Crystallization and Composition of Ni-C/Ti Multilayer with Varied Ni-C Thickness

Abstract

Ni-C/Ti are suitable for the components of neutron supermirrors with high reflectivity because of their excellent optical constant and smoother interfaces compared to Ni/Ti. In this paper, to investigate the mechanism of C doping to the interface, crystallization, and composition of a Ni-C/Ti multilayer with variable Ni-C thickness, four Ni-C/Ti multilayers were prepared by direct current magnetron sputtering, in which the thickness of the Ni-C layers was 1.5 nm, 2.5 nm, 3.5 nm, and 4.5 nm, respectively, and the thickness of the Ti layers was kept at 5 nm. The prepared samples were characterized by XRD, XPS, HRTEM, EDX, and SAED. The XRD and HRTEM results show that Ni-C layers in Ni-C/Ti multilayers translate from amorphous to polycrystal form, with their thickness increasing from 1.5 to 4.5 nm, and the crystallite size in Ni-C layers is equivalent to the layer thickness, respectively. The XPS, SAED, and EDX results illustrate that the enrichment position of C in Ni-C/Ti multilayers evolves from the Ni-C layers to the Ti layers as the respective Ni-C layer thickness increases from 2.5 to 4.5 nm. The enrichment position evolution of C in Ni-C/Ti multilayers could be due to the lower standard Gibbs free energy of TiC (−180.1 KJ/mol) compared with NiTi (−37.3 KJ/mol) and Ni3Ti (−35.9 KJ/mol) at 298 K.

1. Introduction

Neutron-scattering technology has attracted considerable attention in obtaining microscopic information to characterize materials. Neutron reflection is one of the most important scattering techniques, whereby neutrons are transported through a neutron guide coating with a single Ni layer. With the development of the research, it was replaced by an $m$-value neutron supermirror guide, which could yield a gain factor of $m^2$ in neutron flux at the end of the guide. Meanwhile, the application of supermirrors with a large $m$ value could not only improve the neutron flux gain but also allow the realization of beam condensers, such as the anti-trumpet section of the neutron guide [1] and imaging systems such as the Kirkpatrick–Beaz mirror [2].

In the application and development of neutron-focusing systems, J-PARC has developed focusing devices combining the large $m$ supermirrors to realize larger neutron flux [2]. Mühlbauer et al. investigated elliptic neutron guides with $m$ = 3 supermirrors, which proved to be effective in experiments with small samples [3]. Maruyama et al. fabricated Ni-C/Ti supermirrors with $m$ = 3, $m$= 4, and $m$= 6.7 by ion beam sputtering, and they demonstrated that doping carbon atoms to the Ni layer suppressed the interface roughness and led to higher reflectivity than when using Ni/Ti supermirrors [4]. The neutron-focusing supermirrors deposited on the ultra-precision figured substrates were fabricated and proved to be effective [5,6,7]. The measured efficiency and signal-to-noise ratio during the process were dominated by neutron flux for the focusing devices, and the neutron flux was related to the value of m of neutron supermirrors, which is the central component of the focusing devices. In general, neutron supermirrors become indispensable devices for transporting, bending, focusing, and polarizing neutron beams.

The neutron supermirrors are based on bilayer structures with a high neutron scattering length density (SLD) contrast between individual layers. To achieve efficiency and realize high reflectivity, the neutron SLD contrast between layers, the neutron absorption, and the stability of multilayers under high-radiation conditions should be considered when fabricating the neutron supermirror. Ni/Ti multilayers could present good performance in application, and Ni/Ti has become a common material pair in the fabrication of neutron supermirrors. However, Jankowski and Keem found obvious interfacial roughness, which was produced by the appearance of unexpected microstructures in Ni/Ti multilayers [8,9]. Generally, the critical reflectivity is influenced by the microstructures of materials and the interface width between layers [10]. Researchers have proposed three methods to suppress interface roughness: reactive magnetron sputtering, the introduction of spacer material, and co-sputtering using ion beam sputtering.

