Abstract

Nanoparticles of a promising lead-free solder alloy (Sn3.5Ag (wt.%, SnAg) and Sn3.0Ag0.5Cu (wt.%, SAC)) were synthesized through a chemical reduction method by using anhydrous ethanol and 1,10-phenanthroline as the solvent and surfactant, respectively. To illustrate the formation process of Sn-Ag alloy based nanoparticles during the reaction, X-ray diffraction (XRD) was used to investigate the phases of the samples in relation to the reaction time. Different nucleation and growth mechanisms were compared on the formation process of the synthesized nanoparticles. The XRD results revealed different reaction process compared with other researchers. There were many contributing factors to the difference in the examples found in the literature, with the main focus on the formation mechanism of crystal nuclei, the solubility and ionizability of metal salts in the solvent, the solid solubility of Cu in Ag nuclei, and the role of surfactant on the growth process. This study will help define the parameters necessary for the control of both the composition and size of the nanoparticles.

1. Introduction

Interconnect (solder-based) materials are of great importance in the field of electronics and manufacturing, and the most widely used solder is an Sn-Pb-based alloy. However, the very nature of the levels of toxicity with lead in this system has been a persistent issue and is, therefore, being phased out in the electronic industry gradually. The development of viable alternatives has focused on many alloy-based systems, such as Sn-Cu [1], Sn-Ag [2], Sn-Zn [3], Sn-Bi [4], and Sn-In [5]. To improve the properties of these binary alloy solders, a third element can be introduced. A Sn-Ag-Cu system [6] was developed to investigate this potential.

The most promising candidates for Sn-Pb solders are Sn-Ag-based alloys, with benefits extending towards the solderability, mechanical properties, and the reliability [7]. There are, however, some disadvantages in this system. One of them is the higher melting temperature than that of Sn-Pb solder, which may cause some damages during the soldering process. To solve this problem, the size-dependent melting temperature of the pure melt and alloy has been studied [813]. These works show that the nanoparticles melt at a lower temperature than the corresponding bulk ones. Thus, the preparation of nanoparticles is extremely important.

Many methods have been developed to manufacture nanoparticles [1417]. Among them, the chemical reduction method is widely used in the production of nanoparticles of lead-free solder [1820]. There has been a significant focus on the physical properties of products. However, little work has been done on the formation process of the nanoparticles of Sn-Ag during the reaction process. The nucleation and growth mechanism are still not fully understood. Hsiao and Duh studied this issue in the synthesis process of Sn3.5Ag0.5Cu nanoparticles in aqueous solution [21]. However, their explanations about the disappearance of Cu in the XRD pattern are still speculative.

In this paper, nanoparticles of two promising lead-free solders, the Sn3.5Ag (wt.%) alloy and the Sn3.0Ag0.5Cu (wt.%) alloy, were synthesized through a chemical reduction method. We studied the formation process of nanoparticles in an anhydrous ethanol solution during the reaction. Many experimental parameters were considered to help explain the differences in the XRD patterns found in the literature.

2. Experimental

The materials used to synthesize Sn3.5Ag nanoparticles were tin(II) 2-ethylhexanoate (C16H30O4Sn) and silver nitrate (AgNO3). 1,10-phenanthroline (C12H8N2·H2O), sodium borohydride (NaBH4), and anhydrous ethanol (CH3CH2OH) were selected as the surfactant, reducing agent, and solvent, respectively. In terms of the synthesis of Sn3.0Ag0.5Cu (SAC) nanoparticles, copper (II) ethoxide monohydrate (Cu(OC2H5)2·H2O) was added to the reactant.

In a typical synthesis of Sn3.5Ag nanoparticles, 0.2998 g of tin (II) 2-ethylhexanoate, 0.0051 g of silver nitrate, and 0.2775 g of 1,10-phenanthroline were mixed with 60 mL of anhydrous ethanol to prepare the precursor solution. The mixture was stirred for about 2 hours before 0.1892 g of sodium borohydride was added into the stirring solution. The reaction time was the variable to be tested in the experimental procedure, so set reaction times of 15 min, 30 min, 45 min, and 60 min were used. The as-synthesized nanoparticles were precipitated by a centrifuge at a rate of 4000 rpm for 45 min. Due to the low solubility of NaBH4 in ethanol [22], the reaction rate during the separation process was extremely low. The reduction reaction could be stopped at any predetermined time through this method. The nanoparticles obtained were rinsed 3 times with anhydrous ethanol and then dried in the vacuum chamber at 40°C for 8 hours. The synthesis procedure of Sn3.0Ag0.5Cu nanoparticles was similar to that of Sn3.5Ag nanoparticles.

The synthesis parameters of these two lead-free nanoparticles with different reaction times were listed in Tables 1 and 2, respectively.

