Introduction

Photothermal conversion is a process wherein absorbed photons are efficiently transformed into heat1,2. Since the beginning of chemistry, researchers have noticed the importance of controlling temperature in any chemical process. Over the past years, the idea of utilizing photothermal materials to control temperature and introduce gradients to a reaction’s heat profile has been growing, yet remains largely an unexplored strategy1,2. Various types of nanomaterials exhibit photothermal properties with plasmonic nanostructures (e.g., Au, Ag, etc.), and carbon-based materials being among the most prevalent1,2,3,4,5,6,7,8. Each category of materials has distinct features representing advantages or drawbacks depending on the relevant system’s requirements1. For example, plasmonic structures are highly efficient light-to-heat converters with tunable absorption wavelengths, however, to unlock their potential, complicated nanostructures of precious metals must be synthesized6. In contrast, carbon-based PTAs are far less costly and have broad absorptions across the visible spectrum7. Nevertheless, both cases generally necessitate surface modifications to stabilize in solution and achieve photothermal conversion effectively in driving chemical reactions4,5,7. Molecular compounds can also convert light-to-heat through photoinduced non-radiative thermal relaxation1. There are reports on the photothermal activity of homogeneous solutions with compounds such as organic polymers and synthetic chromophores, however they are restricted to low temperatures due to low photothermal conversion and stability concerns1,9,10,11.

At present, the primary focus of photothermal applications lies within the realm of biomedical fields, with limited exploration into synthetic processes, particularly those demanding elevated temperatures under solvothermal reaction conditions2,6,8,12,13. In this work, we harness the photothermal capabilities of five different converters (cobalt chloride molecular complex, cobalt MOF, molecular iodine, carbon black, and PdNPs) to facilitate the synthesis of MOFs and develop a pathway to their self-assembly. MOFs are characterized by their crystalline structures, comprising of metal ions or clusters linked by organic connectors, resulting in highly porous materials with extensive surface areas14,15. The versatility of MOFs has spurred significant research across diverse fields, including gas storage and separation16,17, drug delivery18, sensing19, catalysis20,21, electronic devices22, and membranes23, making MOF synthesis an area of considerable interest. The photothermal approach was previously shown to enable the formation of intricate assemblies of metallic NPs4, herein we present a highly innovative application of this method to achieve stable self-assemblies of MOFs. The assembly of colloidal MOF particles into micro or nano-superstructures is an effective strategy to obtain advanced materials with improved properties24,25. Nonetheless, effective routes to obtain these stable assemblies remain scarce and relatively unexplored, presenting opportunities for innovation. The few existing methods include assembly by capillary force26,27,28, DNA hybridization29, polymer matrix templating30, electric field31, and more recently liquid bridging32. However, these methods are often impeded with extensive surface functionalization, meticulous design, reaction control, and templating24,33.

In our work we uncover the versatility of the photothermal approach by studying the light-assisted synthesis of cobalt-2,5-dihydroxyterephthalic acid MOF (Co-DHBDC) with different light-to-heat converters at two distinct wavelengths. Initially, the photothermal response of a homogeneous Co complex was utilized at 520 and 660 nm light to initiate the formation of individual MOF rods or their assembly depending on the wavelength. The molecular cobalt chloride complex proved to be a highly efficient and stable PTA enabling remote control over the temperature and activation of the reaction. An in-depth analysis of the assembly mechanism revealed that photothermal conversion, produced by photon-activated molecular vibrations of in-situ-generated MOF rods, likely promotes secondary nucleation and growth near MOF surfaces, contributing to the formation of superstructures. In addition, this hypothesis was tested by introducing a competing PTA in the form of molecular iodine, carbon black, and palladium nanocubes (PdNCs). Finally, the in-situ insertion of PdNCs to MOF superstructures was shown to be possible and is wavelength-specific by utilizing the described photothermal method.

Results and discussion

Photothermal synthesis and assembly of Co-DHBDC

Here, we introduced cobalt chloride as an efficient PTA for obtaining assemblies of Co-DHBDC also known as MOF-74-Co (Fig. 1, Supplementary Fig. 1). Initially, we set out to test whether the photothermal method could be utilized to obtain well defined MOF superstructures as we have previously shown for NPs4. We thus envisioned that if the molecular precursor and resulting MOF had an overlap in absorption with substantial contribution from the MOF, we could utilize a single wavelength to both initiate MOF formation and in-situ produce assemblies. Thus, we searched for a suitable synthesis procedure and came across Co-DHBDC34. Based on the absorption spectra, we identified 520 nm to be the suitable wavelength as we hypothesized, thus we irradiated a solution of cobalt chloride in presence of DHBDC using this wavelength, and were excited to observe that assemblies indeed formed (Fig. 2a-h, Supplementary Figs. 2-4).

