Effect of varying N₂ pressure on DC arc plasma properties and microstructure of TiAlN coatings

The IEDFs of Ti, Al, TiN, N and N₂ are shown in figure 3 and they contain multiple peaks. These peaks merge into a single broad peak when the N₂ pressure is increased. At 0 Pa (N₂), Ti¹⁺ is the dominant fraction among all the Ti ions in the plasma. As the N₂ pressure increases, Ti²⁺ becomes the dominant Ti ion fraction. Al¹⁺ is the major fraction of Al ions irrespective of the N₂ pressure. We find that the maximum velocities of Ti and Al are similar, which is consistent with the velocity rule [20].

Zoom In Zoom Out Reset image size

The average ion energies, extracted from the respective IEDFs of Ti, Al and N are shown in figure 4(a). The average energies of Ti, Al and N ions follow a decreasing trend when the N₂ pressure increases, which indicates a similar trend that has been reported previously [21, 22]. The average ion energy of Ti is higher than that of Al at all the N₂ pressures studied. This higher average ion energy of Ti compared to Al is in line with previous observations for a Ti-50 at.% Al cathode [20]. The average charge states of Ti and Al ions are shown in figure 4(b). The average charge state of Al is close to 1+ for all the investigated pressures, whereas the average charge state of Ti is close to 1+ for 0 Pa of N₂ and remains above 1.7+ for all the other N₂ pressures.

Zoom In Zoom Out Reset image size

The following relation is used to calculate the ion saturation density for each species at different N₂ pressures:

$$\begin{equation*}{J}_{\mathrm{s}\mathrm{a}\mathrm{t}}^{s}=\frac{{q}_{S}\int {I}_{s}\enspace \mathrm{d}E}{\sum _{i}{q}_{i}\int {I}_{i}\enspace \mathrm{d}E}{\times}{J}_{\mathrm{i}\mathrm{o}\mathrm{n}\mathrm{s}\mathrm{a}\mathrm{t}}\end{equation*}$$

where ${J}_{\mathrm{s}\mathrm{a}\mathrm{t}}^{s}$ is the ion saturation density of species S, q_S the charge state of species S, I_S the intensity (measured by MEA) of species S, and J_ionsat the total ion saturation density (measured by flat probe). The measured individual ion saturation densities of the plasma species are plotted according to their fraction of the total ion saturation density as a histogram in figure 5. The total ion saturation current density shows a sharp decrease from 0 Pa to 0.07 Pa, while a further increase in the N₂ pressure results in a more gradual decrease. The N, N₂, and TiN ions contribute marginally to the total ion saturation current density, which instead is dominated by the metal ions. Ti ions contribute the most to the total ion saturation current density irrespective of the N₂ pressure. At 0 Pa the ratio Ti-ions/Al-ions is approximately 60/40 and for the rest of the N₂ pressure range the ratio is closer to 70/30. The major fraction of Ti ions is Ti¹⁺ at 0 Pa, while it is Ti²⁺ when N₂ is present in the chamber. Figure 5 also shows that N¹⁺ starts to appear as Al²⁺ decreases, and when Al²⁺ vanishes at 0.2 Pa, an immediate increase in N¹⁺ occurs. With further increase in N₂ pressure, N¹⁺ starts to decrease and ultimately disappears at 1 Pa.

Zoom In Zoom Out Reset image size

The non-Maxwellian distribution of the IEDFs (see figure 3) suggests that the conventional Maxwellian assumptions to determine electron temperature and electron density from the Langmuir probe measurements are not valid. The Druyvesteyn method of Langmuir probe analysis, which assumes non-Maxwellian electron energy distribution, is used here to calculate EEDF from the second derivative of Langmuir probe I–V characteristic curve using the formula [14]:

