Quantitative evaluation of high-energy O⁻ ion particle flux in a DC magnetron sputter plasma with an indium-tin-oxide target

Using the QMA-EA, various negative ions in the magnetron plasma are measured. From mass spectra of negative ions at various kinetic energies, abundant O⁻ was measured with a negligibly small intensity of ${\rm O}_{2}^{-}$. The result is consistent with our previous measurements of negative ions in the case of MgO [19, 20]. Hereafter, we focus on O⁻ ion as the high-energy negative ion in the ITO sputter plasma. Figure 3 shows the O⁻ energy distribution function (EDF) from r  =  1.6 to 4.0 cm at a pressure of 0.2 Pa. Discharge voltage and discharge current are 280 V and 0.15 A, respectively. The O⁻ EDF shows an intense peak at ~280 eV irrespective of the radial position. The measured O⁻ energy (~280 eV) is the same as the energy accelerated by the target potential (280 V). This means that O⁻ ions are produced on the target surface and are accelerated by the electric field of the cathode fall in the vicinity of the target surface. It should be also noted that the peak intensity shows its maximum at r  =  2 cm. This result indicates that the O⁻ ions are mostly produced at the plasma-ring position on the target surface and that the produced high-energy O⁻ ions are ejected from the target perpendicularly from the target surface to the QMA-EA at z  =  10 cm.

Zoom In Zoom Out Reset image size

Figure 4 shows pressure dependence of the O⁻ peak intensity at r  =  2 cm and z  =  10 cm. The O⁻ peak intensity decreases almost exponentially with the Ar pressure. Axial and pressure dependences of high-energy O⁻ flux Γ is expressed by the following equation

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle \Gamma ={{\Gamma }_{0}}\cdot \exp \left( -\sigma {{n}_{{\rm Ar}}}z \right)={{\Gamma }_{0}}\cdot \exp \left( -pz/K \right).\nonumber \end{align} \tag{ 1 }$$

Zoom In Zoom Out Reset image size

Here, Γ₀, σ and n_Ar are initial O⁻ flux at the target surface, scattering cross section of O⁻ by Ar and number density of Ar, respectively. From the result, we can derive the decay parameter K is 1.8  ×  10⁻² Pa m, which is useful to correlate heat flux at different axial position.

Heat flux is measured using the calorimetry method [22]. To evaluate spatial profile of the heat flux using the calorimetry method, the TC is scanned across the vessel at z  =  10 cm and temporal variation of the TC temperature is measured. Figure 5(a) shows an example of the TC temperature after opening the shutter. The TC temperature starts to increase just after opening the shutter and the temperature increase rate gradually decreases to a saturation temperature. Conservation of heat energy in the TC can be expressed in a simplified form as follows:

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle JS=C\frac{{\rm d}{{T}_{1}}}{{\rm d}t}-hA({{T}_{1}}-{{T}_{2}}).\nonumber \end{align} \tag{ 2 }$$

Zoom In Zoom Out Reset image size

Here, J is energy flux to the TC tip per unit area and time, S, C and A are surface area, heat capacity of the TC tip to receive the energy and cross section of the thermocouple, h is heat transfer coefficient between the TC tip at a temperature of T₁ and the root of the TC wire at a fixed temperature T₂. Supposing that the TC is at a thermally equilibrium condition before the plasma exposure, T₁ is considered to be equal to T₂. After opening the shutter (t  >  0), the TC tip starts to receive the heat flux and the time-derivative of the TC temperature (dT₁/dt) at t  =  0 becomes proportional to the heat flux. From this, (dT₁/dt)_t=0 (slope of a broken line tangential to the temperature curve at t  =  0) is measured as relative heat flux. Figure 5(b) shows the relative heat flux as a function of the rotation angle of the TC in the polar coordinate. The TC is positioned at r  =  2 cm and z  =  10 cm. Here θ  =  0° corresponds to the direction where the shield plate of the TC is facing to the target. At θ  =  0°, i.e. the condition that the heat flux directly coming from the target is shielded, decrease in the relative heat flux is clearly observed. The full-width at half-maximum of the heat flux dip Δθ is ~25°, which value is close to a shielding angle of the shield plate (~20°) calculated from geometrical structure of the TC and the shield plate. The result indicates the existence of anisotropic heat flux that reaches the TC from the direction of the target.

