Interface Bonding Properties of CrAlSiN-Coated Cemented Carbides Doped with CeO2 and Y2O3 Rare Earth Oxides

This study performed first-principle-based calculations of the interface adhesion work in interface models of three terminal systems: CrAlSiNSi/WC-Co, CrAlSiNN/WC-Co, and CrAlSiNAl/WC-Co. The results proved that the CrAlSiNSi/WC-Co and CrAlSiNAl/WC-Co interface models had the highest and lowest interface adhesion work values (4.312 and 2.536 J·m−2), respectively. Thus, the latter model had the weakest interface bonding property. On this basis, rare earth oxides CeO2 and Y2O3 were doped into the Al terminal model (CrAlSiNAl/WC-Co). Then, doping models of CeO2 and Y2O3 doped on the WC/WC, WC/Co, and CrAlSiNAl/WC-Co interfaces were established. The adhesion work value was calculated for the interfaces in each doping model. When CeO2 and Y2O3 were doped in the WC/WC and CrAlSiNAl/WC-Co interfaces, four doping models were constructed, each model contains interfaces withreduced adhesion work values, indicating deteriorated interface bonding properties. When the WC/Co interface was doped with CeO2 and Y2O3, the interface adhesion work values of the two doping models are both increased, and Y2O3 doping improved the bonding properties of the Al terminal model (CrAlSiNAl/WC-Co) more significantly than CeO2 doping. Next, the charge density difference and the average Mulliken bond population were estimated. The WC/WC and CrAlSiNAl/WC-Co interfaces doped with CeO2 or Y2O3, with decreased adhesion work, exhibited low electron cloud superposition and reduced values of charge transfer, average bond population, and interatomic interaction. When the WC/Co interface was doped with CeO2 or Y2O3, superposition of the atomic charge densities of electron clouds was consistently observed at the CrAlSiNAl/WC-Co interface in the CrAlSiNAl/WC/CeO2/Co and CrAlSiNAl/WC/Y2O3/Co models; the atomic interactions were strong, and the interface bonding strength increased. When the WC/Co interface was doped with Y2O3, the superposition of atomic charge densities and the atomic interactions were stronger than for CeO2 doping. In addition, the average Mulliken bond population and the atomic stability were also higher, and the doping effect was better.


