Selective multi-nanosoldering for fabrication of advanced solution-processed micro/nanoscale metal grid structures

Solution-processed metal grid transparent conductors with low sheet resistance, high optical transmittance and good mechanical flexibility have great potential for use in flexible optoelectronic devices. However, there are still remaining challenges to improve optoelectrical properties and electromechanical stability of the metallic structures due to random loose packings of nanoparticles and the existence of many pores. Here we introduce a selective multi-nanosoldering method to generate robust metallic layers on the thin metal grid structures (< a thickness of 200 nm), which are generated via self-pining assisted direct inking of silver ions. The selective multi-nanosoldering leads to lowering the sheet resistance of the metal grid transparent conductors, while keeping the optical transmittance constant. Also, it reinforces the electromechanical stability of flexible metal grid transparent conductors against a small bending radius or a repeated loading. Finally, organic light-emitting diodes based on the flexible metal grid transparent conductors are demonstrated. Our approach can open a new route to enhance the functionality of metallic structures fabricated using a variety of solution-processed metal patterning methods for next-generation optoelectronic and micro/nanoelectronic applications.

Transparent conductors (TCs) are essential components for a variety of optoelectronic devices, including organic solar cells 1 , organic light-emitting diodes (OLEDs) 2 , touch screen panels 3 . Recently, there have been increasing efforts in developing alternative nanomaterial TCs based on carbon nanotubes 4 , graphene 5 , metal nanowires 6 and metal grids 7 to replace indium tin oxide (ITO)-based films. Among these nanomaterial-based TCs, the metal grid TCs have been spotlighted for use in flexible optoelectronic devices due to facile control over their grid width and spacing, scalability to large-area application, uniform sheet resistance and low junction resistance [8][9][10][11] . Especially, the metal grid TCs are typically fabricated using a variety of printing technologies of metal nanoparticles (NPs), including inkjet printing 11 , gravure printing 12 , micro-contact printing 13 and direct imprinting [14][15][16] at low costs and in a high throughput manner.
The solution-processed metal grid TCs show the electrical conductivity from merging of neighboring NPs and eliminating of insulating organic complexes by several sintering processes, including thermal sintering 17 , photonic sintering 18 , laser sintering 19 and chemical sintering 20 . They exhibited the poor electrical conductivity compared with the evaporative metal grids obtained using photo-roll lithography [21][22][23] due to random loose packings of NPs, incomplete elimination of organic complexes and the existence of many pores between NPs 24,25 . Oh et al. 16 carried out experiments on temperature-controlled direct imprinting of silver ionic ink, enhancing the optoelectrical properties of metal grid TCs by the porosity reduction. The unsintered metal grid structures with non-uniform shape and thickness (> μm) suffered from a local damage or a crack generation, resulting in locally unstable mechanical, electrical and optical properties during the mold detachment. Especially, these issues hindered the use of large-scale TCs.
Recently, solution-grown (SG) metal grid TCs fabricated using electroless plating or electroplating have been reported to show excellent optoelectrical properties. Sciacca et al. 26 introduced the SG metal grid TCs generated using electroless plating and soft lithography on a glass substrate. After rapid thermal annealing, the SG metal grid TCs showed superior optoelectrical properties than those obtained using vacuum-based metal deposition: sheet resistance (R s ) of 3.5 Ω sq −1 vs 10.7 Ω sq −1 at transmittance (T 550nm ) of 76%, respectively. Jin et al. 27 suggested a method for directly generating the SG metal grid TCs on polyethylene terephthalate (PET) substrates, which were coated using a poly(dopamine) to improve interfacial adhesion between the SG metal NPs and the substrate. The SG metal grid TCs exhibited R s of 8 Ω sq −1 at T 550nm of 91% and a reasonable electromechanical stability under bending stresses. Also, Khan et al. 28 showed the metal grid structures that are embedded and mechanically anchored onto a flexible substrate using electroplating process. This method could provide flexible metal grid TCs with the high aspect ratio, showing R s < 1 Ω sq −1 at T 550nm = 91%. However, all of these methods suffered from a time-consuming and expensive task related to the generation of polymer masks and the need of dry (or wet) etching process.
