Nanoscale characterization of 1D Sn-3.5Ag nanosolders and their application into nanowelding at the nanoscale

The Sn-Ag alloy nanowire solders were fabricated by dc electrodeposition into AAO and PC templates with about 50 nm diameter pores. The AAO templates were fabricated by anodizing 99.999% pure Al sheets in a 0.6 M oxalic acid electrolyte for 12 h at room temperature with continuous stirring. An amalgamation process was adopted to peel off the Al substrates. The peeled AAO templates were then immersed into 0.5 M phosphoric acid for 90 min to remove the barrier layers. A gold film (approximately 50 nm thick) that served as a working electrode was evaporated onto one side of the non-barrier layer AAO template. The electrodeposition was achieved in a three-electrode glass cell on a CHI630B electrochemical workstation at room temperature in which the reference electrode was a saturated calomel electrode (SCE), and the counter electrode was a platinum plate. The electrodeposition electrolyte is composed of 0.30 M Sn(CH₃SO₃)₂, 4.5 mM AgI, 0.60 M K₄P₂O₇, 1.0 M KI, 0.4 M TEA, 6.4 mM heliotropin (HT) and 2 g L⁻¹ hydroquinone. The K₄P₂O₇ is a complexing agent for Sn²⁺ and is also used to prevent Sn²⁺ hydrolysis and oxidation. The KI is a complexing agent for Ag⁺, which contributes to the success of the co-deposition of Sn and Ag, inhibiting the replacement between Sn and Ag. Both K₄P₂O₇ and KI also act as a conductive salt, which increases the conductivity of the plating solution. The TEA acts as an assistant brightener and can decrease cathodic polarization. The HT is a main brightener, and the hydroquinone is an antioxidant. The pH value of the electrodeposition electrolyte was adjusted to 5.5–6.0 by adding methanesulfonic acid. The voltammetric experiments (figure 1(a)) in which a Cu sheet was used as a working electrode were completed to determine the optimal electrodeposition potentials for the co-electrodeposition SnAg alloy. When the potential was swept to negative values, two cathodic peaks were observed at −0.606 V and −0.932 V. An EDX quantification analysis revealed that the atomic weight ratios of Sn and Ag at −0.606 V, −0.77 V and −0.932 V potential were approximately 4.5:95.5, 18:82 and 95.7:4.3, respectively; this indicates that peak A is the initial codeposition stage of the Sn-Ag alloy, and the Ag content within the Sn-Ag alloy decreases as the cathodic potential becomes increasingly negative. In this work, a −1.1 V potential was applied to obtain 1D Sn-3.5Ag alloy nanosolders, of which a typical curve of the time dependence of the observed current during the growth of Sn-3.5Ag alloy nanowires in an AAO template is shown in figure 1(b). The typical deposition time for each specimen was set to 30 min, and the electrodeposition electrolyte was continuously agitated throughout by a magnetic stirrer.

Zoom In Zoom Out Reset image size

The morphological, crystal structural and chemical characterisations of 1D SnAg nanosolder were analyzed at the nanoscale using field emission scanning electron microscopy (FESEM, Hitachi S-4800 II, Japan), a high-resolution transmission electron microscope (HRTEM, TecnaiTM G2 F30, FEI, US) equipped with energy-dispersive x-ray analysis (EDAX, AMETEK Co., LTD), high-angle annular dark-field and scanning transmission electron microscopy (HAADF-STEM) and x-ray photoelectron spectroscopy (XPS, Kratos Axis Ultra DLD, Japan). The AAO templates or PC templates were completely dissolved in a 0.1 M sodium hydroxide solution or in dichloromethane to release the nanowires, which were then rinsed with distilled water and dispersed on Si wafers and holey carbon-coated gold grids for SEM, TEM, STEM, EDX and XPS analysis.

The quality tests performed by applying individual 1D SnAg nanosolders into welding nanopatterns at nanoscale size were carried out using two in situ SEM nanomanipulators (Kleindiek Nanotechnik, Germany) equipped with nichrome nanoprobes/nanotips and a Keithley 6487 picoammeter. The data acquisition and nanowelding signal were automated using a dedicated multifunctional program.

