To reduce the resistance-capacitance delay due to the reduction of the feature size and wiring spacing of the integrated circuit, the SiO₂ thin film was replaced with a low dielectric constant (low-k) thin film [1–3]. Silicon oxycarbide (SiOC) thin films, which are low-k thinfilm materials, have been widely used as passivation layers or interlayer dielectric films [4,5]. Plasma etching is essential for the etching of low-k thin films. However, the properties of low-k thin films degrade because of the changes in the chemical bonds caused by plasma exposure [6–8]. Therefore, it is important to understand the mechanism of thin-film damage and develop a process that reduces damage. Several studies have been conducted on the etching damage to SiOC thin films. Uchida et al. [9] studied a technology for evaluating plasma damage and showed that an increase in the dielectric constant is affected by ions, radicals, and ultraviolet (UV) irradiation. Kim et al. [10] evaluated the damage to a thin film by varying the density of cations using a narrow scan of the X-ray photoelectron spectroscopy (XPS) and X-ray diffraction patterns. Nonetheless, they did not determine the extent and effect of each ion, UV irradiation, or radical on the change in the dielectric constant. On the contrary, Kang et al. [11] reduced etching damage using a neutral beam etching technique. However, neutral-beam technology is not widely applied in industries. To analyze the characteristics of thin films exposed to plasma, it is necessary to separate the plasma species and evaluate the degree of damage caused by each plasma parameter. In our previous work, Ar and CF₄ gases were used to investigate the etching damage caused by plasma exposure on SiOC thin films [12]. A silicon and MgF₂ optical masks were used to isolate the effects of each plasma parameter. It was found that ions were mainly involved in the change in the dielectric constants of the ion intensity and UV irradiation. Nevertheless, the change in the dielectric constant that can be used to evaluate the etching damage is small; therefore, in this study, Fourier-transform infrared (FTIR) measurements were used to confirm the changes in the thin-film properties.

The high aspect ratio oxide etching process has been entirely performed by the capacitively coupled plasma reactive ion etching (RIE) process so far. Recently, we suggested that the etching process can be successfully performed by the inductively coupled plasma (ICP) RIE process using a low frequency (2 MHz) bias power system [13]. However, the etching damage to SiOC thin films has not yet been examined. Accordingly, in this study, the etching damage caused by the 13.56 MHz bias power was also investigated, and these results were compared with the etching damage caused by a 2 MHz bias power.

Plasma etching process was performed using an ICP etcher. The reactor consists of a cylindrical chamber made of anodized aluminum and a five-turn copper coil placed on a horizontal quartz window. The coil was connected to a 13.56 MHz power supply. A 13.56 or 2 MHz RF generator was applied to the bottom electrode to control the negative DC bias [14,15].

To minimize changes in the radicals formed in the plasma, CF₄ was fixed at 50 sccm, and the ratios of Ar and He were adjusted. To maintain a constant ion density, the Ar flow rate was fixed at 50 sccm and the ratios of CF₄ and C₄F₈ were adjusted.

As described previously [12], two types of masks were used: MgF₂ and Si. The support for the optical mask was printed with a 3D printer, and the length of the support was 1.7 mm. As the sidewalls of the mask support were open, some radicals could freely diffuse into the samples. Previous studies [12] also confirmed that the degree of influence of radicals and photons is proportional to the distance between the sample and optical mask.

The SiOC thin films were deposited using a magnetron sputtering system. A 900 × 129 mm² SiC target was used. A 400 nm thick SiOC film was deposited on a Si wafer by applying a power of 3 kW under 300 sccm Ar and 6 sccm O₂.

The dielectric constant was calculated as the square of the refractive index, and the refractive index of the SiOC thin film was measured using spectroscopic ellipsometry (M-2000V, J. A Woollam). The chemical bond changes on the surfaces of the thin films were evaluated using FT-IR spectroscopy (IFS66v/S and Hyperion 3000, Bruker).

To confirm the plasma properties, plasma diagnosis was performed using a dual Langmuir probe (DLP 2000; Plasmart Inc., Korea). Optical emission spectroscopy (OES) (Avaspec-3648, JinYoung Tech, Korea) and a UV detector (YK-35UV) were used to measure the light intensity generated from the plasma.

3.1. Noble gas change in the plasma

The flow of the CF₄ gas was kept constant, and the experiment was conducted by changing the flow rate ratio of Ar to He. Ar has lower ionization energy than He [16]. This indicates that when He is replaced with Ar, the density of ions in the plasma increases.

