To optimize the ALD process, a comprehensive study of process parameters including: (1) precursor pulse time (0.5, 1, and 2 s), (2) H2 plasma pulse time (5, 10, and 20 s), (3) purge time after precursor/H2 plasma pulse (15, 30, and 60 s), and (4) deposition temperature (150, 200, 250, and 275°C), as shown in Fig. 1. The XRR results of the film thicknesses as functions of the pulse time, H2 plasma pulse time, and purge time after the precursor/H2 plasma pulse are shown in Fig. S2, and the results are summarized in Table S1. As shown in Fig. 1 (a)–(c) and Table S1, the carbonyl cyclohexadiene ruthenium precursor showed a constant GPC independent of the precursor pulse time, H2 plasma pulse time, and purge time after the precursor/H2 plasma pulse, which is a well-known deposition characteristic of ALD based on self-limiting growth behavior. In contrast, the carbonyl cyclohexadiene ruthenium precursor is highly sensitive to the deposition temperature. The GPC at 250°C is stable and shows high uniformity independent of the deposition location, whereas the GPC at 150, 200, and 275°C is higher than that at 250°C and exhibits significant fluctuations depending on the deposition location, as shown in Fig. 1 (d) and Table S1. The higher GPC and significant fluctuations in the deposition thickness may be due to the insufficient thermal energy to complete surface reactions at low temperatures and the possible decomposition of surface species with additional reactant adsorption at high temperatures [25]. Thus, it is plausible that an ALD window exists within a limited temperature range close to 250°C.
Based on the results presented in Fig. 1 and Table S1, the standard process condition for PEALD Ru was established: 1 s Ru precursor pulse, 10 s exposure time, 30 s Ar purge time, 10 s H2 plasma pulse time, and 30 s Ar purge time at a temperature of 250°C. The thicknesses of the PEALD Ru films as a function of the number of ALD cycles was investigated, as shown in Fig. 2 (a) and (b). The PEALD Ru films prepared by 50, 100, and 200 cycles exhibited thicknesses of 7.3 ± ~ 0.2, 13.4 ± ~ 0.2, and 34.3 ± ~ 0.3 nm, respectively. The thickness of PEALD Ru increased linearly with the number of ALD cycles, which is a well-known self-limiting ALD growth characteristic. XPS and XRD analyses were conducted to investigate the film purity and crystallinity of PEALD Ru. Figure 2 (c) shows the high-resolution XPS spectra of the Ru 3d and C 1s peaks of the PEALD Ru films based on the XPS survey scan results (Fig. S3). The Ru 3d and C 1s peaks were deconvoluted for carbon analysis because of the binding energy overlap between Ru 3d and C 1s. The detailed data for the fitted single components are summarized in Table S2. PEALD Ru contained 95.54 at% Ru, 3.71 at% C, and 0.74 at% O, indicating that the PEALD Ru prepared using the carbonyl cyclohexadiene ruthenium precursor has low carbon and oxygen impurities. Figure 2 (d) shows the GIXRD results, which indicate that the PEALD Ru film is polycrystalline, and the (002) and (101) orientations of Ru are the most prominent (from the standard ICDD, PDF Card No.: 00-006-0663).
The conformality of PEALD Ru was evaluated by deposition on a ~ 1:4 aspect ratio trench Si substrate (~ 3 µm hole diameter and ~ 12 µm depth) under the optimized process conditions, as shown in Fig. 3. It is recognized that PEALD has relatively low conformality compared with thermal ALD because of the inhibitory effect of the surface recombination of plasma species during the penetration of plasma species into the hole structure [26]. However, optimizing the H2 plasma pulse time allows conformal film deposition, which is possibly due to the plasma species that were not recombined with the substrate entering well into the bottom of the hole structure. In this study, the Ru thin film exhibited a thickness of 39 nm at the top surface, 38.7 nm at the mid-surface, and 35.7 nm at the bottom surface. These results indicated conformal deposition with a thickness difference of approximately 5% between the top and bottom surfaces.
Figure 4 shows the morphologies of the ALD Ru films, as well as the resistivity and thickness of the films, as a function of the number of ALD cycles. The film morphology and electrical resistivity were characterized using HR-SEM and a four-point probe. From the HR-SEM images, nucleation islands (red circles in Fig. 4 (a) and (b)) remained on the ALD Ru film surface after 50 and 100 cycles, which is attributed to island growth instead of ideal layer-by-layer growth during the initial nucleation stage [27, 28]. In contrast, a smooth full-film of metallic Ru without nucleation islands was observed after 200 ALD cycles, as shown in Fig. 4 (c). The electrical properties of the Ru thin films were characterized based on the morphological analysis. Relatively high resistivities (39.2 and 32.6 µΩ·cm) were measured after 50 and 100 cycles owing to the noncontinuous film morphology. However, the Ru film prepared by 200 cycles has low resistivity (28.8 µΩ·cm) and a dense morphology, as shown in Fig. 4 (c) and (d). Electrical resistivity is known to be influenced by scattering effects at the surface, interface, and grain boundaries of thin films, and these scattering effects can become dominant with decreasing film thickness [29]. Additionally, the thin film density is considered a key factor affecting the resistivity [30]. Consequently, the optimized Ru thin film, deposited using carbonyl cyclohexadiene ruthenium and H2 plasma by PEALD, exhibits a resistivity of 28.8 µΩ·cm, which is comparable with the range of resistivity values reported in previous studies using different combinations of precursors and reactants (10–36 µΩ·cm) [21].