Hexamethyldisilazane (HMDS) is commonly used for surface modification and manufacturing of low-dielectric-constant (κ) materials in various fields, such as semiconductors [1–3]. HMDS has nonpolar methyl groups that contribute to the formation of porous thin films with unique properties, such as high hydrophobicity and low permittivity [1,4]. However, synthesizing films from HMDS is challenging because it is a stable compound requiring high activation energy to dissociate and form beneficial radicals such as CH_x and Si-CH_x that are attributed to thin film deposition [5].

Plasma-enhanced chemical vapor deposition (PECVD) is widely used to decompose stable compounds and generate fluent reactive species. The HMDS in plasma is easily decomposed through various pathways, such as electron impact and penning ionization [6]. Thus, PECVD is an efficient and effective alternative to conventional wet chemical processes for thin-film deposition [2,7]. PECVD offers control over the reactive species in the plasma and, consequently, over the deposited film properties by tuning the electron temperature, i.e., adjusting operational parameters such as the input power and plasma gas composition [1,8,9]. Kim et al. [9] reported that increasing the HMDS weight percentage in Ar/HMDS/O₂ plasma increased both the impurities and surface roughness of the deposited films. In a previous study [1], we discovered that the amount of skeleton structure molecules, such as Si- CH₃, in thin films could be tuned by controlling the flow rate of N₂ gas in atmospheric pressure Ar/HMDS/N₂ plasma. However, further investigation of plasma characterization is required to fully understand the influence of additive gas effects on the plasma process.

Owing to the distinct physical properties of N₂, such as its high vibrational temperature and many metastable states, it significantly impacts its plasma characteristics and surface reactions [10–12]. Therefore, it is widely utilized in various processes to control plasma, leading to beneficial changes in processes such as deposition, etching, and gas conversion [11,13]. This indicates that understanding the influence of N₂ on plasma is crucial for understanding plasma characteristics and optimizing plasma processes.

Hence, in this study, we investigated the correlation between the plasma thermal characteristics and reactive species to understand the influence of N₂ on atmospheric-pressure Ar/HMDS plasma. Accordingly, reactive species in the plasma were also investigated using optical emission spectroscopy and various temperatures of plasma, i.e., rotational temperature (T_rot), vibrational temperature (T_vib), and excitation temperature (T_exc), were estimated using the synthetic spectra technique.

The atmospheric-pressure plasma source was composed of a cylindrical electrode (diameter = 12 mm) in a dielectric tube (alumina; thickness = 2 mm) and powered by a 13.56 MHz radio-frequency power generator via a matching network, as illustrated in Fig. 1. A detailed description of the experimental setup is provided [1]. Ar (99.999 %) gas with a flow rate of 15 L⋅min⁻¹ was introduced as the main plasma gas. HMDS (99.9 %, Sigma-Aldrich) vapor was generated and carried to the plasma with a 0.02 L·min⁻¹ Ar carrier gas passing through a bubbler system. The influence of N₂ on the Ar/HMDS plasma was studied by varying the N₂ gas flow rate from 0 to 0.1 L⋅min⁻¹. The emission spectra obtained from the plasma center were analyzed using a spectrometer (Princeton Instruments, IsoPlane SCT320) equipped with a charge-coupled device (Princeton Instruments, PIXIS400B) and a grating (1,800 grooves/mm, blaze wavelength = 500 nm) in the spectral range of 300–900 nm. The wavelength and sensitivity calibrations of the spectrometer were performed using a standard lamp source, and the full width at half maximum of the instrumental broadening of the spectrometer was estimated to be 0.084 nm.

Figure 1. Scheme of the atmospheric pressure-PECVD setup equipped with a visible spectrometer.

Three temperatures (T_rot, T_vib, and T_exc) were calculated by fitting numerically simulated spectra to the observed spectra. The theoretical emission intensity of N₂ molecules, considering the collisional quenching rate between Ar and N₂ [15,16] as follows:

where c is the speed of light, k is the Boltzmann constant, Q represents the partition functions, N₀ is the density of particles, λ₀) is the line position in nm unit, E_e, E_v, and E_r are the electronic, vibrational, and rotational energies, respectively, ${\overline{R}}_e^2$ is the electronic transition moment, q(v′,v″) is the Franck–Condon factor, S(J′,J″) is the Hönl–London factor, and τ_eff is the effective lifetime. Single prime (′) and double prime (″) denote the upper and lower energy levels [14]. Based on Eq. (1), theoretical N₂ spectra at different temperatures were obtained by considering an instrumental broadening of 0.084 nm, as illustrated in Fig. 2(a). Figure 2(b) depicts an example of the measurement of temperatures (T_rot, T_vib, and T_exc) by comparing the best-fitted synthetic N₂ spectra to the observed N₂ spectra of the Ar/HMDS/N₂ plasma (for 0.005 L•min^-1 N₂) at 100 W.

