Atomic Layer Deposition of Ru Thin Film Using a Newly Synthesized Precursor with Open‐Coordinated Ligands

Ru films are grown on Pt, TiN, and SiO2 substrates via atomic layer deposition (ALD) using Ru(II)(η5‐C7H7O)(η5‐C7H9) as the novel Ru metalorganic precursor and O2 as the reactant. The ALD self‐limiting film growth is confirmed at the low temperature of 200 °C by manipulating the injection time and purge time of the Ru precursor and O2, and the saturated growth per cycle is 0.22 Å cy−1. It is confirmed that the combustion reaction occurs during the deposition process from the formation of the H2O and CO2 by‐products. Thin films with a low resistivity of 17–19 µΩ cm are grown at a thickness of ≈15 nm. The Ru incubation times are remarkably short at 200 °C (negligible on Pt, ≈30 cycles on TiN and SiO2), but increase with increasing temperature. The energy dispersive X‐ray mapping image of the Ru film on the pattern in which TiN is formed on SiO2 shows that the Ru film with a novel precursor has the intrinsic substrate selectivity at the deposition temperature of 300 °C. Furthermore, the substrate affects the properties of the Ru film. Particularly because Ti serves as an oxygen reservoir, a relatively large amount of RuOx is produced on the TiN substrate.


Introduction
The continuous scaling down of semiconductor devices has increased the density and operating speed of transistors, whereas the dimensions of the interconnect in semiconductor devices have been reduced to a level <39 nm, corresponding to the electron mean free path (EMFP) in the bulk state of Cu, which causes the resistivity of Cu to increase rapidly due to phonon scattering. [1][2][3] This phenomenon causes performance degradation. To address this issue, Ru, Co, and Mo, which have a shorter EMFP than Cu, have been proposed as potential alternatives to Cu.
Ru is recognized as a next-generation interconnect material due to its excellent thermal, chemical, and physical stability, low resistivity (7.1 µΩ cm), high work function (4.7 eV), and catalytic properties. [4−11] Ru has a shorter EMFP (6.6 nm) and better compatibility with the CMOS process than Cu in 5 nm-level technology because it does not require a barrier layer. [12] In addition, Ru is a suitable barrier metal for replacing TiN because it has low resistivity, low solubility compared to Cu, strong adhesion properties, and a high melting point (2334 °C). [13] As the size of the devices decreases, complicated 3D structures, such as trenches, will require precise thickness control, and excellent step coverage. Furthermore, the low-temperature deposition technology is particularly important for device scaling, as it enables the deposition of thin films without damaging the underlying substrate or the delicate components of the device. Atomic layer deposition (ALD) is an essential technology for designing microelectronic devices because it provides relatively superior step coverage and film uniformity compared to other deposition methods and allows precise thickness control at the atomic level through ALD cycle control. [14,15] In the ALD of films, the characteristics of the precursor may directly affect the quality of the film. Precursors with cyclopentadienyl (Cp) ligands, which have been extensively studied and experienced a nucleation delay, which critically impacts the properties of the film. [13,[16][17][18][19][20][21][22][23] In other words, these precursors are not suitable for application in interconnect processes because the overall long deposition time increases the cost of manufacturing and the surface roughness deteriorates owing to poor nucleation. According to the documentation, Ru ALD Ru films are grown on Pt, TiN, and SiO 2 substrates via atomic layer deposition (ALD) using Ru(II)(η 5 -C 7 H 7 O)(η 5 -C 7 H 9 ) as the novel Ru metalorganic precursor and O 2 as the reactant. The ALD self-limiting film growth is confirmed at the low temperature of 200 °C by manipulating the injection time and purge time of the Ru precursor and O 2 , and the saturated growth per cycle is 0.22 Å cy −1 . It is confirmed that the combustion reaction occurs during the deposition process from the formation of the H 2 O and CO 2 by-products. Thin films with a low resistivity of 17-19 µΩ cm are grown at a thickness of ≈15 nm. The Ru incubation times are remarkably short at 200 °C (negligible on Pt, ≈30 cycles on TiN and SiO 2 ), but increase with increasing temperature. The energy dispersive X-ray mapping image of the Ru film on the pattern in which TiN is formed on SiO 2 shows that the Ru film with a novel precursor has the intrinsic substrate selectivity at the deposition temperature of 300 °C. Furthermore, the substrate affects the properties of the Ru film. Particularly because Ti serves as an oxygen reservoir, a relatively large amount of RuO x is produced on the TiN substrate.
techniques utilizing zerovalent Ru precursors demonstrated superior initial growth characteristics. [24][25][26][27] A Ru precursor with an open-coordinated ligand has been proposed for solving the problems associated with the existing precursors. This precursor is expected to have higher reactivity than the precursor with the Cp ligand, according to the Green-Davies-Mingos rule. [28] Several research groups have documented the creation of an open-coordinated ligand precursor and its implementation in the ALD method, which validates the efficacy of the open-coordinated ligand precursor as an ALD precursor due to its outstanding nucleation performance. [6,29,30] In this study, a new Ru precursor with an open-coordinated ligand, Ru(II)(η 5 -C 7 H 7 O)(η 5 -C 7 H 9 ), is synthesized, and an ALD process is developed. For many noble metal ALD processes, including Ru ALD, O 2 has been used as a reactant to produce CO 2 and H 2 O as combustion by-products. The catalytic surface of Ru has the characteristic to catalyze dissociative chemisorption of O 2 molecules during the O 2 injection time, then hydrocarbon ligands undergo a combustion reaction with O 2 atoms during the Ru precursor injection time. [31,32] In this case, since the new Ru precursor contains oxygen, the precursor itself can act as an oxygen source to accelerate the combustion reaction. The physical properties of the newly synthesized compound are confirmed, and the growth behavior of the Ru thin film is discussed along with the physical, chemical, and electrical properties of the Ru film. Moreover, intrinsic substrate selectivity is observed at high temperatures.