$$\begin{equation*}f\left(\varepsilon \right)=\frac{2{m}_{\mathrm{e}}}{{e}^{2}\cdot A}\sqrt{\frac{2e\left({V}_{\mathrm{p}}-{V}_{\mathrm{b}}\right)}{{m}_{\mathrm{e}}}}\frac{{\mathrm{d}}^{2}{I}_{\mathrm{e}}}{\mathrm{d}{V}_{\mathrm{b}}^{2}}\end{equation*}$$

where f(ɛ) is the EEDF, A the probe area, e the charge of electron, m_e the mass of electron, V_p the plasma potential, V_b the bias on the probe, and I_e the electron current. Digital smoothing of the I–V curves was performed by applying the Blackman window (B.W) filter [23]. The measured and B.W filtered I–V curves recorded at 0.6 Pa N₂ are shown in figure 6.

Zoom In Zoom Out Reset image size

For illustrative purposes, the EEDF is usually represented in the form of the electron energy probability function (EEPF) (see figure 7), computed by dividing EEDF with the square root of electron energy ($\sqrt{\varepsilon }$). At 0 Pa the EEPF appears to have a bimodal distribution which changes to a unimodal distribution at higher N₂ pressures. The energetic tails of the EEPFs diminish with an increase in N₂ pressure, while the population of low energy electrons increases, indicating increased collisions among plasma species that lead to their thermalization.

Zoom In Zoom Out Reset image size

The plasma potential (700 mm away from the cathode) is measured from the inflection point of the second derivative of the probe current, while the floating potential is the point where the probe current becomes zero. The difference between the floating potential and the plasma potential is the sheath potential. The plasma, floating, and sheath potentials are shown in figure 8 as a function of N₂ pressure. The magnitude of the sheath potential decreases with increasing N₂ pressure.

Zoom In Zoom Out Reset image size

The electron density and effective electron temperature [14] are shown in figure 9, which are calculated by the following relations:

$$\begin{equation*}{n}_{\mathrm{e}}=\int f\left(\varepsilon \right)\mathrm{d}\varepsilon \end{equation*}$$

$$\begin{equation*}{T}_{\mathrm{e}\mathrm{ff}}=\frac{2}{3}{\times}\frac{1}{{n}_{\mathrm{e}}}\int \varepsilon .f\left(\varepsilon \right)\mathrm{d}\varepsilon \end{equation*}$$

where n_e represents the electron density, $f\left(\varepsilon \right)$ the EEDF, ɛ the electron energy, and T_eff the effective electron temperature. The electron density increases with increasing N₂ pressure, while the effective electron temperature shows a steep decrease from 0 Pa to 0.2 Pa, and at higher N₂ pressures the decrease is more gradual.

Zoom In Zoom Out Reset image size

Figure 10 shows the morphologies and cross-sections of the coatings deposited at 0.07 Pa, 0.2 Pa, 0.4 Pa, 0.6 Pa and 1 Pa N₂. All the coatings contain macroparticles, which are inherent to the explosive nature of the DC cathodic arc evaporation process. The coating deposited at 0.07 Pa has a glassy structure. A similar glassy structure is observed for coating deposited in the absence of N₂ (not shown here). All the coatings deposited at N₂ pressures ⩾0.2 Pa have a columnar structure. At 0.07 Pa the topography of the coating shows globular grains which transform to faceted grains at 1 Pa. The thickness of the coatings decreases with increasing the N₂ pressure. The thickness reduction by approximately 20%, 33% and 43% is observed for the coatings deposited at 0.4 Pa, 0.6 Pa and 1 Pa, respectively, compared to the coatings deposited at 0.2 Pa.

