Low-temperature nanoredox two-step sintering of gelatin nanoskin-stabilized submicrometer-sized copper fine particles for preparing highly conductive layers

Title Low-temperature nanoredox two-step sintering of gelatin nanoskin-stabilized submicrometer-sized copper fine particles for preparing highly conductive layers Author(s) Yonezawa, Tetsu; Tsukamoto, Hiroki; Matsubara, Masaki Citation RSC advances, 5(75): 61290-61297 Issue Date 2015 Doc URL http://hdl.handle.net/2115/59816 Rights(URL) http://creativecommons.org/licenses/by/3.0/ Type article File Information Yonezawa-RSCA(5-75).pdf


Introduction
The unique properties of metal nanoparticles and ne particles provide possible new product developments. [1][2][3][4] At present, nanotechnology is a technology that cuts across different boundaries, including use in construction and composite materials, 5 as catalysts, [6][7][8][9] antibacterial coatings, 10,11 biosensors, 12 plasmonic photo devices, 13 and electrical devices, [14][15][16][17][18][19] in addition to some consumer products. The steady increase in nanoparticle and ne particle use is accompanied with larger production, requiring development of the handling and processing of such materials.
Moreover, the recent intensive progress in printed electronics has been based on the development of materials, including nanomaterials. [15][16][17][18] In particular, conductive inks and pastes that consist of metal nanoparticles and ne particles are key materials in printed electronics, as they provide a costsaving wet process application. [16][17][18][19][20][21][22][23] This modern printing process technology is expected to open the door for new manufacturing techniques that increase productivity while reducing energy and resource consumption compared to conventional methods. For exible, lightweight, and inexpensive electronic devices, such as smart labels that include radio-frequency identication (RFID), organic light-emitting diodes (OLEDs), exible displays, printed electronics will be frequently used to make cost-effective and eco-friendly products. 24 Noble metals, in particular the silver nanoparticle inks, are highly stable and oen used for producing conductive lms. 16 However, the high material cost still hinders their use in large-scale printed electronic applications. Silver migration is also an important issue, as it can degrade device performance. Conducting polymers are other possible alternatives, but in many cases, their conductivity is insufficient for industrial use. In contrast, copper nanoparticles and ne particles are attractive from a cost and anti-migration viewpoint. Recently, the synthesis of copper nanoparticles and ne particles has attracted considerable interest. [25][26][27][28][29][30][31][32][33][34][35][36][37][38][39][40][41][42] However, due to their poor resistance to oxidation, only a few papers have been reported thus far on their practical application as electro-conductive materials. From the printed electronic material viewpoint, the presence of copper oxides on the particle surface has negative consequences, as it increases the required sintering temperature and reduces the electro-conductivity. To overcome the copper nanoparticle and ne particle oxidation issue, numerous attempts have been made, including new sintering technologies, [26][27][28][29][30][31][32][33] antioxidation, 34 encapsulation of the copper surface with polymers, 35-38 organic molecules, 39 metal oxides, 40 and metal shells. 41 Yabuki et al. applied a two-step sintering method to a copper layer formed from a copper ink containing 20 nm copper nanoparticles covered by a thin carbon layer. 27 A 1 h annealing step in air followed by a 1 h reductive sintering step under 5% hydrogen (H 2 ) in argon (Ar) gas mixture at 300 C provided a copper layer with a low resistivity of 1.4 Â 10 À5 U cm. We previously applied a similar two-step sintering method to a layer obtained from a copper ink containing 46 nm copper nanoparticles covered by gelatin. 43 Aer annealing the copper layer at 200 C under 10 ppm oxygen (O 2 ) in nitrogen (N 2 ) gas ow for 1 h, it was then reduced at 250 C by 3% H 2 in N 2 gas ow for 1 h. A low resistivity of 5.1 Â 10 À6 U cm was obtained. Recently, we have also reported two-step sintering process at 250 C in air and in 3% H 2 in N 2 gas with ecologically prepared aminestabilized copper ne particles and discussing the formation of copper oxide nano-objects on the particle surface and the effect of sintering temperatures on resistivity. 44 However, to our best knowledge, a detailed discussion on the mechanisms of the low-temperature two-step sintering process of polymerstabilized copper ne particles that produces highly conductive copper layers at relatively low temperature as 200 C has not been discussed so far.
