As shown in Fig. 1a, we studied the electrical stability of pristine AgNW network film and chitosan-ascorbic acid (Chi-AsA)/AgNW composite film. The IR thermal images show the broken pristine AgNW film while the composite film is still in good state. To test multiple samples simultaneously, we built up an electrical measurement system as shown in Fig. 1b. The samples were connected in series through a screw type wire connector with current resource, and in parallel with voltage measuring system. In this way, we can keep the same current flow into every sample, and measure the individual voltage from every sample. We fixed the initial resistance of the films around 13 Ω. Constant current mode was applied to maintain the current density at 100 mA cm-1 and 200 mA cm-1, respectively. The lifetime is defined until the moment when the sample lost conductivity under current stress [9, 29]. As shown in Fig. 1c, the pristine AgNW film demonstrated almost linearly increase in resistance from 13 to 32 Ω under 100 mA cm-1 and then spiked up at 729 h, suggesting the disconnection among AgNW networks at this point. In contrast, Chi-AsA/AgNW composite film maintained stable conductivity for more than 24000 h, and its resistance only slightly changed from 13 to 18 Ω. The electrical stability is significantly improved after introduction of Chi-AsA to AgNW networks. We further increased the current density to 200 mA cm-1. The composite film showed resistance change below 30% during a long duration of 846 h, and maintained in good condition until 889 h, while the pristine one quickly exceeded 30% in resistance only after 34.5 h and completely lost its conductivity in 82 h (Fig. 1d). Although lifetime drops predictably with current density, the lifetime of composite at 200 mA cm-1 is still longer than that of the pristine film at 100 mA cm-1. Since patterning technology is essential for electronic devices, we further tested the patterned films in a width of 350 μm at the current density of 200 mA cm-1. It is found the pattern has similar long lifetime in resisting current stress as the bulk composite film as shown in Fig. 1e. As illustrated in Fig. 1f and Table 1, the previously reported composite AgNW films mostly operated less than hundreds of hours at the current density below 100 mA cm-2. Our composite worked normally for more than 24000 h under 100 mA cm-1 (i.e. 125 mA cm-2), exhibiting much better stability under current loading. As most devices operate below 100 mA cm-1, the composite film with improved electrical stability to ~3 years at 100 mA cm-1 could enable long-term stable devices to meet the commercial longevity requirement.
The mechanism behind the improvement of electrical stability is highly interesting. First, we used dual-probe ohmmeter to check the conductivity across the failed film after longevity current loading test. As shown in Fig. 2a, one probe of the ohmmeter was secured at silver paste connection electrode coated on one side of the composite film, and another probe was moved freely to contact with any points. Afterwards, change the fixed probe to another silver paste contact electrode and repeat the check. We found that actually most regions across the failed film still had good conductivities, and the resistances only slightly shifted to ~20-40 Ω. However, there was always one site with resistance larger than 2000 Ω when the probe was moved up and down, perpendicular to the current flow direction. It suggests a disconnection line perpendicular to the current flow. We then used IR thermal camera to track the failure process, and found most samples kept uniform thermal distribution in the early stage, and then local hot spots showed up and grew to a wave line in the latter stage (Fig. 2b). It demonstrates that the network undergoes slow and widespread degradation for a period. After that, the AgNWs in some weak regions degrade to a critical state and begin accelerated degradation, appearing as hot spots. Then the current in the broken area redistributes to the surroundings vertical to current flow. The increased current density and corresponding higher temperature accelerate the degradation of surroundings, and in turn, affect further region. Thus, the crack extends vertically to the current flow in this cascade process. When the remaining AgNWs along the wave break are insufficient to withstand the crowded current, rapid electrical failure occurs in the AgNW film, consistent with the phenomenon in Fig. 1c-d.
The optical and SEM images further confirmed the failure mode. As shown in Fig. 2(c, f, i), an obvious waved break formed in Chi-AsA/AgNW composite network and pristine AgNW network. In the enlarged SEM images, the composite electrode shows smooth silver nanowire without obvious damage in the region aside the waved break (Fig. 2d), and serious silver migration inside the break (Fig. 2e). It suggests that the degradation is only limited along the wave break. Such distinct degradation in different positions has been seldom reported in long-term electrical failure of AgNW composite electrode. In the enlarged SEM images of pristine AgNWs, serious damages are observed inside the waved crack, and across the whole film as well. Disconnections in the cross-linked junctions and hillocks on surface of nanowires appeared both at 200 and 100 mA cm-1 (Fig. 2g-k). So compared to the composite, two main differences are found. First, inside the waved crack, the silver migration was mainly limited along the nanowire and less nanoparticles formed among network voids in AgNW composite, while the silver migrated more freely in pristine one. Nanowires fractured into particles, migrated into aggregations, and moved to network voids. Second, in the region aside the waved crack, silver nanowire kept its original shape and morphology well in composite AgNWs, while degradation appeared everywhere in pristine one. We could hypothesize that the over-coating layer efficiently stopped the silver migration under current. Besides, several local failures similar with the reported hotspots were also found in the AgNW composites (Fig. S2), especially when there is an external contaminant nanoparticle on the surface.