Fig. 1: Molecular-photothermal conversion driven synthesis of Co-DHBDC.
figure 1

Scheme showing molecular-photothermal conversion assisted synthesis of Co-DHBDC MOF assembly. Green and red arrows/circles correspond to the photothermal reactions under 520-nm and 660-nm LEDs, respectively.

Full size image
Fig. 2: Photothermal synthesis of assembly of MOF.
figure 2

a–j Characterization of the Co-DHBDC MOF. a–f, TEM images of MOF nanorods obtained through conventional heating (blue outline), a, and heating under photothermal conditions (using 520 nm LED, green outline) at 120 °C for 2 h, d, with corresponding elemental mapping under STEM mode (b-c: conventional heating, scale bars are 500 nm, e, f: photothermal conditions, scale bars are 1 µm) exhibiting cobalt as metal ion, b, e, and oxygen, c, f, from the DHBDC linker. g, SEM image of the photothermally synthesized MOF at lower magnification exhibiting their morphology. Scale bar is 5 µm. h, UV–vis absorption spectra of MOF precursors (cobalt chloride as metal precursor and DHBDC as ligand) respectively in DMF exhibiting strong absorption in the visible region ( ≈ 450-730 nm, maxima at ≈670 nm) of the metal precursor while insignificant absorption of the ligand. UV–vis absorption spectrum of Co-DHBDC exhibiting absorption in visible region that decreases consistently at longer wavelength. Green and red shadows highlight the absorbance at the corresponding wavelengths. i–j PXRD pattern i, and XPS spectra (Co 2p3/2 ≈ 780.8 eV, Co 2p1/2 ≈ 796.6 eV with satellite peaks at ≈785.1 eV and 801.6 eV respectively for Co (II)) j, of the synthesized MOF structures through regular heating (top), and under photothermal conditions (bottom). Source data are provided as a Source Data file.

Full size image

In previous reports, Co-DHBDC synthesized utilizing a conventional heating method was shown to be formed via the coordination of Co2+ with the carboxyl and hydroxyl groups of DHBDC creating CoO6 octahedra connected by organic linkers (Supplementary Fig. 5)34. In addition, Co-DHBDC is known to possess the highest density of metal sites reported to date in MOFs, and high stability due to the fully coordinated oxygen in the carboxyl and hydroxyl of the DHBDC34. The rod Co-DHBDC structures obtained through conventional heating at 120 °C for 2 h, are ≈1200 nm in length, and 140 nm in width (Fig. 2a, Supplementary Fig. 3). Instead, photothermal reaction utilizing a 520 nm LED under the identical condition yielded assembly of individual nanorods into the superstructure with a length of ≈5000 nm, and width of ≈3800 nm (Fig. 2d). Considering a close-pack hexagonal prism and the homogenous morphology of individual rods, the superstructures consist of nearly 1000-1500 nanorods, with a few individual nanorods protruding outward in a hedgehog-like fashion, as clearly observed in TEM tomography (Supplementary Fig. 6, Supplementary Movie 1). Elemental analysis confirms the presence of cobalt and oxygen as the main constituents of the Co-DHBDC structures both under photothermal and conventional heating conditions (Fig. 2b, c, and e, f). SEM images of the obtained MOF assembly reveal uniform morphological feature. Notably, the assemblies form a reasonably stable dispersion in polar solvents and exhibit morphological robustness under centrifugation and ultrasonication, making them suitable for practical applications (Fig. 2g, Supplementary Figs. 4, 7). The UV–vis absorption spectra shows that the metal precursor and Co-DHBDC both absorb throughout the visible region ( ≈ 450-730 nm), but at ≈520 nm, Co-DHBDC has higher absorption while the molecular precursor’s absorption is significantly lower (Fig. 2h). XRD analysis further reveals the characteristic peaks of the crystalline MOF structures obtained through both of the heating methods as mentioned above (Fig. 2i, Supplementary Fig. 8)34. Moreover, XPS indicated no significant change in cobalt oxidation state in the MOF structures obtained as compared to conventional heating method (Fig. 2j)35. The photothermal synthesis and assembly of Co-DHBDC thus indicate a differential heating method employed during the reaction, as compared to conventional heating, leading to a substantially different synthetic outcome. Therefore, a detailed mechanistic investigation is carried out to comprehend the photothermal synthesis, and assembly process.