Kumar et al. modified the layers by reactive sputtering, using Ar₂ and air mixture gas when Ni layers were sputtered. They prepared the $m$ ≈ 3.5 and $m$ ≈ 4 supermirrors with nearly zero stress and proved that the reflectivity was better with a mixture gas flow of 9 sccm [11]. However, the reactive sputtering method requires precise control of the composition and content of the working gas, and the optimal conditions of the working gas vary with the preparation conditions, which brings difficulties to the preparation work. Ay et al. proposed a new solution for the interface improvement of Ni/Ti multilayers. They found that the reflectivity could be improved by adding a spacer Cr layer around 3 Å between the deposition of the Ni layer and Ti layer [12], whereas the addition of spacer materials has a weaker effect on improving the reflectivity performance of the supermirror. By using the ion beam-sputtering method, Hino et al. fabricated the $m$ = 7 Ni-C/Ti supermirror [13].

From past research, we know that the Ni-C co-sputtering fabricating method can increase the difference in neutron SLD, reduce the size of the Ni crystal grain, and suppress the diffusion between layers compared with Ni/Ti multilayers [14,15,16]. Grimmer et al. explained that the occupancy of C atoms in the octahedral interstitial positions of the Ni fcc crystal caused the strengthening of the Ni 200 peak in NiC_x/Ti multilayers [17,18]. Jiang et al. chose the Ni (6 nm)/Ti (7 nm) and Ni-C (6 nm)/Ti (7 nm) multilayers with the same 100 periods to study the interfaces, and they found that Ni-C/Ti multilayers had sharper interfaces than Ni/Ti by HRTEM, and the grain sizes of Ni crystals were reduced [19]. Casanove et al. found that the incorporation of C atoms into the Ni layer can suppress the diffusion effect between layers and smooth the interfaces [20]. Wood et al. gave the optimal atomic ratio of Ni₈₆C₁₄ and Ni₇₂C₂₈, which was based on the method of measuring the neutron reflectivity [16]. Vidal et al. found that the doping of C atoms in the Ni layer and the inserting of hydrogen atoms in the Ti layer by reactive hydrogen sputtering to prepare Ni-C/TiH multilayers showed a better performance in terms of reflectivity [15]. Zhu et al. concluded that the C-doped layer could make the Ni/Ti interfaces smoother by comparing the Ni/Ti multilayers and Ni-C/Ti multilayers through TEM images [21]. In summary, it has been demonstrated that the co-sputtering of the Ni layer with C has strong effects in terms of improving the interface qualities of Ni-C/Ti neutron multilayers. To investigate the mechanism of doping C into the Ni layer, our team conducted comparative research on Ni/Ti and Ni-C/Ti multilayers, and the results were mainly focused on a relatively thick (around 18.5 nm) bilayer thickness [22]. According to the multilayer structure calculation, the thickness of the Ni-C layer should be thinner than 5 nm for fabricating $m$ ≥ 3 supermirrors to meet the demands for neutron intensity.

Borchers et al. reported that the thickness of the Ni crystalline transition is approximately 2 nm [23]. Many studies have shown that C atoms could inhibit Ni crystallization in Ni-C/Ti multilayers. However, research on the mechanism of evolution of C atoms with varied Ni-C layer thicknesses in Ni-C/Ti multilayers has not yet been conducted systematically. It is important to have a better understanding of the mechanism of action of C atoms in Ni-C/Ti multilayers, which can give us a theoretical direction for conducting research on improving the performance of supermirrors. In the previous work [22], Jiao et al. prepared Ni-C/Ti and Ni/Ti multilayers with a specific periodic thickness (d = 18 nm). They found the enrichment position of C atoms with d = 18 nm and concluded that the presence of C doping can improve the interface because C atoms reduced the crystallization of the Ni-C layer and Ti layer. It was found that the C atoms were mainly concentrated at the interface between the Ni-C layer and Ti layer, while a small amount of C was distributed in the Ti layer, which was considered a very interesting phenomenon, but the reason and physical mechanism of this phenomenon were not explained and analyzed. Based on the phenomenon, we extended the research to explore the evolution and physical mechanism of C enrichment positions. Meanwhile, in the neutron supermirrors, as the m value of the supermirror increases, the minimum period thickness in the supermirror becomes very small. The neutron supermirror is a non-periodic multilayer film, which is formed by stacking Ni-C/Ti film pairs with different period thicknesses. To clarify the evolutional behavior of C atoms in Ni-C layers, especially in very thin Ni-C layers below 5 nm thickness, the Ni-C/Ti multilayer films with variable Ni-C layer thickness are selected in this paper, a series of research works were carried out, and they are reported in this paper as follows, which not only gives an in-depth explanation of the evolution mechanism of C atoms but is also crucial to clarify the fabrication of neutron supermirrors with a larger m value.

