Annealing-induced changes in wear resistance and nanomechanical properties of CuZr metallic glass thin films

Over recent years, metallic glass thin films (MGTFs) have found extensive applications in advanced micro-engineering systems. Consequently, there is a need to thoroughly assess the nanomechanical and tribological behaviors of MGTFs to optimize the design of efficient components. In this study, we employed the nanoindentation technique in various modes to investigate the elastic heterogeneity, tribological response, and mechanical properties of CuZr amorphous films. Before conducting the mechanical tests, annealing treatments at 500 K and 600 K were performed to create samples with different stored energies. The thermal history analysis revealed that the annealing process reduced the stored energy in the microstructure. Furthermore, the pre-annealing treatment resulted in increased hardness and Young’s modulus of the thin films. Additionally, higher annealing temperatures significantly improved the wear resistance of the MGTFs. Observing the serration dynamics in the scratching test, we noticed that the annealing treatment induced larger shear bands on the wear track side. Moreover, the increase in annealing temperature led to a reduction in elastic heterogeneity, which was consistent with the enthalpy relaxation values in the samples. This suggests that the annealing temperature enhanced the densely packed atomic structure, leading to the stabilization of the thin films.


Introduction
Metallic glass thin films (MGTFs) have garnered significant attention owing to their remarkable properties, including high elastic limit, excellent wear resistance, and superior hardness [1][2][3][4][5].The distinctive combination of chemical and mechanical properties in MGTFs renders them exceptionally well-suited for utilization in micro-engineering systems.Their superior wear resistance and robust adhesion characteristics make them promising candidates for a wide range of applications, particularly in biomedical and aerospace industries.[6,7].Consequently, numerous studies have focused on characterizing the nanomechanical and tribological properties of MGTFs, which are essential for the development of micro-mechanical and biomedical devices [8][9][10][11].For instance, as one of the pioneers in this field, Liu et al [12] investigated the tribological behavior of ZrCuAlNi MG films on using micro-scratch techniques.Their results showed good adhesion and ductility of the film with no evidence of film de-bonding, attributed to the favorable interfacial bonding and low residual stress.Another study [13] explored the use of a diamond-coated pyramidal tip in scratching Ni-Fe amorphous thin film surfaces using an atomic force microscope (AFM).It was found that perpendicular tip face orientation to the scratching direction minimizes protuberances and allows better control of the scratched geometry.Marimuthu et al [14] conducted a study on the performance of multilayer MGTFs on various substrates.Their research revealed that both the layer pattern and modulation ratio played a crucial role in determining the failure behavior of the thin films during nanoscratch tests.Utilizing the co-sputtering technique, Onyeagba et al [15] achieved MGTF coatings with adjustable properties, resulting in high-quality polymorphic films on both stainless steel 316 l and titanium alloy substrates.Notably, these MGTFs display remarkable wear resistance and robust adhesion, positioning them as promising candidates for biomedical surface applications.Another work showed that the addition of nitrogen enhances hardness, wear resistance, and adhesion critical loads, with the best combination achieved in Zr-based TFMGs containing 13.8 at.% nitrogen [16].Pandian et al [17] utilized nanoindentation and finite element method to investigate the nanomechanical properties of amorphous multilayer thin films (ZrCuAlNi/ZrCuAg).They observed that film thickness, modulation ratio, and layer patterns significantly influenced the mechanical behavior of the films.In another investigation [18], a fully amorphous CrCoNbB TFMG was fabricated and characterized, which exhibited higher hardness, lower surface roughness, and lower coefficient of friction (COF) compared to sputter-deposited chromium (Cr) coating.Wang et al [19] reported that the ZrCuNiAl metallic glass thin film (MGTF) with a homogeneous structure exhibited higher hardness and lower stiffness compared to its heterogeneous counterpart.Another study demonstrated that a multilayer strategy improved the wear resistance and reduced friction coefficient, attributed to inhomogeneous shear-banding events during nanoscratching [20].Li et al [21] found that crystalline oxide formation on Ni60Nb40 MGTFs led to improved hardness and Young's modulus at the surface.Mamun et al [22] fabricated TiSiNiNb MGTF and highlighted its higher hardness, lower elastic modulus, and good corrosion resistance, promoting its potential as a conventional bio implant.Additionally, the current density of DC magnetron sputtering influenced the surface morphology and topography of MGTFs, with higher current leading to enhanced nanohardness, surface roughness, and fracture toughness in the samples [23].Cao et al [24] successfully fabricated a novel nanostructured metallic glass thin film (MGTF) and found that its excellent tribological and mechanical properties were attributed to the presence of nanoscale zones with amorphous structure and interface regions possessing lower density and additional free volume compared to the core of grains.Song et al [25] demonstrated that TaNbHfZr MGTFs thicker than 50 nm reduced stress on the Si substrate and inhibited indentation-induced phase transition.Additionally, thicker films exhibited sink-in behavior due to higher diffusivity, while thinner ones showed pile-ups owing to inactive thickness-dependent diffusion.
As reported, the increasing significance of MGTFs in high-tech industries has sparked considerable interest in metallurgical investigations.However, a crucial yet often overlooked factor in evaluating MGTFs is the role of the initial stored energy of the films on their nanomechanical properties and tribological behavior.Changes in the energy state of these glassy films can lead to alterations in the atomic arrangement, subsequently affecting their mechanical response.To address this issue, ZrCu amorphous thin films with varying initial stored energies were fabricated using DC magnetron sputtering, and nanoindentation techniques were employed for materials characterization.This study presents a novel and comprehensive investigation of the annealing effects on Zr 55 Cu 45 MGTFs, utilizing various experimental techniques.By analyzing thermal history-dependent properties, including stored energies, elastic heterogeneity, mechanical properties, and wear resistance, this research provides valuable insights not previously explored in the literature.The findings contribute significantly to understanding the annealing behavior in MGTFs and will aid in the design and optimization of MGTFs for advanced micro-engineering applications.

