RGB‐Color Textile‐Based Flexible and Transparent OLEDs Considering Aesthetics

Recently, as state‐of‐the‐art displays, wearable electronic devices and free‐form displays have begun to attract attention to the point that much research on textile‐based displays has been actively conducted as an ultimate platform for wearable displays. However, although many studies have reported organic light‐emitting diodes (OLEDs) as the most suitable type for textile‐based displays due to their high opto‐electronic performance and flexibility, most research on textile‐based OLEDs (textile OLEDs) has left for the future improvements that require the use of transparent OLEDs (trOLEDs) for applications where aesthetics are important, such as fashion displays and automobile interiors. Herein, RGB‐color trOLEDs are designed using a dielectric‐metal‐dielectric transparent electrode to overcome the limitations of previous works on textile‐based non‐transparent OLEDs and to improve the aesthetic elements of textile displays. The opto‐electronic performance and the mechanical characteristics of textile RGB trOLEDs are analyzed, and specifically, this study suggests a new method and a novel equation for an optical analysis to quantify the clarity of printed color images (i.e., information) on a textile substrate when trOLEDs on the textile substrate are not operated, as needed to optimize and maximize the aesthetic elements of textile displays.


DOI: 10.1002/admi.202202114
only for simple one-dimensional information transmissions from devices to human but also to transmit information among people and devices in both directions and for various purposes, display technology has achieved form-factor innovation, resulting in various types of smart devices, especially wearable displays. Among such innovations, clothing-type displays are attracting attention as highly effective methods of communication among humans and devices and as the most human-friendly technology suitable for the movements of the human body. [1][2][3][4][5][6][7] Moreover, clothing-type displays are a form of state-of-the-art technology that includes multiple functions and aesthetics at the same time, applicable to numerous fields, such as smart textiles, military goods, and even the interior walls of automobiles. [8][9][10][11][12][13][14] Given these technological development trends, clothing-type displays that show organic electroluminescence (EL) fabricated with several techniques, such as inkjet printing, [15] screen printing, [16] or thermal evaporation on a polydimethylsiloxane (PDMS) planarization layer, [17] have been reported. However, with the limitations of the light-emitting mechanisms of EL materials, there are disadvantages in that these devices are driven at high voltages of 165-400 V, thus presenting a serious problem when they are applied to wearable displays. Moreover, the maximum luminance achieved in earlier works related to these devices much less than 1000 cd m −2 , a level considered to be insufficient for outdoor displays.
In this regard, Choi et al. reported extile-based organic lightemitting diodes (textile OLEDs) that resolved the limitations of clothing-type EL displays. [18] A maximum luminance value of 7968 cd m −2 and current efficiency of 14.6 cd A −1 demonstrated sufficient electrical and optical performance capabilities for outdoor clothing-type displays. Also, the report showed that the fabricated planarized textile substrate had mechanical flexibility nearly identical to that of a bare textile, resulting in excellent bending and wrinkling performance for textile displays. Furthermore, with this planarized textile substrate, foldable and washable textile-based OLEDs were reported, showing excellent reliability of the textile platform. [19] As such, these novel approaches have provided electrical and versatility advantages for clothing-type displays; however, for better applicability to various fields such as fashion displays, in which aesthetic Recently, as state-of-the-art displays, wearable electronic devices and freeform displays have begun to attract attention to the point that much research on textile-based displays has been actively conducted as an ultimate platform for wearable displays. However, although many studies have reported organic light-emitting diodes (OLEDs) as the most suitable type for textilebased displays due to their high opto-electronic performance and flexibility, most research on textile-based OLEDs (textile OLEDs) has left for the future improvements that require the use of transparent OLEDs (trOLEDs) for applications where aesthetics are important, such as fashion displays and automobile interiors. Herein, RGB-color trOLEDs are designed using a dielectric-metal-dielectric transparent electrode to overcome the limitations of previous works on textile-based non-transparent OLEDs and to improve the aesthetic elements of textile displays. The opto-electronic performance and the mechanical characteristics of textile RGB trOLEDs are analyzed, and specifically, this study suggests a new method and a novel equation for an optical analysis to quantify the clarity of printed color images (i.e., information) on a textile substrate when trOLEDs on the textile substrate are not operated, as needed to optimize and maximize the aesthetic elements of textile displays.
factors are important, the application of transparent OLEDs to textile displays was found to be essential.
Yin et al. reported textile-based transparent green OLEDs that relied on ultra-thin 7 nm Au film as the anode and 9 nm Ag film as the cathode to improve the aesthetic factors of textile OLEDs. [20] However, although they successfully resolved the limitations discussed in previous works, their method was limited to single-color (green) displays, which are insufficient for information displays. In addition, the metal film was so thin that the surface morphology of the electrode had somewhat poor film quality, possibly leading to a low product yield and poor suitability for large-area displays. Also, transparency optimization was not fully realized, and an optical analysis of the textile OLEDs is lacking for full application to clothing-type displays. In textile-based transparent displays (textile trOLEDs), not only the transparency of the trOLEDs but also the clarity of the images on the textile substrate when trOLEDs are not operating are very important factors that should be considered when optimizing textile trOLEDs.
Unlike other transparent displays, with textile trOLEDs, trOLED films of n% transmittance do not guarantee that n% of the images on the textile (i.e., information on the textile substrate) can be identified; this issues arises because trOLEDs directly and entirely cover the textile substrates such that any external light undergoes multiple stages of light loss. When light passes through trOLED film, it is reflected by the substrate and passes through the trOLED film again. Therefore, it is necessary not only to determine the transmittance of the trOLEDs accurately but also to analyze how visible the information on the textile is. This implies that numerical quantification of the information acquisition outcomes from non-operating textile trOLEDs should be done.
Here, to realize a more practical application of textile displays, red, green, and blue (RGB) colored trOLEDs with high transmittance values are realized by optimizing the thicknesses of a dielectric-metal-dielectric electrode based on a transfer matrix formula; these devices were fabricated on textile substrates and showed outstanding electrical performance. Furthermore, with an in-depth analysis and novel equations to quantify the clarity of the images on the textile when the trOLEDs are not operating, a universal optical analysis method for textile trOLED displays is suggested. In addition, using a multifaceted approach to assess the mechanical properties of textile trOLEDs, such as cantilever and bending tests, a complete analysis of the flexibility of textile trOLEDs was conducted.

