An azobenzene-based photothermal energy storage system for co-harvesting photon energy and low-grade ambient heat via a photoinduced crystal-to-liquid transition

Reagents and materials

In this study, nitrosobenzene (C₆H₅NO, 98%), 4-chloroaniline (C₆H₆ClN, 98%), 4-bromoaniline (C₆H₆BrN, 99%), 4-iodoaniline (C₆H₆IN, 98%), para-anisidine (C₇H₉NO, 99%), 4-ethoxyaniline (C₈H₁₁NO, 98%), glacial acetic acid (AR) and a microscope glass slide (24mm × 24mm × 0.17 mm, length × width × thickness) were purchased from Shanghai Aladdin Biochemical Technology Co., Ltd. Dichloromethane (DCM, AR and SP), ethyl acetate (AR) and petroleum ether (AR) were purchased from Tianjin Kemiou Chemical Reagent Co., Ltd. All reagents were used directly without further purification.

Synthesis and characterization

Azo compounds were synthesized via the Mills reaction^[26,27]. The detailed synthesis procedures and characterization data, including the high-resolution mass spectra (HRMS), ¹H and ¹³C nuclear magnetic resonance (NMR) spectra, and Fourier transform-iIR spectra of the products, are included in Supplementary Table 1.

A SepaBean machine automated flash chromatography system (Santai Technologies, Inc.) was used for purification. ¹H and ¹³C NMR spectra were recorded using a 400 MHz spectrometer (INOVA, Varian, USA) with chloroform-d as the solvent and tetramethylsilane as an internal standard. IR spectra were recorded using a Bruker Vertex 70 Fourier spectrometer with the powdered samples pressed into KBr pellets. HRMS (ESI) were recorded on a Bruker solanX 70. The samples were charged/discharged under controlled light irradiation with different wavelengths, intensities, and durations. The intensity of light was measured with an optical power meter (CEL-NP2000, Beijing China Education Au-light Co., Ltd.). Optical microscope images were obtained using a Nikon SMZ745T (Japan). An IR thermal imaging camera (Fluke TiX640, USA) was used to monitor the heat released from the samples.

UV-Vis absorption spectroscopy

The UV-Vis absorption spectra (1 cm path-length quartz cuvettes, DCM) were measured using a UV-Vis spectrophotometer (3600 Plus, Shimadzu, Japan). Baselines were corrected and spectra were normalized using Origin8.6 software.

For the solution state test, trans-isomers were dissolved in DCM. The UV-Vis spectra were then recorded as trans-isomers. After that, the samples were irradiated with 365-nm wavelength light (80 mW/cm²) in solution state until they reached a photostationary state (PSS), which generally takes 10 to 30 s. The UV-Vis spectra were then recorded as the cis-isomers. Since the solution-state Azo compounds have ample free space, the cis-isomer content can exceed 95% in the 365-nm PSS. Therefore, we approximate the 365-nm PPS Azo compounds in the solution state as 100% cis-configurations in this work. The UV-Vis absorption spectra of trans- and cis-Azo compounds in solution state (1 × 10^-4 mol L^-1 DCM solution) can be seen in Supplementary Figures 1, 2A, E and I .

For the solid-state test, a small sample was taken with a glass rod (Φ = 0.5 mm) and dissolved in a cuvette filled with DCM solution. The UV-Vis spectra were then recorded. The obtained data were normalized using Origin 8.6 software. The cis-isomer content in the 365-nm PSS can be estimated as the percentage change of absorbance at the wavelength of the π-π* transition peak following the method reported in the literature^[28,29].

%cis ≈ (A – A_trans)/(A_cis – A_trans) (1)

The 365-nm PSS in dilute solutions was assumed to be ~100% cis, while the thermal-stationary state was assumed to be ~100% trans. A is the absorbance during the trans-to-cis photoisomerization.