Experimental procedure
In this work, the MGTFs with nominal composition (at.%) of Zr 55 Cu 45 were deposited on the Si substrate by the magnetron sputtering technique.The choice of the Zr 55 Cu 45 alloying composition was based on several scientific considerations.Firstly, the use of a binary alloy with just two elements simplifies the magnetron sputtering process, making it easier to control the deposition and ensure a consistent film quality.Additionally, this well-known alloying composition allows for direct comparisons with existing works, providing a valuable reference point for the analysis of our results.Finally, the limited number of elements in the alloy enhances the interpretability of elemental heterogeneities within the thin film, enabling a more focused investigation of the structural variations and their influence on the film's properties.In order to produce a thin film with a coating thickness of ∼ 2.5 μm, the process was conducted in a high-vacuum chamber with a target-substrate distance of 70 mm, a pressure of 0.8 Pa, and a power of 60 W. Additionally, a protective environment of high-purity argon was maintained throughout the process.The prepared samples were also annealed at temperature of 500 K and 600 K to change the stored energy in the glassy structures.In this regard, thermal annealing experiments were conducted using a crystal tube furnace under vacuum conditions, and a protective atmosphere of argon gas was employed initially to prevent premature oxidation of the film.The annealing process was carried out at pressures below 10 −4 Pa.The annealing treatment involved considering two specific temperatures, 500 K and 600 K, both below the glass transition temperature.The annealing time was set at 5 min, a deliberate choice to avoid any crystallization and ensure the absence of any atypical thermal behaviors in the MGs.Hereafter, the as-prepared sample and the annealed ones at 500 K and 600 K will be named as S0, S5 and S6, respectively.To analyze the thermal history, the samples were tested through the differential scanning calorimetry (DSC) under the heating/ cooling rate of 40 K min −1 .As documented in other works, higher heating/cooling rate in the DSC experiment leads to the intensification of exothermic peaks in the thermographs, facilitating the measurement of enthalpy of relaxation in the material [26,27].This event is more sensitive in the thin films with more disordered structure.Hence, the cooling/heating rate of 40 K min −1 was considered for this experiment.Figure 1 illustrates the DSC curves for the MGTFs and the magnified relaxation region for each sample.As demonstrated, the enthalpy of relaxation for S0, S5 and S6 is 0.53 k J /m o l , 0.48 kJ mol −1 and 0.45 kJ mol −1 , respectively.Hence, three samples with different initial energies were provided for the subsequent experiments.Furthermore, the information presented in table 1, derived from the DSC curves, indicates that the annealing process below the glass transition temperature (sub-T g annealing) primarily impacts the relaxation enthalpy, while it does not significantly influence other crucial thermal parameters such as crystallization and glass transition temperatures or crystallization enthalpy.The amorphousness and microstructure of samples were also evaluated by the x-ray diffraction (XRD) with a Cu-Kα radiation, high-resolution scanning transmission electron microscopy (STEM).
To measure the nanomechanical properties of MGTFs, the nanoindentation test was carried out with a Berkovich diamond tip upon a load control mode.It should be noted that the calibration of indentation depth and contact area was also conducted prior to the experiment.The indenting process was conducted at room temperature with 1 mNs −1 loading rate and maximum load of 10 mN.To obtain the dynamic response of samples, the indenting process was performed through the nano-DMA mode under the tip frequency of 200 Hz.The details for dynamic characterization of MGs via the nanoindentation test are given in [28][29][30].Afterwards, the wear properties of samples were studied by the scratching test with a 2D-transducer mode of nanoindentation system, in which the precise data was recorded in the lateral and normal directions.In this experiment, the indenter made a lateral movement with a ramping load increment and a speed of 5 μm.s −1 over a length of 300 μm.

