Ambient Aqueous-Phase Synthesis of Copper Nanoparticles and Nanopastes with Low-Temperature Sintering and Ultra-High Bonding Abilities

Copper nanoparticles (NPs) with an average particle diameter of 50–60 nm were successfully obtained by reducing an aqueous solution of a copper(II)-nitrilotriacetic acid complex with an aqueous hydrazine solution at room temperature under an air atmosphere. Copper NP-based nanopastes were printed onto a glass substrate using a metal screen mask and pressureless sintered under a nitrogen atmosphere at 200 °C for 30 min. The electrical resistivity of the resulting copper electrode was 16 μΩ · cm. For a metal-to-metal bonding test, copper nanopaste was printed on an oxygen-free copper plate, another oxygen-free copper plate was placed on top, and the bonding strength between the copper plates when pressureless sintered under a nitrogen atmosphere at 200 °C for 30 min was 39 MPa. TEM observations confirmed that highly crystalline metal bonding occurred between the copper NPs and the copper plate to introduce the ultrahigh strength. The developed copper NPs could provide promising advances as nanopastes for sustainable fabrication of copper electrodes and die attachment materials for the production of next-generation power semiconductors.

Copper electric circuits on substrates in smart devices, such as tablet PCs and mobile phones, are widely manufactured by a photolithography process. This process is advantageous for fabricating large-scale, high-quality integrated circuits. However, the multistaged masking and etching process requires expensive production systems and the use of harmful and environmentally undesirable chemicals. On the other hand, printed electronics (PE) technology 1-6 has attracted a great deal of attention as an alternative eco-friendly and low-cost process to write electric circuits on flexible substrates with metal inks. Recently, a combination of PE and integrated circuit (IC) production technologies, flexible hybrid electronics (FHE) [2][3][4][5][6][7][8][9][10][11] , is a promising technology for flexible and wearable smart devices because the FHE-based on-demand and eco-friendly process allows the production of a wide variety of devices based on market needs. For practical use of PE and FHE technologies to fabricate on-demand devices, bonding materials that are applicable to flexible substrates are needed. Currently, solder alloys and conductive epoxy polymers have been widely applied as materials. On the other hand, metal inks and pastes have considerable potential as bonding materials to fabricate next-generation power semiconductors (SiC and GaN) and as alternatives to solder alloys for electronic parts in automobiles [12][13][14] . Nano-and micron-sized silver particles have been widely utilized in inks and pastes due to the high electric and thermal conductivity and low-temperature sintering property of silver. For example, intermetallic bonding with silver paste used for sintering is carried out at a low temperature of 300 °C or less, and the obtained bond strength between metals reaches 30 MPa or more [15][16][17][18][19] . To ensure the operational stability of next-generation power semiconductors and automotive electronic components, which are expected to be used in high-temperature environments (200 °C), intermetallic bonding materials are required to have high heat dissipation, high thermal resistance, and low resistance [20][21][22] . The thermal conductivity of bulk silver is 420 W·m −1 ·K −1 , and the advantage of silver has resulted in

Results and Discussion
Ambient aqueous-phase synthesis of copper Nps. To obtain copper NPs with a mean particle diameter of several tens of nanometers, we focused on a reduction of an in situ-prepared organocopper reagent in an aqueous-phase system. After optimizing the reaction conditions, a reduction method enabled us to prepare a large quantity of specifically designed copper NPs with precisely controlled particle size and shape in an atmospheric aqueous solution system. The established reaction procedure is as follows. Initially, a water-soluble copper precursor (copper(II) ions: 0.27 mol/L) was prepared by mixing nitrilotriacetic acid disodium salt (NTA H 2Na) and Cu(OH) 2 powder in water at