Results and Discussion
The most important structural component of transparent OLEDs is the electrodes, and typical materials widely investigated as transparent electrodes include ITO (indium tin oxide), PEDOT:PSS, metal nanowires, graphene, and carbon nanotubes, among others. [21][22][23][24][25] However, because each material has its own disadvantages, such as poor flexibility, low transmittance, or high sheet resistance level, they can be considered as unsuitable for application to flexible wearable displays, especially textile trOLEDs. On the other hand, multilayer electrodes consisting of highly refractive index dielectric layers/thin metal films/high refractive index dielectric layers (DMD) have an advantage in that they exhibit high transmittance, high flexibility, and low sheet resistance. [26] DMD electrodes have excellent flexibility owing to their internal thin metal films. Furthermore, due to their unique structural and optical characteristics, such as suppressing surface plasmon coupling at the dielectric/metal interface, [27,28] the transmittance of thin metal films with relatively low transmittance can be dramatically increased ( Figure S1, Supporting Information). Also, lower dielectric layers acting as metal seed layers to prevent porous metal deposition allow the deposition of continuous thin metal films, resulting in low sheet resistance and improved transmittance. [29] Additional descriptions of the relationship between the transmittance and sheet resistance of DMD electrodes are described in Figure S2 (Supporting Information).
With their suitability for wearable displays, especially for textile displays, DMD electrodes have been used as transparent electrodes for textile RGB trOLEDs. Figure 1a shows the proposed textile trOLEDs when not operating, where the color images on the textile substrate can be clearly identified. Figure 1b shows textile RGB trOLEDs operating at 5 V.

