1. Introduction

Ruthenium has been widely used in electronic devices, chemical industry, medical treatments and other fields due to its good thermal stability, low resistivity, high catalytic activity and other characteristics (1–3). Recently, ruthenium is used for the interlayer or antiferromagnetic coupling layer of magnetic recording devices (4–6), where it plays a crucial role in controlling the growth of the magnetic recording layer of hard disks. In addition, it also has great application potential in next-generation integrated circuit wiring materials, magnetic random-access memory, micro-electromechanical systems and other fields (4–9).

To obtain the best ruthenium film preparation process, many researchers have investigated the influence of different preparation techniques (such as magnetron sputtering technology, electron beam evaporation deposition technology and atomic layer deposition technology), different substrate materials (such as silicon, silica and glass) and different sputtering parameters (such as heating temperature, sputtering power, sputtering gas and others) on the microstructure, resistivity, etching and magnetic properties of ruthenium films. For example, Jhanwar, et al . used direct current magnetron sputtering to deposit a ruthenium film on a silica substrate and studied the influence of different sputtering powers on the structure and properties of the ruthenium film (10). Lower resistivity (12.40 μΩ cm) and higher mobility (4.82 cm² V s^–1) were obtained when the films were deposited at high power. Nagano et al . studied the relationship between the orientation, resistivity, crevice area and density of ruthenium films with change of substrate temperature during sputtering deposition (11). Low resistivity, crack-free and smooth ruthenium films can be obtained at 700°C. However, in the existing literature, there are few studies on the microstructure and performance of ruthenium films with the microstructure of ruthenium targets. Therefore, in this paper, the influence of ruthenium targets on the morphology, orientation and electrical properties of films deposited at different deposition times by RF magnetron sputtering were studied. This provides experimental data for the preparation of ruthenium films and theoretical guidance to optimise the production of ruthenium targets.

2. Materials and Methods

2.1 Preparation of Ruthenium Targets

Two kinds of ruthenium powder (99.95 wt%) with different particle sizes and morphologies were selected, as shown in Figure S1 in the Supplementary Information. The 1^#Ru powder was nearly spherical. 2^#Ru powder was composed of granular and porous lamellar. The ruthenium targets were prepared by vacuum hot pressing furnace (VHP200/20-2100), the pressure was set at 40 MPa and the sintering temperatures of the 1^#Ru and 2^#Ru powders were 1050°C and 1250°C, respectively. Targets prepared using 1^#Ru and 2^#Ru powders are hereafter denoted as 1^#Ru and 2^#Ru targets, respectively.

2.2 Preparation of Ruthenium Films

Ruthenium targets prepared as described above were used as the cathode target source and ruthenium films were deposited onto cleaned SiO₂/Si(100) substrate by RF magnetron sputtering using a magnetron sputtering deposition system (JGP-450B). The background vacuum pressure and sputtering pressure were set to 2 × 10^–3 Pa and 3 Pa, respectively. The power was set to 200 W. High-purity argon (99.999 wt%) was used as the sputtering gas and the substrate temperature was 200°C. Four sputtering times were set: 1 min, 5 min, 10 min and 15 min. The ruthenium films deposited by sputtering of the 1^#Ru and 2^#Ru targets are hereafter denoted as the 1^#Ru and 2^#Ru films, respectively.

2.3 Microstructure Observation and Performance Testing

Ultrasonic C-SCAN flaw-detection system (V100) was used to test the internal defects of the targets. The density of ruthenium targets was measured by an Explorer^TM Quasi-Microbalance EX225DZH (OHAUS International Trading Co Ltd, China) with a density-testing component. The phase, crystal structure, orientation and micro-strain of the ruthenium targets and films were analysed by X-ray diffractometer (X’Pert MRD, Malvern Panalytical Ltd, UK). Tungsten filament SEM (S-3400N, Hitachi Science Systems Ltd, Japan) was used to observe the fracture and surface morphology of the ruthenium targets. High resolution field emission SEM (Versa 3D^TM DualBeam^TM, FEI Company, USA) was used to observe the microstructure of the ruthenium films. A digital conductivity meter (Sigma 2008B, Xiamen Tianyan Instruments Co Ltd, China) was used to test the electrical properties of the ruthenium targets and the test method is shown in Figure S2 in the Supplementary Information. A four-point probe (SZT) was used to test the electrical properties of the ruthenium films.

