The ECD structure and the constituent components of the SPEC gels are illustrated in Fig. 1a-1c. In the stretchable ECD, the photo-patterned SPEC gel, integrated with SRSL and PDL, is sandwiched between two stretchable networked electrodes. Although the bottom electrodes and SPEC layer establish a highly conformal and reliable contact, mainly due to the bottom-up fabrication process where the SPEC is spun over the bottom electrodes, the top electrodes merely contact the SPEC layer, resulting in poor interfacial contact and weak adhesion. The SRSL and PDL are strategically introduced to ensure both stable electrical properties and mechanical durability of the ECD under extreme operating conditions. As shown in Fig. 1b, the SRSL provides a reliable and uniform electrical contact by eliminating the air gaps between the top electrodes and EC layer, while the PDL, which physically (electrically) separates the pixels, enhances mechanical integrity and prevents electrical crosstalk. The SPEC gel formulation comprises an acrylate viologen, PUA, diphenyl(2,4,6-trimethylbenzoyl)phosphine oxide (TPO), and 1,1'-dimethyl ferrocene (DMFc), combined with acetone, propylene carbonate (PC), and ionic liquids ([EMIM][TFSI] or [BMIM][TFSI]) (Supplementary Fig. 1). The RGB acrylate viologens displayed sufficient solubility in both ionic liquids, highlighting their suitability as media for SPEC gels. The use of ionic liquids was carefully guided through EC performance evaluations (Detailed explanation is provided in Supplementary Table 1–4 and Supplementary Note 1–4). PUA and TPO served as a UV-curable polymer matrix and photo-initiators, respectively. DMFc served as an anodic species, effectively suppressing undesired reactions between solvents and electrolytes, thereby reducing the operating voltage of ECDs by facilitating electron transfer during electrochemical reactions.

Red, green, and blue-color SPEC gels were prepared using different acrylate viologen derivatives. The SPEC gel containing red viologen was designated as SPEC-R, and those with green and blue viologens were labelled SPEC-G and SPEC-B, respectively. As shown in Fig. 1d-1e, electrochemical characterization revealed high ionic conductivities of approximately 3.6, 5.05, and 7.2 mS/cm for the SPEC-R, SPEC-G, and SPEC-B, respectively. Also, the Nyquist plots confirm stable charge-transfer kinetics, consistent with their excellent electrochemical performance. Following complete photo-curing, the SPEC gels exhibited high stretchability with maximum strain tolerances of 154% (SPEC-R), 169.3% (SPEC-G), and 120.3% (SPEC-B), with corresponding elastic moduli of 2.54 MPa, 2.190 MPa, and 4.12 MPa, respectively (Fig. 1f). The strain-dependent behaviours (Fig. 1g), highlight uniform deformation across all three layers in both relaxed and stretched states, confirming their robust resilience under dynamic conditions.

More importantly, the SPEC layers can be directly patterned onto PUA substrates, as illustrated in Fig. 1h. This compatibility arises from their shared chemical composition, with both materials featuring methyl acrylate terminal groups. Upon UV irradiation, polymerization is initiated, and polymer chain mobility gradually decreases. Residual unsaturated groups facilitate the formation of strong covalent bonds between the two layers, leading to strong interfacial adhesion. To enhance UV patternability and improve chemical compatibility with the PUA-based substrates, acrylate-functionalized viologens were employed. This chemical modification enables rapid, UV-triggered radical polymerization, as confirmed by Fourier transform infrared (FTIR) spectroscopy (Supplementary Fig. 2a-2c). The resulting covalent interactions, mediated by polymerized acrylate groups, not only strengthen interfacial bonding but also enable the formation of sub-micrometre resolution patterns. This is achieved by selectively dissolving the unexposed EC gel in dimethyl sulfoxide (DMSO) during the development process. A more mediated fine-patterning process was carried out using an optimized EC gel thickness of 60 µm, which was determined based on a balance between ionic conductivity and optical modulation, as displayed in Supplementary Table 4–6 and Supplementary Fig. 3–5.

