Extracting mechanical and microstructural properties of Cu–Zr thin film alloys by MEMS, AFM and ellipsometer

The quantification of the atomic concentration ratios of thin-film metallic alloys having low atomic ordering is challenging, particularly if they are grown on similar metals and possess different surface chemistries. Micromechanical and optical methods have been used to correlate the elemental ratios with the mechanical and optical properties of the films. The room-temperature growth of Cu–Zn thin-film alloys with varying elemental ratios on cosputtered Si substrates was performed to obtain an amorphous film structure. X-ray diffraction patterns confirmed that the grown films exhibited a very short range ordering, suggesting an amorphous structure. The mechanical properties of the films evaluated using microelectromechanical system (MEMS) indicated that the alloy films with moderate Zr concentrations had lower surface stress compared to those with low and high Zr concentrations. Furthermore, spectroscopic ellipsometry was employed to qualitatively assess the relaxation times of free carriers. The results demonstrated a strong correlation between the relaxation times and surface roughness measurements, showing that the microstructure and resistivity characteristics of the alloys align with the Nordheim semiempirical model. The extinction coefficient of the binary alloy film linearly depends on the metallic bulk concentration ratio in a specific metallic ratio range, paving the way for realizing qualitative elemental percentage assessment in the field of metrology.


Introduction
Since the pioneering discovery of Au-Si alloys via the rapid solidification of a liquid in 1959, bulk metallic glasses (BMGs) and their applications have attracted considerable interest.This is primarily due to their exceptional mechanical properties such as high strength and superelasticity [1].BMGs offer a desirable combination of physical and chemical characteristics, including high hardness and corrosion resistance and excellent thermal properties [2].Metallic glasses (MGs) have gained popularity as a distinct material owing to several reasons.First, they exhibit typical metallic properties such as high electrical and heat conductivity.Second, they possess superior mechanical properties compared to their crystalline counterparts.Third, owing to the softening effect observed in the supercooled liquid region above their glass transition temperature, they exhibit self-healing capabilities [3][4][5][6].The atomic structure and the relation between structure and properties in case of MGs have recently garnered significant attention.These materials have been studied for over 40 years owing to their intriguing combination of metallic characteristics (such as electron and heat conductivity and ductility) and glass-like properties (such as hardness) [7,8].The use of MGs (amorphous metallic alloys) as structural materials has increased owing to their inherent toughness, high yield strength, hardness, and magnetic permeability, large elastic strain, temperature-independent electrical resistivity, low wear and corrosion resistance, as well as excellent surface finish and biocompatibility [5,6,9,10].These attributes originate from their disordered atomic structure, which lacks the grain boundaries associated with crystalline alloys having well-defined atomic lattice arrangements [11][12][13].Despite considerable advancements, synthesizing MGs with desirable properties has remained challenging, making it difficult to determine their glass-forming ability [14].The continuous development of BMGs and the unique properties of amorphous alloys have led to the discovery of thin-film MGs (TFMGs) [1].TFMGs exhibit improved plasticity and fatigue resistance compared to their bulk counterparts [15,16].Among various thin films, nanocrystalline alloys and MGs can be easily distinguished [17].Thin films can be deposited with a wide range of different compositions [1].In 1977, Leamy and Dirks employed a sputtering technique to produce primary MGs in the form of thin films [18].Various physical vapor deposition (PVD) processes such as pulsed laser [19], arc plasma [20], and sputtering [21] have been proposed for the generation of amorphous alloys.The development of new materials is crucial for the advancement of science and technology.Amorphous alloys, also known as MGs, have been extensively studied in the fields of physics, chemistry, and metallurgy [22].MGs comprise two or more elements that form metallic bonds and exhibit disordered atomic structures unlike crystalline metals [23,24].TFMGs offer versatility and find applications in various fields, including biomedicine, nanodevices, micro-and optoelectronics, and military components.For example, Mg-Cu TFMGs are suitable for biomedical applications owing to their reduced weight [25].Zr-based TFMGs can enhance the sharpness of surgical blades or dental components in medical devices.The performance of liquid-crystal displays can be improved using a Zr-Cu-based TFMG/ITO bilayer instead of a standard indium tin oxide transparent electrode [26].Zr-Cu alloys have recently attracted interest owing to their BMG properties [23,[27][28][29][30], mechanical characteristics [31][32][33], and superconductivity properties [34].Atomic loyalty modeling of this binary system with Cu as the center atom and Zr as the loyalty atom demonstrated that it likely forms an amorphous structure [35].Another molecular dynamics study performed by An et al investigated its homogenous liquid-(L) and heterogenous solid-like (G) phases with certain Cu-to-Zr concentration ratios; they found that the L-phase is homogeneous at the atomic scale, whereas the Zr-Cu G-phase has locally ordered core regions with an icosahedral short-range order [36].Moreover, Zr-Cu systems have been used as a liquid anode in the electrochemical refining of Zr owing to their low melting points, and a Zr-Cu system with Cr was produced using cold-rolling for yield strength enhancement [37].In addition, TFMGs, which lack grain boundaries found in crystalline materials, can be utilized in microelectromechanical systems (MEMS) by employing Zr-Cu-Al coatings [38].
New methods and techniques are highly required for determining the compositions or weight fractions of thin films; however, research in material science that explores new methods with optimal results are lacking.A commonly used method for detecting the weight fraction of a specimen is energy-dispersive x-ray spectroscopy (EDX), which measures the elemental composition of a sample.However, EDX is expensive and not easily portable and requires skilled personnel to operate [39].Therefore, developing new techniques for determining the weight fractions of Zr-Cu specimens is an urgent requirement.In this research, we investigated methods for identifying the compositions or weight fractions of thin-film amorphous alloys.We propose a novel approach utilizing micromechanical sensors (such as MEMS-based sensors), which have been extensively used in a wide range of applications across different fields owing to their small size, fast response, and high sensitivity, and ellipsometry to determine the compositions of binary Zr-Cu alloys [40,41].