Zoom In Zoom Out Reset image size

X-ray diffractograms of the coatings are shown in figure 11. The coating deposited at 0.07 Pa N₂ pressure has very broad or no significant peaks, while the coatings deposited at N₂ pressures ⩾0.2 Pa show a cubic B1 structure. The crystallographic orientation along the growth direction is predominantly 220 for the coating deposited at 0.2 Pa. At 0.4 Pa the coating shows almost equal intensities of 111 and 220 peaks. At N₂ pressures of 0.6 Pa and 1 Pa, the crystallographic orientation along the growth direction predominantly becomes 111. The decrease in full width half maximum of the 111 peak with the increasing N₂ pressure indicates an improved crystalline quality of the coatings. The residual stress in the coatings decreases when increasing the N₂ pressure. At N₂ pressures in the range of 0.2 Pa to 0.6 Pa, the coatings are in a compressive residual stress state. At 1 Pa the residual stress in the coating is tensile. α (Ti), γ (TiAl) and α₂ (Ti₃Al) phases are observed in the coating deposited at 0 Pa.

Zoom In Zoom Out Reset image size

The compositions of coatings deposited at 0.07 Pa, 0.2 Pa, 0.6 Pa, and 1 Pa were measured by the ToF-E-ERDA and the results are shown in figure 12. The figure shows that the content of N increases with the increase in N₂ pressure, while the content of O and H stays below 4.05 at.% and 1 at.%, respectively, for all the coatings. Another noticeable feature is that the Ti to Al ratio is approximately close to 1 in the coatings deposited at 0.07 Pa and 0.2 Pa, and at higher N₂ pressures the Al content in the coating decreases slightly.

Zoom In Zoom Out Reset image size

The mean free path of electrons and ions decreases as more N₂ is introduced in the chamber. Consequently, the probability for them to suffer inelastic collisions with gas atoms or molecules increases. Such collisions are responsible for the observed reduction in effective electron temperature (see figure 9) and the reduction in average energies of ions (see figure 4). This reduction in effective electron temperature is proportional to the reduction in effective electron velocity in plasma. The reduction in effective electron velocity means that the electrons stay for a longer time in the plasma. Since the arc is continuously operating at a constant current, an increase in electron density is expected. Another factor contributing to the increase in electron density (see figure 9) with the increase in N₂ pressure is the increased probability of electrons being knocked out of the gas during inelastic collisions between electrons and the gas.

The bimodal distribution of the EEPF at 0 Pa is related to the fact that the cathode is the source of thermally emitted and thermo-field emitted electrons. With the increase in N₂ pressure both distributions thermalize and appear as a single distribution.

In the plasma, electrons have a smaller inertia than all the ions and therefore they are easily influenced by perturbations (e.g. chamber walls, plasma probes etc) which can cause the electrons to leave the plasma faster than the ions. As quasineutrality is an inherent property of the plasma, a potential (i.e. sheath) is developed to impede the flow or loss of electrons in order to maintain the quasineutrality.

At 0 Pa the electrons have a higher effective velocity (${V}_{\mathrm{e}\mathrm{ff}}=\sqrt{2{T}_{\mathrm{e}\mathrm{ff}}/{m}_{\mathrm{e}}}=9.1{\times}1{0}^{5}\enspace \mathrm{m}{\mathrm{s}}^{-1}$) compared to at 1 Pa and a higher sheath potential (−4.91 V) is thus required to retard the electron flow to the Langmuir probe (see figure 8). At 1 Pa the effective velocity of electrons decreases by 42% compared to at 0 Pa and thus a low sheath potential of −0.65 V is sufficient to impede the flow of electrons toward the probe.

In the absence of N₂ gas the cathode spot can be expected to have the type 2 behavior [24] that has a characteristic feature of a longer duration (slower velocity) of the cathode spot compared to type 1 [25]. Due to this longer duration, the area around the cathode spot tends to reach a higher temperature and more metal neutrals are thus expected to be emitted [26]. In such a plasma the charge exchange collision reaction Ti²⁺ + Ti → Ti¹⁺ + Ti¹⁺ is likely to occur, which promotes the generation of Ti¹⁺. In the presence of N₂ gas, the cathode surface forms a nitride layer [9, 10] where the cathode spot is expected to have the type 1 behavior [24] (i.e. a high velocity [25]) that leads to a reduction in the emission of metal neutrals [26], and thus the Ti²⁺ + Ti → Ti¹⁺ + Ti¹⁺ reaction becomes less probable. This explains how the dominant Ti ion species transition from Ti¹⁺ to Ti²⁺ as the N₂ pressure is increased from 0 Pa to 0.07 Pa (see figures 2 and 4). This transition is also manifested by a sudden increase in the average charge state of Ti ions between 0 and 0.07 Pa (see figure 4(b)).

