Jiyun
Hu
a,
Hamed
Mehrabi
b,
Yin-Shan
Meng
c,
Maddison
Taylor
a,
Jin-Hui
Zhan
*d,
Qigeng
Yan
e,
Mourad
Benamara
e,
Robert H.
Coridan
a and
Hudson
Beyzavi
*a
aDepartment of Chemistry and Biochemistry, University of Arkansas, Fayetteville, Arkansas 72701, USA. E-mail: beyzavi@uark.edu
bMaterial Science and Engineering Program, University of Arkansas, Fayetteville, Arkansas 72701, USA
cState Key Laboratory of Fine Chemicals, Dalian University of Technology, Dalian, 116024, China
dState Key Laboratory of Multiphase Complex System, Institute of Process Engineering, Chinese Academy of Sciences, Beijing, 100190, China. E-mail: jhzhan@ipe.ac.cn
eInstitute for Nanoscience & Engineering, University of Arkansas, Fayetteville, Arkansas 72701, USA
First published on 11th May 2021
Metalation of covalent organic frameworks (COFs) is a critical strategy to functionalize COFs for advanced applications yet largely relies on the pre-installed specific metal docking sites in the network, such as porphyrin, salen, 2,2′-bipyridine, etc. We show in this study that the imine linkage of simple imine-based COFs, one of the most popular COFs, readily chelate transition metal (Ir in this work) via cyclometalation, which has not been explored before. The iridacycle decorated COF exhibited more than 10-fold efficiency enhancement in (photo)catalytic hydrogen evolution from aqueous formate solution than its molecular counterpart under mild conditions. This work will inspire more functional cyclometallated COFs to be explored beyond catalysis considering the large imine COF library and the rich metallacycle chemistry.
(Photo)catalytic hydrogen evolution reaction (HER) from water splitting is a promising clean energy production technique. COFs have been proved to be potential photocatalysts in promoting HER.44–49 The system is typically comprised of COF photocatalyst, Pt nanoparticle cocatalyst, and a sacrificial electron donor (e.g. ascorbic acid, triethanolamine) in an aqueous solution. Formic acid serves as an alternative high H2 storage density reservoir (4.3 wt%).50,51 The decomposition of HCOOH under basic conditions could release H2 with high purity. However, some iridium-based homogeneous HCOOH dehydrogenation catalysts displayed low stability and were prone to deactivate via nanoparticle formation.52,53 Immobilization of the catalyst on proper solid support is expected to extend the catalyst lifetime.54 Taking all of these considerations together, in this work, we show that cyclometalation at the imine site leads to an iridacycle functionalized imine COF for the first time (Fig. 1C), which exhibits fascinating performance in (photo)catalytic HER from aqueous formate solution.
The successful formation of the iridacycle functionalized COF Py-1P–Ir was confirmed by FT-IR spectroscopy, X-ray photoelectron spectroscopy (XPS), and solid-state NMR characterization, and the retention of the framework crystallinity and porosity was examined by powder X-ray diffraction (PXRD) and N2 adsorption–desorption isotherm analysis, respectively. Py-1P–Ir COF showed strong diffraction peaks at 2θ = 3.74, 5.36, 7.56, 11.40, 23.52°, which are slightly shifted to low 2θ direction (0.04 to 0.18° difference) compared to the diffraction pattern of the parent Py-1P COF (Fig. 2A). This might result from the crystal cell expansion after introducing the bulky Ir organometallic unit. FT-IR analysis revealed a new CN vibration band at 1597 cm−1, corresponding to the metal coordinated imines, in addition to the free ones at 1623 cm−1 for Py-1P–Ir COF and (Fig. 2B). The shifting of CN vibration to lower wavenumber upon coordination to Ir was also observed for N-benzylideneaniline (L1) when forming the model iridacycle complex (L1–Ir, 1626 to 1582 cm−1, Fig. S1†).62 The coordination of imine N to iridium was further proved by XPS analysis (Fig. S2† and 2C). The N 1s XPS spectrum of Py-1P–Ir COF showed two subpeaks with binding energies of 399.0 and 400.0 eV, which were assigned to the free and coordinated imine nitrogen, respectively (Fig. 2C). The increase of N 1s binding energy upon coordination is consistent with that of the model complex L1–Ir (399.2 to 399.8 eV, Fig. S3†). Solid-state cross-polarization/magic angle spinning (CP/MAS) 13C NMR spectroscopy unambiguously confirmed the formation of the iridacycle in Py-1P–Ir COF. Three new signals at 173.7, 89.0, and 7.6 ppm appeared in the CP/MAS 13C NMR spectrum of Py-1P–Ir COF (Fig. 2D). The broad peak at 173.7 ppm is assigned to the iridium bonded carbon and the imine carbon of the iridacycle, while the peaks at 89.0 and 7.6 ppm originate from the aromatic and methyl carbons of Cp* ring, respectively, which matches well with the 13C NMR spectrum of the model complex L1–Ir (Fig. S4†). Meanwhile, the relative intensity of the free imine carbon peak at 157.0 ppm of Py-1P–Ir COF decreased accordingly compared to that of Py-1P COF. N2 adsorption experiment was carried out to investigate the porosity of Py-1P–Ir COF (Fig. 2E). The Brunauer–Emmett–Teller (BET) surface area of Py-1P–Ir COF is 972 m2 g−1, which is lower than 1960 m2 g−1 of Py-1P COF (Fig. S5 and S6†). The pore size is decreased to 1.57 nm for Py-1P–Ir COF, compared to 2.19 nm for Py-1P COF, which is expected for the pore wall metalation (Fig. 2F). The iridium loading was determined to be 11.8 wt% by inductively coupled plasma mass spectrometry (ICP-MS), corresponding to ca. 20% imine metalation. Scanning electron microscope analysis revealed that Py-1P–Ir COF was composed of aggregates of nanometer-sized particles, similar to that of Py-1P COF (Fig. S7†). Energy dispersive X-ray (EDX) analysis revealed the homogeneous distributions of Ir and Cl elements over the framework, and the Ir/Cl ratio is close to 1/1, as expected for the proposed structure (Fig. S8†). Besides, the formation of the iridacycle was also successfully performed on an azine linked COF (see ESI† for details), demonstrating the generality of the cyclometalation modification of imine-based COFs.
