Characterization of self-assembled silver nanoparticle ink based on nanoemulsion method

A well-dispersed self-assembled silver nanoparticles (AgNPs) ink with high purity was synthesized via AgNO3 emulsion prepared by blending an AgNO3 aqueous solution and a liquid paraffin solution of both polyoxyethylene (20) sorbitan monooleate (Tween 80) and sorbitan monooleate (Span 80). The ink remained as an emulsion at low temperatures; however, it produced AgNPs after sintering at about 60°C and showed a high stability at nanoscale sizes (with diameters ranging 8.6–13.4 nm) and a high conductivity. During the whole procedure, Tween 80 acted as a surfactant, reductant and stabilizer. Presumably, Tween 80 underwent an autoxidation process, where a free radical of an α-carbon of ether oxygen was formed by hydrogen abstraction. The mean diameter of emulsion droplets could be reduced by decreasing water content and increasing the ratio of surfactant and concentration of AgNO3 aqueous solution. Consequently, the thermogravimetric analysis and X-ray diffraction result clarified the purity of the produced Ag0. Dynamic light scattering and ultraviolet-visible spectroscopy clarified that an increased concentration of AgNO3 decreased the particle size.

A well-dispersed self-assembled silver nanoparticles (AgNPs) ink with high purity was synthesized via AgNO 3 emulsion prepared by blending an AgNO 3 aqueous solution and a liquid paraffin solution of both polyoxyethylene (20) sorbitan monooleate (Tween 80) and sorbitan monooleate (Span 80). The ink remained as an emulsion at low temperatures; however, it produced AgNPs after sintering at about 60°C and showed a high stability at nanoscale sizes (with diameters ranging 8.6-13.4 nm) and a high conductivity. During the whole procedure, Tween 80 acted as a surfactant, reductant and stabilizer. Presumably, Tween 80 underwent an autoxidation process, where a free radical of an α-carbon of ether oxygen was formed by hydrogen abstraction. The mean diameter of emulsion droplets could be reduced by decreasing water content and increasing the ratio of surfactant and concentration of AgNO 3 aqueous solution. Consequently, the thermogravimetric analysis and X-ray diffraction result clarified the purity of the produced Ag 0 . Dynamic light scattering and ultraviolet-visible spectroscopy clarified that an increased concentration of AgNO 3 decreased the particle size.

Materials
Tween 80, liquid paraffin (JIS Special Grade), ethanol (JIS Special Grade, 99.5%) and silver nitrate (JIS Special Grade) were all purchased from Wako Pure Chemical Industries, Ltd. Sorbitan monooleate (Span 80) purchased from Tokyo Chemical Industry Co. Ltd was used for the preparation of inks. royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 7: 200296 All chemicals were employed without any further purification, and deionized water was used for all experiments.

Preparation and characterization of nanoemulsion
The stability of the nanoemulsion was tested by titrating liquid paraffin solutions of the surfactant blend, as a component with water. This served as a model of the AgNO 3 aqueous solution for determining the maximum emulsifiable limit, which can help in identifying the appropriate dose of AgNO 3 solution, both as a component and a medium. Tween 80 and Span 80 were dissolved in paraffin together, in different ratios (0 : 100, 20 : 80, 30 : 70, 40 : 60, 50 : 50, 60 : 40, 70 : 30, 80 : 20 and 90 : 10), but with a constant surfactant blend-paraffin ratio of 5 : 95 (w/w).
Emulsions were formed by adding water to the liquid paraffin solutions of the surfactant blend and stirring with a mixer (Automatic Labo-Mixer NS-8, PasolinaNS-8, Pasolina, Japan) for homogenization at 800 r.p.m. at room temperature for 30 min. The maximum water solubility was determined by observing the emergence of phase separation.
The droplet size distribution of the W/O emulsion was evaluated by dynamic light scattering (DLS) at 25°C, using a DLS spectrophotometer (DLS-7000, Otsuka Electronics Co., Inc., Osaka, Japan). The data were obtained as a mean value from four or five measurements and conducted immediately after the homogenization.

