Active Brazing of Alumina and Copper with Multicomponent Ag-Cu-Sn-Zr-Ti Filler

Abstract

The study was designed to investigate the synergic effect of Ti and Sn in the active metal brazing of Al₂O₃ ceramic to copper brazed, using the multicomponent Ag-Cu-Zr filler alloy. Numerous fine and hexagonal-shaped rod-like ternary intermetallic (Zr, Ti)₅Sn₃ phase (L/D = 5.1 ± 0.8, measured in microns) were found dispersed in the Ag-Cu matrix of Ag-18Cu-6Sn-3Zr-1Ti alloy, along with the ternary CuZrSn intermetallic phases. An approximate 15° reduction in contact angle and 3.1 °C reduction in melting point are observed upon the incorporation of Ti and Sn in Ag-18Cu-3Zr filler. Interestingly, the interface microstructure of Al₂O₃/Cu joints brazed by using Ag-18Cu-6Sn-3Zr-1Ti filler shows a double reaction layer: a discontinuous Ti-rich layer consisting of (Cu, Al)₃(Ti, Zr)₃O, TiO, and in-situ Cu-(Ti, Zr) precipitates on the Al₂O₃ side and continuous Zr-rich layer consisting of ZrO₂ on the filler side. The shear strength achieved in Al₂O₃/Cu joints brazed with Ag-18Cu-6Sn-3Zr-1Ti filler is 31% higher, compared to the joints brazed with Ag-18Cu-6Sn-3Zr filler. Failure analysis reveals a composite fracture mode indicating a strong interface bonding in Al₂O₃/Ag-18Cu-6Sn-3Zr-1Ti filler/Cu joints. The findings will be helpful towards the development of high entropy brazing fillers in the future.

1. Introduction

Ceramics have a wide range of applications in electric vehicles, aerospace industries, nuclear and chemical-powder plant applications. Irrespective of their potential in various applications, utilizing ceramics in required complex shapes imposes serious manufacturing and economic limitations in industrial sectors [1,2,3,4,5]. The idea of ceramic to metal joining was considered as a promising approach that resolves the existing difficulties in the manufacturing process. Joining metals to ceramics provides complementary properties that cannot be achieved with either ceramics or metal alone; however, achieving a good joint interface between ceramic and metal has a lot of hurdles to overcome [2,3]. Moreover, the overall properties rely on the joint interface strength. Ceramics and metals have distinct differences in their bonding nature and chemical and physical properties. Metals have a metallic bond with free electrons; in contrast, ceramics have a covalent or ionic bond with extremely stable outer-shell electrons. Hence, molten metal does not wet the ceramic surface, thus limiting the prerequisite condition for joining, i.e., establishing intimate contact between ceramic and metal surface [4]. Additionally, distinct physical properties between ceramics and metals, such as elastic modulus, strength, and Coefficient of Thermal Expansion (CTE) mismatch lead to large residual stress during cooling, affecting the joint strength [5].

