Electrodeposition of Gold Alloys and the Mechanical Properties

Abstract

Strengthening of electrodeposited gold-based materials is achieved by alloying with copper according to the solid solution strengthening mechanism. Composition of the Au–Cu alloys is affected by the applied current density. The mechanical properties are evaluated by micro-compression tests to evaluate the mechanical properties in microscale to take consideration of the sample size effect for applications as microcomponents in MEMS devices. The yield strength reaches 1.15 GPa for the micropillar fabricated from constant current electrodeposited Au–Cu film, and the film is composed of 30.3 at% Cu with an average grain size of 5.3 nm. The yield strength further increases to 1.50 GPa when pulse current electrodeposition method is applied, and the Cu concentration is 36.9 at% with the average grain size at 4.4 nm.

1. Introduction

1.1 Application of Au materials in MEMS devices

In recent years, microelectromechanical system (MEMS) capacitive accelerometers have been developed and used in a variety of consumer electronics for acceleration detection in a range of 1–5 G (1 G = 9.8 m/s²) [1, 2, 3]. For applications in medical and health care fields, accurate sensing with sub-1 G detection is necessary to monitor hardly detectable body motions [4, 5]. To detect such low acceleration in a compact sensor module, various types of MEMS accelerometers based on silicon (Si) bulk micromachining have been reported [6, 7]. In order to suppress the thermal-mechanical noise (i.e., Brownian noise (BN) [8]) for the highly sensitive detection, a large proof mass is required. Limited choices of materials for the proof mass and other movable components in a CMOS-MEMS accelerometer have been a major challenge to reduce the BN, which becomes more critical when the parasitic capacitance is reduced in miniaturized devices. Yamane et al. [9, 10, 11] propose a miniaturized MEMS accelerometer by using a post-CMOS process with electrodeposited Au in the main components, which enables further size reduction of the proof mass and the device footprint without compromising the sensitivity. With the application of electrodeposited Au in MEMS accelerometers [9, 10, 11], a wide range of acceleration from 1 mG to 20 G can be achieved and are expected to be used in monitoring of hardly detectable body motions.

However, mechanical strengths of Au are much lower than the values of other commonly used materials in electronic devices. For instance, the yield strength of Au is 50–200 MPa in its bulk state [12], and the fracture strength of Si is 1–3 GPa [13], which is one order larger than the strength of Au. The low mechanical strength of Au raises concerns on the structure stability when employed as movable microcomponents. In a study on long-term vibration test of microcantilever made of electrodeposited Au, an obvious tip defection is reported after 10⁷ cycles of the vibration [14]. Therefore, strengthening of the Au-based material is necessary to ensure high structure stability for applications in MEMS devices.

1.2 Strengthening mechanisms in electrodeposits

There are four strengthening mechanisms in metallic materials, including work (strain) hardening, grain boundary strengthening, precipitation strengthening, and solid solution strengthening. Except for the work hardening, the other strengthening methods are plausible in the electrodeposits by controlling the electrodeposition conditions. For example, Rashidi et al. [15, 16] report that a finer crystalline grain structure is obtained in the electrodeposited Ni by controlling the electrodeposition parameters such as current density, bath temperature, and additive amount in the aqueous electrolyte. Grain boundary strengthening of the electrodeposited gold is therefore applicable by the grain refinement effect. Classically, the mechanical strength is proportional to inverse square root of the average grain size according to the Hall-Petch Equation [17] given by.

$$\sigma ={\sigma }_0+k\cdot {d_{\mathrm{g}}}^{–0.5}$$E1

where σ₀ is the friction stress in the absence of grain boundaries, k is a constant, and d_g is the average grain size. In other words, the yield stress increases as the average grain size decreases because pileups in fine-grained materials contain fewer dislocations, and the stress at the tip of the pileup decreases. Thus, a larger applied stress is required to generate dislocations in the adjacent grain. When the average grain size becomes too small, this mechanism breaks down because the grains could not support the dislocation pileups. Typically, breakdown of the Hall-Petch relationship would occur when the average grain size reaches 10 nm in most metals.

Alloying is also one of the commonly applied methods to increase the mechanical strength in electrodeposits. Solid solution strengthening results from the interaction between dislocation and solute atoms can take place. The solute atoms affect the elastic energy of a dislocation due to both local size and modulus changes and act as obstacles to dislocation motions. The alloys could be electrodeposited from a mixed electrolyte containing different metal salts. Schuh et al. [18] reported that the hardness of Ni increased from 1 to 8 GPa by forming Ni–W alloys. Similar strengthening was also reported in Ni–Co [19, 20], Ni–P [21], and Ni–Mn [22] alloys. In addition, alloying of elements having a large difference in the atomic masses would exhibit pronounced strengthening as demonstrated in Cu-based alloys [23].