Here we introduce a selective multi-nanosoldering (SMN) method for simultaneously enhancing both optoelectrical properties and electromechanical stability of solution-processed metal grid structures without polymer masks and etching process. The thin metal grid structures (< a thickness of 200 nm) can be fabricated using a self-pinning assisted direct inking of silver ions based on the ink capturing. When the mold is detached in a liquid state of ink, the process facilitates the prevention of unwanted structural damages from the metal grids. After the SMN treatment, the metal grid structures based on loosely packed metal NPs were transformed into compact and robust metallic structures. The effect of SMN on the optoelectrical properties of the metal grid TCs was evaluated by considering cycle number, growth time (t G ), growth temperature (T G ), sintering temperature (T S ). Also, the effect of SMN on the electromechanical stability of flexible metal grid TCs after a transfer process was examined under static and dynamic bending stresses.

Results
Self-pinning assisted direct inking of silver ions. The metal grid structures were fabricated on a glass substrate via self-pinning assisted direct inking of silver ions. A schematic illustration of the process is shown in Fig. 1a. Note that the process directly generates a grid-patterned ink, rather than metal NP structures. A silver ionic ink was captured within a grid-patterned mold under heating at 60 °C and pressing at 130 kPa. During solvent evaporation, some of the silver ions were thermally decomposed to silver NPs and organic complexes. After the mold was detached, the meniscus of the grid-patterned ink was kept pinned at contact lines due to the spreading inhibition induced by the confinement of the silver NPs and organic complexes 29 . The grid-patterned ink was thermally decomposed at an elevated temperatures (250 °C) to form the silver NP structures and to eliminate of larger amounts of organic insulating complexes, which cause the generation of many pores inside the metal NP structures 16 . Figure 1b shows a schematic illustration of spreading and self-pinning mechanism of the grid-patterned ink. The spreading behavior of a liquid droplet on substrates is natural phenomenon. In general, it is described in terms of the spreading coefficient (S) that is given by S = σ sg − σ sl − σ lg , where σ is the interfacial tension between two different phases, including solid-gas (σ sg ), solid-liquid (σ sl ) and liquid-gas (σ lg ), respectively. When S < 0, an equilibrium force balance at contact lines is described by Young's equation, given by σ sg = σ sl + σ lg cos θ E , where θ E is an equilibrium contact angle. However, θ E of the liquid droplet was modulated by the NPs confined at contact lines. Weon et al. 29 suggested a simple model to quantitatively describe the self-pinning behavior of colloidal droplet by equating a spreading force (F s = 2πRS) with a capillary force (F c = 2πrNσ lg (cosθ) 2 ) at contact lines, where R is the diameter of droplet, r is the average diameter of NPs, N is the number of NPs at contact lines and θ is the contact angle, respectively. They insisted that the self-pinning behavior of the colloidal droplet should be generated at a critical value of 15° and a critical linear packing fraction of 10%, regardless of the NP size and the initial volume fraction.
Similar to the colloidal droplet, θ E of a silver ionic ink droplet was modulated from the thermally decomposed silver NPs and organic complexes confined at contact lines. Figure S1a shows photographic images for the measured θ E of the ink droplet after ink addition and heating. An ink volume of 5 μl was added to the ink droplet (5 μl). The measured θ E of the ink droplet without heating showed similar (or slightly decreased) values due to solvent evaporation. After heating at 60 °C for 2 min, the ink droplet showed the spreading behavior at θ E = 50°. The measured θ E of the ink droplet increased with an addition of ink due to the thermal reduction of silver ions. The measured θ E of the ink droplet slowly decreased with solvent evaporation. This self-pining behavior is important to guarantee the physical continuity and the rigorous resolution of the grid-patterned ink as well as the capture of large amounts of ink.
