage is applied through these electrodes, the ions are transferred into the electrochromic film, thereby increasing the absorption of the material.
Properties of zinc oxide
Zinc oxide (ZnO) is a wide band gap semiconductor. It is used in the chemical industry for rubber production, paints and agricultural us-age [6, p. 3]. ZnO is also used in the form of micro- or nanoparticles in sunscreen, as its absorption edge is in the UV . Research on ZnO as a semiconductor material has started in the 1950s, however since the 1990s there has been a significant increase in the number of publications on ZnO [6, p. 1]. This increase can be related to the many interesting properties of ZnO, such as its direct band gap, the large exciton bind-ing energy, as well as its piezoelectric and luminescence properties . Other important factors are the possibility to easily fabricate ZnO nano-structures  and the potential of ZnO to be used as a dilute magnetic semiconductor . This section will give an overview on the proper-ties of ZnO with a focus on the possibility of doping ZnO with extrinsic impurities in order to use it as an n-type TCO.
ZnO can occur in several crystal structures, such as the hexagonal wurtzite structure, the cubic zincblende structure and the cubic rock-salt structure . The most stable one of these crystal structures is the wurtzite structure. In the wurtzite structure of ZnO (Spacegroup P63mc) the hexagonal unit cell has the lattice parameters a = 3.2501 Å and c = 5.2071 Å. The structure can be seen as the superposition of two hexagonal closed packed sublattices of Zn and O [48, p. 2]. Each Zn and O atom is surrounded by the respective other type of atom in a tetra-hedral coordination. The crystal structure is shown in Figure 2.3. In ZnO the (0001) planes are terminated by Zn atoms, whereas the (0001¯) surfaces are terminated by O atoms. These planes are therefore polar, which means that ZnO does not show an inversion symmetry and has piezoelectric properties . It is interesting to note, that thin films of ZnO often show a preferential orientation along the c-axis . This preferential orientation depends on the deposition method, but is very common especially in sputter deposited films. However, the exact reas-ons for this preferential orientation are not yet fully understood, but are possibly related to the surface energies of the different planes .
Deposition methods for AZO thin films
There are several methods in order to deposit thin films of ZnO or AZO. These deposition methods can belong to several categories, such as chemical solution based methods, chemical vapour deposition (CVD) and physical vapour deposition (PVD) [7, pp. 201-202].
Chemical solution based methods
Among the chemical solution based methods are techniques, such as spray pyrolysis [64, 65] and sol-gel methods [66, 67]. In spray pyrolysis, fine droplets of a solution of precursors such as zinc acetate and alu-minium chloride are sprayed onto a heated substrate. Typical deposition temperatures are around 450 ◦C, at which the precursors decompose and ZnO is formed. In the sol-gel method, a colloidal solution of polymer particles (the “sol”) is prepared from precursors [68, pp. 2-8]. This sol can be transferred to the substrate by techniques such as dip coating or spin coating. The colloidal particles can then aggregate and form a polymer network (the “gel”). The substrate then needs to be annealed at temperatures between 400 and 850 ◦C in order to remove the solvents and condensate a ceramic compound, that can be crystallized by sinter-ing. Films obtained by these methods are generally polycrystalline and their resistivity is limited to around 1 ×10−3 Ωcm, however there are also reports of resistivities down to 7×10−4 Ωcm in case of spray pyrolysis . The advantage of these methods is their simplicity, as no vacuum system is needed. They also provide a good control over the composition of the films. However, they require quite high process temperatures compared to other techniques.
Another chemical solution based method is electrochemical depos-ition . The deposition temperature in this technique is less than 100 ◦C, as the process takes place in an aqueous solution. The substrate is put into an electrolytic bath and used as a working electrode. The pre-cursors are then electrochemically reduced to form the film. This means, that this technique requires a conductive substrate, so it is not useful for the fabrication of transparent electrodes. However, electrochemical deposition of ZnO allows to deposit nanostructures, such as nanorods, by decreasing the concentration of the zinc precursor [70, 71].
Chemical vapour deposition
AZO films can also be deposited using CVD methods. With the excep-tion of atmospheric pressure CVD, CVD methods are usually vacuum techniques, which means, that they are more expensive than chemical solution based methods. On the other hand, working in vacuum re-duces the amount of unwanted contamination. Generally, CVD works via a chemical reaction of vapours of the precursors, in which the pre-cursors are decomposed and form a film on the substrate [72, p. 147]. The chemical reaction on the substrate depends strongly on the sub-strate temperature. There are several types of CVD, that can be used to deposit transparent conducting ZnO films, such as metal organic CVD (MOCVD) , plasma enhanced CVD (PECVD) , atmospheric pressure CVD  and atomic layer deposition (ALD) .
