Loosening metal nodes in metal-organic frameworks to facilitate the regulation of valence

Abstract

The valence of metal nodes in metal-organic frameworks (MOFs) determines their performance in applications while developing an efficient approach for valence regulation is challenging. Here we present a strategy to make the valence regulation much easier by loosening metal nodes by thermal pretreatment. The typical MOF, HKUST-1, with the tunable valence of Cu nodes, was used as a proof of concept. Thermal pretreatment (producing HK-T) changes the chemical environment and loosens Cu nodes, endowing them with enhanced reducibility. In the subsequent vapor-induced reduction, the yield of Cu⁺ from Cu²⁺ conversion in HK-T (producing HK-T-V) reaches 69%, which is higher than that in pristine HKUST-1 (producing HK-V) with a Cu⁺ yield of 19% as well as the reported yields of target-valence metal nodes in various MOFs (6%–30%). The obtained HK-T-V possessing abundant Cu⁺ sites can capture 0.809 mmol/g thiophene in adsorptive desulfurization, 2.5 times higher than HK-V and superior to most reported adsorbents.

Graphical abstract

1. Introduction

Metal-organic frameworks (MOFs) are a unique class of porous coordination polymers comprising metal nodes connected to organic linkers [1], [2], [3], [4]. Due to their structure variation and high porosity, MOFs have attracted extensive attention in various applications like adsorption [5], [6], [7], [8]. The metal nodes in MOFs can interact with different guest molecules [9], and the performance of MOFs in applications is strongly dependent on the valence of metal nodes [10], [11], [12]. For instance, Cu⁺ sites can form π-complexation with unsaturated compounds (e.g., thiophenic sulfur and olefins) and realize selective capture of these unsaturated compounds. In contrast, Cu²⁺ sites do not have this interaction [13], [14], [15], [16]. Therefore, the valence regulation of metal nodes in MOFs is of great importance for efficient utilization.

Recently, the valence regulation of metal nodes in MOFs has been studied, and the generation of desirable valence was used for gas separation [17,18], liquid-phase adsorption [19], selective oxidation [20], and photocatalysis [21], [22], [23]. Two main techniques have been developed to control the valence of metal nodes in MOFs: direct synthesis and post-modification. The direct synthetic method makes use of a modulator, which can prevent the metal ions from disproportionation during synthesis, to control the valence of metal nodes. With the usage of formic acid as a modulator, 7.6% of Ce³⁺ is formed in UiO-66 [24]. By use of direct synthesis, the amount of metal nodes with desired specific valence is quite limited because the charge unbalances result in difficulty in assembling MOF topology. The post-modification method is generally employed to tailor the valence of metal nodes, including Fe, Cr, and Cu. For instance, 6.5% of Fe²⁺ is formed in MIL-100(Fe) by activation under the atmosphere of helium [17]. By loading KNO₃ on MIL-53(Cr) followed by heat at 300 °C, 9.2% of Cr³⁺ is oxidized to Cr⁶⁺ [25]. Through X-ray irradiation, 12% of Cu⁺ can be generated in HKUST-1 [26]. By reduction using hydroquinone, Cu²⁺ in HKUST-1 can be converted to Cu⁺; however, the reduction proceeds by 30% and self-terminates despite the overwhelming feed of the reducing agent [27]. Although elaborate efforts have been dedicated, the reported approaches suffer from a limited yield of metal nodes with desirable valence. The development of an efficient method to regulate the valence of metal nodes in MOFs has remained a pronounced challenge.

Actually, once a MOF is synthesized, the chemical environment of metal nodes is definite, and the regulation of valence is difficult. If the chemical environment could be adjusted, the reducibility of metal nodes would change, which may facilitate valence regulation. Herein, we demonstrate a strategy, loosening metal nodes of MOFs by facile thermal pretreatment, to make the valence regulation quite easier (Fig. 1). As a proof of concept, HKUST-1 containing Cu metal nodes at a high valence (Cu²⁺) was employed. Thermal pretreatment of HKUST-1 produces HK-T, which adjusts the chemical environment and loosens Cu nodes, endowing them with boosted reducibility. Vapor-induced reduction (VIR) was then utilized to convert Cu²⁺ to Cu⁺, producing HK-V and HK-T-V, respectively, from HKUST-1 and HK-T. Remarkably, Cu⁺ content in HK-T-V can reach 69%, which is much higher than HK-V (19%), as well as the reported yields of target-valence metal nodes in MOFs through various methods (6%–30%). This indicates loosening Cu nodes in MOFs is quite effective for facilitating the regulation of valence. Our results also show that HK-T-V is efficient in adsorptive desulfurization and the thiophene uptake is up to 0.809 mmol/g, which is more than 2.5 times higher than HK-V and superior to most adsorbents reported until now.