RGB-Color Textile trOLEDs: Optimized Transparency and Opto-Electronic Performance
For the fabrication of the proposed textile trOLEDs, as shown in Figure 1c,f,i, RGB trOLEDs were fabricated on an imageprinted polyester textile substrate.
A transparent DMD electrode made of ZnS/Ag/MoO 3 was used for the anode, and an electrode made of ZnS/Cs 2 CO 3 /Ag/ZnS was used for the cathode for the RGB trOLEDs.
In the trOLED structure, and especially in the DMD electrode structure, the layer with the greatest influence on the transparency is the thin metal layer. [30] Here, due to its good conductivity and lowest absorptivity in the visible wavelength band, Ag is used for the thin metal layer in the DMD electrodes for the cathode and anode. [31] A thicker silver electrode can cause a decrease in the transmittance ( Figure S3, Supporting Information), whereas one that is too thin can complicate the fabrication process and decrease the sheet resistance. [32] As such, an optimized minimum thickness of 8 nm Ag is fixed for the DMD anode and cathode.
To design an effective DMD electrode, it is essential to consider a dielectric material with a high refractive index, such as ZnS, MoO 3 , and WO3; among dielectric materials that meet this requirement, ZnS has relatively high surface energy, allowing it effectively to prevent diffusion from the adjacent Ag film, resulting in a smoother Ag thin film on the ZnS layer. ZnS acts as an excellent seed layer for thin Ag film. For the anode, 5 nm of MoO 3 is used as a capping dielectric layer due to its capability as a hole injection layer (HIL) for the anode. For the cathode, if the seed layer of ZnS is too thick, it could interfere with proper electron injections from the Ag cathode to the emitting layer. As such, 5 nm of ZnS is used as the seed layer and 1 nm of Cs 2 CO 3 is used to improve the electron injection capability. [33] For the capping dielectric layer of the cathode, ZnS is used to improve the transmittance.
After the thickness of the other organic/inorganic layers was optimized considering the charge balance, an optical simulation was conducted using the measured refractive index value of each material and MATLAB software (MathWorks) based on the transfer matrix formula; [34] an optical transmittance simulation was conducted in relation to the top and bottom ZnS  Figure 1e h,k. The similarity between the theoretical calculations and the experimental results validates both the optical transmittance simulation and actually measured refractive index values; with high transmittance results for the RGB trOLEDs (63.94%, 77.26%, and 64.99% at 550 nm for RGB trOLEDs), the applicability to textile trOLEDs can be confirmed.
The optimized RGB trOLEDs were fabricated on textile substrates and showed sufficient opto-electronic performance for outdoor displays.  Figure 2d-f shows respective current efficiency values of 6.07, 3.79, and 2.11 cd A −1 for the RGB textile trOLEDs. Slightly different current efficiency values were noted between the textile trOLEDs and the glass-based trOLEDs due to the characteristics of transparent displays, as explained in Figure S4 (Supporting Information).
Furthermore, the textile RGB trOLEDs show a very low microcavity effect due to the 8-nm-thick Ag film [35,36] such that there is almost no change in the peak wavelength, as shown in Figure 3a-c. The peak wavelength approximates Lambertian emission, as shown in Figure 3d-f for the textile RGB trOLEDs. These results demonstrate that the textile display is scarcely distorted by the viewer's angle, which can be considered a strong advantage for a wearable display.
With the well-optimized transparency and outstanding optoelectronic properties, RGB-color textile trOLEDs can become the new mainstream of wearable displays, leading a new IoT era.

Mechanical Characteristics of Textile trOLEDs
The mechanical characteristics of the textile trOLEDs were evaluated by two types of tests: a cantilever test (ASTM:D1388) [37] to evaluate the mechanical flexibility and a bending test to determine the bending reliability.
As shown in Figure 4a, as the textile specimen is slid along the front and back sides in the direction of an incline, one side of the specimen is bent downward by gravity and finally touches the slope. The length of the slid part until one side touches the slope is defined as the cantilever length. The shorter the cantilever length is, the more flexible the textile specimen is. This means less force is required to bend the textile specimen to a certain radius, and there is less wear-induced stress when the device is actually worn. The cantilever lengths of a bare polyester textile, a textile substrate, a textile substrate to which DMD electrodes are applied, a textile substrate to which all trOLED structures are applied, and other specimens of flexible materials (PET films and A4 paper in this case) are shown in Figure 4b. Furthermore, the bending rigidity (flexural rigidity), which is a measure used to evaluate actual mechanical flexibility and considers the weight differences of