MOCVD uses metal organic precursors, that means metal atoms with organic ligands. The oxygen precursor can be O2, H2O or an alcohol [6, pp. 235-236]. The precursors are evaporated and carried into the vacuum chamber by a carrier gas such as Ar. O2 might be introduced separately into the chamber. The substrate needs to be heated to temperatures of several hundred ◦C. Using MOCVD, AZO films with a resistivity of 6 ×10−4 Ωcm have been deposited at 275 ◦C . The deposition tem-perature can be reduced by using PECVD. In this technique, a plasma is used to assist in breaking the chemical bonds of the precursors, so lower substrate temperatures can be used [72, p. 181]. This way, AZO films with a resistivity of 7 ×10−4 Ωcm could be deposited . Another CVD method is atmospheric pressure CVD, which does not require a vacuum system. AZO films with a resistivity in the order of 10−4 Ωcm could be deposited with this technique at a substrate temperature of 400 ◦C .
ALD is another technique that is related to CVD. The main difference is that the metal and oxygen precursors are inserted into the reactor sep-arated in time. This allows to grow films layer by layer, which allows an accurate control over the film thickness. While this technique generally gives lower deposition rates, films deposited by ALD have a great con-formity to any substrate shape. However, ALD is sensitive to the chemical nature of the substrate, as the initial growth can be severely hindered if the precursor does not react with the surface of the substrate . AZO films with a resistivity in the order of 9 ×10−4 Ωcm have been grown at a substrate temperature of 200 ◦C using ALD .
Physical vapour deposition
Physical vapour deposition methods are also vacuum processes, leading to a lower contamination of the substrates and the growing films [79, p. 7]. PVD methods for the deposition of AZO films include evaporation [80, 81], pulsed laser deposition (PLD) [6, p. 303], cathodic arc deposition  and magnetron sputtering .
In thermal evaporation methods, a source of material is heated in va-cuum, so it evaporates and the atoms are deposited onto a substrate [79, p. 11]. There are several possible ways to evaporate the material, such as passing an electric current through the material [79, p. 368] or directing an electron beam of several keV onto the surface of the material [79, p. 400]. Both of these techniques have been used to deposit AZO films with lowest resistivities in the order 10−4 Ωcm using ZnO powder as a source at temperatures between room temperature and 200 ◦C [80, 81].
Another PVD technique that allows the growth of high quality AZO films is pulsed laser deposition. In PLD a target is ablated by laser pulses with high energy [6, pp. 305-309]. As the laser pulse interacts with the vaporized target material a plasma plume is formed, that expands to-wards the substrate. The atoms and ions then condense on the substrate, leading to growth of a thin film. One of the advantages of PLD is the stoichiometric transfer of the compound from the target to the sub-strate. Additionally, the process is free of contamination and the growth of metastable materials is possible. The main disadvantage of PLD is the limitation to substrate sizes of about 1 cm2 [6, p. 304]. However, in laboratory scale, PLD allows the deposition of AZO films with the lowest resistivity known . Using a ceramic ZnO/Al2O3 target, Agura et al  deposited columnar AZO films with a resistivity of 8.5 ×10−5 Ωcm at a substrate temperature of 230 ◦C on glass substrates.
Cathodic arc deposition is another method for the synthesis of AZO films. It is based on an arc discharge that is used to evaporate a target at so called cathode spots and transform the material into a plasma . In these cathode spots, the current density is very high in the order of 1010 to 1012 A/m2, while the voltage is low at around 20 V. The high current density leads to large ionized fraction of the target atoms. However, macroparticles are often produced at the cathode spots, that can be incorporated into the growing film. In order to avoid this, a filtering system can be used, such as a magnetic field to steer only the ions towards the substrate. Using pulsed filtered cathode arc deposition, AZO films with a resistivity in the order of 10−4 Ωcm were deposited at a substrate temperature of 200 ◦C . In these depositions, Zn rods with a few percent of Al were used as targets and oxygen was injected into the vacuum chamber as a reactive gas.
Finally, magnetron sputtering is a PVD method, that is commonly used for the deposition of AZO films . This is due to the advantages of magnetron sputtering, such as the scalability to large areas of up to 20 m2 and the possibility to achieve good film properties at relatively low substrate temperatures due to the plasma assistance [7, p. 202]. This technique is used in this thesis. Therefore, a detailed description of magnetron sputtering will be given in chapter 3 with a focus on the sputtering of AZO films in section 3.7.
Table of contents :
2 Transparent conducting oxides
2.1 Physical properties of TCOs
2.2 Applications of TCOs
2.3 Properties of zinc oxide
2.4 Deposition methods for AZO thin films
3 The sputter deposition process
3.1 Basics of plasma physics
3.2 Sputtering interactions
3.3 Magnetron sputtering
3.4 Reactive magnetron sputtering
3.5 High power impulse magnetron sputtering
3.6 Thin film growth
3.7 Sputter deposition of AZO thin films
3.8 Experimental setup
4 Characterization methods
4.2 Photoluminescence spectroscopy
4.3 Electrical characterization
4.4 X-ray diffraction
4.5 Scanning electron microscopy
4.6 Transmission electron microscopy
4.7 Secondary ion mass spectrometry
4.8 Rutherford backscattering spectrometry
4.9 X-ray photoelectron spectroscopy
5 Summary of the results
6 Contributions and future work