Fig. 3 Catalytic hydrogen evolution from aqueous formate solution. The influence of catalyst (A), temperature (B), pH (C), and formate concentration (D) on the reaction outcome (H2 quantity and purity are presented in orange column and grey line/square symbol form, respectively. Purity refers to the H2 composition in the generated H2/CO2 mixture). Standard reaction condition: the reactions were carried out by heating a 10 mL HCOONa solution (1.0 M, pH = 7.52) containing 6.35 μmol catalyst (based on Ir) at 65 °C for 6 h. The pH effect was studied at room temperature. The concentration effect was studied at 65 °C and the pHs of HCOONa solutions were not adjusted. See ESI† for details. |
As shown in Fig. 3B, Py-1P–Ir COF displayed higher efficiency at higher temperature. An amount of 250.0 μmol H2 was obtained at 85 °C, which is ca. 2.6 times of that obtained at 65 °C. The high reactivity of Py-1P–Ir COF was manifested that 16.5 μmol H2 was formed even at 25 °C. In all the tested temperatures, the H2 purity was no less than 90%. Lowering the pH was found to favor the reaction that 48.9 and 213.8 μmol H2 were afforded at pH = 5.92, 3.72 respectively, while further lowering the pH to 1.94 did not lead to higher H2 production (200.2 μmol, Fig. 3C). The increased H2 production came with compromised H2 purity to ca. 70% under acidic conditions. The concentration of HCOONa on the HER showed a volcano-type effect (Fig. 3D). The H2 yield increased to 163.2 μmol from the reaction of 2 M HCOONa but started to fall back to 137.5 μmol for 5 M HCOONa and 92.6 μmol for 5 M HCOONa respectively thereafter. The H2 purities were all higher than 90% from the tested four HCOONa concentrations. The decreased reaction efficiency might be due to the decreased concentration of water, which is the proton source for H2 production. Py-1P–Ir COF exhibited excellent stability under all the tested conditions. All the recovered COF samples preserved the crystallinity as shown by PXRD analysis (Fig. S20–S22†). Interestingly, FT-IR and XPS indicate that the imine linkage was partially reduced (Fig. S20–S23†). The imine bond reduction is likely mediated by an iridium hydride intermediate via an outer-sphere process.64–66 The involvement of Ir in the imine reduction is supported by the fact that no imine reduction was observed in Py-1P COF during HER catalysis (Fig. S24†) and the catalytic reduction of N-benzylideneaniline as an exogenous substrate in the presence of Py-1P–Ir COF was observed (Fig. S25†). ICP-MS showed that the Ir concentrations in the reaction filtrates were all below 60 ppb (Table S2†), demonstrating the heterogeneous catalysis nature of the reaction. The high stability of Py-1P–Ir COF allowed it to be recycled for at least another four runs (Fig. 4). Interestingly, an initial performance improvement was observed in the second cycle. This improved reactivity of Py-1P–Ir COF is attributed to the presence of more reactive Ir species in recovered material, in which the Cl ligand has been replaced (see mechanism discussion below). This is supported by the fact that no Cl element was detected by EDX in Py-1P–Ir COF after one HER cycle (Fig. S26†). The average H2 production rate during the five cycles was calculated to be 4626 μmol g−1 h−1, corresponding to a TOF of 7.3 h−1. The recovered Py-1P–Ir COF after five cycles was still crystalline (Fig. S28†). No metal nanoparticle formation was observed by XPS (Fig. S29†) and TEM (Fig. S30†) analysis.