Adhesive bonding [6], diffusion bonding [7], active metal brazing (AMB) [8] and transient liquid-phase bonding [9] are some of the well-known ceramic-to-metal joining methods. Owing to its simplicity and good joint properties, AMB is the commonly employed ceramic–metal joining technique. AMB is a joining process wherein ceramic and metal surfaces sandwiched with Active Braze Alloy (ABA) are brazed in vacuum furnace. ABA generally consists of Ag, Cu, Ni or Au based fillers that contain Ti or Zr as active metals to promote their wetting on the ceramic surface in their molten state [10]. Moreover, in AMB there is no necessity to pre-coat the ceramic surfaces with molybdenum-manganese (Mo-Mn) metallizing layer, as the active metals in ABA improves the wetting and assist in the formation of reaction layer on the ceramic surface. Therefore, AMB is a suitable process for joining ceramic metal parts with complex shapes [10]. Commercially used ABAs are Ag-35.25Cu-1.75Ti wt.% (Cusil) and Ag-26.7Cu-4.5Ti wt.% (Ticusil), where Ti is distributed as Cu-Ti intermetallic compounds (IMCs) in the Ag-Cu eutectic alloy matrix [11]. Cu-Ti IMCs with a size range of 50–100 µm formed during solidification of Ag-Cu-Ti alloy deteriorate the mechanical property of the ceramic–metal brazing seam. To overcome this issue, Sn was added to promote the formation of finer Cu-Ti and Sn-Ti IMCs (<20 µm) [12]. Work of adhesion (W_ad), defined as the amount of work required to separate two surfaces, is an important quantity associated with the bond strength of metal/ceramic interactions. Higher W_ad in ceramic/metal system signifies a stronger bonding, which results in good joint strength. As an active element, Zr has higher work of adhesion (W_ad) with Al₂O₃, compared to Ti [13]. In mid-1993, Scandia national laboratories came up with Zr based fillers such as Ag-28Cu-2Zr wt.% (Zrcusil) and Ag-1Cu-2Zr wt.% ABA alloy as alternative candidates for ceramic/metal joining [14]. Ag-Cu-Zr brazing fillers have been employed for joining various ceramics, such as Al₂O₃ [15], AlN [15], MoSi₂ [16], cBN [17] and SiO₂ [18]. However, their usage is limited by the formation of blocky Cu₄AgZr IMC because of the high affinity of Zr with matrix elements Cu and Ag. Formation of Cu₄AgZr IMC in the filler alloy also reduces the Zr availability for the formation of reaction layer, resulting in a thinner reaction layer [14]. Kim et al. [13] observed a reduced affinity of Zr with Cu and Ag upon the addition of Sn, thus reducing the fraction of Cu-Zr IMCs in the brazing seam. Additionally, the addition of Sn decreased the melting temperature of Ag-28Cu-5Zr wt.% filler and increased the shear strength of the Al₂O₃/Ni-Cr steel joints [19]. The shear strength increase upon Sn addition in Ag-Cu-Zr filler is mainly attributed to the reduction of blocky Cu₄AgZr IMC [14,19]. Recently, double-active filler Ag-Cu-Zr-Ti is given attention, as it exhibits high hardness and wear resistance with appreciable wetting on cBN substrates. However, blocky Cu₄AgZr IMCs in double-active filler Ag-Cu-Zr-Ti reduced the shear strength of brazed joints [20]. Considering the advantage of Sn in reducing the affinity of Cu-Zr IMC in Ag-Cu-Zr filler, the work reported here aims to study the interaction of Sn and Ti in Ag-Cu-Zr filler.

In the present era, multi-component alloys having more than three alloying elements, and high entropy alloys having five or more alloying elements wherein maximum and minimum atomic compositions for each elements ranges between 35 and 5 at.%, are widely explored in joining applications [21]. Different elements in multi-component alloys can interact and influence the reactions at the filler/ceramic interface. Hence, it important to understand the microstructure and the phase formation in the fillers. This study can be used to understand the IMCs formed during the interaction of elements Sn, Ti and Zr in the Ag-Cu system. In the present study, two multicomponent filler alloys, namely Ag-Cu-Sn-Zr and Ag-Cu-Sn-Zr-Ti, were studied for their microstructure, phase formation and contact angle on the Al₂O₃ substrate. Then, the interfacial microstructure and the corresponding shear strengths were investigated for Al₂O₃/Cu joints brazed by using Ag-Cu-Sn-Zr and Ag-Cu-Sn-Zr-Ti fillers.

2. Materials and Methods

2.1. Filler Alloy Fabrication and Active Metal Brazing

Braze filler alloys with nominal composition, as shown in

Table 1. Nominal composition of the brazing filler alloy.

S. No.	Composition in Weight %
	Ag	Cu	Sn	Zr	Ti
1.	73	18	6	3	-
2.	72	18	6	3	1

, were prepared by using a vacuum furnace (all the alloy compositions mentioned in this work are in wt.%, unless stated otherwise). Commercially available powders of Ag (99.99%), Cu (99.99%), Sn (99.9%), Ti (99.95%) from Kojundo Chemical Laboratory Co., Ltd., (Saitama, Japan) and Zr powder (99.4%) from GG-JING YAN (Shaoguan, China) were used. The powders were mechanically mixed and melted at 1000 °C, in a vacuum furnace, at 1.3 × 10⁻² Pa. Vacuum melting is necessary to prevent the oxidation of filler alloys. The alloys were homogenized and re-melted at 1000 °C, to achieve compositional uniformity. From the ingot, the braze sheets were then cut, polished into sheet of dimension 15 mm × 15 mm × 0.3 mm and ultrasonically cleaned in ethanol.

Then, 96% pure Al₂O₃ (purchased from Korea electronic material, Gyeonggi, Korea), having a dimension of 15 mm × 15 mm × 5 mm, and 99.9% pure Cu (purchased from Daeduck metals, Gyeonggi, Korea) with a dimension of 20 mm × 20 mm × 5 mm were used as base ceramic and metal substrates respectively. These substrates were polished by using #500-, #800- and #1000-grit SiC abrasive paper and ultrasonically cleaned in ethanol. The as-prepared braze sheets were then sandwiched between Al₂O₃ and Cu substrates. AMB was performed at the vacuum level of 1.3 × 10⁻³ Pa with the heating and cooling rate of 10 and 5 °C/min, respectively. The peak brazing temperature and brazing time were fixed as 840 °C and 15 min, respectively, for both the brazing alloys.

