Microstructure, Mechanical Properties and First Principles Calculations of Mo/VC Multilayers

Abstract

In the present work, Mo/VC multilayers with constant period (10 nm) and different modulation ratios (from 6:4 to 9:1) were deposited by dc magnetron sputtering. We investigated the microstructure and mechanical properties of Mo/VC multilayers. XRD and TEM results indicated that Mo was a bcc structure and VC was a NaCl B1 (fcc) structure. An HRTEM image revealed the coherent interfaces between the Mo and VC layers. Hardness and toughness were studied using the Nanoindent and Microindent methods, respectively. It was found that the multilayers hardness remained as high as ~22 GPa, while the toughness increased from 2.91 to 4.70 MPa·m^1/2 and the modulation ratio varied from 6:4 to 9:1. The enhancement of both hardness and toughness is attributed to the interfaces and the lamellated structure. The first-principles method was also used to determine the most stable interface model and the formation mechanism of the interface between the Mo and VC layers.

1. Introduction

The toughening of hard coatings has drawn increasing research attention the recent decades. However, the antagonism between hardness and toughness makes it a great challenge to realize the synchronous optimization of such coatings [1,2,3,4]. A number of studies have shown that the nacre layer of the shell adopts hard hydroxyapatite and soft collagen to achieve a toughness that is 3000 times higher than its components, in addition to maintaining a high strength through the multilayer structure [5,6]. This suggests that the soft-hard architectural structure can be considered for toughening hard coatings [7,8,9,10]. Metal-ceramic multilayers seem to be the desired design for the preparation of hard and tough coatings. It was reported that enhanced plastic co-deformation was performed in Al/TiN nanolayered coatings, which provided the possibility of high hardness and improved ductility for nanoscale metal-ceramic multilayers [11]. Yuan et al. [12] synthesized the Ti/TiN multilayer coating and showed that the coating had the maximum hardness and Young’s modulus, while retaining good fracture toughness and crack resistance. Shi et al. [13] found that the hardness of W/VC multilayer coatings with a 10 nm bilayer period were measured to be 28~30 GPa, which were harder than pure VC. The toughness of the coatings increased from 3 to 7.5 MPa·m^1/2 with the W fraction from 60% to 90%. All of the coatings exhibited significant hardness and toughness enhancement. In addition to the layer structure, the interfacial characteristic is another key factor with regard to the toughening of hard coatings [14], which is a benefit for the confinement of dislocation gliding, such as dislocation pile-up near the interface and a single dislocation bowing in layers [15]. When the layer thickness is small enough, dislocations slip across the interface [16] and can facilitate energy dissipation. Thus, complex interfaces, avoiding stress concentration, can redistribute the stress at the crack tip over a larger area.

As a hard high-melting point metal, Mo is considered to be one of the best wear resistant coating materials [17]. While VC can reach a hardness of 25 GPa and has a low coefficient of friction [18], especially at high temperatures when V₂O₅ generates [19], it can be used as a solid lubricant in high-speed cutting. Therefore, Mo/VC multilayers, in which the metal phase offered the toughness and ceramic phase provided the hardness, are proposed as a great potentially protective coating in this paper. The relationship between the hardness and toughness of these multilayers was explored. Moreover, we also focused on the interface structure to explore the interface toughening mechanism.

2. Materials and Methods

2.1. Synthesis

Four different Mo/VC multilayers, of which the modulation ratio ƞ_Mo/VC = l_Mo:l_VC (l_Mo represents the Mo layer thickness, l_VC represents the VC layer thickness) which varied from 6:4 to 9:1 with the fixed modulation period Λ(l_Mo + l_VC) = 10 nm, were deposited onto polished silicon (100) substrates using dual-cathode DC magnetron sputtering in an argon (99.99%) environment at ambient temperature. For all multilayers, the deposition started with the growth of a 30 nm VC buffer layer followed by 57 bilayers of Mo and VC for a total thickness of 600 nm. The buffer VC layer was deposited to prevent the reaction between the Si substrate and Mo. The power for two targets of Mo (Φ50 × δ3 mm) and VC (Φ50 × δ3 mm) were 100 and 200 W, respectively. The substrate-to-target distance is 100 mm. The substrate was biased at −100 V during deposition to improve film densification by ion bombardment of the growing film. The base pressure in the chamber was less than 1 × 10⁻⁵ Pa, and the working pressure was set at 0.5 Pa. The deposition rate for Mo was 11.3 nm/min, while that for VC was 6.7 nm/min. The modulation ratio ƞ_Mo/VC was achieved through computer control of the shutters in front of the Mo and VC targets. The opening time of the shutters was calculated from the deposition rate, as shown in

Table 1. The opening time of the Mo and VC shutters.