Results and Discussion
Thermogravimetric analysis (TGA) of the Ru(II)(η 5 -C 7 H 7 O)(η 5 -C 7 H 9 ) complex was acquired from room temperature to 800 °C under a constant flow of nitrogen gas (Figure 1) to avoid exposure to air. The TGA trace of Ru(II)(η 5 -C 7 H 7 O)(η 5 -C 7 H 9 ) shows a single vaporization step at 230 °C and a negligible amount of residue (<1%). The differential thermal analysis (DTA) curve of Ru(II)(η 5 -C 7 H 7 O)(η 5 -C 7 H 9 ) did not exhibit peaks related to thermal decomposition in the temperature range of 25-230 °C, suggesting the excellent evaporation and thermal stability characteristics of the Ru precursor.
By increasing the injection time and purging time of the precursor and reactant, the ALD self-limited growth behavior of the Ru precursor was considered based on the Ru layer density growth per cycle (GPC) on the Pt substrate at a deposition temperature of 200 °C. Figure 2a,b shows a typical ALD saturation curve, in which the GPC of the Ru film increases as the precursor and reactant injection step times increase. The thickness no longer increased at ≈27 ng cm −2 cy −1 , which corresponds to a step time increase of 0.22 Å cy −1 . When only the Ru precursor was injected on the substrate by setting the oxygen injection time to 0 s, ruthenium atoms were not detected, and it was confirmed that the precursor was not deposited through thermal decomposition, but rather by the ALD reaction between the precursor and the reactant to form a thin film. Consequently, the number of ALD cycles for the Ru film was adjusted to 70 s (Ru precursor injection), 5 s (Ru precursor purge), 10 s (O 2 injection), and 5 s (O 2 purge). Although the injection time of the precursor was set to be relatively long at 70 s, it can likely be reduced by increasing the canister temperature. Time-of-flight mass spectroscopy (ToF-MS) was used to confirm the chemical reaction during the Ru ALD process. Here, the pulse sequence 9-20-20-20, which is shorter than the saturation condition, was adopted for the ALD process. As depicted in Figure 2c, the formation of H 2 O (m/z = 18) as a by-product when the Ru precursor and reaction gas were injected was confirmed. Clear peaks were confirmed when O 2 was injected, and a signal with higher intensity than the background level was detected, although no peak was observed during precursor injection owing to the low vapor pressure. Although CO 2 (m/z = 44) interfered with Ar (m/z = 40), it was confirmed that CO 2 was generated when O 2 was injected, based on comparison of the peak intensities of Ar and CO 2 ( Figure S1, Supporting Information). Considering that H 2 O and CO 2 are generated as reaction by-products, it is inferred that the hydrocarbon ligand undergoes the combustion reaction, and Ru forms a thin film, as reported in previous studies. [32] Notably, a small but persistent reaction by-product was generated during Ru injection after the 2nd cycle, indicating the involvement of a slow reaction. According to previous studies, this process is the diffusion of oxygen incorporated into the Ru thin film. Figure 2d shows the change in the film thickness and resistivity as a function of the deposition temperature in the range of 100-300 °C. The GPC values were highest at 200 °C (27 ng cm −2 cy −1 ). The GPC values decreased to 1.6 ng cm −2 cy −1 at a low temperature of 100 °C, and also dropped to 5.1 ng cm −2 cy −1 at a higher temperature of 300 °C. The decrease in the GPC at temperatures below 200 °C is due to insufficient thermal energy for the combustion reaction. When the temperature surpasses 200 °C, the desorption rate increase of physisorbed precursors exceeds the increase of their deposition rate via the combustion reaction. Consequently, nucleation was delayed, and the average deposition rate decreased. As described later, this effect on the rate of deposition of Ru on the substrate disappeared after nucleation, resulting in a higher deposition rate at high temperatures. The change in resistivity as a function of the deposition temperature was also measured, where a low resistivity of 17-19 µΩ cm was confirmed in the

