Effect of UV Radiation on Oxidation for Ru CMP

All polishing experiments were carried out on an E460E polisher from Alpsitec Inc., equipped with a politex pad (purchased from the Dow chemical company). The polishing slurry under investigation was composed of colloidal silica (particle size between 60 nm to 70 nm, pH = 9.5–10.0), Hydrogen peroxide ((H₂O₂, the mass fraction of 30 wt%, semiconductor grade) and potassium persulfate (K₂S₂O₈, 99.5% purity). Finally, the pH of the slurry was adjusted to 8.0 by using diluted KOH or HNO₃. A 200 nm Ru film was deposited on a silicon substrate with a diameter of 12 inches by plasma enhanced atomic layer deposition, and a 76.2 mm diameter Ru samples was cut out for the CMP experiment. An ultraviolet lamp (UV, purchased from Shanghai Yichen) is used for CMP experiment and subsequent experiments. A Schematic diagram of CMP process is shown in Fig. 1, and the experimental process parameters are shown in Table II. In order to reduce the influence of the irradiation position, the UV lamp is covered with a hood during the polishing process, so that the UV lamp can only irradiate to the upstream of the polishing head. All the CMP experiments were carried out at room temperature (25 ± 1 °C). The removal rates of Ru were determined by measuring the difference in film thickness before and after polishing, using four-point probe (4D Model 333 A) test the mean value of 81 different points along any diameter. To ensure the accuracy of the data, each experiment was repeated three times and the average value was recorded.

Zoom In Zoom Out Reset image size

All electrochemical experiments were conducted in the CHI660E electrochemical workstation produced by Shanghai Chenhua. Electrochemical workstation with a three-electrode glass cell, (100 ml volume, Pt counter electrode and saturated calomel electrode). a Ru coupon (10 × 20 × 2 mm, 99.99% purity) was used as a working electrode. Before experiment, the working electrode was abraded with sandpaper, then dipped into 1 wt% citric acid solution to ensure the removal of naturally oxidized products.

Open circuit potential (OCP) was continuously measured at 600 s and a dynamic potential curves (Tafel curve) were obtained by the potentiodynamic polarization curve test. The tests of electrochemical impedance spectroscopy (EIS) were performed at premeasured OCP by superimposing a sinusoidal AC potential perturbation of 5 mV in the frequency range of 0.01 Hz to 10 kHz. The EIS data was applied to complex nonlinear least square (CNLS) analyses by the ZSimpWin software to fit the electrode equivalent circuit (EEC) of the reaction interface between the Ru electrode and the slurry.

The surface characteristics and chemical modifications of Ru films were analyzed by X-ray photoelectron spectroscopy (PHI 250Xi, ESCA System). The Ru square samples (10 × 10 mm, cut from Ru blanket wafer) were immersed in a 500 ml solution (without abrasives) for 10 min and then tested. Finally, the surface chemicals of the Ru square sample were analyzed using CASA XPS software.

Surface roughness and surface quality of Ru films treated with various solutions (without abrasives) for 6 min were measured using Atomic force microscopy (AFM, Agilent 5600LS) and Scanning electron microscopy (SEM, ZEISS SIGMA 500/VP). Three 10 × 10 μm positions were randomly measured on the surface of the Ru square sample (10 × 10 mm, cut from Ru blanket wafer). All experiments were repeated at least three times for each sample and the average value was reported.

In the presence of potassium ions, the removal rate of Ru could be increased, which is due to the reduction of the electrostatic repulsion on the surface of silica and Ru. ¹⁷ Figure 2 shows the effect of different potassium salts on the removal rate of Ru under the conditions of 0.15 wt% H₂O₂ and 5 wt% SiO₂ at pH = 8.0. From Fig. 2, we can see that under the same potassium ion concentration conditions, the removal rate of Ru in the slurry containing K₂S₂O₈ is almost the same as that of the polishing slurry containing potassium sulfate (K₂SO₄) or potassium nitrate (KNO₃). This may be due to partly K₂S₂O₈ and H₂O₂ react with each other to cause the self-consumption of the oxidant and reduce the oxidation capacity. ²⁰ The reaction should be as follows.

$$\begin{eqnarray}&&4{H}_{2}{O}_{2}+2{S}_{2}{{O}_{8}}^{2-}\to 4S{O}_{{\rm{4}}}^{{\rm{2}}-}+2{H}_{2}O+3{O}_{2}+4{H}^{+}\end{eqnarray} \tag{ 1 }$$

