Electrically tunable infrared optics enabled by flexible ion-permeable conducting polymer-cellulose paper

npj | flexibleelectronics Article Published in partnership with Nanjing Tech University https://doi.org/10.1038/s41528-024-00339-7 Electrically tunable infrared optics enabled by flexible ion-permeable conducting polymer-cellulose paper Check for updates Chaoyang Kuang1, Shangzhi Chen 1,MingnaLiao 1,2, Aiman Rahmanudin 1,2, Debashree Banerjee1,2, Jesper Edberg 3, Klas Tybrandt 1,2,DanZhao 1& Magnus P. Jonsson 1,2 Materials that provide dynamically tunable infrared (IR) response are important for many applications, including active camouflage and thermal management. However, current IR-tunable systems often exhibit limitations in mechanical properties or practicality of their tuning modalities, or require complex and costly fabrication methods. An additional challenge relates to providing compatibility between different spectral channels, such as allowing an object to be reversibly concealed in the IR without making it appear in the visible range. Here, we demonstrate that conducting polymer-cellulose papers, fabricated through a simple and cheap approach, can overcome such challenges. The papers exhibit IR properties that can be electrochemically tuned with large modulation (absolute emissivity modulation of 0.4) while maintaining largely constant response in the visible range. Owing to high ionic and electrical conductivity, the tuning of the top surface can be performed electrochemically from the other side of the paper even at tens of micrometer thicknesses, removing the need for overlaying electrode and electrolyte in the optical beam path. These features enabled a series of electrically tunable IR devices, where we focus on demonstrating dynamic radiative coolers, thermal camouflage, anti-counterfeiting tags, and grayscale IR displays. The conducting polymer-cellulose papers are sustainable, cheap, flexible and mechanically robust, providing a versatile materials platform for active and adaptive IR optoelectronic devices. Mid-infrared (mid-IR) light, with wavelengths between 3 and 30 μm, represents a spectral range of high relevance for both civilian and military applications1,2. Mid-IR light carries vital information spanning from fingerprints of molecules3and features of localized charge carriers4,5,tothermal profiles of physical objects6, enabling multiple important applications from biosensors to thermal camouflage and radiative cooling. The latter utilizes spontaneous thermal emission to passively cool objects on Earth by radiatively transferring thermal energy to cold outer space7–9. Currently, the majority of IR devices are passive and therefore only provide pre-designed properties and functions10–13. This restricts their use from many applications, such as dynamic camouflage that can adapt to changing environments2,14,15, or tunable radiative coolers for smart thermal management16–19. Besides realizing IR tunability through functional device integration (e.g., thermoelectric devices15), most approaches utilize dynamically tunable IR materials, such as phase change materials20–23 (e.g., Ge 2 Sb 2 Te 5 ). Such materials provide modulation through thermal stimulation, including via light-induced heating24,25 or Joule heating20,21,26.Forthe2DMXeneTi 3 C 2 T x it was possible to tune the IR response in a large range via reversible water uptake through variations in the humidity27, however, with challenges related to tuning speed and large-scale homogeneity28. Another interesting approach is to use mechanically tunable materials, including systems inspired by cephalopods29,30 or mechanically stretchable conductors31.Also for these systems, challenges relate to low switching speeds, and also in difficulty in gradually varying the IR response with high accuracy. In that sense, micro-electromechanical systems (MEMS) form a related category that can provide fast switching speed and convenient control via electrical signals, but they rely on sophisticated and expensive microfabrication techniques32. An alternative approach is to use materials whose properties can be electrically tuned via reversible electrochemical doping, often also providing low-power consumption due to low operating voltages and bistability or hysteresis effects. Examples of materials include electrochromic polymers33–40 and graphene41–43. Importantly, electrochemical 1Laboratory of Organic Electronics (LOE), Department of Science and Technology (ITN), Linköping University, Campus Norrköping, Norrköping, Sweden. 2Wallenberg Wood Science Center, Linköping University, Campus Norrköping, Norrköping, Sweden. 3RISE Research Institutes of Sweden, Digital Systems, Printed-, Bioand Organic Electronics, Norrköping, Sweden. e-mail: [email protected];[email protected] npj Flexible Electronics | (2024) 8:55 1 1234567890():,; 1234567890():,; doping and dedoping requires the tunablematerialtobeincontactwithan electrolyte, which typically is opaque in the IR. To not supress or completely cancel the IR modulation of the device, the electrolyte therefore needs to be positioned below the tunable material (not in the beam path). However, most redox-tunable materials also require a supporting electrode to facilitate switching due to modest electrical conductivity. In turn, even thin electrodes are typically opaque in the IR range due to high density of free charge carriers, and would therefore also strongly limit the IR modulation if placed in the beam path44. Recent progress in the development of far-IR transparent conductors is highly interesting in this respect44, but so far the common approachhasbeentoinsteadpositionstructured electrodes (e.g., microporous electrode) that are permeable to ions under the tunable material. This ensures contact of the tunable material with both the electrode and the electrolyte, but often requires complex and costly fabrication procedures45–47. Another alternative is to develop IR-tunable materials that are sufficiently electrically conducting to be used as their own electrodes, such as multi-layer graphene41,42 or highly conducting polymer thin films based on PEDOT (poly[3,4-ethylenedioxythiophene])33,34,48. This removes the need for a supporting electrode in contact with the tunable material. This approach has been successful, but the modulation of IR properties of such thin film devices often also led to large changes in their visible response33,41,47. This is less desired for applications such as dynamic camouflage and anticounterfeiting. For example, it would be problematic if a concealed object in the IR range becomes apparent in the visible upon tuning. In this study, we demonstrate that a flexible and mechanically robust paper material made by blending cellulose with a conducting polymer can be used to overcome the above challenges. These papers possess high electrical conductivity (above 500 S/cm) and also provide vertical transport of ions. This allows for electrochemical tuning of the PEDOT redox states across the whole paper, for thicknesses as large as 150 μmandwithoutany additional top electrode. In turn, modest electrical biases within ±1.5 V could tune the IR emissivity of paper-based devices in a range of about 0.4, which is higher than for