In our previous study, we proposed the use of gelatin as a stabilizer and antioxidation reagent for copper nanoparticles and ne particles prepared by a chemical reduction of copper(II) oxide (CuO) 35 and a copper salt. 36 CuO is a solid copper source containing oxygen as the only counter anion. From the contamination point of view, CuO is the ideal metal source candidate for copper ne particles used in conductive inks and pastes. Furthermore, this process can be performed at a copper concentration of 1 mol dm À3 (i.e., 64 g of Cu per 1 dm 3 of water), 35 which is signicantly higher than the usual chemical reduction process. Once these gelatin-stabilized copper ne particles were obtained, conductive pastes were prepared and used as the inner electrode material in a multi-layer ceramic capacitor (MLCC). 14 We also formed a conductive copper layer by a two-step sintering process at 200 C and 250 C using the Cu nanoparticles coated with a gelatin layer, 43 however, this previous study could not reveal the mechanism of this lowtemperature sintering. Moreover, this copper nanoparticle ink did not contain any sintering binders, which are usually used in industrial processes.
In this paper, we describe the low-temperature sintering process of copper lms made from the gelatin-stabilized copper ne particle ink using ethyl cellulose as an organic binder, via the two-step annealing method. The detailed mechanism of this two-step annealing was examined through X-ray diffraction (XRD) and electron microscopy analyses. The oxidative annealing produces smaller prominences of Cu 2 O, which play a key role in the sintering process. The resultant conductive copper lms show a low resistivity of 8.2 Â 10 À6 U cm. Compared to our previous studies, we performed a sintering process that simulated an industrial process due to using a copper ink that included organic binders.

Materials
The CuO microparticles used as the metal source were supplied from Nisshin Chemco (Kyoto, Japan) and had an average diameter of approximately 4 mm. Cu 2 O was purchased from Kanto (Japan). Hydrazine monohydrate, used as a reducing reagent, and aqueous ammonia (28%) for pH adjustment were purchased from either Wako or Junsei Chemical (Japan). Gelatin (cow origin, MW ¼ ca. 20 000) and a polyamine type dispersed agent were supplied from Nitta Gelatin NA Inc. (Osaka, Japan). EC vehicle, containing 10 wt% of ethyl cellulose, was purchased from Nisshin-Kasei Co., Ltd., (Japan). Water was puried by an Organo/ELGA Purelabo ultra (>18 MU).

Preparation of the copper ne particles
The detailed preparation procedure is described in Fig. 1. The copper ne particles were prepared following a method described previously. 35 32 g of gelatin was dissolved into hot water (950 cm 3 , 60 C), and then the CuO microparticles (80 g) were introduced into this solution. Then, aqueous ammonia (28%) was added to adjust the dispersion pH to ca. 11. The reduction of CuO (80 g, 1 mole) to metallic copper using the hydrazine monohydrate (120 cm 3 ) was carried out at 80 C in air with stirring for 2 h. The ne particles were then collected by decantation aer introducing a saturated aqueous citric acid solution to reduce the dispersion pH to ca. 8.5 and washed with water twice and ethanol twice again, followed by drying the collected copper ne particles at 60 C under N 2 atmosphere.
Preparation of the copper ne particle paste and conductive copper lm In order to prepare a paste of the copper ne particles, the collected copper particle powder was rst ground by an automatic mortar for 30 min, pulverized by a blender, and mixed with 5 wt% of a dispersing agent. Then the mixture was dispersed into 50 wt% of commercially available paste vehicle using a mixer and an emulsier. The obtained copper paste was deposited on a glass and an aluminum oxide (Al 2 O 3 ) substrates (supplied by Furuuchi Chem., Japan) by the doctor blade method for a lm with 40 mm thickness and 20 mm width. Aer drying in N 2 at 60 C for 1 h, the Al 2 O 3 substrate was cut into 20 mm Â 20 mm pieces and annealed at 200 C in air for 4 h, and subsequently reduced in 3% H 2 in N 2 gas. The reductive annealing times were varied from 1 to 4 h to determine the reduced state of copper. The gas ow rates were adjusted to 2 dm 3 min À1 for both air and 3% H 2 in N 2 gas during annealing.