2.2. Microstructural and Thermal Analysis

The brazed samples were cut perpendicular to the bonded interface, and metallographic specimens were prepared. Microstructures of the filler alloy and brazed joints were examined, using analytical SEM (JEOL JSM-6010PLUS, Tokyo, Japan) attached with an energy-dispersive spectrometer (EDS). The microstructure specimens were etched in a solution containing 25 mL H₂O + 50 mL H₂O₂ + 25 mL ammonia water prior to the microstructural examination. The reaction phase at the Al₂O₃/filler interface was identified by XRD, using the Cu K-α spectrum having a wavelength of 1.5406 A (Bruker AXS GmbH-D8 DISCOVER, Karlsruhe, Germany). For XRD, the brazed samples were cut parallel to the bonded interface and polished until the reaction phases at the interface were revealed. The melting point of the filler alloys was analyzed, using differential scanning calorimetry (DSC, SDT 650, TA instruments, New castle, DE, USA) at a heating rate and cooling rate of 10 °C/min under N₂ atmosphere. The sample weight was maintained at 10.5 ± 0.2 mg in all the DSC tests.

2.3. Contact Angle Measurement and Shear Test

The spreading tests for contact angle measurement were carried in accordance with the Japanese Industrial Standard (2003: JIS-Z-3198-3). To measure the contact angle, spreading of 0.3 g of filler alloy on Al₂O₃ substrate was carried in a vacuum furnace, at a vacuum level 1.3 × 10⁻³ Pa, with a heating and cooling rate of 10 and 5 °C/min, respectively. The temperature and time were 840 °C and 15 min, respectively. After spreading, the samples were mounted in epoxy, cut perpendicular to the spreading interface, and the contact angle is measured by following Young’s equation: [22]

$${\mathsf{\gamma }}_{\mathrm{SL}}{=\mathsf{\gamma }}_{\mathrm{SV}}{-\mathsf{\gamma }}_{\mathrm{LV}}\cos \mathsf{\theta }$$

(1)

where ${\mathsf{\gamma }}_{\mathrm{SL}}$, ${\mathsf{\gamma }}_{\mathrm{SV}}$ and ${\mathsf{\gamma }}_{\mathrm{LV}}$ represent the solid–liquid, solid–vapor and liquid–vapor surface tension, respectively, and θ represents the contact angle. The measurement procedure is schematically illustrated in Figure 1. Assuming the solidified solder having a spherical shape, a best fit circle for solder surface was drawn. A tangent was drawn at the point P, which is the intersection of the circle and the top surface of the Al₂O₃ substrate. The contact angle is then given by Equation (2) [23]:

$${\mathsf{\theta }=\sin }^{-1}(\frac{\sqrt{{\mathrm{R}}^2{-\mathrm{y}}_{\mathrm{c}}^2}}{\mathrm{R}})$$

(2)

where R is the radius of the circle and y_c is the y-axis coordinate of center of the circle.

The reported contact angle measurement was the average of three specimens. The shear test was carried at a shear speed of 0.5 mm/min using UTM WDW-20 (Chenda tester, Jian, China). Each reported shear test value is an average value of three tested specimens. After the shear test, the fractured Al₂O₃ samples were mounted and sectioned to study the crack propagation.

3. Results and Discussion

3.1. Microstructure of Ag-18Cu-6Sn-3Zr and Ag-18Cu-6Sn-3Zr-1Ti Brazing Filler

Figure 2 shows the microstructure and EDS analysis of as-cast filler alloys. The microstructure of the Ag-18Cu-6Sn-3Zr filler alloy as shown in Figure 2 consists of a bright Ag-rich and dark Cu-rich matrix region. It reveals a bright lamellar phase marked as A and dark blocky phase marked as B dispersed In the Ag-Cu matrix. EDS analysis on points A and B are shown in Figure 2. Based on the at.% calculations from the EDS analysis and the references listed in Table 2, the bright lamellar phase indicated by point A refers to the ternary CuZrSn phase [23] and the blocky phase indicated by point B refers to a ternary Cu₄AgZr IMC [24].