Figure 2. (a) Simulated synthetic N₂ spectra for different temperature cases. The instrumental broadening for spectra simulation is 0.084 nm. (b) Comparison of the observed spectra (BLACK) with the synthetic spectra of N₂ (RED) for a 100 W Ar/HMDS plasma with the addition of 0.05 L⋅min⁻¹ N₂.

Optical emission spectroscopy was employed to monitor the reactive species and determine the characteristics of Ar/HMDS/N₂ plasma. Typical optical emission spectra of Ar/HMDS/N₂ discharge at 100 W with Ar (15 L⋅min⁻¹) and HMDS (0.1 L⋅min⁻¹) in the ultraviolet to visible (VIS) ranges are illustrated in Fig. 3. For clarity, the spectra are presented in two ranges, 380–900 and 300–550 nm, with different exposure times. Strong Ar atomic lines and N₂ first positive system (FPS, $B^3{\Pi }_g\to A^3{\Sigma }_u^{+}$) were mainly observed in the range of 550–900 nm, whereas N₂ second positive system (SPS, $C^3{\Pi }_u\to B^3{\Pi }_g$), OH 3064 Å system ($A^2{\Sigma }^{+}\to X^2\Pi$), and various carbon-containing diatomic molecular bands such as CH 4300 Å system ($A^2\Delta \to X^2\Pi$), C₂ swan system ($A^3{\Pi }_g\to X{\text{'}}^3{\Pi }_u$) and CN violet system ($B^2\Sigma \to X^2\Sigma$) were observed in the shorter wavelength region of 380–550 nm. The observed reactive species originated from the dissociation of gases (N₂, HMDS, and ambient gas) and the recombination of their fragmented species. For example, CH primarily originated from the dissociation of HMDS in plasma, whereas CN resulted from the recombination of dissociated carbon and nitrogen.

Figure 3. (a) Emission spectrum in the range of 380–900 nm and (b) 300–550 nm of Ar/HMDS/N₂ plasma at 100 W with 0.005 L⋅min⁻¹ of N₂.

Based on the synthetic spectra technique [14–16], the plasma temperatures were obtained by analyzing the observed spectra. Rotational and vibrational motions deliver energy and facilitate chemical reactions in the plasma [17]. In addition, the energy distribution of the excitation states enables the estimation of changes in the energy path channel and/or electron temperature [14,18]. Consequently, the temperature information must be obtained to understand the plasma characteristics. The T_rot was determined by analyzing the rotational band systems of the N₂ SPS (1,4), while the T_vib was obtained by observing three N₂ SPS rovibrational bands (Δv = 3, v′ = 0, 1, and 2). The T_exc was evaluated using two electronic band systems: N₂ SPS and N₂ FPS. As observed in Fig. 4, as the N₂ flow rate increased from 0 to 0.1 L⋅min^-1, T_rot, T_vib, and T_exc reduced from 696–538 K, 2372–1978 K, and 5659–5584 K, respectively. The presence of N₂ provided additional electron energy loss channels, such as molecular vibrations and rotations, which could have decreased the temperature in the Ar/HMDS plasma. Notably, the addition of trace amounts of N₂ (≤0.01 L⋅min^-1) substantially influenced the thermal characteristics of the Ar/HMDS plasma, as well as the composition ratio of reactive species in it [Fig. 5(a)]. Conversely, T_rot and T_exc at an N₂ flow rate of over 0.01 L⋅min^-1 increased marginally, which could be attributed to the charge exchange between Ar ions and the increased number of N₂ molecules [18]. In general, the additional N₂ molecules in the plasma can be converted into $N_2^{+}$ via charge transfer with Ar ions (Ar⁺ + N₂ → Ar + $N_2^{+}$ ) owing to the approximately equal ionization potentials of Ar (15.76 eV) and N₂ (15.58 eV). Because the produced $N_2^{+}$ ions convert into two nitrogen atoms by exothermic dissociative recombination, the temperature can be increased [18].

Figure 4. Influence of N₂ on the rotational temperature (triangles), vibrational temperature (stars) and excitation temperature (circles).

Figure 5. (a) Normalized intensities of molecular and atomic emission spectra and (b) their intensity ratios as a function of N₂ flow rate in Ar/HMDS/N₂ plasma.