The nanowire solder reflow experiments were carried out in a programmable high-temperature tube furnace using the standard industrial reflow profile strategy, which includes preheating, a thermal soak, reflow and cooling steps. The solder nanowires were dropped onto Si wafers and holey carbon-coated copper grids. A glass of no-clean liquid soldering flux was used to clean up the surface oxides of the nanosolders during the reflow experiments, which were placed to the nearby 1D Sn-3.5Ag nanosolders. Nitrogen purging was used to avoid further oxidation of the nanosolders during this process. The temperature was firstly increased to the approximate range of 150–180 °C for 5 min to preheat the wires; then, the temperature was briefly raised to 240 °C for 5 min to ensure a complete reflow. The Si wafer was then brought out of the furnace and cooled down to room temperature.

The morphologies of 1D Sn-3.5Ag nanosolders were observed by SEM and HAADF-STEM. Figure 1(c) shows a representative SEM image of a bundle of flower-like SnAg solder nanowires sticking to a gold template replica at the end, from which the AAO template have been completely dissolved. One can see that the individual SnAg nanowires have a continuous structure and virtually uniform diameter. The size statistics reveal that the average diameter of the prepared SnAg nanowires is approximately 50 nm, which corresponds to the pore diameter of AAO templates, as shown in the inset of figure 1(c). The average length for most of the nanowires is around 10 μm. As shown in figure 1(b), the time of the nanowire growth in the pores (the stable stage) is approximately 1500 s; so, the growth rate of the SnAg nanowires is estimated to be 6.7 nm s⁻¹.

Figure 1(e) shows a representative HAADF-STEM image. The contrast of incoherent high-resolution HAADF-STEM images directly depends on the atomic number and thickness of the materials; in the nanowire images, the SnAg nanowires show a continuous structure and uniform size, which is in agreement with the SEM observation. Their surfaces are smooth and do not show obvious defects, revealing good quality. The magnified HAADF-STEM image (inset of figure 1(e)) shows that individual SnAg nanowires have an even contrast, revealing an even chemical composition along the nanowire's length. Figure 1(d) shows a representative SEM image of a field of wheat-like Sn-Ag nanowires prepared from PC template. The magnified HAADF-STEM image (figure 1(f)) shows that individual SnAg nanowires have uneven diameter-like cylinders. The size statistics reveal that the top diameter is approximately 50 nm, and the bottom diameter is approximately 120 nm. The average length for most of the nanowires is around 5 μm.

Individual SnAg nanowires have been used in the application of welding nanopatterns at the nanoscale to test their quality as nanosolder. Figure 2 illustrates the whole process. Two gold nanowires that are 55 nm in diameter and about 6 um in length were mechanically manipulated and assembled into a simple nano-pattern such as Chinese number eight '八' on a SiO₂(100 nm)/Si wafer; then, a sacrificial Sn-3.5Ag nanowire 100 nm in diameter and 2 um in length, which was used as nanosolder (figure 2(a)), was placed onto the assembled '八' nanopattern at the weld point, as shown in figure 2(b). Before welding, the resistance of the assembled nanostructure was measured as isolating due to the structural disconnection.

Zoom In Zoom Out Reset image size

Prior to welding, a moderate current (about 75% of the maximum current density of a single 120 nm SnAg nanowire) was applied to soften the SnAg solder nanowire for 1–2 min, which was done to induce the first join to the pattern (figure 2(c)). The softening of the sacrificial solder nanowire is due to Joule heating and can greatly improve the controllability and reliability of the nanowelding procedure. A linear pulse −1.5 V signal 300 ms in width was then run through the solder to utterly form the weld. The short voltage pulse causes significant Joule heating and associated rapid material diffusion of the solder nanowire onto the chosen junction (figure 2(d)).