Ag-Cu system is an interesting combination having the same crystal structure (face-centered cubic) and positive enthalpy of mixing (10.8 KJ·mol⁻¹ [23]). During solidification, to attain equilibrium Cu and Ag atoms tend to segregate and forms Cu-rich and Ag-rich matrix [25]. As there are no thermodynamic investigations available for a quaternary Ag-Cu-Sn-Zr alloy, the present work characterizes the intermetallic phases based on the ternary systems of Ag-Cu-Zr [26,27,28] and Cu-Sn-Zr [24,29]. Zr has a maximum solid solubility of 0.1 at.% at 900 °C in Ag [26] and 0.12 at.% at 972 °C in Cu [27]. Liquid-phase separation is observed in the ternary Ag-Cu-Zr system in Ag-rich area, where the liquid melt (L) is separated into Ag-rich liquid (L₁) and Cu-Zr-rich liquid (L₂) [28]. During solidification, precipitation of Cu₄AgZr IMC and occurs at 898 °C by the reaction (3). Cu₄AgZr is a high-temperature IMC with a congruent melting temperature of 1080 °C [22].

$${\mathrm{L}}_1\left(\mathrm{Ag}\right)\mathrm{rich}\to {\mathrm{L}}_2{\left(\mathrm{Cu}\text{-}\mathrm{Zr}\right)\mathrm{rich}+\mathrm{Cu}}_4\mathrm{AgZr}+\mathsf{\alpha }\text{-}\mathrm{Ag}$$

(3)

Romaka et al. [29] experimentally investigated the isothermal section of the ternary Cu-Sn-Zr system at 397 and 497 °C. Apparently, from the investigation, a ternary equiatomic CuZrSn phase (designed as τ₁) with TiNiSi structure type and cell parameters of a = 0.66279 nm, b = 0.43679 nm, c = 0.76791 nm has been identified. Yuan et al. [24] experimentally observed the presence of CuZrSn (τ₁) phase in the Cu-Sn-Zr system at 700 °C. The atomic percentages of the CuZrSn (τ₁) phase as reported by Romaka et al. [29] and Yuan et al. [24] is found to be good in agreement with the EDS in the present work as shown in Table 2. From the microstructure, the lamellar CuZrSn (τ₁) phase either exists along with the Cu-rich inclusions or formed from Cu₄AgZr IMC. The CuZrSn (τ₁) phase could have transformed from the Sn-enriched Cu-Zr liquid (L₂) during solidification. However, an in-depth analysis is required to understand the transformation mechanism.

Figure 3a–h shows the SEM microstructure, elemental mapping and EDS analysis of as-cast Ag-18Cu-6Sn-3Zr-1Ti filler alloy. Similar to Ag-18Cu-6Sn-3Zr filler, CuZrSn (τ₁) phase (marked as C) is observed in Ag-18Cu-6Sn-3Zr1Ti filler as shown in Figure 3b. Since Ti and Zr are soluble over the entire composition range, presence of Ti in CuZrSn (τ₁) phase is inevitable as seen from the EDS analysis of points C in Figure 3h. Blocky Cu₄AgZr phase is not seen in Ag-18Cu-6Sn-3Zr-1Ti filler. Instead, a fine and hexagonal rod-like phase (L/D = 5.1 ± 0.8 measured in microns) was observed. EDS analysis indicated by point D confirms the presence of Ti, Zr and Sn in the rod-like phase. The at.% of the elements indicates that this phase is identical to (Zr, Ti)₅Sn₃ phase composition in the literature, as shown in Table 2 [30,31]. Elemental mapping of the magnified microstructure as shown in Figure 3c–g confirms the coexistence of Zr, Ti and Sn in (Zr, Ti)₅Sn₃ phase. Nowotny et al. reported that Zr₅Sn₃ and Ti₅Sn₃ IMCs can substitute Ti and Zr atoms to form a hexagonal (Zr, Ti)₅Sn₃ phase [31]. Several other researchers also support the formation of (Zr, Ti)₅Sn₃ phase in the Sn-Zr-Ti system over a wide range of Ti and Zr compositions [31,32,33]. Fine (Zr, Ti)₅Sn₃ phase formed upon the addition of Ti and Sn in Ag-Cu-Zr filler could be more beneficial in mechanical properties as compared to the blocky Cu₄AgZr IMC in Ag-Cu-Zr and Ag-Cu-Zr-Ti fillers [22,34].

Table 2. Quantification of possible phases formed in Ag-18Cu-6Sn-3Zr and Ag-18Cu-6Sn-3Zr-1Ti alloy.

Spot	Nominal Composition from EDS (at.%)					Nominal Composition from the Literature (at.%)					Possible Phase	Reference
	Ag	Cu	Sn	Zr	Ti	Ag	Cu	Sn	Zr	Ti
A	-	33.9	32.7	33.4	-	-	33.3	32.1	34.7	-	CuZrSn (τ1)	[23]
B	13.5	68.2	-	18.3	-	18	64.4	-	17.6	-	Cu4AgZr	[22]
C	-	33.7	30.9	28.2	7.2	-	33.3	32.1	34.7	-	CuZrSn (τ1)	[23]
D	-	-	36.8	46.4	16.8	-	-	37.1	46.8	16.1	(Zr, Ti)5Sn3	[29,30,31,32]

Table 2. Quantification of possible phases formed in Ag-18Cu-6Sn-3Zr and Ag-18Cu-6Sn-3Zr-1Ti alloy.