The crystal structure of the synthesized SnAg alloy nanoparticles was characterized by X-ray diffraction (XRD), from 20° to 90° (2θ), using Cu Kα radiation (λ = 0.154056 nm) on a Rigaku D/Max-2550 X-ray diffractometer. The electron accelerating voltage and accelerating current used in the detection process were 40 kV and 200 mA, respectively.

The SAC particle samples were measured on a Rigaku D/Max-2200 X-ray diffractometer from 20° to 90° (2θ). The wavelength of the X-ray from the Cu Kα radiation was 0.154056 nm. The electron accelerating voltage and accelerating current used in the detection process were 40 kV and 40 mA, respectively. The surface features of the synthesized particles were studied by using a scanning electron microscope (SEM, model JSM-6700F).

3. Results and Discussion

3.1. Formation Process of SnAg Nanoparticles

Figure 1 shows the XRD results of the synthesized SnAg nanoparticles. There were two clear phases, the β-Sn and Ag3Sn, indicating the successful alloying of Sn and Ag during the synthesis process. This was further confirmed through their crystal structures by high resolution transmission electron microscopy (HRTEM) [23]. The relative intensity of the diffraction peaks of β-Sn increased with the reaction time, while the Ag3Sn phase showed the opposite trend.

The relative intensity of the three strong peaks from the Ag3Sn phase, obtained from the XRD pattern, was shown in Figure 2. It was obvious that the relative intensity of the strongest peak declined from 100% to 10%. A similar trend could be found in the other two strong peaks. It tended towards stability when the reaction time increased from 40 min to 60 min.

The relative intensity of the diffraction peak of the Ag3Sn phase indicted the formation process of the two phases. When the reaction time was 15 min, the relative intensity of the diffraction peak of the Ag3Sn phase was at the maximum. At the same time, the diffraction peak of β-Sn phase could also be found, as shown in Figure 1(a). This implied that the product was mainly in the Ag3Sn phase. When the reaction time increased to 30 min, the relative intensity of the diffraction peak of Ag3Sn phase in the synthesized nanoparticles decreased dramatically, as shown in Figure 2. The maximum relative intensity corresponds to the main peak of the β-Sn phase, and therefore, the amount of the β-Sn phase increased in the product. The result was the same for the nanoparticles obtained when the reaction time was further increased to 45 min. The reaction was near completion when the reaction time was set to 60 min and the main phase changed to β-Sn with the relative amount of the Ag3Sn phase decreasing to a very low level, which could be derived from Figures 1 and 2.

Hsiao and Duh [21] had studied the formation process of Sn3.5Ag0.5Cu alloy nanoparticles by means of a chemical reduction method in aqueous solution. The precursor was made by mixing SnSO4, AgNO3, and Cu(NO3)·2.5H2O together in the solvent. The reducing agent (NaBH4/NaOH) was added to the precursor and the products and reaction times were investigated. In their work, the formation process of Sn3.5Ag0.5Cu nanoparticles was divided into stages. Firstly, the Ag-Cu nuclei formed in the solution before being transformed to (Ag, Cu)4Sn by adsorbing tin atoms. The second stage involved the growth of the (Ag, Cu)4Sn nanoparticles. It was noted that (Ag, Cu)3Sn formed when the tin atoms in the (Ag, Cu)4Sn phase exceeded the maximum amount.

Another change from the reported literature procedure was the use of anhydrous ethanol as the solvent in our experiment. There were Sn2+ and Ag+ ions in the aqueous solution, and due to the different standard reduction potential of Sn2+/Sn4+ (0.151 V versus the standard hydrogen electrode, SHE) and Ag+/Ag (0.8 V versus SHE), the following reaction would occur [24]

The (Ag, Cu)4Sn phase could be generated without the addition of a reducing agent in reference [21]. However, the origin of Ag atoms was not specified. Using (1), it can be seen that the silver atoms came from two reaction mechanisms at the first stage. Both and NaBH4 could reduce .

If the solvent was changed to anhydrous ethanol, which was studied in the present work, the solubility of AgNO3 would be low [25, 26]. Although AgNO3 could be ionized in ethanol [27, 28], the amount was possibly small due to the difference of molecular polarity between solvent and solute. The reaction in (1) did not take place, which could be confirmed by the experimental phenomena that no change occurred during the preparation process of metal precursor.

When the reducing agent NaBH4 was added into the precursor, black particles precipitated rapidly. In the homogeneous metal ion solutions, electron transfer from NaBH4 to the metal ions occurred during the redox reaction [29, 30]; thus the metal ions turned into atoms. This process could be described as follows [31]

The transition of electrons from the reducing agent to the metal ions was determined by their standard redox potentials. The reduction rate of cations increased with the standard redox potentials increase [32]. The standard redox potential of Ag/Ag+ and Sn/Sn2+ was 0.8 V (versus SHE) [24, 33] and 0.14 V (versus SHE) [34], respectively. The principles were

It could be seen that the standard redox potential of Ag/Ag+ was much higher than that of Sn/Sn2+. Thus, the oxidizing ability of Ag+ was larger compared with Sn2+. It was the same for their reduction rate in the reaction system, which was consistent with the XRD results. Therefore, the formation process of Sn3.5Ag nanoparticles in this work could be summarized.