Energy charging process

In a dark room, trans-crystal powder samples were set on a 24 × 24 mm glass slide. The slide was set on a constant temperature heating platform that simulated the ambient heat (T₁). The sample was then irradiated with 365-nm wavelength light (80 Mw/cm², 5 cm away) until the trans-crystal was converted into the cis-liquid through photoisomerization. After that, the heating platform was adjusted to the lowest temperature (T₂) so that the cis isomer could maintain a liquid state to maximize the trans-to-cis photoisomerization and obtain as much cis-isomer as possible. Images of the samples in the crystal and liquid states were recorded using a smartphone (Honor 30 Pro). During the irradiation, the UV-Vis absorption spectra of the samples were recorded at regular intervals. The obtained data were normalized by software. The T₁ and T₂ of the five Azo compounds are presented in Supplementary Table 2.

DSC measurements

DSC analysis was conducted on a DSC Q20 (TA Instruments, USA) with cooling from an RCS90. If not otherwise stated, the cooling and heating rates were set to 10 °C/min. In the phase transition property experiments, the cis-isomer samples were prepared using the solvent-assisted charging procedure to obtain the values of T_mand T_cas precisely as possible. The specific procedure is as follows. Each trans-isomer was dissolved in DCM and irradiated with 365-nm wavelength light (80 mW/cm²) until reaching the PSS. The solutions were then concentrated and dried in a vacuum before being transferred to DSC pans for analysis. In the energy density measurements experiments, the preparation procedure of the cis-isomer samples was the same as that for the energy charging process.

Lifetime measurements

The cis-liquid samples of the 365-nm PSS were transferred to a dark room at a constant temperature (35 °C for Azo-Cl, 45 °C for Azo-Br, 65 °C for Azo-I and 25 °C for Azo-OMe and Azo-OEt). The UV-Vis absorption spectra of the samples were recorded at regular intervals until the samples reverted to their initial state, i.e., the trans-crystal state. The obtained data were normalized by software.

Photoinduced liquid-to-crystal phase transition experiments

Cis-liquid isomer thin films were prepared by placing 8 mg of trans-crystal isomer sample on a clean glass slide (24 × 24 mm) and melting it on a heating platform. The trans-liquid sample was flattened by pressing with another glass slide to fill the entire space between them and cooled to T_{1 [}Supplementary Table 2] to simulate the ambient heat. The sample was then irradiated with 365-nm wavelength light (80 mW/cm²⁾ to induce the trans-to-cis photoisomerization. The temperature was then reduced to T_{2 [}Supplementary Table 2] using an ethanol bath with a circulatory system and the light irradiation was removed. Half of the sample was covered with cardboard and irradiation was applied with 420-nm wavelength light (80 mW/cm²) to trigger the cis-to-trans isomerization and induce crystallization. After crystallization, the cover was removed and images were taken with an optical microscope at room temperature. The UV-Vis absorption spectra of the samples were recorded at regular intervals during the photoinduced crystallization. The obtained data were normalized by software.

Ir thermal imaging

The cis-liquid samples were set in a dark room at a constant temperature (35 °C for Azo-Cl, 45 °C for Azo-Br, 65 °C for Azo-I and 25 °C for Azo-OMe and Azo-OEt). The samples were then irradiated with 420-nm wavelength light (80 mW/cm², 5 cm away) under an IR thermal imaging camera.

Theoretical calculations

The structures of molecules 1 to 5 were first optimized using the density functional theory (DFT) at the ωB97XD/def2-SVP level. Among all the possible conformations for both the trans- and cis-isomers, all geometry optimizations were performed using the Gaussian 16a package. Harmonic frequency calculations were performed using the same theory to help verify that none of the structures have imaginary frequencies. The most stable structures were selected to analyze the direction and magnitude of the molecular dipole moment.

Molecular design

In order to achieve PCLTs, the Azo molecules must have a large melting point gap (ΔT_m) between the trans- and cis-isomers so that the trans-crystal and cis-liquid phases can be formed under the specific temperature T_phase(cis-T_m< T_phase< trans-T_m). Grafting functional groups on the benzene ring of the Azo molecules disrupts the molecule symmetry, resulting in a large ΔT_mbetween two isomers. In addition, the DFT calculation result shows that the cis isomers of the unilateral-functionalized Azo derivatives have significantly higher polarity than their trans-forms [Figure 1B]. The high polarity of the cis-isomers effectively disrupts the π-π interactions between the aromatic groups, eventually reducing their ability to pack and form ordered crystals.^[20] Hence, we synthesized five Azo photoswitches that crystallize in the trans-form and liquefy in the cis-form at or slightly above RT for co-harvesting photon energy and ambient heat via PCLTs [Figure 1A].