Spot	Nominal Composition from EDS (at.%)					Nominal Composition from the Literature (at.%)					Possible Phase	Reference
	Ag	Cu	Sn	Zr	Ti	Ag	Cu	Sn	Zr	Ti
A	-	33.9	32.7	33.4	-	-	33.3	32.1	34.7	-	CuZrSn (τ1)	[23]
B	13.5	68.2	-	18.3	-	18	64.4	-	17.6	-	Cu4AgZr	[22]
C	-	33.7	30.9	28.2	7.2	-	33.3	32.1	34.7	-	CuZrSn (τ1)	[23]
D	-	-	36.8	46.4	16.8	-	-	37.1	46.8	16.1	(Zr, Ti)5Sn3	[29,30,31,32]

3.2. Thermal Analysis and Contact Angle of Ag-18Cu-6Sn-3Zr and Ag-18Cu-6Sn-3Zr-1Ti Filler on Al₂O₃ Substrate

Figure 4a represents the DSC curves for Ag-18Cu-6Sn-3Zr and Ag-18Cu-6Sn-3Zr-1Ti filler. One endothermic peak is observed in both the filler alloys. The peak melting temperature of Ag-18Cu-6Sn-3Zr filler is 782.9 °C slightly higher than the Ag-Cu eutectic melting temperature of 779 °C. Addition of 1 wt.% Ti in Ag-18Cu-6Sn-3Zr filler has slightly reduced the peak melting temperature to 779.8 °C. Reactive wetting and spreading are considered an important process in ceramic–metal brazing. Though distinctly defined, spreading and wetting are interrelated concepts and are often interpreted with a certain degree of ambiguity. Wetting is a necessary precursory condition required for spreading [35]. Landry proposed a model wherein the formation of a new compound at the interface leads to wetting in metal/ceramic systems. The final contact angle and spreading kinetics depend on the newly formed interfacial compound [36]. The lower contact angle is a consequence of a higher spreading rate and wettability. Brazing atmosphere has a substantial influence on the wettability of filler metal on the ceramic surface. Low oxygen potential in brazing atmosphere can prevent the active elements (Ti and Zr) in the filler from reacting with the oxygen, which may result in a loss in the wettability [22]. Figure 4b shows the contact angle of Ag-18Cu-6Sn-3Zr and Ag-18Cu-6Sn-3Zr-1Ti filler on the Al₂O₃ substrate, at 840 °C, for 15 min. A contact angle of 70.1 ± 0.4° was obtained for the Ag-18Cu-6Sn-3Zr filler. This value is considerably higher than the contact angle of 54.2 ± 0.5° observed for Ag-18Cu-6Sn-3Zr-1Ti filler. In metal–ceramic systems, the spreading rate is limited by the rate of interfacial reaction process namely: diffusive transport of reacting species and local reaction kinetics at the interface [36]. Although Ti and Zr are both active elements, low atomic mass and small atomic radius of Ti assist in rapid diffusion through the Ag-Cu alloy [17].

The activity coefficient is an important parameter that quantifies the activity (χ_Ti) of titanium at the filler alloy and substrate interface during brazing and is given by Equations (4) and (5) [37]:

$${\mathsf{\chi }}_{\mathrm{Ti}}{=\mathsf{\gamma }}_{\mathrm{Ti}}{·\mathrm{X}}_{\mathrm{Ti}}$$

(4)

where X_Ti is the concentration of Ti and ${\mathsf{\gamma }}_{\mathrm{Ti}}$ is the activity coefficient described as follows [17].

$${\ln \mathsf{\gamma }}_{\mathrm{Ti}}\propto \frac{{\overline{{\mathrm{G}}_{\mathrm{i}}}}^{\mathrm{E}}}{\mathrm{RT}}$$

(5)

where ${\overline{{\mathrm{G}}_{\mathrm{i}}}}^{\mathrm{E}}$ is the excess Gibbs free energy of mixing, R is the gas constant and T is the temperature. The higher the activity coefficient, the more there will be segregation of active elements at the ceramic–metal interface. The activity coefficient of Ti in infinitely diluted Cu-Ti alloy is −1.4 [38], whereas the activity coefficient of Zr in infinitely diluted Cu-Zr alloy is −4 [39]. Due to its high activity, Ti facilitates the formation of reaction products on the Al₂O₃ substrates at a faster rate, as compared to Zr. Consequently, the spreading enhances reducing the contact angle in Ag-18Cu-6Sn-3Zr-1Ti filler.