Next, the radial profile of the relative heat flux is measured both in the cases of with-shield (θ  =  0°) and without-shield (θ  =  180°) conditions, as shown in figure 6. When the shield is turned to the direction of the ITO target (θ  =  0°), heat flux is rather uniform along r direction up to 2 cm and gradually decreases at r  >  2.5 cm. Plasma density was measured by a Langmuir probe scanning along the radial direction at the same discharge condition, and the plasma density profile was similar to that of the radial profile of the relative heat flux. When the shield is turned around to the lower position (θ  =  180°), localized increase of the heat flux is clearly observed at r ~ 2 cm.

Zoom In Zoom Out Reset image size

In magnetron plasmas, the solid surface is irradiated with various particle- and photon-fluxes including high energy O⁻ ions. Total energy flux per unit solid surface and unit time J is composed of following heat fluxes.

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle J={{J}_{{\rm i}}}+{{J}_{{\rm e}}}+{{J}_{{\rm rec}}}+{{J}_{{\rm radt}}}+{{J}_{{\rm radp}}}+{{J}_{{\rm sp}}}+{{J}_{{\rm bs}}}+{{J}_{{{{\rm O}}^{-}}}}.\nonumber \end{align} \tag{ 3 }$$

Here, J_i, J_e, J_rec, J_radt, J_radp, J_sp, J_bs and ${{J}_{{{{\rm O}}^{-}}}}$ are fluxes of ion impingement, electron impingement, ion recombination on the surface, radiation from the target, radiation from the plasma, aggregation and chemical reaction of sputtered particles on the substrate, backscattered Ar atoms and high energy O⁻ ions, respectively. These heat fluxes are classified into two components according to the directionality of the heat flux, i.e. isotropic and anisotropic ones. J_i, J_e and J_rec are considered as isotropic components, because the plasma surrounds the TC even in the case of the with-shield condition (θ  =  0°). J_radt, J_radp, J_sp, J_bs and ${{J}_{{{{\rm O}}^{-}}}}$ are considered as anisotropic components, because all fluxes are originated from the target. Furthermore, the anisotropic component can be classified into two. In the cases of J_radt, J_radp, J_sp and J_bs, their emission angle with respect to the target surface normal is broad and the resultant radial profile at z  =  10 cm becomes rather smooth. ${{J}_{{{{\rm O}}^{-}}}}$, however, is different from J_radt, J_radp, J_sp and J_bs because ${{J}_{{{{\rm O}}^{-}}}}$ is the only anisotropic flux where all of the emitted O⁻ ions have strong directionality perpendicular to the target surface due to the accelerating E-field in front of the target surface. This behavior results in strong radial (r) dependence of the heat flux only in the case of the O⁻ heat flux even when the axial position is far from the target surface. Considering the radial profile of high-energy O⁻ (figure 3) and the present discussion, the peak of the heat flux at r ~ 2 cm is considered to be the heat flux originated from the high-energy O⁻ ions. In figure 6, however, the difference of the relative heat flux is observed not only at r  =  1.5–2.5 cm originated from O⁻ ions, but also at other radial positions. We consider that the difference is due to the suppression of the plasma by the shield plate even though the distance between the shield plate and the TC is 10 mm. Taking account for the influence of the reduced plasma density, subtraction of the relative heat flux with the shield from that without the shield is made by multiplying a factor to the relative heat flux with the shield before the subtraction so that the relative heat flux at r  ⩽  1 cm to be suppressed.