Finding the photothermal activity of the molecular heater

Initially, to investigate the photothermal mechanism, we first conducted the Co-DHBDC synthesis using a 660 nm LED, which matches the maximum absorption of the molecular precursor (Fig. 2h). As evident from the electron microscope images, the introduction of a 660 nm LED during the Co-DHBDC synthesis at 120 °C for 2 h produced MOF rods measuring ≈2800 nm in length and ≈1200 nm in width (Fig. 3a-c, Supplementary Fig. 9). In contrast, as previously discussed, under the same reaction conditions, a photothermal reaction utilizing a 520 nm LED resulted in the assembly of the MOF (Fig. 2). To investigate further, we started by assessing the photothermal response of the individual components at the relevant wavelengths to identify the active photothermal sources. Thus, we prepared four separate batches of MOF precursors solution, the first two consisted of DMF solutions of cobalt chloride, while the second two were DMF solutions of 2,5-dihydroxyterephthalic acid (DHBDC) with identical concentrations as those used for Co-DHBDC synthesis. The photothermal response was evaluated using these formulations under 520 and 660 nm light irradiation (Fig. 3d–g). Expectedly, heat generated by the ligand solutions were insignificant upon irradiation with either of the green ( ≈ 520 nm) or red LEDs ( ≈ 660 nm) (Fig. 3d, f). In contrast, the metal precursor (CoCl2) solution readily heated up as evident from the temperature profile under green or red-light illumination (Fig. 3d, f). The temperature profiles under green and red-light irradiation demonstrated markedly different heating ability of the cobalt precursor. Under green light, the cobalt chloride solution gradually heated up to a maximum of 140 °C, while under red light, it rapidly heated up, reaching reflux temperature under identical conditions (Fig. 3d, f). To enable a quantifiable comparison, temperature profiles were generated within a set range of 100 to 120 °C (Fig. 3e, g). These profiles revealed that under green light, the cobalt chloride precursor completed ≈65 heating cycles lasting about 61 ± 2 s each (n = 5), whereas under red light, it completed around 97 cycles lasting 25 ± 1 s (n = 7) each during 2 h reactions (Supplementary Figs. 10, 11). These experiments unequivocally establish cobalt chloride as a stable molecular photothermal converter throughout the reaction duration (Supplementary Fig. 12). The photothermal response is due to absorption of red and green light by the cobalt precursor that could be attributed to the formation of a blue-colored tetrahedral cobalt chloride-solvent complex (Fig. 2h). UV–vis absorption spectra demonstrate that this complex exhibited a stronger visible light absorption due to d-d transition of Co (II) which heated up through non-radiative relaxation. Then again, for Co-DHBDC, exhibiting typical broad absorption in visible region due to periodic arrays of linked molecules with discrete molecular absorption modes that decreases consistently at longer wavelength (Fig. 2h). The photothermal response of Co-DHBDC under green and red light was evaluated under similar conditions as for the molecular precursor (Fig. 3h). The temperature profiles illustrated the photothermal conversion ability of Co-DHBDC consistent to the light absorption, and also probed using Raman (Supplementary Fig. 13)36,37,38. Under 660 nm LED, Co-DHBDC gradually heated up to 120 °C, albeit at a slower rate compared to green light irradiation with 520 nm LED, contrary to observations for the molecular precursor. The heating cycles for Co-DHBDC under green light indicated ≈28 cycles lasting about 67 ± 3 s each (n = 11), while under red light, around 8 cycles lasting 363 ± 25 s each (n = 3) were observed during 1 h reactions (Supplementary Figs. 14, 15). These observations under varying light irradiations highlight the photothermal conversion mechanism, with the temperature profiles revealing distinct photothermal responses in the solution. (Fig. 3i). The capability of Co-DHBDC or CoCl2 to undergo photothermal conversion, driven by the wavelength of light used (520 or 660 nm), has facilitated light-assisted orthogonal synthesis, allowing the selective isolation of either rods or superstructures under identical conditions (Figs. 2d–g, 3a–c, Supplementary Fig. 16).

Fig. 3: Photothermal activity of the molecular heater.
figure 3

a, TEM image of MOF rods obtained through heating under photothermal conditions at 120 °C for 2 h (using 660 nm LED), along with b-c, SEM images at high b, and, low c, magnifications exhibiting their morphology. Scale bars are b, 5 µm, and c, 10 µm. d, Photothermal response of the DMF (control), DHBDC, and metal precursor under green light exposure (maximum temperature set at 200 °C for 30 min). e, Photothermal stability test for the metal precursor under green light for 2 h (maximum and minimum temperature set at 120 °C and 100 °C respectively). Similarly, f, photothermal response of the DMF (control), DHBDC, and metal precursor under red light exposure (maximum temperature set at 200 °C for 30 min). g, Photothermal stability test for the metal precursor under red light for 2 h (maximum and minimum temperature set at 120 °C and 100 °C respectively). h, Photothermal activity of the MOF at various wavelength (typically at 520 nm or 660 nm) showing correspondences with the light absorption (Fig. 2h). i, Photothermal response of the molecular heater and the MOF at two different wavelengths employed (under green LED: 520 nm and red LED: 660 nm, with SD represented as error bars). Source data are provided as a Source Data file.