In this paper, four Ni-C/Ti multilayers with varied thicknesses of the Ni-C layer, from 1.5 to 4.5 nm, were fabricated by a direct current magnetron sputtering machine and characterized by XRD, XPS, SAED, and HRTEM to investigate the mechanism of C-doped Ni-C/Ti multilayers. The XRD and HRTEM results show that the grain size of Ni crystals in the Ni-C layer was almost the same as the thickness of the Ni-C layer. The XPS, EDX, and SAED results illustrate that when the thickness of the Ni-C layer was increased, C atoms tended to bind with Ti atoms to form C-Ti bonds, which could be due to the lower Gibbs free energy of TiC compared with NiTi and Ni₃Ti. The physical models of the multilayers with varied Ni-C layer thicknesses were built to make the process of evolution in the multilayers clearer. To summarize, the results reveal the evolution mechanism of C atoms in the Ni-C/Ti multilayers. C atoms entered the Ti layer and formed TiC crystals, and the number of TiC crystals increased with the Ni-C layer thickness increasing from 1.5 to 4.5 nm. The diffusion between Ni-C and Ti layers was inhibited, and the interface width became smaller.

2. Experimental Details

2.1. Sample Preparation

All the samples in this article were prepared by a domestic high-vacuum direct-current (DC) magnetron sputtering machine, which is equipped with a 2 in (1 in = 25.4 mm) circular DC magnetron sputtering source produced by Lesker Company in the United States. Four circular magnetron sputtering target guns are installed in the sputtering vacuum chamber. On the circumference with the center of the vacuum chamber, each target gun is separated by 90 degrees and installed vertically upward, which can be seen in Figure 1a. The four magnetron sputtering targets are all flat with a diameter of 100 mm. The magnetron sputtering target guns A and C are mainly used in this experiment. Among them, target gun A is a super-strong magnetic sputtering target gun, which is used to deposit the Ni-C layer, and target gun C is used to deposit the Ti layer. Each target gun is equipped with a shielding cylinder to prevent electromagnetic interference between different target guns and mutual pollution between target materials during coating. An independent and programmable mechanical shutter baffle is equipped between each shielding cylinder and the substrate holder. When the substrate holder moves and starts to remain for a set time on the target gun, the shutter opens, and then, the film of the target used for the target gun is deposited on the substrate. Four circular sample trays are installed on the cover of the vacuum chamber. The rotation is controlled by a stepping motor, so that the substrate holder can stay on different target guns under program control. The positioning of the substrate holder is completed by four magnetic sensors, and the positioning accuracy is less than 1 mm. Through the DC motor, the rotation of the substrate holder can be controlled.

During the deposition process, the control of the film thickness was achieved by controlling the residence time of the substrates when facing the sputtering source, as shown in Figure 1b. The principle of the process of depositing thin films by a magnetron sputtering system is shown in Figure 1c. It has a simpler system and is more cost-effective compared to the ion beam sputtering, and the thin film could grow in a low-temperature environment compared to the chemical vapor deposition methods [24]. At the same time, to improve the uniformity of the thin films, the substrates maintained high-speed rotation during the deposition process. Four Ni-C/Ti multilayers with varied thicknesses were fabricated to study the mechanism of evolution of doping C atoms into the Ni layer and the effect of C with different Ni-C layer thicknesses. The Ni-C mixed target used in this article was made up of two equal-area semicircular Ni and C targets, which were spliced and bonded on a round copper plate. We first prepared a Ni-C single-layer film with a thickness of 12.5 nm using the Ni-C target to determine the atomic ratio of the Ni-C layer based on the splicing structure of the target and the sputtering parameters. According to the test results of X-ray photoelectron spectroscopy (XPS), the atomic ratio of Ni-C was about 86:14, which met the ratio we expected. Ni-C will appear below, as a simplified name, instead of Ni₈₆C₁₄ [16]. The design structures of the Ni-C/Ti samples are shown in

Table 1. The design structure of Ni-C/Ti multilayers.

	M1	M2	M3	M4
Ni-C Thickness (tNi-C)	1.5 nm	2.5 nm	3.5 nm	4.5 nm
Ti Thickness (tTi)	5 nm	5 nm	5 nm	5 nm
Number of the Period	30	20	15	15

. All multilayers were deposited on single-sided polished Si (100) substrates. The super-polished silicon wafer was of the dimensions 20 mm × 30 mm. The base pressure before the deposition was better than 1.0 × 10⁻⁴ Pa. Argon as the sputtering gas with a purity of 99.999% was utilized during the process of deposition, and the working pressure was maintained at 0.266 Pa. The samples were deposited at room temperature. The magnetron cathodes were operated while the powers were fitted at 30 W and 40 W for Ni-C and Ti targets, respectively. The effective deposition rates of Ni-C and Ti were 0.047 nm/s and 0.051 nm/s, respectively.