1.3 Electrodeposition of metallic materials

In metal electrodeposition, current density is often used to control the characteristics in electrodeposits, in particular, grain size. Metal electrodeposition generally follows Butler-Volmer Equation [24], which indicates the current density applied to the electrode is interrelated to the overpotential η:

$$j=j\left\{exp\left({\mathrm{\alpha }}_a\mathit{zF}\mathit{\eta }/\mathit{RT}\right)-exp\left({\mathrm{\alpha }}_c\mathit{zF}\mathit{\eta }/\mathit{RT}\right)\right\}$$E2

$$\eta =E-E_{\mathrm{eq}}$$E3

where j is the current density, j₀ is the exchange current density, z is the number of electrons involved in the electrochemical reaction, F is the Faraday constant, R is the universal gas constant, T is the absolute temperature, α_a and α_c are the anodic and cathodic transfer coefficient, η is the overpotential, E is the electrode potential, and E_eq is the equilibrium potential. On the other hand, the nucleation rate (ν) of the metal deposited on the electrode is expressed by the following Equation [25]:

$$\mathrm{\nu }=a\cdot exp\left(-{\mathit{b}\mathit{\epsilon }}^2/{\mathit{qk}}_{\mathrm{b}}T\left|\eta \right|\right)$$E4

where a is a proportionality constant, b is the geometrical factor, ε is the surface energy, and q is the required charges for formation of a monolayer. Combining Eqs. (2) and (4), the nucleation rate can be promoted by an increase in the current density, leading to electrodeposits having a finer-grained structure. In addition, electrodeposition can produce not only pure metals but also alloys with controlled compositions. Alloy electrodeposition can realize further enhancement of the mechanical strength based on the solid solution strengthening mechanism [26, 27].

Pulse electrodeposition is a versatile method that has been proven to produce nanocrystalline materials [28, 29]. Pulse electrodeposition parameters (current on-time, current off-time, and pulse current density) play important roles in controlling the electrodeposition process and hence the microstructure and properties of the electrodeposits [30, 31].

1.4 Mechanical properties of small-scale materials

Microcomponents used in MEMS such as microsprings, cantilevers, and structural support could suffer mechanical straining during employment, and mechanical property evaluation of the specimen in microscale is needed. Conventional indentation or wear tests are widely used to characterize mechanical properties of the electrodeposited metallic materials [32]. However, the obtained results are often affected by the substrate, which may not represent the real information of the microcomponents. Moreover, mechanical properties of materials in microscale are much different from those of bulk materials due to the sample size effect [33]. Since Uchic et al. [34] firstly introduced the uniaxial compression testing of micropillars, a new wave of studies of small-scale plasticity has been explored in numerous materials [35, 36, 37, 38, 39, 40, 41]. Therefore, micromechanical tests using specimens (i.e., micropillars [40], microcantilevers [42]) in microscale are recognized as the most reliable method to provide reliable information on the mechanical properties for design of MEMS microcomponents.

2. Electrodeposition of Au–Cu alloys from noncyanide electrolyte

Electrodeposition of Au-based alloys is reported for the uses of decorative jewelry, conductive materials in electronic devices, magnetic materials, or catalysts. For applications in MEMS accelerometers, it is particularly important to have properties such as high mechanical strength, high electrical conductivity, and high density. Au–Ni [43, 44] and Au–Co [45] alloys are reported to show improved mechanical strength, but their magnetic properties may cause the undesired effects in the MEMS devices. Au–Sn alloys are reported to be soft materials and mainly used for soldering [46]. Among these solute elements, Cu has high electrical conductivity and is widely used in electronic devices. Besides the difference of atomic masses between Au and Cu is large, a pronounced effect of solid solution strengthening is expected. The Au–Cu alloys are usually electrodeposited from the alkaline cyanide electrolyte due to the electrolyte stability [47, 48]. However, such strong alkaline electrolyte cannot be used in the lithography process for fabrication of MEMS components, which would cause damage of the photoresists. In this chapter, we utilize the noncyanide electrolyte to electrodeposit Au–Cu alloys and characterize their properties.