Full size image

Photothermal transformation to Co-DHBDC superstructure

To further investigate the mechanism, we conducted time-dependent analyses of the products formed during the photothermal reactions. Our results show that the light-assisted transformation of Co-DHBDC to superstructures is a complex, multi-step process driven primarily by photothermal effects (Fig. 4, Supplementary Table 1). Under 520 nm LED (8 W cm-2) at 120 °C, rapid nucleation occurs within the first 25–30 min, leading to the formation of large hexagonal structures ( ≈ 5.27 ± 0.12 µm by 4.26 ± 0.32 µm, based on 50 particles) alongside numerous smaller MOF structures (Fig. 4a, Supplementary Fig. 4d, e). Efforts to isolate early nucleation products at less than 20 min were largely unsuccessful. By 45 min of irradiation, a structural reorganization becomes evident from large hexagonal structures to elongated hexagonal rod-like morphologies with higher aspect ratios ( ≈ 4.24 ± 0.31 µm by 2.01 ± 0.19 µm, based on 100 particles) (Fig. 4b). After an hour, tightly packed bundles of smaller, uniform hexagonal MOF-rods emerge (Fig. 4c, d, Supplementary Fig. 16). Two potential mechanisms were considered to explain this transformation: first, light absorption by Co-DHBDC enhances photothermal effects, promoting secondary nucleation and subsequent growth of smaller rods on pre-existing structures to form uniform bundles. Second, light absorption introduces internal stress within larger rods, driving photomechanical effects that split the rods and promote growth along specific crystallographic planes. Control experiments were conducted, where the reaction temperature under 520 nm was reduced to 30 °C, after 45 min at 120 °C where only Co-DHBDC rods formed, and continued under the same light (520 nm, 8 W cm-2) for over an hour (Supplementary Fig. 17). The absence of superstructure formation ruled out light-induced stress or photochemical contributions, indicating the primary role of photothermal conversion. Further experiments confirmed that a photothermal temperature below 110 °C utilizing a 520 nm LED irradiation (8 or 3-5 W cm-2) was insufficient for MOF formation within 2 h, and no MOF formed under conventional heating at 110 °C for the same duration, highlighting the critical role of photothermal temperature in MOF synthesis (Supplementary Fig. 18).

Fig. 4: Photothermal synthesis mechanism.
figure 4

Time-dependent TEM images (green arrow) showing the formation of Co-DHBDC under LED irradiation (8 W cm-2) at 520 nm (green LED) a-d, and 660 nm (red LED) e-h, across various time intervals at 120 °C: 30 min a, 45 min b, 60 min c, 90 min d, 30 min e, 45 min f, 60 min g, and 180 min h. SAED patterns are highlighted in inset b, and f. i-n, Temperature profile of photothermal reactions at 120 °C triggered by sequential LED activation at 520 and 660 nm: i (45 min at 520 nm followed by 120 min at 660 nm), j (60 min at 520 nm followed by 60 min at 660 nm), and reverse LED activation order, starting with 60 min at 660 nm followed by 60 min at 520 nm, is shown in k along with corresponding SEM images of the resulting products (black outline) shown in l (corresponding to i), m (corresponding to j), and n (corresponding to k). Scale bars: 10 µm. The temperature drop in i-k corresponds to the LED switching during the reactions. Source data are provided as a Source Data file.

Full size image

When 660 nm LED irradiation (8 W cm-2) was applied, the process differed significantly. Large hexagonal structures ( ≈ 3.0 ± 0.6 µm by 2.0 ± 0.30 µm, based on 50 particles), apparently composed of bundled rods, appeared within 30-40 min (Fig. 4e). By 45-50 min, less uniform large single-crystal rods ( ≈ 5-6 µm in length × ≈2 µm in width) formed (Fig. 4f). Rod formation continued over 1-3 h, resulting in Co-DHBDC of size ≈1600-2800 nm × 400-1200 nm (Fig. 4f, g, and Supplementary Figs. 16, 18). Heating profiles revealed that reaching 120 °C took ≈7 min under conventional heating, 6.8 min under 520 nm light, and 2 min under 660 nm light. Thus, initial ramp effects were negligible in deriving assembly under 520 nm. Intriguingly, when conventional heating was adjusted to replicate the rapid ramp of photothermal conversion under 660 nm (2 min to 120 °C), hexagonal Co-DHBDC structures formed at 45-50 min, but rod growth remained significantly less pronounced than those obtained under identical reactions but with 660 nm irradiation (Supplementary Fig. 19). Additionally, extended conventional thermal heating (1-5 days at 120 °C) yielded no substantial morphological changes to the Co-DHBDC nanorods (Supplementary Fig. 20), highlighting photothermal conversion’s role in directing Co-DHBDC growth, and assembly. Close morphological similarities between nucleation-growth patterns under 660 nm, conventional thermal heating, and to that of early nucleation of Co-DHBDC under 520 nm light (40-50 min), followed by superstructure formation later observed only under 520 nm light hinting to the hypothesis of a possible two-step mechanism.