The self-pinning of the grid-patterned ink was easily controlled by a heating time at a fixed temperature of 60 °C to minimize unwanted residual layers within grid spacings. In Fig. 1c(i-ii), microscope images show the spreading and self-pinning behavior of the grid-patterned ink after two heating times of 30 and 300 sec, respectively. Figure 1c(i) shows that the continuous grid-patterned ink was divided into irregular droplets, so called dewetting behavior, due to the spreading behavior of the ink and its surface instability. The silver NP structures, generated from the broken droplets, did not have the electrical conductivity regardless of the sintering process. Figure 1c(ii) shows that the grid-patterned ink was kept to an original shape due to the spreading inhibition of confined NPs at contact lines. In Fig. S2a, a wetting behavior of silver ionic ink after the mold detachment was compared at a heating time. After the thermal decomposition and sintering process, the metal grid TCs showed reasonable optoelectrical properties, which is approximately R s of 20 Ω sq −1 at T 550nm of 88.0%. In Fig. S2b, R s and T 550nm of the metal grid TCs for different heating times were compared. Figure 1c Effect of SMN on optoelectrical properties. As the metal grid structures include random loose packings of silver NPs with many pores, they usually have larger electrical resistivity (~1.50 × 10 −7 Ω m) than that of bulk www.nature.com/scientificreports www.nature.com/scientificreports/ silver (1.59 × 10 −8 Ω m). Especially, when the size of the metal grid structures decreases to micro/nanoscale, the resistivity can be considerably increased and locally unstable due to electron scattering at grain boundary of loosely packed NPs 26,30 . Figure 2a shows a schematic illustration of SMN process transforming random loose NP packings of surface of metal grids into robust metallic structures. The SMN method is based on a repetitive process of selective nanosoldering (SN), which consists of solution-growth of silver NPs on the metal grids by electroless plating, elimination of residual silver NPs within grid spacings by cleaning process, and merging of the SG silver NPs and the metal grid structures by thermal sintering. Note that the metal grid TCs treated using SMN or SN are defined as the SG metal grid TCs. Figure 2b(i) shows a scanning electron microscope (SEM) image of the SG metal grid structures after SMN of 4 cycles for t G = 1 min (SMN 1). The SG silver NPs were deposited on the metal grid TCs with minimum residual layers within grid spacings, and then were merged at 250 °C for 1 min. In Fig. 2b(ii), energy dispersive X-ray analysis was taken from the grid line (A) and grid spacing (B) of the SG metal grid TCs after SMN 1, respectively. The inset shows magnified spectra of EDX taken from the grid spacing (B) of the metal grid TCs and SG metal grid TCs, respectively. A silver peak in the spectrum of B obtained using EDX was not observed. This result means that the SG silver NP-based residual layers within grid spacings is negligible. Figure 2b(iii-v) shows the SEM images of the top surface of the metal grid structures, and the SG metal grid structures after SMN of 2 and 4 cycles for 1 min, respectively. The SG silver NPs were merged with loosely packed metal NPs presented on the top surface of the metal grid structures. After SMN 1, the top surface of the SG metal grid structures was changed to more compact and denser metallic structures. Figure 2b(vi-viii) shows the cross-sectional focused ion beam (FIB)-SEM images of the metal grids, and SG metal grids after SMN of 2 and 4 cycles for 1 min, respectively. In Fig. 2b(vi), the metal grid structures exhibited many grain boundaries of NPs and pores, which are main sources of the poor electrical resistivity of metal structures. After SMN, the SG metal grid structures showed more interconnected and continuous connection of metal NPs in Fig. 2b(vii) and (viii).
In SMN, the cleaning process played a critical role in decreasing the SG silver NP-based residual layers within grid spacings, while it led to gradually decreasing R s of the SG metal grid TCs due to a slight loss of the SG silver www.nature.com/scientificreports www.nature.com/scientificreports/ NPs on the metal grid structures. In Fig. 3a, R s and T 550nm of the SG metal grid TCs treated using SN for t G = 1 to 30 min were compared by considering the cleaning process. For SN within t G of 10 min, R s of the SG metal grid TCs decreased with keeping T 550nm a constant. Figure 3b shows R s and T 550nm of the SG metal grid TCs during SMN 1. R s of the metal grid TCs decreased from 20.6 to14.5 Ω sq −1 over T 550nm of 87.9%. When t G increased to 5 min, R s of the metal grid TCs decreased to 12.4 Ω sq −1 over T 550nm of 87.8%. In Figure S3a and b, transmittance spectra of the SG metal grid TCs treated using SMN 1 and 5 are shown. During SMN, a significant reduction in the spectral intensity over a wide wavelength range was not observed. A slight reduction in the spectral intensity near wavelength of 400 nm was generated due to residual metal NPs 9 . Transmittance spectra of the SG metal grid TCs treated using a multi-nanosoldering are compared, as shown in Figure S3c and d. After multi-nanosoldering of several cycles, a significant reduction in the spectral intensity over a wide wavelength range was observed. It is caused by an incomplete elimination of the remaining NP and solution within grid spacings.