2.1 Fabrication of Au–Cu alloys by constant current electrodeposition

The Au–Cu electrolyte used in this work is a commercially available electrolyte provided by MATEX Co., Japan, which contained 17.3 g/L of X₃Au(SO₃)₂ (X = Na, K), 1.26 g/L of CuSO₄, and EDTA as the additive with pH of 7.5. A potentiostat (Solartron SI1287) is served for applying the constant current. The electrodeposition is carried out at 50 °C, and the current density is varied from 2 to 9 mA/cm². A piece of Pt plate and Cu plate with the same dimensions of 1 × 2 cm² is used as the anode and the cathode, respectively. Two thicknesses of the films are prepared for the characterization. Thin films with a thickness of ~3 μm are used for surface characterization, and thick films with a thickness of ~50 μm are used only for fabrication of the microcompression specimens.

Figure 1 shows surface morphology of the Au–Cu alloy films electrodeposited at various current density. The films deposited at lower current densities exhibited nodular-like structures as shown in Figure 1(a) and (b). When a higher current density is used (4–7 mA/cm²), the surface morphology gradually changes to smooth surface condition as shown in Figure 1(c)–(f). Large agglomerates of bump clusters are observed on the surface when the current density is higher than 8 mA/cm², as shown in Figure 1(g) and (h). Similar morphology is reported for the Au-based alloys electrodeposited at high current density [48].

Figure 1.

SEM micrographs of the Au–Cu alloy films electrodeposited at current density (a) 2, (b) 3, (c) 4, (d) 5, (e) 6, (f) 7, (g) 8, and (h) 9 mA/cm².

2.2 Crystalline structure and chemical composition of electrodeposited Au–Cu alloys

Figure 2 shows XRD patterns of the Au–Cu alloys electrodeposited at current densities ranging from 2 to 9 mA/cm². Both the (111) and (200) peaks shift continuously to a higher diffraction angle as the current density increases. For instance, the (111) peak shifts from 2θ = 38.79° at the current density 3 mA/cm² to 2θ = 40.09° at the current density 8 mA/cm². The peak shift is suggested to be a result of the increase in the Cu content since the lattice constants of Cu are larger than that of Au. No diffraction peaks from intermetallic nor other ordered phases are observed in the electrodeposited films. Relationships between the current density with average grain size, Cu concentration, and the lattice constant are summarized in Figure 3. The average grain sizes are estimated from the XRD results and the Scherrer equation. The grain size is reduced from 8.8 nm to a minimum value of 5.3 nm when the current density is increased from 2 to 6 mA/cm². Grain size of electrodeposited materials is highly dependent on the overpotential, in which grain refinement is observed as the overpotential increased [49]. Based on the Butler-Volmer equation, the overpotential is interrelated to the current density, in which the overpotential increases as the current density increases. Therefore, it is expected to see a reduction in the grain size as the current density increases.

Figure 2.

XRD patterns of the Au–Cu alloy films electrodeposited at varied current density.

Figure 3.

Plots of the current density versus grain size and Cu concentration.

On the other hand, an increased in the grain size is observed when the current density increases beyond 6 mA/cm². Increasing the current density also promotes side reaction(s), such as hydrogen evolution. Because of this, overpotential of the main reactions, which are reduction of Au and Cu in this study, would be lowered when the side reaction(s) is promoted [50]. This should be the cause of the grain coarsening observed when the current density is higher than 6 mA/cm². Meanwhile, a sustained increase of the Cu concentration from 12.2 to 46.7 at% is observed when the current density is increased from 2 to 9 mA/cm². The results can be interpreted by the difference in the standard reduction potential between Au and Cu [51]. The standard reduction potential of Cu is more negative than that of Au. An increase in the cathodic current density would make the applied potential to be more negative; hence, reduction of Cu is gradually favored and leads to an increase in the Cu concentration.

2.3 Fabrication of Au–Cu micropillars and micromechanical properties

Micromechanical properties of the Au–Cu alloys are evaluated using micropillars fabricated from the thick Au–Cu films by focus ion beam (FIB, Hitachi FB2100). Fabrication process of micropillar is shown in Figure 4. The Au–Cu micropillars have square cross section of 10 × 10 μm² and height of 20 μm. The microcompression tests are conducted with a testing machine specially designed for microspecimens. The compression is conducted at a constant displacement rate of 0.1 μm/s using a piezoelectric actuator.

Figure 4.

Fabrication process of micropillars by FIB.

