For Reviewers

Transparent conductive electrodes are very important in many modern applications such as photovoltaic devices [1-3], light-emitting diodes (LEDs) [4, 5], and flexible displays [6]. Indium tin oxide (ITO) is the most common material used as transparent conductive electrodes nowadays; however, the price of ITO has been increasing owing to the limited supply of indium. In terms of fabrication, ITO involves time-consuming processes under high temperature and vacuum conditions. However, the brittleness of ITO makes it unsuitable for use in flexible and bendable applications. Many candidates have been reported as prospective replacements for ITO, such as conductive polymers, carbon nanotubes, graphene, metal grids, and silver nanowire random networks. Amongst these candidates, silver nanowires have drawn much attention because of their excellent electrical conductivity and high optical transparency [7]. Furthermore, silver nanowire random networks can be fabricated through the solution process, which facilitates low cost and scalable roll-to-roll manufacturing. Prototypes of devices using silver nanowire random networks as transparent conductive electrodes have been reported, such as LEDs, touch screens, and solar cells [7-9].

Silver nanowire random networks have also been used as transparent film heaters, following the Joule heating characteristics of electrically conductive materials. Many researchers have attempted to utilize silver nanowire film heaters in real-life applications, such as for window defogging [10], in thermoacoustic speakers [11], and in thermochromic devices [12]. However, silver nanowire random networks breakdown under the flow of electric current and are unsuitable for long-term use [13]. Khaligh et al. concluded that silver nanowire transparent electrodes failed at elevated temperatures, owing to the instability of the silver nanowire and breakdown of the silver nanowire network [13]. However, the main reason for this instability was unknown, although some reasons were hypothesized. The temperature of the silver nanowires could have been higher than the average surface temperature, particularly, at the junction of the silver nanowires. The corrosion caused by the reaction with the sulfur in silver nanowires could be another possible reason, as silver sulfide (Ag₂S) was observed in the nanoparticles formed on the surfaces of the failed silver nanowires at elevated temperatures. Moreover, as all the sidewalls of the silver nanowires were {110} planes, which were not the lowest energy planes in the FCC structure, high temperatures could induce atomic diffusion at the sidewalls of the silver nanowires. Li et al. tried to conduct a numerical analysis of the Joule heating of the silver nanowire networks and hypothesized that the failure could have been caused by the melting of the nanowires [14].

In this work, we studied the Joule heating and failure of silver nanowire random networks embedded in polymers. Numerical analyses were performed to compare the temperature of the silver nanowire with the average surface temperature.

Figure 1 (a-b) shows the scanning-electron microscope (SEM) image of the silver nanowires synthesized by the polyol method. The average length and width were 11±3.7 μm and 67±17.7 nm, respectively. The silver nanowire random network embedded in hybrimer is shown in Figure 1 (c-d). Typically, silver nanowires have poor adhesion to polymer substrates and thus, can be easily detached from the substrate surface. The embedding of silver nanowires in the hybrimer improved the adhesion between the silver nanowires and substrate and simplified the handling of the heater sample.

Fig. 1. 
SEM images of silver nanowire synthesized by polyol method (a-b) and silver nanowire random network embedded in hybrimer substrate (c-d)

Figure 2(a) shows a sample of the silver nanowire random network embedded in hybrimer. The optical transparency of the silver nanowire random network embedded in hybrimer is good, as the word “KAIST” underneath the sample can be seen clearly using naked eyes. The actual optical transmission spectrum of the silver nanowire random network embedded in hybrimer, measured by ultraviolet–visible (UV-Vis) spectroscopy, is shown in Figure 2(b). The optical transparency at wavelengths greater than 500 nm was ~80 % while the sheet resistance measured by a four-point probe was ~50 Ω/sq. When a voltage bias was applied to the silver nanowire random network, the surface temperature increased, according to the Joule heating effect. The time-dependent surface temperature profiles of silver nanowire random network (R_sh = ~50 Ω/sq.) with different voltages applied are shown in Figure 2(c). The result shows that the higher the voltage is applied, the higher the final surface temperature is obtained. The response time (T₉₀) and recovery time (T₉₀) are estimated as 170 s and 240 s, respectively.