Synthesis and purification of silver nanoparticles
First, 2 ml aqueous AgNO 3 solutions with different concentrations (0.10, 0.20, 0.30 and 0.50 g ml −1 ) were added into liquid paraffin solutions of the surfactant blend containing 6 g paraffin, 1 g Span 80 and 1 g Tween 80, and mixed with a magnetic stirrer at 500 r.p.m. at room temperature for a few minutes to form an Ag + nanoemulsion. Brackets [] will be used to indicate the concentration of the AgNO 3 solution in g ml −1 in the text and figures hereafter.
For the synthesis of AgNPs, the AgNPs nanoemulsion was heated at 80°C and stirred at 500 r.p.m. for 15 min with a hot stirrer.
Purified solid AgNPs were obtained by a subsequent purification procedure with ethanol three times to remove surfactants and paraffin. Ethanol was added to each aqueous suspension of unpurified AgNPs; the sample was shaken thoroughly with a mixer and then centrifuged at 9060g RCF (relative centrifugal force) at 20°C for 20 min. Sedimentary solid particles were recovered by decantation, and further purification was carried out by repeated centrifugation. The sample was dried under N 2 gas for the later characterization.

Characterization of silver nanoparticles
The dried solid AgNP samples were analysed by X-ray diffraction (XRD, AXS D8 Advance, Bruker, Germany) in order to identify the crystal structure in a 2θ range from 10 to 90°at a 2θ scan speed of 0.2°s −1 at 40 kV and 40 mA using Cu-Kα radiation (1.5418 Å).
The degradation behaviour of the AgNPs was investigated by thermogravimetric analysis (TGA, Mettler Toledo TGA/SDTA 851, Gießen, Germany). A degradation diagram was obtained by placing 6.000 mg purified AgNPs into a platinum pan and heating from 40°C to 1100°C at the rate of 20°C min −1 . The Ar gas flow rate was 200 ml min −1 .
Fourier transform infrared (FTIR) spectroscopy was conducted to characterize the AgNPs with a FTIR spectrophotometer (FT/IR-6100, JASCO Corporation, Tokyo, Japan) at wavenumbers ranging from 600 to 3700 cm −1 at room temperature. An unpurified liquid sample of Tween 80 was subjected to measurement by pasting a trace amount on the surface of a KBr single pellet. Solid AgNP samples were kept in a desiccator for drying for 12 h, followed by measurement with a pellet containing AgNPs and KBr (1 : 100).
Transmission electron microscopy (TEM, H-7650, Hitachi, Japan) was conducted to determine the static particle diameter (d S ) distribution of AgNPs at an acceleration voltage of 200 kV. Samples were prepared by placing a drop of the purified AgNP ethanol dispersion on a carbon-coated formvar Cu royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 7: 200296 grid and drying under nitrogen gas flow for approximately 10 min at room temperature. ImageJ was used for particle diameter analysis in a projected area range from 10 to 500 nm 2 .
The ultraviolet (UV)-visible spectrum of the unpurified AgNP aqueous solution was acquired using a spectrophotometer (UV-3100 PC, Shimadzu, Kyoto, Japan) with a 3 ml solution in a quartz cell, with a light path length of 10 mm, at a wavelength interval of 200-800 nm, at room temperature.
The zeta potential was measured to demonstrate the stability of the AgNPs, with a zeta potential analyser (Zetasizer Nano-ZS, Malvern Instruments Ltd, UK) at 20°C, after sonication for 20 min at room temperature. The produced AgNPs were diluted 20 times. A 1.0 ml of the diluted AgNPs was shaken well, poured into a cuvette with a capacity of 1.0 ml and set to a sample chamber of the instrument (DTS1070, Malvern Instruments Ltd, UK). The equilibrium time chosen was 30 s. The viscosity of water was set to 1.0031 mPa s, and the refractive index was set to 1.330 for the calculation in the application software. A mean value was calculated from more than six measurements; during each measurement, zeta potential was measured for three 20 s runs without any interval between runs.
As one of the common methods to obtain an average particle diameter, hydrodynamic diameter (d H ) of AgNPs was measured by Zetasizer equipped with the DLS technique [34].
The relationship between the speed of Brownian motion [35] of a particle and that particle's d H is defined by the Stokes-Einstein equation [35,36], where D is the diffusion coefficient, k B is the Boltzmann constant, η is the dynamic viscosity and T is the absolute temperature. The resistance was measured with a digital multimeter for determining the conductivity of ink spots manually dropped on ink-jet paper and subsequently sintered at different temperatures.