room temperature with stirring. Then, an aqueous solution of hydrazine monohydrate (4.9 mol/L) was rapidly added in one portion under atmospheric conditions to the precursor solution to obtain copper NPs. The resulting copper NPs were purified by dispersing them in water after centrifugation. To investigate the effect of the [NTA H 2Na]/[Cu(OH) 2 ] molar ratio on the copper NP preparation, the ratio was changed in the range from 0 to 2.4. Figure 1(a,b) show the X-ray diffraction (XRD) patterns and scanning electron microscopy (SEM) images, respectively, of the solid particles that were formed by the hydrazine treatment with different molar ratios of [NTA H 2Na]/[Cu(OH) 2 ] ((i) 0, (ii) 1.69, (iii) 2.0, and (iv) 2.4). All diffraction peaks in Fig. 1(a,i) can be assigned to the formation of an fcc-type metallic copper crystal structure (JCPDS No. 00-4-836), indicating the obtained copper particles were in a single phase. In contrast, when NTA H 2Na was used as the chelating agent, metallic copper was the predominant phase, and a slight amount of Cu 2 O phase (JCPDS No. 00-5-667) was observed ( Fig. 1(a)(ii)-(iv)). The diffraction peaks due to the Cu 2 O phase became stronger as the   Fig. 1(b). The copper particles obtained in the absence of NTA were submicrometer in size with an irregular, coagulated shape ( Fig. 1(b)(i); these particles are abbreviated as Cu1). Based on the XRD patterns ( Fig. 1(a)(ii),(iii), and (iv)), the particles with a uniform spherical shape, as seen in Fig. 1(b)(ii),(iii), and (iv) and abbreviated as Cu2, Cu3, and Cu4, respectively, could be assigned as metallic copper NPs. The results suggest that NTA plays an important role in controlling the size and shape of copper NPs, and Cu1, which was obtained without using NTA, was not applicable for copper nanopastes due to its large diameter. Characterization of the frontier surfaces of Cu3 by diffuse reflectance infrared Fourier transform (DRIFT) and X-ray photoelectron spectroscopy (XPS) methods revealed that NTA adsorbed on the surfaces, and thermogravimetric and differential thermal analysis (TG-DTA) indicated 2.1 wt% NTA was present (see, supplementary information (SI)). Yonezawa et al. reported that a small amount of low-molecular-weight gelatin on a Cu 2 O layer on the surface of copper NPs prevented further oxidation of the copper NPs 36 . Based on our results, the NTA on the surface of the copper NPs also prevents oxidation of the copper NPs in the present study. Next, the average particle diameter (D ave ) and standard deviation (σ) of Cu2, Cu3, and Cu4 were determined by counting more than 150 NPs in the corresponding SEM images, and the diameters were 75.1 (13.0) nm, 59.2 (12.6) nm, and 51.4 (8.3) nm, respectively. The diameters decreased with an increase in the [NTA H 2Na]/ [Cu(OH) 2 ] molar ratio. The decrease in diameter resulted in an increase in the surface area of the copper NPs. The increase in the surface area might accelerate partial oxidation on the surfaces under atmospheric conditions. Furthermore, the surface roughness of the copper NPs (Cu2-Cu4) might also enhance partial oxidation, resulting in the formation of the Cu 2 O phase on the surface of the copper NPs. The increase in the peak intensities of the Cu 2 O phase with the decrease in D ave observed in Fig. 1(a)(ii)-(iv) was in good agreement with the oxidation behavior on the surfaces. Here, the initial crystallite sizes (S i ) of Cu2, Cu3, and Cu4 before sintering were calculated using the half width of the (111) diffraction peak and Scherrer's equation 46 and were (ii) 15.5 nm, (iii) 14.0 nm, and (iv) 11.8 nm, respectively. The large difference between D ave and S i strongly suggests that the present copper NPs have an internal polycrystalline structure. Figure 2 presents the characterization results for the internal structure and surface state of Cu3 obtained using a high-resolution transmission electron microscope (HR-TEM) and a high-angle annular dark field scanning TEM (HHADF-STEM) equipped with an energy dispersive X-ray spectroscopy (EDS) system. Figure 2(a) shows the HR-TEM images of Cu3, and lattice fringes and grain boundaries due to the formation of polycrystalline structures are clearly observed inside the particles. The image also suggests