Figure 7.7a shows the energy distributions of the high energy population of O− ions for different total operating pressures (p = 0.33, 0.67, 1.34 and 2.00 Pa), measured at a constant axial position d = 100 mm. The average discharge power was maintained constant at Pd = 100 W and the pulse parameters were also constant; τ = 100 µs and
f = 100 Hz. The intensity of the high energy O− ion signal is observed to decrease rapidly with increasingp. The peak position of the high energy O−ions in figure 7.7a is observed to shift to lower energies for increasing pressure. This is attributed to a lower absolute target potential during the HiPIMS pulse on-phase being required to maintain a constant discharge power for operation at higher pressures. A significant decrease in the intensity of high energy O− ions was also observed for operating at a constant pressure (p= 1.34 Pa) but varying the target-to-substrate distance d, as shown in figure 7.7b.
The rapid decrease in the number of O−ions detected for both an increasing pressure at a fixed distanced= 100 mm and for a constant pressure at increasing distance away from the target can be understood by considering the transport of O−ions formed at the target surface through the background gas to the EQP probe. Using a simple binary collision analysis similar to that discussed in [184], it is possible to approximate the amount of high energy O− ions impacting a substrate at a given distance away from the target and given a certain pressure and gas mixture. For the simple treatment presented here, there
7. NEGATIVE ION ENERGY DISTRIBUTIONS IN REACTIVE HIPIMS
Figure 7.7. The high energy region of the energy distributions of O− ions measured for (a) different pressures at a constantd= 100 mm, and for (b) different target-to-substrate distances,d, for a constant p= 1.34 Pa.
are several approximations made:
1. Negative ions are hard spheres with a radius equal to that of the Pauling radius.
2. Negative ions interact only with neutral Ar gas atoms and O2 molecules via elastic
collisions. This assumption is justified due to the low particle ionization fraction outside of the magnetic trap (typically 10−2 to 10−4 [28]).
3. The background gas behaves like an ideal gas in thermal equilibrium at temperature Tgas, whereby the neutral particle density can be calculated by ng =p/kBTgas. The
7. NEGATIVE ION ENERGY DISTRIBUTIONS IN REACTIVE HIPIMS
background gas particles are also hard spheres with a radius equal to the van der Waals radius.
4. Gas density is uniform between the target and substrate; hence gas rarefaction effects and the sputtered vapour are not accounted for. The latter is justified by the low sputtering yield of reactive magnetron sputtering operated in poisoned mode.
5. Thermal motion of gas is neglected as Tgas is taken to be uniform and equal room
temperature (Tgas = 300 K), thus the background gas can be considered stationary
relative to the fast moving high energy negative ions (Teff,O− Tgas).
6. Any elastic collision deflects the negative ion sufficiently to make it undetectable. This approximation is employed due to the small acceptance angle of the EQP probe for detecting high energy ions [139].
Given the above approximations, it is possible to calculate the collisional cross section σ =πb2 wherebis the impact parameter equal to the sum of the radii of the two colliding
particles. The effective cross sections for a singly-charged negative atomic oxygen ion colliding with neutral argon atoms and oxygen molecules are given by
σAr =π(rAr+rO−)2
σO2 =π(rO2 +rO−)
2
(7.3)
whererO− = 176 pm,rAr = 188 pm andrO
2 = 152 pm. With the cross sections from equa-
tion 7.3, the effective total cross section of an Ar/O2 gas mixture where pO2/ptotal = 0.2 is
calculated to be 3.9×10−19 m2. By neglecting the motion of the background gas relative
to the high energy negative ions, the mean free path can be found straightforwardly by
λ = 1
nArσAr+nO2σO2
(7.4)
wherenArandnO2 are the particle densities of Ar and O2, respectively. ForpO2/ptotal = 0.2
and ptotal = 0.67 Pa, from equation 7.4 the mean free path is calculated to be approxi-
mately 16 mm. Since the negative ions that make it to the detector are assumed to be the uncollided fraction of initial flux, the probabilityχof detecting the negative ions with the orifice at a distance d above the target surface is
χ= exp −d λ (7.5)
As well as varyingng for a constant d = 100 mm, the position of the orifice was also
varied from d = 50 to 100 mm for two total gas pressures, p = 0.67 and 1.34 Pa. The HiPIMS discharge parameters were again maintained constant (f = 100 Hz, τ = 100 µs
7. NEGATIVE ION ENERGY DISTRIBUTIONS IN REACTIVE HIPIMS
and Pd = 100 W). The energy distribution was integrated for E > 425 eV and plotted
against the pressure-distance product (pd) in figure7.8, and is used to gain insight to the effects of both pressure and distance on the intensity of high energy negative ions arriving at the orifice.