In the device, the SPEC-R/G/B layers are deposited onto poly(3,4-ethylenedioxythiophene) : poly(styrene sulfonate) (PEDOT:PSS)/AgNWs electrodes, where the AgNWs are partially embedded within the PUA matrix. Notably, the PEDOT:PSS network does not completely encapsulate the PUA/AgNWs interface, leaving portions of the methyl acrylate-terminated PUA surface exposed (Supplementary Fig. 6). This residual chemical reactivity enhances intrinsic interfacial cohesion, thereby reinforcing the structural integrity of the layered assembly. To elucidate the nature of the interlayer bonding, FTIR spectroscopy was employed to analyze the chemical interactions between SPEC and PUA, with a focus on the acrylate functional groups (Supplementary Fig. 6a-c). For the SPEC/PUA sample, the PUA was partially cured to retain acrylate double bonds before deposition and complete curing of the SPEC layer. This sequential curing allowed additional covalent bonding between SPEC and the residual acrylate groups in PUA. FTIR spectra of individual SPEC and PUA layers, compared to the combined SPEC/PUA layer, revealed distinct spectral changes—most notably a peak at 1639 cm^− 1—indicative of new covalent bonds formed at the interface. This interaction, driven by the shared methyl acrylate, substantially improves the cohesion between the layers. To quantitatively evaluate the adhesive strength of the SPEC layers, peel tests were conducted using various substrate configurations (Supplementary Fig. 6d). Polydimethylsiloxane (PDMS)- and polystyrene-block-poly(ethylene-ran-butylene)-block-polystyrene (SEBS)-based substrates exhibited immediate delamination, whereas thermoplastic polyurethane (TPU) substrates showed partial delamination and possible interfacial recombination via hydrogen bonding. PEDOT:PSS on PUA substrate showed higher adhesion, while PUA substrate exhibited a stable force throughout the displacement range, underscoring the exceptional interfacial bonding between SPEC and PUA. Furthermore, FTIR spectroscopy examined whether methyl acrylate groups facilitate covalent bonding between the electrode and SPEC in the presence of PEDOT:PSS and AgNWs. (Supplementary Fig. 6e-g). The SPEC layer was cured entirely on both AgNWs/PUA and PEDOT:PSS/AgNWs/PUA sides, evidenced by reduced at 1639 cm^− 1 FTIR spectra. This result confirms the successful formation of covalent bonds, as the AgNW network, which was well embedded in the PUA, did not interfere with the bonding process. Consequently, the chain-like morphology and thus incomplete encapsulation of PEDOT:PSS is not likely to prevent direct covalent linkage between SPECs and PUA. These results collectively demonstrate that acrylate crosslinking proceeds effectively across the interfaces, enabling the formation of a continuous polymeric network even in the presence of electrodes.

To evaluate the electrochemical properties of SPECs, cyclic voltammetry (CV) was performed using a sandwiched structure with indium tin oxide (ITO) glass electrodes (Fig. 2a-2c and Supplementary Note 5). The anodic/ cathodic peak currents showed a linear relationship with the square root of the scan rate, indicating a diffusion-controlled mechanism³⁷. Incorporating ionic liquids improved ionic conductivity, enhancing the redox characteristics and enabling reversible color switching of SPEC-R, SPEC-G, and SPEC-B to reveal red, green, and blue colored states (Supplementary Fig. 7). In spectroelectrochemical analyses, SPEC-R initiated coloration at -1.3 V, exhibiting a prominent absorption peak at 515 nm (Fig. 2d and Supplementary Fig. 3j). The transmittance contrast (ΔT) between bleached and colored states was 85.2% at 470 nm and reached 90.2% at 515 nm. SPEC-G showed coloration starting at -1.4 V, characterized by strong absorption peaks at 620 nm (visible region) and 851 nm (near-infrared (NIR) region) (Fig. 2e and Supplementary Fig. 4j). The corresponding ΔT values were 69% and 82.2%, respectively, showing NIR absorption capability due to the extended absorption range of the four pyridinium groups³⁸. SPEC-B demonstrated prominent absorption at 539 nm and 602 nm upon applying − 1.3 V, achieving a ΔT of 83% at 602 nm (Fig. 2f and Supplementary Fig. 5j). The color changes of these SPECs were quantitatively analyzed through color coordinates (Fig. 2g-2h). In the bleached state, the xy coordinates of SPEC-R, SPEC-G, and SPEC-B were (0.33, 0.35), (0.31, 0.34), and (0.30, 0.36), respectively, corresponding to pale yellow. Upon electrochemical activation via applied voltage, these coordinates shifted to (0.46, 0.34), (0.29, 0.40), and (0.31, 0.27), clearly representing red, green, and blue coloration, respectively.