Fig. 4 Recyclability of Py-1P–Ir COF. The reactions were carried out in 1 M HCOONa at 85 °C for 6 hours. |
Pyrene-based COFs have exhibited excellent photophysical properties which benefit photocatalysis.67–70 Diffuse reflectance UV-vis absorption spectrum of Py-1P–Ir COF displayed prominent absorption in 200–500 nm and a long tail to near-IR range (500–800 nm, Fig. 5). The incorporation of iridium complex into Py-1P COF did not change the absorption in 200–500 nm band but strongly improved the absorption ability in the 500–800 nm region. Encouraged by the excellent light absorption ability across the whole UV-visible region of Py-1P–Ir COF, we further tested its photocatalytic HER performance from formate decomposition. The reactions were carried out in 1 M formate solution in the presence of 6.35 μmol catalyst (based on Ir) under 460 nm light irradiation. As shown in Table 1, only 0.7 μmol H2 was detected in the absence of catalyst (entry 1). Py-1P–Ir COF catalyzed the generation of 84.5 μmol H2 in high purity (93%) in 6 hours (entry 3), which is ca. 34 times higher than the molecular counterpart L1–Ir did (entry 2). The corresponding H2 generation rate of 1358 μmol g−1 h−1 (TOF 2.2 h−1) is comparable to the typical COF photocatalyst/Pt cocatalyst/sacrificial electron donor system.71Py-1P COF failed to exert any catalytic effect showing the critical role of iridium for the reaction (entry 5, 0.6 μmol H2). Under the dark condition, Py-1P–Ir COF produced 31.6 μmol H2 in 92% purity, proving it to be a photocatalytic process (entry 4). A physical mixture of Py-1P COF and L1–Ir displayed lower H2 production (19.5 μmol) and poorer selectivity (71% purity), highlighting the importance of covalent hybridization of Ir catalytic centers within the COF (entry 6).72 The recovered Py-1P–Ir COF almost retained its crystalline structure as evidenced by PXRD analysis (Fig. S31†). Similarly, reduction of the imine bond was observed (Fig. S31†). ICP-MS analysis of the Py-1P–Ir COF catalyzed reaction solution revealed a low Ir concentration of 0.99 ppm, corresponding to 0.8% of the total Ir in Py-1P–Ir COF. Besides, the reaction filtrate only produced 2.0 μmol H2 under the photocatalysis conditions (entry 7), demonstrating the heterogeneous catalytic process by Py-1P–Ir COF.
Entry | Catalyst | H2 (μmol) | Purity (%) |
---|---|---|---|
a Reaction conditions: 10 mL formate solution (1.0 M) containing 6.35 μmol catalyst was irradiated with 460 nm LED light for 6 h. b Without light at 45 °C (the photothermal effect raised the reaction temperature to ca. 42 °C). c Physical mixture of Py-1P COF and L1–Ir with the same Ir and total mass loading as Py-1P–Ir COF. d Reaction filtrate of entry 3. | |||
1 | — | 0.7 | 10 |
2 | L1–Ir | 2.5 | 82 |
3 | Py-1P–Ir | 84.5 | 93 |
4b | Py-1P–Ir | 31.6 | 92 |
5 | Py-1P | 0.6 | 68 |
6c | Py-1P/L1–Ir | 19.5 | 71 |
7d | — | 2.0 | 58 |
Based on literature reports,52,53 a proposed mechanism is shown in Fig. 6A. The catalytic cycle is initiated by Cl ligand replacement of IM-1 with formate to give HCOO− coordinated intermediate IM-2. CO2 extrusion from IM-2 affords the key iridium hydride species IM-3. Then two pathways account for hydrogen production depending on the reaction conditions (pH). Under neutral or basic conditions, IM-3 reacts with one molecular H2O to generate H2 and OH− bonded Ir intermediate IM-4, which could be reconverted to IM-1 or IM-2via ligand exchange to close the catalytic cycle. While under acidic conditions, IM-3 reacts with HCOOH to release H2 and regenerate IM-2. The involvement of iridium hydride intermediate was supported by NMR studies on L1–Ir in solution. Mixing L1–Ir and excess HCOONa in CD3CN/D2O (v/v, 4/1) generated an iridium hydride complex showing a hydride signal at −12.45 ppm in the 1H NMR spectrum (Fig. S32†).52,73 To further support the proposed mechanism, we carried out DFT calculations. The computed Gibbs free energy profile of the prosed reaction pathway is shown in Fig. 6B. IM-2 is slightly lower (−1.56 kcal mol−1) in energy than IM-1, suggesting the first step ligand exchange is feasible. The transformation of IM-2 to IM-3via CO2 extrusion is thermodynamically favored (ΔG = 1.88 kcal mol−1) with a small energy barrier of 21.85 kcal mol−1 (TS1). This is consistent with our experimental findings that the iridium hydride formed immediately at room temperature. The reaction of IM-3 with H2O to give IM-4 and H2 is an energy demanding process (ΔG = 18.07 kcal mol−1) with an energy barrier of 50.81 kcal mol−1 (TS2). In contrast, the reaction of IM-3 with HCOOH to give IM-2 and H2 is energetically less demanding (ΔG = 2.11 kcal mol−1) and the activation energy is also significantly reduced to 28.68 kcal mol−1 (TS2′). These results match well with the observed higher reactivity of the catalyst in lower pH conditions.
Fig. 6 (A) Proposed mechanism for Py-1P–Ir COF catalyzed HER from formate; (B) calculated Gibbs free energy profile. |
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/d1sc01692j |
This journal is © The Royal Society of Chemistry 2021 |