Fig. 1.

Schematic illustration of loosening metal nodes to facilitate the reduction of Cu²⁺ in HKUST-1. (a) Loosening Cu nodes in HKUST-1 via thermal pretreatment to form HK-T; (b) conversion of Cu²⁺ to Cu⁺ using VIR, producing HK-V and HK-T-V.

2. Experimental methods

2.1. Chemicals

Copper nitrate trihydrate [Cu(NO₃)₂·3H₂O], N,N-dimethylformamide (DMF), dichloromethane, ethanol, and absolute methanol were purchased from Sinopharm Chemical Reagent Co., Ltd. 1,3,5-Benzenetricarboxylic acid (H₃BTC), thiophene and isooctane were supplied by Aladdin Industrial Co., Ltd. All chemicals were used directly without further treatment. Deionized water was used for all of the experiments.

2.2. Preparation of HKUST-1

HKUST-1 was prepared according to the previous report [27]. Cu(NO₃)₂·3H₂O (2.08 g) was dissolved in H₂O (15 mL). In a separate flask, H₃BTC (1.00 g) was dissolved in the mixed solvent of ethanol (15 mL) and DMF (15 mL). After the Cu(NO₃)₂·3H₂O solution was added to the H₃BTC solution at room temperature, the mixture was kept in the oven at 100 °C for 10 h. After cooling down, HKUST-1 was obtained by washing with DMF and methanol.

2.3. Thermal pretreatment of HKUST-1

The prepared HKUST-1 was transferred to a tube furnace and heated at the target temperature for 2 h under N₂ at a heating rate of 10 °C/min to loosen the metal nodes. Considering the thermal stability of HKUST-1 and the post-treatment which is enough to loosen the nodes in HKUST-1, 270 °C, 275 °C, and 280 °C were selected to treat the MOF. After cooling down, HK-T(n) is obtained (where n is the pretreatment temperature varied from 270 °C to 280 °C).

2.4. Valence regulation of metal nodes in MOFs

The reduction of Cu²⁺ to Cu⁺ was conducted by the VIR strategy. Typically, 100 mg materials (HKUST-1 or HK-T) and 2 mL methanol were put in two separate open vials. Then two vials were placed inside an autoclave and heated at 200 °C for 8 h. After cooling down, the vial with powder was taken out immediately and dried under vacuum before use. The obtained materials were denoted as HK-V and HK-T(n)-V from HKUST-1 and HK-T(n), respectively (where n is the pretreatment temperature varied from 270 °C to 280 °C).

3. Results and discussion

3.1. Thermal treatment of HKUST-1 to loosen metal nodes

To adjust the chemical environment and loosen the metal nodes in HKUST-1, thermal pretreatment under relatively mild conditions was employed so that the structure of MOF can be retained. HKUST-1 was treated under N₂ at the temperature of 270 °C, 275 °C, or 280 °C for 2 h, leading to the formation of HK-T(n), where n denotes the temperature. The structure and properties of HK-T were studied in detail.

The crystalline structure of HK-T was first investigated. SEM images of HK-T(270) and HK-T(275) display octahedral morphology, which is similar to HKUST-1 (Fig. 2a). The diffraction peaks in XRD patterns of HK-T(270) and HK-T(275) are sharp (Fig. 2b), indicating the structural integrity after thermal pretreatment. In the case of HK-T(280), the crystal surface becomes rough in SEM image, and the diffraction peaks in XRD pattern turn weak, which suggests partial degradation of MOF structure with increased pretreatment temperature. Further increasing the pretreatment temperature to 300 °C, the MOF structure is destroyed completely, and metallic Cu and oxide are formed (Fig. S1). Water adsorbed in MOFs could be determined by the relative intensity of two XRD diffraction peaks at 6.7° and 9.5° [28]. After thermal pretreatment, the water in the structure is obviously decreased and the color of MOFs turns dark green (Fig. S2). The TG and DTG results of pristine and pretreated HKUST-1 are shown in Fig. S3. For pristine HKUST-1, free water and crystallization water are removed at around 100 °C and 300 °C, respectively. The decomposition of organic ligands, which signals the collapse of frameworks, causes the steep DTG peak at 330 °C. The DTG peaks appear at 341 °C, 343 °C, and 345 °C for HK-T(270), HK-T(275), and HK-T-(280), respectively. The thermal stability of HK-T is better than that of pristine HKUST-1. These results indicate that thermal pretreatment at suitable temperatures does not damage the MOF structure.