The position of wave break in pristine samples depends on the role of Joule heating and electromigration in the failure mechanism as we reported [44]. The wave break locates in the middle region of the film when rapid thermal fusing dominates the short-term failure, and moves towards anode when electromigration and electrochemical effects play a role in the long-term failure [44]. We sum up the locations of the wave break of failed samples with initial resistance of 10~15 Ω at 200 mA cm-1 in Fig. S3. For pristine samples, the wave break mainly occurs near the anode. By contrast, the composite samples show a more complicated failure locations, which is regarded as the result of the effective inhibition on electromigration by Chi-AsA coating. Due to the suppression of electromigration near the electrode in composite samples, the influence of other factors, such as high temperature in the center area, local inhomogeneity of the network, and impurities in the film, might be prominent, resulting in various break locations.
We then studied the function of Chi-AsA coating via changing the covering position shown in Fig. 3a. First, the Chi-AsA was only coated on the central area of AgNW networks and exposed the area near connection electrodes (Sample I). The lifetime at current density of 200 mA cm-1 was drastically reduced from 889 to 116 h compared with fully-covered composite film (Fig. 3b). The wave break appeared at the exposed AgNWs area next to covered area in the side towards positive connection electrode, exhibiting obvious migration of silver to form spheres (Fig. 3c). Second, the Chi-AsA coating was extended to the whole area of AgNW networks, but the part beneath connection electrodes was still the bare AgNWs (Sample II). The sample exhibited improved lifetime to 561 h, and the wave break located at the vicinity of the positive connection electrode (Fig. 3e). It is similar with Sample I, in which the wave break both happened near the boundary of Chi-AsA coating. Third, only around 30% of the AgNW network in the central area was exposed, and all other area including the part beneath connection electrodes was coated with Chi-AsA (Sample III). It showed a lifetime of 382 h, which is much higher than the first sample type, but lower than the second one. The wave break line appeared at the border between the covered and the exposed networks with serious silver migration (Fig. 3g). Pristine AgNWs inside exposed area also exhibited obvious degradation with disconnected nanoparticles on the surface of AgNWs and gaps among network, and some big cluster of silver agglomeration at the contact junctions (Fig. 3h). For all three types of samples, AgNWs covered with Chi-AsA remained intact network, good contact junctions and smooth surface after the electrical failure of the samples (Fig. 3d, f, i).
It demonstrates the effectiveness of Chi-AsA protection on electrical stability. The main failure always occurred at the boundary region of Chi-AsA coating no matter where it is. When the location of Chi-AsA covering layer moves, the wave break line follows instead of only appearing near the connection electrode. Inside the failure area of wave break, rut voids along the deposition path of silver nanowire left after silver migration, which is similar with the morphology of fully-covered composite samples. It is considered that the silver atoms must break through the obstacles of Chi-AsA to migrate. Since the Chi-AsA polymer is coated on the surface and fills the voids of AgNW network, silver atoms are strongly suppressed from moving along the AgNWs or migrating to the surrounding area under current stress. Thus, the electrical and thermal stability of Chi-AsA/AgNW film is significantly enhanced. In addition to electromigration, the degradation caused by concentrated Joule heating at the junctions also contributes to the electrical failure [22, 36]. Since Chi-AsA layer helps form a tighter contact and smaller contact resistance between AgNWs [14], the risk of contact junctions being molten with Joule heating is reduced compared with pristine AgNW networks. Besides, it is learned from sample II that even all AgNWs are covered with Chi-AsA, the lifetime under current stress is still lower than that of the fully covered one in which the part beneath connection electrode is also protected with Chi-AsA. It indicates that the protection of AgNWs beneath conductive connection points is also important. It requires the coating material be thin enough to transfer electron transfer between AgNW networks and connection electrodes, but strong enough to provide sufficient protection on the AgNWs from current stress. Some traditional insulating polymers such as polydimethylsiloxane or epoxy can also protect AgNW networks as overcoatings, but generally make the composite surface insulate from the top connection electrode. In contrast, Chi-AsA layer with tens of nanometers can deposit on AgNWs beneath connection points with unchanged conductivity and good protection.