previously reported systems based on thin PEDOT films33,34. Furthermore, the visible response of the devices remained largely constant during tuning, allowing for dynamic tuning of thermal radiation and apparent temperatures without changing or revealing information in the visible range. As depicted in Fig. 1, we demonstrate that these papers enable a range of applications, focusing on tunable radiative coolers, dynamic thermal camouflage, dynamic anti-counterfeiting tags, and grayscale UV-patterned IR displays. Our demonstrations include the integration of the papers with various functional films to achieve more complex and unique functions. The papers can be fabricated in a very simple way by blending and drying and can be easily scaled up to arbitrary sizes or thicknesses. In terms of large-scale use in commercial settings, we also Voltage controller CPC paper 2 CPC paper 1 Anti-counterfeiting tag Thermal camoulfage Radiative cooler Infrared display Gel electrolyte Thermal radiation Min. voltage Intermediate voltage Max. voltage Infrared camera Fig. 1 | Concept illustration of conducting polymer-cellulose paper and their device applications for infrared optics. Operation of the devices utilizes the mixed ionic-electronic transport properties of the conducting polymer paper, which enable efficient vertical transport of ions across the paper. Applying different electrical biases modulates the infrared properties of the papers, such as emissivity and surface reflection. In combination with additional functional layers, this opens for various applications ranging from radiative coolers and thermal camouflages to dynamic displays and anti-counterfeiting devices. For radiative coolers, varying the voltage could change the emissivity of the top paper and thus control the radiative cooling power and temperature of the overall device. The changes in emissivity also lead to changes in the apparent temperature captured by an infrared camera, enabling applications such as dynamic thermal camouflage and infrared displays. Anticounterfeiting applications could further show visible patterns (e.g., letter “Y”) using an infrared-transparent dye and contain concealed information (e.g., letter “N”) that can be retrieved and observed in the infrared range by tuning the voltage. https://doi.org/10.1038/s41528-024-00339-7 Article npj Flexible Electronics | (2024) 8:55 2 emphasize that biobased cellulose is non-toxic, cheap, and more sustainable compared to commonly used synthetic fluorinated binders. We believe our conducting polymer-cellulose papers can find important use for various electrically tunable IR optical devices and applications. Results Conducting polymer-cellulose papers and their basic properties We fabricated conducting polymer-cellulose (CPC) papers using poly(3,4ethylenedioxythiophene):polystyrene sulfonate (PEDOT:PSS) and cellulose nanofibrils (CNF) as the main ingredients. The plasticizer glycerol and secondary dopant ethylene glycol (EG) were also added to improve the ionic and electrical conductivities of the paper as well as its mechanical properties49 (see chemical structures in Supplementary Fig. 1 and more details on the fabrication process in “Methods”section). We prepared papers with different thicknesses (in the range from tens to hundreds of micrometers) by varying the volume of the blend solution. The intended IR applications require the papers to not only be electrically conducting butalso ion permeable to allow forelectrochemical tuning across the whole thickness of the paper, including the top interface that is not in direct contact with the electrolyte layer. To examine this, we used the basic paper device structure illustrated in Fig. 1, consisting of an electrolyte layer sandwiched between a top and a bottom CPC paper. The electrolyte was an ionic gel made of an ionic liquid [1-ethyl-3-methylimidazolium bis(trifluoromethylsulfoniyl), EMIM:TFSI] in the matrix of poly(vinylidene fluoride-co-hexafluoropropylene) (PVDF-HFP, see chemical structures in Supplementary Fig. 1). This solid-state gel layer can be used as a “cut-and-stick”electrolyte for various electrochemical devices due to its excellent ionic transport properties and good mechanical properties and stability in air50,51. Figure. 2a, b presents the specular reflection at different electrical biases for a device with a 47 µm thick top CPC paper. At 0 V bias, the device exhibited specular reflection of only ~5% between 400 and 1500 nm, while it increased to around 50% for longer wavelengths in the mid-IR. Applying a positive bias to the top paper did not change the overall spectral line shape but slightly increased the reflection in the near IR. These results agree with the PEDOT:PSS in the paper being in a high-conducting oxidized state after fabrication49, which leads to optically metallic properties and high reflectance at frequencies below its plasma frequency35. The results indicate that the paper is sufficiently conducting and smooth to provide high specular IR reflectance in a broad range from around 2500 nm to at least 17 µm. Applying a negative bias drastically decreased this broad IR reflectance. This is due to redox switching of the PEDOT to its neutrallow-conducting state, for which the material is no longer optically metallic or reflective in the IR. Indeed, the dependence on bias potential of the IR surface reflection correlates with the non-monotonic variation in electrical conductivity (Fig. 2c) and the results also agree with the complex refractive index dispersion of PEDOT thin films in highand low-conducting states (Supplementary Fig. 2). The de-doping process upon application of negative bias is governed by the following scheme: PEDOTþ:PSSþMþþe!PEDOT0þMþ:PSSð1Þ Tensile strain (%) Norm. resistance 4 1.0 1.4 1.8 0.6 812 16 20 24 Sample 1 Sample 2 Sample 3 Time (s) Wavelength (nm) 0 1000 400 800 1200 1600 1500 2000 2500 40 30 20 10 0 Surface reflection (%) d Wavelength (nm) Surface reflection (%) 500 80 40 01500 2500 +1.5 V +1.0 V +0.5 V 0 V -0.5 V -1.0 V -1.5 V a Wavelength (μm) Surface reflection (%) 4 80 40 0 +1.5 V +1.0 V +0.5 V 0 V -0.5 V -1.0 V -1.5 V 81612 b Electrical bias (V) Electrical conductivity (S/cm) -1.5 100 300 500 -0.5 0.5 1.5 c g Number of bending cycles Norm. resistance 0 0.8 1.6 2.0 20 40 60 80 100 83th cycle Bending test 1.2 h Voltage (V) Current (mA) 0 -10 0 -20 10 20 -1 1 Scan rate: 50 mV/s 1st cycle 10th cycle 44th cycle 100th cycle i Paper thickness (μm) Surface reflection response time (s) 23 40 20 47 68 92 150 Reduction time Oxidation time 80 60 0 10 5 20 15 0 f λ = 3000 nm Time (s) Normalized reflection 300 100 0600 900 1200 1500 +1.5 V -1.5 V e Time (s) 960 01000 100 90% 10% Reduction 17.3 s Time (s) 660 700 90% 10% Oxidation 5.5 s Fig. 2 | Optical, electrical, electrochemical, and mechanical properties of CPC papers. a Specular reflectance in the visible and near IR for a 47 µm paper device at different electrical bias. bSpecular reflectance in the mid-IR for the 47 µm paper device at different electrical bias. cBulk electrical conductivity of a 47 µm paper at different electrical bias. dIn-situ electrochemical specular reflectance spectra with the application of electrical bias in the form of periodic square waves (±1.5 V and 180 