Although an increase of t G led to sharply decreasing R s of the SG metal grid TCs, it required a considerable processing time. Instead of SMN for long t G , T G was used to increase the solution-growth rate of metal NPs. Figure 3c shows R s and T 550nm of the SG metal grid TCs treated using SMN 1 at different values of T G . When T G increased from 26 °C to 30 °C, the SMN for 1 min facilitated a considerable decrease in R s of the SG metal grid TCs with less than 1% of T 550nm . The SMN 1 treated at T G = 34 °C led to sharply decreasing R s of the SG metal grid TCs, while it sacrificed approximately 4% of T 550nm . This result means that large amounts of the SG metal NPs, rapidly generated at the increased T G , could be insufficiently eliminated during the cleaning process. Also, R s of the SG metal grid structures was compared with controlling T S after the solution-growth of metal NPs. Figure 3d shows R s of the SG metal grid TCs generated using SMN 1 at different values of T S ranged from 150 to 300 °C. R s of the SG metal grid TCs did not decrease at T S = 150 °C. However, when T S increased from 200 °C to 300 °C, R s of the SG metal grids rapidly decreased to lower values.
Optoelectrical properties of the SG metal grid TCs could be evaluated using the following equation; T = (1 + 188.5(σ Op /σ DC )/R s ) −2 , where σ Op and σ DC are the optical and direct current conductivity of the nanostructure-based TCs, respectively. The term σ DC /σ Op is used as a figure of merit (FoM), whose larger value www.nature.com/scientificreports www.nature.com/scientificreports/ represents better TC performance. Figure 3e shows FoM values of the SG metal grid TCs according to SMN 5. The FoM value of the SG metal grid TCs increased from 150 to 252 by SMN. On the other hands, multi-nanosoldering for 5 min without the cleaning increased the FoM value to 193, and then decreased to 19 due to a significant loss of T 550nm . Figure 3f shows R s and T 550nm of the SG metal grid TCs with different widths after SMN 1. When a linewidth decreased with keeping T 550nm of 90% by a geometrical consideration, the metal grid TCs showed the higher sheet resistance due to electronic scattering and local defects. Especially, the metal grid TCs with a linewidth of 3 μm showed R S of ~kΩ, which lowered to few hundred Ω after SMN 1. This result means that the SMN treatment leads to enhancing optoelectrical properties of micro/nanoscale metallic structures fabricated using a variety of solution-based patterning processes.

Effect of SMN on electromechanical stability. The electromechanical stability of metal grid structures
against bending stresses is one of the most important issues for use in next-generation flexible optoelectronic devices. The SG metal grid TCs were transferred, and embedded into transparent and UV-curable polymer films of Norland Optical Adhesive 81 (NOA 81) to evaluate the electromechanical stability. In general, a transfer process of the metallic TCs leads to lowering a surface roughness, to giving a flexibility and to protecting them from mechanical stress and surface oxidation. After the transfer process, a transmittance spectrum of the flexible SG metal grid TCs was compared with those obtained from a commercially available ITO-coated PET film in Fig. 4a. A transmittance spectrum of bare NOA 81 film was shown as a reference. It should be noted that the transmittance spectra through the substrate are measured. Optoelectrical properties of the flexible SG metal grid TCs (R s of 13.1 Ω sq −1 and T 550 nm of 78.8%) showed superior than those obtained using the ITO-coated PET film (R s of 15.1 Ω sq −1 and T 550 nm of 75.4%). The transmittance spectrum of the ITO-coated PET film decreased with significant fluctuations over the wavelength range of 400-1500 nm. On the other hand, the transmittance spectrum of the flexible SG metal grid TCs remained constant. The inset shows a photograph of the SG metal grids TCs attached on a test tube, demonstrating its flexibility and transparency.  4 Ω, respectively) showed superior electromechanical stability than the ITO-coated PET film under static bending stresses. However, R/R o of the flexible metal grid TCs significantly increased at r ≤ 2.5 mm due to serious damage. R/R o of the flexible metal grid TCs treated using SN decreased with increasing t G at r = 1 mm. The flexible metal grid TCs treated using SMN 1 showed lower R/R o than those obtained using both SN 1 and SN 5. When the flexible metal grid TCs were bent up to r = 1 mm, the values of ΔR/R o increased by 376% and 36%, respectively. Also, ∆R/R o of the flexible metal grid TCs were evaluated under an extremely localized bending stress that is so called folding. Figure 4c shows ∆R/R o of the flexible metal grid TCs under folding (r < 0.3 mm), respectively. The metal grid TCs underwent a ∆R/R o of more than 1360%. The flexible metal grid TCs treated using SN 1 and SN 5 showed a considerable decrease of R/R o with increasing t G (∆R/R o of less than 460% and 