Before the addition of reducing agent, the metal salts were dissolved homogeneously by stirring for 2 hours in anhydrous ethanol. The potential for ionization was relatively low as the polarity was different from that of solvent.

When the reducing agent was added and mixed into the precursor, black particles precipitated immediately. At this stage, most of the Ag+ ions were reduced by NaBH4. Only a few Sn2+ ions were reduced owing to the relatively lower oxidization potential and reduction rate. Silver atoms and tin atoms derived from the redox reaction formed the intermetallic compound (Ag3Sn) through the mechanical stirring. Because of the low silver content, some of the tin atoms did not combine with silver atoms and formed the β-Sn phase.

The reduction of Sn2+ ions continued till the completion of the reaction until the Sn3.5Ag alloy nanoparticles were obtained.

3.2. Formation Process of the Primary Nanoparticles of the SnAgCu Alloy System

The influence of Cu added to the SnAg alloy for the formation process of nanoparticles was studied in this section. Different reaction times were chosen, as shown in Table 2, and the corresponding XRD results were shown in Figure 3.

Figure 3 indicated that the products formed in the initial 15 min were mainly Ag3Sn, Cu6Sn5, and β-Sn. Only 5 diffraction peaks of β-Sn emerged in this sample. Sn2+ ions in the solution were not sufficiently reduced at this reaction stage. After 30 minutes of reaction, the XRD pattern of the SAC2 sample showed no obvious difference with that of SAC1, except for the change in the maximum relative intensity of the β-Sn phase from (200) to (101). When the reaction duration was set to 45 min, the diffraction peaks of β-Sn emerged, as shown in Figure 3(c). The relative intensity of the diffraction peaks of Ag3Sn and Cu6Sn5 decreased further. There was no significant change in the XRD patterns between SAC4, with a reaction time of 60 min, and SAC3.

As for the Sn3.0Ag0.5Cu alloy, the standard redox potential of Ag/Ag+, Cu/Cu2+ and Sn/Sn2+, was 0.8 V (versus SHE) [24, 33], 0.339 V (versus SHE) [35, 36], and −0.14 V (versus SHE) [34], respectively. So, the oxidizing abilities of the metal cations in order were Ag+, Cu2+, and Sn2+. The reduction reaction of Cu2+ followed (6), and Cu atom formed through the electrons transmission from the reducing agent to the Cu cation

It was thought that the Ag-Cu nuclei formed because the standard redox potentials of these two species were similar. The Sn atoms reduced in the solution adsorbed on the surface of Ag-Cu nuclei, which acted as the core, and formed a shell through the diffusion mechanism. The omission of Cu in the XRD pattern was explained through crystallography, which showed that the Cu atoms dissolved in the Ag matrix and formed a solid in solution.

In order to understand the origin of the difference, the synthesis conditions and the mechanism of nucleation and growth were considered. Firstly, the solubility and ionizability of the reactant AgNO3 in anhydrous ethanol solution were much smaller than those of aqueous solution [2427], and that would lead to the relatively low reaction rate (the relative intensity of the XRD pattern of Sn phase was strong enough when the reaction time was 10 seconds [21], compared with that in this paper), so the metal cations could be reduced to atoms at a relatively slow rate, compared with that in aqueous solution. Secondly, Ag and Cu in their metallic form could form nuclei separately in the initial stage of the reaction. There were three main nucleation or growth mechanisms in the colloidal system. The classic Lamer mechanism [37] proposed the idea that fast nucleation and slow growth took place in the supersaturated solution. The growth process was controlled by diffusion. However, this mechanism was considered to be suitable to the sulfur sols and other analogous systems [38]. The Finke mechanism [38], which was reasonable for the unsaturated solutions, showed that nanoparticle formation was dominated by low, continuous nucleation and fast, autocatalytic surface growth. The last growth mechanism was described as the seeded growth [39]. The seeds, which act as the nuclei for the growth process, could be either the same or different composition as the absorbed atoms. This growth mechanism was used to fabricate the core-shell structures in the bimetallic alloy system [40]. In the case of the present work, when the reducing agent was added into the precursor solution, the nucleation and growth process of intermetallic compound nanoparticles could be described by the Finke mechanism.