www.advmatinterfaces.de
200-250 °C range. Here, the thickness of each film was controlled in the range of 10-15 nm. This is similar to or superior to that of previously reported Ru ALD thin films. [7,9,24,26,33,34] The Ru film showed a higher resistivity of 146 µΩ cm at a low temperature of 175 °C, which is expected to be due to the high impurity concentration due to insufficient surface reactivity, and a value of 58 µΩ cm at a high temperature of 300 °C, which is expected to be due to high roughness due to the inferior nucleation properties.

Figure 3a
shows the variations in the layer density of the Ru films as a function of the number of cycles in the range of 10-150 cycles at a deposition temperature of 200 °C. Three types of substrates (Pt, TiN, and SiO 2 ) were used to confirm the dependency on the substrate material. Although no incubation cycles occurred on the Pt substrate, ≈30 cycles of incubation occurred on the SiO 2 and TiN substrates, where the quality of these films is expected to be inferior to that of the thin film deposited on Pt. The ALD incubation cycle is dependent on the

www.advmatinterfaces.de
binding properties of the substrate. The bonding characteristics of the substrate play a decisive role in the initial stage of adsorption of the precursor molecules. In the case of Pt, adsorption of the precursor molecules on the metallic surfaces was enhanced by metallic bonding. In contrast, the TiN and SiO 2 substrates formed relatively weaker covalent bonds, resulting in an incubation cycle. [35,36] After the incubation cycle, the layer density of the thin films on the TiN and SiO 2 substrates increased linearly, and the saturated GPC of the Ru film was 0.2 Å cy −1 regardless of the substrate. As shown in Figure 3b, the layer density of the thin films on the Pt and TiN substrates increased linearly after 100 cycles at a deposition temperature of 300 °C. The GPC was 0.3 Å cy −1 , indicating a faster growth rate than the GPC at 200 °C. This result shows that as the deposition temperature increased to 300 °C, nucleation became inefficient due to increased desorption of the precursor, thereby inhibiting the initial growth of the thin film. On the other hand, nucleation of the Ru film on the SiO 2 substrate did not occur, and the substrate showed selectivity, wherein no film was deposited on the SiO 2 even after 300 cycles. These substrate-selective characteristics were confirmed by energy-dispersive X-ray (EDX) analysis on the pattern in which TiN is formed on SiO 2 ( Figure S2, Supporting Information).
Ru, O, and C in the films were quantified by time-of-flight secondary ion mass spectrometry (ToF-SIMS) depth profiling. Figure 4 presents a comparison of the profiles of the Ru thin film deposited by sputtering and the Ru thin film deposited by ALD, which was used to identify the C impurities. Here, the Ru thin film deposited by sputtering was used as a reference sample because it is considered to have a very low C impurity from the characteristic that has no C source in the thin film deposition process. As shown in Figure 4a, no difference in the C impurity content of the sputtered Ru and ALD Ru was observed at the top of the thin film on the Pt substrate, whereas a relatively large amount of C impurity was observed at the bottom of the film. It can be inferred that the reaction of the precursor was promoted by the catalytic effect of the Pt substrate in the initial stage of deposition, but hydrocarbon residues were formed because of insufficient oxygen supply. In particular, it was confirmed that less C impurity was present in the film deposited on the TiN substrate by ALD. This seems to be because the partially oxidized TiN surface serves as a reservoir for supplying the oxygen necessary for the combustion reaction. [37,38] Because C is removed in the form of CO x upon contact with oxygen on the Ru surface, the amount of C impurities decreases in an oxygen-rich environment. As a result, the Ru thin film formed on the TiN substrate has a high concentration of oxygen. Figure S3 (Supporting Information) shows the Ru thin film formed on the TiN substrate has a high O/Ru ratio compared to that of the other substrates. This is attributed to the facile redox reactions of the TiN surface. On the SiO 2 substrate, there was no significant difference in the concentration of C impurities in the sputterdeposited film and the film deposited by ALD (Figure 4c).
The chemical compositions of the Ru films deposited on the Pt, TiN, and SiO 2 substrates were evaluated by X-ray photoelectron spectroscopy (XPS). Figure 5a shows the Ru 3d XPS profiles of the thin films deposited on the Pt, TiN, and SiO 2 substrates. The Ru 3d 5/2 peak appeared at a binding energy of ≈280 eV, indicating successful growth of the metallic Ru film. The O 1s XPS profiles of RuO 2 and RuO x /Ru (Figure 5b) showed peaks at binding energies of 529.4 and 531.2 eV, respectively. The peak corresponding to RuO 2 was evident only in the spectrum of the Ru film grown on the TiN substrate. These XPS data demonstrate that the Ru film grown on the TiN substrate contained more of the RuO 2 phase than the Ru film grown on the Pt and SiO 2 substrates, consistent with the ToF-SIMS analysis. Figure 6a shows the X-ray diffractometer (XRD) patterns (obtained by the θ-2θ method) of the as-deposited 10 nm-thick Ru films grown on different substrates at a temperature of 200 °C. The diffraction peaks of the thin films grown on both TiN and SiO 2 substrates were observed at 38.3°, 42.2°, and 44.0°, corresponding to the (100), (002), and (101) planes of Ru metal, respectively. No crystal peak corresponding to RuO 2 was observed for the films on any of the substrates because RuO 2 was present in a relatively small amount or in an amorphous state.
Only one peak at 42.2° was observed for the Ru film grown on the Pt substrate. This is because the crystals of the Ru thin film grew in the preferred orientation along the vertical orientation of the Pt (111) plane. Because the lattice mismatch between the Pt (111) and Ru (002) planes was only 2.52%, preferential formation of the Ru film on the Pt substrate was observed. [36,39] Therefore, no peak corresponding to Ru metal on the Pt substrate was observed in glancing angle mode XRD ( Figure S4, Supporting Information).