240%, respectively). The flexible SG metal grid TCs treated using SMN 1 showed ∆R/R o of less than 85%. The inset shows a photographic image of LED integrated circuit operation connected with the flexible SG metal grid TCs under folding. This result reveals that the SMN treatment leads to enhancing the electromechanical stability of the flexible metal grid TCs even under an extremely localized bending stress. Figure 4d plots ∆R/R o obtained from the flexible metal grid TCs (no treatment, SN 1, SN 5 and SMN 1 of 4 cycles, respectively) under dynamic bending stresses with r = 7.5 mm. ΔR/R o of the ITO-coated PET film was evaluated as a reference. Although all of the flexible metal grid TCs showed similar values of ∆R/R o under static bending stresses with r = 7.5 mm, they exhibited different aspects of ∆R/R o under dynamic bending stresses. After 1000 cycles of repeated bending/relaxation, ∆R/R o obtained from the flexible metal grid TCs increased by more than 1000%. The flexible metal grid TCs treated using SN 1 and SN 5 showed a considerable decrease of ∆R/R o with increasing t G (∆R/R o of l than 700% and 500%, respectively). On the other hand, the flexible metal grid TCs treated using SMN 1 showed ∆R/R o of less than 30%. The SMN treatment results in reinforcing the mechanical strength of the metallic structures, which leads to enhancing the electromechanical stability under static bending, folding and dynamic bending stresses.
Applications to flexible OLEDs. Using the flexible metal grid TCs, the green phosphorescent OLEDs were fabricated as shown in Fig. 5a. The highly conductive PEDOT:PSS (PH 1000 with 5 wt% of DMSO, 50 nm, 242 Ω sq −1 ) was used as a buffer transparent electrode on top of the flexible metal grid TCs. The low conductive PEDOT:PSS (PVP AI4083, 50 nm) and MoO 3 were used as a hole injection layer 31 . On top of the MoO 3 , organic layers and metal electrode were sequentially deposited using thermal evaporator, and the detailed configuration is described in the experimental section. The flexible metal grid TCs-based OLEDs exhibited superior electro-luminous properties than those obtained using the flexible metal grid TCs due to excellent optoelectrical properties. Also, since the flexible metal grid structures is relatively compact and robust, the flexible metal grid TCs-based OLEDs showed leakage-free performances due to a minimal breakage during the transfer process. On the other hand, the flexible metal grid TCs-based OLEDs showed a large leakage current due to a breakage of the loose and weak metal grid structures during the transfer process that leads to rough bottom surface of the device (seen below 3 V in Fig. 5b). For this reason, the proposed device exhibited external quantum efficiency (η EQE ) and power efficiency (η PE ) as high as 18.9%, and 47.6 lm/W at 103.7 cd/m 2 , while the flexible metal grid TCs-based OLEDs showed the limited performance as low as 15.2% and 33.8 lm/W at 95.0 cd/m 2 , as shown in Figs. 5c and d. Figure 5e shows electroluminescence (EL) characteristics of the proposed and ITO-coated device. The proposed devices show the color stable performance with respect to viewing angle, while the devices based on ITO-coated PET exhibit the weak angular color shift due to its high refractive index that is prone to make weak microcavity effects in the device 32 . Moreover, the flexible OLEDs was evaluated to verify performance stability under dynamic bending tests with r = 4.3 mm. As a result, the flexible metal grid TCs-based OLEDs (tensile strain (ε f ) of 1.97%, d Sub-NOA : 170 μm) shows relatively superior luminance characteristics even after 1000 cycles of repeated bending/ relaxation, however, the devices based on the flexible metal grid TCs and ITO-coated PET exhibit degraded performances under the same conditions (r and applied voltage), as shown in Fig. 5f.

Discussion
In summary, we introduced the fabrication of the metal grid structures based on self-pinning assisted direct inking of silver ions, and the demonstrated the SMN treatment for transforming the loose and weak NP networks of the metal grid structures to compact and robust metallic structures. SMN led to not only decreasing R s of the SG metal grid TCs without degrading T 550nm , but also enhancing the electromechanical stability of the flexible metal grid under static bending, folding and dynamic bending stresses. SMN can be used with chemical or photonic sintering to enhance optoelectrical properties and electromechanical stability of various metallic structures fabricated on flexible substrates in roll-to-roll system. Also, SMN facilitates the reinforcement of optoelectrical properties and electromechanical stability of random loose micro/nanoscale metallic structures. This method is particularly a promising candidate for producing next-generation optoelectronic and micro/nanoelectronic devices based on advanced metallic structures with high transparency and conductivity, reinforced mechanical strength and flexibility.