Characterization of water in oil nanoemulsion
Our results show that an increase in the ratio of Tween 80 leads to an increase, then a decrease of the maximum emulsifiable water content (figure 2). Consequently, the highest water content of 5.12 wt% on the total blend was attained at a ratio of Tween 80 : Span 80 = 50 : 50.
Studies were carried out using the above ratio of Tween 80 : Span 80 to determine zones where nanoemulsions were formed. It is already established that the mixture becomes a W/O nanoemulsion with a high paraffin content, high surfactant blend content and low water content. Meanwhile, with an increase in the water phase in the system, the emulsion transforms from a W/O nanoemulsion to a W/O emulsion and finally becomes an O/W emulsion, causing water droplets to grow in diameter, as predicted according to figure 3. Namely, the emulsion system with larger amounts of surfactant is more likely to stably form a W/O nanoemulsion with smaller water droplets and has the capability to hold more water, if added, despite decreasing stability and shifting to W/O emulsion with larger water droplets. As another variable, if a AgNO 3 aqueous solution at as high a concentration as possible is contained instead of only water, the total amount of silver in the emulsion accordingly reaches a maximum, as is expected to provide a silver nano-ink with a maximum content of AgNPs.
Furthermore, as shown in figure 4, the water droplet diameters of the nanoemulsion measured by DLS increased systematically with an increase in the water content of the emulsion. With the constant ratio of water content, nanoemulsion with higher surfactant tends to have smaller droplet diameters. As mentioned above, the diameters of AgNPs can be controlled by adjusting the ratio between oil, surfactant and AgNO 3 aqueous solution.