3.3. Interface Analysis of Al₂O₃/Cu Joint

3.3.1. Interfacial Study of Cu/Al₂O₃ Joints Brazed with Ag-18Cu-6Sn-3Zr Filler

Figure 5 shows the characteristic features of Al₂O₃/Cu joint brazed with Ag-18Cu-6Sn-3Zr filler alloy. The chemical reactions between Al₂O₃ and Ag-18Cu-6Sn-3Zr filler resulted in a reaction layer with a thickness of 5.2 ± 0.4 µm at the Al₂O₃/Cu interface, as shown in Figure 5a. XRD analysis of reaction zone shown in Figure 5b displays the peaks corresponding to α-Al₂O₃, Ag, Cu and monoclinic ZrO₂. Moreover, α-Al₂O₃ peak originates from the substrate, while Ag and Cu peaks originated from the filler alloy. EDS analysis of the interface for the individual elements Al, O, Zr, Cu, Ag and Sn in Figure 5c, shows the presence of Zr and O in the ZrO₂ reaction layer. A copper-rich region is also observed in the immediate vicinity of the Al₂O₃ substrate ahead of ZrO₂ layer. Formation of ZrO₂ reaction layer at Al₂O₃/filler interface is considered as more complex when compared to a simple Al₂O₃ reduction-Zr oxidation reaction [15]. During brazing, Zr in the molten filler diffuses towards Al₂O₃ substrate, facilitates wetting and penetrates along the grain boundaries of Al₂O₃. Meanwhile, Al₂O₃ gets reduced to Al and O ions by Reaction (6) [40]:

$${\mathrm{Al}}_2{\mathrm{O}}_3\to 2\mathrm{Al}+3\mathrm{O}$$

(6)

Zr in the molten filler reacts with O ions and gets oxidized to ZrO₂ by Reaction (7):

$$\mathrm{Zr}+2\mathrm{O}\to {\mathrm{ZrO}}_2$$

(7)

This leads to the oxidation-reduction reaction at the interface described below:

$$2{\mathrm{Al}}_2{\mathrm{O}}_3+3\mathrm{Zr}\to 3\mathrm{Zr}{\mathrm{O}}_2+4\mathrm{Al}$$

(8)

However, Al₂O₃ is thermodynamically more stable than ZrO₂ and the Gibbs free energy (${\mathsf{\Delta }\mathrm{G}}_{\mathrm{R}}^0$) of Equation (8) is positive (396.8 kJ at 1127 °C) [15]. Loehman et al. proposed that the formation of Cu-Zr-Ag-Al precipitates on Al₂O₃ can lower the Gibbs free energy of Equation (7) and favors the formation of ZrO₂ at the interface [15]. Therefore, after the filler metal is melted, Cu-Zr-rich liquid infiltrates along the Al₂O₃ grain boundaries and the high oxidation potential of Zr could have led to redox reaction within Al₂O₃. Thus, the ZrO₂ reaction layer form and continue to grow by the outward diffusion of oxygen ions, leaving behind Cu at the Al₂O/reaction layer interface to form Cu-rich precipitates [40]. Although a possible mechanism of ZrO₂ formation is supported based on Equation (8), sufficient thermodynamic evidence on the formation of ZrO₂ on Al₂O₃ is still too complex to comprehend.

3.3.2. Interfacial Study of Cu/Al₂O₃ Joint Brazed with Ag-18Cu-6Sn-3Zr-1Ti Filler

Figure 6 shows the characteristic features of Al₂O₃/Cu joint brazed with Ag-18Cu-6Sn-3Zr-1Ti filler. The microstructure of the Al₂O₃/Cu joint, as shown in Figure 6a, displays a reaction layer of thickness 7.3 ± 0.3 µm at the interface. This is ~1.9 µm thicker, as compared to the reaction layer obtained by using Ag-18Cu-6Sn-3Zr filler. Furthermore, unlike the joints brazed by using Ag-18Cu-6Sn-3Zr filler, no blocky IMCs are found in the brazed seam. XRD analysis of the reaction layer shown in Figure 6b displays the peaks of α-Al₂O₃, Ag, Cu, Cu₃Ti₃O, CuTi₃, TiO and m-ZrO₂. High-magnification FESEM images and the elemental mapping, as shown in Figure 6c,d show two distinct composition of reaction layers (marked as I and II). A discontinuous Ti-rich region I is present immediately adjacent to Al₂O₃ substrate, followed by a continuous Zr-rich region II. Figure 6d displays the EDS analysis of the phases in the reaction layer of the magnified image. The atomic percentage of elements from EDS analysis corresponding to the spots E, F and G are given in

Table 3. Average elemental composition of reaction products in Figure 6c.