To calibrate the relative heat flux obtained by the TC, a stainless-steel substrate (0.5  ×  0.5 cm², 0.1 mm in thick) is placed on a substrate holder at r  =  2 cm and z  =  10 cm and its temperature variation after opening the shutter is measured by the radiation thermometer. To eliminate influence of the film deposition on the emissivity and the resultant temperature measurement, the radiation thermometer is installed inside a substrate holder and the substrate temperature is measured from the backside of the substrate through a BaF₂ window. A spacer is inserted between the substrate holder and the substrate to suppress thermal contact between them. Based on the temperature increase rate measurement, absolute heat flux is measured with the known substrate surface area and the heat capacity of the stainless-steel substrate. After calibration of relative heat flux (figure 6) and subtraction heat fluxes except for high-energy O⁻ ions, absolute heat flux density of high-energy O⁻ ions is obtained as shown in figure 7. From the result with an error observed in the experiment, a heat flux density of ~80  ±  20 W m⁻² at r  =  2 cm is obtained as the heat flux of high-energy O⁻ ions. In the figure, the spread of the heat flux along radial position is about 1 cm at FWHM. Considering plasma ring radius (2 cm), plasma ring width (1 cm), target erosion radius (2 cm), erosion ring width (~1 cm) and erosion depth (~0.2 mm) the spread of the O⁻ heat flux in radial position is due to the size of the plasma and slight influence of deflected high-energy O⁻ ion from eroded target [23].

Zoom In Zoom Out Reset image size

From the heat flux measurement of O⁻ ions, particle flux of high-energy O⁻ ions can be evaluated. In the course of the O⁻ impingement on the surface, however, not all their kinetic energies are converted into the heat and ratio of the heat flux to the incident O⁻ energy flux must be discussed. Total energy $\left\langle E \right\rangle$ that is not used for the surface heating per incident O⁻ ion is composed of (i) electron detachment energy from incident O⁻ ion: E_ED, (ii) secondary electron emission by O⁻ impingement: E_SE, (iii) average kinetic energy of backscattered O atom: $\left\langle {{E}_{{\rm BS}}} \right\rangle$, and (iv) energy for ITO re-sputtering on the surface by O⁻ ion impingement: $\left\langle {{E}_{{\rm RS}}} \right\rangle$, accordingly, $\left\langle E \right\rangle$ is expressed as

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle \left\langle E \right\rangle = ~ \left\langle {{E}_{{\rm BS}}} \right\rangle +\left\langle {{E}_{{\rm RS}}} \right\rangle +{{E}_{{\rm ED}}}+{{E}_{{\rm SE}}}.\nonumber \end{align} \tag{ 4 }$$

Next, each energy values are discussed as follows. Firstly, energy for (i) the electron detachment E_ED is obtained from the literature [24] to be 1.5 eV. Secondly, (ii) energy lost by the secondary electron $\left\langle {{E}_{{\rm SE}}} \right\rangle$ is expressed as γ_SEE_W, where γ_SE and E_W are secondary electron emission probability and work function of the ITO film, respectively. E_W is obtained from [25] to be 4.5 eV. However, γ_SE is not known for the ITO and γ_SE ~ 2 is adopted in the calculation, considering higher secondary electron emission for oxide surface [26]. Accordingly, E_SE ~ 9 eV is obtained.

Thirdly, (iii) average energy loss due to backscattered O atom per one incident O atom is expressed as follows:

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle \left\langle {{E}_{{\rm BS}}} \right\rangle = ~ {{\gamma }_{{\rm B}}}\mathop{\int }^{}{{E}_{{\rm B}}}f\left( {{E}_{{\rm B}}} \right){\rm d}{{E}_{{\rm B}}}.\nonumber \end{align} \tag{ 5 }$$

