Abstract
The aim of this research was to study the direct bonding of ceramic materials, mainly Al2O3 and selected metals, with primary attention given to Cu substrate. Soldering was performed with Sn-based solder alloyed with 2% La. We found that the bond formation between Sn2La solder and Al2O3 occurs at the activation of lanthanum phases in solder by ultrasound. Lanthanum in the solder becomes oxidised in air during the soldering process. However, due to ultrasonic activation, the lanthanum particles are distributed to the boundary with ceramic material. A uniformly thin layer containing La, 1.5 µm in thickness, is formed on the boundary with Al2O3 material, ensuring both wetting and joint formation. The shear strength with Al2O3 ceramics is 7.5 MPa. Increased strength to 13.5 MPa was observed with SiC ceramics.
1. Introduction
Sn-based solders are the most used solders in the electronic industry for surface mount technology and similar applications [1–3]. However, industrial production often requires also joining parts made of different material combinations such as ceramics/metal [4–6]. In soldering such joints, it is inevitable that the ceramic material is wetted with metallic solder.
Research is at present oriented toward direct bonding of ceramic materials by application of active solders [7–10]. This approach reduces the time required for joint fabrication, the hygiene of the working environment is improved, and the economy of production is also enhanced.
The direct bonding of copper with Al2O3 ceramics was established [11]. The joint was fabricated by ultrasonic soldering with application of Zn14Al solder. High quality joints without visible defects were achieved at optimised soldering parameters. The highest shear strength of Al2O3/Zn14Al/Cu joint, 80 MPa, was achieved at a soldering temperature of 480°C and soldering time of 30 s.
In another study [12], the direct soldering of SiC ceramics was performed with ultrasound assistance. The SiC ceramic substrates were soldered in air with Zn8.5Al1Mg solder at a temperature of 420°C. The highest strength (148.1 MPa) was achieved with ultrasound acting for 8 s. A new amorphous layer 2 to 6 nm in thickness was formed in the interface between the solder and substrate. The strong bond between SiC substrate and Zn-Al-Mg solder is attributed to transfer of SiO2 mass to Zn-Al-Mg solder due to cavitation erosion.
The subject of study [13] was oriented to direct soldering of sapphire by ultrasound with application of Sn10Zn2Al solder. It was found that ultrasound supported the oxidation reaction between Al from the solder and sapphire substrate. A nanocrystalline layer of α-Al2O3 (2 nm in thickness) was formed in the Sn-Zn-Al/sapphire interface at soldering in air. This layer allowed the bond formation. The shear strength of joints achieved 43 to 48 MPa, which is a relatively high value when compared to other joints of Al2O3 ceramics fabricated with active Sn solders with addition of Ti and/or lanthanides.
The work [14] deals with direct soldering of Al2O3 by use of an active Sn solder type Sn3.5Ag4Ti(Ce,Ga). Soldering was performed in air at a temperature of 280°C. Ultrasound with a frequency of 40 kHz was employed for solder activation. The shear strength of Al2O3/Sn-AgTi/Al2O3 joint was 24 MPa. A reaction layer, 4–7 μm in thickness, was formed on the interface of Al2O3/Sn-AgTi joint, formed with Ti oxides, mainly TiO. This layer ensured the wettability of Al2O3 ceramics.
The direct soldering of ITO (indium tin oxide) ceramics with a copper substrate, performed with the same Sn3.5Ag4Ti(Ce,Ga) solder in air at temperature of 250°C, was solved in work [15]. The molten solder was agitated by mechanical activation for 30 s. Line scanning has shown that Ti was segregated in the boundary between the ceramics and solder. This new layer is responsible for bond formation. The shear strength of ITO/Cu joints was 3.4 MPa. The Cu/Cu and ITO/ITO joints attained shear strengths of 14.3 MPa and 6.8 MPa, respectively.
