As semiconductor devices become integrated and their scale becomes increasingly smaller, even minor changes in the plasma processing chamber can induce significant process shifts in process-toprocess or tool-to-tool matching [1–4]. Therefore, interest in the maintenance of plasma processing chambers has been increasing [1–4]. In particular, in plasma etching or deposition processes using fluorocarbon (C_xF_y) based gas discharges, chamber wall contamination, which can induce severe process deviation, often occurs by the formation of an unwanted thick fluorocarbon polymer film on the wall [5,6]. It is therefore important to investigate the effect of the wall-coated polymer film on the plasma properties such as the formation of heavy polymer particles near the chamber wall via plasma-wall interactions [7]. These polymer particles are released back into the plasma, resulting in a change in the plasma properties; this in turn can affect the characteristics of the plasma such as radical distribution and/or process quality on the substrate [8–10]. Many studies therefore have been conducted to understand the impacts of wall contamination on the plasma properties [11–13]. However, there is still insufficient information to comprehend the correlation quantitively between the wall-deposited film and plasma processing according to long-term operation.

In this study, we conducted a long-term C₄F₈ plasma deposition process to elucidate the variation of plasma characteristics depending on the level of chamber wall contamination. The level of innerwall contamination was determined by measuring the thickness of the polymer film deposited on the wall using a scanning electron microscope (SEM). Furthermore, the chemical properties of plasma were studied by estimating the density of radicals such as C₂, CF₂, and F through optical emission and absorption spectroscopy.

Figure 1 presents a schematic illustration of the experimental setup. The experiment was performed in a radio frequency inductively coupled plasma (RF-ICP) chamber equipped with a 2-turn planar-shaped antenna connected to a 13.56 MHz RF generator (Youngsin RF, YSR-10AF) with a matching network. Octafluorocyclobutane (C₄F₈) gas [50 standard cubic centimeter (sccm)] and argon (Ar) gas (30 sccm) were uniformly introduced to the processing chamber by mass flow controllers (Linetech, M3030V). The plasma was generated with 350 W input power at 2.67 Pa sustained by a rotary pump (Woosung Vacuum, MVP24) and a turbo molecular pump (Edwards EXT 300D). For the long-term deposition process, the total operation time of the plasma process was 12 h and the deposition properties were analyzed every 2 h. Stainless steel (stainless steel sheet, Nilaco Corporation) samples were attached to the inner wall and p+ type silicon wafer (Si wafer, Prulong) samples were placed on the bottom substrate, as seen in Fig. 1. The thickness of the deposited thin films was measured using SEM (Thermo Fisher Scientific, Apreo 2 S). Furthermore, several samples were attached on the wall under a clean-chamber condition. In this study, two wall-deposited films with different thicknesses are referred to as ‘accumulated thickness’ and ‘summed thickness.’ Samples were detached from the pre-attached specimens on the wall every 2 h for thickness measurement, and a clean sample was reattached to continue the deposition process alongside the existing samples. Subsequently, samples deposited for 2 and 4 h were removed for thickness measurement. This process was repeated for a total of 12 h. Finally, after the 12 h process, the samples deposited for 2 and 12 h were removed from the wall for thickness measurement. Each thickness, referred to as the process thickness, deposited on the clean samples every 2 h was summed to calculate the total ‘summed thickness.’ On the other hand, the thickness of the sample deposited continuously every 2 h represented the total ‘accumulated thickness’ over that period.

Figure 1. A schematic diagram of a 13.56 MHz RF-ICP chamber with plasma diagnostics tools and a cross-sectional SEM image of the wall-deposited film.