	ƞMo/VC = 6:4	ƞMo/VC = 7:3	ƞMo/VC = 8:2	ƞMo/VC = 9:1
Mo shutter/s	32	37	42	48
VC shutter/s	36	27	18	9

. Prior to deposition, all silicon (100) substrates were cleaned in an ultrasonic bath of acetone and methanol sequentially for 30 min, and then ion cleaned at –500 V bias in a 6.5 Pa argon plasma pressure for 15 min to remove surface contaminants.

2.2. Characterization

The crystalline structure of the coatings was determined by grazing incidence X-ray diffraction (GIXRD) using the Cu Kα line. The multilayer structure, bilayer periods and interface details were confirmed by transmission electron microscopy (TEM). Nanoindentation tests were conducted using the Hysitron nanoindentor with a Berkovich diamond tip. The hardness (H) and elastic modulus (E) of the coatings were measured by the continuous stiffness measurement (CSM) method. The micro-indentation method as described by Xia et al. [20] was used to determine coating toughness. In this method, the coating is deposited on the Si substrate through a mask to create a sharp edge. A Vickers indentation was performed near the edge on the uncoated Si substrate. Figure 1 shows the indentation geometry. The coating toughness K_C is calculated by the following equation:

$${\mathrm{K}}_{\mathrm{c}}={\left\{{\mathrm{K}}_{\mathrm{s}}^2{\left[1+0.45\frac{\left(\varnothing \mathrm{b}-\mathrm{a}\right)}{\mathrm{t}}\right]}^2\pm {\left[1.9{\mathsf{\sigma }}_{\mathrm{r}}\sqrt{\mathrm{t}\frac{{\mathrm{E}}_{\mathrm{c}}^{*}}{{\mathrm{E}}_{\mathrm{s}}^{*}}}\right]}^2\right\}}^{1/2}$$

(1)

where K_S is Si substrate toughness, a and b are the lengths of indentation cracks growing into and away from the coating, respectively, ϕ is the slope of a versus b, t is coating thickness, σ_r is residual stress, ${\mathrm{E}}_{\mathrm{c},\mathrm{s}}^{*}$ is the effective elastic modulus given by Ec,s/(1 − ${\mathsf{\nu }}_{\mathrm{c},\mathrm{s}}^2$), and subscript c and s describe the coating and substrate, respectively, and ν is the Poisson ratio. The + and − signs are for tensile and compressive residual stresses, respectively.

A standard Vickers indentation was performed to obtain the indents. Here, 0.5, 1 and 2 N loads were used to make sure the crack propagates into the coating with no delamination, and 20 available indentations were obtained for each load. The associated crack patterns were observed using a scanning electron microscope (SEM). The residual stress was calculated from the Stoney equation via measurements of the wafer curvature by stylus profilometry before and after deposition. It should be emphasized that three replicates for the coatings toughness test were performed in order to establish the accuracy of the results.

2.3. First-Principles Calculations

The interface theoretical calculations presented in this paper were carried out using the Cambridge serial total energy package (CASTEP) code in the framework of the first-principles pseudo-potential method. The generalized gradient approximation (GGA) of Perdew, Burke and Ernzerhof (PBE) [21] was employed as an exchange-correlation function to optimize the geometry of the unit cells of bulk Mo and VC. The K-points were set as 11 × 11 × 9 and 11 × 11 × 11 for Mo and VC unit cells, respectively, as the plane wave cutoff energy of 550 eV was used in all calculations. The calculated lattice parameters for Mo (0.31469 nm) and VC (0.41675 nm) are comparable with the available experimental values [22,23].

A Mo(110)/VC(111) interface was constructed by a supercell approach with periodic boundary conditions, and a vacuum layer with 10 Å was employed to eliminate the periodic interaction between the free surfaces of Mo and VC along the z-axis. K-points were set as 11 × 11 × 1 for the Mo(110)/VC(111) interface. The SCF convergence threshold was set to be 5.0 × 10⁻⁷ eV/atom to obtain the ground state by solving the Kohn-Sham equation [24]. The atoms were relaxed to achieve the minimum total energy of the system to fulfill the geometry optimization by using the Broyden-Fletcher-Goldfard-Shanno (BFGS) algorithm, while the convergence tolerances were 5.0 × 10⁻⁶ eV/atom for energy and 0.01 eV/Å for the maximum force [25].