Alternating voltages of -1.5 V and 0.5 V were applied in 20-second intervals to evaluate the stability and coloration efficiency (CE) of the ECDs (Fig. 2i-2k). The CE, a key figure of merit, relates optical modulation to the charge involved in the redox process, and was calculated as

where is the exchanged charge density and , with ​ and denoting transmittance in bleached and colored states.

Initially, SPEC-R showed a ΔT of 83.1%, slightly decreasing to 78.9% after 5 cycles. Within the first 30 cycles, T_min increased from 1.2% to 5.4%, stabilizing at 3% after 300 cycles (12,000 s), while maintaining a ΔT of 75.2%. The CE for SPEC-R was 52.0 cm²/C, with estimated coloration (τ_c) and bleaching (τ_b) switching times of 17 and 11 s, respectively. Moreover, SPEC-G exhibited a ΔT of 69.2%, which reduced to 54.7% after 300 cycles, representing a CE of 289.3 cm²/C and switching times of 6 s (τ_c) and 4 s (τ_b). SPEC-B maintained a ΔT of 82.4% over 300 cycles with a CE of 253.5 cm²/C, and the switching times (τ_c and τ_b) of 18 and 9 s.

The equivalent circuit of the stretchable ECD is illustrated in Fig. 4c, where R_e represents the resistance of the stretchable electrodes, R_ct denotes the charge transfer resistance, Z_w corresponds to the Warburg impedance, and C_dl signifies the double-layer capacitance. Under direct current bias, C_dl behaves as an open circuit, allowing the equivalent resistance (R_eq) to be expressed as R_eq = 2(R_e + R_ct + Z_w). Notably, all of these resistive components are highly strain-sensitive. In stretchable electrodes, resistance typically increases in proportion to the square of the applied strain. Similarly, our AgNW-based composite electrodes exhibit a severe increase in resistance with strain, primarily due to the percolation-dependent nature of randomly distributed AgNW networks (Fig. 4c). This increased resistance directly impacts device operation by reducing the voltage drop across the SPEC layer, thereby limiting the effective bias available for electrochromic switching. Such phenomena typically introduce a critical hurdle to implementing strain-resilient, high-performance stretchable ECDs. However, the device architectures we proposed in this research can address these challenges throughout optimized electrode design and interface engineering, such as PEDOT:PSS integrated AgNWs and SRSL and PDL layers, which mitigate resistance growth under strain.

EC performance under mechanical deformation was evaluated, as shown in Fig. 4d-4f. The transmittance spectra and switching behaviours of red, green, and blue ECDs were measured under strains of up to 50%. In the bleached state, a slight increase in transmittance was observed with increasing strain, attributed to the expanded spacing within the AgNW network, which allows more light to pass through. In the colored state, a comparable increase in transmittance was observed, attributed to the strain-induced increase in resistance that reduces the effective coloration bias. Despite these optical variations, the devices maintained strong EC functionality. Even at 50% strain, the transmittance differences between the bleached (T₀) and colored (T_color) states were 57.77%, 41.53%, and 49.57% for red, green, and blue ECDs, respectively—values that remain comparable to their pristine (0% strain) conditions of 42.35%, 43.54%, 46.85% (Fig. 4g-4i). To further validate mechanical robustness, we conducted fatigue tests involving 1500 stretch–relaxation cycles between 0% and 30% strain. The corresponding T₀ - T_color values over these cycles are shown in Fig. 4j-4l. A slight decline in optical contrast was observed within the first 500 cycles, after which performance remained relatively stable. After 1500 cycles, the T₀ - T_color values were retained at 40.36%, 51.40%, and 44.60% for red, green, and blue devices, respectively. These results demonstrate that the current ECDs not only survive but maintain functional integrity under sustained and challenging mechanical strain, highlighting the success of material and structural design strategies.