Characterization
Scanning electron microscopic (SEM) images of the particles were performed using a JEOL JSM-6701F eld emission type SEM with an acceleration voltage of 15 kV. In this study, secondary electron observation images (SEI) and back scattered electron images for observation of the composition of samples (BEI-COMPO) were taken. SEI gives topological information of sample surfaces. BEI-COMPO was used to understand the composition of the surface of the samples. Heavier elements show brighter contrasts. XRD patterns were obtained with a Rigaku Mini Flex II (Cu Ka). Transmission electron microscopic (TEM) images were obtained with a Hitachi H-9500 with an acceleration voltage of 300 kV. The TEM samples included the annealed copper layer that was detached from the Al 2 O 3 substrate by scratching, then dispersed in ethanol and applied to carbon-coated copper grids. The thermal properties of the copper ne particles and copper oxide particles were measured by thermogravimetric (TG) analysis with a Shimadzu DTG-60H. The TG measurements of the copper paste and the sintered copper lm were carried out under air. Aer the measurement up to 600 C, all metallic copper was oxidized to CuO and organic molecules were supposed to be totally decomposed and eliminated. The sheet resistances of the copper lms were obtained using a four-point probe method (Loresta-GP, MCP-T610, Mitsubishi Chemical Analytech Co., Japan). In order to obtain reproducible data, six points were measured with a four point probe per a copper layer, as shown in Fig. S1, † and the average values are displayed here.

Results and discussion
Preparation and characterization of the gelatin-nanoskin stabilized copper ne particles Copper ne particles with uniform sizes could be obtained readily and reproducibly by our process. 35 Aer the reaction and pH control by citrate addition, the obtained copper ne particles were collected by decantation, washed by water and ethanol to remove excess gelatin molecules at room temperature, and dried under N 2 ow at moderated temperature (60 C). The powders showed no obvious particle oxidation, and no color change of the dried particles was observed for several weeks, as reported elsewhere. 35 The XRD pattern of the as-prepared copper ne particles is shown in Fig. S2. † As shown in Fig. 2a, the particles were uniform and nearly spherical, with an average particle size and a size distribution of 120 AE 35 nm. The surface of the particles was covered by the gelatin layer, 36 forming a smooth surface with no small dots observed. Gelatin molecules have amino groups and mercaptan groups in their side chains, and they are coordinated to the particle surface. If surface of metallic copper particles is oxidized, small dots should be observed because of the difference of density between metallic Cu and copper oxide (CuO or Cu 2 O). The copper ne particles were formed by the gradually supply of Cu 2+ ion from the surface of CuO microparticles in the presence of excess amount of hydrazine. Thus, supply rate of Cu 2+ ion limited the reaction, resulting in the formation of submicron sized uniform copper ne particles.
Preparation of the copper ne particle paste and printing Aer purication of the collected copper ne particles, they were then re-dispersed to form a viscous electro-conductive paste. The liquid component of the paste vehicle evaporated at around 120 C and the ethyl cellulose decomposed at temperatures over 210 C (Fig. S3 †). The obtained copper paste was stable and could be printed with a uniform thickness onto an Al 2 O 3 substrate using doctor blading. Aer printing, the copper layer was dried at 60 C under N 2 ow to avoid oxidation. Fig. 2b-d show the surface and cross-section SEM images of the as-printed copper particle layers. Copper particle layers with a at surface and a relatively uniform thickness of ca. 4.0 mm, ranging from 3.0 to 5.0 mm, were obtained (Fig. 2b). Fig. 2c and d are the surface SEM images at the same position taken with a secondary electron detection (SEI) mode and back scattered detection for composite observation (BEI-COMPO) mode. In a BEI-COMPO image, the image contrast is affected not only by the topological structure but also by the atomic number of the materials. Therefore, the bright contrast in BEI-COMPO images shows copper metallic cores of the particles (Fig. 2d). By comparing these two images, one can clearly understand that the copper ne particles are completely covered by polymer layers, including the gelatin, dispersant, and ethyl cellulose. Copper metallic cores did not contact directly to others due to these polymer layers. According to these polymer layers, these as-printed copper particle layers were not electro-conductive. Copper layers aer annealing in air and hydrogen-nitrogen mixed gas The copper thin layers on substrates were rst annealed at 200 C under airow to effectively remove the organic components, and then subsequently annealed under 3% H 2 in N 2 gas. Aer the oxidative annealing in air at 200 C for 4 h, the surface of the copper layer was also observed by SEM (Fig. 3). The images in Fig. 3 are quite different from those shown in Fig. 2. In Fig. 3a (the secondary electron image), a rough surface of large spherical particles and small prominences (which appear to be spherical particles) was observed. The large particles were connected via the small 10 nm size prominences. The BEI-COMPO image (Fig. 3b) indicated that the smaller prominences, with their bright contrast, consisted of copper. However, a comparison of these two images revealed that the organic components were not completely removed by this oxidative annealing. The areas marked by arrows in Fig. 3b, for example, are considered as organic components based on a comparison of the two SEM images.