Magnetron sputtering deposition
Seven thin-film samples with varying compositions were deposited, including pure Zr, pure Cu, Zr (82.59 wt%)-Cu (17.41 wt%), Zr (69.03 wt%)-Cu (30.97 wt%), Zr (53.89 wt%)-Cu (46.11 wt%), Zr (39.43 wt%)-Cu (60.57wt%), and Zr (24.78 wt%)-Cu (75.22 wt%).The deposition process took place at room temperature in a chamber equipped with two magnetron cathodes.Zr-Cu thin-film alloys with different compositions were grown on silicon substrates (intrinsic, oriented (100), thickness: 1892-29 nm) using DC or pulsed DC sputtering of metallic zirconium and copper targets.The films were deposited in a confocal configuration without external heating.The deposition pressure was set at 0.3 Pa, and the target-to-substrate distance was 9 cm.A discharge current of 0.3 A was applied to the Zr target, whereas the discharge current applied to the Cu target varied between 0 and 0.21 A depending on the desired composition.The PVD cosputtering technology proved efficient for depositing Zr-Cu thin-film MG across a wide composition range.

Picomeasure system (PM3)
The PM3 system (Fourien Inc., Edmonton Canada) was employed to perform measurements in the static and dynamic modules of MEMS, particularly for vibrating elements such as microcantilevers.The system has been discussed in detail elsewhere [42,43] 2.2.1.Microcantilever chips Octo 500 S/Si microcantilever chips were obtained from Micromotive (Germany).Each chip comprised an array of eight microcantilevers with specific dimensions (length: 500 ± 5 μm, width: 90 ± 2 μm, and thickness: 1 ± 0.3 μm).Prior to performing the measurements, the samples were immersed in ethanol for 5 min and then rinsed with deionized water to remove any contaminants.

Results and discussion
Herein, we aimed to develop methods for detecting different compositions or weight fractions of thin-film amorphous alloys.We focused on developing a microelectromechanical-based sensing system for surface composition detection, and two experimental setups were employed for this.In the first setup, a Si wafer substrate coated with thin films was employed, where the effects of different compositions on the mechanical, structural, and optical properties of the films were examined before and after heating.The second setup utilized Si microcantilevers coated with the same thin films, and the deflections resulting from these coated microcantilevers were studied before and after heating.Deflection was performed using the PM3 system, along with various techniques such as x-ray diffraction (XRD), atomic force microscopy (AFM), and spectroscopic ellipsometry, which was used to analyze the optical properties.
Three different cantilevers were selected from each chip for testing, and each cantilever was tested for 1 h (continuous mode for static measurements and taken at 5-min intervals for the dynamic mode).