Another observation of the impact of N₂ pressure is on the population of Al²⁺ in the plasma. The stark decrease in the Al²⁺ signal with increasing N₂ pressure (see figures 2 and 4) can be understood from the fact that Al²⁺ has an ionization energy of 18.82 eV [27], which is sufficient to ionize N₂ (ionization energy of 15.58 eV [27]) and N (ionization energy of 14.53 eV [27]), while Ti²⁺ has a lower ionization energy of 13.57 eV [27].

Studies of single element cathodes show that Al ions have a wider angular spread than Ti ions [28]. The angular spread of Al ions becomes even wider when a compound Ti–Al cathode is used since lighter species in the plasma from a dual element cathode tend to have a wider spread compared to its angular spread from a single element cathode [29]. One can thus expect the Al ions to have a substantially wider angular spread in front of the cathode compared to that of Ti ions in the plasma from a Ti-50 at.% Al cathode. This wider angular spread of Al ions causes a decrease in the density of Al ions and a lower flux of Al ions is thus expected to be collected by the MEA probe. This is also the reason for the apparent dearth of Al ions in the present work (see figure 5).

The ion composition of the plasma in the presence of N₂ and at 700 mm away from the Ti-50 at.% Al cathode consists of Ti ions (the majority species), Al ions and a small fraction of TiN, N₂ and N ions (see figure 5). However, the coatings deposited with such plasma contain almost an equal atomic percentage of Ti and Al with a significant fraction of nitrogen (see figure 12) which suggests that significant flux of Al and N (and possibly Ti) neutrals reaches the substrate and contributes to the growth of coatings (see figure 12).

A significant source of neutrals in the cathodic arc plasmas is the surfaces inside the deposition chamber [30]. Due to low sticking coefficient of cathodic arc plasma ions, the ions colliding with the surfaces do not stick and return to the plasma as neutral vapor [31, 32]. As stated earlier that Al ions have a wider spread compared to Ti ions, it can be expected that these Al ions when striking the walls of the chamber return to the plasma as Al neutrals and contribute to the growth of TiAlN coatings. Other sources of Al neutrals can be hot craters left by cathode spots (due to a lower boiling point of Al compared to Ti, it is more easily evaporated), evaporating macroparticles [33], and sputtering from the cathode [34].

As stated earlier the plasma species suffer more inelastic collisions as the N₂ pressure is increased. This increase in number of inelastic collisions and the fact that Al and N₂ have similar masses lead to the reduction in the Al flux at the coatings growth front. This is the reason why a slight decrease in Al content is observed in the coatings deposited at higher N₂ pressures (see figure 12).

The IEDFs of the metal ions show three distinguishable distributions, which become indistinguishable with increasing N₂ pressure (see figures 3 and 13(a) and (b)). These distributions can be categorized as low, mid and high energy distributions. The low energy distribution in the IEDFs (figures 13(a) and (b)) represents a thermalized ion flux with a peak ion energy roughly corresponding to the plasma potential (see figure 8). The high energy distribution in the IEDFs may represent ions that did not suffer any significant energy loss before reaching the analyzer. The origin of these high energy ions (⩾30 eV for Al, ⩾55 eV for Ti, for 0 Pa [35]) is explained by Anders [36] and Tanaka et al [37] as a combination of two factors: (i) the effective depletion of electrons in the vicinity of the cathode spot causes a double layer or a potential hump to form which accelerates the ions into the chamber (potential hump model); (ii) the high density in the cathode spot generates a pressure gradient with the surroundings, and this pressure gradient plus electron–ion coupling contribute to the high energy attained by ions (gas-dynamic model). The mid energy distribution in the IEDFs stems from charge-exchange collisions. Since a range of different charge-exchange collisions may occur, multiple peaks can be seen in the IEDFs [38, 39]. The charge-exchange collisions are expected to yield ions whose energy is higher than the completely thermalized ions but lower than the ions generated inside the cathode spot.