X-ray diffractograms of the copper layers before and aer annealing are displayed in Fig. 4. As discussed in our previous paper, 35 the copper ne particles were not oxidized in their powder form. Aer they were re-dispersed in a copper particle paste and atly printed on an Al 2 O 3 substrate, a small, broad peak corresponding to Cu 2 O (111) (2q ¼ 36.4 ) was observed. Aer annealing in air, this peak became larger but remained broad, suggesting that a small quantity of Cu 2 O was generated on the copper particle surface, corresponding to the SEM images in Fig. 3. Aer annealing in air, the copper layer contained 29 wt% Cu 2 O and 71 wt% metallic copper, calculated using the reference intensity ratio (RIR) method.
The annealed substrates were then again annealed at 200 C, but the gas was changed to 3% H 2 in N 2 gas mixture (reductive annealing). The reductive annealing time was varied from 1 to 4 h. As observed in Fig. 4, the XRD peak intensity corresponding to Cu 2 O decreases with the reductive annealing time. First, the surface of the lm containing Cu 2 O was partially reduced to form copper under reductive annealing for 1 h (Fig. 4c). Compared with the XRD patterns of oxidative annealed lms (Fig. 4b), the very small change in intensity of Cu 2 O peaks can be observed. A resistivity of the oxidative annealed lm was 25 U cm, which indicates that the oxidative annealed lm was not conductive. The resistivity dramatically changed, however, aer 1 h of reductive annealing, to 7.4 Â 10 À5 U cm. Thus, reductive annealing makes the lm electro-conductive. This is probably due to the connecting of surface metallic copper which was generated by the reductive annealing. The surface metallic copper formed necking during reductive annealing and increased the conductivity. But this reduction of Cu 2 O to Cu cannot be observed in the XRD patterns. Aer additional 1 h of reductive annealing, more than half of Cu 2 O was reduced to metallic copper ( Fig. 4c and d). The resistivity and the average thickness of the lm reduced for 2 h were 2.7 Â 10 À5 U cm and 3.3 mm (Fig. 6c).
The XRD pattern (Fig. 4e) indicates that Cu 2 O was completely reduced by annealing for more than 3 h at 200 C, and a low resistivity of 8.2 Â 10 À6 U cm was achieved. A very small peak observed at 36.88 in Fig. 4e and f (which cannot be found in the Al 2 O 3 pattern of JCPDS card) can be attributed to the Al 2 O 3 substrate used for sintering (Fig. S4 †). This resistivity is almost 5 times of that of the bulk copper (1.7 Â 10 6 U cm). Thus, the complete reduction of the inside Cu 2 O is necessity to achieve the low resistivity. Additional 1 h of reductive annealing (total 4 h) did not decrease the resistivity. Resistivities of the reductive annealed lms are summarized in Fig. 5. On the other hand, TG measurement of commercially available Cu 2 O particles was performed in this mixed reductive gas, and the change in mass versus time is shown in Fig. S3. † The Cu 2 O had a slight change in mass under 3% H 2 in N 2 during the reduction at 200 C for 2 h. Thus, $2 mm size Cu 2 O particles (SEM image shown in Fig. S5 †) are not usually reduced by these conditions. However, the XRD patterns of the reduced lms revealed that the Cu 2 O was completely reduced at 200 C for 3 h of annealing time, forming metallic copper. We will discuss below about the mechanism of Cu 2 O reduction at a lower temperature.  The surface and cross-sectional SEM images aer this reductive annealing step for 2 h are shown in Fig. 6. Necking [45][46][47][48][49] of the copper particles and connections between particles were clearly observed in SEI image of Fig. 6a. The cross-sectional image (Fig. 6c) clearly indicates that the necking and connections were observed not only at the surface region but also near the substrate. Fig. 6b shows the BEI-COMPO image of the copper layer surface. Compared with Fig. 6a, the areas marked by arrows can be considered the organic components.