www.advmatinterfaces.de
individual specimens, was calculated (based on Equations (S1) and (S2), Supporting Information). [38,39] These results, shown in Figure 4c, indicate that the textile trOLEDs showed superior flexibility compared to the other flexible substrates, with mechanical flexibility nearly identical to that of the bare textile, confirming that the trOLEDs scarcely affect the flexibility of bare textiles such that wear-induced stress can be minimized when the device is actually worn. Figure 4d shows the textile trOLEDs when they are bent and wrinkled. To evaluate the bending reliability of the textile trOLEDs, a bending test was conducted in two ways; the normalized luminance and sheet resistance of DMD electrodes were analyzed after the bending test with a bending radius in a range from 10 to 1.5 mm and repetitive bending cycles of 1 to 1000 times at a bending radius of 3 mm. As shown in Figure 4e, the textile trOLEDs can work stably up to a bending radius of 3 mm. They are still able to emit light at a radius of 2 mm but had many parallel vertical cracks in the bending direction, as can be seen in the right side inset figure in Figure 4e. With a bending radius of 3 mm, the textile trOLEDs work stably after up to 1000 repetitive bending cycles, as shown in Figure 4f. These results confirm that textile trOLEDs can sufficiently endure not only ergonomic flexion and movement but also the harsh deformation that could occur in everyday life.

Information Acquisition from Non-Operating Textile trOLEDs
To quantify printed color images (i.e., information) on nonoperating textile trOLEDs numerically, the spectrally defined reflectance values observed in the visible light wavelength band by a UV-vis spectrometer (UV-2550) were judged as numerically quantified values of information. [40] For example, as shown in Figure 5a for various bare colored textiles (white, red, green, and blue textiles), the reflectance results show how much information the bare textiles contain at each wavelength from 400 to 700 nm. It was found that the white textiles have information over the entire wavelength range; each differently colored textile has information within a narrow wavelength band (400-470 nm for blue, 480-580 nm for green, and 590-700 nm for red textiles); there is little information in other wavelength bands. However, when trOLEDs are fabricated directly on textile substrates, the reflectance values of the textile trOLEDs contain not only information on the textile substrates but also light reflected by the thin trOLED films, as shown in Figure 5b. Therefore, it is necessary to remove the reflected light at the thin trOLEDs films from the reflectance values of the entire surface of the textile trOLEDs so as to allow a calculation of how much information on the textile substrate can be acquired. Here, by subtracting the reflectance of the thin trOLEDs film from that of textile trOLEDs, [41] it is possible to determine the information acquisition rate from non-operating textile trOLEDs, which can be expressed as follows: where R to denotes the reflectance of textile trOLEDs and R of is the reflectance of the trOLED film, as shown in Figure 5c-e. The information acquisition rate represents the spectrum of information on the textile substrate when the trOLEDs are applied. To calculate how much information is acquired compared to the information the bare textile originally had   : 1388), bending directions of the textile OLED specimen (right). b) Cantilever lengths from the cantilever test. c) Bending rigidity calculated from the results of the cantilever test. d) Textile trOLEDs under deformation when not operated (left) and when operated (right). e) Normalized luminance of textile trOLEDs and normalized sheet resistance of a DMD electrode on a textile substrate after the bending test in relation to the bending radius, and cell images (inset). f) Normalized luminance of the textile trOLEDs and normalized sheet resistance of a DMD electrode on a textile substrate after the bending test in relation to bending cycles with a 3 mm radius, and bending images (inset).

www.advmatinterfaces.de
(the acquired rate of information considering the information of the bare textile as 100%), it is necessary to normalize this result according to the information spectrum of the bare textile, which can be expressed as where R bt is the reflectance of the bare textile, as shown in Figure 5a. Figure 5f-h shows the calculated normalized information acquisition rate from RGB textile trOLEDs when the textile colors are white, red, green, and blue. These results have two implications: the first is that not only the transmittance of the trOLED film but also information acquired from the non-operating textile trOLEDs can be expressed as numerically quantified values in all visible wavelength bands. This suggests which colored textiles are most advantageous in terms of information Figure 5. a) Reflectance of white, red, green, and blue textiles. b) Mechanism schematic of information acquisition from textile trOLEDs. c) Reflectance of red trOLED film and textile red trOLEDs on white, red, green, and blue textiles. d) Reflectance of green trOLED film and textile green trOLEDs on white, red, green, and blue textiles. e) Reflectance of blue trOLED film and textile blue trOLEDs on white, red, green, and blue textiles. f) Normalized information acquisition rate from the textile red trOLED on white, red, green, and blue textiles. g) Normalized information acquisition rate from textile green trOLEDs on white, red, green, and blue textiles. h) Normalized information acquisition rate from the textile blue trOLED on white, red, green, and blue textiles.
www.advmatinterfaces.de acquisition when trOLEDs are fabricated on textile substrates; hence, optimization to select the proper textile color for textile trOLEDs is possible. Second, the normalized information acquisition rate from the red, green, and blue textile trOLEDs showed tendencies similar to that of the white textile. This result shows that the normalized information acquisition spectrum does not change significantly depending on the color of the textiles, even when the trOLEDs are fabricated. In other words, this result means that only the transmittance performance of the trOLED film affects the information acquisition spectrum, whereas the textile color does not. In the approximate wavelength range of 520-610 nm, it can be seen that the information acquisition rate is lower than that of the white textile, which can be interpreted as following the tendency in which the amount of information the bare textile originally had decreased in that wavelength range, as shown in Figure 5a. To extract information from textile trOLEDs, the bare textile used for the substrate must have more than a certain amount of information. Through these two implications, it is expected that calculating the normalized information acquisition rate will become a standard step in the universal optical analysis method for textile trOLEDs applied to various fields in which aesthetic factors are very important.