3. Results and Discussion

3.1. Microstructure

Figure 2 shows the grazing incidence X-ray diffraction (GIXRD) spectra for Mo/VC multilayers with various modulation ratios ƞ_Mo/VC from 6:4 to 9:1. The results indicate that the multilayers are polycrystalline. All peaks of Mo(110), (200) and (211), which are preferred to the bcc structure [26], are expected to be sharper in this 2θ range as the Mo concentration increases. In addition, the B1 NaCl type peaks of VC(111), (200) and (220) [27] are identified at 2θ = 38.1°, 43.4°and 62.8° for multilayers with ƞ_Mo/VC = 6:4 and 7:3. As ƞ_Mo/VC changes to 8:2 and 9:1, the VC peaks are broadening and virtually indistinguishable from the background due to the lower concentration and smaller layer thickness of the VC [13].

The bright-field cross-sectional TEM micrograph of Mo/VC multilayers with modulation ratios η_Mo/VC = 8:2 is displayed in Figure 3a. The multilayers, showing a dense columnar structure, disclosed alternating bright (VC) and dark (Mo) layers in a growth direction. The average thickness of the Mo layer is 7.8 nm, while that of the VC layer is 2.2 nm, consistent with the controlled value. The corresponding electron diffraction pattern of the cross-section sample is shown in Figure 3b. Mo and VC reflections are clearly visible, which matches the typical bcc and fcc (B1) polycrystalline structures, respectively. The lattice spacing of Mo was 0.315 nm and the VC was 0.418 nm, supporting the GIXRD results.

3.2. Mechanical Properties

3.2.1. Hardness and Modulus

Figure 4a,b depict the variation in hardness (H) and elastic modulus (E) as a function of displacement for Mo/VC multilayers with different modulation ratios, respectively. One can see that plateaus occur for the hardness curves in the displacement range from 60–90 nm, which is a 10%–15% coating thickness (600 nm), and is assumed to be the coating hardness with negligible substrate effects, as well as the modulus curves. When the displacement goes beyond 90 nm, the hardness and modulus begin to reduce due to substrate effects.

For the Mo/VC multilayers with modulation ratio from 6:4 to 9:1, as Figure 5 shown, the hardness values were measured to be 22.93, 22.57, 22.47 and 22.02 GPa, While the elastic modulus values were measured to be 261, 254, 249, and 239 GPa.

3.2.2. Toughness

Figure 6 shows the toughness of Mo/VC multilayers determined by the microindentation method. It was noted that as toughness for Mo/VC multilayers increased by 62% from 2.91 to 4.70 MPa·m^1/2 with modulation ratios varying from 6:4 to 9:1, the hardness (in Figure 5) is almost sustained, with only 6% reduced. The trade-off between hardness and toughness was apparent, although the mechanical properties of Mo/VC multilayers have not reached the high level of biomaterial properties. This result is due to a combination of several factors, such as the nano-scale grain size [28], the different modulus between the two neighboring layers [29], and interfaces [30,31,32,33,34,35]. Particularly, coherent interfaces not only improve the hardness of the multilayers by alternating the stress field acting as obstacles slipping, but also enhance the toughness by coherency stress activating the dislocation motion near the interfaces. In general, heterogeneous fcc/bcc coherent interfaces produce higher drag force to block the threading dislocation migration compared with a coherent interface with same crystal structure [36]. It is possible for coherent interfaces to absorb or accommodate dislocations due to stress induction, which also mediates dislocations across the interfaces, thus the multilayer toughness improves [30]. Figure 7 shows the indentation with crack pattern, while

Table 2. Parameters required for toughness of Mo/VC multilayers.

Modulation Ratios	6:4	7:3	8:2	9:1
Thickness/nm	600	600	600	600
Poisson ration	0.26	0.27	0.28	0.29
Residual Stress/GPa	−0.75	−0.64	−0.47	−0.11
Φb − a	1.6	2.6	3.4	4.4

lists the parameters required for toughness calculation.

3.3. Interface Details

3.3.1. Interfacial Structure

One can believe that interfaces play an important role for the mechanical properties as discussed in the preceding paragraphs of this paper. Thus, a high-resolution TEM analysis was carried out to reveal the interfacial structure, as Figure 8 presents. Interfaces between the multilayers exhibit a strong Kurdjumov–Sachs (K–S) orientation relationship in the growth direction: Mo{110}//VC{111}, Mo < 111 >//VC < 110 >. This means that coherent interfaces were formed between Mo and VC. The corresponding regions revealed the coherent interfaces across the layers, asshown in Figure 6b–e.