To realize highly pixelated stretchable ECDs, three SPEC gel arrays were photo-patterned onto a stretchable electrode substrate. Sub-pixelated ECDs were fabricated by sequentially patterning the SPEC gels through a drop-casting and UV-curing process, as illustrated in Fig. 5a. A critical consideration during the SPEC gel patterning process is avoiding exposure of the EC solution to oxygen, which inhibits the photopolymerization of acrylate-based resins^44,45. Supplementary Figs. 13 and 14 show fabrication details and the PDL with SPEC pixels. The incorporation of PUA in both the SPEC gels and the electrode substrate also significantly enhanced the adhesion of the cured SPEC gels, contributing to robust structural integrity under deformation. The proposed device architecture implies three essential advantages: 1. Enhanced interfacial bonding. The SRSL ensures secure bonding between the top electrode and underlying layers. During UV curing, polymer chains form covalent bonds with adjacent layers, effectively integrating all device components into a cohesive assembly (Fig. 5b). 2. Enhanced electrical connectivity. The SRSL fills air gaps between the multi-layered SPEC pixels and the top electrode, enabling conformal electrical contact that facilitates efficient ion transport. This design leads to enhanced EC switching behaviours. To suppress signal crosstalk arising from ion diffusion in the SRSL, the PDL provides pixel confinement (Fig. 5c). All functional materials—stretchable electrodes, SPEC layers, SRSL, and PDL—are physically confined within defined pixel areas, eliminating undesired lateral diffusion. The mechanism is schematically illustrated in Fig. 5d. Notably, as shown in Fig. 5e, ECDs fabricated without the SRSL showed low and diminished current densities, indicating ineffective electrochemical contact and switching behaviours. In contrast, SRSL-integrated devices exhibited high and stable current densities, confirming reversible electrochemical reactions facilitated by robust electrical pathways. 3. Mechanical stress release. The SRSL and PDL also serve to redistribute mechanical stress across the device surface. Horizontal and vertical air gaps between SPEC sub-pixels and SPEC and electrodes, respectively, can be considered as zero modulus areas, resulting in significant modulus mismatch and localized stress concentration at pixel boundaries and electrode interfaces. This mismatch is a common cause of cracking or delamination under strain. Therefore, devices without the SRSL fractured immediately upon strain, whereas those with SRSL maintained electrical and mechanical stability up to 50% strain (Supplementary Fig. 12d-i). Finite element simulations (COMSOL) confirmed that placing the SRSL and PDL in these regions effectively redistributes stress, releasing the imposed stresses and thus preventing localized failure (Fig. 5f). This stress-balancing strategy allows the device to maintain structural cohesion and stable coloration performance even under substantial mechanical deformation. To ensure the scalability and practicality of the developed SPECs and device architectures, we implemented large-area pixelated passive matrix (PM) stretchable ECDs with 10 × 10 RGB pixels, verifying the efficacy of the materials and devices for high-quality stretchable applications. These free-standing, multicolor pixel arrays demonstrated excellent mechanical durability and reliable coloration for 1,500 mechanical strain cycles at strains up to 30%. (Fig. 5g-h and Supplementary Fig. 15). Notably, no degradation or crosstalk was observed between adjacent pixels under stretched states. All RGB colored characters and images were implemented, and full-pixel activation of blue arrays was successfully demonstrated.

In this article, we presented a general and scalable strategy for realizing highly pixelated intrinsically stretchable full-color ECDs that outperform the state-of-the-art stretchable counterparts. To address the above requirements, we proposed a novel material design and facile implementation methods based on direct photo-patternable acrylate-substituted viologens and PUA mediated stretchable electrodes. This approach is further integrated with ion gel-based stretchable components such as SRSLs and PDLs to achieve system-level robustness and scalability, validating significantly improved interfacial adhesion and overall device performance via CMOS-compatible processes. These intrinsically stretchable, high-resolution ECDs could be of use for a wide range of applications in large-area skin-worn displays, healthcare monitoring and professional outdoor platforms, offering a solid foundation for full autonomy of next-generation intrinsically stretchable electronic systems.