Zoom In Zoom Out Reset image size

Figure 3 shows the effects of UV and K₂S₂O₈ concentration on Ru removal rate under the condition of 0.15 wt% H₂O₂ and 5 wt% SiO₂ at pH = 8.0. In the absence of UV, the polishing removal rate of Ru increases with the increase of K₂S₂O₈ concentration and reaches the saturation maximum (about 620 Å min⁻¹) at 40 mM K₂S₂O₈. In the presence of UV, the Ru removal rate is increasing from 132 Å min⁻¹ to 786 Å min⁻¹ with the increase of K₂S₂O₈ concentration from 0 mM to 80 mM. In contrast to the absence of UV, when the K₂S₂O₈ concentration is less than 60 mM, the addition of UV will inhibit the Ru removal rate, but exhibit a promoting effect at 80 mM K₂S₂O₈, indicating that UV can promote the activation of K₂S₂O₈ and accelerate the formation of the passivation layer on the Ru surface. From Fig. 4, when only potassium persulfate is contained, there are fewer surface oxides and have the ability of UV to increase Ru removal rate (from 85 Å min⁻¹ to 102 Å min⁻¹). Moreover, when the H₂O₂ concentration is greater than 0.15 wt%, the removal rate of Ru will decrease, and the removal rate of Ru will further decrease under UV irradiation. The results show that when the concentration of H₂O₂ is higher than 0.15 wt%, UV has the obvious ability to inhibit the removal rate of Ru.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figure 5 shows the OCP and Tafel plots of Ru in the solutions which contain various concentrations of K₂S₂O₈ and with and without UV. From Fig. 5a, it can be seen that the OCP increases with the concentration of K₂S₂O₈ in the solution increases. Moreover, without UV, the OCP of 0 mM K₂S₂O₈, 40 mM K₂S₂O₈ and 80 mM K₂S₂O₈ were 0.119 V, 0.168 V and 0.460 V, and they increased to 0.143 V, 0.498 V and 0.508 V, respectively, with UV. This means that the addition of UV can make the solution more likely to generate RuO₂/RuO₃ oxide layer on the Ru surface. There is an OCP difference of 0.04 V between Only 80 mM K₂S₂O₈ with and without UV, indicating that UV can enhance the activity of the Ru surface. This coincides with the polishing experiment. Similarly, it can be seen from Fig. 5b that the corrosion voltage (E_corr) also shows a rising trend with the increase of K₂S₂O₈ concentration, and the presence of UV will also increase the corrosion voltage. It is worth noting that under the condition of 40 mM K₂S₂O₈, the E_corr with UV is 0.33 V higher than that without UV, implying the reactions occurring on Ru surface were strengthened, but their polishing rate is inhibited, which will be explained in the following experiment. The reaction equations (Eq) that may occur during the surface corrosion and CMP of Ru are shown in the below following.

$$\begin{eqnarray}&&{H}_{2}{O}_{2}+2{e}^{-}=2O{H}^{-}\end{eqnarray} \tag{ 2 }$$

$$\begin{eqnarray}&&R{\rm{u}}+4O{H}^{-}=R{\rm{u}}{O}_{2}+2{H}_{2}O+4{e}^{-}\end{eqnarray} \tag{ 3 }$$

Zoom In Zoom Out Reset image size

After UV, the corrosion current of only-K₂S₂O₈ (without H₂O₂) increased by 2.955 μA cm⁻², indicating that K₂S₂O₈ was sensitive to UV and produced substances with stronger oxidation ability, as shown in Table III. The reasonable explanation are as follows: persulfate was more stable at normal temperature (25 °C). However, once it is affected by some external conditions such as transition metal ions, light, heat, etc., it will break the oxygen-oxygen single bond (–O–O–), and produce a more oxidizing sulfate radical (SO₄ ^·−, E₀ = 2.5–3.1 V vs NHE). ^21,22 Therefore, persulfate (S₂O₈ ²⁻) could be photochemically (UV used in this paper) activated, this is the Fenton-like method, and the reaction equation should be as follows.

$$\begin{eqnarray}&&{S}_{2}{{O}_{8}}^{2-}\mathop{\longrightarrow }\limits^{UV}S{{O}_{4}}^{2-}+S{{O}_{4}}^{\cdot -}\end{eqnarray} \tag{ 4 }$$

The formation of strong oxidizing SO₄ ^·− promotes the formation of Ru oxide film, which explains the increase of Ru OCP and E_corr in the UV-containing solution.