The electron temperature and ion density were also measured with a floating-type harmonic probe (P&A Solutions, Wise probe system). Furthermore, optical emission and absorption spectroscopy were carried out using a spectrometer (Princeton Instruments, IsoPlane SCT 320) equipped with a ultraviolet-sensitive grating (1,800 grooves/mm) and a charge coupled device (Princeton Instruments, PIXIS400B). The F density was calculated via the actinometry method based on the emission intensity ratio [14,15]. On the other hand, the CF₂ and C₂ densities were measured using absorption spectroscopy with a Xe-arc lamp (Hamamatsu, L2273) as the light source. The densities were estimated by measuring the fractional transmittance of the light source with and without plasma, as described by following Eq. (1) below [16].

where I0 is the light source (Xe-arc lamp) emission intensity considering the background intensity B(λ) without plasma emission, IT is the transmission of the light source with plasma emission, Lp(λ) is the measured spectrum with both plasma and the Xe-arc lamp on, P(λ) is the emission intensity of the plasma, and L0(λ) is the Xe-arc lamp emission intensity. From the experimentally obtained absorbance, A(λ), the radical density (n) can be determined with the given absorption cross-section (σ) and absorption path length (l). For example, Figs. 2(a) and 2(b) depict the absorbances of CF₂ and C₂ radicals, respectively. In this study, the absorption cross-section of CF₂ radicals was obtained by referring to the results of Sharpe et al. [17]. However, the absorption cross-section of the C₂ radicals was calculated from the summation of all the rotational line oscillator strengths [18] based on Eqs. (2) and (3).

Figure 2. Absorbances of the (a) CF₂ and (b) C₂ molecules.

where W_ν is the equivalent width of the rotational absorption line, A(λ) is the absorbance of the C₂ molecule according to the wavelength, re is the radius of the electron, c is the speed of the light, f_J′J″ is the rotational line oscillator strength, N_J′ is the rotational population density, and l is the optical absorption path length [19]. The equivalent width of each rotational absorption line was summed over the known rotational lines to consider the all-rotational population density of the C₂ molecule, as given by Eq. (3) below:

In this work, the plasma operational parameters were fixed at 350 W of RF input power and 2.67 Pa with 50 sccm of C₄F₈ and 30 sccm of Ar. Thus, the fundamental parameters of the plasma such as the electron temperature and density remained unchanged during longterm plasma operation, as shown in Fig. 3. This implied that the physical properties of the plasma did not vary significantly with operation time. However, as depicted in Fig. 4, the emission spectra of the C₄F₈/Ar plasma exhibited varying intensities depending on operation time. In all cases, the molecular emission spectra of CF (B-X and A-X), CF₂ (A-X), C₃ (A-X), and C₂ (d-a) transitions, along with the atomic emission spectra such as F (3p-3s) and Ar (2p-1s) transition, were observed and their intensities varied with operation time, as seen in Fig. 4 [10,20–23]. Therefore, considering the almost constant electron temperature and ion density during the process, the density of the active species in the plasma could be changed according to the plasma operation time.

Figure 3. Electron temperature (-○-) and ion density (-□-) of the C₄F₈/Ar plasma with respect to process operation time.

Figure 4. Optical emission spectra of the C₄F₈/Ar plasmas after 0, 6, and 12 h of operation.