Characterization of silver nanoparticles
During heating, the colour of the solution turned dark brown. In order to understand which components are involved in the chemical change, every possible combination of the components contained in the AgNP nanoemulsion was mixed and heated. The result showed that chemical change only occurred using the AgNO 3 aqueous solution and Tween 80.
The speed of the reaction increased with an increase in temperature ( figure 5); however, a stable condition with no chemical change was maintained at room temperature or lower for a long period of time.
The Gibbs free energy is expressed as where ΔG is the Gibbs free energy, ΔH is the enthalpy change, T is the absolute temperature and ΔS is the entropy change.
During the reaction, a solid phase is produced from the liquid phase, which indicates that ΔS < 0. As the reaction occurs spontaneously (ΔG < 0), ΔH is negative with a large absolute value. Furthermore, this reaction occurs spontaneously and slowly at a relatively low temperature. Figure 6 shows the XRD patterns of the solid AgNPs prepared from an AgNP nanoemulsion containing 6 g paraffin, 1 g Span 80, 1 g Tween 80 and 2 ml aqueous 0.30 g ml −1 AgNO 3 solutions, which was then purified and sintered at 80°C.
The  Figure 7 shows TG curves of Tween 80, Span 80 and the H 2 O/paraffin nanoemulsion containing 2 g H 2 O, 6 g paraffin, 1 g Span 80 and 1 g Tween 80. For the H 2 O/paraffin nanoemulsion, a weight loss period at approximately 100°C is attributed to water evaporation, followed by the volatilization of paraffin at approximately 120-205°C with only one step. The surfactants lost weight at 350-450°C, depending on the type of surfactant, compared with the TG curves of Tween 80 and Span 80 [37]. Figure 8 shows a TG curve of dried AgNPs, that exhibits a similar degradation tendency to the H 2 O/ paraffin nanoemulsion at 40-500°C, losing 1.97% in weight. The degradation peak at 158°C is related to the residual functional groups from Tween 80, which can also be observed in the DTG curve with a rapid weight loss (15.8 mg min −1 ) as shown in electronic supplementary material, figure S1. For later heating period from 500-900°C, the weight of dried AgNPs presented a similar decreasing speed, but a rapid decrease at approximately 960-968°C. This degradation period in the TG and DTG curves corresponds to the melting point of Ag 0 , where DTA shows a sharp endothermic peak.
This sharp peak has a reversal that means a momentary weight gain (only 0.09% in weight), then a very sharp decrease due to explosive decomposition with the recoil effect. Some references also showed a similar phenomenon, where Muhamad et al. applied two weight gain peaks for eutectic and monotectic reactions (melting) temperatures for the mixture of Ag and CuO [38], while Tsuzuki and McCormick did not point out a tiny reversal peak for ZnO nanoparticles [39]. Jagadeesh Babu et al. linked the sharp exotherm peak to the decomposition/dissociation of the salt as free temozolomide [40]. Although the TG curve rapidly changes, there is only one endothermic peak in the DTA curve (91.3 µV s mg −1 at In the FTIR spectra of Tween 80 shown in figure 9, an O-H peak appears at approximately 3439 cm −1 , and the -CH 2stretching vibrations are present at approximately 2932 and 2866 cm −1 (asymmetrical stretch and symmetrical stretch, respectively) [41]. The sharp and symmetric characteristic peak at 1746 cm −1 is presumably due to the stretching vibration of C=O in an ester carbonyl group [42,43]. The peak at approximately 1647 cm −1 is regarded as a stretching vibration of   royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 7: 200296 C=C [42,44]. The peak at 1465 cm −1 is mainly attributed to the bending vibrations of -CH 2 , although the asymmetrical bending vibrations of -CH 3 are involved as well [41,45]. The 1360 cm −1 peak is attributed to the symmetrical bending vibrations of -CH 3 [46]. The C-O peak of appears at 1307 cm −1 (symmetrical stretch) and 1110 cm −1 (asymmetrical stretch), respectively, while the stretching vibration of C-O is visible at 1259 cm −1 [47][48][49].
Based on comparison with the FTIR results for Tween 80, several functional groups, such as −CH 3 , C=C and C=O, are found to be present in the purified solid AgNPs.
Furthermore, the transmittance of C-O-C decreased after mixing Tween 80 and AgNO 3 , while that of C=O increased with increased [AgNO 3 ] in the emulsion system.
Some studies have demonstrated that surfactants with specific functional groups could reduce metal ions into metal via redox reaction [50][51][52][53][54]. For example, free radicals of the α-carbon of ether oxygen in nonionic polyoxyethylene surfactants (Tween series surfactants) are able to reduce Ag + into Ag 0 as shown in figure 10 [54,55].
In Tween 80 with a chemical structure shown by four sets of parentheses marked with symbols from 'w' to 'z', there are several -R-O-CH 2 -R'-groups. Namely, it is believed that Tween 80 undergoes a redox reaction, in which a free radical is formed from the α-carbon of the ether oxygen via hydrogen abstraction in the initial stage [54]. Afterwards, the radicals can be further oxidised into esters, or degraded to form aldehydes.