As can be seen from Table 1, we kept the thickness of the Ti layer at 5 nm during the preparation process, and the thickness of the Ni-C layer was changed from 1.5 to 4.5 nm. The total layer number of the sample layers is 60 layers (M1), 40 layers (M2), 30 layers (M3), and 30 layers (M4). After deposition, the multilayers were characterized by grazing incidence X-ray reflectometry (GIXRR) at the Cu Kα radiation (λ = 0.154 nm). The thickness of the Ni-C layer and Ti layer was determined by fitting measured curves with the Bede Refs software (genetic algorithm).

2.2. Characterization

2.2.1. X-ray Diffraction (XRD)

X-ray diffraction (XRD) was performed to conduct the phase analysis of crystallization in layers. XRD was performed on a D8 Advance X-ray diffractometer (Bruker AXS GmbH, Karlsruhe, Germany) using Cu Kα radiation (λ = 0.154 nm) with theta–2theta scanning. The voltage and current during the process of measurement were 40 kV and 40 mA. The scanning range of the detector angle was from 32° to 50°, and the step was 0.019°. The crystallization was analyzed by comparing the angular positions of diffraction lines with the powder diffraction file (PDF) of the International Centre for Diffraction Data (ICDD) in the JADE software.

2.2.2. X-ray Photoelectron Spectroscopy (XPS)

Based on the results of XRD measurement, M1 and M4 were selected for X-ray photoelectron spectroscopy (XPS) measurement to analyze the chemical state of elements in the Ni-C/Ti multilayers. The XPS measurement of the monolayer and multilayers was carried out on a Thermo Fisher Scientific K-Alpha+ spectrometer (Thermo Fisher Scientific, Waltham, MA, USA) with an anode Al Kα source, which was manufactured by Kratos. The samples were irradiated by a 4 keV Argon ion beam. The full and narrow spectrum scanning power of the deep etching area was 120 W and 150 W, respectively. The XPS result in each sample exhibits a wide scanning spectrum for realizing qualitative analysis and a narrow scanning spectrum for analyzing the chemical state. The etching area was approximately 3 mm × 3 mm, and the rate of the etching process was approximately 0.03 nm/s.

2.2.3. Transmission Electron Microscopy

High-resolution TEM (HRTEM), selected area electron diffraction (SAED) measurement, and energy dispersive X-ray spectroscopy (EDX) linear scan were performed on M1, M3, and M4.

HRTEM, SEAD, and EDX linear scanning processes were provided by Materials Analysis Technology Inc. to analyze the crystalline state of elements. The HRTEM analysis was conducted using FEI G2 F20 Tecnai (FEI, Hillsboro, OR, USA) operated at the acceleration voltage of 200 kV. Selected area electron diffraction (SAED) measurements were used to characterize the crystallization of the multilayers, and a large electron beam was used to cover the whole multilayer stack during the measurement. The EDX technique was performed with the same FEI Tecnai G2 F20 instrument used for TEM. A well-focused electron beam (1–2 nm size) fell normally onto the multilayer cross-section and moved perpendicularly to the interfaces, allowing the composition distribution of the chemical elements in the depth of a multilayer structure to be determined.

3. Results and Discussion

3.1. XRD Analysis

We first need to determine and analyze the crystallization and crystal phase to compare the microstructural states in Ni-C and Ti layers with varied Ni-C thicknesses. The XRD was performed on the samples, and the obtained diffraction image is shown in Figure 2a. It can be seen that all four samples have a diffraction line positioned around 37 degrees, and M2, M3, and M4 have a diffraction line positioned near 44.5 degrees. There exist Si 211 and Si 220 at around 33 degrees and 48 degrees from the signal of the Si substrate. We will analyze these two diffraction lines one by one which should correspond to the crystal structures in the Ni-C layer and the Ti layer. The crystal structure in the Ti layer was analyzed first. Figure 2a shows that the diffraction lines in the Ti layers of all samples appeared between Ti 002 at two-theta values of 38.4° (PDF#44-1294) and TiC 111 at two-theta values of 35.9° (PDF#32-1383), which means the diffraction lines in the Ti layer are mainly contributed by Ti 002 and TiC 111. It can be seen as the thickness of the Ni-C layer increased, the diffraction line position in the Ti layer gradually deviated from Ti 002 and was close to the line position of TiC 111.

