To synthesize WO₃, sodium tungstate (Na₂WO₄) was dissolved in deionized water without further purification. Its aqueous solution was passed at a fixed rate through a glass column packed with protonated cation exchange resin (TRILRTE SCR-BH), and the yellow effluent containing tungstic acid (H₂WO₄) was collected in a beaker, as shown in Figure 12728. The tungstic acid solution was mixed with isopropyl alcohol to improve the coating quality and then spin-coated on a glass substrate. The resulting thin layer was baked at 200°C to convert the tungstic acid to WO₃. Subsequently, Ag (99.99%) was thermally evaporated on the WO₃ layer in a high-vacuum chamber with a base pressure of 2 × 10⁻⁶ Torr. Finally, the tungstic acid solution was spin-coated on the Ag layer again and baked to complete the WAW multilayer as shown in Figure 2a. The thicknesses of the layers were measured by a surface profiler (Ahpha-Step 500, KLA-Tencor, Milpitas, CA, USA). The sheet resistance and optical transmittance were measured by the four-point probe technique (CMT-SR2000N, AIT, Gyeongi, South Korea) and a UV-VIS-NIR spectrophotometer (V-570, JASCO Inc., Easton, MD, USA), respectively.

The PLEDs on the WAW multilayer anode were composed of layers of glass, the WAW multilayer, poly(3,4-ethylenedioxythiophene) poly(styrenesulfonate) (50 nm, PEDOT:PSS; Clevios P VP AI4083, Heraeus Precious Metal, Hanau, Germany), yellow light-emitting polymer (60 nm, ‘super yellow’ (SY), Merck & Co., Inc., Whitehouse Station, NJ, USA), ZnO NPs and ionic interlayer (20 nm, polyethylene oxide + tetra-n-butylammonium tetrafluoroborate), and Al, as shown in Figure 2b. The layers of WO₃, PEDOT:PSS, super yellow, and ZnO NPs and the ionic interlayer were all fabricated by spin-coating, whereas the Ag and Al electrodes were thermally evaporated. PLEDs on a conventional ITO anode were fabricated as control devices for comparison. The resulting active area of each PLED is approximately 4.4 mm². After the fabrication, the devices were measured without further encapsulations in air. The device characteristics of the current density, voltage, and luminance (J-V-L) curves were recorded with a source-measure unit (Keithley 2400, Keithley Instruments Inc., Cleveland, OH, USA) and a calibrated photodiode (CS-100A, Konica Minolta Optics, Inc., Tokyo, Japan).

We passed an aqueous solution of Na₂WO₄ through a protonated cation exchange resin. The Na⁺ ions were replaced with H⁺ ions by the exchange resin, and we collected tungstic acid consisting of H⁺ and WO₄²⁻ ions. The tungstic acid was converted to WO₃ ·2H₂O as a primary product, and then WO₃ ·2H₂O was calcined at 200°C for conversion to WO₃ as a secondary product 28.

The WO₃ ·2H₂O particles were slowly synthesized from tungstic acid at room temperature, and the slow synthesis increased the particle size. Hence, the WO₃ derived from these WO₃ ·2H₂O particles had irregular square plates with widths of 500 to 1,000 nm and thicknesses of 200 to 500 nm. Figure 3a shows a FESEM image of the WO₃ particles. These large WO₃ particles cannot form thin dielectric layers; a DMD multilayer is generally required to be thinner than 100 nm. To form a thin WO₃ layer, we devised a new method that synthesized tiny WO₃ NPs. The tungstic acid solution was coated on a glass and then baked at 200°C for rapid conversion and calcination. The rapid conversion obstructed the growth of WO₃ ·2H₂O particles, and WO₃ NPs were synthesized from the WO₃ ·2H₂O particles after a calcination process. Figure 3b shows a TEM image of WO₃ NPs less than 2 nm in size synthesized by the new synthesis method. Because these WO₃ NPs are sufficiently small, the thickness of the WO₃ layer can be controlled by adjusting the concentration and the coating condition of the tungstic acid solution in an air environment. Therefore, these WO₃ NPs can form a thin dielectric layer to optimize the optical properties of the WAW multilayer.

This method has another advantage. Because the WO₄²⁻ ions in tungstic acid were completely dissolved in deionized water, this aqueous solution could coat a uniform thin layer on the substrate without aggregation. The morphology of the WO₃ layer was investigated by atomic force microscope (AFM; XE-100, Park Systems Corporation, Suwon, South Korea). Figure 4a shows that the surface of the WO₃ layer was homogeneous and flat, with a low roughness (root mean square (rms) roughness = 0.457 nm, peak-to-valley roughness = 3.778 nm]. This WO₃ layer with excellent surface quality can form WAW multilayer which has lower surface roughness than conventional transparent anode.