Figure 1. (A) Concept and molecular design for co-harvesting of ambient heat and photon energy via the trans-crystal-to-cis-liquid transition of photoliquefiable Azos. (B) Dipole moment calculation (Debye) of photoliquefiable Azos.

Energy charging: photoinduced trans-crystal-to-cis-liquid transition

The photoliquification behaviors of all five Azos forms were observed and presented in Figure 2 and Supplementary Figure 2. To absorb ambient heat above RT, we selected halogen-grafted Azos with photoinduced phase transition temperatures above RT. Such Azos liquify under UV light at ambient temperatures above RT and store the ambient heat in liquid-state molecules as latent heat [Figure 2A, Supplementary Figure 2D and H]. Taking the Azo-Cl molecules as examples, as shown in Figure 2A, an ambient temperature of 60 °C is first simulated. The trans-crystal sample is then irradiated with 365-nm wavelength light to trigger the trans-to-cis photoisomerization and the simultaneous crystal-to-liquid phase transition, which harvests the photon energy and ambient heat. During light irradiation, Azo-Cl was sampled and dissolved, and its isomerization degree was monitored by UV-Vis spectroscopy.

Figure 2. Energy harvested from ambient heat and photons, i.e., the energy charging process. First column: images of Azo-Cl (A) and Azo-OEt (D) samples on glass slides, showing the trans-crystals at the specific temperature, cis-liquid after UV light irradiation (365 nm) and cis-liquid after cooling, as indicated by the arrows. Second column: time-evolved UV-Vis absorption spectra of Azo-Cl (B) and Azo-OEt (E) under UV light of 365 nm. The spectra were normalized with respect to the isosbestic point at 279 nm for Azo-Cl and 302 nm for Azo-OEt, respectively. The arrows indicate the order of tests performed. Third column: the cis-isomer percentage of Azo-Cl (C) and Azo-OEt (F) vs. irradiation time.

For the purpose of maximizing the trans-to-cis photoisomerization and obtaining as much cis-isomer as possible, the cis-liquid sample is cooled to the lowest temperature (T₂ in Supplementary Table 2) at which it could remain in the liquid state while maintaining light irradiation. As shown in Figure 2B and C, the Azo-Cl sample reached the 365-nm PSS after a 60 min irradiation and the cis-isomer proportion reached 72%. It should be noted that prolonging the irradiation could reduce the cis-liquid content. The main reason for this is that the relatively high temperature of ambient heat causes the reverse conversion of trans-to-cis isomerization [Supplementary Figure 2B, C, F and G]. For ambient heat at RT, we selected alkoxy-grafted Azos with photoinduced phase transition temperatures at RT. Taking the Azo-OEt molecules as examples, as displayed in Figure 2D, the trans-isomer of Azo-OEt is photoliquified into a reddish-orange liquid under RT. The Azo-OEt sample could reach the 365-nm PSS within 18 min at RT and exhibit a high trans-to-cis photoisomerization degree with a cis-isomer content of 92% [Figure 2E and F]. This is approaching the maximum level achievable in diluted solutions (91%-96%).^[25] Moreover, The 365-nm PSS irradiation time and cis-isomer content of Azo-OMe sample are 10 min and 86% [Supplementary Figure 2J and K]. Through such high effective trans-to-cis isomerization during photo liquefication, energy harvesting could be more efficient and could result in high energy densities. Furthermore, the alkoxy-grafted cis-isomers remained in their liquid phase at -20 °C, indicating that the phase state of the cis-isomers is highly reliant on the para-substituent [Figure 2D and 2L]^[30,31].