3. Results and Analysis

3.1 Structure Analysis of Ruthenium Targets

The results of ultrasonic C-scan non-destructive testing showed that the interiors of the two ruthenium targets were free of inclusions or porosity defects (as shown in Figure S3 in the Supplementary Information), meeting the requirement of downstream customers. It indicated that the two types of ruthenium powder with different particle sizes and morphologies matched with the two sintering temperatures could be used to obtain ruthenium targets without inclusions or porosity. The density measurements showed that the 1^#Ru and 2^#Ru targets had densities of 12.29 g cm^–2 and 12.31 g cm^–2, respectively. Their respective densification values were 98.71% and 98.88% and the density of the 1^#Ru target was slightly lower than that of the 2^#Ru target. This is related to the fact that the sintering temperature of the 2^#Ru target (1250°C) was higher than that of the 1^#Ru target (1050°C). Firstly, the use of a higher temperature will annihilate the pores in the sintered body, improving its density. Secondly, the densification process of vacuum hot pressing sintering is a process of atomic diffusion. The atomic diffusion coefficient directly affects the speed of atomic motion and the sintering temperature is an important factor determining the atomic diffusion coefficient. Therefore, using a higher sintering temperature could improve the density of the material. However, the 1^#Ru powder had small near-spherical particles and good fluidity, which is beneficial to the merging of voids and the elimination of defects in the sintering process.

The fracture morphology of the ruthenium targets is shown in Figure 1. Under the action of external force, the two ruthenium targets show brittle fracture characters and the fracture is flat and perpendicular to the direction of external force, showing a rough surface with grain shape. When the grain boundary binding force is lower than the grain internal binding force, the grain boundary is weakened and becomes the priority channel for crack propagation under the action of external force. The fracture presents different degrees of the crystal shape of a grain polyhedron and the ruthenium targets show intergranular fracture (12, 13).

From Figure 1 the microstructure of 1^#Ru target is uniform, with grain size in the range of 0.73–2.25 μm and an average diameter of 1.33 μm. However, the microstructure of 2^#Ru target is obviously uneven, with a grain size range of 0.32–3.99 μm and an average diameter of 1.32 μm. Many circular micro-holes (as shown by white arrow in Figure 1(b)) and cleavage steps (as shown by the red dotted box in Figure 1(b)) can be observed in 2^#Ru target, which might be caused by the appearance of transgranular fractures between grains when the ruthenium targets were fractured by external force, indicating that the interfacial bonding strength in local areas was improved during sintering. In addition, the effect of high temperature enhanced the atomic vibration frequency and diffusion migration ability, so that the internal pores of the sintered body merged, the volume shrank and balled. Atoms interacted on the grain-contact surfaces to form into bonding surfaces and these expanded into sintering necks during the sintering process. The particle interfaces were transformed into grain interfaces and extruded each other, causing the grains to grow continuously. Moreover, the 2^#Ru powder is mainly composed of sheet particles with poor fluidity, and the atomic diffusion rate is significantly lower in the normal direction of the sheet powder than in the transverse or longitudinal direction, so that 2^#Ru target with uneven grain size was finally obtained.

Figure 2 shows XRD patterns of the two ruthenium targets. The diffraction peaks of the (0002) and (101) planes had high intensities, while the diffraction peak of the (100) planes was low intensity. By comparing these with the PDF standard card of ruthenium (Joint Committee on Powder Diffraction Standards (JCPDS) No. 65-1863), according to the intensities of the three strong peaks of ruthenium, the crystal plane texture coefficient (TC_hkl) was calculated (Equation (i)):

(i)

where I _{(hkl )} and I _{0(hkl )} represent the diffraction-peak intensity of the crystal plane (hkl) of the sample and the PDF standard, respectively; and n is the number of diffraction peaks (14). If the TC value of a certain (hkl) plane is greater than the average, this indicates that growth along this crystal plane is preferred; the larger the TC value, the higher the degree of preference. The results for the ruthenium targets are shown in Table I. The two ruthenium targets grew preferentially along the (0002) plane and the degree of the preferred orientation of (0002) plane of the 2^#Ru target was slightly larger than that of the 1^#Ru target.

