The resistivity of commercial ITO is about 10-4 Ω·cm [14], which indicates that the use of the Zr-doped ZnO films as its replacement requires further resistivity reduction by at least one order of magnitude through controlling the deposition parameters. Based on a review covering the maximum carrier mobility and density published, the lowest resistivity limit for doped ZnO is at 1×10-4 Ω·cm due to limitations by ionised impurity scattering [145].
0 5 10 1.8 1.9 2.0 Refr active ind ex Zr content (at. %)
Figure 69: Film thickness dependence of carrier concentration and mobility for 4.8 at.% doped films.
The two ways of reducing the resistivity is the increase of either the carrier concentration or the carrier mobility. With doping the carrier density was shown to increase up to 4.8 at.% doping, which suggests that this is the maximum limit of extra carriers by Zr doping. The resistivity can be reduced through increasing the mobility by measuring the scattering. The dominant scattering effect found earlier (ionised impurity scattering) cannot be reduced as the doping must be kept fixed to maintain the carrier density. As a result, the microstructural related scattering could be reduced through thickness control, which is the most precise parameter controlled in ALD. This was investigated by systematically increasing the film thickness through controlling the number of ALD cycles ranging from 300-1500, and by keeping the recipe fixed for all the other parameters. By increasing the overall film thickness and keeping everything else fixed, the grains increased in size (see chapter
0 100 200 300 1x1020 2x1020 3x1020 4x1020 5x1020 6x1020
Carrier concentration Carrier mobility
Thickness (nm) Carr ier con cen tra tion (cm -3 ) 0 5 10 15 20 25 Carr ier mo bility (cm 2 /Vs)
4) resulting in the simultaneous decrease of dislocation scattering (reduction of dislocation density), interfacial scattering and grain boundary scattering. The surface roughness increase was the only parameter that might cause increase in carrier scattering.
The carrier mobility and density as a function of thickness are shown in Figure 69. The mobility trend starts as linear and then sublinear as thickness increases. This is expected as carrier mobility theoretically follows an asymptote line until it gets to the bulk value. This increase shows that the scattering is reduced, but as the grain size and thickness simultaneously increase it is not possible to determine experimentally which scattering mechanism is driving the mobility increase. Nevertheless, based on Vancea and Hoffmann studies [336], the grain boundary scattering is the most dominant when the crystals are in the order of a few nanometres, while the interfacial scattering based on Boltzmann formalism [149] is generally responsible for the resistivity reduction in terms of overall film thickness (i.e. ρ=ρ∞[1+G(t‒1, mean free path)] [149]). As a result, the resistivity reduction is often achieved through thickness increase as a method of interfacial scattering reduction, and it is reported to follow an asymptote trend line [36],[149],[330],[337]- [338].
The carrier density was expected to be fixed as the thickness increased, due to the fixed number of impurities added to the films. However, as shown in Figure 69, the carrier density follows an asymptote line with large changes at low thickness and very small changes between the thicker films. This could be associated to strain changes within the films. Strain within the film may shift the distribution of the density of states that affects the arrangement of electrons within the energy states [73]. As a result, compression strain in a lattice could cause electron rearrangement into higher energy states and increase the energy gap while a tensile strain will have the
opposite effect. Therefore, large grains which induce tensile strain in the film are related to a bandgap reduction, while reduction of their size results in compression and the increase of the bandgap. The energy gap changes may affect the population of the bands as the carrier density depends on the energy gap (i.e. n exp(-Eg)) [73]. As a result, in the current films the gap is expected to be reduced as the tensile strain increases with thickness, causing an increase of the carrier density. The strain increase shown in chapter 4 supports this theory as it follows the same trend as the Hall effect measurements shown in Figure 69.
Figure 70: Film thickness dependence of resistivity for 4.8 at.% doped films.
The resistivity decreases asymptotically by a factor of 3 as thickness increases from 50 nm to 250 nm (Figure 70), in a manner similar to the carrier density increase. The
0 100 200 300 1x10-3 2x10-3 3x10-3 4x10-3 Four-point-probe Van der Pauw
Resistivity (
cm)
drop-off effect for very thin films (<100 nm) is attributed to the strain sensitivity of very thin films before reaching the bulk thickness. The resistivity recorded is within the required order of magnitude at 7.5×10‒4 Ω·cm, while the carrier density reaches as high as 4.16×1020 cm-3 and the mobility up to 19.6 cm2V‒1s-1. In comparison to literature, the resistivity found in the current study was even lower than comparable thickness ALD films doped with Al having resistivity of 7.7×10‒4 Ω·cm [34] and 2.4×10‒3 Ω·cm [186], as well as for Ga-doped ZnO films having resistivity of 8×10-4 Ω·cm [38].