To further understand the Ru surface oxidation mode and the passivation film structures, the Nyquist and Bode plots were used for different k₂s₂o₈ concentrations slurries in 0.15 wt% hydrogen peroxide. Figure 6 shows that the Nyquist impedance plots and corresponding electrode equivalent circuit (EEC). The equivalent circuit fitting is divided into the following parts. In the ultra-high frequency region, the electrolyte transfer can be represented by a slurry resistance (R_s). In the high frequency region, the electrolyte/metal oxide interface to form a semi-circle capacitive loop. This process can be represented by an passivation film resistance (R_pf) and passivation film capacitance (CPE_pf). In the intermediate frequency region, a semi-circle capacitive loop related to the charge transfer process, which can be represented by charge transfer resistance (R_ct) and double-layer capacitance (CPE_dl). In the low-frequency region, an approximately 45-degree line related to the solid diffusion process of ions inside the Ru passivation layer. This process can be represented by Warburg impedance (W) describing diffusion. Therefore, the equivalent circuit should consist of two time constant and a Warburg impedance (W), as shown in Fig. 7, and the EIS fitted parameters is shown in Table II by the Zsimpwim analyzed software. Here, due to the non-uniformity of the surface, the ideal capacitor is replaced by a constant-phase element (CPE). ^5,23

$$\begin{eqnarray}&&{Z}_{CPE}={\left[\left.Q{\left(j\omega \right)}^{n}\right)\right]}^{-1}\end{eqnarray} \tag{ 5 }$$

where Q is a constant, ω is the angular frequency, j is the imaginary part, the value of n represents the extent of frequency dispersion in the capacitive component of double layer charging (n = 1 for no dispersion). As can be seen from Table IV, under the condition of not containing UV, the R_s, the R_pf and the R_ct decreases with the increasing of K₂S₂O₈ concentration, the Warburg impedance is almost negligible. So, the Ru surface is mainly affected by charge transfer at this time, the internal charge transfer speed of Ru increases, the chemical reaction is enhanced. This is consistent with polishing experiments.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Under the condition of containing UV, compared with UV-free conditions, R_s and R_pf are all slightly decreased, it may be that the Ru electrode is partially dissolved in the solution, which reduces the resistivity of the solution. Simultaneously, the R_ct of 40 mM K₂S₂O₈ decreased from 515.4 Ω to 447.2 Ω, and the W increased from 10.31 × 10⁻⁴ Ω⁻¹·cm⁻²·s^0.5 to 495 × 10⁻⁴ Ω⁻¹·cm⁻²·s^0.5 (the increase is about 48 times); The R_ct of 80 mM K₂S₂O₈ decreased from 886 Ω to 715.4 Ω, and the W increased from 53.56 × 10⁻⁴ Ω⁻¹·cm⁻²·s^0.5 to 386 × 10⁻⁴ Ω⁻¹·cm⁻²·s^0.5 (the increase is about 7 times). The decrease of R_ct can prove that the charge transfer can be accelerated in the presence of UV, and the chemical reaction is enhanced, while the increase of W indicates that the reaction is more affected by diffusion-control. Under the condition of 40 mM K₂S₂O₈, W increases by 48 times, while 80 mM K₂S₂O₈ only increases by 7 times, indicating that the passivation layer under 40 mM K₂S₂O₈ condition is denser. Therefore, it can be explained that the removal rate decreased under the condition of 40 mM K₂S₂O₈ and the removal rate increased under the condition of 80 mM K₂S₂O₈ in the polishing experiment.

Bode diagrams can also be applied to study the corrosion behaviors of slurries. ^24,25 Bode Amplitude graph was given in Fig. 7a. Impedance modulus (IM) changes with frequency, and different frequencies represent different interface responses. It can be seen from Fig. 7a, the impedance decreased with the increase of K₂S₂O₈ concentration regardless of any frequency, this indicated that K₂S₂O₈ can effectively reduce the impedance of Ru, which accelerated the corrosion of Ru. Similarly, under the same K₂S₂O₈ concentration, the impedance with UV is smaller than the impedance without UV, indicating that UV can effectively decrease the impedance and increase the corrosion voltage of Ru. Bode-phase graphs was given in Fig. 7b. Two obvious capacitance and resistance loops can be observed, the high-frequency peak shifted to a high phase angle, indicating the increase in impedance.