Table I

Crystal Plane Texture Coefficients TC_hkil of Different Ruthenium Targets

Ruthenium PDF JCPDS No. 06-0663		1#Ru target		2#Ru target
(hkil)	I0(hkil)	I(hkil)	TChkil	I(hkil)	TChkil
(100)	40.0	3	2.32%	2.8	2.19%
(0002)	35.0	100	88.43%	100	89.45%
(101)	100.0	29.9	9.25%	26.7	8.36%

3.2 Electrical Properties of Ruthenium Targets

The resistivity results of the ruthenium targets are shown in Table II. The resistivity of 1^#Ru target was lower than that of 2^#Ru target and lower than the theoretical resistivity of ruthenium bulk (7.1 × 10^–6 Ω cm), but the uniformity of resistivity was worse than that of 2^#Ru target. According to analysis using the equivalent resistor network method (15), if a circular tube material with balls is regarded as a ‘columnar crystal’, the change of tube-to-ball diameter ratio will directly affect the crystal stacking structure and contact mode between the spheres, thus affecting the resistivity of the material; the resistivity will also increase gradually with increasing diameter ratio (16). Therefore, the resistivity of a material is closely related to its crystal structure. Here, ruthenium has a hexagonal close-packed structure and its close-packed plane is {0001}. XRD analysis showed that the ruthenium targets grew preferentially along the (0002) plane, indicating that the proportion of the grains on the close-packed plane was higher than that of the standard. Therefore, when the density of the ruthenium targets was nearly close to the theoretical density, its resistivity was close to or less than the theoretical resistivity. According to the formation mechanism of resistivity, the resistivity of a metal is proportional to the average collision frequency; thus, the average collision frequency is only related to the electron-scattering mechanism. The resistivity of the ruthenium targets results mainly from a linear superposition of the resistivity generated by phonon scattering, impurity scattering, defect (dislocation) scattering and interface scattering. Since the test samples were ruthenium targets with a purity of 99.95 wt% at the same temperature, the resistivity generated by phonon scattering and impurity scattering will not be discussed here. According to the morphology analysis, the grain size of 1^#Ru target is uniform and fine, and the interface spacing is relatively consistent, so that the scattering effect of the interface was relatively smaller during the movement of electrons. In addition, the two ruthenium targets were close to the bulk density and the scattering effect of the defects was very small. The combined effect of these factors made the resistivity of 1^#Ru target lower than that of 2^#Ru target and lower than the theoretical value.

3.3 Microstructure Analysis of Ruthenium Films

The microstructure of ruthenium films is shown in Figure 3. At the initial deposition stage, the ruthenium films formed uniformly distributed circular particles, as shown in Figure 3(a) and 3(b). With the increase of the deposition time, the particles gradually diffused and migrated, their distances were constantly reduced and began to merge to form large-size islands. The blank areas were gradually filled and the island-like circular particles were extruded and connected to form worm-like particles with a series of thermodynamic and kinetic processes such as recrystallisation, particle growth, merging of grain orientation and defects. As the deposition process continued, the worm-like particles grew up gradually, as shown in Figure 3(c)–3(h). The surface roughness of ruthenium films also increased at first and then tends to be constant, as shown in Table III. The particle size of 1^#Ru films was larger than that of 2^#Ru film. Its surface roughness changed from the same as that of the 1^#Ru film at the initial deposition stage to gradually less than that of the 2^#Ru films. Since the deposition process of atoms was controlled by the activation energy of the process, the formation of the film structure would be closely related to the relative substrate temperature (Ts/Tm) and the energy of the deposited atoms (17). The relative substrate temperature of the ruthenium films was less than 0.3 (i.e., Ts/Tm < 0.3), and the sputtering pressure was relatively low, so the crystal zone T-shaped fine fibre structure was formed on the cross-section, as shown in Figure 3(i)–3(j). Moreover, the surface fluctuation of 2^#Ru films was greater than that of 1^#Ru films, as shown in the red line in Figure 3(i)–3(j). This led to the defect density inside the grains being larger. Hence, the density of 2^#Ru films was lower than that of 1^#Ru films.