Energy storage

The notable difference in crystallinity between the trans- and cis-isomers of the five as-prepared Azos, i.e., Azo-Cl [Figure 3A], Azo-Br [Supplementary Figure 3A], Azo-I [Supplementary Figure 3B], Azo-OMe [Supplementary Figure 3C] and Azo-OEt [Figure 3B], was measured by DSC. The trans-forms of all five Azos exhibit sharpened melting peaks upon heating and crystallization peaks upon cooling. The same melting and crystallization characteristics are significantly reduced (Azo-Cl, Azo-Br, and Azo-I molecules) or absent (Azo-OMe and Azo-OEt compounds) for the cis-form counterparts. This intriguing result could be attributed to the significantly higher polarity of the cis-isomers than the trans-isomers, which effectively disrupts the π-π interactions between aromatic groups, eventually reducing their ability to pack and form ordered crystals.^[20,32] The cis-forms of halogen-grafted Azo-Cl [Figure 3A], Azo-Br [Supplementary Figure 3A] and Azo-I [Supplementary Figure 3B] maintained their liquid phase at temperatures above 24, 46, and 64 °C, respectively. These temperatures were slightly lower than the observed ones for maintaining the cis-liquid phase. This discrepancy may be attributed to the different preparation methods (solvent-assisted or direct irradiation in the solid state), resulting in different cis-isomer contents [Supplementary Figures 3C-E]. Additionally, Azo-Cl exhibits a cold-crystallization peak at ~45 °C, indicating that it is more prone to crystallization than other compounds [Figure 3A].

Figure 3. Phase transition of isomers and energy storage properties. First row: DSC plots of trans (top black curve)- and cis (bottom red curve)-isomers upon heating and cooling with Azo-Cl (A) as a representative example of halogen-grafted Azos and Azo-OEt (B) as a representative example of alkoxy-grafted Azos. T_m, T_c, and T_cc represent the melting, crystallization, and cold-crystallization temperatures, respectively. Second row: ¹H NMR spectra of Azo-Cl (C) and Azo-OEt (D) isomers for DSC test. Third row: DSC curves of typical cis-liquid Azos for Azo-Cl (E) as a representative example of halogen-grafted Azos and Azo-OEt (F) as a representative example of alkoxy-grafted Azos. Fourth row: time-evolved UV-Vis absorption spectra of the cis-liquid form of Azo-Cl (G) and Azo-OEt (H) reversion in darkness (the spectra were normalized with respect to their corresponding isosbestic point). The arrows indicate the test order.

The cis-forms of the alkoxy-grafted Azos displayed remarkable liquid-phase stability at temperatures below 0 °C, thereby providing opportunities for low-temperature heat release [Figures 3B and Supplementary Figure 3C]. As summarized in Supplementary Table 3, these data match the observed phenomena, suggesting that direct irradiation in the solid state has the same effect on energy storage as the solvent-assisted charging method at RT [Figures 3D and Supplementary Figure 3F]. Therefore, a significant difference can be observed between the trans- and cis-isomers for all Azos. Specifically, the trans-isomers are highly crystalline, while the cis-isomers are liquid at the same temperature. This phase state discrepancy indicates an immense opportunity for co-storing the photon energy and ambient heat in the cis-liquid form of molecules and releasing them as high-grade thermal energy through the cis-liquid-to-trans-crystal transition.