According to electrical failure theory of bulk metal film, electromigration highly depends on the current density and temperature. The current density on single AgNW is related to both applied current value and network density. We fabricated pristine and composite samples with various AgNWs deposition densities by different coating times. The higher deposition density represents lower transmittance. As shown in Fig. S4, the transmittance at 550 nm decreases linearly by 1.1% on average for each additional coating time. We investigated the electrical stability of the composite films with 2~7 coating times and corresponding transmittance of 96.8~91.6%. As shown in Fig. 4a-b, the lifetime of the composite samples increases obviously with AgNWs deposition density. For example, composite film with 96.1% of transmittance only subsisted for 223 h at current density of 200 mA cm-1, which expanded to 1582 h for the one with 92.5% of transmittance. Pristine samples also exhibit the similar relationship between lifetime and transmittance, but the lifetime is below 10% of the composite sample with comparable deposition density. Higher deposition density means lower current density on single AgNW, resulting in slighter electromigration, less Joule heating, and hence better electrical stability. In addition, denser AgNW networks benefit more uniform distribution of current, and thus reduce the occurrence of hotspots [21, 23, 29, 45]. The AgNWs deposition density can be characterized with SEM, and the current density along single AgNW is calculated via dividing the current by the sum of cross-sectional area of all AgNWs on the line perpendicular to current flow as shown in Fig. S5. The calculated current density on single AgNW from SEM images is inversely proportional to the coating times (Fig. S6), which verifies the accuracy of the calculation from SEM images. As shown in Fig. 4c, it demonstrates a power function relationship with a power index of -3.7 between lifetime and current density on single AgNW. It illustrates that the lifetime could exceed 1000 h if the current density of single AgNW is less than 8.4 × 105 A cm-2, which is approximately corresponding to the films with 94% of transmittance at 200 mA cm-1, or 97% of transmittance at 100 mA cm-1. Therefore, as long as the transmittance is among the application requirement, increasing the AgNWs deposition density would significantly improve the lifetime under current stress.
We then studied the performance of composite and pristine AgNWs under elevated current densities, and compared their tolerance threshold to current shock. The AgNW networks were fabricated on polyimide (PI) in this experiment because it can withstand higher current density than glass before brittle fracture induced by thermal stress. Meanwhile, the network on PI substrate exhibits higher temperature than that on glass under similar current stress because of different heat dissipation and thermal capacity, which results in more resistance change. The pristine and composite films with initial resistance of 14 Ω were tested under current density that started from 50 mA cm-1 and increased step by step of 25 mA cm-1 per 10 min. As shown in Fig. 4d, the resistance increased with current density because of higher temperature due to Joule heating. When the current density was below 175 mA cm-1, the composite and pristine films exhibited similar resistance. However, the composite film started to exhibit obvious smaller resistance than the pristine film at 200 mA cm-1, meaning slower aging process in this condition. The pristine film failed at the current density of 250 mA cm-1 when the substrate was intact. By contrast, the Chi-AsA/AgNW composite film performed well until 350 mA cm-1 before the rupture of substrate. It illustrates that the Chi-AsA/AgNW composite film itself can withstand current shock above 350 mA cm-1 on a stronger substrate, much superior to pristine AgNW film.
To understand the influence of temperature on the electrical reliability of composite films, we studied the effect of Chi-AsA coating on the heat dissipation of AgNW network. The film temperature increased after current applied due to Joule heating, and reached steady stage in several minutes (Fig. 5a). The temperature increase is defined as the difference between the room temperature (T0) and steady temperature. As shown in Fig. 5b, the temperature increase changes linearly with the resistance. Both the composite and pristine samples meet the same linear fit at 100 mA cm-1 and 200 mA cm-1, respectively. It suggests that the Chi-AsA coating shows no adverse effect on the heat dissipation of AgNW network. Tens of nanometers thickness of the coating layer is negligible relative to the substrate and shows minimal obstacle to surface heat dissipation. Besides, Chi-AsA enables stronger adhesion between AgNWs and substrate [14], which enhances the heat dissipation through substrate.
Both the internal Joule heating and external thermal environment contribute to the temperature of AgNW films, and thus influence their electrical stability. Chi-AsA/AgNW composite samples of 13 Ω were applied with current from a DC power and heated on a hot plate simultaneously. The sample temperature was controlled by the current density and/or hot plate temperature. Compared with the composite film at 200 mA cm-1 without external heating (~ 50 oC due to internal Joule heating), the lifetime of composite film was shortened from 889 h to 450 h at 200 mA cm-1/60 oC, and further to 136 h at 200 mA cm-1/85 oC (Fig. 5c). Most area except the wave break of the network at 200 mA cm-1/85 oC showed degradation with breakage and aggregations (Fig. 5c-Ⅰ), and that at 200 mA cm-1/60 oC showed some fractures (Fig. 5c-Ⅱ). It is expected that high temperature not only enhances thermal melting, but also provides energy for silver atoms to migrate and accelerates the corrosion, causing poor electrical stability of the film.