the existence of a rough shell structure on the surface of Cu3 with a thickness of ca. 5 nm. Based on Fig. 1(a)(iii), the Cu3 copper NPs contain some Cu 2 O phase. The detailed observation of the thin layers on the surfaces of Cu3 revealed some lattice fringes in the layer, and the lattice interval of 0.248 nm could be assigned as the (111) plane of the Cu 2 O phase ( Fig. 2(b)). Rietveld analysis of the XRD ( Fig. 1(a)(iii)) pattern of Cu3 revealed that that the weight ratio of the Cu 2 O/Cu was assigned as 17/83. The above results indicate that Cu3 has a copper metal core and Cu 2 O shell structure, and NTA molecules were adsorbed on the Cu 2 O shell layer to prevent oxidation of the copper metal core. The STEM-EDS images shown in Fig. 2(e,f) are the mapping images of carbon and oxygen atoms, respectively. Both elements are distributed on the surfaces of the NPs and can be recognized as carbon atoms in the NTA molecules and oxygen atoms in the Cu 2 O phase and the NTA molecules. The images also support the formation of a unique NTA-and Cu 2 O-covered polycrystalline structure. This unique structure is the possible reason for the high oxidation resistance, long-term stability, and low-temperature sintering, and ultrahigh bonding abilities of the copper nanopastes mentioned in the following section. Details on the extensive efforts to clarity the effects of the chelation agent, reductant, and initial pH on the copper NP synthesis applicable to the present copper NP-based pastes will be reported elsewhere by the authors.
sintering behavior of the copper Nps. To investigate the pressureless sintering ability of the present copper NPs, temperature-variable XRD measurements were carried out with pressureless heating. The experimental details are summarized in the Methods section. The copper (111) diffraction peaks of Cu1 and Cu3 at 30, 100, 200, 300, 400, and 500 °C are summarized in Fig. 3(a,b), respectively. Here, the peak shift to lower angles was due to thermal expansion of the crystal lattice with the increase in the temperature. For Cu1, the half width of the (111) diffraction peaks slightly sharpened with the increase in the heating temperature. The lattice intervals of the (111) planes at 30 °C and 500 °C were calculated as 0.209 nm and 0.211 nm, respectively. The expansion was good agreement with the reported thermal expansion of Cu 47 . In contrast, the corresponding half width of Cu3 drastically narrowed under the same treatment, and the C s of Cu3, which has a much smaller D ave than Cu1, increased from 14.0 nm to 62.1 nm as the temperature increased from 30 °C to 500 °C. To clarify this behavior, Fig. 3(c) summarizes the relationship between the C s ratio, S T /S i , and the sintering temperature ( T ). Here, S T represents C s at T °C. The temperature range was 30 °C to 500 °C, and the heating rate was 1 °C/min. The interval to measure the XRD profile was 10 °C. All the XRD patterns were taken after holding the samples at T °C for 1 min. In the case of Cu3 (S i of Cu3: 14.0 nm), a rapid increase in the S T /S i value was observed at ca. 170 °C, and the S T /S i value gradually increased from 200 to 300 °C. It is worth noting that such a rapid increase in the value at ca. 170 °C was not observed in the case of Cu1 (S i of Cu1: 37.0 nm). In addition, a further increase in the sintering temperature above 300 °C resulted in an increase in the S T /S i values for both Cu3 and Cu1. In previous studies 28 , copper pastes were heat-treated at temperatures higher than 300 °C to fabricate the corresponding sintered copper electrodes by adhering copper particles through decomposition of organic modifiers on surfaces and grain boundary diffusion. Based on the results shown in Fig. 3(c), the increase in the S T /S i values in the range from 170 °C to 300 °C, which was only observed for Cu3, could indicate growth of C s by diffusion on the rough surfaces of Cu3 to connect the Cu3 NPs. This behavior, that is, the increase in the S T /S i value of Cu3 at low temperature, is the most plausible reason for the remarkable low-temperature pressureless sintering ability of the Cu3-based nanopastes, as mentioned below.
sintering behavior of the copper nanopaste and resistivity of the resulting copper electrodes.