Zoom In Zoom Out Reset image size

To observe the relative changes in the three distributions of Ti¹⁺ and Al¹⁺, a shifted-Maxwellian function, as used by Bilek et al [21] and Atiser et al [40], is fitted to the distributions. The shifted-Maxwellian function $f\left(E\right)$ is:

$$\begin{equation*}f\left(E\right)=\sum _{i=1}^{3}{C}_{i}\left(E-{V}_{\mathrm{p}}\right)\mathrm{exp}\left[-\frac{{\left({\left(E-{V}_{\mathrm{p}}\right)}^{1/2}-{\left({E}_{\mathrm{d}\mathrm{i}{\mathrm{r}}_{i}}\right)}^{1/2}\right)}^{2}}{{E}_{\mathrm{r}\mathrm{a}{\mathrm{n}}_{i}}}\right]\end{equation*}$$

where C_i is a scaling constant for the ith distribution, V_p is the plasma potential (measured by Langmuire probe) close to the entrance of the analyzer, ${E}_{\mathrm{d}\mathrm{i}{\mathrm{r}}_{i}}$ is the directional energy (center of mass of the distribution) of the ith distribution, and ${E}_{\mathrm{r}\mathrm{a}{\mathrm{n}}_{i}}$ is the random energy (full width at half maximum of the distribution) of the ith distribution. The best fits of $f\left(E\right)$ to the IEDFs recorded at 0.6 Pa are shown in figure 13(a) and (b). The directional (E_dir) and random (E_ran) energies of all the distributions are plotted as a function of N₂ pressure in figures 13(c)–(f). The exponential functions fitted to the E_dir and E_r data, shown by solid lines, are intended to highlight the trends. With an increase in N₂ pressure, the directional energy of all the three distributions of Ti¹⁺ and Al¹⁺ portrays a decreasing trend. The random energy of both Ti¹⁺ and Al¹⁺ decreases with increasing N₂ pressure for the low and mid energy distributions, but an increasing trend is visible for the high energy distribution with the change in N₂ pressure.

The shifts of the directional energies (E_dir) for low, mid, and high energy distributions (see figures 13(c) and (d)) toward lower energies is an indication of increasing inelastic collision rates with increasing N₂ pressure. An increase in N₂ pressure results in a decrease in the random energies (E_ran) for the low and mid energy distributions, however it increases E_ran for the high energy distribution (see figures 13(e) and (f)). These changes indicate that the high energy distribution suffers more collisions and a significant portion of this distribution is transformed into the mid and low energy distributions.

The aforementioned loss of a significant fraction of the high energy distributions to the mid and low energy distributions at higher pressures of N₂ influences the microstructure of deposited coatings. An energetic ion flux is known to favor a 220 growth texture in TiN coatings since this surface has the lowest resputtering yield compared to other crystallographic orientations [41]. At 0.2 Pa, the ion flux consists of a considerable fraction of the high energy ion distribution compared to 0.4 and 0.6 Pa. Thus, the coating's growth front in the case of 0.2 Pa experiences bombardment by a higher energy flux compared to the coatings deposited at 0.4 and 0.6 Pa, which results in a 220-texture and a higher compressive stress (see figure 11). When the N₂ pressure increases or the energetic bombardment decreases, a 111-texture becomes favorable (see figure 11).