s duration for each step). eNormalized reflectance at the wavelength of 3000 nm for a periodic square wave electrical bias (±1.5 V and 300 s duration for each step) with a few cycles. The bottom panel also indicates the method to extract oxidation and reduction times. fResponse times for devices made from CPC papers with different thicknesses. gNormalized resistance-strain curves measured at a stretching speed of 0.5 mm/s for a CPC paper with thickness of 47 µm, length of 10 mm, and width of 8 mm. hNormalized resistance-bending cycle curves of a CPC paper (same dimensions as in (g)) with a bending radius of 3 mm and a bending cycle rate of 0.5 mm/s. iCyclic voltammograms of a complete device, measured after the 1st, 10th, 44th, and 100th bending cycle. The CPC papers in the device had a thickness of 47 µm, length of 10 mm, and width of 30 mm. The bending radius was 10 mm and the bending cycle rate was 2 mm/s. https://doi.org/10.1038/s41528-024-00339-7 Article npj Flexible Electronics | (2024) 8:55 3 where M+is the cation provided by the electrolyte and e−is an electron. The high-conducting state of the polymer can be retrieved back by re-doping via oxidation using a positive electrical bias: PEDOT0þMþ:PSS!PEDOTþ:PSSþMþþeð2Þ Importantly, the results confirm that switching occurs also at the outer interface of the device where reflection takes place (the optical penetration depth is far smaller than the paper thicknesses, as estimated for PEDOT:PSS in Supplementary Fig. 2d). Hence, we can conclude that the paper provides sufficient vertical ionic and electric mobility to enable switching across its whole thickness. Measurements for paper devices made using CPC papers with other thicknesses show successful switching up to at least 150 µm thicknesses (Supplementary Fig. 3). In the visible and near IR range (below 1500 nm), the tuning led only to slight changes with reflectance remaining below 5% for both states (Fig. 2a). This makes the material suitable for applications such as dynamic IR camouflage and anticounterfeiting with minimal changes in the visible52.Figure2c and Supplementary Fig. 4 also confirm effective tuning of the electrical conductivity of the paper. The bulk conductivity of a 47 µm CPC paper was about 30 S/cm at negative bias of −1.5 V and increased to 560 S/cm at +1.0 V. It slightly decreased (to 523 S/ cm)atayethigherbiasof+1.5 V, which might be related to mild degradation from over-oxidation53,54 or anti-ambipolar behavior of the conjugated polymer55. To further understand the vertical ionic transport dynamics of the CPC paper, we carried out time-resolved in-situ spectroelectrochemical characterization where the electrical bias was applied in the form of periodic square waves (i.e., +1.5 V for 180 s and −1.5Vforanother180s).Figure2d presents the results for a device with 47 µm thick CPC top and bottom papers, showing simultaneous changes in the spectra upon every change in bias potential. The specular reflectance values agree well with the static measurements in Fig. 2a. We extracted the switching times for oxidation (from neutral state to oxidized state) and reduction (from oxidized state to neutral state) as the time to change the reflectance between 10% and 90% of the initial and final values (bottom panel in Fig. 2e). The oxidation time for a 47 µm CPC paper device was about 5.5 s, while the reduction time was 17 s. Figure 2f and Supplementary Fig. 5 present response times for devices with CPC papers of different thicknesses. Both oxidation time and reduction time increased with increasing paper thickness, following an approximate linear relationship (Fig. 2f). The switching times are remarkably short considering the large thicknesses of the papers, indicating high vertical ionic mobility. In fact, the switching times are not far from that of electrically tunable vertical devices based on thin films (with thickness of only a few hundred nanometers, such as multi-layer graphene41 and PEDOT:Tosylate films33), which exhibited switching times on the order of a few seconds (e.g., 4.6 and 3.3 s33,41). The results for our devices based on much thicker CPC papers (e.g., thickness of 47 µm) corresponds to about two orders of magnitude faster thickness-normalized switching. Recent studies on high-performance kirigami-type Au/polyaniline electrodes45 exhibited similar switching times (~16 s) as for our devices, but for lower thicknesses (<1 µm). It should also be stressed that the presented CPC paper devices possessed good optical memory function, in particular for the oxidized high-conducting state for which the normalized reflectance remained within 93% of the original value formorethan4hafterremovingtheappliedbias(SupplementaryFig.6).If turning off the device in the low-conducting state, the normalized reflectance increased within around 20 min to around the middle point between the lowest and highest values (50% normalized reflectance). This means that the stable reflectance was distinctly different if leaving the device unbiased for long times from the high or the low conductivity states. The corresponding optical memory effect does not only make our papers suitable for low-power operation, but also open for other applications such as neuromorphics56,57. We further note that these type of devices showed good durability upon repeated electrical modulation. The tuning range of the reflectance decreased <20% after more than 100 cycles between −1.5 and +1.5 V (Supplementary Fig. 7). Most applications rely on robust mechanical and electromechanical properties, and many benefit also from stretchability or bendability. Using a custom-built four-point probe tensile strain setup, we could study the behavior upon stretching via the variation in electrical resistance across different CPC papers (see “Methods”section). As shown in Fig. 2g, the papers showed minor increases in their normalized resistance up to 22% strain. By comparison, previously reported CPC papers49 had an elongation at break of 13.4%, indicating that the addition of EG as a plasticizer enhanced the stretchability of the papers. To further test the mechanical properties of the papers, bending tests were performed at a bending radius of 3 mm for 100 cycles (see photographs in Supplementary Fig. 8). The normalized resistance fluctuated marginally during the first 83 cycles until partial mechanical failure at the 84th cycle (Fig. 2h). Encouraged by the mechanical robustness of the CPC papers, we tested the bending properties of a whole device as a function of its electrochemical performance (see Supplementary Fig. 9). Tape was applied around the edges of the device to improve adhesion between the layers (see details in “Methods”section). Cyclic voltammograms (CV) were measured in the flat state of the device after the 1st, 10th, 44th, and 100th bending cycle. The device showed good electrochemical stability with only a minor 4% variation in capacitance between the 1st (17.7 mF) and 100th (16.9 mF) bending cycle. Discussion Electrical modulation of emissivity for tunable radiative cooling The large variation in specular reflection (Fig. 2b) implies that the devices may also provide variations in their thermal emissivity, which was confirmed through measurements of the total reflectance (specular +diffuse, Rtotal) using an integrating sphere (Fig. 3a). Kirchhoff’slawofthermal radiation tells us that the thermal emissivity (ελðÞ) is equal to the absorption atthesamewavelength(AλðÞ).SincetheCPCpapershavenegligibleIR transmittance (Supplementary Fig. 10), the thermal emissivity of the devices is then obtained as ελðÞ¼1Rtotal λðÞ.Figure3aexemplifies the results for a paper device with 47 µm thick CPC papers, showing successful broadband electrical tuning of the thermal emissivity by more than 0.38. For example, the emissivity varied between 0.29 and 0.67 at the wavelength of 10 µm. For both redox states, these emissivity values correspond to considerably higher reflectance than the specular reflection (Fig. 2b), highlighting the importance to account for diffuse back scattering and not only specular reflection when determining the absorption and emissivity. Figure 3b presents a summary of electrical emissivity modulation for devices with different paper thicknesses (at the wavelength of 10 µm), all exhibiting the same trend of the emissivity decreasing with increasing bias voltage. We found no clear dependence on thickness on the absolute emissivity or tuning range, but instead attribute the observed variations to natural differences between devices. Indeed, devices with the same paper thickness could present variations in the exact emissivity tuning range (Supplementary Fig. 11), but still showing the same tuning trend with the electrical bias. Papers with thickness of 47 µm were chosen as model system for the rest of the study. The power of a radiative cooler is proportional to its thermal emissivity7,58, which has been utilized in different types of switchable or controllable radiative coolers45,48,59. We recently demonstrated temperature control of devices at ambient conditions via electrical redox tuning of a thin conducting polymer film, which provided emissivity modulation of around 0.2534. Motivated by the much larger modulation for our current system (around 0.4), we here studied the performance of CPC paper devices in terms of electrically tunable radiative cooling. Night-time outdoor cooling performance was first characterized for pristine CPC paper devices with three different thicknesses (23, 47, and 92 µm). All three devices showed clear sub-ambient cooling, with a temperature decrease of 1.0 to 1.5 °C compared to the surrounding temperature. Applying electrical biases did not result in clear changes in the absolute device temperatures. We stress in this respect that the absolute device temperature was monitored by thermocouples and not by a thermal camera, because a thermal camera also https://doi.org/10.1038/s41528-024-00339-7 Article npj Flexible Electronics | (2024) 8:55 4 measures the change in emissivity itself 41,59. Indeed, the electrically controlled variation in apparent temperature obtained by a thermal camera reached much larger values of above 12 °C for our devices (see sections below). Since outdoor measurements can be affected by factors such as wind or moving clouds, we also characterized our devices under more stable conditions in a sky simulator (see illustration in Fig. 3d). Our sky simulator was described in detail in our previous report60.Inbrief,itusesanIRabsorptive material cooled by liquid nitrogen to mimic the low emissive and absorptive IR behavior of the cold universe. Emitted infrared light from devices placed (upside down) at the top of the setup is reflected by sidewall mirrors and absorbed by the cold source (“the sky”). Although the setup cannot provide identical properties as the real sky34, it can provide a similar but stable environment for evaluating and comparing the radiative cooling performance of our devices. Figure 3e presents tunable cooling of a device with 23 µm thick CPC paper upon periodically switching the electrical bias between ±1.5 V. The results show clear and reversible tuning of the device temperature. At the oxidized state, the temperature was around 0.72 °C lower than the surroundings while the corresponding decrease in temperature from the surroundings was 1.04 °C for the reduced state. This gives a temperature tuning range of about 0.32 °C, which is higher than that reported for devices based on thin PEDOT films at ambient conditions34. This improvement is consistent with the superior emissivity modulation for the CPC paper devices. Indeed, the 47 µm CPC paper device (with yet higher emissivity modulation) showed an even larger tunable temperature modulation of 0.42 °C (Fig. 3f). Interestingly, the 47 µm paper device also showed transient upward temperature peaks for each switch in electrical bias, which is different from the device with thinner CPC paper and PEDOT thin film devices34. We attribute this behavior upon increasing CPC thickness to increased role of Joule heating in defining the temporal evolution of the device temperature. Joule heating is proportional to the square of the electrical current and the resistance, both of which increases with paper thickness. The bottom panel of Fig. 3fexemplifies this as a much higher current required to switch the 47 µm paper device compared with the 23 µm paper device, with maximal currents exceeding 50 mA/cm2.Thisisanatural consequence of the volumetric capacitance and more PEDOT material being switched for the thicker system. The influence of oxidation state on radiative cooling performance was also investigated by gradual chemical reduction of CPC papers by exposure to vapor from a reducing agent (branched poly(ethylenimine), or (PEI))61. Previous reports verified that this chemical tuning method can achieve d Wavelength (μm) Diffused reflectance 4 0.6 0.4 0.2 812 +1.5 V +1.0 V +0.5 V 0 V -0.5 V -1.0 V -1.5 V 0 0.8 0.6 0.4 Emissivity 1.0 16 0.8 0.2 a 23 μm 47 μm 68 μm 92 μm Paper thickness Electrical bias (V) Diffuse reflectance -1.5 0.6 0.4 0.2 -0.5 0.5 0 0.8 0.6 0.4 Emissivity 1.5 0.8 0.2 01.0 1.0 b λ = 10 μm Outdoor Real temperature (ºC) 4.0 6.0 8.0 23 μm 47 μm 92 μm Reference Outdoor Temperature difference (ºC) 0.5 Time (hour) 1.0 1.5 2.0 2.5 -2 -1 023 μm 47 μm 92 μm c Thermal insulation Liquid nitrogen Thermoflask containing black Al foil IR-reflective Al foil wall Thermal couple over the sample Thermal couple inside sky simulator 1.5 m height Temperature (ºC) 17.2 18.0 18.8 -1.5 V +1.5 V Reference 0Time (min) 30 60 90 120 23 μm paper 47 μm paper e Temperature (ºC) 15 16 18 -1.5 V+1.5 V Reference 17 19 f Current density (mA/cm2) 0Time (min) -50 0 30 60 90 120 50 Wavelength (nm) Surface reflection (%) 500 30 1500 01000 2000 2500 3000 20 10 Original PEI 20 min PEI 60 min PEI 40 min g Wavelength (μm) Diffuse reflectance 4 0.6 0.4 0.2 0 0.8 0.6 0.4 Emissivity 81216 0.8 1.0 0.2 Original PEI 20 min PEI 60 min PEI 40 min h Temperature difference (ºC) 20 Time (min) 40 60 -3 -1 0 -2 PEI 20 min PEI 60 min PEI 40 min i Real temperature (ºC) 60 Time (min) 120 180 -5 -1 -3 Original PEI 60 min Current density (mA/cm2) -25 0 25 Device Original Fig. 3 | Infrared tunable emissivity and application for dynamic radiative cooling. aDiffuse reflectance, and emissivity measurement of a CPC paper device at different electrical bias (47 μm thick CPC paper). bRelation between the emissivity and electrical bias for paper devices with