High-resolution TEM (HRTEM) observation of copper layers aer annealing
For further investigation, we carried out a HRTEM observation of the annealed copper thin lm scratched from the Al 2 O 3 substrate (Fig. 7). As shown in Fig. 7c, aer heating under airow at 200 C, small prominences less than ca. 10 nm in size were evident on the surface of the copper particles. These nanosized Cu 2 O were clearly observed in our previous report about the in situ heating TEM observation of Cu oxidation. 45 The lattice fringe distances (2.5Å and 3.1Å) observed in the HRTEM images shown in Fig. 7e and f allowed us to identify these prominences as Cu 2 O (111) and (110), respectively. These images strongly support a formation mechanism of the oxidative annealing in air at 200 C producing small prominences of Cu 2 O (<10 nm) covering the metallic copper particles. Fig. 7g-j show the TEM images of the copper lm annealed under the reductive gas for 2 h (g and h) 4 h (i and j). Fig. 7g and h are displaying the necked particles. During the oxidation of the copper particle surface, the diffusion of oxygen ions and copper ions into the surface of copper particle rst occurs, resulting in formation of copper oxides. 45 Then, the particle swelled due to the difference in density between the copper (8.94 g cm À3 ) and copper oxide (Cu 2 O: 6.04 g cm À3 ). This swelling allowed the particle surface to partially contact nearby particles, resulting in an aggregate formation suitable for sintering. The prominences that were below 10 nm in size could lead to the necking between particles. Thus, sintering of the particles progressed due to the melting point depression resulting from the Cu 2 O nanoscale prominences (generated in the heating under air) reducing to form copper again. The selected area electron diffraction (SAED) image of the copper layer aer the reductive sintering is shown in Fig. 7j. The   diffraction spots indicate that the obtained layer was consisted of metallic copper.
Low-temperature sintering mechanism of the submicrometersized copper ne particles that produces electro-conductive layers Fig. 8 shows the TG result of copper particles paste. The measurement was carried out under air ow (200 cm 3 min À1 ) with the heating rate of 0.5 C min À1 . The liquid component of paste (48 wt%) was rstly evaporated below 120 C. Aer measurement, the copper was completely oxidized and changed into CuO. However, the value of 56.3 wt% includes the weight loss of decomposed polymer components such as the gelatin, the dispersing agent, and the ethyl cellulose. The initial copper ratio of the paste can be calculated as 0.563/1.25 ¼ 0.450 with use of the ratio of MW(CuO)/MW(Cu) ¼ 79.545/63.546 ¼ 1.25. Thus, the amount of the polymer is calculated as 52.0 À 45.0 ¼ 7 wt%. It contains not only gelatin but also vehicles, ethyl cellulose, required to obtain a stable paste. The detailed oxidation behavior of copper particles can be observed between 100 and 250 C. The gelatin-stabilized metallic copper particles can be oxidized at higher temperatures than ca. 120 C under air. Cu 2 O forms initially on the copper particle surface then. The mass in the TG thermogram increased above 120 C, before decreasing again above 165 C. As shown in the TG thermogram of the paste vehicle used in this study (Fig. S3 †), the decomposition of ethyl cellulose in the commercial paste vehicle starts at 160 C and lasts at temperatures higher than 200 C. Therefore, annealing at 200 C is insufficient to eliminate all the organic components from the copper paste. Fig. 9 shows TG result of the sintered copper lm with 3 h of reductive annealing. TG measurement was also carried out under air ow (200 cm 3 min À1 ) and the heating rate of 5 C min À1 . The oxidation of Cu started at ca. 120 C. Above 160 C, the decomposition of polymers occurred at the same time. Aer measurement, the copper was completely oxidized and changed into CuO. However, the value of 117 wt% also includes the weight loss of decomposed organic components. The initial copper ratio of sintered sample can be calculated as 1.17/1.25 ¼ 0.94. From this result, we can estimate that the copper layer aer sintering contained 6 wt% of organic components. The copper paste used to prepare copper layer contained 7 wt% of polymers, including gelatin, dispersant and ethyl cellulose.