In work [16], the authors have dealt with soldering of Al2O3/Al2O3 and Al2O3/Cu at a temperature of 250°C in air. The molten Sn3.5Ag4Ti(Ce,Ga) solder was agitated for 30 s for wetting on bond surfaces and then a copper or alumina specimen was placed on the molten solder to be joined with an alumina specimen by rubbing together for 30 s. The affinity of Ce to oxygen prevents Ti from oxidising and thus Ti can react with Al2O3 at a low temperature. The shear strength of alumina/alumina, copper/copper, and alumina/copper joints was 13.5 MPa, 14.3 MPa, and 10.2 MPa, respectively.
In spite of a great number of positive properties, which Sn solder attains by the addition of rare earth elements to alloy, some negative phenomena also occurred, namely, the formation of Sn whiskers. A wide group of authors [19–23] dealt with this issue. The growth of whiskers is, by the opinion of many authors, caused by the oxidation on solder surfaces and due also to internal stress. Author [24] stated that the stress induced by formation of intermetallic Cu6Sn5 compound forces the Sn atoms out of the outer surface oxide of Sn layer. By author [25], the growth of Sn whiskers is very slow in most cases. Regarding the fact that the rare earth elements have a high chemical potential and that they react much easier with oxygen, they thus enhance the growth of Sn whiskers in solders containing these elements.
As is obvious from the studies mentioned, the active element is an essential component of the solder, since it ensures wettability and bond formation between the metallic solder and ceramic material.
Many research works dealt with solder where titanium was used as an active element [26–29]. Active solders containing Ti (up to 4 wt.%) and small trace amounts of lanthanides such as Ce [18, 30–32] were also studied. Cerium and other lanthanides in those solders contained around 0.1 to 0.2 wt.% in trace amounts. The authors in the mentioned works did not study whether the presence of lanthanides without the presence of Ti can ensure the wettability of ceramic substrates such as Al2O3 and thus create a direct bonding with the ceramic material.
The aim of our study was oriented toward the direct bonding of Al2O3 ceramics with copper substrate with the application of Sn2La solder. We studied whether Sn-based solder, alloyed with La, can wet Al2O3 ceramics and form a strong bond. We also studied whether La can substitute Ti in the active Sn solders.
For this purpose, the analyses were performed to reveal the mechanism of bond formation and the shear strength of fabricated joints was also measured. Lanthanum, which exerts a high affinity to oxygen, was applied as an active element in this case. The amount of 2 wt.% La was selected for comparison, since in the previous study [26] we dealt with interactions of Sn2Ti solder with the surface of Al2O3 ceramics. Soldering was performed at a low temperature in air with application of high-power ultrasound.
2. Experimental
Sn solder with 2 wt.% La was used in the experiments. The solder was manufactured by casting in form of an ingot. Weighing single solder components was done after setting the weight ratio of prepared alloys. Components with high purity from 3 N to 5 N were used for solder fabrication. The manufacture was performed in horizontal tube vacuum furnace with resistance heating. The working temperature used during manufacture was 900°C and vacuum of 10−4 Pa was also employed. At this temperature, held up within the time of 20 min, homogenization of soldering alloy took place. Cooling down in vacuum furnace was slow. The cooling rate was 14°C/min.
The substrates of the following materials were used in experiments:(i)Metallic Cu substrate of 4 N purity in the shape of rings in dimensions 15 × 1.5 mm.(ii)Ceramic Al2O3 substrate of 5N purity in the form of 15 × 2 mm rings (manufacturer Glynwed, GmbH, designation Degussit Al23).(iii)Ceramic SiC substrate in form of 15 × 3 mm rings (manufacturer CeramTec, GmbH, designation Rocar SiC).For a more detailed analysis, the material combinations given in Figure 1 were selected.
Soldering was performed on Hanuz UT2 ultrasonic equipment with parameters given in Table 1. Solder activation was realised by using an encapsulated ultrasonic transducer consisting of a piezoelectric oscillation system and titanium sonotrode with an end diameter of 3 mm. The scheme of ultrasonic soldering through the layer of molten solder is shown in Figure 2. The soldering temperature was 20°C above the liquid temperature of the solder. Soldering temperature was controlled by a continuous temperature measurement on hot NiCr/NiSi plate by a thermocouple.