Figure 4 shows the XRD patterns of different films at 200°C for 1 min. A broadening peak appears near 2θ = 44°, indicating that the film was in an amorphous state, as shown in Figure 3(a) and 3(b). As the deposition time increases to 5 min, the diffraction peak of ruthenium gradually appears, indicating the formation of a crystalline film. The crystal plane texture coefficient TC_hkil was calculated from the peak intensity, as shown in Figure 5. At the initial deposition stage, the films first grew preferentially along the (101) plane, and the preferred orientation degree of the 2^#Ru films (101) was greater than that of the 1^#Ru films. The preferred growth of the (101) plane gradually weakened while the preferred growth of the (100) plane gradually increased. According to Wulff theory, the crystal plane with high surface energy that could reduce the total surface energy would grow preferentially and gradually be covered over time, thus showing the crystal plane with the lowest surface energy (17). Hence, the two ruthenium films showed a composite growth mode of first layer and then island (i.e. Stranski-Krastanov type). The distances of most ruthenium film crystal plane were larger than that of the standard sample, showing tensile stress, as shown in Figure 6, indicating that the growth of the films was relatively loose.

To further analyse, Williamson-Hall method was used to calculate the average grain size and micro-strain of ruthenium films deposited by Gaussian function n = 2 with NBS Silcon-2 sample as reference. The results are shown in Figure 7. As the films did not form a crystalline state at 1 min, further comparison will not be made here. At 200°C, the grain size and micro-strain of the films show the same trend with time. From 5 min to 10 min, the grain size of 2^#Ru films increases with increasing micro-strain, mainly because the energy obtained by the deposited atoms at this temperature promotes diffusion, which causes grain growth. The different grains also squeeze and compete for growth, which also increases the micro-strain. For the deposition process of 1^#Ru films at different times and 2^#Ru films at 10 min to 15 min, the grain size tends to decrease with decreasing micro-strain. The nucleation and growth behaviours compete with each other. When the energy required for further growth gradually exceeds that for embryo nucleation, the nucleation rate increases and the grain size decreases gradually. The weakening trend of grain growth reduces the micro-strain, so there is a trend of simultaneous reduction of micro-strain and grain size (18, 19). It is worth noting that the micro-strain of 2^#Ru films was always greater than that of 1^#Ru films.

The films’ thickness gradually increases with time (Figure 8). The thickness of 1^#Ru films is greater than that of 2^#Ru films. The average deposition rate of the two ruthenium films was calculated according to the average deposition rate every 5 min. The average deposition rates of 1^#Ru and 2^#Ru films were 11.20 nm min^–1 and 9.35 nm min^–1, respectively. The faster the deposition rate of the films, the smaller the critical nucleation radius and critical nucleation free energy of the films, thus forming a finer film structure (17). Hence, the average grain size of 1^#Ru films was smaller than that of 2^#Ru films, consistent with the grain size analysis results in Figure 7.

3.4 Deposition Time on Electrical Properties of Ruthenium Films

The resistivity of the two ruthenium films with time is shown in Figure 9. With increasing deposition time, more atoms are deposited on the substrate, lattice defects are reduced, the densities of the films increase, the charge transmission is less impeded, the resistivity of the films gradually decreases and tends to become constant. The resistivity of 1^#Ru films was lower than that of 2^#Ru films; however, the gap between the resistivity of the two films gradually reduced with time. At the initial stage, the resistivity inhomogeneity of 1^#Ru films was larger than that of 2^#Ru films, but the resistivity inhomogeneity of the films decreased and tended to be constant with time.

4. Discussion and Inference

4.1 Influence of Ruthenium Targets on Crystal Structure of Ruthenium Films

When high frequency voltage was loaded on the cathode target, electrons e and Ar⁺ in the plasma region bombard the surface of ruthenium targets in turn under a high frequency electric field. When electrons and ions bombard the surface of the targets at the same speed, atoms of smaller grain are easier to bombard out of the targets and deposit on the substrate than atoms of larger grain, making the initial particles that deposit on the substrate relatively small. However, atoms of larger grain are subjected to more multi-stage collisions of electrons and ions in the plasma during the deposition process to reach the substrate surface due to obtaining smaller energy. Therefore, the deposition rate of 2^#Ru films was relatively slow.