Introduction
CrAlSiN-coated cemented carbide tools are widely known for their high hardness, corrosion resistance, wear resistance, and high-temperature oxidation resistance. These features have made them especially suitable for machining difficult-to-process materials, such as titanium alloys [1][2][3][4]. However, CrAlSiN-coated cemented carbide tools are more commonly associated with interface problems, such as coating disbondment and sticking [5]. Some scholars have doped additives into the coating or matrix to improve the interface properties of the coated cutting tools. For example, Lu et al. [6] prepared a diamond coating with a level of 8000 ppm doped B on the WC-Co cemented carbide matrix and found that, compared with undoped coated cutting tools, the residual stress was lower and the coating-matrix bonding strength was higher. Wang et al. [7] prepared a CrBN coating containing Ni or Cu doping on a 45 steel matrix. It was found that the coating hardness decreased after adding Ni or Cu, but the coating resistance to circular cracks was significantly enhanced due to good bonding properties between coating substrates. Yu et al. [8] utilized a cathodic arc evaporation system to prepare AlTiN coatings with various B/C doping ratios. The indentation method was used to evaluate the bonding strength between the coating and substrate. The results showed that the bonding strength between the coating and substrate was the best when the B/C ratio was 1:1.
Rare earth elements have high chemical activity and low electronegativity [9,10]. Extensive studies have been conducted worldwide on the reinforcement of material performance using rare earth element doping, which has been found to improve the interface bonding strength. Li et al. [11] employed the sol-gel method to prepare WC-10 (Co, x-Re) composite powders with various rhenium content levels. Microstructural analysis showed that adding rhenium resulted in a more regular WC grain shape and more equiaxed grains. In addition, the interface between the Co binder phase and WC was smooth and without pores, and the interface bonding was satisfactory. Liu et al. [12] reported that doping rare earth elements into WC-Co cemented carbides inhibited grain growth, refining the grains and increasing the interface bonding strength. Zou et al. [13] proved that doping with appropriate amounts of rare earth borides LaB 6 and CeB 6 purified the grain and phase boundaries and improved the WC/Co interface wettability, thereby increasing the interface bonding strength. Wang et al. [14] manufactured WC-11Co cemented carbides with and without CeO 2 doping by vacuum hot-pressing sintering. Their study showed that CeO 2 addition significantly improved the cutter's fracture toughness. Wen [15] doped nano-CeO 2 into WC-10Co cemented carbide and reported that CeO 2 addition improved the fracture toughness of the cemented carbide. The reason was that CeO 2 was enriched at the WC grain boundaries and reacted with impurity elements to purify the grain and phase boundaries. As a result, the wettability and bonding strength of the WC/Co interface were improved, finally increasing the grain and phase boundaries' strength and the cemented carbide's fracture toughness. Guo et al. [16] prepared a WC-6Co cemented carbide doped with Y 2 O 3 . During solid-phase sintering, the doped Y 2 O 3 effectively inhibited the growth of WC grains, while Y 2 O 3 located in the WC/Co grain boundary separated WC and Co, thereby inhibiting the dissolution and reprecipitation of WC and increasing the interface bonding strength. Wang et al. [17] doped Y 2 O 3 into the WC-10Co cemented carbide. They found that the Y 2 O 3 particles were pinned to the WC grain and phase boundaries, hindering the diffusion, dissolution, and growth of WC particles and hence refining the grains. Huang [18] used a wet grinding style to study the effects of CeO 2 and Y 2 O 3 doping on WC-10Co cemented carbides. The results showed that CeO 2 and Y 2 O 3 doping increased the fracture toughness of the cemented carbides from 12.8 MPa·m 1/2 before doping to 16.7 MPa·m 1/2 after doping. Yang et al. [19] employed spark plasma sintering (SPS) technology to prepare a WC-8Co-Y 2 O 3 cemented carbide. It was found that Y 2 O 3 addition increased the WC/Co interface strength, thereby improving the hardness and fracture toughness of the cemented carbide. Qin et al. [20] employed solid-liquid doping and SPS technology to prepare a Y 2 O 3 -doped WC-12Co cemented carbide. The results showed that the semicoherent interphase boundary between Y 2 O 3 and WC increased the hardness and fracture toughness of the Y 2 O 3 -doped cemented carbide by 2.1 and 9.2%, respectively, compared with that before doping. Deng et al. [21] employed the in situ synthesis and spray drying process to prepare a CeO 2 -doped WC-10Co cemented carbide. Their study showed that CeO 2 doping decreased the surface energy differences between the crystal planes while increasing the wettability of the interface between the Co binder phase and WC grains. For this reason, the degree of densification, hardness, and toughness of the cemented carbide were improved.
The above brief survey of existing studies in the relevant field reveals that most of them have focused on the performance of WC-Co cemented carbides doped with rare earth oxides and achieved this purpose experimentally. However, few studies have been conducted on the interface performance of CrAlSiN-coated cemented carbides. From an atomic perspective, there still needs to be more investigations into the interface bonding mechanism of CrAlSiN-coated cemented carbides doped with rare element oxides. Therefore, in this paper, based on the first-principles method, from the microscopic atomic point of view, the present study focused on the bonding performance of the coating-matrix interface in the interface models of CrAlSiN-coated WC-Co cemented carbides with different terminal atoms. First, we determined the interface model with the worst interface bonding performance. CeO 2 and Y 2 O 3 were doped into the WC/WC, WC/Co, and CrAlSiN Al /WC-Co interfaces of CrAlSiN-coated cemented carbides. The adhesion work was calculated for each interface model, and the charge density difference and average Mulliken bond population were estimated. On this basis, the influence pattern and the nature of the interface bonding performance of CrAlSiN-coated cemented carbides doped with rare earth oxides were revealed. These research findings are instrumental in the design optimization, popularization, and application of coated cemented carbide cutting tools with improved interface bonding properties.