To verify the conversion of WO₃ ·2H₂O to WO₃, we observed the crystallization of the WO₃ NPs using an X-ray diffractometer (RIGAKU D/MAX-2500, Rigaku Corporation, Tokyo Japan). Solution-processed WO₃ should be pure without WO₃ ·2H₂O. WO₃ ·2H₂O reduces the hole and the electron mobility of the dielectric layer of the WAW multilayer anode as impurity, and the low hole and electron mobility cause high turn-on voltage of PLEDs 16. Figure 4b shows the XRD pattern of the solution-processed WO₃. The typical sharp peaks of the monoclinic phase of WO₃ are recognizable in the XRD pattern. This suggests that most of the WO₃ ·2H₂O is converted to WO₃, and solution-processed WO₃ is formed as a crystalline phase. Figure 4c shows the real part (n) and imaginary part (k) of complex refractive indices of the solution-processed WO₃ layer. A 45-nm thick WO₃ layer has higher refractive index than 2.0 in the wavelength range of visible light as in the case of a WO₃ layer deposited using the evaporation process 19.

We fabricated a WAW multilayer with solution-processed WO₃ and investigated its optical and electrical properties as a transparent electrode. The transmittance of the multilayer is a function of the thicknesses and refractive indices of the materials 21. Since the refractive indices of WO₃ and Ag are fixed, the transmittance is determined by the thicknesses of the layers. It has been reported that a Ag layer has optimal electrical and optical properties when its thickness is 15 nm 141619. Therefore, the transmittance of the WAW multilayer should be optimized by controlling the thickness of the bottom and top WO₃ layers.

To optimize the structure of the WAW multilayer, we calculated its transmittance by a transfer matrix method. Figure 5a shows the calculated transmittance at a wavelength of 580 nm for different thicknesses of the bottom and top WO₃ layers. The WAW multilayer had a maximum transmittance of 85% when the bottom and top WO₃ layers were 49 and 45 nm, respectively. Table 1 shows the measured sheet resistance and transmittance of the Ag, ITO, and WAW multilayers. As the thickness of the top WO₃ layer increased from 0 to 60 nm with the thickness of the bottom WO₃ layer kept constant, the optical properties of samples 1 to 5 changed; however, all the samples had the same sheet resistance. These results indicate that the WO₃ layers improve the transmittance without decreasing the Ag conductivity by the tungstic acid solution. Figure 5b shows the transmittance spectra of the samples of Table 1 in the visible range, and Figure 5c shows the transmittance variation of the calculated and experimental results at a wavelength of 580 nm. Sample 4 with WO₃ (45 nm)/Ag (15 nm)/WO₃ (45 nm) had the best transmittance (85%). This result indicates that the experimental transmittance is consistent with the calculated result.

Figure 6 shows the morphologies of the WAW multilayer and ITO measured by AFM. The rms surface roughness and peak-to-valley roughness of the WAW multilayer were 0.588 and 5.290 nm, respectively, and those of ITO were 2.773 and 22.548 nm, respectively. The WAW multilayer had a homogeneous and flat surface which is quite smooth compared to the surface of ITO. The surface roughness of the WAW multilayer anode is an important factor limiting the performance of PLEDs because thin layers of PEDOT:PSS and a light-emitting polymer (SY) must be in direct contact on the anode 6.

Consequently, the solution-processed WAW multilayer had a sheet resistance of 4 Ω/sq, an rms of 0.588 nm, and a transmittance of 85%, which is considerably high as compared to the 50% transmittance of a Ag layer. As a transparent anode for PLEDs, the solution-processed WAW multilayer is superior to ITO. This is because the WAW multilayer has better electrical conductivity and surface roughness than ITO, although its transmittance is slightly lower.

To evaluate the electrical and optical properties of the WAW multilayer as a transparent anode, we fabricated PLEDs based on the optimized WAW multilayer anode and the conventional ITO anode. The layer structure of the PLEDs was as follows: WAW multilayer/PEDOT:PSS/Super Yellow/ZnO NPs and ionic interlayer/Al. Figure 7 shows J-V-L characteristics of the WAW-based and ITO-based PLEDs. The turn-on voltages were identical (2.1 V). This implies that the hole-injection behavior of the solution-processed WAW multilayer was as good as that of ITO with the vacuum process because the turn-on voltage is one of the factors measuring the hole-injection property of an electrode 1520. The maximum luminance of the WAW-based PLEDs reached 35,550 cd/m² at 8 V, whereas ITO-based PLEDs had a maximum luminance of 15,095 cd/m² at 8 V. The maximum luminous efficiency of the WAW-based PLEDs (6.24 cd/A) was also higher than that of the ITO-based PLEDs (6.10 cd/A) as shown in Figure 8. These results indicate that the solution-processed WAW multilayer transparent anode is superior to ITO anode. The high conductivity of the WAW multilayer and the good contact quality of the WAW/PEDOT:PSS interface resulting from the low surface roughness of the WAW multilayer improve the performance of PLEDs 14.