Since the samples were coated under the same high pressure, even though there is a small number of residual oxygen-water elements in the chamber, due to the tiny content and the consistent working pressure, the residual elements are not the cause of peak position shift. In general, two reasons cause the peak position of crystal Ti to shift from the diffraction line position of pure Ti crystals. One is the existence of residual stress in the thin film, and the other is the doping of other elements in the Ti element. If the shift of peak position was due to the existence of residual stress, the residual stress in the film should reach a few GPa in order to cause a peak shift such as that which appeared in this paper. The compressive stress test was performed on M3, and the stress was 608.7 MPa. Therefore, the stress of the whole film is small, and the stress of the multilayer is determined by the sum of the stress of the Ni-C layer and Ti layer. From the XRD results (Figure 2b), the diffraction line position of Ni did not drift, and it can be referred to that the stress of the Ni-C layer is relatively small; from this point of view, the stress of Ti layer should not be very large. Above all, the peak position in the Ti layer deviating from Ti 002 was not caused by the residual stress. It could be inferred that the C atoms entered the Ti layer and combined with Ti atoms to form TiC crystals during the process. The phenomenon described above became more obvious when the thickness of Ni-C increased and the number of sputtered C atoms in a single layer became larger. With the increasing thickness of the Ni-C layer, more C atoms enter the Ti layer to form the TiC crystal, causing the contribution of TiC 111 reflection to become more compared to the Ti 002 reflection. Therefore, the diffraction line tends to be closed to the TiC 111 line position.

We then analyzed the diffraction line position of the Ni-C layer after analyzing the diffraction line position of the Ti layer. Figure 2a shows that when the thickness of the Ni-C layer was 1.5 nm, there was no Ni crystal detected in the Ni-C layer. As the thickness of the Ni-C layer increased from 2.5 to 4.5 nm, Ni 111 (PDF#04-0850) appeared and gradually became stronger. Meanwhile, no crystal peaks of Ni_xC and Ni_xTi were observed, which is a result consistent with that found by Jiang et al. [19].

As shown in Figure 2b, the results indicate that the crystalline grain sizes of the Ni 111 phase increased with the Ni-C layer thickness, which was demonstrated by the shape of the peaks, and the grain size was calculated based on the Scherrer formula [14]:

$$D_{hlk}=\frac{0.94\lambda }{\mathsf{\Delta }{\theta }_{B}cos{\theta }_B}$$

(1)

Here, $D_{hlk}$, $\lambda$, ${\theta }_B$, and $\mathsf{\Delta }2{\theta }_B$ represent the crystalline sizes, the X-ray wavelength, the Bragg angle, and the FHWM of the peak (which was fitted by the Gauss–Lorentz linear fitting function and obtained in JADE), respectively.

Table 2. The Ni crystal grain sizes in M1–M4.

Sample	2Theta (deg)	FHWM (deg)	Ni-C Thickness (nm)	Crystal Size (nm)
M1	/	/	1.67	0
M2	44.25	3.24	2.42	2.60
M3	44.32	2.44	3.49	3.50
M4	44.31	2.03	4.42	4.40

shows that the crystal grain sizes of Ni in M2-M4 were 2.6 nm, 3.5 nm, and 4.4 nm, respectively. From the calculated results of the grain size, it can be seen that Ni-C grain size increased with the increase in Ni-C film thickness. Comparing the actual Ni-C layer thickness with the Ni-C grain size, the grain size was almost the same as the actual thickness of the Ni-C layer.