Parameter Selection for Simulation Analysis
The main research in this paper was carried out in the CASTEP module of the Materials Studio software. Geometric optimization was implemented for all models (primal cells such as WC, Co, CrN, Al, Si, CeO 2 , and Y 2 O 3 and all models before and after doping) based on the first principle and density functional theory. The simulation parameters were chosen as follows.
Based on the Monkhorst-Pack algorithm, the k-point grid was set to 5×5×1. The energy of each unit cell was determined under different cutoff energies, and it was found that the energy of the unit cell tended to converge at the cutoff energy of 400 eV; therefore, Ecut was set to 400 eV. The structure of the original cell was optimized under different exchange correlation functions; we found that when the exchange association function is GGA-PBE, the calculated lattice constants of the optimized cell have the least deviation from the experimental values; therefore, GGA-PBE was selected as the exchange association function. Ultrasoft pseudopotential was chosen to describe the interactions between valence electrons and the nuclei of ions.
The Broyden-Fletcher-Goldfarb-Shanno (BFGS) algorithm was used to optimize the model. The optimization parameters were as follows: the SCF convergence threshold was specified as 1.0 × 10 −5 eV/atom; the maximum interatomic interaction was 0.03 eV/Å; the maximum intracrystalline stress was 0.05 GPa; the maximum atomic displacement was 0.001 Å; and the number of iteration steps was 100 [22].

Parameter Calculation of Interface Bonding Properties
After the geometric optimization of the interface models, simulation calculation was conducted again using the above parameters to obtain the energy on surfaces α and β and the total energy and area of the α/β interface involved in the optimized interface models. Finally, the adhesion work at the interface was estimated using the relevant formula. Adhesion work is an important parameter characterizing interface bonding properties. It is defined as the reversible work required to separate two phases from each other. The higher the adhesion work, the stronger the interatomic bonding at the interface; hence, the stronger the interface bonding properties and the more stable the interface structure. The adhesion work at the interface between α and β can be calculated as follows [23]: where W ad is the adhesion work, J/m 2 ; E α and E β are the energies of surfaces α and β, eV; E α/β is the total energy of the α/β interface system, eV; and A is the area of the interface, Å 2 .

Construction of the WC-Co Cemented Carbide Matrix Model
The WC(0001) crystal face with a W terminal atom is the most stable, and Co can replace the C atom on the WC(0001) surface [24,25]. Considering this, we used the Co atoms to replace the C atoms on the WC(0001) surface with W terminal atoms and added a 20 Å vacuum layer to build the WC-Co model with Co content of 10.4 wt%. As shown in Figure 1, this model was used to approximately represent the cemented carbide matrix YG10.

Construction of the CrAlSiN Coating Model
CrAlSiN and CrAlN grew preferably in the (111) orientation [26]. Al replaced some Cr atoms to become the solid solution in CrN. Si entered the CrAlSiN coating by replacing the Al atoms [27,28]. In addition, the hardness values of the CrAlN and CrAlSiN coatings were higher when Al and Si atoms accounted for 31 and 4.88 wt%, respectively [29,30]. Then, the CrAlN(111) model with Al content of 33 wt%. was built by replacing the Cr atoms in the supercell CrN(111) with Al atoms. Next, Si atoms were used to replace the Al atoms in CrAlN(111), and a 20 Å vacuum layer was added to obtain a CrAlSiN model with Si content of 4.9 wt%. Figure 2 shows a CrAlSiN Al coating model with a vacuum layer.

Construction of the CrAlSiN/WC-Co Models with Different Terminal Atoms
The interface bonding strength between the WC-Co matrix and the CrAlSiN coating directly impacted the usability of CrAlSiN-coated cemented carbides. There are three terminal atoms on the CrAlSiN crystal surface: Si, N, and Al. Next, interface models with different terminal atoms were built using the WC-Co cemented carbide matrix and the CrAlSiN coating. The parameters selected for simulation analysis as above were used for the geometric optimization of the interface models Figure 3 shows the CrAlSiN/WC-Co models with Si, N, and Al terminal atoms after geometric optimization.