CPC papers of different thicknesses. cOutdoor radiative cooling tests for paper devices with CPC papers of different thicknesses (top: real temperature, bottom: temperature difference). dSchematic of the sky simulator and device structure (inset). Sky simulator radiative cooling tests for paper devices at different electrical bias for CPC papers with thicknesses of 23 µm (e) and 47 µm (f). In the top panels, the light blue lines are the raw data, while the dark blue lines are locally weighted smoothed data (LOESS, points of window 30). gSpecular reflection of a 47 µm thick CPC paper after different PEI vapor treatment times. hDiffuse reflectance and emissivity of CPC papers after different PEI vapor treatment times. iOutdoor (bottom) and sky simulator (top) radiative cooling tests of PEI-treated 47 µm thick CPC papers. https://doi.org/10.1038/s41528-024-00339-7 Article npj Flexible Electronics | (2024) 8:55 5 highly similar effects as electrochemical tuning and this chemical process is reversible35–37. We treated 47 µm thick CPC papers with PEI vapor for different time durations (20, 40, and 60 min) and characterized their specular and diffuse reflectance (Fig. 3g, h). The spectral variations upon chemical reduction are essentially the same as to those obtained by electrochemical tuning (Figs. 2aand3a), providing an emissivity modulation at 10 µm wavelength from 0.33 (original without PEI vapor treatment) to 0.70 after 60 min PEI treatment. Sky simulator measurements show that increasing the PEI vapor treatment time led to better cooling power, with a difference in temperature decrease of around 1.2 °C between the pristine paper and that treated by PEI vapor for 60 min (top panel of Fig. 3i). Also for outdoor measurements, a 60 min PEI-reduced CPC paper showed a clear lower temperature compared to a non-reduced CPC paper, agreeing well with the measurement results from the sky simulator (bottom panel of Fig. 3i). Dynamic thermal camouflage The large emissivity modulation of the CPC paper devices makes them suitable for dynamic camouflage applications, as here investigated by thermal imaging using an IR camera (Fig. 4a). Electrical redox tuning led to drastic changes in the apparent temperature, from an average of −9.6 °C at −1.5 V to −23.2 °C at +1.5 V (Fig. 4b, 47 µm CPC paper device). The temperature of the surrounding environment was around 0 °C. This large modulation of more than 13 °C results from the large emissivity variation, which could also be gradually controlled by the electrical bias (Fig. 4a, b). Switching the device for multiple cycles showed no observable degradation of the performance (Fig. 4c). As shown in Fig. 4d, increasing the CPC paper thickness lowered the average modulation range only slightly, from 13.7 °C (23 µm CPC paper device) to 11.4 °C (150 µm CPC paper device). The results are in line with the slight reduction in emissivity modulation for devices with thick CPC papers (Fig. 3b). Figure 4e presents the switching response times of the corresponding devices. The trend of increasing response times with increasing CPC paper thickness is the same as for emissivity modulation. The absolute numbers were slightly different in these experiments, likely due to variations between batches of devices. We further note that the switching response times are comparable to those reported for other IR electrochromic devices33,34,41,42. To demonstrate the application of our devices for active thermal camouflage, we placed a 47 µm paper device on a Peltier element (as a heater) with varying temperature. This allowed us to test the ability of the device to conceal an object with temperature that does not only differ from that of the surrounding environment but also varies over time (Fig. 4f). We note here that the temperature of the paper device will be affected by the temperatures of both the Peltier element and the surrounding environment. Varying the temperature of the Peltier element could therefore provide an effective working range for adaptive camouflage (i.e., apparent temperature difference between the concealed object and the surroundings). We first set the apparent temperature of the Peltier element to around 45 °C, while the surrounding apparent temperature was 16.8 °C. This gives a difference in temperature between the concealed object and the surrounding of 28.2 °C, making the (non-concealed part of the) Peltier element clearly stand out from the surroundings in thermal images (Fig. 4f). Without electrical bias, the apparent temperature of the area covered by the paper device was considerably lower even than that of the surroundings (close to 0 °C), making the hidden object visible in the thermal image. This result is attributed to the low thermal emissivity of the high-conducting CPC paper in combination with lower absolute temperature than the Peltier element due to heat exchange with the surroundings (Fig. 4f, g). Reducing the top -35 ºC 5 ºC Apparent temperature Time (s) Apparent temperature (ºC) 10 0 -10 -30 300 600 900 1200 1500 Sponge reference Metal reference ΔT = 13.6 ºC -1.5 V -1.0 V -0.5 V 0 V +0.5 V +1.0 V +1.5 V -9.6 ºC -23.2 ºC-12.0 ºC -15.2 ºC -17.5 ºC -18.4 ºC -19.7 ºC Paper thickness (μm) Apparent temperature variation (ºC) 16 12 8 04080120160 Electrical bias (V) Apparent temperature (ºC) -8 -1.5 -12 -16 -20 -24 -0.5 0.5 1.5 Paper thickness (μm) Apparent temperature response time (s) 23 0 80 40 47 68 92 150 Reduction time Oxidation time 0 10 5 ab cde -20 ºC 60 ºC Apparent temperature Device Peltier element Surrounding Peltier element Surrounding Time (s) Apparent temperature (ºC) 0 -20 80 60 120 Peltier element Surrounding (16.5 ºC) Metal reference (-18.2 ºC) 180 240 60 30 10 ΔT = 11.7 ºC -1.5 V bias +1.5 V bias fgh Thermal camera Device Peltier element Corss-section Time (s) Apparent temperature (ºC) 0 -20 0 10 40 50 120 240 360 Peltier element Surrounding (16.8 ºC) Metal reference (-18.2 ºC) ΔT = 12.3 ºC Device -1.5 V +1.5 V -1.5 V +1.5 V -1.5 V +1.5 V Fig. 4 | IR camouflage applications. a IR camera images of a device (size: 2.1 cm × 1.5 cm) at different electrical bias, based on CPC paper with 47 µm thickness. The temperature of the surrounding environment was around 0 °C. bExtracted average apparent temperature at different electrical bias of the paper device shown in (a). cTemporal variation in apparent temperature of the paper device for multiple cycles of square wave electrical bias (±1.5 V). dModulation range of the apparent temperature for devices made from CPC papers of different thicknesses. eResponse times for oxidation and reduction extracted from thermal camera images. fSchematic of camouflage measurement setup (left) and IR images of a paper device at negative (middle) and positive (right) electrical bias. Device size: 1.8 cm × 1.6 cm. The device was placed on a Peltier element (heated to 45 °C, measured by a thermal camera), which in turn was placed on metal plate (deep blue background color in the thermal images). A plastic foam with emissivity of about 1 was used to represent the temperature of the surrounding environment (located in the top part of the thermal image in yellow–green color). gReal-time apparent temperature variation of the paper device upon redox-switching for a concealed