DSC was also used to investigate the amount of energy stored in the cis-liquid isomers of all five Azos: Azo-Cl [Figure 3E], Azo-Br [Supplementary Figure 4A], Azo-I [Supplementary Figure 4B], Azo-OMe [Supplementary Figure 4C] and Azo-OEt [Figure 3F]. Each Azo compound has two separate exothermic peaks. The broad exothermic heat flows in the first heating segment were attributed to the heat-induced cis-to-trans isomerization. This portion of energy came from the photon energy captured during the energy charging process. The sharp exothermic peaks in the cooling segment were ascribed to the crystallization of the trans-liquid state. This portion of energy came from the thermal energy absorbed from the ambient environment during the energy charging process. Therefore, these Azo molecules harvested and stored the hard-to-collect ambient heat through their PCLT behaviors without notably compromising the storage capacity of photon energy. The photon energy and ambient heat were tactfully integrated into the single-component small molecule via photochemistry and thermophysics, resulting in a high energy density (172-335 J/g). The energy densities of all five Azo compounds are presented in Table 1. Furthermore, the time-evolved UV-Vis absorption spectra of cis-liquid Azos in darkness were plotted to survey their storage lifetimes (Azo-Cl in Figure 3G, Azo-Br in Supplementary Figure 4D, Azo-I in Supplementary Figure 4E, Azo-OMe in Supplementary Figure 4F and Azo-OEt in Figure 3H. The results are summarized in Table 1. Alkoxy-grafted Azos have a significantly longer storage lifetime than halogen-grafted Azos because the cis-form halogen-grafted Azos require higher ambient temperatures to sustain their liquid phase. However, the high temperature could also cause thermally-induced cis-to-trans isomerization. Therefore, it is necessary to select suitable photoliquefiable Azos to accommodate various application scenarios. For example, halogen-grafted Azos are potentially suitable for applications that need to accumulate heat in a short period of time or within a small area, while alkoxy-grafted Azos are more favorable candidates for long-term energy storage.

Energy discharging: photoinduced cis-liquid-to-trans-crystal transition

Further verification of the photoinduced crystallization is achieved via selectively exposing the cis-liquid films to visible light (420 nm) using an optical mask (typically cardboard), as shown in Figure 4A. The 420-nm visible light irradiation can help the cis-isomers overcome the energy barrier and accelerate the cis-liquid-to-trans-crystal isomerization. As shown in Figure 4B, C and Supplementary Figure 5A-C, the cis-liquid films are partially covered with cardboard on the left and the entire films are exposed to light irradiation at 420 nm. The phase of the covered part (the left side) has no variation, whereas the right sides of the cis-liquid films crystallize due to light irradiation. These optical microscope images clearly visualize the crystalline features of each trans-isomer and the boundary between the original liquid phase and the generated crystalline phase. This indicates that the crystallization is induced exclusively by light irradiation and not artifact nucleation. Therefore, temporal and spatial control over the heat release can be achieved with light triggers. Supplementary Figure 6 shows the decrease in cis content in the liquid films under the 420-nm wavelength light irradiation. After 20 s of the 420-nm light irradiation, the cis content of the halogen-grafted Azos drops below 40% and crystallizes [Figure 4D, Supplementary Figure 5D and 5E]. In contrast, the alkoxy-grafted Azos require at least 10 min to achieve the same effects [Figure 4E and Supplementary Figure 5F], indicating a higher liquid phase stability.

Figure 4. Experimental demonstration of selective crystallization of liquid sample and the heat release monitoring via photoinduction. (A) Schematic illustration of the selective crystallization of liquid sample via photoinduction. Selective photoinduced crystallization process of (B) Azo-Cl as a representative example of halogen-grafted Azos and (C) Azo-OEt as a representative example of alkoxy-grafted Azos, which were irradiated with visible light (420 nm) at ambient temperatures of 35 and -20 °C, respectively. The left side of the liquid film is covered with cardboard to preserve the cis-isomers in the stable liquid state. Optical microscope images of the left side, right side, and center of the film, indicating cis-isomers at the liquid phase, solid phase, and the liquid and crystalline solid boundary, respectively. The images on the top corners of (B) and (C) depict the cis-liquid samples (left) before and (right) after the 420-nm light irradiation. Normalized UV-Vis spectra of the uncovered parts for (D) Azo-Cl and (E) Azo-OEt during the 420-nm light irradiation. The arrows indicate the test order. IR thermal images of (F) Azo-Cl and (G) Azo-OEt under the 420-nm light irradiation. The insets show the samples before the light irradiation. The number in the upper right corner of each figure indicates the highest temperature of the materials. Temperature-time curves of (H) Azo-Cl and (I) Azo-OEt during heat release. ΔT represents the temperature difference before and after heat release.