Interface Bonding Property Analysis
Based on the abovementioned parameters, the CrAlSiN/WC-Co models with three different terminal atoms after geometric optimization were used to calculate the total energy E α/β on the two surfaces E α and E β and the area of the interface A α/β . The calculated results were substituted into Formula (1) to obtain the adhesion work at the interfaces, as shown in Table 1. , the most unstable model, was doped with rare earth oxides CeO 2 and Y 2 O 3 to build doped models. The adhesion work at this interface was calculated. The charge density difference and the average Mulliken bond population were analyzed. On this basis, we discussed the effects of doping rare earth oxides on the interface bonding properties of CrAlSiN-coated cemented carbides from the perspectives of charge transfer and bonding mode.

Geometric Optimization of the Doped Models
The Al terminal models doped with CeO 2 and Y 2 O 3 were subjected to geometric optimization using the above parameters for simulation analysis to obtain an interface model with a stable structure. Figure 6a The models doped with CeO 2 or Y 2 O 3 all contained several interfaces. For example, in the CrAlSiN Al /WC/CeO 2 /Co model, there were three interfaces: CrAlSiN Al /WC-Co, CeO 2 /Co, and WC/CeO 2 . All interface adhesion work values were calculated, as shown in Table 2, compared with the adhesion work values at the CrAlSiN Al /WC-Co interface before doping. As long as the adhesion work value of any of the multiple interfaces included in the doping model was smaller than that of the CrAlSiN Al /WC-Co interface before doping, CeO 2 or Y 2 O 3 doping decreased the interface bonding properties of the Al terminal model. Only when the adhesion work of all interfaces in the doped model was greater than that of the CrAlSiN Al /WC-Co interface before doping could the doping of CeO 2 or Y 2 O 3 improve the interface bonding performance of the Al terminal model.   As shown in Figure 7, doping CeO 2 or Y 2 O 3 into the WC/WC and CrAlSiN Al /WC-Co interfaces reduced the adhesion work compared to non-doped interfaces-that is, doping impaired the interface bonding properties of the Al terminal model. Compared with the CrAlSiN Al /WC-Co interface adhesion work not doped with rare earth oxides, doping CeO 2 or Y 2 O 3 into the WC/Co interface increased the adhesion work-that is, doping improved the interface bonding properties of the Al terminal model. After doping rare element oxides into the WC/Co interface, further analysis revealed that of the two models, CrAlSiN Al /CeO 2 /WC-Co and CrAlSiN Al /Y 2 O 3 /WC-Co, the adhesion work at the CrAlSiN Al /WC-Co interfaces was 4.216 and 4.297 J·m −2 , respectively. The increase was 1.68 and 1.761 J·m −2 , respectively, compared with those before doping. Compared with CeO 2 doping, Y 2 O 3 doping more significantly improved the bonding properties for the Al terminal model.