With our low temperature sintering, all organic components were not removed from the copper layer by sintering. However, formation of prominences of Cu 2 O by oxidative sintering followed by the reductive sintering at a low temperature generated metallic copper neckings to form a conductive copper layer. These results indicate that the polymers used for the copper paste can play also an important role of antioxidizing stabilizers of the sintered metallic copper layer under ambient conditions. From the results and discussion above, the sintering mechanism of our two-step annealing process can be illustrated, as shown in Fig. 10. First, the copper ne particles were surrounded by an organic layer. Then, the diffusion of oxygen ions and copper ions into the surface of copper particle occurred 45 and small prominences of Cu 2 O, ca. 10 nm in size, which were attached to the copper ne particles, were produced via the oxidative annealing. As the grass transition temperature (T g ) of gelatin is about 80 C, gas molecules can be penetrate and diffused easily in the polymer layers and can react with copper and copper oxide surfaces. Finally, these Cu 2 O prominences were reduced again to form the particle necking. Kim et al. reported the time-dependent reduction of Cu 2 O powder under 5% H 2 in Ar gas at 200 C. 50 They also showed that a higher reduction temperature gives a shorter sintering period and the reduction of Cu 2 O powder at 200 C completed in 450 min. Our reduction completed much faster than their result. In our system, the Cu 2 O particle size is less than 10 nm as shown in   9 TG chart of the sintered copper film with 3 h of reductive annealing. The sample was scratched from a glass substrate. TG measurement was carried out under air flow (200 mL min À1 ) and the heating rate of 5 C min À1 . After measurement, the copper changed into the copper oxide (CuO), so that the initial copper ratio was MW(Cu)/MW(CuO) ¼ 117/125 ¼ 0.94. The organic component ratio was estimated as 6 wt%. Fig. 7b and in the previous in situ TEM observation. 46 Compared with micro-and nano-particles, the Cu 2 O particle size may strongly affect their reduction. If the particle size of Cu 2 O is larger than 1 mm, the reduction temperature increases above 400 C. 51 At 200 C, only 3 wt% of Cu 2 O micro-particles ($2 mm) were reduced under 3% H 2 in N 2 gas for 20 h of annealing ( Fig. S5 and S6 †). This result indicates that only 10 nm of depth from the surface were reduced. Thus, the Cu 2 O particles size of 10 nm is enough small to enable an easy reduction. Our sintering mechanism can be explained that the nano-sized Cu 2 O prominences enable and accelerate the reduction at lowtemperature. Using this two-step annealing, we successfully achieved the low resistivity due to formation of particle necking without removing all polymer layers.

Conclusions
In conclusion, we performed an oxidative-reductive two-step annealing at 200 C of 130 nm sized copper ne particles to obtain a highly conductive copper layer, even though the organic components included in the copper ne particle paste did not fully decompose at this temperature. Aer annealing the copper layer in air, Cu 2 O prominences with the size of less than 10 nm were generated on the copper particle surface, connecting together through the organic layers. These Cu 2 O nanoprominences could then be reduced to generate metallic copper nanoparticles, connecting the copper cores. Necking of the copper ne particles were observed along with their connections aer the reductive annealing. The obtained lm resistivity was as low as 8.2 Â 10 À6 U cm. Our results clearly indicate that using small copper nanoparticles is not the only solution for a low-temperature sintering process.