Figure 1 shows the SiOC etch rates and V_DC(self-bias voltage) at bias power supply frequency of 13.56 or 2 MHz. In the case of a 2 MHz bias power, the process was performed under process conditions in which the source power, bias power, and pressure were 150 W, 600 W, and 6 mTorr, respectively. Meanwhile, at 13.56 MHz, the source power, bias power, and pressure were 500 W, 100 W, and 10 mTorr, respectively. Figure 1(a) shows that in the 2 MHz system, much higher etch rates were observed compared to the 13.56 MHz system. As shown in Fig. 1(b), the difference in etching rate is largely due to the difference in V_DC [13].

Figure 1. (a) Etch rate and (b) V_DC at Ar fraction in Ar+He+CF₄ plasma.

Figure 2 shows the positive ion density at bias power frequency of 13.56 and 2 MHz. In the 13.56 MHz system, much higher ion densities were observed compared to the 2 MHz system. This phenomenon seems to be due to the higher source power in the 13.56 MHz system compared with that in the 2 MHz system. However, as the Ar fraction increased, the lower ionization energy of Ar (1520.571 kJ/mol) compared with He (2372.322 kJ/mol) increased the density of ions formed in the plasma [16].

Figure 2. Positive ion density at Ar fraction in Ar+He+CF₄ plasma.

Figure 3 shows the change in the dielectric constant ϵ with the Ar fraction. As expected, no change in the dielectric constant was observed for the Si mask sample. When the Ar ratio changed from 0 to 100 % at 2 MHz bias power, the ion density changed from 3.10⁹ to 3.59 × 10⁹ cm⁻³ by approximately 10.6 %. This is due to the low ionization energy of the Ar gas. In other words, the ion density increased as the Ar fraction increased. In this study, the intensity of UV also increased from 2.05 to 2.25 mW/cm² by approximately 9.7 %, showing that by changing the Ar fraction, not only the density of ions but also the UV intensity changed. Without a mask, that is, when samples were exposed to ions, UV irradiation, and radicals, the change in the dielectric constant increased from 2.937 to 2.943 as Ar fraction increased from 0 to 100 % at a bias power of 2 MHz. When using the MgF₂ mask, that is, for samples exposed to UV irradiation and radicals, the dielectric constant increased from 2.915 to 2.917. For the Si mask, as the ratio of the Ar gas varied, there was small change in the dielectric constant. At 13.56 MHz, the dielectric constant of the no mask samples changed by approximately 0.23 %, whereas it changed by approximately 0.2 % at 2 MHz. A slight change was identified; however, the difference was so small that it was difficult to clearly identify the change. However, at 13.56 MHz, the dielectric constant was larger. This can be interpreted as a result of the higher ion density in the case of 13.56 MHz, as shown in Fig. 2. This implied that the change in the dielectric constant value was less when using 2 MHz compared with that at 13.56 MHz. Although the etching rate of the oxide film at 2 MHz was higher than that at a 13.56 MHz bias frequency, the damage to the SiOC thin film at 2 MHz was lower than that at 13.56 MHz.

Figure 3. Dielectric constant according to Ar fraction in Ar+He+CF₄ plasma.

The change in the dielectric constant due to plasma exposure was very small. To further clarify the changes in SiOC damage owing to plasma exposure, FT-IR analysis was performed. To quantify the degree of etching damage, FT-IR spectra were extracted in the range of 800–1,250 cm⁻¹. Using the C–O (1,030 cm⁻¹) and Si–O bonds (1123 cm⁻¹) functional groups [4,14,17,18] from the FT-IR spectra, the area % of the Si-O and C-O bonds were calculated. Figure 4 shows the deconvolution of the FT-IR spectra into Si-O and C-O bonds.

Figure 4. Deconvolution of the FT-IR spectra into Si-O and C-O bond.