Results and discussion
The XRD analysis of MGTF samples is presented in figure 2. As can be seen, the patterns include a diffuse scattering peak without formation of sharp Brag peaks, implying that the thin films are fully amorphous.The STEM images also reveal a distinctive contrast pattern characterized by alternating dark and bright regions at the nanometer scale in the investigated samples of amorphous thin films (See figure 3).Furthermore, with the  application of annealing process and as the annealing temperature rises from 500 K to 600 K, there is a tendency for the extent of heterogeneous contrast to diminish.However, the STEM EDS mappings demonstrate no discernible chemical variation associated with the heterogeneity at the nanoscale.It should be emphasized that high-resolution STEM has been validated as responsive to the local chemical composition and density, as evidenced by the heterogeneous image contrast that reflects a non-uniform distribution of chemistry or density in MGs.As a result, the variation in elasticity observed in our investigation of thin film samples can primarily be attributed to their non-uniform spatial structure at the nanometer scale, which involves fluctuations in the local atomic packing density.Figure 4(a) represents the load-displacement curves of MGTF samples attained from the nanoindentation test.Considering the shape of curves, it is derived that the hardness and elastic modulus of specimens are different in each state.Using the Oliver-Pharr technique, it is possible to accurately measure the values of hardness (H) and of Young's modulus (E) from the curves of indenting test [31,32], as given in figure 4(b).The results showed that the E values of S0, S5 and S6 are 91.2,96.1 and 99.3 GPa, while the H values are measured as 4.8, 5.9 and 6.9 GPa, respectively.It is concluded that both of hardness and elastic modulus show increasing trends with the rise of annealing temperature; however, the rate of hardness increment is more apparent.In general, it is suggested that the annealing process facilitates the atomic rearrangement to more stable states, leading to the formation densely packed structures and enhancement of angstrom-scale orders [33].Hence, the  E and H values increase with the rise of annealing temperature.Nevertheless, it is worth mentioning that the annealing temperature is not high enough to form nanocrystals in the structure, as confirmed in the XRD results.
A typical three-process of nano-scratching, i.e. topography-scratch-topography procedure, was conducted to characterize the tribological properties of the MGTFs.In the nano-scratch experiment, the profiles of scratch track were recorded in three different states, i.e. before scratch (D pre ), post scratching (D post ) and within scratch (D in-situ ), where the difference between D post and D in-situ is indicative of wear depth [29].Figure 5(a) illustrates the D pre , D in-situ and D post curves for sample S0.It should be noted that the wear depth can be obtained from the difference between the D in-situ and D post values.According to the results, the constant load increment is accompanied with the rise of scratch depth.On the other hand, there exists no linear relation between the scratch distance and the scratch depth, which is similar to that reported in other works [29].Figure 5(b) represents the scratch depth as a function of scratch distance for all the specimens.It is detected that the final scratch depth for samples S0, S5 and S6 is 71 nm, 54 nm and 41 nm, respectively, implying that the primary  annealing treatment significantly changes the tribological behavior of thin films.Using following equations, it is possible to quantitatively analyze the wear volume and wear resistance of samples, respectively [29,34]: Where x and θ are the scratch distance and the half-included angle of the indenter, respectively.Moreover, the wear depth at the scratch length of x is defined by h x .As given in figure 5(c), the wear volume for S0, S5 and S6 is 31, 26 and 23 μm 3 , while the wear resistance stands at 2.4 ´10 8 , 2.9 ´10 8 and 3.2 ´10 8 Pa, respectively.Previous studies showed that there was a correlation between the hardness and the wear resistance of thin films, defined by Archard-type equation [35]: In which K is a dimensional wear coefficient.Numerous works have indicated that there is linear correlation between the hardness and wear resistance in a wide range of metallic glasses with different alloying systems [29,36].Hence, it is concluded that the tribological behavior is similar in the MGTFs so that a same mechanism for the removal of material in the test can be identified.Some researchers have proposed that the inhomogeneous plasticity of MGs, mainly caused by the localized shear bands, plays a crucial role in the wear properties [37,38].For instance, Lahiri et al [39] did some tribological experimentations and reported that the propagation of shear bands in the wear track leads to creation of some serrations or fluctuations in the lateral force.These serrations are indicative of inhomogeneous plastic deformation and appearance of shear bands on the MG surfaces, which is also seen in our work (See figure 5(d)).Moreover, the lateral force-time curves showed that the serrations become sharper with the increase of annealing temperature, meaning that the higher hardness/wear resistance in the treated samples is accompanied with the formation of bigger shear bands in the structure.On the other hand, the weakening of serrations in sample S0 implied that the plastic deformation at the wear sides is consistent with the generation of fine shear events, which is consistent with a more homogeneous plastic deformation in this state.