Figure 5 shows SIM images of the Au–Cu alloy micropillars fabricated from the thick Au–Cu alloy films before and after the microcompression tests. Barrel-shape deformations are observed in the micropillars fabricated from the films electrodeposited at current density 3, 5, and 6 mA/cm², which are typical deformation behaviors for polycrystalline metallic materials [52, 53]. When the current density is further increased to 8 mA/cm², brittle fractures indicated by the cracks along boundaries of the agglomerates are observed after the compression test. The Au–Cu alloy film electrodeposited at 8 mA/cm² is composed of nano-grains, which is similar to the film electrodeposited at lower current density of 3 mA/cm²; however, formation of the bump-clustered agglomerates at high current density might be the main cause of the brittle deformation. Au–Cu alloys are known to be highly ductile materials. To the best of our knowledge, this is the first report on brittle fracture of Au–Cu alloys, and this information is essential for the design of components used in MEMS devices.

Figure 5.

SIM images of the Au–Cu alloy micropillars fabricated from the films electrodeposited at current density (a,b) 3, (c,d) 5, (e,f) 6, and (g,h) 8 mA/cm². (a,c,e,g) Before and (b,d,f,h) after microcompression tests.

Engineering strain-stress (SS) curves obtained from the microcompression tests are shown in Figure 6. Generally, all the pillars exhibit extremely high yield stress (σ_y, determined by the cross-point of the SS curve and 0.2% offset line of the elastic deformation region) ranged at 1.00–1.15 GPa, which are far larger than the yield stress obtained from micromechanical tests of pure Au and pure Cu reported in the literature [52, 54]. Flow stresses (σ_f) at 10% plastic strain of all the micropillars are higher than 1.3 GPa except for the micropillar prepared from the film electrodeposited at current density of 8 mA/cm², in which the crack-induced brittle fracture should be the reason of the lowered flow stress.

Figure 6.

Engineering strain-stress curves of the micropillars fabricated from the films electrodeposited at current density of (a) 3 mA/cm², (b) 5 mA/cm², (c) 6 mA/cm², and (d) 8 mA/cm². The yield strength (σ_y) in each curve is marked by a horizontal bar.

The enhanced yield stress in the Au–Cu alloys is mainly attributed by the following two mechanisms: (i) grain boundary strengthening [17] and (ii) solid solution strengthening [26, 27]. As shown in Figure 3, the grain refinement effect goes along with an increase in the Cu concentration as the current density increases. According to the grain boundary strengthening mechanism, the strength of metallic materials increases as the total amount of grain boundary in a specimen increases, which is also understood as a decrease in the average grain size. Moreover, the solid solution strengthening mechanism could restrict the dislocation movement due to interaction of the dislocations with the strained lattice surrounding the solute atoms, which then leads to a stacked strengthening beyond the grain boundary strengthening mechanism.

3. Pulse current electrodeposition of ultrahigh strength nanocrystalline Au–Cu alloys

3.1 Fabrication of Au–Cu alloys by pulse current electrodeposition

The Au–Cu alloys are electrodeposited on cold-rolled Cu substrates with a commercially available electrolyte (see Section 2.1). Temperature of the electrolyte is maintained at 50 ± 1°C using a water bath. The pulse current electrodeposition is carried out using a pulse power supply (plating electronic GmbH, type pe86CB-20-5-25-S/GD). For all experiments, the current on-time (t_on) is fixed at 10 ms, while the pulsed current density (J_p) and the current off-time (t_off) are varied. The parameters are summarized in Table 1. Thin Au–Cu alloy films with a thickness of 3–5 μm are used for characterization of the composition, grain size, and morphology. Thick films (thickness ~50 μm) are prepared for fabrication of the microcompression specimens.

Operating parameters	Range
Pulse current density (mA/cm^2)	5–60
Current on-time (ms)	10
Current off-time (ms)	5–600
Electrolyte temperature (°C)	50

Table 1.

Parameters for pulse electrodeposition of Au–Cu alloy films.