Figure 5(a) illustrates the intensities of the emitted species, such as N₂ (C-B), CN (B-X), CH (A-X), C₂ (A-X), OH (A-X), and Ar I (2p₃→1s₅, 706.7 nm), plotted for various N₂ flow rates. The N₂ emission intensity IN₂ increased with increasing N₂ flow rate up to 0.05 L⋅min^-1 and then declined. Likewise, the CN emission intensity ICN soared at a low N₂ flow rate of ≤ 0.005 L⋅min^-1 and plummeted at higher rates. The emission intensities of CH, OH, C₂, and Ar (denoted by I_CH, I_OH, I_C2, and I_Ar, respectively) decreased by 63, 70, 89, and 64 %, respectively, as the flow rate increased from 0.0 to 0.1 L·min^-1. The overall emission intensities, except for that of the N₂ molecule, decreased owing to the weaker ionization (lower electron density) caused by N₂, which provided additional electron energy loss channels and reduced the dissociation of HMDS-generating atomic carbons. Because of this reduced dissociation of the molecules, the number of carbon-containing molecules (CN and C₂) rapidly decreased, as shown in Fig. 5(a). The OES results closely matched the variations in the chemical compositions of the deposited films [1]. The addition of N₂ led to changes in the degree of dissociation of the HMDS precursor forming more Si-CH₃ in the deposited films.

To compare the relative concentrations of reactive species in the plasma, the intensity of the Ar atom (706.7 nm) was employed as a base species because the Ar gas flow rates did not change and Ar was not involved in the chemical reactions. The emission intensity ratios of I_CN, I_CH, I_C2, and IOH to I_Ar as functions of the N₂ flow rate are presented in Fig. 5(b). Notably, we observed an inflection point for each emission intensity at 0.01 L⋅min^-1, which was the same for the changes in temperature (Fig. 4). Interestingly, we confirmed that N₂ flow dependence differed for each species. For example, when the N₂ flow rate was less than 0.01 L⋅min^-1, the intensity ratio ICN/I_Ar increased with increasing N₂ flow rate, while I_CH/I_Ar and I_OH/I_Ar decreased. Conversely, when the N₂ flow rate was above 0.01 L⋅min^-1, the intensity ratios I_CH/I_Ar and IOH/I_Ar exhibited an upward trend and eventually saturated. However, the ratios of carbon-containing species, such as ICN/I_Ar and I_C2/I_Ar, rapidly decreased as the N₂ flow rate increased. This indicates that the CN and C₂ species originating from the recombination between the highly dissociated components of HMDS significantly decreased because the electron energy dispersed to various loss channels by N₂ introduction was insufficient to form carbon atoms from HMDS, as shown in Fig. 4.

In this study, the influence of N₂ as an additive on the temperature and reactive species in Ar/HMDS plasma was studied. Plasma properties such as chemical composition and thermal characteristics were investigated using optical emission spectroscopy. In particular, by simulating the synthetic spectra of N₂ molecules, such as N₂ FPS and N₂ SPS, plasma temperatures such as the excitation (T_exc), vibrational (T_vib), and rotational temperature (T_rot) were estimated. From the results, we discovered that each energy state showing different temperatures revealed that the plasma was not in thermal equilibrium. In addition, even in small quantities, N₂ addition cooled the Ar/HMDS plasma. Interestingly, a drastic change in the composition ratio of the plasma was observed when the additive N₂ flow rate was below or above approximately 0.01 L⋅min^-1. For example, at lower N₂ flow rates (0.005–0.010 L⋅min^-1), more CN molecules were observed. Conversely, at higher N₂ flow rates (>0.05 L⋅min^-1), CH and OH became the majority molecules, while the number of CN and C2 species progressively decreased because of the reduction in the electron temperature caused by N₂ addition. The presence of N₂, which provides more energy loss channels, reduced the decomposition rate of HMDS, which explains the effects of additive N₂ on the properties of thin films, such as changes in morphology and composition ratio [1,9,19]. Furthermore, we determined that N₂ could modulate the plasma temperature, playing an important role in the production and loss of species, which can be useful for efficiently controlling the properties of the deposited film.

This work was supported by a Korea Basic Institute (National Research Facilities and Equipment Center) under a grant funded by the Ministry of Education (2021R1A6C101B383), by the R&D program of Plasma Equipment Intelligence Convergence Research Center (No. 1711121944), and by SAMSUNG Electronics (No. IO210129-08345-01).

The authors declare no conflicts of interest.