Here, E_B is kinetic energy of each backscattered atom, f(E_B) is energy distribution function of backscattered O atoms. In equation (5), backscattering yield γ_B is also taken into account. To evaluate energies lost by the backscattering $\left\langle {{E}_{{\rm BS}}} \right\rangle$, the TRIM (transfer of ion in matters) code simulation is used [27]. The TRIM is a simulation based on the Monte Carlo method and simulates the trajectory of incident high energy atoms in the solid. From the simulation, sputtering yield and backscattering yield are obtained with energies and directions of backscattered and sputtered atoms. In this study, the O atom is supposed as the incident species, and 10⁴ atoms are injected to a target of 36% In, 3% Sn and 61% O in the atomic ratio that corresponds to the ITO composition (In₂O₃/SnO₂  =  0.9/0.1). Then, trajectory of each incident atom in the ITO film is calculated and the number of backscattered atoms is counted with their backscattering energies. Figure 8 shows simulated backscattering probability γ_B and average backscattering energy ($\left\langle {{E}_{{\rm BSC}}} \right\rangle$  =  $\mathop{\int }^{}{{E}_{{\rm B}}}f\left( {{E}_{{\rm B}}} \right){\rm d}{{E}_{{\rm B}}}$) per backscattered O atoms as a function of the incident O energy. In figure 8(a), the γ_B is ~14% at incident energies above 100 eV. In figure 8(b), the $\left\langle {{E}_{{\rm BSC}}} \right\rangle$ is almost proportional to the incident O energy E_i and from the slope, $\left\langle {{E}_{{\rm BSC}}} \right\rangle$ is approximately expressed by $\left\langle {{E}_{{\rm BSC}}} \right\rangle$  =  0.25E_i. From these results, $\left\langle {{E}_{{\rm BS}}} \right\rangle$(=γ_B$\left\langle {{E}_{{\rm BSC}}} \right\rangle$) is calculated to be 8.6 eV at an O⁻ incident energy of 244 eV.

Zoom In Zoom Out Reset image size

Finally, (iv) energy lost by re-sputtering $\left\langle {{E}_{{\rm RS}}} \right\rangle$ is also calculated by the TRIM simulation. In the simulation, sputtered In, Sn and O atoms by the incident O atom is counted with their kinetic energies. Energy lost by the re-sputtering $\left\langle {{E}_{{\rm RS}}} \right\rangle$ is expressed as

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle \begin{array}{@{}llllllllllllllllllll@{}} & \left\langle {{E}_{{\rm RS}}} \right\rangle = ~ \underset{i}{\mathop \sum }\,{{\gamma }_{{\rm RSi}}}\left[ \mathop{\int }^{}{{E}_{{\rm RSi}}}f\left( {{E}_{{\rm RSi}}} \right){\rm d}{{E}_{{\rm RSi}}} \right] \nonumber \\ &\qquad\qquad +\mathop{\sum }^{}\left[ {{\gamma }_{{\rm RSIn}}}\left( {{R}_{{\rm I}{{{\rm n}}_{{\rm 2}}}{{{\rm O}}_{{\rm 3}}}}}{{E}_{{\rm DI}{{{\rm n}}_{{\rm 2}}}{{{\rm O}}_{{\rm 3}}}}} \right)+{{\gamma }_{{\rm RSSn}}}\left( {{R}_{{\rm Sn}{{{\rm O}}_{{\rm 2}}}}}{{E}_{{\rm DSn}{{{\rm O}}_{{\rm 2}}}}} \right) \right]. \end{array}\nonumber \end{align} \tag{ 6 }$$