Static-mode measurements before heating
Figure 1 illustrates the average surface stress behaviors of the microcantilevers.It was observed that the surface stress behavior remained relatively consistent at 10, 30, and 50 min with minor variations.At 10 min, an increase in surface stress was observed between pure Cu and Zr (24.78 wt%)-Cu (75.22 wt%), while a considerable decrease was observed for Zr (39.43 wt%)-Cu (60.57wt%).Subsequently, stability in surface stress was observed between Zr (39.43 wt%)-Cu (60.57wt%) and Zr (53.89 wt%)-Cu (46.11 wt%).The surface stress increased for Zr (69.03 wt%)-Cu (30.97 wt%) and decreased for Zr (82.59 wt%)-Cu (17.41 wt%).At 30 min, the surface stress behavior was similar to that observed at 10 min, except for near stability between pure Cu and Zr (24.78 wt%)-Cu (75.22 wt%).At 50 min, the surface stress behavior was comparable to those at 10 and 30 min, with near stability observed for the alloys with 24.78 wt% Zr.In addition, the surface stress decreased between pure Cu and Zr (24.78 wt%)-Cu (75.22 wt%) and between Zr (39.43 wt%)-Cu (60.57wt%) and Zr (53.89 wt%)-Cu (46.11 wt%).These behaviors were consistent at different time points, and deflection occurred due to surface stress variations at the nanoscale level.Overall, the surface stress decreased as the Zr content increased.This trend is supported by the AFM results shown in figure 2, where roughness decreased with increasing Zr content, indicating the role of roughness in surface stress behavior.The high magnitudes of the roughness of pure Cu and Cu-rich alloy films explain the high variations of error bars observed in these samples in figure 1.This is because high roughness contributes more to the misalignment of the laser beam in the measurements of cantilever bending used to extract the surface stress.A similar case is considered for higher Zr contents and pure Zr because the error bars correlatively vary with the surface roughness extracted using AFM.
The general trend that the addition of Zr (HCP) to Cu (FCC) created smoother surfaces as Zr content increased has been observed in other alloys.Feng et al [44] reported that the grain size and surface roughness decreased with increasing Zr content (x) for (CrTaNbMoV)Zrx.They observed structural evolution from BCC to an amorphous structure, whereas in our study, the films evolved from FCC (Cu) to an amorphous structure.Both studies demonstrated the effects of adding an HPC crystal to a cubic structure, resulting in order-disorder transformation and reduction in grain size and surface roughness.
Contact-mode AFM was used to study the surface roughness of the Zr-Cu thin-film alloys with different compositions and pure Zr and Cu.Two-and three-dimensional AFM images were collected over a scanning Figure 3 shows the AFM surface texture and topography for all samples deposited on a Si substrate.Figure 3  the increased grain size was evident.As shown in figure 3(g), for the deposited thin film comprising pure Cu, the grain size was larger, and particles of different sizes were observed.
The data obtained using AFM image analysis include root mean square (RMS) roughness values used to describe surface morphology.
As shown in figure 2, pure Cu had the largest roughness value, and it dramatically decreased with increasing Zr content.Pure Zr exhibited large roughness but not as large as that of pure Cu.Notably, the surface roughness in Zr (44.87 wt%)-Cu (55.13 wt%) was slightly different from those observed in the rest of the alloys.

Static-mode measurements after heating
Three alloys with different compositions, Zr (69.03 wt%)-Cu (30.97 wt%), Zr (53.89 wt%)-Cu (46.11 wt%), and Zr (39.43 wt%)-Cu (60.57wt%), were selected for analysis.These alloys were subjected to 200 °C in an autoclave for 30 min.The choice of these concentrations was based on the distinctive behavior observed in Zr (53.89 wt%)-Cu (46.11 wt%) and the samples with lower (Zr (39.43 wt%)-Cu (60.57wt%)) and higher (Zr (69.03 wt%)-Cu (30.97 wt%)) Zr contents.Figure 4 shows the surface stress-Zr percentage relation, and the average deflection was separately measured for each sample to capture the overall changes.The annealing  induced changes in the mechanical properties.Notably, an increase in hardness was observed with increasing Zr content [45].

XRD measurements
Thin-film XRD measurements were performed using a wavelength value (λ) of ∼0.15406 nm.The detector was positioned in the 2θ range 20°-80°to cover all possible peaks originating from the materials.XRD measurements were performed on all samples to examine the effects of different alloy compositions.Figure 5 presents the diffraction patterns of the deposited thin films of pure metals and alloys.
The peaks corresponding to Zr and Cu were observed only for the pure metals, confirming previous findings in the literature [46].Furthermore, when Zr was added to Cu or Cu was added to Zr, the XRD spectra of the Zr-Cu alloy films with different compositions exhibited an amorphous atomic structure [47].In contrast, the pure metals displayed a crystalline atomic structure.The dashed line represents the peak observed in all samples, which may be attributed to Si.