Luo et al. synthesized ZnPd nanoparticles through a chemical reduction method and found that the coreduction of Zn2+ and Pd2+ could happen when the reducing agent was in excess [41]. The conditions were similar in this work, and the metal cations (Ag+, Cu2+, and Sn2+) with different reduction potentials could be reduced simultaneously. Meanwhile, the reduction rate of Ag+ and Cu2+ was faster than that of Sn2+ due to their reduction potentials [32]. There were more Ag and Cu atoms than Sn atoms in the solution at the initial reaction stage, and as a result of this, the coreduction of intermetallic compounds, that is, Ag3Sn and Cu6Sn5, could take place. Due to the relatively low content of Ag and Cu, the Sn in the solution was not consumed in the formation process of the intermetallic phase, and the Sn phase in the XRD was made up of these Sn atoms. The growth of the formed Sn nuclei might combine the autocatalytic surface growth mechanism and homogeneous seeded growth mechanism.

In addition, the surfactant used in our experiment, which was different from that used in literature, may play an important role in the detection of Cu. The capping effect of 1,10-phenanthroline on the formation of metal nanoparticles has already been well documented [4253]. The passivation layer could effectively prevent the growth of the primary particles or the formation of the secondary particles, that is, aggregation and Ostwald-Ripening process [54, 55]. A typical SEM image of the SAC nanoparticles (SAC4) synthesized at room temperature with a reaction time of 60 min was shown in Figure 4. The particles were basically spherical with a narrow diameter distribution of  nm. Despite the low content of Cu in the core-shell structure, the relatively small particle size could make it detectable. The formation process of SAC nanoparticles was illustrated in Figure 5.

There was another question concerning whether the Ag-Cu nuclei could actually be formed. Hume-Rothery rules were employed to prove the existence of nuclei of this specie in the literature [21]. In fact, the Ag-Cu system was almost absolutely immiscible at room temperature, which violated the Hume-Rothery rules [56, 57]. Moreover, the lattice constant of the Ag-Cu solid solution should be smaller than that of Ag owing to the lattice contraction caused by the substitution of Cu atoms to Ag atoms. However, the contrary result was obtained when the Ag-Cu nanoparticles were annealed at 300 K [58]. When the temperature increased to 493 K, the diffusion of Ag and Cu was rather difficult, which indicated poor solubility [59]. A study on the equilibrium solubility of Cu in Ag showed that the substitution of a solid solution could be formed only when the samples was annealed at temperature near 653 K [60]. So, the Ag-Cu nuclei and the following (Ag, Cu)3Sn phase were not recommended considering this result. One of the possible reasons for the disappearance of Cu in Duh’s work was the relatively large agglomeration rate of Sn atoms at the surface of Cu nuclei. The Sn shell in the nanoparticles was too thick for the X-ray to detect the Cu phases. Other reasons should be taken into consideration. For example, the aggregation took place during the growth process because of the absence the surfactant acting as a protective medium.

In summary, the formation process of SAC nanoparticles could be divided into three stages, similar to that of SnAg nanoparticles. First, the metal salts were mixed in anhydrous ethanol. The mixture was intensely stirred for 2 hours to produce a uniform precursor solution. Second, black particles precipitated when the reducing agent was added into the precursor. Most of the Ag and Cu cations were reduced to their metallic form and formed nuclei, respectively. At the same time, a few Sn atoms were generated in the solution. With mechanical agitation, the nuclei of Ag and Cu acted as the heterogeneous seed, absorbed Sn atoms and formed the intermetallic phases, Ag3Sn and Cu6Sn5. The rest of the Sn atoms, that were not used in the formation process of the intermetallic phases could nucleate and grew into the Sn phase. The capping effect of the surfactant and the size of the synthesized nanoparticles can be well controlled. This stage can also be described by the following reaction equations:

Finally, the Sn cations in the solution continued to be reduced until the termination of the reaction, which can be described as (3).

4. Conclusions

A chemical reduction method was used to synthesize lead-free alloy nanoparticles for use as a solder. The nucleation and growth mechanism in the anhydrous ethanol solution was discussed, and due to the difference in the redox potential, the Ag and Cu cations were reduced faster than Sn cations. The limited solubility of the reactant in the solvent ensured that the redox reaction took place. The immiscible nature of Cu and Ag at room temperature meant that these two species nucleated through the Finke mechanism separately. The nucleus acted as the heterogeneous seed during the growth process of the intermetallic compound. The nucleation mechanism of the Sn phase was similar to that of Ag and Cu, while the growth followed the homogeneous seeded mechanism. Moreover, the surfactant could control the size of the synthesized nanoparticles during the growth process, through a capping effect. The relatively thin Sn shell surrounded the intermetallic core and the light aggregation of the primary nanoparticles made the Cu detectable in the XRD pattern.

Acknowledgment

This work is supported by the National Natural Science Foundation of China (Grant nos. 50971086 and 51171105).