Soldering proceeded as follows: solder layer was deposited on the substrate heated at a soldering temperature. The liquid solder was then subjected to ultrasound activation in air for 5 s. After ultrasonic activation, the excessive layer of molten solder and formed oxides were removed from the substrate surface. Both joined surfaces were prepared in this way. The substrates with deposited layer of molten solder were laid on each other to obtain contact with the liquid phase. Then, they were centred and the joint formed by their slight compression. Graphical representation of this procedure is shown in Figure 3.
Metallographic preparation of specimens from soldered joints was realised by the standard metallographic procedures. Grinding was performed by use of SiC emery papers with granularity of 240, 320, and 1200 grains/cm2. Polishing was performed by use of diamond suspensions with grain sizes: 9 μm, 6 μm, and 3 μm. Final polishing was made by use of polishing emulsion OP-S (Struers) with 0.2 μm grain size.
Solder microstructure was studied using(i)light optical microscope type Neophot 32 with the application of image analyser NIS-Elements, type E,(ii)electron scanning microscopy (SEM) on FEI Quanta 200 FEG microscope,(iii)qualitative and semiquantitative chemical analysis of Sn2La solder performed on JEOL 7600 F equipment with microanalyser type Microspec WDX-3PC.X-ray diffraction analysis was used for identification of phase composition of solder. It was realised with a solder specimen in dimensions of 10 × 10 mm on XRD diffractometer type PANalytical X’Pert PRO.
DSC analysis of Sn2La solder was performed on equipment type Netzsch STA 409 C/CD in shielding with Ar gas of 6 N purity.
Shear test was carried out for determination of shear strength of soldered joints. Measurements were realised on two ceramic (Al2O3 and SiC) and five metallic (Al, Ni, Ti, Cr-Ni steel, and Cu) materials with Sn2La solder. The shear strength was measured on versatile tearing equipment type LabTest 5.250SP1-VM. An especially developed shearing jig was used for changing the direction of tensile loading forces acting on the specimen. This shear jig assures a uniform loading of the specimen in shear in the plane boundary between the solder and substrate (Figure 4). The dwell time at a soldering temperature during specimen fabrication was 30 s and the ultrasound period was 5 s.
3. Experimental Results
3.1. Analysis of Sn2La Solder
Figure 5 shows the macrostructure of Sn2La solder on ingot cross section. Clearly visible La phases, uniformly distributed in tin matrix, may be seen. The uniform distribution of La phases in tin matrix can be seen also on solder microstructure shown in Figure 6. A detailed view of La phase may be seen in Figure 7. No La was observed in the matrix of solder studied. Solder matrix is formed of pure tin. This was proved by EDS analysis.
The results of XRD analysis in Figure 8 prove that the solder matrix is formed of pure tin, where lanthanum phases, LaSn3, also occur. Similarly, the presence of LaSn3 phase was proved by EDS analysis. The point analysis of LaSn3 phase in the solder is documented in Figure 9.
The authors’ opinion in [17] and the binary La-Sn system also proves the presence of LaSn3 phase at lanthanum content 2 wt.% La (Figure 10).
DSC analysis was performed to identify the melting point of Sn2La solder. Measurements were taken in three subsequent cycles called RUN1, RUN2, and RUN3. DSC record of Sn2La solder is shown in Figure 11. From the curve course, it is obvious that the start of solder melting is at temperature 232.1°C, which approximately corresponds to the melting point of pure tin. The temperature peak is at 237.2°C. According to the binary La-Sn diagram in Figure 10, the eutectic phase transformation at temperature °C is concerned. The solder is already fully molten at temperature 243.6°C. The presence of 2 wt.% La in the tin solder matrix resulted in a shift of melting temperature of Sn2La by approximately 5°C, compared to the melting point of pure tin.