The details of all synthetic procedures Synthesis of 4,4’-(1,4-phenylene)bis(1-(2-(acryloyloxy)ethyl)pyridin-1-ium) (AC-R), Synthesis of Synthesis of 4-(2,4,5-tri(pyridine-4-yl)phenyl)-bis(1-(2-(acryloyloxy)ethyl)pyridine (AC-B), Synthesis of 1,1'-bis(2-(acryloyloxy)ethyl)-[4,4'-bipyridine]-1,1'-diium (AC-G) are available in Supplementary Note I, Supplementary Figs. 1–3, Supplementary Table 4.

A flask charged with polyol (25 g, 100 mmol) and Methyl ethyl ketone (MEK) (25 mL) was heated at 60 ^oC until the solution turned clear, after which Isophorone Diisocyanate (IPDI) (8.33 g, 150.0 mmol) and dibutyltin dilaurate (cat.) were added. This mixture was then heated at 60 ^oC for 3 hours, and after adding HEA (2.9 g, 99.9 mmol), it was stirred for an additional 3 hours at 60 ^oC before cooling to room temperature. It was then mixed with an Irgacure 184 at a ratio of 10:0.36.

The polyurethane acrylate (6 g), [BMIM][TFSI] (1.5 g), [Li][TFSI] (1.5 g), ferrocene (Fc) (20 mg), TPO (216 mg), and 2-propanone (1.5 g) were mixed and stirred for 24 hrs at 80 ^oC.

A dispersion of AgNWs in water (concentration of 1 mg ml^− 1) was coated on bare wafer substrates using a #5 Meyer rod until sheet resistance was under 30 Ω/sq and annealed at 120 ^oC for 1 min. PEDOT:PSS ink was prepared for the electrode solution using PEDOT:PSS (1.3 wt% dispersion in water, Clevios pH 1000) (10 g). Then, DMSO (2.5 g) and glycerol (1 g) were added and stirred at room temperature for 2 hrs. After that, ethylene glycol (0.5 mL) and DBSA (0.05 mL) were added and stirred for 15 min. The prepared PEDOT:PSS solution was spin-coated on the AgNWs layer using a consecutive two-step coating process with different spin speeds (500 rpm, 10 s, and 1500 rpm, 20 s) and subsequently annealed at 120°C. The obtained transparent conductive coating on the wafer substrates was then coated with a PUA solution. The PEDOT:PSS and AgNW layers were patterned by conventional photolithography and wet etching with Cr etchant (ETCR-300). The PUA solution was spin-coated at 300 rpm for 60 s. The coated AgNWs wafer was placed in a convection oven at 80 ^oC for 1 hr to evaporate residual solvent and level the plane, then cured with a UV lamp. After that, the patterned electrodes were heated at 80°C for 1 hr and then immersed in water for 1 hr to peel off.

Polystyrene (Sigma Aldrich average M_w~350,000) (25 g) and toluene (25 mL) were stirred for 12 hrs at 80 ^oC. The polystyrene solution was spin-coated on bare glass at 1000 rpm for 60 s, and then heated at 100°C for 10 min. Unpatterned PEDOT:PSS-AgNW-PUA composite electrodes were transferred to polystyrene-coated carrier glass. The EC solutions were drop-cast onto the composite electrodes and cured under UV irradiation (350 nm, 145 W) through a PDMS-coated glass, resulting in a final layer thickness of 60 µm. After curing, the PDMS glass was removed. The SRSL solution was drop-cast onto the SPEC gel/electrode surface. The second electrode was then aligned in reverse and placed on top of the SRSL-coated SPEC gel. UV light was applied through the second electrode to cure the SRSL solution in place. After curing, the assembled device was carefully removed from the carrier glass.