3.2 Effects of the pulse current density

Figure 7(a) shows XRD patterns of the Au–Cu alloys electrodeposited at the J_p of 5–20 mA/cm² with the t_on and t_off both fixed at 10 ms. All electrodeposits show the same crystal structure, in which all of the peaks could be indexed to the face-centered cubic (fcc) reflection. With an increase in the J_p, the peaks shift to larger diffraction angles due to the lattice shrinkage caused by the increase in the Cu concentration. Effects of the J_p on the Cu concentration and grain size are plotted in Figure 7(b). The copper concentration linearly increases from 15.3 to 41.2 at% as the J_p increases from 5 to 20 mA/cm². These results are similar to the Au–Cu alloys prepared by the constant current electrodeposition, in which an increase in the J_p leads to a higher Cu concentration. The standard reduction potential of Cu²⁺ ions is more negative than that of Au⁺ ions [51]. An increase in the J_p makes the applied potential to be more negative; therefore, the reduction of Cu²⁺ is gradually favored and leads to an increase in the Cu concentration. In the meanwhile, the grain size decreases from 8.0 nm to a minimum value of 5.2 nm as the J_p increases from 5 to 10 mA/cm², which is attributed to the increase of the nucleation rate as the current density increases [49]. On the other hand, after reaching the minimum value, the grain size increases to 5.8 nm as the J_p increases to 20 mA/cm². The grain growth at high J_p could be attributed to the promoted side reactions (i.e., hydrogen evolution) as the applied potential becomes more negative, which then lowers overpotential of the main reaction(s) (reductions of Au⁺ and Cu²⁺ in this case). The grain size of electrodeposits is highly dependent on the overpotential, and the grain size increases when the overpotential is lowered [49].

Figure 7.

(a) XRD patterns of the Au–Cu alloys electrodeposited at the J_p varied from 5 to 20 mA/cm² with t_on and t_off both fixed at 10 ms. The straight dash line indicates the center of (111) diffraction peak in the J_p of 5 mA/cm² sample. (b) Plot of the J_p versus the grain size and Cu concentration.

3.3 Effects of current off-time

Effects of the t_off on crystal structure and alloy composition of the Au–Cu alloys are discussed in this section. Figure 8 shows XRD patterns of the Au–Cu alloys electrodeposited at the J_p of 20 mA/cm² and the t_off varied from 20 to 120 ms. Similar to Figures 3 and 7(a), all the XRD patterns show the fcc reflections, and no other diffraction peak is observed. The major (111) peak gradually shifts from 2θ = 40.2° to 38.8° when the t_off increases from 20 to 120 ms, which indicates a decrease in the Cu concentration. It is known that the t_off plays an important role in controlling the alloy composition due to the galvanic displacement reaction occurred on the substrate surface [55, 56, 57, 58]. During of the off-time period, nobler metals continue to be deposited on the substrate surface, and less noble metals on the substrate surface would be oxidized and dissolved away. The displacement reaction leads to a decrease in concentration of the less noble component in the alloy. In the present Au–Cu system, the standard reaction potential of Au is more positive than Cu; hence, nobleness of Au is higher. Therefore, the displacement reaction occurred during the off-time period causes a decrease in copper concentration in the Au–Cu alloy, which is consistent with the lattice swelling observed from the XRD results.

Figure 8.

XRD patterns of the Au–Cu alloys electrodeposited at the J_p of 20 mA/cm² with the t_off varied from 10 to 120 ms. The t_on was fixed at 10 ms. The straight dash line indicates center of the (111) diffraction peak in the t_off of 10 ms sample.

Dependence of the Cu concentration and grain size on the t_off as the J_p varies from 10 to 60 mA/cm² is shown in Figure 9. Several trends are observed as the J_p and the t_off change. Firstly, a decrease in the Cu concentration is observed as the t_off increases at all of the J_p. The results correspond well with those observed from the XRD patterns, in which more Cu is replaced by Au as the t_off increases. Secondly, when the t_off increases, decreasing rate of the Cu concentration shows a transition from high to low as indicated by the change in slope of the curves inFigure 4(a). The J_p of 50 mA/cm² curve indicates this point clearly. Cu concentration of the Au–Cu alloy shows a steep decrease from a short t_off to t_off of 240 ms, and the slope becomes less steep at t_off longer than 240 ms. The slope is suggested to be related to the displacement reaction or dissolution rate of Cu component in the Au–Cu alloy, in which a steep slope indicates a high Cu dissolution rate. Again, the result is expected since the Cu dissolution rate is directly related to concentration of Cu component at the surface of the film, and the Cu concentration is higher at the moment when the electrodeposition just entered the off-time period. Then the Cu concentration gradually decreases and leads to a lower Cu dissolution rate. Thirdly, the decreasing rate of Cu concentration is slowed down as the J_p increases. The evidence can be clearly seen in the J_p of 20 mA/cm² and J_p of 60 mA/cm² cases. The Cu concentration decreases from 36.9 to 18.4 at% when the t_off is increased from 30 to 100 ms in the case of the J_p of 20 mA/cm². A much longer t_off is required, from 290 to 590 ms, for the Cu concentration to decrease from 37.1 to 18.9 at% in the case of the J_p of 60 mA/cm². The alloys electrodeposited at higher J_p contain higher Cu concentration. Although the dissolution rate is highly dependent on the Cu concentration, diffusion of Cu²⁺ away from surface of the substrate and Au⁺ from the bulk to the reaction site could also affect the dissolution rate. Hence, a longer t_off is needed to reach the same Cu concentration when the J_p is high.