Figure 5(a) shows the ratio of the Si-O and C-O area % with the increasing Ar fraction in the plasma. When no mask was used, that is, the samples were exposed to ions, radicals, and UV irradiation at 2 MHz bias power, the change in ratio of the Si-O and C-O area % increased from 0.821 to 0.881 by approximately 7.3 %. In the case of the MgF₂ mask, that is, when the samples were exposed to radicals and UV irradiation, the change in ratio of Si-O and C-O area % increased from 0.603 to 0.626 by approximately 3.8 %. As shown in Fig. 3, the dielectric constant of the no-mask sample changed by approximately 0.2 %, whereas the FT-IR results changed by approximately 7.3 %. In the MgF₂-masked sample, the dielectric constant changed by less than 1 %, but the Si-O and C-O area % changed by approximately 3.8 %. These results indicate that FT-IR analysis is more effective than dielectric constant measurements for detecting damage to SiOC thin films caused by plasma exposure. Meanwhile, at 13.56 MHz, the Si-O / C-O ratio of the no-mask and MgF₂-masked samples changed by 8.6 and 5.0 %, respectively. The SiOC thin film was less damaged by plasma exposure at 2 MHz than that at 13.56 MHz, even though the etch rate of SiOC films at 2 MHz was much higher than that at 13.56 MHz. For the samples with no mask, the change in the Si-O/C-O ratio was greater than that of the MgF₂-masked sample because UV irradiation, radicals, and ions affected the change in the Si-O/C-O ratio. To separate the effects of UV irradiation and ion exposure on the dielectric constant changes, the ion and UV effects were recalculated. The UV effect was calculated using the difference between the Si-O/C-O ratios of the Si-masked and MgF₂-masked samples, and the ion effect was determined using the difference between the Si-O/C-O ratios of the MgF₂-masked and unmasked samples [12]. For instance, the impact of ion bombardment in Ar plasma can be evaluated using k_ion = k₀ + (k − k_UV), where k_UV is the Si-O/C-O ratio for the SiOC surface under the MgF₂ mask. Figure 5(b) shows the change in the Si-O/C-O ratio as a function of the UV intensity and ion density. The Si-O/C-O ratio by ions at 2 MHz system increased from 0.22 to 0.26 by approximately 18 % and that by UV increased from 0.17 to 0.19 by approximately 12 %. This is in agreement with previous results [12]. As pointed out in the previous research results, the effect of ions was greater than that of UV. In contrast, at 13.56 MHz, it was shown to increase by approximately 20 % by ions and approximately 17 % by UV. These results show that when exposed to plasma using a 2 MHz bias power, the change in the Si-O/C-O ratio is by no means larger compared to 13.56 MHz, as mentioned above.

Figure 5. (a) Area % of Si-O/C-O and (b) ion and UV effects according to optical mask and Ar fraction in Ar + He + CF₄ plasma.

3.2. Fluorocarbon gas change in plasma

Next, we investigated the effect of the polymer on the Si-O/C-O ratio. CF₄ and C₄F₈ gases, which are fluorocarbons, were used to confirm the effect of the polymer. While the total flow rates of CF₄ and C₄F₈ gases were fixed at 10 sccm, the amount of polymer was varied by changing the flow rate ratio of the CF₄ and C₄F₈ gases. The flow rate of Ar gas was fixed at 50 sccm.

When the CF₄ gas ratio changed from 0 to 100 % at 2 MHz bias power, the ion density increased from 5.61 × 10⁹ to 5.68 × 10⁹ cm⁻³ by approximately 1.2 % and UV intensity increased from 9.63 to 9.72 mW/cm² by approximately 0.93 %. This indicates that when the Ar flow rate was fixed, even when the CF₄/C₄F₈ ratio increased, the ion density and UV intensity in the plasma did not increase significantly. In contrast, a change in the CF₄/C₄F₈ ratio influenced the radical species formed in the plasma [19]. OES was performed to confirm the changes in these radicals. Figure 6 shows the extracted I_CFx (x = 1, 2, and 3) and I_F as a function of the CF₄ ratio in a 2 MHz bias power system using OES spectra. The F radical density (I_F), increased almost linearly, while CF_x species (I_CFx), significantly decreased. This implied that when CF₄ increased, fewer CF_x species were formed in the plasma, resulting in fewer formation of polymers by CF_x on the sample surface.

Figure 6. F and CF_x intensities and CF_x/F ratio as CF₄ fraction in CF₄+ C₄F₈+Ar plasma at 2 MHz.