TEM images of AgNPs prepared from AgNP nanoemulsions with different [AgNO 3 ] are shown in figure 11. The morphology of AgNPs was found to be spherical. The d S was estimated by ImageJ software and found to vary as 8. When the number of nuclei remained constant, or increased slower than the total ions due to the high Ag + ion concentration, the atoms formed at the latter period were used for the growth of particles and resulted in the formation of larger particles [56,57]. In addition, those particles had a smaller diameter than the droplets of the H 2 O nanoemulsion itself (approx. 73 nm), as presented in figure 4. This finding also indicates that this nanoemulsion system could provide an extremely tiny microreactor to produce nanomaterials, especially by reduction.
The d H was found to be 16.9, 18.4, 20.4 and 24.2 nm with different [AgNO 3 ] of 0.10, 0.20, 0.30 and 0.50 g ml −1 , as shown in electronic supplementary material, table S1. Compared with the d S results obtained by TEM, the average d H measured by Zetasizer showed almost two times.
In this study, the synthesized AgNPs have been confirmed with some functional groups stabilized on the surface, which provides extra shells to the AgNPs, leading to a larger d H through DLS method than the d S through TEM [58], as shown in electronic supplementary material, figure S2, and similar results have been reported in the previous studies [34,59,60]. Figure 12 shows the UV-visible spectra of the AgNP dispersions synthesized with different concentrations of AgNO 3 . A red shift from 407 to 415 nm occurred with an increased [AgNO 3 ]. According to Mie's theory [61], this phenomenon indicates the crystallization of AgNPs with increased diameter because of the increased initial [AgNO 3 ]. However, the negative absorbance values should be related to the instability of the AgNP solution, or the C=O conjugation that exists on the surfaces of the particles generated from surfactant groups after the reaction.
For a stable colloid system, surface effects play a more important role than bulk effects. As for the most accepted Derjaguin-Landau-Verwey-Overbeek (DLVO) theory [62,63], the stability of a colloid can be reflected by the sum of van der Waals attractive forces and electrostatic repulsive forces within the electrical double layer model [64]. The zeta potential reflects the stability by electrostatic repulsion, as it measures the effective charge of the particle surface [65]. In general, nanoparticles with absolute Tween 80 redox reaction zeta potential values greater than 30 mV have high degrees of stability [66]. Figure 13 shows that all the AgNP samples had highly negative zeta potentials varying from −45.8 to −50.6 mV, implying that they were all stably dispersed. The high absolute zeta potentials ensure the high dispersion stability of the AgNP ink, due to the residual functional groups (-OH, -O-C=O, etc.) from Tween 80 present in the product (FTIR). These functional groups provide electrosteric repulsion, which is the combination of electrostatic repulsion arising from the negative charge condition of the mentioned organic functional groups on the AgNP surface and steric repulsion derived from the long-chain molecules on the AgNP surface. The synthesized AgNP inks practically maintained a stable condition, even after long-term royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 7: 200296 storage (five months) at room temperature, with only slight sedimentation ( figure 14). Therefore, the negative values appearing in the UV-visible spectrum are probably not caused by the instability of the AgNP solution, but mainly resulted from the C=O conjugation. As shown in table 1, the electrode created by drying the AgNP nanoemulsion at room temperature had high resistance; however, it became conductive after sintering the nanoemulsion at approximately 50°C, which accelerates the reduction of Ag + to solid Ag 0 . The sintering temperature should be set as high as possible for higher conductivity, but in a reasonable range that causes no damage to the substrate. The agglomerated AgNPs obtained by centrifugation of a nanoemulsion after heating at 60°C showed excellent conductivity (0 Ω).

Conclusion
A self-assembled AgNP nanoemulsion ink was prepared successfully through a one-step nanoemulsion method, where Tween 80 can act not only as a surfactant, but also as a reducing agent to obtain AgNPs, and as a stabilizer to prevent aggregation after the reduction. The mechanism of the reaction between Tween 80 and AgNO 3 is proven to be a redox reaction caused by a free radical of an α-carbon of ether oxygen and Ag + . The ink remains as a nanoemulsion at low temperature; however, it produces AgNPs with diameters of 8.6-13.4 nm after a sintering procedure, ensuring a high conductivity and good stability. Regulation of the particle size is feasible, and smaller particles are easily obtained by adjusting certain factors: (i) water content, (ii) surfactant-to-paraffin ratio, and (iii) AgNO 3 concentration. The AgNP inks had high absolute values of zeta potential varying from −45.8 to −50.6 mV, implying that they were all stably dispersed, although there was no significant difference in [AgNO 3 ]. The high stability of the ink proves that the high absorbance values in the UV-visible spectrum are related to C=O conjugation. Competing interests. There are no conflicts to declare.  Funding. This work was supported by JSPS KAKENHI grant no. 17KT0069 and the operating budget of University of Tsukuba for language editing and the article processing charge for an open access article.