One of the most commonly used dopants is Al, introduced by different growth methods such as RF magnetron sputtering [176]-[179], CVD [180]-[181], spin coating technique [182], ionised deposition [183], sol-gel method [184], spray pyrolysis [185], and ALD [33]-[34],[167]-[169],[186]-[188]. The selection of Al as a dopant is based on its low cost over large scale applications and on its electrical and optical properties when added to ZnO. In more detail, Al3+ provides one extra carrier when ionised and substitutes to Zn2+ leading to high carrier density and thus to lower resistivity. In addition, Al3+ has smaller ionic size (0.57Å) than Zn2+ (0.72Å) [185] and thus could occupy both interstitial and Zn positions in the film [40].
The resistivity values of ALD films [33],[186]-[187] were slightly higher than spray pyrolysis [185], magnetron sputtering [176]-[177] and much higher than PLD [161] as shown in Table 2. However, the resistivity of the ALD films was reported lower when using sapphire as the substrate (i.e. from 3.0×10-3 Ω·cm on glass to 7.7×10-4 Ω·cm on sapphire), attributed to the c-axis textured grains formed on sapphire resulting in less scattering than the random oriented grains on glass [34].
Gallium is another widely used dopant for ZnO films in TCO replacements of ITO, since Ga is economical and more abundant than indium. It also offers less grain reduction in comparison to other dopants such as In, Zr and Sn [29], and it is placed as
substitutional to Zn in the lattice [40]. The Ga-doped films often show lower resistivity and higher transparency than Al-doped films, and are preferred due to their resistance to oxidation and the fact that Ga is less reactive to Zn rather than Al [189]
. Comparable studies between Al and Ga dopants showed higher carrier density for Ga (5×1020 cm‒3) compared to Al (3×1019 cm‒3) [40]. ALD deposited films showed comparable resistivity (8.0×10-4 Ω·cm [38]) to the Al-doped films [34], but still much higher than the one reported for PLD films (8.12×10-5 Ω·cm) [162].
Table 2: List of published doped ZnO films using different dopants and growth techniques. Growth method Resistivity (Ω·cm) Carrier density (×1020cm-3) Carrier mobility (cm2/Vs) Transp arency Optical gap (eV) Growth T (°C) Film thickness Al SP1[185] 7×10-4 3.6 25 88% 3.7 450 1.4 μm RF MS2[176] 4.6×10-4 7.2 18.8 90% - 250 130 nm RF MS (H- annealed at 300°C) [177] 8.3×10-4 8.86 9.7 90% 3.7 - - PLD [161] 8.5×10-5 15.4 47.6 88% - 230 280 nm ALD [33] 3.2×10-3 1.4 14.3 - - 200 45 nm ALD [186] 2.4 ×10-3 - - >80% 3.5 150 180 nm ALD [187] 4.4×10-3 1.7 8 92% 3.73 150 100 nm Ga RF MS (annealed at 550°C) [189] 1.5 ×10-3 1.4 29.4 - - - 1.4 μm PLD [162] 8.1×10-5 146 30.96 90% 3.5 300 200 nm ALD [38] 8.0×10-4 1.1 18 - - 300 - In SP [22] 5.6×10-1 - - 90% 3.3 450 600 nm RF MS [23] 4.4 ×10-3 18 10.1 >80% - - 1.5 μm ALD [192] 3×10-3 6 20 >85 - 200 180 nm Ti ALD [171] 8.9 ×10-4 6.2 10 >80% 3.4 200 100 nm Ni Sol-gel [173] 4.8×10-4 - - 91.2% - 200 - Hf ALD [28] 6×10-4 3.7 20 >80% 3.56 200 200 nm 1 Spray pyrolysis 2 RF magnetron sputtering
A direct comparison between Al, Ga and In showed that In resulted in the highest carrier density [190], possibly due to the formation of oxides, as In substitutes to Zn
and combines to O interstitials [191]. However, indium is not preferred as a dopant for ZnO as the aim for using doped ZnO as TCOs is to replace ITO due to In scarcity.
2.4.1.2 Co-doping
Co-doping of ZnO films has been developed in the recent years, to either enhance n- type conductor or to explore p-type ZnO. Most of the reported studies are theoretical predictions, as it is very difficult to experimentally achieve stable and improved properties with co-doping. The experimentally reported studies are under dispute for their accuracy of films with stable properties. An approach for p-type ZnO films is the use of dual acceptor co-doping [193], such as Li-Ni co-doped by PLD [195], or even donor-acceptor co-doping, such as Al-N co-doped ZnO by ALD [194], and Li-Zr co- doped ZnO by aqueous solution [196]. On the other hand, donor-donor doping could enhance further the electrical n-type properties by improving structural defects caused by a single dopant. An example is the surface degradation caused by Ga doping, which was shown to be avoided with the introduction of In, showing reduced resistivity of 7.4×10‒4 Ω·cm compared to GZO with resistivity of 1.04×10‒3 Ω·cm [197]
. Other combinations of donor co-dopants include Co- Al [198] and Co-Ga [199].
2.4.2 Zirconium doped ZnO films by different techniques
Zirconium was chosen as the dopant in this work due to its high abundance in Earth‟s crusts and because it acts as a shallow donor in ZnO providing two extra free
electrons when substitutes to Zn2+. Zirconium is also a transition element and it is considered to be a good doping candidate based on its comparable ionic size to Zn [43]
(i.e. 0.745 Å for Zr and 0.74 Å for Zn [42]). This helps to minimize lattice distortion usually observed with other dopants when substituting for Zn2+ (e.g. Al) [29]
that could affect the mobility, the electron distribution and the band structure. The minimisation of the lattice changes does not correspond to zero changes, in contrast, Zr-doped ZnO studies showed a small increase in the d-spacing of the lattice when adding Zr [26],[56]. An additional advantage is the reluctance of Zr to bond with Zn atoms, avoiding the formation of secondary (inter-metallic) phases and therefore enhancing the stability of the target phase [44].