The XRD results explain the evolution mechanism of C atoms in the Ni-C layer; it can be confirmed that Ni 111, Ti 002, and TiC 111 exist in the Ni-C/Ti multilayer. The diffraction line of Ni 111 is located at 44.5°, and the position of the diffraction line contributed by Ti 002 and TiC 111 gradually approaches from 37° to 36.5° with the Ni-C layer thickness increasing. The evolution of the diffraction line position of the Ti layer is explained as follows: with the thickness of the Ni-C layer increasing, the content of C increases, and the number of C atoms entering the Ti layer increases, causing the crystallization of TiC 111 to be strengthened. As a result, the contribution of TiC 111 to the diffraction line increases, and the diffraction line approaches the line position of TiC 111 (about 36°). For the Ni 111 crystallization that appears in the Ni-C layer, it can be seen that Ni evolves from an amorphous state to a crystalline state. The grain size of Ni 111 gradually increases with the increase in the thickness of the Ni-C layer, and the grain size remained consistent with the film thickness. Therefore, it can be concluded that C atoms play a delaying role in the crystallization of Ni by entering into the Ti layer to form crystalline TiC. The combination of C and Ti could reduce the interpenetration of elements between the Ni and Ti interfaces. Compared with Ni/Ti multilayers prepared under the same conditions, the doping of C causes Ni crystallization to occur in a thicker Ni-C layer and the grain size to be smaller, being almost the same size as the thickness of the Ni-C layer. Under a similar roughness, the thicker the film, the weaker the effect of roughness on the film. Meanwhile, because of the smaller grain size in the film, the interface is clearer and the roughness is smaller. Therefore, the quality of the Ni-C/Ti multilayer is better than that of the Ni/Ti multilayer under the same preparation conditions.

3.2. XPS Analysis

M1 and M4 were chosen to take XPS measurements to confirm the XRD analysis results and determine the composition of the chemical elements in the multilayer. Firstly, depth etching was carried out to obtain the elemental distribution. Then, the qualitative analysis of the chemical bond of the C element obtained by narrow-area high-resolution scanning was performed. Finally, the quantitative analysis of TiC strength was carried out by calculating the area ratio of C-Ti bonds.

The depth etching process was performed by using Ar⁺ ions to sputter the multilayers. After each etching process, the concentrations of elements were deduced from XPS measurements. The surface layer of samples is Ni-C, and there exist NiO-Ni₂O₃ and Ni(OH)₂ on the surface region due to the oxidation and the interaction of H₂O with NiO [25]. The thicknesses of NiO and Ni(OH)₂ layers on the nickel single crystal samples exposed for long times to oxygen–water mixtures at high pressure are 0.82 nm and 0.34 nm, which is thinner than the thickness of the surface layer, and the hydroxides are very stable reaction products to passivate the surfaces [26]. We chose the position of 18 nm away from the surface layer to start the XPS analysis and did not consider the distribution analysis of the oxygen element. Figure 3 shows the distribution of the element concentration in M1 and M4 with the etching depth changed. As shown in Figure 3a, the thickness of Ni-C in M1 is 1.5 nm, and the contrast between the Ni and Ti layers is not obvious in the results due to the low thickness of the Ni-C layer. However, the concentration of C shows an opposite trend to that of Ti from Figure 3a. It shows that when the thickness of the Ni-C layer is small, C atoms mainly exist in the Ni-C layer. Figure 3b presents the distribution of the element concentration in M4. It shows that as the thickness of Ni-C increases to 4.5 nm, the contrast of the distribution trend between Ni and Ti becomes more obvious. The distribution trend of C in the multilayer changes from being opposite to that of Ti in M1 to have the same distribution trend in M4. The measurement results indicate that as the thickness of Ni-C increases, C atoms are more likely to diffuse into the Ti layer.

The spectrum of C was fitted by Gauss–Lorentz linear fitting by using the weighted least squares method to further analyze the chemical state of C elements. All the peaks were analyzed using the nonlinear Shirley background. Due to the existence of C atoms in the samples, the peak position of Ar 2p3/2 was used to realize the calibration. The identification of the spectral position during the fitting process came from the data of the NIST XPS database and was supported by some references to the literature [26,27,28]. The data in terms of M1 etching at 13.5 min and M4 etching at 13 min were selected to analyze to compare the C chemical state in M1 and M4.

Based on the previous study and the above XRD results, it can be concluded that C and Ni do not form compounds, which is also consistent with the phase diagram of C and Ni. Therefore, in the C spectrum fitting, we only introduced the C-C bond and C-Ti bond. As shown in Figure 4a,b, two components were introduced into M1 and M4 to fit the spectrum: The C-Ti bond with a binding energy of 281.7 eV and the C-C bond of amorphous carbon with a binding energy of 283.4 eV. The two fitting results were in good agreement with the experimental results. The data show that TiC compounds exist in both samples. The calculation shows that the area ratios of C-Ti in M1 and M4 are 59.6% and 76.5%, respectively, which shows that the binding ability of C atoms and Ti atoms in M4 is stronger. The Ni 2p spectra of M1 and M4 were compared, and the results show that the position of Ni 2p3/2 spectra were both at 853.0 eV, which could prove that Ni atoms in M1 and M4 scarcely combine with other atoms.