The XPS technique can be used to analyze the surface chemical changes, ²⁶ when the Ru material is immersed in different solutions. The Fig. 8 shows The XPS Ru (3d, hybridized orbital) spectra of Ru wafers after being immersed in the slurries containing (a) 40 mM K₂S₂O₈ without UV; (b) 40 mM K₂S₂O₈ with UV; (c) 80 mM K₂S₂O₈ without UV; (d) 80 mM K₂S₂O₈ with UV. The binding energy of metal Ru is 280.0 eV and 284.2 eV at 3d5/2 and 3d3/2, respectively. ^27,28 It was found that the binding energy of RuO₂ was 280.6 eV ²⁹ and 285.0 eV. Since the RuO₂·2H₂O conductive tape is only partially filled, it is likely that nuclear-hole coupling will occur on the surface. Therefore, RuO₃ is considered to be a phase of RuO₂·2H₂O with a characteristic peak of 282.1 eV. ³⁰ By further calculation of Ru/RuO_x (x=2,3) content, shown in Table V. The results showed that in the absence of UV, the concentration of K₂S₂O₈ increased from 40 mM to 80 mM, and the oxide RuO_x (x=2,3) on the surface of Ru increased from 42.08% (calculated according to each peak area) to 63.13%, indicating that the addition of K₂S₂O₈ can increase the oxide content on the Ru surface. When UV is present, the content of RuO₂ at 40 mM K₂S₂O₈ and 80 mM K₂S₂O₈ increases by 12.72% and 8.61%, respectively. However, RuO₃ hardly changed under 40 mM K₂S₂O₈ conditions, and reduced by 5.4% under 80 mM K₂S₂O₈ conditions. This shows that RuO₃ hardly reacts under the condition of 40 mM K₂S₂O₈, and UV only promotes the formation of RuO₂. The possible reason for this phenomenon is that on the one hand, K₂S₂O₈ and H₂O₂ react with each other to cause the self-consumption of the oxidant and reduce the oxidation capacity. On the other hand, UV activated K₂S₂O₈ causes the deposition of Ru dioxide on the Ru surface, which consumes most of the K₂S₂O₈, it cannot continue to react with RuO₃. Combined with the analysis of the polishing data of 40 mM K₂S₂O₈ in Fig. 3, we reasonably speculate that the Ru surface oxide grows uniformly, which makes the surface RuO₂/RuO₃ oxide layer denser, the impedance becomes higher, and the removal rate decreases. The reactions should be as follows.

$$\begin{eqnarray}&&Ru+{S}_{2}{{O}_{8}}^{{\rm{2}}-}+6O{H}^{-}\mathop{\longrightarrow }\limits^{UV}2Ru{O}_{2}+2S{O}_{{\rm{4}}}^{{\rm{2}}-}+2{H}_{2}O\end{eqnarray} \tag{ 6 }$$

Under the condition of 80 mM K₂S₂O₈, RuO₃ will participate in the reaction process and dissolve in the solution, making the surface oxide layer looser and more porous, so the removal rate of Ru is significantly improved under mechanical action. This is in good agreement with the polishing data of 80 mM K₂S₂O₈ in Fig. 3. The possible reactions process should be the following Fenton-like reaction. ^31–33

$$\begin{eqnarray}&&Ru{O}_{3}+{S}_{2}{{O}_{8}}^{2-}+{\rm{2}}O{H}^{-}\mathop{\longrightarrow }\limits^{UV}Ru{{O}_{{\rm{4}}}}^{-}+S{O}_{4}^{\cdot -}+S{O}_{4}^{2-}+{H}_{2}O\end{eqnarray} \tag{ 7 }$$

Zoom In Zoom Out Reset image size

In order to show the effect of mechanical action more clearly, we characterize the effect of UV under different abrasive particle size conditions, as shown in Fig. 9. The results show that when the particle size increases from 40 nm to 80 nm, under the action of ultraviolet rays, the increase in Ru removal rate increases from 20.59% to 26.36%, indicating that the increase in mechanical action can significantly increase the removal rate. Therefore, at 40 mM K₂S₂O₈, the addition of UV made the Ru removal rate exhibits a suppressed phenomenon due to the rapid generation of a large amount of robust, dense RuO₂/RuO₃ oxide layer on the surface. The promotion phenomenon under 80 mM K₂S₂O₈ with UV conditions is due to the excessive K₂S₂O₈ that makes RuO₃ also participate in the reaction, which makes the Ru surface loose and porous, so it is easier to remove under mechanical action.

Zoom In Zoom Out Reset image size

Figure 10 represents SEM images of the Ru surface after immersion in each condition's solution at pH = 8.0 for 10 min, it can be seen that the surface corrosion is weak in the absence of UV. After adding UV, loose surface of oxidation corrosion could be clearly seen on the surface of Ru, and the corrosion degree of 80 mM K₂S₂O₈ is stronger than 40 mM K₂S₂O₈. Figure 11 represents AFM images of the Ru surface after immersion in each condition under 0.15 wt% H₂O₂ at pH = 8.0 for 10 min It can be seen from Fig. 11 that the surface roughness (Rq) of Ru decreases from 0.418 nm to 0.340 nm when the 40 mM K₂S₂O₈ solution is irradiated with UV, while the surface roughness of Ru increases from 0.421 nm to 0.533 nm when the 80 mM K₂S₂O₈ solution is irradiated with UV. This is consistent with the previous analysis. A small amount of K₂S₂O₈ makes a dense RuO₂/RuO₃ oxide film on the Ru surface. However, in the excessive K₂S₂O₈, the Ru surface becomes loose and porous due to the consumption of RuO₃. Therefore, the corrosion diagram of Ru surface should be as shown in Fig. 12.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size