Charge Density Difference Analysis
Geometric optimization of the doped models caused charge redistribution among the atoms. The charge density difference maps allowed for the more intuitive observation of interatomic bonding in the system. In addition, the polarity of interatomic bonds could be assessed based on the spatial distribution of charge aggregation and charge transfer. Therefore, the interface bonding properties were characterized by the bonding strength.
The charge density difference was calculated for the doped models after geometric optimization, with the results shown in Figure 8, where the regions with electron loss, gain, and zero transfer are indicated by red, blue, and white colors, respectively.   Figure 8c,d depict the charge density difference maps after doping CeO 2 or Y 2 O 3 into the WC-Co interface. As shown in Figure 8c, the interatomic distance decreased at the CrAlSiN Al /WC-Co interface while the superposition of charge densities was enhanced. This phenomenon was more pronounced in the blue region near the Co atom, indicating the loss of many charges. Charge aggregation was more pronounced near the Cr and Al atoms, resulting in the gain of many charges. As a result, atoms at the CrAlSiN Al /WC-Co interface exhibited strong covalency, and the interface bonding strength increased. At the WC/CeO 2 /Co interface in (c), charge transfer in Ce-W and O-Co atom pairs was significant, accompanied by increased charge density, enhanced interatomic interactions, high covalency, and high interface bonding strength. At the CrAlSiN Al /WC-Co interface in (d), the atomic positions were changed, and charge transfer occurred between Al, Cr, and Co atoms. Interatomic interactions and covalency were strengthened, being manifested as increased bonding strength at the CrAlSiN Al /WC-Co interface. At the WC/Y 2 O 3 /Co interface in (d), the high charge density in the O-W atom pair suggested strong attraction in this atom pair and high interface bonding strength. Figure 8e,f present the charge density difference maps after doping CeO 2 and Y 2 O 3 into the CrAlSiN Al /WC-Co interface, respectively. In the CrAlSiN Al /CeO 2 /WC-Co model, the electron clouds were superposed between Al, O, and Ce atoms, leading to strong interatomic interactions. Charge transfer occurred in the Co-O atom pair at the interface. There were shared charges between Co-O and Ce atoms, accompanied by decreased charge density and weak interatomic interactions. In the CrAlSiN Al /Y 2 O 3 /WC-Co model, O and Y atoms moved away from the CrAlSiN Al coating. There was a greater distance between Al and O atoms, less superposition of electron clouds, and lower interatomic interactions. In contrast, the superposition of electron clouds was greater between the Co atom and O-Y, which meant stronger interatomic interactions. That is, doping CeO 2 or Y 2 O 3 into the CrAlSiN Al /WC-Co interface deteriorated the interface bonding properties of the Al terminal model. The results of the charge density difference analysis agreed with those of the adhesion work analysis.