Here, the subscript i denotes each atomic species (In, Sn, O). γ_RS is sputtering yield of In, Sn and O per one incident O atom, E_RS is kinetic energies of sputtered atoms, f(E_RS) is energy distribution function of sputtered atoms. ${{R}_{{\rm I}{{{\rm n}}_{{\rm 2}}}{{{\rm O}}_{{\rm 3}}}}}$ and ${{R}_{{\rm Sn}{{{\rm O}}_{{\rm 2}}}}}$ are composition ratio of In₂O₃ and SnO₂ in the ITO, respectively. In this study, composition ratio of In₂O₃/SnO₂ is 9/1 in weight for the sputter target, and the composition of the deposited film is supposed to be the same with that of the ITO target. ${{E}_{{\rm DI}{{{\rm n}}_{{\rm 2}}}{{{\rm O}}_{{\rm 3}}}}}$ and ${{E}_{{\rm DSn}{{{\rm O}}_{{\rm 2}}}}}$ are energies to dissociate In₂O₃ and SnO₂, and to produce one indium and tin atoms in the gas phase, respectively. From simulated result (figure 9), sputtering yields of In, Sn and O are 0.30, 0.03 and 0.50, respectively. ${{E}_{{\rm DI}{{{\rm n}}_{{\rm 2}}}{{{\rm O}}_{{\rm 3}}}}}$ and ${{E}_{{\rm DSn}{{{\rm O}}_{{\rm 2}}}}}$ are calculated to be 8.0 eV and 11.7 eV, using enthalpies of formation for In₂O₃ (922 kJ mol⁻¹) [28] and SnO₂ (577 kJ mol⁻¹) [29], heat of vaporization for In and Sn atoms and O₂ dissociation. From these values, energy required to release ITO atoms from the solid to the gas phase per one incident O atom is calculated to be 2.7 eV and considering kinetic energy of sputtered atoms (~5 eV) in average, $\left\langle {{E}_{{\rm RS}}} \right\rangle$ is calculated to be ~8 eV. From the above discussion with some ambiguity of the data used for the calculation, total energy that is not used for the surface heating is obtained from equation (4) to be ~27  ±  5 eV per one incident O⁻ ion at an incident energy of 244 eV. This means that almost 89  ±  2% of incident kinetic energy is used to heat the solid.

Zoom In Zoom Out Reset image size

From this result, particle flux of the incident high-energy O⁻ ions is finally obtained to be 2.3  ×  10¹⁸ ions m⁻² s⁻¹ with a random error of 25% and a systematic error of 2% at a position of z  =  10 cm and r  =  2 cm, an Ar pressure of 0.4 Pa and a discharge current of 0.3 A. As shown in figure 4, the O⁻ flux decays exponentially with the Ar pressure and the result suggests that the O⁻ flux decays also with respect to the axial position z. Using the parameter K obtained from figure 4, we can estimate the O⁻ flux at the target position using equation (1) and its value becomes 2.0  ×  10¹⁹ ions m⁻² s⁻¹. In the present study, FWHM of the erosion ring width on the ITO target is ~5 mm with a ring radius of 2 cm. From the erosion area, current path area on the target is estimated, and discharge current density is estimated from the current path area and the discharge current. As the resulting discharge current density at r  =  2 cm is roughly estimated to be 960 A m⁻². Secondary electron emission coefficient for the ITO with the Ar ion impact is not known and it is difficult to evaluate exact value of the incident Ar ion flux to the target, as well as the O⁻ production probability per incident Ar ion. If neglecting the secondary electron emission current from the evaluated discharge current density, the maximum value of O⁻ production probability per one incident Ar ion can be roughly evaluated to be 3.3  ×  10⁻³ at an Ar ion impact energy of 244 eV. Although actual O⁻ production probability is more than this value depending on the secondary electron emission coefficient, the O⁻ production yield on the target is much lower than that of O sputtering yields (~0.5) or In, Sn sputtering yields. However, it should be noted that the flux ratio of high-energy O⁻ to sputtered species varies depending on the axial position. As was mentioned, the O⁻ ions are strongly accelerated by the cathode fall and are directed to the deposition surface with sharp directionality. Considering rather broad angular distribution of sputtered atoms from the target, flux ratio of the high-energy O⁻ to the sputtered particles becomes much larger than the ratio of the production yield on the target with increasing the distance from the target.