Thus, to shed more light on the photothermal reaction mechanism, and due to the versatile options offered by the photothermal conversion, our interest naturally leaned towards photo-switching experiments achieved simply by toggling the light source during reactions. We started by conducting the reaction under 520 nm LED light for 45 min where only Co-DHBDC rods formed, then switched to 660 nm light to continue the reaction for an additional 2 h (Fig. 4i). Interestingly, only rod-like morphology for Co-DHBDC were observed (Fig. 4l). Next, we focused on two additional photo-switching experiments first, photothermal reaction using 520 nm light for 60 min followed by irradiation with 660 nm light to continue the reaction for an additional hour, and vice versa (Fig. 4j, k, Supplementary Figs. 16, 21). In the former scenario, a Co-DHBDC assembly was formed, resembling a sheaf of wheat composed of individual MOF rods, as evidenced by SEM imaging (Fig. 4m, Supplementary Fig. 22). Intriguingly, in the reverse reaction under 660 nm light, the Co-DHBDC rods formed during the initial 60 min were no longer significantly observed after photo-switching to 520 nm light and continuing the reaction for an additional 60 min. Instead, the assembled Co-DHBDC structures now exhibited a well-packed mixed morphology of individual rod-like features of larger sizes (Fig. 4n, Supplementary Fig. 23). These experiments corroborate that the photothermal activity of the MOF structures could be leveraged under specific conditions to control the morphological of the assembly (Fig. 4). Possibly, under photothermal condition, initially, kinetically controlled morphology as hexagonal rod-like structures form as intermediates. These rods under photothermal condition at 520 nm light then undergo substantial heating, possibly initiating structural defects that lead to secondary nucleation and promoting the growth of smaller rods on existing structures (Fig. 4b-d)4,39,40. This process leads to the formation of thermodynamically stable morphologically uniform superstructures (Supplementary Table 2). Raman spectroscopy further corroborated these findings, revealing that vibrational modes of Co-DHBDC were most effectively excited using lasers around 532 nm as compared to 633 nm, consistent to the light absorption and aligning with photothermal conversion efficiency at this wavelength (Supplementary Fig. 13)38. This high heating ability likely drives secondary nucleation and growth including intergrowth and twining in selected crystals near MOF surfaces (Supplementary Figs. 4c, 8), contributing to superstructure formation. The reaction conducted under simultaneous irradiation at 660 nm and 520 nm LED also produced Co-DHBDC superstructures, further supporting the observations described earlier (Fig. 5a-c, Supplementary Fig. 24a-c). Interestingly, similar superstructures were obtained when white light (8 W cm⁻²) was used, reinforcing these findings (Fig. 5d-f, Supplementary Fig. 24d-f). To explore the generality of the molecular-photothermal conversion in synthesis, similar reactions were conducted with other MOF systems, including copper-benzene-1,3,5-tricarboxylic acid (Cu-BTC known as HKUST-1) and iron-fumaric acid (Fe-FA known as MIL-88A) (Supplementary Figs. 25-28). MOF-on-MOF aggregates were formed under specific photothermal conditions, where the light absorption of MOFs induced significant photothermal conversion capabilities, resulting in comparable superstructure formation as observed for Co-DHBDC (Supplementary Figs. 25-28). Definitely, the insights gained utilizing molecular precursors for photothermal reactions in MOF synthesis and assembly could significantly impact multidisciplinary applications, enabling the formation of complex structures otherwise difficult to synthesize. This study highlights the potential of light-assisted photothermal processes as a foundation for developing such structures for advanced applications, and we would like to develop photothermal chemistry further based on the insights gained to such applications24,25,26,27,28,29,30,31,32,33,39.

Fig. 5: SEM images.
figure 5

SEM images of Co-DHBDC superstructures captured at various magnifications: a-c, formed under simultaneous irradiation with 520 nm (8 W cm⁻²) and 660 nm (8 W cm⁻²) LEDs, and d-f, formed under a white light LED (8 W cm⁻²), both at 120 °C for 2 h.