3.2. Analysis of Boundary of Al2O3/Sn2La Soldered Joint
Solder microstructure obtained by SEM after soldering with power ultrasound on Al2O3/Sn2La solder boundary is shown in Figure 12. Since the soldering was performed in air, the La containing phases oxidised. Lanthanum is concentrated in great, considerably oxidised phases in size from 5 to 60 μm, as is clearly visible in Figure 12. Some phases are with sharp edges (Figure 14). However, this phenomenon was not observed in case of soldering Cu/Sn2La solder/Cu joint. In this case, La remains uniformly distributed in the solder matrix structure. Great oxidised La phases scarcely occur. These facts suggest that formation of great oxidised La phases is related to La interaction with the surface of Al2O3. Comparison of microstructures of soldered Al2O3/SiC and Cu/Cu joints from optical microscopy is documented in Figure 13.
During the course of study with bounced electrons (BSE), lanthanum should glow, but due to the fact that it is enriched by oxygen, the La containing particle is dark (Figure 14).
Oxidised lanthanum particles from Figure 14 contain 9.56 wt.% O2, 18.20 wt.% Sn, and 72.24 wt.% La. Point analysis of sharp-edged lanthanum particle is documented in Figure 15.
Solder matrix in Al2O3/Sn2La soldered joint is composed of almost pure tin (the accuracy of EDS analysis should be taken into account). The point analysis of matrix is documented in Figure 16. The matrix is composed of fine-grained tin structure.
As was shown, considerable amount of lanthanum is oxidised during soldering in the air. The lanthanum particles are distributed to the boundary with ceramic material owing to the effect of ultrasonic activation (Figures 17 and 18). The concentration line of La in Figure 17 proves increased La concentration on the boundary with Al2O3 ceramics. In Figure 18, one can see a uniform continuous layer of lanthanum particles on the boundary with ceramic material, which ensures the joint formation. This layer is around 1.5 μm thick. In spite of this layer, the solder is more or less adhered to the ceramic substrate. The bond with ceramic material is of adhesive character. No formation of intermetallic phases was observed, which also causes lower shear strength of joints with ceramic materials.
3.3. Boundary Analysis of Cu/Sn2La/Cu Soldered Joint
When soldering copper substrate with Sn2La solder, it was found that the solder matrix is also formed of pure tin, similarly as in cases of joints with Al2O3 ceramics. Redistribution of greater constituents containing lanthanum (Figure 19) was caused by ultrasound activation. The solder structure in Cu/Sn2La/Cu joints differs from the Al2O3/Sn2La/Al2O3 joints. Lanthanum in the case of the Cu/Sn2La/Cu joint was uniformly distributed on the grain boundaries of tin in the form of very fine LaSn3 phases (Figures 19 and 20). The particles of this phase are very fine with their size ranging from 0.25 to 0.35 μm. It is surprising that due to La presence in solder matrix the copper is not dissolved in tin matrix. Normally, the Cu concentration in Sn should be almost 1 wt.%.
The intermetallic Cu6Sn5 and Cu3Sn phases were formed on the boundary of copper substrate and tin solder during ultrasonic soldering. Fine arches of those intermetallic phases with maximum thickness of 4 μm are seen in Figure 20. The Cu3Sn phase is very thin and therefore less visible. It was identified by EDS analysis. From these results, we may conclude that La in the solder prevents dissolution of Cu in tin solder and hinders the growth of intermetallic Cu6Sn5 and Cu3Sn phases, which is advantageous from the viewpoint of the life of joints subjected to thermal cycling.
Contrary to Al2O3/Sn2La joint, also greater constituent of oxidised lanthanum phases seldom occurs in cases of Cu/Sn2La joints in solder matrix. The incidental oxidised La phase may be observed in Figures 20 and 21. The concentration profiles of La, Cu, Sn, and O elements in Figure 21 show that the greatest oxygen amount, thus oxides, occurs in the lanthanum particle. In Figures 19, 20, and 21, purely visible uniformly distributed fine particles of LaSn3 phase are seen.