www.advmatinterfaces.de
The measured X-ray reflectometry (XRR) curve and simulated curve of the Ru films on the three types of substrates are shown in Figure 6b. The densities were 12.2, 10.35, and 12.19 g cm −3 , respectively, which are 97%, 83%, and 97% of the theoretical density of metallic Ru (12.45 g cm −3 ). Notably, the density of the Ru film on the TiN substrate was lower than that of the other substrates, and the low-density RuO x formed in the Ru thin film on the TiN substrate contributed to lower the overall density. Figure 7a shows the surface morphologies of the thin films of Ru on the Pt, TiN, and SiO 2 substrates, as observed by atomic force microscopy (AFM). The thickness of the Ru film was ≈10 nm. The root-mean-square (RMS) roughness increase of the Ru films on the Pt, TiN, and SiO 2 substrates was 0.22, 0.02, and 1.26 nm, respectively (Figure 7b). The difference in the RMS roughness of the Ru films grown on the Pt and TiN substrates before and after the ALD process was negligible, whereas on the SiO 2 substrate, the RMS roughness increased by 1.2 nm due to poor nucleation properties. Figure 8 shows a cross-sectional transmission electron microscopy (TEM) image of the Ru films grown on the trench structure (opening diameter ≈115 nm, depth ≈430 nm, and aspect ratio ≈3.73). The thickness of the Ru film at the top, middle, and bottom of the trench structure was 14.86, 14.2, and 10.98 nm, respectively, further demonstrating that the side step coverage was 0.95 and the bottom step coverage was 0.7, which is the advantage of ALD.

Conclusion
A new Ru precursor, Ru(II)(η 5 -C 7 H 7 O)(η 5 -C 7 H 9 ), was synthesized. TGA indicates that the complex shows promising evaporation characteristics and thermal stability for use as an ALD precursor. Ru thin films were grown on Pt, TiN, and SiO 2 substrates by ALD at a deposition temperature of 100-300 °C using a Ru precursor and O 2 . ALD-specific self-limiting growth was demonstrated, with a saturated growth rate of 0.22 Å cy −1 at  www.advmatinterfaces.de the relatively low temperature of 200 °C, where the resistivity of the 10 nm Ru film was also low (18 µΩ cm). There were no incubation cycles on the Pt substrate, and ≈22 cycles occurred on the TiN and SiO 2 substrates. As the temperature increased, the reaction rate and steady-state growth rate increased, but the incubation cycle increased rapidly, retarding the overall growth of the film. Ru metal was formed on the Pt and SiO 2 substrates, whereas on the TiN substrate, a relatively large amount of RuO x was formed because Ti acted as an oxygen reservoir. Furthermore, the Pt and TiN substrates enabled facile formation of the Ru film owing to the excellent nucleation properties of the substrates, which led to a less pronounced increase in the roughness; however, the Ru film formed on the SiO 2 substrate exhibited increased surface roughness due to the poor nucleation properties. The grown films showed high step coverage, even in the trench structure.