Fig. 2.

Characterization of HKUST-1 before and after thermal pretreatment. (a) SEM images; (b) XRD patterns; (c) Cu-K-edge XANES spectra; XPS spectra for (d) Cu 2p_3/2 and (e) O 1s.

The chemical environment of Cu in MOFs was then investigated by X-ray absorption near-edge structure (XANES), X-ray photoelectron spectra (XPS), and Fourier transform infrared spectroscopy (FTIR) spectra. For pristine HKUST-1, the near-edge peak at 8997 eV, the shoulder at 8985 eV, and the pre-edge peak at 8977 eV are all assigned to Cu²⁺ (Figs. 2c and S4), which are caused by the 1s→4p transition (continuum), 1s→4pz transition (shakedown), and the 1s→3d transition, respectively [29]. All of the HK-T samples show similar edge energies to that of HKUST-1, and hence the estimation of edge positions suggests the presence of Cu²⁺. The extended X-ray absorption fine structure (EXAFS) spectra can be used to further describe the bond information. The Fourier transform (FT) of the EXAFS spectra (Fig. S5) shows that Cu-K-edge EXAFS spectra of HK-T are in accordance with that of pristine HKUST-1. The predominant peak around 1.5 Å corresponds to the first coordination shell of Cu‒O. The unique Cu‒Cu bond of Cu₂ dimeric units results in the typical peak at about 2.1 Å. The effect of thermal pretreatment is further studied with XPS for Cu 2p_3/2 and O 1s (Fig. 2d, e). For HKUST-1, the characteristic peak at 934.2 eV is ascribed to Cu²⁺, accomplished by its characteristic shakeup satellite peaks (937−946 eV). After thermal pretreatment, the binding energy (BE) of Cu 2p_3/2 slightly shifts to higher values. Meanwhile, a negative shift of the O 1s BE is observed as the treatment temperature increases. This is attributed to the change in the chemical environment of Cu centers and is indicative of the partial breakage of Cu−O bonds. As a result, Cu nodes in HKUST-1 are loosened, endowing them with enhanced reducibility. FTIR spectra presented in Fig. S6 reveal more details of the Cu chemical environment. For pristine HKUST-1, the peak around 760 cm⁻¹ is attributed to aromatic ring bending vibration, and the peak located at 1110 cm⁻¹ originates from stretching vibrations of C−O−Cu [30,31]. After thermal pretreatment, their intensity decreases progressively with increasing temperature. For HK-T, a new band at 1560 cm⁻¹ is assignable to the coordination change of carboxylate ligands [32,33]. At lower temperatures, this band becomes intense while turning weak at 280 °C. This suggests partial breakage of C−O−Cu bonds under appropriate temperatures. In addition, the emerging cracks and holes on the surface of HK-T declare the departure of partial building blocks of MOFs (Fig. S7), which agrees with the above results.

Regarding the aforementioned results, it is obvious that the structure of HKUST-1 is preserved after thermal pretreatment at suitable temperatures. Despite the mild pretreatment conditions, the chemical environment of Cu nodes is adjusted due to the cleavage of some coordination bonds and the departure of partial ligands. Hence, Cu nodes in HKUST-1 are loosened, which endows them with enhanced reducibility and makes the valence regulation much easier, as shown below.

3.2. Valence regulation of metal nodes in MOFs

The strategy of VIR was employed to conduct the valence regulation of metal nodes in terms of our previous research [34,35]. MOF and methanol as the reducing agent were deposited in the autoclave without direct contact. With temperature increasing, methanol was evaporated and gradually diffuses into MOF triggering the redox. The regulation of Cu²⁺ to Cu⁺ in HKUST-1 and HK-T(n) leads to the formation of HK-V and HK-T(n)-V, respectively, where n denotes the temperature ranging from 270 °C to 280 °C.