Figure 9.

Plots of the t_off versus (a) grain size and (a) Cu concentration at varied J_p.

Effects of the t_off on the grain size show similar trends at various J_p, in which the grain size initially decreases to a minimum value of ca. 4.40 nm and then the grain size reversely increases when the t_off increases. The displacement reaction occurred during the off-time period can initiate rearrangement of atoms in the alloy, which could induce nucleation or grain growth of the reduced metals in the alloy. The driving force of the rearrangement is dependent on the dissolution rate. When the driving force is high, the rearrangement is more vigorous, and nucleation is induced. As shown in Figure 9(a) and (b), the grain size reduces with a decrease in the Cu concentration until reaching ca. 35 at%. On the other hand, the rearrangement is less vigorous, and grain growth is favored when the Cu concentration is low. This is why grain coarsening is observed when the Cu concentration is lower than ca. 35 at% as the t_off increases.

As a result, a wide Cu concentration ranging from 10.1 to 53.0 at% is attained by adjusting either or both the J_p and the t_off. In addition, the critical point observed at the Cu concentration of ca. 35 at% indicates the grain size is interrelated to the alloy composition. Figure 10 shows the grain size as a function of the Cu concentration. The Cu concentration and the grain size basically follow a monotonic relationship. Similar behavior is reported in other pulse current electrodeposited alloys [58, 59]. When compared to the constant current electrodeposited Au–Cu (square symbols), the constant current electrodeposited Au–Cu also shows the same monotonic relationship. Furthermore, the pulse current electrodeposition allows fabrication of Au–Cu alloys with a wider range of the Cu concentration and a much finer-grain size than those of the constant current electrodeposition, which are both advantageous for applications as movable microcomponents.

Figure 10.

A plot of relationship between grain size and Cu concentration for Au–Cu alloys electrodeposited with varied J_p and t_off.

3.4 Morphology of pulse electrodeposited Au–Cu alloys

Effects of the pulse current electrodeposition parameters on morphology of the Au–Cu films are observed by the SEM as shown in Figure 11. The overview of the Au–Cu alloys electrodeposited at the J_p of 15 mA/cm² shows bright surfaces when the t_off is between 20 to 50 ms. From Figure 11(a), the alloy film electrodeposited at the t_off of 20 ms shows pebble-like structures, and size of the pebble-like structures shrinks gradually as the t_off increases to 50 ms as shown in Figure 11(b) and (c). The surface becomes dull when the t_off increases to 100 ms, and the pebble-like structures are still observed as shown in Figure 11(d). When a lower J_p at 5 mA/cm² is used, two alloy films deposited at the t_off of 30 and 100 ms both show dull surface. The size of the pebble-like structures increases as the t_off increases to 100 ms (Figure 11(f)). The surface condition becomes very rough when the J_p is increased to 20 mA/cm². As shown in Figure 11(g), the alloy film deposited at the J_p of 20 mA/cm² and the t_off of 20 ms shows large agglomerates of colony-like clusters, and dull surface is observed. Then the surface becomes bright, and the size of the pebble-like structures decreases as the t_off increases to 50 ms, shown in Figure 11(h). An interesting conclusion could be made here, in which Au–Cu alloys with similar surface morphology and similar Cu concentration could be fabricated using different pulse parameters (Figure 11(b): J_p = 15 mA/cm², t_off = 30 ms; Figure 11(h): J_p = 20 mA/cm², t_off = 50 ms). This result demonstrates that not only the grain size but also the surface morphology is interrelated to the Cu concentration. The morphology, composition, grain size, and electrodeposition parameters of the Au–Cu alloy thick films are summarized in Table 2.

Figure 11.

SEM micrographs of the Au–Cu alloy thick films. The alloys electrodeposited at the J_p of 15 mA/cm² with the t_off of (a) 20 ms, (b) 30 ms, (c) 50 ms, and (d) 100 ms; J_p of 5 mA/cm² with (e) 30 ms and (f) 100 ms; J_p of 20 mA/cm² with (g) 20 ms and (h) 50 ms. The t_on is fixed at 10 ms.