Spot	Nominal Composition from EDS (at.%)					Possible Phase
	Cu	Ti	Zr	Al	O
E	14.4	33.9	12.8	5.1	33.5	(Cu, Al)3(Ti, Zr)3O oxide
F	73.8	5.6	14.2	5.2	-	Cu-(Ti, Zr)
G	-	-	47.7	-	52.2	ZrO2 oxide

. The stoichiometric ratio of elements Zr, Ti, Cu, Al and O in spot E corresponds to Cu₃Ti₃O, where Ti is substituted by Zr, and Cu is substituted by Al. During reduction of Al₂O₃, Al ions diffuses towards the filler and participates in the reaction products formed during the brazing. Earlier research also confirms the solubility of Al in (Cu, Al)₃Ti₃O compound [41]. The presence of elements Cu, Zr, Ti and Al in the EDS analysis of spot F and their stoichiometric composition matches with the Cu-(Ti, Zr) IMC with Al substituted in Cu. The EDS analysis in the spot G indicates the presence of Zr and O that coincides with the stoichiometric percentage of ZrO₂. By combining EDS, XRD and elemental mapping analysis, it is found that the phases present in region I could be (Cu, Al)₃(Zr, Ti)₃O, TiO and in situ Cu-(Ti, Zr) IMC and the region II mainly composed of ZrO₂.

As explained in Section 3.2, a high activity coefficient, low atomic mass and radius of Ti enable faster diffusion towards Al₂O₃ substrate, compared to Zr. Segregation of Ti results in the formation of Cu₃Ti₃O and TiO oxides in the immediate vicinity of Al₂O₃. Apparently, many works in the literature support the formation of interfacial reaction products, such as TiO₂, TiO, Ti₂O and Cu₃Ti₃O during the AMB between alumina and Ag-Cu-Ti brazing filler [37,42,43,44]. The established mechanism of reaction layer in Al₂O₃/Ag-Cu-Ti system is the formation of nm-thick TiO layer followed by µm-thick Ti₃(Cu+Al)₃O on Al₂O₃ [43]. However, the formation mechanism can be affected by the composition of the filler alloy and joining conditions. After melting, Ti in the filler metal diffuses and reacts with alumina to form TiO. Upon cooling, TiO reacts with Cu-Ti intermetallic compounds to form Cu₃Ti₃O and Ti₂O as expressed by the Equations (9) and (10) [44]

$${\mathrm{Al}}_2{\mathrm{O}}_3+3\mathrm{Ti}\to 3\mathrm{TiO}+2\mathrm{Al}$$

(9)

$$3{\mathrm{Cu}}_4\mathrm{Ti}+2\mathrm{TiO}\to {\mathrm{Ti}}_3{\mathrm{Cu}}_3\mathrm{O}+{\mathrm{Ti}}_2\mathrm{O}+9\mathrm{Cu}$$

(10)

As explained in Section 3.2, with low atomic mass and radius, Ti diffuses towards the Al₂O₃ at a faster rate, and the high activity coefficient of Ti facilitates the formation of reaction products, such as TiO and Cu₃Ti₃O, on the Al₂O₃ substrates. Hence, region I in the immediate vicinity of Al₂O₃ substrate is Ti-rich. However, it is complex to understand the formation mechanism of in situ Cu-(Ti, Zr) IMC. With higher atomic mass and radius, Zr diffuses towards the Al₂O₃ at a smaller rate and establish the formation of ZrO₂ in region II. This results in the formation of double reaction layers at the interface.

3.4. Effect of Ag-Cu-Sn-Zr-Ti Filler on the Shear Strength of Al₂O₃/Cu Joints

Figure 7a displays the shear strength of Al₂O₃/Cu joints brazed at 840 °C for 15 min, using Ag-Cu-Sn-Zr filler and Ag-Cu-Sn-Zr-Ti filler as a function of Zr content. The average shear strength of Al₂O₃/Cu joints brazed with Ag-18Cu-6Sn-3Zr-1Ti filler is higher (18.4 MPa), compared with Cu/Al₂O₃ joints brazed with Ag-18Cu-6Sn-3Zr filler (14.1 MPa). The reliability of the Al₂O₃/Cu joints depends on the residual stresses due to CTE and elastic modulus mismatch developed during cooling [5]. Crack propagation in the fractured Al₂O₃/Cu joints brazed with Ag-18Cu-6Sn-3Zr and Ag-18Cu-6Sn-3Zr-1Ti filler are shown in Figure 7b,c, respectively. In the specimens brazed with Ag-18Cu-6Sn-3Zr filler, cracks having initiated at the ZrO₂ reaction layer/filler interface propagated through the ZrO₂ layer, indicating a weak interface. Besides this, cracks are also seen in Al₂O₃ ceramic substrate adjacent to the reaction layer.