As reported in a previous study, the energy applied to the sample surface is reduced by the polymer formed [20]. To confirm this effect, the thickness of the polymer formed on the sample surface was calculated through XPS analysis. Calculations were performed using the following equation [21]:

The thickness of the fluorocarbon layer (d_{Cx Fy}) was calculated from the integrated C 1s photoemission intensity $[I_{C\left(1s\right)}^d]$, using the intensity [I_C(1s)] of the thick passively deposited CxFy layer as the reference intensity [22]. Here, λ_C(1s) is the electron mean free paths of the C 1s photoelectrons. For our calculations, λ_C(1s) was set to 2.5 nm [23,24]. For samples without mask, when CF₄ ratio changed form 0 to 100 %, polymer thickness reduced from 1.32 to 0.69 nm by approximately 48.0 %, as shown in Fig. 7 or MgF₂-masked samples, the ratio reduced from 2.12 to 1.60 by approximately 24.5 %. For the Si-masked samples, the polymer was the thickest. This indicated that the CF_x species diffused freely through the space formed under the mask. In the Simask sample, the polymer thickness also decreased from 3.37 to 2.76 by approximately 18.0 %. This decrease in polymer thickness is related to the concentration of the CF_x species formed in the plasma, as described above. This is consistent with the decrease in the density of the CF_x formed in the plasma, as shown in Fig. 6. However, the polymer formed on the sample surface was removed by UV irradiation and ions, and the polymer thickness was reduced.

Figure 7. Polymer thickness according to CF₄ fraction in CF₄+ C₄F₈+Ar plasma.

Figure 8(a) shows the change in the Si–O/C–O ratio with the change in the CF₄/C₄F₈ ratio. When 2 MHz was used, the change in Si-O/CO ratio with changes in etching conditions was smaller compared to that of 13.56 MHz. Overall, the Si-O/C-O ratio of the samples exposed to 2 MHz plasma was lower than that of the samples exposed to 13.56 MHz plasma. Although a vertical etched profile could be obtained when the process was performed at 2 MHz, the damage to the SiOC thin film formed was not larger compared to that at 13.56 MHz, as mentioned above. To isolate the effect of the plasma parameters on etch damage, the method shown in Fig. 5 was applied, as shown in Fig. 8(b). At 2 MHz system, UV effect increased the Si-O/C-O ratio from 0.21 to 0.22 by approximately 5 % and ion effect increased it from 0.23 to 0.25 by approximately 9 %. Moreover, as the mixing ratio of CF₄ gas changed from 0 to 100 %, the UV intensity and ion density varied by almost 1 %, that is, almost no changes were observed. Based on this result, it could be confirmed that the increase in the Si-O/C-O ratio was relatively large compared to that when the ratio of Ar and He was changed. Although the change in the ion density and UV intensity decreased by approximately 1/10 in the CF₄ ratio change experiments compared to that in the Ar ratio; the change in the Si-O/C-O ratio was almost similar. These results cannot be explained by the changes in ion density and UV intensity. This indicates that the thickness of the polymer layer formed on the sample surface influences the Si-O/C-O ratio. As the ratio of CF₄ gas increased, the ion density and UV intensity slightly increased, and the thickness of the polymer formed on the sample surface decreased. Consequently, the energy of the ions and UV absorbed by the polymer decreased, and the Si-O/C-O ratio formed on the sample surface was relatively large. Furthermore, the Si-O/C-O ratio change in the ion effect at 2 MHz was greater than that at 13.56 MHz. This seems to be due to the thinner polymer formed at 2 MHz compared with that at 13.56 MHz. The thinner polymer formation appears to be related to the higher V_DC at a bias power of 2 MHz.

Figure 8. (a) Area % of Si-O/C-O and (b) ion and UV effects according to optical mask and CF₄ fraction in CF₄+ C₄F₈+Ar plasma.

The properties of SiOC thin films damaged by plasma exposure are investigated using a CF₄/C₄F₈/Ar/He mixed gas plasma in an ICP system with a bias power of 2 MHz. To evaluate the etching damage, FT-IR spectra are measured, and the ratio of the Si-O/C-O area % is calculated.

The change in the Si-O/C-O ratio extracted from the FT-IR analysis was more sensitive than the dielectric constant measurement. Although a vertical etching cross section could be obtained when the process was performed at 2 MHz, the damage to the formed SiOC thin film was not larger than that at 13.56 MHz. However, as the ratio of CF₄ to C₄F₈ gases changed, the density of ions and UV intensity changed slightly, whereas the CF_x polymer thickness changed considerably. As the proportion of CF₄ gas increased, polymer formation on the sample significantly influenced the Si-O/C-O ratio. This phenomenon occurred as the thick polymer absorbed the energy caused by the UV and ions.

This work was partly supported by the Korea Agency for Infrastructure Technology Advancement (KAIA) grant funded by the Ministry of Land, Infrastructure and Transport (Grant RS-2020-KA157018) and ‘Regional Innovation Strategy (RIS)’ through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (MOE)(2021RIS-004).

The authors declare no conflicts of interest.