Conclusion
In summary, RGB-color transparent OLEDs using DMD electrodes were realized to improve the aesthetics of textile displays. With sufficient consideration of the trOLEDs, the transparency was optimized using the Fresnel equation and designed to maximize the electrical properties, resulting in transmittance values of 63.94%, 77.26%, and 64.99% at 550 nm and CE values of 6.07, 3.79, and 2.11 cd A −1 for the RGB trOLEDs, respectively. Furthermore, the maximum corresponding brightness values of 5511, 4534.9, and 1388 cd m −2 demonstrate feasibility for truly wearable clothing-type displays for use outdoors. In addition, multifaceted mechanical tests, here a cantilever test and a bending test, were conducted, showing stable operation of the devices after 1000 trials of repetitive bending, confirming that the textile trOLEDs can fully withstand human flexion or repetitive movements. Furthermore, by suggesting a new optical analysis method (i.e., the normalized information acquisition rate) considering the special characteristics of textile trOLEDs, more flexible designs are now possible by optimizing the aesthetic elements.
With these results of textile displays, in which various elements are perfectly analyzed in a suitable manner, we hope that the era of truly wearable displays will come closer.

Experimental Section
Textile Substrate Fabrication: The planarized textile substrate was fabricated based on the ultra-thin planarization layer implanting method. More details about the basic physical parameters of the bare textile substrate, fabrication method, and fabricated textile substrate are provided in Figures S5 and S6 (Supporting Information).
TrOLEDs Fabrication: DMD electrodes and RGB organic materials (phosphorescent red and fluorescence green and blue trOLEDs) were fabricated on an image-printed polyester textile substrate by thermal evaporation at 5 × 10 -6 torr. The following structure was utilized for red: ZnS (16 nm)/Ag (8 nm Figure S7 (Supporting Information).
Refractive Index Measurement for Transmittance Simulation: The refractive index (n, k) values used in the transmittance simulation were determined from ellipsometry measurements at the National NanoFab (NNFC). The refractive index values of ZnS, Ag, and MoO 3 are plotted in Figure S8 (Supporting Information).
Simulations: A custom-made MATLAB code based on the Fresnel transfer equation was used to calculate the transmittance values of the DMD electrodes and the trOLEDs.
The commercial software LightTools (Synopsis Inc.) was used to determine the visual light path of the trOLEDs for a supplementary explanation of the current-efficiency differences between the textile trOLEDs and the glass-based trOLEDs.
Measurement: A UV-vis spectrophotometer (UV-2550, Shimadzu Inc.) was used to measure the transmittance and reflectance values of the textile substrate, the DMD electrodes, the trOLEDs film and the textile trOLEDs. A spectroscopic radiometer (CS2000, Konica Minolta Inc.) equipped with a close-up lens (CS-A35) and a source meter (2400 series, Keithley Inc.) were used to measure the opto-electrical performances of the textile trOLEDs. For the textile trOLEDs, the opto-electrical performances were analyzed from top side only because, for the bottom side of the textile trOLEDs, the sum of the absorbance and reflectance of textile substrate is much larger than the corresponding transmittance, making it difficult properly to observe and analyze the luminous effect at the bottom sides of textile trOLEDs. For the glass-based trOLEDs, the opto-electrical performances were analyzed on both the top and bottom sides.
A bending test machine (Sciencetown Inc.) was used to analyze the mechanical performance of the textile trOLEDs. A custom-made cantilever test machine was used to measure the cantilever length of the aforementioned components (i.e., the textile substrate, textile trOLEDs) based on American Society for Testing and Materials standards (ASTM D: 1388).

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.