The immediate check of the nanoweld's quality, as shown by the green curve in figure 2(f), reveals a 116 Ω resistance of the whole welded structure before the 55 Ω calibrated tip-to-tip contact resistance is subtracted. As measured in our previous report [25], the resistivity of the 55 nm Au nanowire is 2.26 μΩ·cm. The resistance of the gold nanowire for the same length as the whole welded structure is theoretically calculated to be 57.1 Ω (27.8 Ω for nanowire A and 29.3 Ω for nanowire B in figure 2(e)). Therefore, the resistance of the welded junction is only 3.9 Ω, which reveals that a very good ohmic contact was formed. This nanoweld resistance is quite lower than that of 1D Sn₉₉Au₁ and pure Au nanosolders, which were used in our previous work [6]; this indicates a better wettability property than the nanosolders of Sn₉₉Au₁ and Au. In addition, the 1D SnAg nanosolder also shows a better mechanical strength, controllability and reliability during the nanowelding process. The mechanical strength of the SnAg solder nanojoint is strong enough to enable welded structures to be directly lifted by a SEM nanomanipulator. The whole welding processes and quality checks are similar to those in our previous report [6]. This result proves that 1D SnAg alloy nanosolder is a reliable and high-performance nanosolder material when applied in nanoscale size.

The electrical properties of 1D SnAg alloy nanosolder were also characterised. The resistance of the sacrificial SnAg solder was measured at 110 Ω, as shown in figure 2(f) (blue curve). This experimentally measured resistance includes the 1D SnAg solder's intrinsic resistance (R₀), the resistance of all contacts (R₁) and the feed line resistance of circuitry (R₂). The calibrated tip-to-tip resistance should be the sum of the resistance of all contacts (R₂) and the feed line resistance of circuitry (R₃). According to our previous experiences [6, 25], the values of the contact resistance between the nichrome tips and the 1D SnAg solder were not found to be detectably larger than that of the tip-touching-tip during our experiments; so, we take the simple treatment that the contact resistance of the tips/nanowire is equal to that of the tip-touching-tip. Therefore, the intrinsic resistance of this 1D SnAg solder is determined to be 55 Ω after the calibrated tip-to-tip contact resistance (55 Ω, the red curve in figure 2(f)) is subtracted.

As determined by the morphologicial and structural characterisation, the individual 1D SnAg solders prepared in this work are polycrystals and possess numerous grains with an average diameter of about 20 nm. Considering this surface scattering, the grain boundary and the impurity scattering, In order to accurately analyse their resistivity, Matthiessen's rule [26], which combines the Dingle [27] and MS [28] models, was adapted to analyse the intrinsic resistivity of our individual 1D SnAg solders under κ ≫⃒ 1:

$$\begin{eqnarray}&&\begin{array}{lllllllllllllll} \rho ={{\rho }_{0}}\left[ 1+\frac{3}{4\kappa }\left( 1-P \right) \right. \\ \quad \quad \left. +\frac{1}{1-\frac{3}{2}\alpha +3{{\alpha }^{2}}-3{{\alpha }^{3}}\;{\rm ln} \left( 1+\frac{1}{\alpha } \right)} \right] \\ \end{array}\end{eqnarray} \tag{ 1 }$$

$$\begin{eqnarray}&&\kappa =\frac{d}{\lambda }\end{eqnarray} \tag{ 2 }$$

$$\begin{eqnarray}&&\alpha =\frac{\lambda }{D}\frac{R}{1-R}\end{eqnarray} \tag{ 3 }$$