object with temperature of 45 °C (the Peltier element, same measurement as in (f)). hSame as in (g) but upon increasing the temperature of the concealed object in the range from 55 to 65 °C. https://doi.org/10.1038/s41528-024-00339-7 Article npj Flexible Electronics | (2024) 8:55 6 CPC paper by applying a −1.5 V bias led to around 12 °C increase in the apparent temperature. Importantly, this modulation range covers the surrounding temperature, making it possible to select an appropriate electrical bias to conceal the hot object by matching the apparent temperature with that of the surroundings (Fig. 4f, g). We further increased the temperature of the Peltier element gradually to around 70 °C while monitoring the apparent temperature of the device upon switching (Fig. 4h). It was still possible to conceal the hot object until its temperature was beyond 65 °C (corresponding to a difference in temperature between concealed object and surroundings of 48.5 °C). Thus, our paper devices could offer dynamically tunable thermal camouflage capable of adaptively concealing objects with temperature that differs from around 20 to 50 °C from that of the surrounding, fulfilling the requirements of many application scenarios for adaptive camouflage15. Anti-counterfeiting tags The above sections focused on a simple three-layer device architecture. From this section, we will discuss integrating the papers with additional functional films to realize more complex functions, including dynamic anticounterfeiting tags and grayscale IR displays (see Section “Greyscale dynamic IR displays”). Anti-counterfeiting tags widely exist in our daily life, such as watermarks, holograms, and barcodes, all of which can provide information that cannot be seen by the naked eye and cannot be forged without understanding their working principles. Typical materials used for anti-counterfeiting tags include luminescent materials62,opticalmicrostructured materials63, and hydroor thermo-chromic materials64,65. Here, we will use our CPC paper devices with electrically tunable IR emissivity for such applications. As illustrated in Fig. 5a, we used a device structure similar to those used above but with the gel electrolyte layer patterned with an intended hidden information to display on request. In the example, the intended information was “LOE”, which abbreviates “the Laboratory of Organic Electronics”.The information could not be discerned by the naked eye (or a normal camera) with or without electrical bias, thanks to the dark opaque top CPC paper in both redox states (Fig. 5b). When no bias or positive bias was applied to the top CPC paper, also the thermal camera showed a uniform image without revealing the hidden information. By contrast, a clear pattern of “LOE” emerged in the thermal image after applying a negative bias to the top CPC paper. The results are due to differences in redox state and thermal emissivity between non-patterned regions and patterned regions. In patterned regions (without electrolyte), the CPC paper could not be switched and therefore presented its low-emissivity oxidized state regardless of bias potential (and also a low apparent temperature, see Fig. 5b and Supplementary Video 1). As a result, the concept can provide hidden information that cannot be viewed in the visible and that only appears in thermal camera images upon an applied potential. An additional feature for anticounterfeiting can be to include fake information in the visible range. To realize this, we introduced IR transparent dyes with visible color. Prussian blue (PB, blue color) mixed withzinc oxide nanoparticles (ZN, white color) in different ratios was previously shown capable of producing blue colors with various brightness while maintaining hightransparency in mid-IR66. We here deposited a paint made of PB mixed with ZN onto CPC papers by spin coating, and prevented deposition on half of the area using tape. The painted area (depicted by white dashed lines in the top right panel of Fig. 5c) showed a clear bright blue reflected color while the non-painted area remained dark in the visible. By contrast, the painted and non-painted areas presented very similar apparent temperatures, including upon redox-switching (bottom panel of Fig. 5c). The painted area showed only slightly higher average temperature of about 1 °C compared with the area without paint, confirming sufficient IR transparency of these dyes. We measured the response times of these devices and found that the apparent temperature changes were reversible and as rapid as without paints (Fig. 5d). This is promising for using these IR transparent dyes to create visible information that is independent of the information hidden in the dynamic IR image. To test this concept, we painted a cross sign with the IR transparent dye paint on the CPC paper surface and also introduced a combination of cross and tick signs in a patterned tape spacer layer between the top CPC paper and the electrolyte layer in the device (see configuration in the left panel of Fig. 5e). The patterned spacer layer limits the contact between the electrolyte and CPC Apparent temperature -25 ºC 5 ºC Conducting polymer-cellulose paper Gel electrolyte + - Patterned electrolyte Visible image at +1.5 V Visible image at -1.5 V Thermal image at +1.5 V Thermal image at -1.5 V ab Apparent temperature -40 ºC 0 ºC IR transparent paint Conducting polymer -cellulose paper Visible image Thermal image +1.5 V bias -1.5 V bias cd Time (s) 0600 1200 1800 Average apparent temperature (ºC) 0 -16 -32 Metal reference Air reference IR transparent paints Conducting polymer-cellulose paper e Visible image (no bias) Paint pattern Tape pattern × Tape -40 ºC 10 ºC +1.5 V -1.5 V Thermal image (no bias) Thermal image (+1.5 V) Thermal image (-1.5 V) Fig. 5 | Anti-counterfeiting tags based on CPC paper and IR transparent paints. aSchematics of the device configuration, including a gel electrolyte layer with hollow patterns (details for fabrication can be found in Methods Section). bSample images at ±1.5 bias obtained by a normal camera (left) and by a thermal camera (right). Device size: 3.0 cm × 1.2 cm. cA paper device with half of its area covered by an IR transparent blue paint. The panel shows a schematic (top left), visible image (top right), and thermal images at +1.5 V bias (bottom left) and −1.5 V bias (bottom right). The area covered by the dye has been marked by white dashed lines. Device size: 1.8 cm × 1.6 cm. dTemporal variations in the apparent temperature upon electrical switching of the device shown in (c), showing temperatures variations for both the region painted by the IR transparent dye and the bare paper region. eAnticounterfeiting device configuration (top left) with both a paint pattern on the paper surface (top center) and a patterned tape spacer layer (top right). The panel presents a visible image of the device (middle left), and thermal images without (middle right) and with electrical biases (bottom). Device size: 2.4 cm × 1.1 cm. https://doi.org/10.1038/s41528-024-00339-7 Article npj Flexible Electronics | (2024) 8:55 7 paper to only certain regions, which therefore will be easily switched while others will not. (A different configuration with patterned electrolyte instead