For the nanopaste copper NPs, Cu3 was used due to its low-temperature sintering property. Details about the preparation of the copper nanopaste are shown in the Methods section. To investigate the sintering characteristics, the copper nanopaste was printed on a glass substrate with a thickness of 55 µm and pressureless sintered at 180, 200, 230, and 260 °C for 30 min under a nitrogen atmosphere to obtain copper electrodes. Figure 4 summarizes the printing process of the copper electrodes and the characterization results for the copper electrodes on a glass substrate. A schematic illustration and photos of each step in the preparation of the copper electrodes using the copper nanopaste are shown in Fig. 4(a). As shown in Fig. 4(a) on the right, a metallic-colored   Fig. 3(c)). Note that there is a large difference between the S pT /S i of Cu3 and S T /S i of Cu3 values at their respective T values. The results clearly indicate that the low-temperature pressureless sintering ability to grow crystallite Cu3 was drastically enhanced by the pasting. The SEM images show that an increase in the sintering temperature enhanced the melting of the surfaces of the NPs, which was followed by connections forming between the NPs. The images also shown that the domain sizes increase with the increase in the sintering temperature. However, the domain sizes in the SEM images were much larger than the corresponding S pT because all the S pT values in the range from 180 to 260 °C were several tens of nanometers. This result indicates that each domain in the copper electrodes still has a polycrystalline structure. The results also support the conclusion that the sintering observed in the present study mainly proceeds by the diffusion of copper atoms on the rough surfaces of Cu3 to connect the Cu3 NPs. Controlling the surface morphology of metal pastes might be a powerful technique to introduce a low-temperature sintering ability. Next, the sintering behavior of the Cu3-based nanopastes was compared with the corresponding Cu3 powder. As mentioned above, Cu3 has a Cu 2 O shell on its surface. On the other hand, the copper nanopastes contain triethanolamine (TEA) as a dispersion medium. Here, TEA reduced the heating requirement 48 . In the case of the present copper nanopaste, reduction of the Cu 2 O shell on the surface of the copper NPs by the TEA in the nanopaste might strongly enhance the sinterability of these copper NPs at low temperatures.
preparation of electric circuits on flexible films using the copper nanopaste. Copper nanopaste-based electric electrodes on polyimide (PI) and polyethylene terephthalate (PEN) flexible films were printed by the procedure shown in the Methods section. Figure 5(a,b) show photos of the copper electrodes on PI and PEN flexible films, respectively, which were prepared by screen printing and pressureless sintering at 200 °C under a N 2 atmosphere for 30 min. Cracks and detachment of the electrodes from the films were not observed after repeated bending of the films (also see, Fig. S4 in SI), and the copper nanopaste was found to be a widely applicable wiring material for not only on a glass substrate but also plastics films. Next, copper electric circuits were screen printed on a PI film using a fine pattern screen mask. The line (L) and space (S) of the mask, L/S, were 40/40 μm/μm. The printed films were pressureless sintered at 200 °C under a N 2 atmosphere for 30 min to obtain conductive circuits. The resulting L/S values and thickness of the copper electric circuits after sintering were measured by an optical microscope. Figure 5(c) exhibits an optical microscopic image of the copper electric circuits. Uniform and straight copper electrodes were fabricated on the PEN film. The average L/S could be assigned as 50/30 μm/μm. The 3D and depth profiles are also shown in Fig. 5(d,e), respectively. The average height was 8 μm, and thick copper electrodes were readily obtained by a simple printing process.