ZrO₂ exhibits three polymorphs: monoclinic (m-ZrO₂), tetragonal (t-ZrO₂) and cubic (c-ZrO₂). Moreover, m-ZrO₂ is a stable structure at room temperature and pressure. During cooling in AMB, polymorphic transformation of t-ZrO₂ to m-ZrO₂ is accompanied by a volume increase of approximately 3%–5% [41]. As a consequence of volume change, residual stresses can be generated to this volume change and can result in micro-cracks formation in the m-ZrO₂ reaction layer [5]. Therefore, the fracture toughness of m-ZrO₂ becomes weak and is highly susceptible to crack propagation. Moreover, volume change and CTE mismatch can together generate large residual stresses that further weaken the Al₂O₃ ceramic substrate ahead of the reaction layer. Additionally, presence of blocky IMC in the filler must have created plastic strain localization at the IMC/matrix interface.

In the specimens brazed with Ag-18Cu-6Sn-3Zr-1Ti filler, the crack initiated at the ZrO₂/filler interface, primarily propagated in ZrO₂ reaction layer and then partially shifted towards the Al₂O₃ substrate. This composite fracture mode implies the existence of a strong bonding at the interface. The CTE values of Al₂O₃, ZrO₂ and Ag-Cu filler are 8 × 10⁻⁶/°C [43], 10.5 × 10⁻⁶/°C [10] and (19.2–22) × 10⁻⁶/°C [43], respectively. At the Al₂O₃/Ag-18Cu-6Sn-3Zr-1Ti filler interface, the presence of the Ti-rich region consisting of Cu₃Ti₃O (15.2 × 10⁻⁶/°C [43]), TiO (9.1 × 10⁻⁶/°C [43]) oxides and Cu-(Ti, Zr) ahead of ZrO₂ reaction layer may have alleviated the residual stresses and micro-cracks in m-ZrO₂ reaction layer generated due to the CTE mismatch and t-ZrO₂ to m-ZrO₂ transformation, respectively. Additionally, in situ formed Cu-(Ti, Zr) IMC dispersed in the oxide reaction layer can block the crack propagation in the reaction layer. This results in a strong interface with high fracture toughness. Thus, multicomponent Ag-Cu-Sn-Zr-Ti filler has shown potential to explore the possibilities of upgrading the existing commercial fillers to join a wide range of ceramics and metals.

4. Conclusions

Al₂O₃/Cu joint was successfully brazed at 840 °C, for 15 min, using multicomponent Ag-Cu-Sn-Zr and Ag-Cu-Sn-Zr-Ti fillers. The results can enhance the understanding of elements interaction in the multi-component fillers, as well as their interaction in ceramic/metal interface. Conclusions are drawn as follows:

• The microstructure of Ag-Cu-Sn-Zr filler consists of CuSnZr ternary phase and blocky Cu₄AgZr IMC in the Ag-Cu matrix, whereas the microstructure of Ag-Cu-Sn-Zr-Ti filler consists of CuSnZr ternary phase and micron-sized ternary (Zr, Ti)₅Sn₃ IMC dispersed in the Ag-Cu matrix.
• The peak melting temperature of Ag-18Cu-6Sn-3Zr is 782.9 °C, which is 3.1 °C higher than Ag-18Cu-6Sn-3Zr-1Ti filler. The contact angle of Ag-18Cu-6Sn-3Zr-1Ti filler is 15° lower compared to Ag-18Cu-6Sn-3Zr filler. The presence of Ti in Ag-18Cu-6Sn-3Zr filler proved to enhance the spreading.
• Al₂O₃/Cu joint brazed by using Ag-18Cu-6Sn-3Zr-1Ti filler at 840 °C for 15 min produced a Ti-rich region on the Al₂O₃ side, followed by m-ZrO₂ reaction layer on the filler side (Al₂O₃/Ti-rich region consisting of Cu₃Ti₃O, TiO and Cu-(Zr, Ti) /m-ZrO₂/filler/Cu), whereas the Ag-18Cu-6Sn-3Zr filler resulted in a typical m-ZrO₂ reaction layer.
• Al₂O₃/Cu joints brazed by using Ag-18Cu-6Sn-3Zr-1Ti filler exhibits a higher shear strength and stronger bonding interface, as compared to Al₂O₃/Cu joints brazed by using Ag-18Cu-6Sn-3Zr filler. The presence of the Ti-rich region reduces the residual stress build-up on Al₂O₃ substrate and presence of consisting of Cu₃Ti₃O, in situ Cu-(Ti, Zr) IMC dispersed in reaction layer enhances the fracture toughness of m-ZrO₂ reaction layer.