of patterned spacer also worked but with less good results, see Supplementary Fig. 12.) We first note that the visible information (painted dye tick) was clearly seen in the visible for both states, while only barely resolved in the thermal images. By contrast, the hidden information below the paper could be made appear or disappear in the thermal images by varying the electrical bias (Fig. 5e), while they remained invisible in the visible. The results demonstrate that the combination of IR transparent dyes and our paper devices are excellent candidates for anti-counterfeiting tags. Grayscale dynamic IR displays IR displays are useful for not only anticounterfeiting, but also various situations, such as smart road signs forautonomousvehiclesordriversat nighttime33 wherevisionindarknessisdemanded.Above,wedemonstrated only digital IR displays and with rather low resolution. As a final demonstration, we combined the CPC paper devices with the concept of UV patterning of conducting polymer thin films67–69 to realize dynamic grayscale IR images with high resolution. The UV patterning method is based on vapor phase polymerization of PEDOT thin films with an added UV exposure step of the polymerization precursors through a photomask67–69. The process flow is depicted in Fig. 6a and can be used to produce microsizedgrayscalepatternsinthinPEDOTfilms. As demonstrated in our previous studies, UV patterning can lead to large differences in various film properties, including thickness, visible and IR absorption, and complex refractive index69. Here, we polymerized a UV-patterned thin PEDOT:Tosylate layer directly on the CPC paper (see details in the “Methods” section). Thermal camera imaging of regions exposed to different UV doses presented large differences in apparent temperatures (Fig. 6b). Tuning the UV transmittance of the photomask from 70% to 1% led to variation of the apparent temperature by almost 10 °C, from 14.5 to 5.6 °C (Fig. 6b). The results agree with previous reports showing decreased electrical conductivity of conducting polymer films prepared at increasingly higher UV exposure doses67,69, which makes them less reflective and more emissive in the IR (Fig. 6c). We further printed grayscale photomasks of the portrait of one of the co-authors (Fig. 6d), and created patterned PEDOT thin films on top of a CPC paper. The portrait could hardly be seen in the visible except for at certain angles (attributed to differences in reflection caused by variations in thickness). By contrast, the pattern was perfectly reproduced in the thermal image. To demonstrate electrical tunability, we applied ±1.5 V bias and monitored changes in the thermal image for a similar UV-patterned device (Fig. 6e and Supplementary Video 2). The thermal image disappeared at negative bias and re-appeared with the positive bias. The different parts of the images presented different modulation of the apparent temperatures upon redox tuning. For example, the forehead part made through high UV exposure dose only showed modulation range of 2.1 °C, while the hair part corresponding to low UV dose presented a higher modulation of 13.5 °C (marked with circles in the left panel of Fig. 6e, f). This can be explained by UV exposure leading to lower electrical conductivity and thereby higher IR emissivity also for the oxidized state of the material70. Indeed, both regions (high and low UV exposure doses) exhibited similar apparent temperature of about 10.0 °C in the reduced state, making it possible to (reversibly) erase the image by negative bias (Fig. 6f). Besides adding high resolution grayscale images, UV patterning therefore provides additional functionality which may be relevant for anti-counterfeiting systems. In summary, this paper has demonstrated that CPC paper can be used as ion permeable and dynamically tunable electrodes for active IR optical devices. The CPC papers possess good mechanical properties and show outstanding out-of-plane ionic transport properties with short response UV light exposure Exposure to EDOT vapour 234 UV patterned PEDOT thin films with varying thicknesses Oxidant film Substrate 1 S OO Conducting polymer -cellulose paper Greyscale photomask Designed photomask Thermal image of the sample Apparent temperature -5 ºC 20 ºC AB CD AB CD UV dose Emissivity Wavelength (μm) 4 8 12 16 1.0 0.6 0.8 0.4 0.2 Without exposure Pixel A Pixel B Pixel C Pixel D Time (s) Apparent temperature (ºC) 20 10 -10 0 600 1200 1800 +1.5 V -1.5 V Forehead Hair 0 Designed photomask Printed photomask Visible image Thermal image Apparent temperature 10 ºC -23 ºC Apparent temperature 15 ºC -30 ºC +1.5 V bias -1.5 V bias +1.5 V bias ab cde f Fig. 6 | Grayscale dynamic IR displays. a Process flow of UV patterning method to create well-defined patterns on CPC paper devices. bGrayscale photomask design (left) and observed thermal image of a sample containing a UV-patterned PEDOT:Tosylate film made using that design. In the design, square pixels with different UV transmission were used (A: 1%, B: 3%, C: 22%, and D: 70%). The average apparent temperatures for pixels A–D were 5.6, 9.9, 14.1, and 14.5 °C, respectively. Device size: 2.4 cm × 2.4 cm. cMeasured IR emissivity of the UV-patterned PEDOT:Tosylate films (without electrical bias) on the CPC paper with different UV doses (pixels A–D corresponding to the sample shown in b). dCPC paper with a UVpatterned PEDOT:Tosylate portrait (the original photo used was taken by Thor Balkhed, Linköping University. The copyright of the photo belongs to D.Z.). Device size: 3.0 cm × 2.5 cm. The bottom panels show visible and IR images and the top panels show the designed pattern and printed mask. Electrically tunable paper displays based on CPC paper with a portrait pattern at different electrical bias (e) and its real-time apparent temperature curve (f) for the hair and the forehead parts of the portrait (the original photo used was taken by Thor Balkhed, Linköping University. The copyright of the photo belongs to M.P.J.). Device size: 3.0 cm × 2.5 cm. The thickness of the CPC papers for these devices was 47 µm. The UV photomask was made using an office printer with resolution of about 100 µm. https://doi.org/10.1038/s41528-024-00339-7 Article npj Flexible Electronics | (2024) 8:55 8 times for electrical redox state tuning of CPC paper devices. Specular IR reflectance and thermal emissivity could be modulated by about 40% using low electrical bias of only ±1.5 V, representing one of the best IR-tunable materials. These properties enable multiple promising IR applications, where we demonstrated tunable radiative coolers, active thermal camouflage, dynamic anti-counterfeiting tags, and grayscale IR displays. We believe this special type of ion-permeable electrodes will become a powerful materials platform for IR optical devices, including not only the applications demonstrated herein but also concepts such as electrically tunable metasurfaces and neuromorphics. Methods Fabrication of CPC papers Conducting polymer PEDOT:PSS (PH 1000, 1.3 wt%) was purchased from Heraeus Clevios. Cellulose nanofibrils (CNF, carboxymethylated, 0.52 wt% aqueous solution) were purchased from RISE Innventia AB. Other chemicals were purchased from Sigma-Aldrich. PEDOT:PSS, CNF, glycerol, and