Experimental Section
Synthesis of Ru(II)(η 5 -C 7 H 7 O)(η 5 -C 7 H 9 ): All chemicals and solvents were of reagent grade and were employed as received without additional purification. Using a Bruker DPX 500 MHz FT-NMR spectrometer, proton ( 1 H NMR) and carbon-13 ( 13 C NMR) nuclear magnetic resonance spectra were obtained. All samples were placed in sealed NMR tubes, and the signals were referenced using a benzene-d 6 standard. Fouriertransform infrared (FT-IR) spectra were obtained using a Nicolet Nexus FT-IR spectrophotometer. TGA was conducted under N 2 condition at a scan rate of 10°C min −1 with a Thermo Plus EVO II TG8120 series thermogravimetry and differential thermal analysis instrument from Rigaku. Elemental analysis was conducted using a Thermo Scientific Flash 2000 instrument. Thermogravimetric and elemental analyses were performed by the Chemical Analysis Center at the Korea Research Institute of Chemical Technology.
The synthesis of the Ru precursor is shown in Scheme 1.
[Ru(cymene) Cl 2 ] 2 (6.12 g, 10 mmol), tert-butanol (150 mL), and Na 2 CO 3 (10 g) were placed in a round-bottom flask. After adding cycloheptatriene (7.37 g, 40 mmol), the mixture was refluxed for 15 h. The solution was filtered to get rid of salts and cooled to room temperature. The filtrate was concentrated by sublimation under vacuum (80 °C, 10 −1 Torr) to obtain a pure product as a yellowish solid (5.12 g, 85% ALD of Ru Film: Ru films were grown in a 4'' (inch)-scale traveling wave-type chamber (Custom Built, Isac Research, Korea) at temperatures ranging from 100 to 300 °C. Ru(II)(η 5 -C 7 H 7 O)(η 5 -C 7 H 9 ) and 500 sccm of O 2 were used as the Ru precursor and oxidant, respectively. The Ru precursor was vaporized in a bubbler-type canister at 75 °C and carried to the chamber by Ar gas at 500 sccm. The unreacted chemicals and by-products were purged with 500 sccm of Ar gas. The temperature of the outer lining was maintained at 100 °C to prevent condensation of the precursors. The throttle valve was modulated to keep the process pressure at 1.1 Torr. To analyze the substrate selectivity, specifically the initial growth features, sputter-deposited 150 nm thick Pt/Ti/SiO 2 /Si was used, which was compared to sputter-deposited TiN and thermally grown SiO 2 . Ru precursor injection, Ru precursor purge, O 2 injection, and O 2 purge are the four sequential processes in each ALD cycle. To discover the ALD's self-limiting behavior, various Ru precursor and O 2 feeding times were tested. The resistivity of the thin film was confirmed by the formation of a 10 nm Ru film on the SiO 2 substrate. The trench pattern wafer size is 40 nm(space)/40 nm(width) and Si trench depth is 400 nm, which was fixed at the whole pattern.
Step coverage was performed with an ALD cycle of 270 at a deposition temperature of 200 °C.
The by-products of the ALD process were evaluated by ToF-MS (PM-ToF, JJCNS). The layer density of the Ru atoms was measured  www.advmatinterfaces.de using X-ray fluorescence (XRF, ARL QUANT'X, Thermo Scientific). Further, the density of the Ru films was measured by XRR (SmartLab, Rigaku). The crystallographic structures of Ru and RuO 2 were examined using an XRD (SmartLab, Rigaku) in θ-2θ and glancing angle modes. The film morphology was evaluated by AFM (multimode8, Bruker). The chemical composition and binding states of the deposited films were examined using XPS (PHI Quantera-II, Ulvac-PHI), and the impurity concentrations, especially oxygen and carbon, in the films were examined using ToF-SIMS (ToF-SIMS 5, IONTOF). The resistivity was determined by measuring the sheet resistance using a 4-point probe (CMT-SR1000N, AIT). The step coverage in the trench structure was examined using TEM (Tecnai G2 F30 S-TWIN, FEI), where the sample was prepared using focused ion beam etching (FIB, Helios NanoLab, FEI). The substrate selectivity at 300 °C was evaluated using scanning electron microscopy (SEM, Vega II LSU, TESCAN) and EDX elemental line mapping.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.