The XPS measurement results demonstrate the XRD measurement results and analysis. The combination of C and Ti that we inferred in the XRD analysis was confirmed. Meanwhile, the analysis also provides the reason why the peak position of Ti deviates more from Ti 002 with the thickness of Ni-C increasing: more C atoms in the Ni-C layer enter the Ti layer to combine with Ti and form TiC crystals, causing the contribution of TiC 111 reflection to become more compared to the Ti 002.

3.3. TEM Analysis

In TEM analysis, we selected the bright-field (BF) image and dark-field (DF) image to observe the morphology and defects of the thin film, the selected area electron diffraction (SAED) to analyze the crystallization and phase in the sample, and the linear scanning of energy dispersive X-ray spectrometers (EDX) to analyze the chemical composition distribution in the longitudinal direction of the film.

Samples M1, M3, and M4 were subjected to TEM measurements in order to observe the film’s growth morphology and structural defects. High-resolution TEM images in the middle position of multilayers are shown in Figure 5a,d,g. The periodic structure of layers can be seen in all samples, where the dark layers are Ni-C layers and the bright layers are Ti layers. The dark-field images for observing crystals and the defects of layers are shown in Figure 5b,e,h. The results of SAED patterns used to investigate the crystalline state and existing phases for the samples are presented in Figure 5c,f,i. Si 200 diffraction spots were used as a marker to calibrate the camera distance from samples.

Figure 5b shows that no crystals were found in the Ni-C layer of M1, and there exist some isolated crystals in the Ti layer. The inter-diffusion phenomenon exists between the interface layers, which weakens the interface clarity. Figure 5c proves the existence of TiC 111 and/or Ti 002. Ti diffraction rings appear broader and relatively darker due to the weaker scattering intensity of Ti [8]. The black area appearing in the Ni-C layer in Figure 5d indicates the presence of microcrystals in the Ni-C layer of M3, which is proved in the dark-field image of Figure 5e. Figure 5f proves the existence of a Ni 111 crystal of fcc structure in the Ni-C layer of M3. Compared with M1 and M3, Figure 5g,h show that the crystalline state of Ni-C is enhanced in M4. The interface is clearer, and the contrast between the Ni-C layer and the Ti layer is enhanced. From Figure 5c,f,i, it can be proved that with the increasing Ni-C layer thickness, the crystal phase transformation of Ni evolves from an amorphous state to a crystallization state, and the crystallization gradually increased. According to the calculations, the Ni grain size is almost equal to the thickness of the Ni-C layer in M3 and M4, which is in the agreement with the calculated results of the XRD measurement.

An EDX linear scan was conducted on the middle positions of M1, M3, and M4 to confirm the composition distribution of the multilayer. Due to the low sensitivity for light element detection, the relative distribution trend of the chemical element could be determined. The intensity was used to describe the tendency of each element concentration as the ordinate, as shown in Figure 6a–c.

Figure 6a shows that the distribution trend of C in M1 is opposite to that of Ti and is consistent with that of Ni. On the contrary, Figure 6c shows that the distribution trend of C in M4 is opposite to that of Ni element distribution and is consistent with that of Ti element distribution. Figure 6c shows that the peak position of the C element is distributed between the peak value of Ni-C and Ti elements. The distribution of C changes from being mainly distributed in the Ni-C layer to mainly distributed in the Ti layer with the thickness of Ni-C increasing, which reveals the gradual evolution of the C distribution in multilayers.

The TEM analysis results can demonstrate the analysis of XRD and XPS. The results of the dark field image prove that Ni does not crystallize in the Ni-C layer in M1, but with the increase in Ni-C layer thickness, there existed grains in the Ni-C layer and the crystalline state is enhanced. In the SAED results, the existence of Ni 111 in the Ni-C layer and the crystallization of Ti 002 and/or TiC 111 in the Ti layer can be proved, and the calculated grain size is also consistent with the XRD test results. The element distribution results of EDX show the enrichment mechanism of C element with the change in the thickness of the Ni-C layer. The C atoms are enriched from Ni-C layer to Ti layer as the thickness of Ni-C layer increases, which also corroborates that the bonding of Ti-C is enhanced with the increase in Ni-C layer thickness.