Beside the formation of transient metallic Cu6Sn5 and Cu3Sn phases, we also observe planar distribution of La, O, and Sn elements in sharp-edged oxidised particle with high lanthanum contents on the planar analysis of soldered Cu/solder Sn2La joint in Figure 22.
3.4. The Results of Shear Strength in Soldered Joints
Research in this study was primarily oriented to soldering ceramic Al2O3 substrate and copper substrate. These experiments determining shear strength of soldered joints were also extended to other metallic materials (Al, Ni, Ti, and CrNi steel) and SiC ceramics in order to show the wider applicability of Sn2La solder.
Measurements were performed on 4 specimens of each material. The resulting average shear strengths are documented in Figure 23. The lowest shear strength was observed on Al2O3 ceramics (7.5 MPa). Somewhat higher strength (13.5 MPa) was observed on SiC ceramics and the highest was achieved on Al and Ni. The shear strength on copper substrate was 26.0 MPa.
The Sn2La solder showed relatively wide variables in shear strength on metallic and ceramic materials. We found the shear strength of joints in metallic materials to be nearly 3 times higher than that in case of ceramic materials. For more correct identification, the fractured surfaces of Cu/Sn2La and Al2O3/Sn2La joints (Figures 24 and 25) were also analysed.
The Cu/Sn2La joint showed formation of a typical ductile failure by shear mechanism (Figure 24). The fracture morphology clearly shows the motion of the shearing tool. Figure 25 shows the documented fractured surfaces of Al2O3/Sn2La joint. Fracture surface is without visible tracks of shear tool, so the failure occurred by tearing without a shearing mechanism. 100% Sn2La solder covering remained on the Cu substrate after the test. On the opposite, Sn2La solder was torn from the ceramic substrate. The Al2O3 substrate remained only partially covered by Sn2La.
4. Discussion
Lanthanum present in form of LaSn3 in the volume of Sn2La solder is inefficient at soldering temperature 290°C and therefore it does not wet the ceramic material. Powerful ultrasound is thus necessary for its activation. At direct ultrasound activation through the layer of molten solder (Figure 2), the solder is extremely vibrated by titanium sonotrode. This allows significant diffusion of active La element from the solder matrix to boundary with ceramic and/or metallic material. This depends on mechanism of joint formation. In the zone of ultrasound impact, the temperature may be locally increased even up to 1000°C and more [33]. This enhances the physicochemical processes at joint formation.
The experiments have also shown that pure tin without addition of active element (La) does not wet Al2O3 material. The solder composed of pure tin is after melting first shrunk into globular form. Only fine particles of tin and tin oxides are formed by long-time acting of ultrasound, which are then sputtered over the surface of ceramic material without bond formation.
The results of direct soldering of ceramic and metallic materials have proven that Sn solder containing La ensures their wettability at application of ultrasonic activation, meaning that Sn2La solder is suitable for practical applications.
We suppose from the results of analysis performed in transition zone of soldered joints that the joint with a metallic material is of metallurgical-diffusion character. However, the joint with ceramic material, namely, Al2O3, is of adhesive character, when soldered with solder containing La.
A new transition layer with thickness around 1.5 μm was observed on the boundary with the ceramic material. For example, in study [10, 11], a transition layer 2 to 6 nm was formed at application of Zn-Al (Zn-Al-Mg, Sn-Zn-Al) based solders. At application of Sn-Ag-Ti based solders, described in work [12], the transition layer in thickness from 4 to 7 μm was formed. The studies [12–14] proved that this transition layer ensures the joint formation at direct soldering of ceramic materials.
In case of application of Sn2La solder for soldering Al2O3 ceramics, we observed no formation of soldered joint of diffusion character with creation of new products. The shear strength of Al2O3/Sn2La joint was 7.5 MPa, shear strength of SiC/Sn2La joint was 13.5 MPa, and shear strength of Cu/Sn2La joint was 26.0 MPa.