Jp, mA/cm^2	toff, ms	[Cu], at%	dg, nm	Morphology
20	20	46.4	4.8	Colony-like clusters and dull surface
	50	33.9	4.7	Pebble structure and bright surface
15	20	36.9	4.7	Pebble structure and bright surface
	30	34.2	4.8	Pebble structure and bright surface
	50	29.5	4.9	Pebble structure and bright surface
	100	21.2	6.2	Pebble structure and dull surface
5	30	18.0	7.0	Pebble structure and dull surface
	100	12.1	9.1	Pebble structure and dull surface

Table 2.

A summary of pulse parameters, Cu concentration ([Cu]), grain size (d_g), and morphology of the Au–Cu alloys.

Effects of the J_p and the t_off on the morphology and the Cu concentration are summarized and illustrated in Figure 12. In general, roughness of the surface is affected by the current density, and smoothness of the surface is related to the displacement reaction, i.e., dissolution of the Cu component in the Au–Cu alloy. In other words, an increase in the J_p leads to roughening of the surface, and promotion of the displacement reaction causes smoothening of the surface. For example, when a high J_p and a short t_off are applied, a rough surface would be formed during the on-time period because of the high J_p, and the smoothening effect caused by the displacement reaction would be insufficient because of the short t_off. In this case, a rough surface condition is obtained as shown in Figure 11(g). When a high J_p and a long t_off are applied, although the high J_p would give a rough surface, but with a long enough t_off, the displacement reaction could cause enough smoothening effect to produce a smooth surface. On the other hand, when a low J_p is used, the surface would be less rough than the one using high J_p. However, the Cu concentration is low when a low J_p is used, and this limits the displacement reaction, that is, the surface smoothening effect. As observed in the alloys electrodeposited at the J_p = 5 mA/cm² in Figure 11(e) and (f), the surface condition does not become smoother as the t_off increases from 30 to 100 ms.

Figure 12.

Illustration of the morphology change with the pulse electrodeposition parameters.

3.5 Micromechanical properties of pulse electrodeposited Au–Cu alloys

Micromechanical properties of the pulse current electrodeposited Au–Cu alloys are evaluated by microcompression tests to demonstrate the potential for applications in microelectronic devices. The micropillars with the same dimensions of 10 × 10 × 20 μm³ are fabricated from the thick Au–Cu films by FIB. Figure 13 shows SIM images of 6 Au–Cu micropillars with different alloy compositions after the microcompression tests. Typical polycrystalline deformation (barrel-shape) is observed in the micropillars at the Cu concentration below ~35 at% (Figure 13(a)–(d)). As the Cu concentration increases to ~37 at%, the deformation behaviors change into brittle fracture (Figure 13(e)). For the Cu concentration of 46.4 at% pillar (Figure 13(f)), the brittle fracture occurs from the crack boundaries originating from the large agglomerates as observed in Figure 11(g). The large agglomerates and the brittle fracture are also observed in the constant current Au–Cu alloys electrodeposited using a high current density, in which the brittle fracture is observed when the Cu concentration is higher than 37 at% (Figure 5(g) and (h)).

Figure 13.

SIM micrographs of the pulse current Au–Cu micropillars after compression tests. The micropillars were fabricated from the thick Au–Cu with the Cu concentration of (a) 12.1 at%, (b) 15.6 at%, (c) 21.2 at%, (d) 34.2 at%, (e) 36.9 at%, and (f) 46.4 at%.

Engineering strain–stress (SS) curves obtained from the microcompression tests are shown in Figure 14. The σ_y’s are estimated from the 0.2% offset line of the elastic deformation region. Similar to the constant current Au–Cu micropillars, the σ_y increases with an increase of Cu concentration and a decrease of grain size until the Cu concentration reaches ~34 at% (Figure 14(a)–(d)). For the Cu concentration of 37 at% micropillar, the σ_y reaches the highest value of 1.50 GPa. However, the subsequent flow stress behaviors are different with the lower Cu concentration micropillars, which are attributed to the difference in the deformation behaviors. The brittle fracture in the Cu concentration of 36.9 at% and 46.4 at% micropillar (Figure 11(e) and (f)) leads to the stagnant and trembling flow stress after the yielding points.

Figure 14.

Engineering SS curves of the micropillars containing the Cu concentration of (a) 12.1 at%, (b) 15.6 at%, (c) 21.2 at%, (d) 34.2 at%, (e) 36.9 at%, and (f) 46.4 at%. The yield strength σ_y is marked by a horizontal bar.