Figures 5(a) and 5(b) show the relative and absolute densities of active species, respectively. The relative density was calculated from the intensity ratio (I_x/I_Ar) based on the actinometry method [14,15], while the absolute density was calculated via absorption spectroscopy, as described in Section 2. As shown in Fig. 5(a), the relative density shows a quite similar trend to that of absolute density according to the operation time. In C_xF_y-based gas plasmas, it has been well reported that CF₂ and C₂ contribute to the wall-deposited fluorocarbon polymer, while F contributes to the plasma-wall interaction, resulting in a decrease of the wall-deposited fluorocarbon polymer film thickness [7,24–26]. Namely, the CF₂, C₂, and F can simultaneously contribute to the fluorocarbon polymer deposited on the inner chamber wall. Figure 5(b) displays the absolute density of the F, CF₂, and C₂ measured by the absorption spectroscopic method. The densities of C₂ and F slightly increased but the CF₂ density decreased by almost 20 % according to the processing time. The different time variation of radical densities might result from the change of plasma properties depending on the wall conditions [12]. Since the physical properties of plasma such as electron density and temperature remained almost constant throughout the long-term process (Fig. 3), changes in plasma-active species densities should be attributed to other factors. The analyses revealed that the only variation in wall conditions over time was the thickness of the film deposited on the walls, as shown in Fig. 6. Notably, Fig. 6(a) shows a discrepancy between the accumulated thickness and the summed thickness, despite that the average film thickness deposited every 2 h on both the walls and the bottom substrate (referred to as process thickness) was nearly the same as the thickness (1.960 ± 0.157 μm) on the walls and that (1.850 ± 0.177 μm) on the substrate. Because of the constant deposition rate per processing time, the accumulated thickness should be the same as the summed thickness when there are no other factors such as wall temperature variation [27], and both total thicknesses gradually and continuously would increase [27]. Thus, the summed thickness (blue bar) increased almost linearly for 12 h, with a constant deposition rate (D.R.) of 16.30 ± 0.13 nm/min during each 2 h interval [Fig. 6(b)]. In contrast, the accumulated thickness increased steadily until 8 h, after which it plateaued. This indicates that the deposition rate for the accumulated thickness case was not constant and decreased from 16 nm/min to 0 nm/min over time, as shown in Fig. 6(b). Since the accumulated thickness represents the actual thickness of the film continuously deposited on the wall during the 12 h plasma process, the difference between the actual and theoretical (summed) thicknesses highlights the influence of plasma-wall interactions under the contaminated-wall condition. Figure 6(c) displays the thickness difference (Δ Thickness), between the accumulated thickness and the summed thickness, suggesting the occurrence of simultaneous deposition and etching or suppression of deposition through the plasma-wall interaction. Therefore, as shown in Fig. 5(b), the density of the CF₂ continuously consumed until 8 h, but the loss of CF₂ was subsequently depressed due to the compensation of loss of CF₂ by the production of CF₂ via plasma-wall interaction [10,28]. Additionally, the density of C₂ radicals, one of the main constituents of the deposited film, was almost constant but increased after 8 h due to suppression of deposition on the wall.

Figure 5. (a) Relative densities of C₂ (-●-), C₃ (-△-), F (-■-), CF (-□-), and CF₂ (-○-), and (b) absolute densities of the CF₂ (-○-), C₂ (-●-), and F (-■-) over the process operation time.

Figure 6. (a) The accumulated thickness (red bar) and summed thickness (blue bar) of the polymer films deposited on the chamber wall, and process thickness (patterned bar) deposited on the bottom substrate. (b) The deposition rates for the accumulated thickness (-○-) and summed thickness (-□-). (c) The thickness difference between the accumulated and summed thickness.

The impact of chamber wall contamination on plasma variation was investigated during a long-term C₄F₈/Ar plasma deposition process. To assess wall-contamination levels, both the accumulated thickness and summed thickness of the deposited film on the wall were evaluated. During the 12 h operation, the total film thickness (i.e., summed thickness) on the wall (i.e., the summation of the deposited thickness measured at each 2 h interval) increased to 11.75 μm with a constant deposition rate of 16.30 ± 0.13 nm/min. However, the actual film thickness (i.e., accumulated thickness) continuously deposited on the wall steadily increased until 8 h, after which it saturated at around 5.78 μm, indicating the deposition process on the wall was suppressed. For example, at the end of the 12 h-process, the suppression rate of deposition on the wall was almost 56 % compared to the summed thickness. Interestingly, during the plasma process, while the electron temperature and ion density were not significantly changed, the density of chemical species was closely aligned with the wall-contamination levels. This suggests the interaction between the plasma radicals and the contaminated-wall played a significant role, influencing the production-loss balance of CF₂ and/or suppressing deposition through the polymer formation on the wall.

This work was supported by a Korea Basic Science Institute (National Research Facilities and Equipment Center) grant funded by the Ministry of Education (2021R1A6C101B383), a National Research Council of Science & Technology (NST) grant funded by the Korean government (MSIT) (CRC20014-100), and by SEMES.

The authors declare no conflict of interest.