Full size image

Testing the effect of a competing PTA

Following a comprehensive investigation of the photothermal mechanism governing the assembly process, our next objective was to test how the addition of an external PTA would affect the dynamics of the system. To achieve this goal, we opted for PdNPs with an absorption spectrum that closely resembles that of Co-DHBDC, which could form intriguing materials when incorporated into a composite. Thus, we synthesized cubic palladium nanocrystals (PdNCs) measuring between ≈10-12 nm that exhibited absorption across the visible region with an increase towards shorter wavelengths (Fig. 6a, b)41.

Fig. 6: Imparting functionality, and effect of non-molecular heater on assembly of Co-DHBDC.
figure 6

a, TEM image of the palladium nanocubes (blue outline), scale bar is 50 nm, and b, UV–vis absorbance spectra of two different concentrations of palladium. Green and red shadows highlight the absorbance at the corresponding wavelengths. c-d, SEM images exhibiting composite structures of Co-DHBDC obtained heating under photothermal conditions at 120 °C for 2 h utilizing c, 520 nm (green outline), and d, 660 nm LEDs (red outline) respectively in presence of 0.5 OD (measured at 520 nm) of PdNPs. Scale bars are 5 µm. e, Temperature profiles of Co-DHBDC, and two different concentrations of palladium (0.5, and 1 OD measured at 520 nm) heated photothermally in DMF between 100-120 °C utilizing 520 nm LED. f, SEM image exhibiting non-assembled Co-DHBDC NP composite structures obtained heating under photothermal conditions in presence of 1 OD (measured at 520 nm) PdNPs utilizing 520 nm LED (green outline). Scale bar is 5 µm. Source data are provided as a Source Data file.

Full size image

Initially a 0.5 OD (OD at 520 nm) concentration of PdNCs was introduced to the standard photothermal synthesis of Co-DHBDC previously described and irradiated at 520 nm. This resulted in MOF assembly akin to previous observations, with the NPs found to be incorporated to the MOF assemblies in a composite manner (Fig. 6c, Supplementary Figs. 29-32). When utilizing 660 nm light to initiate the same reaction, thick rod-like structures were obtained with palladium predominantly on the surface of the MOF particles (Fig. 6d, Supplementary Figs. 33, 34). Further examination of the photothermal conversion capacity revealed that at a concentration of 0.5 OD PdNCs, the system underwent ≈14 heating cycles (each lasting an average of 189 ± 9 s, n = 4) under 520 nm light irradiation during a 1 h reaction period (Fig. 6e, Supplementary Fig. 35). Increasing the palladium concentration to 1 OD at 520 nm resulted in 32 heating cycles under 520 nm light (each lasting ≈61 ± 3 s, n = 10) bringing its photothermal conversion capacity to be comparable with the Co-DHBDC in the solution (Fig. 6e, Supplementary Fig. 36). Interestingly, in presence of 1 OD of palladium NCs under 520 nm LED, no assemblies of Co-DHBDC were observed, as confirmed by SEM (Fig. 6f, Supplementary Figs. 37-41). These observations demonstrate that under photothermal conditions at 520 nm, the presence of high concentrations of PdNPs suppresses the assembly of Co-DHBDC. A photo-switching experiment also attempted for Co-DHBDC synthesis in presence of 0.5 OD PdNCs with sequential activation using 660 nm and then 520 nm light. However, in this case, a complete uniform assembly of the palladium-Co-DHBDC larger rods formed using 660 nm light, were not isolated after irradiation with 520 nm light (Supplementary Fig. 42). This observation also points to the requisite of optimum size, and surface functionality of the initial MOF structure generated in situ for the assembly process. These experiments not only reinforce the proposed photothermal conversion mechanism but also lead to the synthesis of NP-MOF composites (Fig. 7). These composites have potential applications in diverse disciplines42, and the developed concept could be further leveraged to create a variety of morphologies, impacting the intended applications. In addition, we selected iodine (I2) and carbon black (CB) as a molecular and non-metallic competing PTAs, respectively, for the photothermal synthesis of the Co-DHBDC superstructure, and copper hydroxide as a competing PTA for the synthesis of Cu-BTC superstructure (Supplementary Figs. 43-51). Our study further revealed that the use of various PTAs under specific wavelengths of light, combined with the significant photothermal contributions from in situ-generated MOFs, enables the synthesis of MOF superstructures. This finding points out the versatility of the presented approach under different reaction conditions. While different systems are expected to exhibit unique behaviors under photothermal conditions due to the interplay of multiple PTAs, the insights gained from this work can be further applied to design superstructures, advancing both the field of photothermal chemistry and MOF superstructure synthesis.