The Mo/VC interface structure with an Mo{110}//VC{111} orientation relationship was also simulated by a first-principles calculation. As shown in Figure 9, considering two terminations of VC(111) (V- and C-termination) and a three stacking sequence (top-, center- and hollow-site), six different Mo(110)/VC(111) interface geometries were built, where top-, center-, and hollow-site means that the Mo is on the top of the surface atoms, the second layer atoms, and the third layer atoms of the VC layer, respectively.

3.3.2. Work of Adhesion

The work of adhesion W_ad is an important thermodynamic quantity to characterize the interfacial bonding of the interface. It can be calculated as follows [37]:

W_ad = (E_layer,Mo + E_layer,VC − E_Mo/VC)/A

(2)

where E_layer,Mo and E_layer,VC are the total energies of the fully relaxed surface layers, E_Mo/VC is the total energy of the interface model, and A is the area of the interface.

Table 3. Interfacial separation (d₀) and calculated work of adhesion (W_ad) of different interfacial geometries after full relaxation.

Termination	Stacking Site	d0 (Å)	Wad (J·m−2)
C-termination	Top	1.314	5.66
	Center	1.311	9.22
	Hollow	1.250	10.64
V-termination	Top	2.488	3.77
	Center	2.372	5.43
	Hollow	2.318	5.76

summarizes the result of d₀ and calculated W_ad for different interface geometries.

Table 4. Relevant data required for W_ad calculation.

Termination	Stacking Site	Elayer,Mo/eV	Elayer,VC/eV	EMo/VC/eV	A/Å2
C-termination	Top	−17,434.5	−10,816.9	−28,254	7.52
	Center			−28,255.7
	Hollow			−28,256.4
V-termination	Top	−17,434.5	−12,643.4	−30,079.7
	Center			−30,080.5
	Hollow			−30,080.6

lists the relevant data required for W_ad calculation. The W_ad value of C-terminated top-, center- and hollow-site interfacial geometry, which increases with optimal interfacial separation reduction, are 5.66, 9.22 and 10.64 J·m⁻², respectively. The W_ad value of V-terminated top-site, center-site and hollow-site interfacial geometry are 3.77, 5.43 and 5.76 J·m⁻², respectively, which are smaller than those of the C-terminated interfaces. This may be because the interface between the Mo layer and the C-terminated VC layer generates more polar covalent bonds, which appear to be stronger than metallic bonds. In particular, the C-terminated hollow-site interface has the highest value of W_ad (10.64 J·m⁻²) among all the considered interface geometries. Meanwhile, the d₀ values of the C-terminated interfacial geometries are less than that of the V-terminated interfacial geometries. It can be seen that the termination of VC(111) plays a vital role in the interfacial bonding of the Mo(110)/VC(111) interface compared to the stacking sequence, and the Mo-C bonding at the interface is presumably formed during the fabrication of Mo/VC multilayers.

3.3.3. Density of States

Because the C-terminated hollow-site interface has the strongest adhesion, the partial density of states (PDOS) has been calculated to further comprehend the bonding nature of the Mo(110)/VC(111) interface. The plots are displayed in Figure 10. The PDOS curves of the interfacial layers are significantly different from the interior layers. The interfacial Mo layer in Mo(110) shows new peaks around −5, −6.5, and −11.7 eV, which are resonant with the peaks of the interfacial C layer and the second V layer in VC(111). These resonant peaks are caused by the hybridization between the interfacial C-s orbits, C-p orbits and Mo-d orbits. Moreover, the height of the PDOS curve for the interfacial Mo layer is lower than that of the interior atoms. This indicates the strong Mo-C interfacial interaction and the formation of the strong Mo-C covalent bonds.

4. Conclusions

A set of Mo/VC multilayers was deposited onto Si(100) substrates using dual-cathode dc magnetron sputtering. The XRD and TEM results indicated that the multilayers were polycrystalline, as Mo was a bcc structure and VC was a NaCl B1(fcc) structure. The coherent interface along the Mo{110}//VC{111} orientation was also observed. With modulation ratios of Mo/VC multilayers varying from 6:4 to 9:1, hardness was almost sustained, with only a 6% decrease, while toughness increased by 62% from 2.91 to 4.70 MPa·m^1/2 The trade-off between hardness and toughness was determined. We attribute this enhancement to the coherent interfaces. The first-principles calculation result indicated that the C-terminated hollow-site interface was the most possible stacking position for the atoms near the interface. The W_ad value and PDOS results implied that a strong bonding was produced between the Mo and VC.