The use of Zr provides the extra electrons required to increase the carrier density and so enhance the n-type conductivity. However it is expected to result in a small mobility reduction due to its tendency to cluster into the grain boundaries which reduces the grains size [29],[48],[57]. This suppression of grain growth is more intense at high doping when ZrO2 molecules are formed and placed at grain boundaries [49],[57]. The mobility and grain size reduction could both be used as an indication of the dopant segregation at grain boundaries in the current study, which can exists in epitaxial films without the formation of clusters due to the dopants mobility within the grain boundaries.
A number of publications using Zr-doped ZnO were reported using different deposition techniques, such as spray pyrolysis [26],[45]-[47], low temperature coprecipitation method [48], sol-gel method [29],[49],[50],[200], DC magnetron sputtering [51]-[54]
, RF magnetic sputtering [56], microwave irradiation [55], low temperature gel- combustion [201], PLD [58], and ALD [25]. The number of related studies is much lower
than for other dopants such as Al and Ga, and thus various gaps remain in the understanding of Zr doping. In particular for ALD growth, there is only one published study for Zr-doped ZnO thin films [25] reporting conductive and transparent films, but also leaves scope for further research which is addressed in the current study. The findings of the published ALD study and the research gaps identified are discussed in the zirconium in ALD section (2.5.3).
A theoretical study by Wang et al. [43] using first principle calculations, reported the most likely position of Zr in the ZnO lattice based on the calculated formation energies. This study suggests Zr is likely to be substituted onto Zn sites as this arrangement has the lowest formation energy and causes the lowest lattice distortion (due to ionic radius similarity). According to those calculations, ZrZn will result in Fermi level shift into the CB and thus the conductivity will be increased.
Experimentally, Zr-doped ZnO films were shown to reduce the resistivity compared to the un-doped ZnO films, with the most effective doping being typically between 1-5 at.% Zr. Films prepared by sol-gel showed resistivity of 7.2×10‒2 Ω·cm at a low doping level of 1.5 at.%, and increased again as the doping increases further [49]. In the same study, possibly due to the low carrier density (1.5×1019 cm-3), the optical measurements showed no Burstein-Moss effect and the bandgap was found to reduce with increasing doping. In another study using the same deposition technique, ZnO nanoparticles with Zr doping, showed enhanced oxygen vacancies concentration with doping that resulted in room temperature ferromagnetism while pure ZnO showed paramagnetism [200]. Different nanostructures with Zr doping have also been reported using a low temperature coprecipitation method [48]. It showed that the
bandgap can be manipulated by controlling the morphology of the material (i.e. nanoflakes, nanorods, spherical nanoparticles).
DC magnetron sputtered films were also reported, including a study of thick doped films (450 nm) that showed high transparency up to 92% with resistivity of 9.8×10-4 Ω·cm after vacuum post-annealing (300°C) [53], more conductive than a study using RF magnetron sputtering of 300 nm thick films (2.1×10‒3 Ω·cm) [57]. Polymer substrates of polyethylene terephthalate (PET) were also used by DC magnetron sputtering, showing low resistivity up to 1.8×10‒3 Ω·cm and transparency up to 86% [51]. Spray pyrolysis method resulted in higher resistivity films (100 nm) of 6.7×10-2 Ω·cm, reduced to 2×10‒3 Ω·cm after vacuum annealing at 400°C [26]. The most conductive Zr-doped films were reported using PLD, with resistivity of 5.6×10-4 Ω·cm for a 200 nm film, and optical transmittance of 84% [58].
According to Paul et al. [49], the grain orientation was initially random for the un- doped films, then formed a-axis preferred orientation (10 ̅1) at 1 at.% before shifting to c-axis orientation (0002) at 2 at.%, which is enhanced as doping increased further to 3 at.% [49]. The increase of c-axis preferred orientation with increasing doping was also observed by Bahedi et al. using spray pyrolysis [45]-[47]. Other studies of Zr- doped ZnO films, reported increase of amorphization as doping increases using RF magnetron sputtering [56] and spray pyrolysis [26]. This trend was explained by Gokulakrishnan et al. as the effect of Zr being placed in interstitial positions and not to zinc sites [26]. This suggests that increase of c-axis alignment corresponds to Zr being placed in zinc sites, with the cause being discussed in chapter 4.
The larger ionic size of Zr atoms was reported to result in larger lattice constants according to Tsay et al. [50] and Zhang et al. [200] using sol gel method [50], Khan et al.
using gel-combustion method [201], Wang et al. using RF magnetron [56],and Gokulakrishnan et al. using spray pyrolysis [26]. As a result, it can be said that chemical mixing methods result in increased lattice constants with the addition of Zr atoms.
The structural properties of Zr-doped ZnO films compared to other dopants such as Ga, In and Sn were studied by Tsay et al. using sol-gel [29]. This study showed that Ga and In doped ZnO exhibited the lower resistivity due to the formation of larger grains in comparison to Zr and Sn doped ZnO, ultimately resulting in lower grain boundary scattering [29]. The same effect of grain size difference was also reported when compared to Al-doped films, where Zr was reported to show slightly higher resistivity (6.1×10‒4 Ω·cm) than Al-doped films (5.7×10‒4 Ω·cm) deposited by DC magnetron sputtering [54]. The reduction of grain size due to Zr doping is yet to be published for ALD, thus this is also investigated in this thesis.