Fig. 7: Photothermal synthesis of MOF and NP-functionalized MOF superstructures.
figure 7

Schematic illustration depicting the photothermal synthesis process involving competing PTAs, leading to the formation of Co-DHBDC MOF rods, superstructures, and palladium NP-functionalized MOF rods and superstructures. Red glows represent active PTAs under synthesis conditions.

Full size image

In summary, our current research has facilitated the development of a concept of molecular-photothermal conversion for synthesizing and assembling of MOFs. The light-assisted synthesis utilizes a molecular photothermal complex to achieve a relatively high temperatures (120 °C) for synthesis, and thus has potential to be incorporated into the repertoire of existing photothermal materials for practical applications. The distinctive photothermal conversion of cobalt’s molecular complex by specific light wavelengths (520 nm or 660 nm) results in the formation of Co-DHBDC assemblies or rods depending on activation wavelength. Additionally, we demonstrated the versatility of using photothermal conversion through orthogonal synthesis in operando, which surpasses the capabilities of regular conventional heating methods. The insights gained from this study were leveraged to introduce PdNP functionality to the synthesized structures. A deeper understanding of the dynamics between light-to-heat converters within the reaction enabled control over the assembly of Co-DHBDC simply by adjusting NP concentrations. Naturally, we aim to expand upon this knowledge and leverage the assembled structures, whether functional or non-functional, for various chemical reactions.

Methods

Photothermal synthesis of Co-DHBDC

A stock solution of the reaction mixture was prepared in DMF (dried over molecular sieve) based on previous reported literature with slight modification34, specifically, by dissolving 2,5-dihydroxy-1,4-benzenedicarboxylic acid (DHBDC) (0.14 mmol, 27.1 mg) in 2 mL DMF, polyvinylpyrrolidone (PVP, Mw = 55 kg mol−1) (250 mg) in 2 mL DMF, and CoCl2·6H2O (0.30 mmol, 71.4 mg) in 1 mL DMF separately, followed by sequentially mixing to obtain a 5 mL solution. 1 mL of the reaction mixture was withdrawn in a 4 mL glass vial (N13, LLG screw-neck vial, outer diameter 14.75 mm, outer height 45 mm, clear, flat bottom) for each reaction under conventional regular heating (using a RCT 5 digital IKA plate) or photothermal conditions (using 520 nm, 660 nm, or white light LED; 2.5 cm × 2.5 cm, 8 W cm⁻²) at 120 °C for 1–2 h. The surrounding temperature was room temperature maintained at 25 °C for all the reactions, and none of the MOFs synthesized in any of the reactions whether conducted using conventional heating or under photothermal conditions were stirred (Supplementary Fig. 52). The specifications for the LEDs, reaction vial, and surrounding temperature were consistent across all the reactions tested, unless otherwise specified. The product was isolated via centrifugation at 9000-10,000 × g for 2 min, washed with DMF and methanol at least 3-4 times, and finally dispersed in desired solvents (e.g., methanol, ethanol, DMF etc.) or dried in a vacuum for 1 h to get as powder form for further studies.

Photothermal response and stability

To investigate photothermal response of the molecular heaters under the light irradiation, 0.30 mmol of CoCl2·6H2O or 0.14 mmol of DHBDC were dissolved separately in 5 mL of DMF, followed by withdrawing of 1 mL solution and heated photothermally under 520 nm or 660 nm light for 30 min (maximum temperature was set to 200 °C). Next, for the stability test, either the molecular heaters or purified Co-DHBDC synthesized through conventional heating were put under the photothermal conditions in presence of 520 nm or 660 nm light with minimum and maximum temperature set between 100-120 °C, and the reaction was monitored for 1-2 h.

Imparting NP functionality through orthogonal photothermal synthesis

Palladium nanocubes were synthesized as described in Ref. 41. A stock solution of the reaction mixture for Co-DHBDC was prepared in DMF as described in above section. 0.5-1 mL of the reaction mixture was withdrawn in a 4 mL glass vial for each reaction and mixed with purified palladium nanocubes (50–100 µL from a stock solution of 1–2 mg mL-1 corresponds to 0.5–1 OD measured at 520 nm (Supplementary Fig. 41). The reaction mixture was heated under photothermal conditions using 520 nm, or 660 nm LED (8 W cm⁻²) at 100-120 °C for 2 h. The product was isolated via centrifugation at 9000–10,000 × g for 2 min, washed with DMF and methanol at least 3–4 times, and finally dispersed in desired solvents (e.g., methanol, ethanol, DMF etc.) or dried in a vacuum for 1 h to get as powder form for further studies.