Spectroscopic ellipsometry for optical properties measurements
Several challenges were encountered during electrical resistivity measurements.As an alternative approach, ellipsometry measurements were performed to utilize the Drude model, which is associated with grain size (mean free path) and free carrier absorption.
A one-layer model was employed to fit the ellipsometry experimental data, particularly Psi (Ψ) and Delta (Δ), in the wavelength range 300-2100 nm.The samples had a thickness of ∼200 nm, which is considered quite thick for ellipsometric measurements in metals.Because of the limited light penetration caused by light absorption in metals for this thickness, direct measurement of the thickness was impractical [48].Consequently, we assumed that the material was bulk, and the reflected light solely originated from the metal surface.The optical parameters, namely the refractive index (n) and extinction coefficient (k), were directly extracted using the software and Fresnel equations of polarized light reflection.

Optical parameters
Figure 9 shows the refractive indices (n) extracted from all the samples.Except for a specific range for pure Cu, the refractive indices generally increase with wavelength (λ) over the investigated range.
Figure 10 displays the extinction coefficient (k), which increases with increasing λ.This behavior aligns with the metallic absorption characteristics resulting from the relatively strong interaction between free carriers and light.
Notably, for samples containing Zr within 31.2-76.77at.%, the extinction coefficient at longer wavelengths appears to increase as the Zr percentage increases.Figure 11 demonstrates a linear relation for k at λ = 1969.3nm, which can find practical applications.
The Drude model provides a widely used equation for describing the free carrier concentration in metals or highly doped semiconductors.By considering the Drude equation of the semifree electron model, the Drude absorption coefficient relation can be obtained as follows [49]:  where a represents the absorption coefficient, N represents the concentration of free carriers, l represents the free-space wavelength, q represents the fundamental charge, e  represents the vacuum permittivity, m e represents the conductivity effective mass of the free carrier, c represents the speed of light, n represents the real component of the metal refractive index, and t represents the relaxation time.Figure 12 shows the plot of n * k versus Zr content.
τ is directly linked to the mean free path length (L), and L is constrained by the grain size represented by the surface roughness observed using AFM.By comparing the n * k versus Zr% plot shown in figure 12 with the roughness versus Zr% plot shown in figure 2, it becomes evident that n * k and RMS are inversely correlated (correlation factor = −0.96508).This correlation arises because an increase in grain size (RMS) increases L and consequently τ.Thus, n * k decreases and vice versa.
The fact that pure elements exhibit larger grains and higher τ aligns with the Nordheim semiempirical behavior, indicating the low electric resistivity (high τ) of pure elements compared to their alloys.

Conclusions
In this study, we demonstrated novel approaches to investigate the thin-film metallic alloys of a Zr-Cu binary system.The mechanical method using MEMS indicated the reduction of the surface stress for the thin-film alloys with moderate Zr concentrations.Moreover, a correlation between surface roughness obtained using AFM and the error bars of surface stress was observed.The Drude equation of the semifree electron model explained the relation between the optical measurements of the Zr-Cu alloys and microstructures of the films extracted using AFM.Interestingly, a linear relation was observed between the metallic concentration ratios of the alloys and extinction coefficients in the Zr concentration ratio in the range of 31.34-77.67.This study can pave the way for utilizing such empirical observations for elemental content determination in other binary alloy systems.

Figure 1 .
Figure 1.The average surface stress as a function of the Zr content at three different times 10, 30 and 50 min.

Figure 2 .
Figure 2. RMS values of Zr-Cu alloys with varying composition.

Figure 3 .
Figure 3. (Continued.) (a) shows the structure of pure Zr.As shown in figure 3(b), the Cu content was increased to 23.23 wt%, and an increase in the grain size of Cu was observed.Figure 3(e) shows a clear grain size.As shown in figure 3(f),

Figure 9 .
Figure 9. Refraction indices of alloys and pure metals.

Figure 10 .
Figure 10.Relation between the extinction coefficient (k) and wavelength.

Figure 11 .
Figure 11.Linear relation between the optical extinction coefficient (k) and Zr concentration.

Figure 12 .
Figure 12.Relation between the Zr concentration and (n * k)