Fig. 2. 
(a) Sample of silver nanowire random network embedded in hybrimer. (b) Optical transmittance spectrum of silver nanowire random network embedded in hybrimer measured by UV-Vis spectroscopy. (c) Time-dependent surface temperature profile of silver nanowire random network (Rsh = ~50 Ω/sq) for different applied voltages.

Figure 3(a) shows the scattering plot for the relationship between the electric power and increase in the surface temperature of the silver nanowire random network embedded in hybrimer. The result shows that the relationship between the electric power and increase in surface temperature is linear, by fitting with a linear equation (R² = 0.978). In order to observe the consistency of Joule heating of the silver nanowire random network, voltage cycles were applied to the silver nanowire random network (R_sh = ~50 Ω/sq) repeatedly. Figure 3(b) shows the time-dependent surface temperature profile of the silver nanowire random network when 5 V on/off cycles were applied. The result obviously shows that the silver nanowire random network was consistent in terms of Joule heating. In each voltage cycle, the surface temperature of the silver nanowire random network increased to ~75 °C and returned to the initial temperature after the voltage was turned off. Moreover, the surface temperature profiles for each on–off cycle were almost identical, with a rise time of~134 s and decay time of ~171 s.

Fig. 3. 
(a) Relationship between electric power and the increase in surface temperature of silver nanowire random network embedded in hybrimer substrate. (b) Time-dependent surface temperature profile of silver nanowire random network when 5 V on/off cycles are applied.

However, the silver nanowire random network failed to operate for a long period. As the period increased, the sheet resistance of the silver nanowire random network increased and the surface temperature decreased until electrical breakdown of the network occurred. The broken down sample was investigated using a scanning electron microscope, as shown in Figure 4. There was no sign of silver nanowire melting at the surface. However, it should be noted that the melting point of silver is ~ 961.8 °C, although it had been reported to be lower for silver nanowires [16]. If the melting of the silver nanowire had interfered with the breakdown of the silver nanowire random network, traces of polymer melting should have been observed. However, we could not observe any sign of polymer melting on the surface of the silver nanowire random network embedded in the hybrimer (the melting point of the hybrimer was 400 °C) [15]. Instead of the silver nanowire melting, we observed morphological changes at the locations where the silver nanowires were exposed to the environment. The sidewall surfaces of the silver nanowires became rough and nanoparticles were observed to form on the sidewall surfaces. This result agreed well with the result of Khaligh et al. [13], which stated that silver nanowires could be unstable with the formation of nanoparticles at the sidewall surfaces at elevated temperatures, leading to disconnection of the electrical network.

Fig. 4. 
SEM images of silver nanowire random network embedded in hybrimer after breakdown. Red arrows indicate the regions of physical breakdown within nanowires

In order to confirm the hypothesis, a numerical analysis was performed. As explained in the experimental section, the numerical simulation model shown in Figure 5(a) was utilized. The surface temperature of the numerical analysis model was raised up to 373 K from 273 K. The surface temperature profile of the numerical analysis model was investigated and is shown in Figure 5 (b-c). The result shows that the difference between the silver nanowire temperature and substrate temperature was very small (~0.1 °C). Therefore, the surface temperature during Joule heating could be considered highly uniform, which implied that melting may not have been the reason for the electrical breakdown of the silver nanowire network. As explained above, the formation of nanoparticles at the exposed areas of the silver nanowires could be the main cause of the electrical breakdown of the silver nanowire network.

Fig. 5. 
(a) Numerical analysis model of silver nanowire network for Joule heating experiment. (b) Numerical analysis temperature distribution result for Joule heating of silver nanowire network. (c) Temperature profile along the red line in figure (b).

We studied the Joule heating of silver nanowire random networks embedded in hybrimer substrates. The silver nanowire random networks embedded in hybrimer showed good response to Joule heating and consistent changes in temperature. However, an electrical breakdown could be observed during the long-term operation of Joule heating. The SEM images agreed well with the literature that stated that the silver nanowires became unstable at elevated temperatures and no signs of melting were observed. Numerical analyses were performed to estimate the local temperature of the silver nanowire and the substrate temperature; however, the results showed uniform temperature distribution. Therefore, it was unlikely that the melting of the silver nanowire caused the breakdown of the electrical network. Instead, nanoparticle formation could be the main cause of the breakdown.