Photothermal synthesis of Cu-BTC

The photothermal synthesis of Cu-BTC (HKUST-1) was performed based on previous reported literature with slight modification43. Briefly, 87.5 mg Cu(NO3)2·3H2O and 15 mg PVP (Mw = 30 kg mol-1) were mixed in 2 mL DMF. To this added, 42 mg benzene-1,3,5-tricarboxylic acid (BTC) dissolved in 1 mL DMF solution. The resulting solution was stirred for 10 min, and heated under photothermal conditions at 80 °C using a 660 nm LED (8 W cm⁻²) for 2 h. The solid MOF was collected after the reaction by centrifugation, washed with fresh DMF for twice and ethanol for three times.

Photothermal synthesis of Fe-FA

Light-induced synthesis of Fe-FA (MIL-88A) was based Ref. 44 with slight modifications. Initially, 4 mmol of FeCl3·H2O and 4 mmol of fumaric acid were dissolved individually in 1 mL of DMF. Then, the two solutions were mixed, heated with a 660 nm LED (8 W cm⁻²) for 4 h at 100 °C. The solid MOF was collected after the reaction by centrifugation, washed with fresh DMF for twice and methanol for three times.

Materials

Cobalt chloride hexahydrate (BioReagent), sodium tetrachloropalladate (II) (98%), iron (III) chloride hexahydrate (FeCl3 · 6H2O, ACS reagent, 97%), polyvinylpyrrolidone (PVP, average Mw 10–55 kg mol-1), potassium chloride ( ≥ 99.0%), iodine flakes (reagentplus ≥99%), and copper(II) hydroxide (technical grade) were purchased from Sigma-Aldrich. N,N-dimethylformamide (99.8%, extra dry over molecular sieve, AcroSeal®) was purchased from Thermo Scientific–ACROS. 2,5-dihydroxyterephthalic acid (98%) and potassium bromide ( ≥ 99%, ACS) were purchased from Strem chemicals. Copper nitrate trihydrate (Cu(NO3)2 · 3H2O, 99% for analysis) was purchased from ThermoScientific. 1,3,5-benzenetricarboxylic acid (98%) was purchased from Thermo Scientific-Alfa Aesar. Fumaric acid ( ≥ 99%) was purchased from Acros Organics. Carbon black powder (EMPEROR® 2000) was purchased from Cabot. All the solvents unless noted otherwise were of analytical reagent grade.

Instruments

A Thermo-Scientific Evolution 220 UV–visible spectrophotometer was used to record the absorbance. A Tecnai T12 G2 TWIN TEM Thermo Fisher Scientific (former FEI) Transmission Electron Microscope was mostly used for general imaging purposes using Gatan CCD MultiScan camera. A Thermo Fisher Scientific (FEI) Talos F200C transmission electron microscope operating at 200 kV was used for both imaging and tomography data accusation. The images were taken with Ceta 16 M CMOS camera. The tilt series were acquired with Thermo Fisher Scientific Tomography software (version 5.9) between angles varying from −60° to 60° with 1° steps. Before each acquisition of an image and before the movement of the stage to the next tilt angle, a cross-correlation of tracking and focusing were made. The post-processing and the stack alignment were done with Thermo Fisher Scientific Inspect3D software (version 4.4). The JEM-2100F, a Field Emission gun Transmission Electron Microscope was employed to measure Energy-dispersive X-ray spectroscopy (EDS) elemental mapping under scanning transmission electron microscope (STEM) mode using Oxford EDS system, and obtain high-resolution images. The post-processing of the images in selected cases was performed using ImageJ software. Scanning electron microscope (SEM) images were taken using a Thermo Fisher Scientific Verios 460 L FEI scanning electron microscope, and using LVEM5 benchtop electron microscope. X-ray diffraction (XRD) was recorded using a Panalytical Empyrean II Diffractometer system-equipped with three position sensitive detectors: X’celerator 1D, 1der (0D and 1D applications), and PIXcel3D detector (with pre-mounted diffracted beam monochromator). The ESCALAB Xi+ Thermo-Fisher Scientific ultrahigh vacuum (1 × 10–9 bar) with Al Kα X-ray source was used to investigate the surface chemical composition, and high-resolution core-level spectra were utilized to probe the corresponding valence states of each component element. All the spectra were charge corrected by using the C 1 s line at 284.6 eV, which appears due to the presence of carbon on the sample surface. The XPS spectra of the samples were further analyzed using the Thermo Scientific AVANTAGE software. Raman measurements were performed with a confocal Horiba LabRam HR Evolution system, equipped with a Syncerity charge-coupled device (CCD) detector, by focusing on a sample film on a silicon wafer substrate using lasers of 532 nm and 633 nm.