where ρ₀ is the resistivity of the bulk material, and 12.3 μΩ·cm is the value for the SnAg alloy [29]; P is the fraction of electrons scattered specularly at the surface, which takes values between 0 and 1; d is the diameter of the wire; λ is the mean free path of the bulk material, and 5.366 nm is the value for the SnAg alloy [30]; R denotes the scattering coefficients at the grain boundary, which takes values between 0 and 1; D is the mean grain size. In our case, the diameter and mean grain size of each individual 1D SnAg solder is 100 nm and 20 nm, respectively, as determined by the above and below morphogical analyses. As far as we know, there is no report on the fraction of the P electrons and the scattering R coefficients of the SnAg alloy nanowires. Steinhögl et al [31] reported that the P and R of the electrodeposited copper nanowire is 0.2–0.6 and 0.5, respectively. If we assume that the P and R of the 1D SnAg nanosolder is 0.5 and 0.5, a value of 28.9 μΩ·cm was obtained for the resistivity of the 1D SnAg nanosolder. This value is higher than the above standardized bulk value of 12.3 μΩ·cm [29]. Two main reasons were deduced to have caused this difference. Firstly, a layer of 4–7 nm SnO₂ oxide covers the outer surface of individual SnAg nanosolders, according to the detailed characterization of the 1D SnAg nanosolders below; this forms a metal-semiconductor contact and may cause fringing field effects on the electrical resistivity of the nanosolder. Secondly, there are at least three phases: β-Sn, Ag₃Sn and SnO₂ in a 1D SnAg nanosolder. Also, there are different grain sizes and crystallographic properties that may change from sample to sample. More detailed investigations of the electrical properties of 1D Sn_xAg_1 − x nanosolders are ongoing and will be reported soon. Nevertheless, this result shows that the electrical property of the 1D SnAg is good enough to be used as a high-quality nanoscale solder. The detailed electrical testing method of the individual nanowires is described in our previous report [6, 25]. In addition, the current versus voltage behaviors of both the probes and the nanowire circuit are linear, as shown in figure 2(f), which indicates remarkably low-resistance ohmic contacts.

The chemistry of the 1D SnAg nanosolder was characterized in detail by EDX, XPS and STEM mapping on a 300 kV HRTEM. Figure 3(b) shows a representative EDX spectrum of three 1D SnAg nanosolders, which are shown in figure 3(a). The Sn and Ag peaks come from the 1D SnAg nanosolders. The O peak should originate from the 1D nanosolder, which is deduced to be due to the partial oxidization of the 1D SnAg nanosolders. The C and Mo peaks are believed to derive from the holey carbon-coated molybdenum grid, which was confirmed by the measurement of an EDX baseline of an empty holey carbon-coated molybdenum grid. A quantitative analysis of this spectrum reveals that the chemical composition of individual 1D SnAg nanosolders is a 96.5:3.5 weight ratio of Sn:Ag, which is the same as the standardized eutectic phase of bulk SnAg alloy [32].

Zoom In Zoom Out Reset image size

The chemical distributions of the 1D Sn-3.5Ag nanosolder were subsequently studied by an EDX elemental line-scan and a mapping analysis technique. Figure 3(c) shows the EDX line-scan spectra of Sn-Kα/Lα and Ag-Kα/Lα taken across the 1D nanosolder, acquired using a 3.5 μm scan line parallel to the nanowire axis (as marked by the red line in figure 3(a)) and a 1.5 nm beam spot-size. We determined that the intensities of Sn and Ag are stable in error and are equal along the wire axes, revealing a uniform chemical phase. In addition, the intensity of Sn is higher than that of Ag, revealing that the chemical content of Sn is higher than Ag. Figures 2(d)–(f) further show the EDX elemental mappings of Sn-Lα (3.44 keV), Ag-Lα (2.98 keV) and O-Kα (0.52 keV), respectively. The elements of Sn and Ag are nearly evenly distributed throughout the whole nanosolder, indicating a uniform Sn-3.5Ag chemical phase. The even distribution of the oxygen element reveals that individual 1D Sn-3.5Ag solders are slightly oxidized after they are released from the AAO or PC templates.