share strength of die attachment using the copper nanopaste. To investigate the performance of the Cu3-based copper nanopaste as a die attach material, the copper nanopaste was printed on an oxygen-free copper substrate (5 × 5 × 1 (mm)) in a 1 mm square using a metal screen mask, and then, an oxygen-free copper plate (3 × 3 × 1 (mm)) was mounted on the paste as a model IC chip. The resulting IC-mounted model substrate with a die attach structure was pressureless sintered at 180 °C, 200 °C, and 260 °C for 30 min under a N 2 atmosphere. Figure 6(a) illustrates a schematic and images of the preparation of the mounted die using the copper nanopaste and the die share stress measurement procedure. The resulting share strength, which was measured by an XYZTEC bond tester, was 25 MPa, 39 MPa, and 38 MPa, respectively. The share strengths at 200 °C and 260 °C were approximately the same, indicating 200 °C was a sufficient temperature to ensure a stable adhesion between copper substrates. It is worth noting that the share strength observed here is much higher than 35 MPa, and copper nanopastes with high performance by pressureless sintering at 200 °C have not been reported thus far [15][16][17]49 . This result means that the present copper nanopaste has an ultrahigh bonding ability and is practically applicable as a die attach material with a low-temperature sintering ability. Figure 6(b-e) shows the results of the HAADF-STEM observations of the bonding state between the copper NPs and the copper substrate. The inter-fringe distance of 0.206 nm in Fig. 6(d) was assigned as the distance of the (111) planes of the metallic copper phase. The results confirmed that metallic bonding with uniform lattice fringes spontaneously occurred between the copper particles and the copper substrate by sintering despite the absence of a reducing gas, such as hydrogen or formic acid (Fig. 6(d)). Although the remarkable low-temperature pressureless sintering mechanism to form single-crystalline metallic bonding at 200 °C under a nitrogen atmosphere was unclear, a highly crystalline, connected structure with uniform lattice fringes between the copper NPs and the copper substrate was the reason for the unprecedented ultrahigh bonding property with a strength of 38 MPa achieved in the present study.

Methods
Reagents. Unless    was further stirred for 1 h at the same temperature. The obtained dark purple solid particles and the aqueous phase were separated by centrifugation (10,000 G, 30 min), and the sediments were washed two times with water and three times with ethanol by dispersal and centrifugation to collect the copper NPs as an ethanol slurry (70 wt.% Cu). For the characterization of the copper NPs, the slurry was dried under reduced pressure to obtain the copper NPs as a powder.
preparation of the copper Np-based pastes. The copper NP-based pastes used in the present study were prepared as follows. TEA (5.29 mL) was added to the obtained ethanol slurry of copper NPs (30.0 g) at room temperature. The ethanol in the mixture was completely removed under reduced pressure with stirring. Then, to prevent to form cracks, a methanolic solution of 3-glycidoxypropyltrimethoxysilane (63 wt.%, 2.18 mL) was added to the mixture to obtain the copper NP-based pastes (80 wt.% Cu). The obtained copper paste was stable at room temperature for 1 day and was confirmed to be stable for 1 year by cryopreservation. The viscosity of the copper paste was 45 Pa·s (10/s), 7 Pa·s (100/s). preparation of a copper-bonded body for the share strength measurement. The oxygen-free copper plate (i-ject Co., Ltd., 5 × 5 mm; t = 1 mm) was polished with a rotary polishing machine (mesh size: # 4000) and a copper plate with a fresh and smooth surface was used for the test. The copper NP-based paste was printed in a 1 mm square on a fresh oxygen-free copper plate using a metal screen mask (Tokyo Process Service Co., Ltd.). Then, the paste was sandwiched by an oxygen-free copper plate (i-ject Co., Ltd., 3 × 3 mm; t = 1 mm). The resulting bonded bodies were sintered at 180, 200, 230, and 260 °C under a pressureless and N 2 atmosphere for 30 min for adhesion.

Characterization equipment.