For investigating the heat release performance, 420-nm light was used to irradiate the liquid films of the cis-isomers and the temperature changes were observed using an IR thermal imaging camera in real time. As indicated in Figure 4F, G and Supplementary Figure 5G-I, the samples became a source of heat in comparison to their surroundings while under light irradiation. The liquid films exhibited a rapid temperature increase until reaching the highest temperature, indicating that the stored photon energy and ambient heat were completely released. Afterwards, the sample temperature slowly reduced via dissipating heat into the surrounding environment [Figure 4H, I and Supplementary Figure 5J-L]. The temperature changes of the as-obtained Azos are presented in Table 1. Based on these results, it is clear that photoliquefiable Azos have excellent energy-storage capabilities and controlled on-demand heat release, thereby potentially contributing to ambient heat and solar energy utilization.

Mechanism of photoinduced transition between trans-crystal and cis-liquid

This new energy conversion/storage principle takes advantage of the different melting points of Azo molecules with different configurations, which is a typical PCLT. Figure 5 presents the PCLT mechanism. Generally, the melting point of trans-Azo molecules (trans-T_m) is higher than that of the cis-Azo molecules (cis-T_m). Assuming that the Azo molecules irradiated with UV light are at a temperature (T_phase) between trans-T_mand cis-T_m, the trans-Azo would absorb the photon energy for photoisomerization from trans-crystal to cis-crystal and consume ambient heat spontaneously for phase transition from cis-crystal to cis-liquid. The ambient thermal energy can come from the waste heat from industrial production, the heat produced by refrigeration equipment, and the heat dissipation of electronic devices and vehicle exhausts. Consequently, photon energy and ambient heat are co-stored as N=N bond energy and phase change latent heat, respectively. If needed, the stored energy can be released as high-grade heat energy via cis-liquid-to-trans-liquid isomerization (ΔH_isom) and trans-liquid-to-trans-crystal crystallization (ΔH_cryst). Eventually, a system is formed with ambient heat and photon energy as inputs and high-grade thermal energy as output.

Figure 5. PCLT mechanism of photoliquefiable Azo molecules.

Photoinduced ambient heat conversion

We have proved that photoliquefiable Azo molecules can co-harvest and convert photon energy and ambient heat into chemical bond energy and latent heat in molecules, eventually releasing them as utilizable high-grade heat. This new model of renewable energy utilization is illustrated in Figure 6.

Figure 6. Schematic illustration of photoinduced ambient heat conversion model.

From the perspective of thermophysics, conventional PCMs are temperature-dependent and cannot convert ambient heat to a higher temperature. In contrast, photoliquefiable Azo molecules are light-dependent, which can convert ambient heat between two different phase transition temperatures (the high temperature trans-T_m and the low temperature cis-T_m). In the trans-crystal-to-cis-liquid conversion, the phase transition temperature decreased from trans-T_m to cis-T_m alongside the trans-to-cis photoisomerization. Thus, the crystal-to-liquid phase transition occurred at a temperature below trans-T_m, allowing low-grade ambient heat to be harvested. In the cis-liquid-to-trans-crystal conversion, the cis-to-trans isomerization raised the phase transition temperature back (from cis-T_m to trans-T_m), which enabled the crystallization to occur at a temperature above cis-T_m, thus releasing high-grade heat. In short, Azo molecules can liquefy and store thermal energy at cis-T_m but crystallize and release it at trans-T_m. Because trans-T_m is higher than cis-T_m, the heat output can exceed the input. Under such circumstances, designing Azo molecules with a large T_m difference between the trans and cis forms may be the key to such photothermal energy storage systems.

From the perspective of photochemistry, the essence of this hybrid energy system is the photoinduced ambient heat conversion. Whether it is the charging or discharging process, photon energy is indispensable to converting the difficult-to-use ambient heat into utilizable high-grade heat. During the charging process, the additional photon energy induced configuration change of the Azo molecules, thus harvesting or extracting the low-grade heat from the environment and storing them together. During the discharging process, the photon energy enabled the release of the stored energy as high-grade heat. Compared with conventional heat pumps that convert ambient heat by spending high-grade energy (mechanical work or electricity), such photothermal energy storage systems are especially advantageous in terms of sustainability, environmental friendliness, and economics.