Figure 7.8. The integral of the high-energy population (E > 425 eV) of O− ions plotted against the pressure-distance product (Ti target).
From equation 7.5 and assuming ideal gas behaviour, the probability χ for a high energy negative ion to be transported uncollided over a distance din an 80:20 Ar/O2 gas
mixture can be shown to be
χ= exp − σt kBTgas pd (7.6)
where σt is the effective total cross section. When equation 7.6 is compared with the
fit to the experimental data in figure 7.8, the yielded effective collisional cross section is σt = 2.2×10−19 m2.
To investigate any effects caused by the metal vapour, a similar investigation to that detailed above was performed using a Nb target. Measured O− energy distributions for p = 0.33, 0.67 and 1.34 Pa at a fixed d = 100 mm, and for d = 50−100 mm at a fixed p = 0.67 Pa are shown in figures 7.9a and 7.9b, respectively. The energy distributions were integrated for E > 550 eV and plotted against the corresponding values of pd and illustrated in figure 7.10. Comparing the exponential fit to equation 7.6, the effective collisional cross section for high-energy O− ions traversing an 80:20 Ar/O2 discharge
7. NEGATIVE ION ENERGY DISTRIBUTIONS IN REACTIVE HIPIMS
Figure 7.9. The high energy population of the energy distributions of O− ions mea- sured for (a) different pressures at a constant d = 100 mm, and for (b) different target-to-substrate distances, d, for a constant p = 0.67 Pa ob- tained during reactive HiPIMS of Nb.
during reactive HiPIMS of Nb was found to be σt,Nb = 2.1×10−19 m2. As the effective
cross sections are approximately equal for Ti and Nb targets, it is suggested that the gas- phase transport of O− is independent of the target material. This may not be the case for higher power densities or longer pulse widths where the displacement of the process gases in-front of the target surface and its subsequent replacement by metal gas is more pronounced.
Both effective cross sections are almost a factor of 2 smaller than the collisional cross section as calculated by assuming the approximations as outlined above (3.9 × 10−19
m2). The measured total effective cross section is found to be smaller for a number of
possible reasons. In the treatment above, no angular dependence is taken into account. It is possible that some O− ions are still detected by the EQP after experiencing very small angle deflections, resulting in a lower measured cross section. Furthermore, local gas heating in the vicinity of the target and gas rarefaction effects (see [185]) have been ignored.
Gas heating and rarefaction in DCMS has been widely reported [185–189] but inves- tigations in HiPIMS are not as fully developed. Due to the higher sputtered flux and stronger ‘sputter wind’ during the pulse on-phase in HiPIMS, it is expected that the gas heating and rarefaction effects may be exaggerated in comparison to DCMS. Simulations by Kadlec [64] predicted the transfer of momentum from sputtered Ti atoms to Ar gas atoms can result in heating of the neutral gas up to approximately 1 eV (≈11,000 K) in the vicinity of the target erosion track. This supposition has been supported by spatio- temporally resolved optical imaging of plasma emissions as performed by Liebiget al.[65] and Hecimovic et al. [66], where both sets of authors observed a decrease in the Ar0 neu-
7. NEGATIVE ION ENERGY DISTRIBUTIONS IN REACTIVE HIPIMS
Figure 7.10. The integral of the high-energy population (E > 550 eV) of O− ions plotted against the pressure-distance product (Nb target).
Furthermore, Horwat and Anders [59] used current-time waveforms as evidence for strong gas compression and rarefaction effects in the pre-target region for HiPIMS of copper in argon. It was also suggested by the same authors that another, more straightforward effect may contribute to gas heating. An elevated target surface temperature due to previous HiPIMS pulses may be responsible for local gas heating and density reduction. Indeed, this effect has been reported by Anders [40] for a hot Nb target. It is possible that the decrease in the number of high energy O− ions detected when increasing the target-to-orifice distance is less pronounced than anticipated due to the assumption of a uniform gas density and temperature distribution.
Knowledge of the attenuation of high energy negative ions at the substrate with re- spect to controlled discharge conditions (i.e. pressure and target-to-substrate distance) may be utilized in order to suppress or even eliminate the detrimental effects caused to growing films by the bombardment of high energy negative ions. One possible method would be to increase the effective collisional cross section via the introduction of heavier rare gases such as Kr or Xe, which is discussed in§7.3.6. However, in addition to reducing the impingement of high energy negative ions at the substrate this may also have conse- quences, detrimental or otherwise, for the deposition rate. Deposition rate measurements in reactive HiPIMS in different inert gases mixed with O2 are discussed further in chapter
7. NEGATIVE ION ENERGY DISTRIBUTIONS IN REACTIVE HIPIMS