In order to accurately determine the chemical states, the as-grown 1D SnAg nanosolders were further examined by XPS while they were still embedded in situ in the PC templates by Ar ion beam sputtering. The templates were cross-sectioned in parallel to the 1D SnAg nanosolder length axes by sputtering away (layer by layer) the top surface of the PC template. Figure 3(g) shows the relevant XPS spectra of Ag 3d and Sn 3d after Ar ion beam sputtering for 0 s, 50 s, 100 s, 150 s, 200 s, 300 s, 400 s, 500 s and 600 s, respectively; the corresponding depth profile analysis is shown in figure 3(h). The binding energies were calibrated using the C 1 s peak of 284.6 eV as the standard. The SnAg alloy 3d spectrum has two peaks because the spin–orbit of the 3d state is coupled with a spin-orbital separation of 8.5 eV for Sn and 6.0 eV for Ag [33]. For simplicity, only Sn 3d_5/2 and Ag 3d_5/2 were studied. The published binding energy of SnAg 3d_5/2 indicated that the peak, which was in the range of 484.5 eV−485.2 eV, is attributed to Sn, the range of 486.0 eV−486.8 eV is attributed to SnO, the range of 486.4 eV−486.9 eV is attributed to SnO₂, the range of 368.1 eV−368.3 eV is attributed to Ag and the range of 367.3 eV−368.0 eV is attributed to Ag oxides [33]. One can see that the Ag 3d_5/2 peaks at 368.3 eV bonding energy in figure 3(g) can be attributed to metallic Ag, and there is no peak shift before and after the Ar ion beam sputtering; this indicates a stability of the Ag chemical phase along with the nanowire length. The Sn 3d_5/2 spectra consist of two peaks located at 484.3 eV and 486.4 eV, indicating a formation of metallic Sn and SnO_x, respectively. A quantitative analysis of the Sn3d_5/2 spectrum before the Ar ion beam sputtering (purple spectrum) reveals the area fractions of Sn and SnO_x to be 24.8% and 74.2%. This result indicates that the top end of the 1D SnAg alloy solders, which were exposed to air, were partly oxidized. After Ar ion beam sputtering at 50 s, the peak of the Sn drastically increases, and a quantitative analysis indicates a 65.2:34.8 area ratio of Sn:SnO_x. The proportion of the Sn metallic state continues to grow over the sputtering time. After the sputtering time increased beyond 500 s, the quantitative analysis of the XPS wide spectrum revealed that the 1D SnAg nanosolders reached a stable 96.54:3.46 mass ratio of Sn:Ag and the oxygen peak could not be detected. This result is the same as the above EDX measurement in error. Therefore, the chemical composition of the 1D SnAg nanosolders in this work is inferred to be Sn-3.5Ag.

Figure 4(a) shows a TEM image of two single 1D Sn-3.5Ag nanosolders, which were exposed to air for 1 day after they were dispersed on the holey carbon-coated copper grid. One can see that a 5–7 nm oxide layer is covered on the surface, which is due to be oxidized and is consistent with the above EDX measurement. Due to the similar binding energies of SnO and SnO₂, it is impossible to experimentally distinguish them by XPS at this stage. However, it is known that the Gibbs free energy of formation for SnO₂ (−515.8 kJ mol⁻¹) is much lower than that of SnO (−251.9 kJ mol⁻¹) [34], which indicates that SnO₂ is more thermodynamically stable. Therefore, it is deduced that the observed tin oxide film covered on the outer surfaces of the two 1D SnAg nanosolders should be SnO₂.

Zoom In Zoom Out Reset image size

The crystalline structures of individual 1D Sn-3.5Ag nanosolders were characterized in detail using selected area electron diffraction (SAED), convergent beam electron diffraction (CBED) and high-resolution TEM (HRTEM). Figure 4(b) shows a magnified TEM image of the area marked by a red square in figure 4(a). Its corresponding SAED pattern is shown in figure 4(c), which reveals that the individual 1D Sn-3.5Ag nanosolder is a multi-crystal. In order to obtain clearer details of the crystalline phases in a 1D Sn-3.5Ag nanosolder, an HRTEM technique was employed. Figure 4(d)–(f) show lattice images of the areas marked by the square 'D', 'E' and 'F' in figure 4(b), respectively. The interplanar distances are measured as 2.37 Å and 2.67 Å for area D (figure 4(d)), which matches the (200) and (101) planes of tetragonal SnO₂; as 2.61 Å for area E (figure 4(e)), which matches the (002) planes of orthorhombic Ag₃Sn; and as 2.05 Å and 1.48 Å for area D (figure 4(d)), which matches the (−200) and (−112) planes of body-centered tetragonal (BCT) β-Sn.