The pH values during the copper NP synthesis were monitored using a pH electrode (HORIBA Ltd., 9615S-10D). The size and shape of the copper NPs were observed by an FEI XL 30SFEG scanning electron microscope (SEM) with an acceleration voltage of 5 kV. High-resolution transmission electron microscopic (HR-TEM) and high-angle annular dark field scanning TEM (HAADF-STEM) images were taken by an FEI TITAN 80-300 instrument at 200 kV and a JEOL ARM200F instrument at 200 kV equipped with an EDS analyzer, respectively. The diffuse reflectance infrared Fourier transform (DRIFT) for the characterization of the surface properties of the copper NPs was measured with a Thermo Scientific, Inc. NICOLET 6700 spectrometer equipped with an ATR system under a nitrogen atmosphere. The surface state of the copper NPs was evaluated by ULVAC-PHI PHI 5000 VersaProbe II X-ray photoelectron spectroscopy (XPS) with AlKα radiation (50 W, 15 kV). The crystal system and crystallite size of the copper NPs were measured by a Rigaku Intelligent X-ray diffraction (XRD) SmartLab system equipped with a PILATUS3 R 100 K detector using CuKα radiation (40 kV, 40 mA). Temperature-variable XRD measurements were also carried out by the SmartLab system under a nitrogen atmosphere for the evaluation of the sintering temperature dependency of the copper NPs. The temperature range and heating speed were 30-500 °C and 1 °C/min, respectively. The XRD patterns were taken at intervals of 10 °C in the range from 30 to 500 °C. The heating rate was fixed at 1 °C/min, and the holding time before the measurement was 1 min. The crystallite size of the NPs was calculated by Scherrer's equation 46 (Scherrer constant: 1.33) using the half width of the (111) diffraction peak. Thermogravimetric and differential thermal analysis (TG-DTA) of the copper NPs was measured using a Rigaku Thermo plus EVO2 TG8121 high-temperature model at a heating rate of 1 °C/min in a nitrogen atmosphere. For the electrical conductivity of the copper-pattern electrodes, the resistivity value was measured using a Mitsubishi Chemical Analytech Loresta-GX MCP-T 700 instrument with a four-point probe method. The share strength of the joined body was measured and calculated by an XYZTEC bond tester Condor Sigma at a shear tool speed of 50 μm s −1 . The bonding states between the copper paste and the copper plate of the bonded body were observed using a HAADF-STEM system.

Conclusions
In the present study, we designed and synthesized copper NPs that are applicable for practical copper nanopastes for the first time, and the resulting nanopastes exhibited remarkably low-temperature pressureless sintering behavior for the fabrication of thick copper electrodes on not only glass substrates but also flexible films, such as PI and PEN. Furthermore, the copper nanopastes thus obtained were successfully applied as a die attachment material. The strength of the bonding between a model IC chip and a copper substrate was 39 MPa after pressureless sintering at 200 °C for 30 min. The remarkable ultrahigh bonding strength and low-temperature pressureless sintering ability of this material make the material applicable as a promising alternative to soda alloys for constructing SiC-and GaN-based IC chips for next-generation power devices. Our copper NP-based nanopastes have the following advantages: (i) the copper NPs can be readily obtained in a large quantity from an aqueous solution system under ambient atmospheric conditions without the use of expensive production equipment, such as an autoclave and heater; (ii) the water-based copper NP preparation system does not require the use of organic solvents, surfactants, and dispersants, resulting in a system free from waste, environmental, and production-cost problems; and (iii) the copper NPs have a polycrystalline copper metal core and a thin NTA-modified Cu 2 O shell structure. The unique Cu 2 O-covered polycrystalline structure resulted in high oxidation resistance, long-term stability and low-temperature pressureless sintering and ultrahigh bonding abilities into the copper nanopastes. Our established water-based ambient NP synthesis through mild reduction of an in situ-prepared metal complex using a chelating agent might become a promising and powerful technique for the production of key materials essential for sustainable progress in PE technology as well as next-generation power device fabrication processes.