ethylene glycol were mixed at weight ratio of 130:130:1:7 or absolute weight ratio of 1.69:0.68:1:7. The blend solution was then homogenized using an ULTRA-TURRAX T-10 dispenser for a few minutes, followed by a degassing step of 1 hour using a vacuum desiccator or a vacuum oven. The solution was then poured into a container (e.g., petri dish), and dried either in a fume hood for 2 days or in a normal oven at 60 °C for 6 h. When the solvent was removed, the CPC paper could be easily peeled off from the container. We varied the volume of the mixed solution to obtain papers with different thicknesses. Fabricating the IR paper devices The standard three-layer device configuration is depicted in Fig. 2a. The gel electrolyte [EMIM][TFSI] in PVDF-HFP was produced following previous reports50,51. In brief, PVDF-HFP, [EMIM][TFSI], and acetone were mixed at a weight ratio of 1:4:7 and stirred in a sealed vial overnight to obtain a homogeneous solution. The solution was drop-casted onto a clean glass substrate followed by a heating step at 60 °C on a hot plate for 2 h to remove residual solvent. The drop-casted layer could then be easily detached from the substrate and transferred onto a CPC paper to be used as the electrolyte layer. Another CPC paper was then placed on top of the electrolyte layer to form the three-layer device. To realize complex functions, the electrolyte layer and/or the top paper layer can be patterned. To create the type of devicesshowninFig.5a, the patterns can be produced manually or through aflatbed cutter. We used the flatbed cutter to remove the “LOE”pattern in the electrolyte while the remaining parts formed a continuous gel film. Additional layers can be deposited on top of the CPC paper device, whereweinthisstudyusedIRtransparent dyes and UV-patterned PEDOT:Tosylate layers. The IR transparent dyes were based on Prussian Blue (PB) and zinc oxide nanoparticles (ZN), as purchased from Sigma-Aldrich. The dyes were dissolved in ethanol and mixed to create the colored paint. The blue color used to produce the sample imaged in Fig. 5c were produced by a 20:1 mixture of PB and ZN. The dye was painted on the CPC papers by spin coating and using cleanroom scotch tape to prevent deposition onto the unwanted area (e.g., to create patterns). For the UV-patterned PEDOT:Tosylate layers, we followed the procedure reported in our previous studies69,followingthesimplified process flowshowninFig.6a(withUV exposure time of 20 min and polymerization time of 1 h). The UV mask was made using an office printer with resolution of about 100 µm. To operate the device, the top paper was connected to the positive electrode and the bottom paper to the negative electrode of the power source. Electrical conductivity measurements Bulk electrical conductivities of CPC papers were measured via the four-line probe method. Four-line gold/titanium electrodes with thickness of 100 nm/5 nm, width of 1 mm, and length of 10 mm were deposited onto glass substrates through thermal evaporation. Due to the hydrophilicity of the CPC papers, they could be easily attached to the glass substrates with Au electrodes, showing good adhesion. To measure the conductivities of the papers at different redox states, we directly attached a three-layer device onto the gold electrode glass substrate. The top and bottom papers were connected to the working electrode and counter/reference electrode of a potentiostat (Bio-logic SP-200). The gold lines were directly connected to four probe stations of a SCS parameter analyzer (Keithley 4200) in the configuration of four probe method. For each electrical bias, a stabilization time of 60 s were used to ensure the paper had reached a stable redox state. The electrical conductivity was then calculated via the formula of σ¼1=ðRstÞ,whereR s is the sheet resistance measured by the four-line probe method and tis the paper thickness. Specular reflectance measurements Static specular reflectance spectra at different electrical bias were collected using a Lambda 900 spectrometer (Perkin Elmer) for the visible and near IR and using a Spectrum 3 FT-IR (Perkin Elmer) for the mid-IR. For static measurements with electrical bias, a stabilization time of 60 s were used to ensure the paper had reached a stable redox state. For dynamic specular reflectance (shown in Fig. 2d), a fiberspectrometer(AvaLight-DH-S-BAL for light source and AvaSpec-NIR256-2.5 for spectrometer, Avantes) was used. To extract response times (Fig. 2e),thetimedrivemodeofLambda900 was used with the resolution of 1 s. Diffuse reflectance measurements Diffuse reflectance or emissivity measurements of CPC papers were carried out using a Bruker Vertex 70 FT-IR spectrometer with a downward-looking diffuse gold-coated integrating sphere (Labsphere A562). The angle of incidence for the integrating sphere was 9°. Since the paper is opaque in IR spectral range, we could calculate the emissivity of via ε¼1R,whereRis the diffuse reflectance of the paper. Thermal image characterizations Thermal images were obtained through a ThermoVision A320G camera (FLIR Systems). The spectral range for this camera was 7.5–13.0 µm with a resolution of 320 × 240 pixels. In this study, all the images were captured outdoors during nighttime. For the apparent temperature measurement, we used a highly reflective metal and a highly absorptive polymer foam as the calibration targets. Radiative cooling measurements and the sky simulator We utilized our home-built setup for outdoor radiative cooling measurement, as reported in our previous studies71,72. In brief, the samples were placed in a chamber coated with Al tape with its opening covered by a piece of PE film. Three thermocouples (K type, Pentronic AB) were used to monitor the real-time temperature variation of the sample, reference objects, and air (surrounding). For indoor measurements, we used a sky simulator as described in our previous study34. A cross-section schematic of the setup is shown in Fig. 3d, where the inside wall of the setup chamber was covered by IR-reflective Al foils to guide the emitted thermal radiation into a thermoflask. This thermoflask was filled with liquid nitrogen and had an IR-absorbing layer at the bottom, to mimic the cold outer space that could absorb the emitted thermal radiation from the sample. The sky simulator resembles but does not perfectly simulate the real sky, as discussed in more detail in our previous study34,60. During measurements, the sample was embedded in a thermal insulation layer and two thermocouples were used to monitor the temperature variations on the backside of the sample and the front side (inside the chamber) of the sample, as shown in the inset of Fig. 3d. Spectroscopic ellipsometry measurements PEDOT:PSS thin film samples were prepared by spin coating onto sapphire substrates and measured without or after 15 min PEI vapor treatment. The ellipsometry measurements were carried out at room temperature under normal ambient conditions, as reported in our previous studies5,36,37.RC2 and IR-VASE ellipsometers (J. A. Woollam Co.) were used for measurement in the visible-NIR range (400–1690 nm) and mid-IR range (2.0–30.0 μm). https://doi.org/10.1038/s41528-024-00339-7 Article npj Flexible Electronics | (2024) 8:55 9