To further verify the accuracy of the crystal measurements, CBED using a 0.5 nm spot size configuration was used to analyze these three detected areas. The corresponding CBED patterns of the 'D', 'E', and 'F' areas are shown in figures 4(g)–(i), which can be well indexed into the [010] crystallographic orientation of the tetragonal structure (inset of figure 4(g)), the [100] orientation of the orthorhombic structure (inset of figure 4(h)) and the [110] orientation of the body BCT structure (inset of figure 4(i)), respectively. These results are well matched with the HRTEM observations. We then concluded that the 1D Sn-3.5Ag nanosolder is comprised of matrix β-Sn and intermetallic compound Ag₃Sn, which is similar with the crystal phases in a bulk Sn-3.5Ag [32].

The melting point and melting behavior are the critical factors that determine the quality of a solder. In this work, we further investigated the melting point and melting behavior of the 1D Sn-3.5Ag nansolders. Figure 5(a) shows a representative DSC curve of 1D Sn-3.5Ag nansolders within the PC template, which was heated from 50 °C to 550 °C at 10 °C min⁻¹; the inset displays the magnified view of the red region in figure 5(a). The curve trend rises due to the presence of the PC template. Also, three peaks can be seen that correspond to the three melting points 215.5 °C, 220.7 °C and 390.6 °C. With consideration to the above crystal phase analysis and the Sn-Ag phase diagram [32], these three peaks could be attributed as melting points of eutectic β-Sn plus Ag₃Sn and monovariant Sn and Ag₃Sn, respectively. Considering that the eutectic β-Sn plus Ag3Sn is the dominant phase in the individual 1D Sn-3.5Ag nanosolders, as detected by the above chemical and structure analyses, we concluded that the melting point of the 1D Sn-3.5Ag nanosolders is about 215.5 °C. This value is lower than the bulk Sn-3.5Ag solder of 221 °C [32], which indicates that the 1D Sn-3.5Ag nanosolders prepared in this work have good flow and wettability. This result confirms that the 1D Sn-3.5Ag nanosolders can be used as a high-performance nanoscale solder. The melting behavior of the tin oxide shell was not reflected in the curve due to the melting point of the SnO and SnO₂, which are 1080 °C and 1630 °C, respectively. Therefore, it was deduced that the eutectic composition of the 1D Sn-3.5Ag naosolder should be shifted to the Sn rich corner, which is in agreement with the study of Sn-Ag nanoparticles [35, 36].

Zoom In Zoom Out Reset image size

To further understand the melting behavior of the 1D Sn-3.5Ag nanosolders, reflow experiments were carried out. Figure 5(b) shows a representative SEM image of a bundle of 1D Sn-3.5Ag nanosolders after preheating at 150 °C–180 °C. One can see that the nanowires nearly retain their original shapes but stick together more closely. The magnified image reveals the formation of a few protuberances along the naowire due to the movement of the molten Sn. Figure 5(c) is a representative SEM image of the 1D Sn-3.5Ag nanosolders after they were reflowed in the N₂ atmosphere under the usage of no-clean liquid flux, which shows a dramatic morphological change from a wire shape to a spheroid shape due to the minimized surface energy at the liquid phase. The size of the reflowed Sn-Ag solders varied from hundreds of nanometers to several microns; this is is believed to be due to the random and nonuniform distribution of the nanowires and the merging of the liquid droplet. The inset of figure 5(c) shows a magnified view of one sphere, revealing that there is a translucent thin film covered on its surface.

The chemical distributions of the reflowed Sn-3.5Ag nanosolder were subsequently studied by the EDX elemental mapping technique using 300 kV HRTEM. Figure 5(d) shows a HADDF-STEM image of 1D Sn-3.5Ag nanosolders after complete reflowing; this image clearly reveals a cluster of spheres enclosed in translucent thin shells, as observed above. Figures 5(e)–(i) shows the corresponding EDX elemental mappings of oxygen, silver and tin. One can clearly see that oxygen exists on the whole structure, indicating that the surfaces of the reflowed solders have been oxidized; these solders may be formed during the cooling or storage process after reflowing. The Sn mappings (figures 5(i) and (h)) show a uniform distribution in each sphere. However, the Ag (figures 5(f) and (g)) appears segregated in the sphere. One can derive this result from the lower cooling rate (naturally cooled to room temperature), which prolongs the solidification period and thus makes Ag₃Sn IMCs grow into a massive block [37].