For comparison of results of shear strength, we also give the results from similar studies, while it must be taken into account that different researchers are using different test methods, shape of specimens, and loading rate during testing. They are also using different compositions of soldering alloys and different soldering parameters. For example, at application of solders type Zn-Al (Zn14Al) in work [11] on Al2O3/Zn14Al/Cu joint, the achieved shear strength was 80 MPa at ultrasound power of 200 W. In study [12], the joint of ceramic SiC substrates soldered with Zn8.5Al1Mg solder achieved shear strength of 148.1 MPa at ultrasound acting for 8 s.
In study [13], sapphire was soldered with ultrasound assistance at application of Sn10Zn2Al solder. The shear strength of joints in that case attained 43 to 48 MPa. Several studies dealt with type Sn-Ag-Ti solders. The new metallic, ceramic, and nonmetallic materials were also tested. For example, in study [16], the Al2O3/Sn-Ag-Ti/Al2O3 joint exerted the shear strength of 24 MPa. In [15], the following strength values were achieved: Cu/Cu (14.3 MPa), ITO/ITO (6.8 MPa), and ITO/Cu (3.4) MPa. Similar examples are also mentioned in study [14], where shear strength of alumina/alumina joint was 13.5 MPa, copper/copper 14.3 MPa, and alumina/copper 10.2 MPa.
A brief overview of shear strength of Al2O3/solder joints achieved by mentioned authors is documented in Table 2.
5. Conclusions
The aim of this work was to study direct bonding of Al2O3 ceramics with copper substrate. We studied whether Sn-based solder alloyed with La can wet the Al2O3 ceramics and other ceramic materials and thus form strong joints. For this reason, analyses were performed to reveal the mechanism of joint formation and also measure the shear strength of fabricated joints. The following results were achieved:(i)DSC analysis has shown that Sn2La solder has a melting point of 273.2°C, which is approximately 5°C more than the melting temperature of pure tin.(ii)The matrix of Sn2La solder after soldering Al2O3 ceramics as well as the copper substrate is formed of almost pure tin, where fine particles of SnLa3 phase are segregated along the grain boundaries.(iii)The bond of Sn2La solder with Cu substrate at ultrasound application occurs due to formation of fine bridges of Cu6Sn5 and Cu3Sn phases on the Cu substrate/Sn2La solder boundary. Thickness of these transition phases is extremely low (max. 4 μm). Comparison has shown that the growth of layer of transition intermetallic phases is prevented by fine La phases precipitated along the grain boundaries.(iv)Bond formation between Sn2La solder and Al2O3 substrate occurs at activation of lanthanum phases by ultrasound. Lanthanum contained in the solder volume is oxidised on the air during soldering. The lanthanum particles are distributed to boundary of ceramic material owing to ultrasonic activation. Lanthanum particles are then bound with surface oxides on ceramic material. A thin uniform layer containing La is formed on the boundary with ceramics.(v)The lowest shear strength was observed on Al2O3 ceramics (7.5 MPa). Somewhat higher strength was achieved on SiC ceramics (13.5 MPa). The highest strength within the metallic materials was achieved on Al and Ni (around 35 MPa). The strength attained on copper substrate was 26.0 MPa.(vi)From the results of analysis of transition zones in soldered joints, measurements of shear strength, and fractographic analysis of fractured surfaces, we suppose that the bond with metallic material is of metallurgical-diffusion character. The bond with ceramic material, namely, Al2O3, soldered with La containing solder is of an adhesive character.
Conflict of Interests
The authors declare that there is no conflict of interests regarding the publication of this paper.
Acknowledgments
The contribution was prepared with the support of APVV-0023-12, research of new soldering alloys for fluxless soldering with application of beam technologies and ultrasound, and VEGA 1/0455/14, research of modified solders for fluxless soldering of metallic and ceramic materials. The authors thank Ing. Marián Drienovský, Ph.D., for DSC analysis; Associate Professor Ing. Maroš Martinkovič, Ph.D., for shear strength measurements, Ing. Jiří Faltus, CSc., for EDS analysis, and Ing. Ivona Černičková, Ph.D., for EDX analysis.