Methods
Materials. The silver ionic ink (TEC-IJ-060, Inktec) with a metal concentration of 12 wt%, consisted of silver ions (silver alkyl carbamate complexes), a base solvent (methanol and toluene), and additives. Before self-pinning assisted direct inking of silver ions, the silver ionic ink of 0.2 ml was heated at 100 °C for 5 min to evaporate the base solvent. The silver alkyl carbamate complexes were decomposed to silver NPs, carbon dioxide, and the corresponding alkyl amines by heating of 60 °C for a few minutes 33 . Fabrication of the PDMS mold. The PDMS solution (Sylgard 184, Dow Corning) was formed by mixing the silicon elastomer kit and a curing agent (10:1), and this mixture was poured onto a SU-8 master. After PDMS curing at 100 °C for 1 hour, the PDMS replica of the grid-patterned mold was carefully released from the SU-8 master. The grid-patterned cavity was designed with a width (w) of 3, 5, 10, 15 μm and a spacing (s) by considering the geometrical shadow zone, T = s 2 /(s + w) 2 . The value of h is a half of w to prevent destruction of the silver NP structures during the detachment of PDMS mold.
Self-pinning assisted direct inking. The silver ionic ink (~10 μL) was dispensed onto a glass substrate. It was squeezed and filled inside a grid-patterned mold under low pressures (P < 120 kPa) and low temperatures (T = 60 °C). During solvent evaporation, some of the silver ions were thermally decomposed to silver NPs. Even after the mold detachment, the grid-patterned ink was maintained due to packings of silver NPs at contact lines of the meniscus. The grid-patterned ink was completely decomposed at heating of 150 °C to eliminate large amounts of organic complexes derived from the thermal decomposition and to improve the thermal decomposition rate of the silver ions. The silver NP structures were sintered at T S = 200-300 °C for 1 min.
Selective multi-nanosoldering. The SN process consisted of an electroless plating, a cleaning process and a sintering process. The electroless plating was used to generate the SG silver NPs on the metal grid TCs from the reaction of a solution (1:1.6) containing silver nitrate (Tollen's reagent) and a reducing sugar (glucose) at 30 °C. After the cleaning process, the SG silver NPs were merged with the metal grid structures by a thermal sintering. transfer process. A glass substrate (sodalime) was treated using 1 H, 1 H, 2 H, 2H-perfluorooctyltrichlorosilane (448931, Sigma-Aldrich) for 20 min in a vacuum chamber. The SG metal grid TCs on the glass substrate were re-treated with the fluorinated silane for 5 min. The NOA 81 solution was poured onto the SG metal grid TCs on the glass substrate, and it was covered by another fluorinated glass substrate (borosilicate) to generate a uniform, thin and transparent NOA 81 film (below 200 μm). A thickness of the NOA 81 films was controlled using a stack of 3 M magic tape. The NOA 81 solution were cured to the NOA 81 film under UV exposure. After the glass substrate (borosilicate) was firstly detached from the NOA 81 film due to relatively weak adhesion energy, the flexible metal grid TCs were separated from the glass substrate (sodalime).
Cross-sectional profiles of the SG metal grid TCs were measured using FIB-SEM (Helios Nanolab 600, FEI), and the surface roughness of the SG metal grid TCs was measured using AFM (XE-100, Park Systems). The transmittance spectra were measured using a UV-VIS-NIR spectrophotometer (Lambda 1050, Perkin-Elmer). The R s of the SG metal grid TCs was measured using the two-terminal method and four-point probe method (4200-SCS, Keithley). Two electrodes between the metal grids, separated by a square area (25 mm 2 ), were fabricated using conductive pens of CW2200MTP and CW2900 (ITW Chemtronics). The electro-luminance characteristics of OLEDs were measured using a source meter (2400, Keithley), a fiber-optic spectrometer (EPP2000, StellarNet) and a calibrated Si-photodiode (FDS-100, Thorlab) in N2-filled glove box. The motorized goniometric system was used to characterize angular electroluminescence profile of OLEDs. The cyclic bending test of OLEDs was performed using a custom-made bending tester that can control the number of cycles and a bending radius.