3.6 Strengthening mechanisms in electrodeposited Au–Cu alloys

The σ_y ranges from 0.90 to 1.50 GPa in the constant current- and pulse current electrodeposited Au–Cu micropillars, which can be understood as synergistic effects of the grain boundary strengthening [17] and the solid solution strengthening mechanisms [26, 27]. According to the grain boundary strengthening mechanism [17], the strength of metallic materials increases as total amount of grain boundary in a specimen increases, which is also understood as a decrease in the average grain size. Moreover, the solid solution strengthening mechanism [26, 27] is considered to restrict the dislocation movement due to interaction of the dislocations with the strained lattice surrounding the solute atoms, which then leads to a stacked strengthening beyond the grain boundary strengthening mechanism. In Figure 15, the grain boundary strengthening mechanism can be summarized as the Hall-Petch plots (σ_y vs. (grain size)^–0.5) using the results presented in this study and the literature [47, 48, 60]. Due to lack of the literature on the yield stress of Au–Cu alloys, the results obtained from Vicker microhardness (HV) tests are adopted and converted to the yield stress by dividing the microhardness value to a Tabor coefficient of 4 (σ_y = HV/4 [61]) for the comparison.

Figure 15.

Plots of (a) inverse square root of the grain size (d_g) and (b) Cu concentration versus the σ_y.

Overall, the values reported in the literature all follow the Hall-Petch relationship. However, softening caused by the inverse Hall-Petch effect occurs when the grain size scales down to ~6 nm. The results presented in this work also follow the Hall-Petch relationship well. Most importantly, the results obtained in this study are much more reliable than those of Vicker microhardness tests since the hardness results are often affected by the substrate, which cannot reflect real strength of the electrodeposited films. A number of theories for solid solution strengthening have proposed that the strength is proportional to the solute concentration with order of 1/2 [Fleischer (1963)] or 2/3 [Labusch (1970)], which depends on the solute concentration. It is worth noticing that the highest σ_y at 1.38 GPa obtained in the pulse current electrodeposited Au–Cu micropillar is higher than that of constant current electrodeposited micropillar with the same Cu concentration (σ_y = 1.15 GPa), demonstrating the capability to further refine the grain size and enhance the strength by pulse current electrodeposition.

4. Conclusions

In the present study, high-strength Au–Cu alloys with nanocrystalline structure are successfully fabricated by electrodeposition techniques in order to be applied in fabrication of movable microcomponents in MEMS devices. The Au–Cu alloys are first fabricated by constant current electrodeposition. Surface morphology of the Au–Cu alloy films shows a wide variation from smooth surface to bump-clustered agglomerates as the current density varies from 2 to 9 mA/cm². A reduction in the grain size and an increase in the Cu content are observed with an increase in the current density. The film with the finest grain size at 5.3 nm is obtained when current density 6 mA/cm² is used. For the microcompression tests, the specimens evaluated are micropillars with dimensions of 10 × 10 × 20 μm³ fabricated from the electrodeposited Au–Cu alloys. The highest σ_y at 1.15 GPa is achieved for the Au–Cu micropillar having the grain size of 5.3 nm and the Cu concentration of ~30 at%. The yield strength is higher than the values reported in the literatures and suggested to be a synergistic effect of the grain boundary strengthening mechanism with the solid solution strengthening mechanism.

Furthermore, effects of the pulse current parameters on the alloy composition, grain size, surface morphology, and micromechanical property of the Au–Cu alloys are investigated. A wide Cu concentration in the Au–Cu alloys ranging from 10 to 54 at% is obtained. An increase in the Cu concentration is observed by using either or both of a high pulsed current density and a short current off-time. The smallest grain size of ca. 4.4 nm is achieved in films having the Cu concentration ranging from 30 to 40 at%. Grain refinement is achieved with a high J_p, and promoting the displacement reaction could also reduce the grain size. A high J_p results roughening of the surface, and enhancing the displacement reaction leads to a surface smoothening effect. Deformation behavior of the Au–Cu micropillar is affected by the Cu concentration, in which brittle fraction is observed when the copper concentration is higher than 37 at%. An ultrahigh yield strength of 1.50 GPa is obtained in the micropillar having the Cu concentration of 37 at% and the grain size of 4.7 nm. As a result, Au–Cu alloys developed in the present study are suggested to fulfill the requirements to fabricate more sensitive and miniaturized next-generation MEMS devices.

Acknowledgments

This work was supported by JST CREST Grant Number JPMJCR1433 and by the Grant-in-Aid for Scientific Research (S) (JSPS KAKENHI Grant number 26220907).

Conflict of interest

I confirm there are no conflicts of interest.