Mid-Infrared Electrochromics Enabled by Intraband Modulation in Carbon Nanotube Networks

Tuneable infrared properties, such as transparency and emissivity, are highly desirable for a range of applications, including thermal windows and emissive cooling. Here, we demonstrate the use of carbon nanotube networks spray-deposited onto an ionic liquid-infused membrane to fabricate devices with electrochromic modulation in the mid-infrared spectrum, facilitating control of emissivity and apparent temperature. Such modulation is enabled by intraband transitions in unsorted single-walled carbon nanotube networks, allowing the use of scalable nanotube inks for printed devices. These devices are optimized by varying film thickness and sheet resistance, demonstrating the emissivity modulation (from ∼0.5 to ∼0.2). These devices and the understanding thereof open the door to selection criteria for infrared electrochromic materials based on the relationship between band structure, electrochemistry, and optothermal properties to enable the development of solution-processable large-area coatings for widespread thermal management applications.

(a) shows Raman spectra of the nanotube films using various laser lines. The radial breathing modes are plotted in figure S1(b) and fitted to show the contributing peaks. The observed estimated diameters from the radial breathing modes Raman shift were extracted using the formula ω RBM = A/d t + B where ω RBM is the shift of the peak corresponding to a particular tubes radial breathing mode, d t is the diameter of the carbon nanotubes, and A and B are constants 1 . These diameters can be superimposed on the Kataura plot 2 for each relevant laser line. Due to the effect of the bundle shift on S-2 observed diameter as well as the inability of the laser line to be exactly in resonance, the points often do not match up exactly with a particular point on the Kataura plot. To this end, the plot can be used qualitatively.
Analysis of the AFM in figure S2a allowed to look at individual and bundled tubes. The diameters vary from low (indicating single tubes) to large indicating bundles. AFM image shows variation in nanotubes diameter from large bundles up to 20 nm in diameter to an individualised nanotube of diameter around 0.5 nm. The statistical distribution of different diameters illustrated through histogram in figure S2b.
Cyclic voltammetry of the device at elevated temperature (100 °C) shows drift in the current above 2 V with each cycle at low scan rates (10 mV/s, figure 2a) this is less observable at higher scan rates (100 mV/s, figure 2b) suggesting this is a faradaic process. These processes could be degradation of the electrolyte due to electrolysis or the removal of water through electrolysis. Since the current is

S-3
increasing it is more likely to be electrolyte degradation as the current from hydrolysis should decrease as the cycles progress. The 3 cycles were produced after 20 cycles at 100 mV/s. The influence of the different potential bias on nanotubes electrode were investigated through Raman spectroscopy by monitoring changes in the D-peak (1322 cm -1 ). The higher intensity of Dpeak indicates an increasing number of defects in the graphitic structure. 3 Figure S4a shows the increment in the D-peak intensity and its gradual broadening with increasing biases. Qualitatively, The D-peak undergoes a significant increase in intensity, width (FWHM), and background scattering after exceeding +2.5 V. The broad peak features of D-peak feature are associated with the presence of amorphous carbon, 4 revealing the transformation of some sp 2 carbon atoms into sp 3 through structural disorder. It indicates electrode damage through the amorphization of carbon structure under electrical stimulation with a threshold voltage of +2.5 volts. This trend is reproducible with several laser energy (1.58 to 2.54 eV), figure S4b). Interestingly, the higher energy laser shows more instability which corresponds to the semiconducting tubes of lower diameter. With the consideration of cyclability of devices, the performance was monitored through cycles between -1 V and 1.5 or 2 V for 1000 cycles. They show reasonable stability with comparable effective temperature drop to that of the device at optimum performance. Figures S4c, d and e characterise this stability showing a drop in the maximum observed temperature (T max ) with cycle number as well as a smaller drop in minimum observed temperature (T min ). The overall fraction of the device performance after 1000 cycles over the initial cycle was approximately 0.65".

S-11
Figure S12(a) plots the effective temperature of sprayed SWCNT films on PET against the body temperature and Figure S12(b) plots the emissivity of the films against the optical transparency at 550 nm, the more opaque in the visible, the lower the emissivity in the IR for the SWCNT films. Figure S12(c) shows the effective temperature of the PET substrate against body temperature and it is worth noting that the emissivity in Figure S11b tends towards the emissivity of the substrate. Therefore, the transparency of the device, as well as the changes in transparency in Figure 4 contribute to the effective emissivity of the device. These devices operated in air maintain approximately 66% of the temperature differential over 1000 cycles (figure S3).