Figure 6(a) shows the COF as a function of scratching time.As can be seen, after some sharp fluctuations in the initial state, the scratching curves stabilize, indicating a steady-state behavior.Moreover, it is observed that the intensity of serrations in the steady state of curves is similar in the samples, suggesting that the surface roughness is comparable among the different specimens.However, a significant finding is the substantial improvement in wear resistance for the annealed metallic glass samples.This improvement can be attributed to the structural changes induced by the annealing process.During annealing, the high-energy defects and atomic rearrangements present in the as-deposited samples are relaxed, leading to a more ordered and dense atomic configuration.As a result, the annealed samples exhibit improved resistance to plastic deformation and a reduced propensity for crack initiation and propagation during scratching, leading to lower wear rates and COF values.The reduced COF is indicative of the reduced frictional forces and improved lubricity, highlighting the significance of the structural modifications achieved through annealing in enhancing the wear performance of the metallic glass thin films.This enhanced wear resistance is particularly beneficial in applications where low friction and high wear resistance are critical, such as in advanced micro-engineering systems.Furthermore, the results from the nanoindentation test also reveal that the annealing process leads to a decrease in plastic deformation and displacement upon indenting, providing further evidence of the improved mechanical properties of the annealed samples (See figure 6(b)).The reduced plastic deformation indicates a higher hardness and increased elastic modulus in the annealed samples, contributing to their enhanced wear resistance and mechanical integrity, which is consistent with the scratching test and COF results.
To understand the dynamics of wear process, it is necessary to evaluate the shear-proliferating and burst of shear activity, which is manifested by the alteration of lateral force rate, i.e. |dFL/dt|.
in which k and A are scaling exponent and normalization constant, respectively.The parameter 'Sc' defines the cutoff occurred in the slip sizes, which is correlated to the intermittent plastic deformation in the glassy structure [42].As can be seen, samples S5 and S6 show a single trend for the data distribution, while the cumulative probability is in the form of a bimodal pattern in sample S0.By fitting the data distribution, it is possible to estimate the values of Sc and k for each state.For all the samples, the K value is very close to 1.5, implying that the exponent is mainly irrelevant to the thermal history and atomic structure of the MGTFs.On the other hand, the Sc values are estimated to be 0.041, 0.047 and 0.052 for samples S0, S5 and S6, respectively.This result shows that the annealing treatment leads to increment of Sc value as the characteristic parameter for evaluation of microstructure and shear banding in the amorphous films.In other words, with the application of annealing treatment, the serration dynamic is intensified in the system, which is resulted from the propagation of main shear bands in the microstructure [29,42].
The 2D maps of elastic modulus of thin films along with the Gaussian distribution of data are given in figure 8.The window size of analyzed region is 12 μm ´12 μm and the experiment was carried out 5 times for each state.It can be seen that the annealing treatment significantly affects the elastic heterogeneity in the microstructure so that the heterogeneous structure with a wide range of data distribution in sample S0 changes to a microstructure with weaker fluctuations of elastic regions (S5 and S6) and a narrower distribution of elastic modulus.Furthermore, Gaussian distribution of sample S0 shows an asymmetric profile with an extra tail at the right side, which is just fitted by a superimposition of two distinct curves.This bimodal behavior of elastic properties is consistent with the cumulative probability function of sample S0, while it is opposed to the elastic response of S5 and S6, in which the glassy structure shows a simple trend with a unimodal feature.The results also indicate that the rise of annealing temperature markedly increases the mean elastic modulus and reduces the elastic heterogeneity on the MGTF surface.As confirmed in previous works [43][44][45], this event is due to the fact that the annealing process induces the structural relaxation, annihilation of free volumes and contraction of loosely packed regions in the amorphous material.As given in figure 8, the dynamic excitation of MG surfaces results in appearance of structural heterogeneity.This means that the soft or loosely packed regions with lower elastic modulus are embedded and distributed within a rigid and dense glassy matrix [46][47][48].In samples S5 and S6, annealing treatment reduces the loosely packed regions and changes the atomic arrangement to the lower energy states, i.e. short and medium range orders.In this condition, no crystalline phases are formed in the microstructure; however, the densely packed regions with angstrom-scale orders are intensified in the system.On the other hand, sample S0 shows a wide distribution of elastic properties on the surface, meaning that the loosely packed regions are dominant in the structure and randomly distributed in the rigid matrix.Moreover, the bimodal behavior of Gaussian distribution suggests that the glassy structure may include a phase-separated configuration impeding the sharp propagation of shear events and triggering small shear bands under an external loading.