3.4. Physical Models

Based on XRD, XPS, and TEM analysis results, we established the Ni-C/Ti multilayer models to show the internal composition structure with varied thickness of the Ni-C layer more clearly, which can be seen in Figure 7. As shown in Figure 7a, the thickness of Ni-C in M1 is 1.5 nm, which is not yet at the critical thickness for the nucleation of the Ni crystal. Ni atoms and C atoms are randomly distributed in the layer, which does not combine into Ni₃C due to the positive standard Gibbs free energy of +50.7 kJ/mol at 298 K [29]. Compared with Ni atoms, C atoms bond more easily with Ti atoms. Therefore, there is a certain amount of TiC crystal in M1. The standard Gibbs free energy of Ni at 298 K is −6.67 KJ/mol, and graphite C is −3.41 KJ/mol [30]. With the increase in Ni-C thickness, Ni atoms are prone to forming a crystalline structure because of the lower standard Gibbs free energy, and C atoms are squeezed to the interface between layers, as shown in Figure 7b,c. Since TiC (−180.1 KJ/mol) [31] has a lower standard Gibbs free energy than NiTi (−37.3 KJ/mol) and Ni₃Ti (−35.9 KJ/mol) at 298 K [32], the C atoms located at the interface tend to diffuse into the Ti layer and combine with Ti atoms to form TiC crystallites. The preferred binding of Ti atoms and C atoms inhibits the binding of Ti atoms and Ni atoms at the interface. From the HRTEM and EDX results of M3, the blurred composition and defects in the Ni-C layer proved the diffusion of C atoms into the Ti layer. There are two possible reasons for this phenomenon. One is that the diffusion of C atoms to the Ti layer is relatively easier, and the other is probably due to the secondary discharge during the film coating process [33], sputtering C atoms into the Ti layer. Since the thickness of the Ti layer is relatively thin, the measurement cannot show whether there is a gradient change with C atoms in the Ti layer. Therefore, with the thickness of the Ni-C layer increasing, the depth quantitative evolution law of C atoms penetrating into the Ti layer cannot be obtained from this experiment. This is also a very interesting research point, which we can study in the future.

As the thickness of the Ni-C layer increases, most of the C atoms in the Ni-C layer diffuse to the Ti layer and combine with Ti atoms. With the thickness of the Ni-C layer increasing, in addition to the increase in the number of sputtered C atoms, the transition of Ni from the amorphous state to the crystalline state has an impact on the diffusion of C atoms into the Ti layer.

4. Conclusions

The experimental results reveal the mechanism of evolution of C atoms in the Ni-C/Ti multilayer with varied thicknesses of the Ni-C layer and explain the physical mechanism of the improved performance when C atoms are doped into the Ni layer. It can be proved that by increasing the thickness of the Ni-C layer, the distribution of C element changes from in the Ni-C layer in M1 to in the Ti layer in M4, combining the measurement results of XRD, XPS, and TEM. The existence of crystalline Ni was not found in M1, and Ni crystals began to appear as the thickness of the Ni-C layer increased. The size of Ni crystal grains increased, which was limited by the thickness of the Ni-C layer, the C-Ti bond increased, and the TiC crystallization in the Ti layer was enhanced.

In this work, we found that the C atoms distribution evolved from the Ni-C layer to Ti layer with the Ni-C layer thickness increasing under the small Ni-C/Ti periodic thickness. Due to the lower standard Gibbs free energy of TiC (−180.1 KJ/mol) compared with NiTi (−37.3 KJ/mol) and Ni3Ti (−35.9 KJ/mol) at 298 K, C atoms are easier to crystallize with Ti than Ni atoms. C atoms move to the Ti layer to bond with Ti atoms, thereby inhibiting the binding of Ti and Ni and reducing the diffusion between Ni and Ti interfaces. The presence of C atoms delays the crystallization of Ni and limits the grain size of Ni crystals, and the interfacial properties of Ni-C/Ti multilayers are improved.

However, the current research in this paper still has some limitations. Firstly, we used the coating method of magnetron sputtering to prepare C doping Ni-C/Ti multilayer films to suppress the interface penetration, but this method cannot suppress the interface roughness very well, so we plan to use the coating method of ion beam sputtering in the future research. Secondly, the Ni-C sputtering target we used in this work is a spliced target with half Ni and half C; we plan to optimize the target in the future and use a mixed Ni-C target to explore whether the change of the target will affect the performance of the film.