The current-voltage-time waveforms for different average discharge powers (Pd = 25, 50
and 100 W) are presented in figure7.5. In each instance, other discharge parameters were maintained constant at f = 100 Hz, τ = 100 µs and p = 0.33 Pa while the target-to- orifice distance was also constant atd= 100 mm. The time-averaged energy distributions of high energy O− ions as obtained using these different average discharge powers were compared with two models outlined below.
Figure 7.5. The time variation of the HiPIMS discharge current Id(t) and target po-
tential Vd(t) for different average discharge powers Pd = 25, 50 and 100
W during HiPIMS of Ti in an Ar/O2 discharge where pO2/ptotal = 0.2.
Inset is a magnified plot of the target potential fort= 0 to 10 µs.
7. NEGATIVE ION ENERGY DISTRIBUTIONS IN REACTIVE HIPIMS
are sputtered from the target surface with a Thompson energy distribution [32] before being accelerated in the cathode fall, it is possible to describe the theoretical negative ion energy distribution in a pulsed plasma by
f(E)∼ Z pulse g(t) E−E0(t) [(E−E0(t)) +Us]3 dt (7.1) g(t) = 1 for E0(t)≤E 0 for E0(t)> E,
where E is the energy of the sputtered particle,Us is the surface binding energy and E0
is the energy offset, defined as E0(t) = qeVd(t) where q is the charge state and e is the
elementary charge. The energy of the sputtered particle is offset due to subsequent accel- eration in the cathode sheath, which is in-turn determined by the time-dependent target potential Vd(t). The full energy distribution in equation 7.1 is obtained by integrating
over the entire on-phase of the pulse; Vd(t) was obtained directly from the oscilloscope
and the surface binding energy Us of oxygen on a titanium oxide surface was taken to be
7 eV as proposed by Kubart et al. [182].
Additionally, the high energy population of the measured O− ion energy distribution was also compared to the target potential distribution function (as discussed in [116]) weighted by the discharge current waveform Id(t), hence assuming the intensity of high
energy O−ions is proportional to the discharge current, but that their energy is unaffected by sputtering. This distribution function was calculated as follows;
f(qeVi)∼F t<τ X t=0 ξ(t)Id(t), Vi∈ hVmin, Vmaxi (7.2) ξ(t) = 1 for Vd(t)∈ Vi− ∆2Vd, Vi+∆2Vd 0 for Vd(t)∈/ Vi− ∆2Vd, Vi+∆2Vd
whereVi,Vmin,Vmaxand ∆Vd are the accelerating potential difference, absolute minimum
and maximum values of the target potential and the potential difference interval used to separate the measured target potential range into discrete values (in this case ∆Vd = 5
V), respectively. The distribution was also multiplied by a scaling factor F for normal- ized comparison with the measured intensity of high-energy O− ions. The theoretical Thompson distributions with energy off-set (from equation 7.1), current-weighted target potential distributions (from equation 7.2) and the measured O− ion energy distributions for Pd = 25, 50 and 100 W are plotted in figure 7.6, with the corresponding HiPIMS
waveforms displayed in figure 7.5.
As illustrated in figure 7.6, the high energy population of O− ions is not well repre- sented by the modified Thompson energy distribution, consistent with the results reported
7. NEGATIVE ION ENERGY DISTRIBUTIONS IN REACTIVE HIPIMS
by Mr´az and Schneider [116]. It is thought that the disagreement between the theoretical distribution and experimental data arises primarily due to neglecting any collisional effects occurring between the target and substrate. By using the expected energy distribution of sputtered particles as calculated by SRIM [183] and subsequently accounting for gas phase transport, Mahieu et al.[110] have shown much better agreement with experimental data for the energy distributions of O− ions in reactive DCMS.
Figure 7.6. The measured energy distribution of high energy O− ions (black line) alongside the modified Thompson distribution (green line) and current- weighted target potential distribution (histogram).
7. NEGATIVE ION ENERGY DISTRIBUTIONS IN REACTIVE HIPIMS
Although the target potential distribution function seems to match the high energy region of the measured energy distribution of O− ions well for Pd = 25, this is not the
case for Pd = 50 or 100 W. The implication of these results is that the energy distribution
of high energy O− ions created in reactive HiPIMS cannot be readily determined by the target potential distribution function, particularly at higher average discharge powers. This is in contrast to the results reported by Mraz and Schneider [116], whereby the authors concluded that the high energy range of the O− ion energy distribution can be determined exclusively by the target potential distribution function during the reactive pulsed DCMS of Al in an Ar/O2 atmosphere. It was further speculated that negative ion
generation by desorption from the target surface must then be the dominant formation mechanism. However, the results presented here suggest that in reactive HiPIMS of Ti in an Ar/O2 gas mixture, the high energy O− ions are formed as a consequence of electron
tunnelling or attachment to O atoms at the target surface. The surface O− ions formed at the target surface are subsequently sputtered and accelerated in the cathode fall region and hence possess additional energy provided by the sputtering event. This additional energy, however, is not explained by the theoretical Thompson distribution alone and the gas transport phase may need to be considered, in-line with the more rigorous treatment presented by Mahieu et al. [110].