Mulliken Average Bond Population Analysis
Mulliken bond population (MBP) analysis is a widely used method to calculate atomic charges. The MBP characterizes the interatomic bonding strength. A positive MBP usually indicates covalency; the higher the value, the stronger the covalency and the interatomic interactions. A negative MBP indicates antibonding, and the higher the absolute value, the more unstable the bonding, the smaller the interatomic interactions, and the greater the repulsion. Calculating the Mulliken average bond population (MABP) allows one to analyze the interatomic bonding strength better. The MABPs were calculated at the WC/WC, WC/Co, and CrAlSiN Al /WC-Co interfaces doped with CeO 2 and Y 2 O 3 , as shown in Figures 9-11.  Figure 9 compares the MABPs between different atoms at the WC/WC, WC/Co, and CrAlSiN Al /WC-Co interfaces doped and not doped with CeO 2 . It is easy to see that when the MABP was above zero, doping CeO 2 into the WC/WC and CrAlSiN Al /WC-Co interfaces did not dramatically increase the MABP. In addition, the MABP was generally higher if CeO 2 was doped into the WC-Co interface than if CeO 2 was not doped or doped into the WC/WC and CrAlSiN Al /WC-Co interfaces. Increased MABP values indicated stronger covalency and higher interatomic forces. At negative MABP values, their absolute magnitudes were smallest when CeO 2 was doped into the WC/Co interface, as shown by the comparison between the four data groups. This indicated that doping CeO 2 into the WC-Co interface more significantly reduced interatomic repulsion. Compared with no doping or doping into the WC/WC and CrAlSiN Al /WC-Co interfaces, doping CeO 2 into the WC/Co interface led to higher interatomic stability. In other words, doping CeO 2 into the WC/Co interface improved the interface bonding strength more significantly in the Al terminal model.   Figure 10 compares the MABPs between different atoms after doping Y 2 O 3 into the WC/WC, WC/Co, and CrAlSiN Al /WC-Co interfaces. As shown by the figure, positive MABPs were generally higher after doping Y 2 O 3 into the WC/Co interface than after not doping or doping Y 2 O 3 into other interfaces. This result indicated that doping Y 2 O 3 into the WC/Co interface significantly increased the interatomic forces, strengthening the interatomic bonding. When the MABPs were negative, the absolute values after doping Y 2 O 3 into the WC/Co interface were generally smaller than those after not doping or doping Y 2 O 3 into other interfaces. This result indicated that doping Y 2 O 3 into the WC/Co interface led to higher interatomic stability than not doping or doping into other interfaces. In the meantime, the interatomic repulsion decreased, and the interface bonding properties were improved more significantly in the Al terminal model. Figure 11 compares the MABPs after doping Y 2 O 3 or CeO 2 into the WC/Co interface. When the MABP was positive, doping Y 2 O 3 resulted in a higher MABP than doping CeO 2 . When the MABP was negative, the absolute value after doping Y 2 O 3 was smaller than that after doping CeO 2 . The above results indicated that Y 2 O 3 doping more significantly improved the interatomic bonding and interatomic forces than doping CeO 2 . For this reason, doping Y 2 O 3 into the WC/Co interface more dramatically improved the interface bonding properties of the Al terminal model.
In summary, the charge density difference analysis and MBP analysis results agreed with those of the adhesion work analysis. There is good consistency between these results and those of available experimental studies.
In particular, Wen [15] prepared a CeO 2 -doped WC-10Co cemented carbide using a gas-pressure sintering furnace. The WC grain size and fracture toughness were measured in the cemented carbides doped and not with CeO 2 using a Scanning Electron Microscope (SEM) and the press-in method, respectively. Their results showed that in the CeO 2 -doped cemented carbide, the average WC grain size decreased from 530 to 410 nm after CeO 2 doping. This means that CeO 2 doping refined the WC grains. Moreover, the fracture toughness of the CeO 2 -doped cemented carbide increased from 9.32 to 10.6 MPa·m 1/2 after doping. The reason was that CeO 2 was enriched at the WC grain boundaries, reacting with impurity elements to purify the grain and phase boundaries. As a result, the wettability of the WC/Co interface was enhanced, as was the bonding strength of the WC/Co interface. This result agrees with our finding that doping CeO 2 into the WC/Co interface improved the bonding properties in the Al terminal model.
Wang et al. [17] prepared a Y 2 O 3 -doped WC-10Co cemented carbide using vacuum sintering. The cemented carbide's surface morphology and Rockwell hardness were measured by SEM and the press-in method before and after doping, respectively. The bending strength of the cemented carbides was determined using an electronic universal testing machine. The analysis revealed that Y 2 O 3 doping decreased the WC grain size in the cemented carbide, thus refining the grains. Due to its high affinity, Y 2 O 3 reacted with the impurities at the grain boundaries, transforming the existing form of the impurities and improving the bonding strength of the WC/Co phase. As a result, the hardness of the cemented carbide increased from HRA 92.3 before doping to HRA 94.5 after doping. The bending strength increased from 1988 MPa before doping to 2250 MPa after doping. These results agree with our finding that doping Y 2 O 3 into the WC/Co interface improved the bonding properties in the Al terminal model.
Huang [18] prepared WC-10Co cemented carbides doped with Y 2 O 3 and CeO 2 using a wet grid style. The surface morphology of cemented carbides before and after rare earth oxide doping was observed by SEM. The results showed that the WC grains in the cemented carbides doped with rare earth oxides were more rounded than those in nondoped cemented carbides. This was because the rounded WC grains had a larger contact area with Co, which affected the mechanical performance of the cemented carbides. The fracture toughness of the cemented carbides was determined using the Palmqvist toughness test. The fracture toughness of the Y 2 O 3 -doped cemented carbide (16.7 MPa·m 1/2 ) was found to be larger than that of the CeO 2 -doped cemented carbide (15.2 MPa·m 1/2 ). In addition, the fracture toughness of Y 2 O 3 -or CeO 2 -doped cemented carbides exceeded that of non-doped ones (12.8 MPa·m 1/2 ). These results confirm our finding that doping Y 2 O 3 into the CrAlSiN/WC-Co interface model improved the bonding properties in the Al terminal model.

Conclusions
(1) The adhesion work values were calculated for three interface models with various terminal atoms, namely CrAlSiN Si /WC-Co, CrAlSiN N /WC-Co, and CrAlSiN Al /WC-Co. The analysis showed that the adhesion work was the highest at the CrAlSiN Si /WC-Co interface (4.312 J·m −2 ) and the lowest at the CrAlSiN Al/ WC-Co interface (2.536 J·m −2 ).