Interestingly, although temperature shows significant effect on electrical stability, the lifetime of the sample at same temperature is heavily dependent on current stress. As shown in Fig. 5c, the composite film at 100 mA cm-1 /85 oC showed a lifetime of 866 h, about 6.4 times that at 200 mA cm-1 /85 oC, and only slight damages were observed in most area (Fig. 5c-Ⅲ). Meanwhile, the composite film at 85 oC without current stress only increased 5 Ω in resistance after more than 2000 h, which maintained intact network without obvious fragmentation (Fig. 5c-Ⅳ). It means that even at the same macroscopic temperature, the electrical stability still depends on the applied current density. The effect of current stress is not only to generate Joule heating, but also causing electromigration due to momentum exchange and electrochemical action. In addition, as the source of Joule heating, AgNW network only accounts for a small percentage of the entire film surface area. Thus, the temperature of individual AgNW, particularly at the contact junctions, is expected to be higher than the apparent macroscopic temperature. Even at the same macroscopic temperature by current stress and external heat simultaneously, higher current density actually results in more Joule heating on individual AgNW and thus reduced lifetime.
The importance of the current density to the electrical stability is further confirmed by the comparison of different samples at same temperature. The composite films of 28.8 Ω /100 mA cm-1 and 8.5 Ω /200 mA cm-1 both reached the apparent temperature of 43 oC. However, the 8.5 Ω /200 mA cm-1 film showed a lifetime much longer than that of the 28.8 Ω/100 mA cm-1 film as shown in Fig. 5d. To achieve the same macroscopic temperature, individual AgNW in sparser network (28.8 Ω) should produce more Joule heat, which means higher current density in single AgNW and thus shorter lifetime. Therefore, the current density and temperature in single AgNW are the more important parameters in evaluating the long-term electrical reliability of the network than the apparent macroscopic temperature of the film.
In addition to electromigration and Joule heating, the effect of corrosion on the electrical stability cannot be ignored as well. Ambient substances such as sulfides, oxygen and moisture have been reported to cause corrosion of AgNWs, and sulfidation has been proven to be the dominant effect [24-26]. The corrosion is expected to function synergistically with electromigration and Joule heating, and even probably brings about complex electrochemical effect, thereby accelerating the electrical failure. XPS was taken from the failed pristine and composite films in Fig. 1c-d. As shown in Fig. 6, the ratio of silver to sulfur content was 82: 18 in the pristine AgNW film that sustained for 729 h under 100 mA cm-1. The ratio drops to 96: 4 for the pristine AgNW film under 200 mA cm-1 for 82 h. It suggests that corrosion by sulfur took place under current stress and developed over time. The corrosion of silver decreases the nanowire diameter and thus degrades the network conductivity, causing elevated temperature and enhanced current density in AgNWs, which plays a role in long-term failure process under current. Coating is an effective way to inhibit the corrosion. Remarkably, even though the Chi-AsA/AgNW composite film underwent longer time and higher temperature test at 200 mA cm-1 than the pristine film at 100 mA cm-1, XPS showed no obvious signal of sulfur, demonstrating good resistance to corrosion by ambient chemicals.
AgNW networks are usually integrated with other materials in optoelectronic devices. Some materials are harmful to the stability of AgNW network. AgNW film is often compounded with PEDOT:PSS when used in OLEDs and solar cells as electrode. We imported PEDOT:PSS onto the top of AgNW film and Chi-AsA/AgNW film, both with initial resistance of 10 Ω, and then tested the electrical stability at 200 mA cm-1. As shown in Fig. 7, the lifetime of PEDOT:PSS/AgNW film decreased to merely 109 h compared to 156 h of pristine AgNW film. The result is consistent with the previously reported work that the acidic property and water-absorbing tendency of commercial PEDOT:PSS lead to declining stability of AgNW network [41]. By contrast, the PEDOT:PSS/Chi-AsA/AgNW film subsisted for 751 h. The results illustrate that the anti-permeation property of Chi-AsA coating not only benefits resisting ambient corrosion, but also contributes to suppressing the destructive effect of composite materials on AgNW network. AgNW network is often packaged for protection such as in film heaters. Here we used PDMS as a protective cover on the top of the sample. The PDMS/AgNW sample exhibited improved lifetime of 792 h at 200 mA cm-1. Meanwhile, the lifetime was significantly elevated to 4193 h for the PDMS/Chi-AsA/AgNW composite film. It demonstrates that no matter what kind of materials are composited with, the Chi-AsA/AgNW composite films always show superior advantages in electrical stability in contrast to pristine AgNW films.