According to nucleation thermodynamics theory, under non-spontaneous conditions, deposited atoms with lower energy fluctuations tend to gather to form an embryo with relatively uniform size larger than the nucleation size. However, the energy fluctuation of the deposited atoms produced from 2^#Ru target with uneven microstructure was large during sputtering deposition, resulting in a slow nucleation rate. In addition, atoms with larger grain had lower kinetic energy when they reached the substrate. Their diffusion ability was limited after collision with the substrate atoms. Ultimately, this led to the slow rate of film nucleation, growth and recrystallisation and the overall change trend lagged behind that of 1^#Ru target. In addition, the films deposited by the 2^#Ru target had poor crystallinity, large average grain size and surface roughness. Therefore, the microstructure of the targets directly affects the average deposition rate, average grain size, surface roughness, crystallisation and the growth rate of nucleation, growth and recrystallisation of the films at the initial deposition stage.

According to the XRD analysis results, the ruthenium targets grew preferentially along the (0002) plane and the preferred orientation of 2^#Ru targets along the (0002) plane was better than that of 1^#Ru targets. However, ruthenium films grew preferentially along the (101) plane and the preferred orientation of 2^#Ru film along the (101) plane is better than that of 1^#Ru films. Therefore, there is no corresponding relationship between the crystal orientation of the targets and that of the films.

4.2 Influence of Ruthenium Targets on Electrical Properties of Ruthenium Films

The film resistivity depends largely on the motion of charge carriers (mainly electrons in this paper) (20). It is mainly affected by phonon scattering caused by thermal vibration of the lattice, interface or surface scattering of electrons, scattering of defects or impurities and other factors (21). Comparing the resistivity of the two films, the resistivity of 2^#Ru films was higher at the initial deposition stage, gradually decreasing with time close to that of 1^#Ru films. The main reasons were as follows: first, at the initial deposition stage, 2^#Ru target with non-uniform grain size meant that relatively few atoms deposited on the substrate and the crystallisation performance was poor, resulting in defect scattering of the films more strongly than that of the 1^#Ru films. The average deposition rate of 2^#Ru films was less than that of 1^#Ru films, the density was lower and the defect density was higher, resulting in strong defect scattering of this film. Secondly, with increasing sputtering time, the average grain size of the films deposited by 2^#Ru target gradually increased to greater than the electron mean free path of ruthenium at room temperature, λ_RT = 6.7 ± 0.3 nm, while that of the films deposited by 1^#Ru target is less than λ_RT, resulting in smaller grain boundary scattering of 2^#Ru films (22). Thirdly, the surface roughness of the films deposited by 2^#Ru target increased with time and the surface scattering of electrons increased. When the average grain size of the films was close to λ_RT, the scattering probability of the grain boundary and surface is gradually enhanced, but the defect scattering effect still exists. Finally, the resistivity of 1^#Ru films was lower than that of 2^#Ru film, but the gap between the resistivity of the two films gradually reduced with time. In addition, the resistivity uniformity of 1^#Ru films was worse than that of 2^#Ru film, especially in the early deposition stage. It can be seen from Section 3.2 in this paper that the resistivity of 1^#Ru target was lower than that of 2^#Ru target, but the uniformity is worse than that of 2#Ru target. Therefore, the resistivity of the targets directly affects the resistivity of the films and is positively correlated. The resistivity uniformity of targets only affects that of films deposited at the initial stage.

5. Conclusions

Ruthenium films presented the Stranski-Krastanov growth mode and formed a crystal zone T-shaped fine fibre on cross-section as prepared. With increasing deposition time, the crystallisation of the films increased gradually and the resistivity of the films decreased gradually. Parameters of ruthenium films, such as the average deposition rate, surface roughness, crystallisation and growth rate of nucleation, growth and recrystallisation of ruthenium films at the initial stage were directly related to the homogeneity of the microstructure of ruthenium targets. The more uniform and finer the microstructure of ruthenium targets, the faster the average deposition rate and growth rate of ruthenium films, the smaller the surface roughness and the better the crystallisation performance. However, no correlation was found between the crystal orientations of the films and the targets. Moreover, the resistivity of the ruthenium films was positively correlated with that of the ruthenium targets. The resistivity uniformity of targets only affects that of films deposited at the initial stage.