Conclusions
In this work, the tribological and nanomechanical properties of ZrCu MGTFs were studied through the nanoindentation test.Prior to the mechanical tests, the thin films were annealed at 500 K and 600 K to create the samples with different stored energies.Based on the DSC outcomes, the enthalpy of relaxation for samples S0, S5 and S6 were estimated to be 0.53 k J /mol, 0.48 kJ mol −1 and 0.45 kJ mol −1 , respectively.It was also detected that the final scratch depth for samples S0, S5 and S6 is 71 nm, 54 nm and 41 nm, respectively, indicating that the primary annealing treatment significantly changed the tribological behavior of thin films.Using DMA mode of nanoindentation, it was found that the annealing treatment markedly affected the elastic heterogeneity in the microstructure so that the heterogeneous structure with a wide range of elastic properties in sample S0 changed to a microstructure with weaker fluctuations of elastic regions (S5 and S6) and a narrower distribution of elastic modulus.

Figure 1 .
Figure 1.(a) The DSC curves for the MGTFs, (b) magnified relaxation region for each sample.

Figure 2 .
Figure 2. XRD patterns for the MGTF samples with different stored energies.

Figure 3 .
Figure 3. STEM images along with the EDS maps for the samples (a) S0, (b) S5 and (c) S6.

Figure 4 .
Figure 4. (a) The load-displacement curves for the MGTFs and (b) hardness and Young's modulus for each sample.

Figure 5 .
Figure 5. (a) Surface profiles attained for S0 at D pre and D in-situ and D post as a function of the scratch distance, (b) scratch depth as a function of scratch distance for all the samples, (c) wear resistance and wear volume of samples, (d) lateral force of wear test as a function of time.

Figure 6 .
Figure 6.(a) COF as a function of time, (b) a comparison between average COF in the scratching process and d f in the nanoindentation test.

Figure 7 (
a)-(c) represents the lateral force rate as a function of time, while figure 7(d)-(f) illustrates the cumulative probability distribution, which is explained by a squared exponential decay function [40, 41]:

Figure 7 .
Figure 7. (a)-(c) The rate of lateral force as a function of time, (d)-(f) the cumulative probability distribution of slip fluctuation size for samples S0, S5 and S6, respectively.

Table 1 .
Thermal parameters for the samples obtained from the DSC test.