The structure of HK-V and HK-T-V obtained from VIR was first studied. XRD patterns show that the diffraction peaks of HK-V and HK-T-V ascribed to MOF structure are comparable to that of materials before VIR (Fig. 3a), suggesting that the crystalline structure is preserved during VIR. Two new diffraction peaks at 36.4° and 42.3° attributed to Cu₂O emerge on HK-V and HK-T-V, and no peaks ascribed to Cu⁰ appear [36]. This suggests that VIR resulted in the formation of only Cu⁺. For HK-T-V, the characteristic peaks ascribed to Cu₂O become intense with increasing pretreatment temperature. Meanwhile, the blue and dark green particle turns to yellow-green, accompanied by the same octahedral crystal as HKUST-1, which can be observed through digital photos, SEM, and TEM images (Fig. S8, S9). Distinguished from parent HKUST-1 and HK-T, some flocculi are observed on the surface after reduction, and their sizes decrease uniformly with the rise of pretreatment temperature. The flocculi should originate from the surface recrystallization of Cu⁺ species generated during VIR. The elemental mappings suggest the homogenously distributed elements in the sample. The porosity of samples was studied by N₂ adsorption and the isotherms are shown in Figs. 3b and S10. HKUST-1 gives an isotherm of type-I, indicating its microporous character. Upon VIR, the shape of the isotherm for HK-V is quite similar to that of HKUST-1, with the appearance of a small hysteresis loop at high relative pressure, indicative of mesoporous structure. The isotherms of HK-T-V decrease at low relative pressure with appearance of H2 type hysteresis loops at high relative pressure, indicating partial conversion of micropores to mesopores. Fig. S11 shows TG and DTG results of HK-V and HK-T(n)-V. For HK-V, the DTG peak occurs at 285 °C, which is lower than that for HK-T-V with DTG peaks higher than 330 °C. This agrees well with the aforementioned results that thermal pretreatment enhances the thermal stability of the frameworks.

Fig. 3.

Characterization of the materials after VIR. (a) XRD patterns; (b) N₂ adsorption-desorption isotherms; (c) Cu-K-edge XANES spectra; (d) Cu-K-edge EXAFS spectra in R space; XPS spectra for (e) Cu 2p_3/2 and (f) O 1s.

Various methods, including XANES and XPS, were employed to verify the valence regulation of Cu²⁺ to Cu⁺ (Fig. 3c-f). The corresponding normalized Cu K-edge XANES spectra show a new absorption peak appearing at ∼ 8982 eV, indicating the formation of Cu⁺ [27,37]. With an increase in the pretreatment temperature, the peak assigned to Cu⁺ becomes more intense, confirming the enhancement of the reduction of Cu²⁺ nodes in the structure. The FT of EXAFS can indicate more information. After reduction, the intensity of Cu‒Cu distance at about 2.1 Å becomes weaker with the increase of loosened Cu²⁺ nodes. The dominant peak around 1.5 Å shifts to the position of Cu‒O distance in Cu₂O, declaring the reduction of Cu²⁺ to Cu⁺. In addition, XPS results evidently display that the yield of Cu⁺ is improved with the pretreatment. The observed Cu 2p3/2 BE at 932.5 eV corresponds to Cu⁺. HK-V has a tiny Cu⁺ peak; in contrast, HK-T-V shows a progressively intense peak of Cu⁺, which increases with the defection introduced by the pretreatment. For instance, Cu⁺ content in HK-T(280)-V achieves 69% according to XPS analysis, while that in HK-V is only 19%. Such a high yield of 69% is obviously higher than that of metal nodes, including Ce (7.6%) [24], Cr (9.2%) [25], Fe (6.5%) [17], and Cu (12%-30%) [26,27] with target-valence in MOFs through various methods reported until now. XPS of O 1s was also performed to compare the chemical states of oxygen. After VIR, the positive shift of the O 1s BE as the pretreatment temperature increases further confirms the relatively high yield of Cu⁺. Table S1 shows the titration-based quantification of Cu⁺, which agrees well with the XPS and XANES results. The first-principle calculation based on DFT proves that the VIR is available to modulate the Cu²⁺ to Cu⁺ herein, in view of the negative free energy change (‒9.4 kcal/mol) of the main reaction path and its mild energy barrier (33.4 kcal/mol) of the rate-determining step (Fig. S12). Such a mild energy barrier enables the VIR and meanwhile controls the reaction rate. As a result, tuning VIR duration by the hour can clearly distinguish the VIR effects.

These results indicate that the VIR can regulate the valence of Cu in HKUST-1 as well as HK-T, and the structure of MOFs is well maintained in the process of VIR. Interestingly, the valence regulation of HK-T with loosened Cu²⁺ nodes via thermal pretreatment is much easier, and the yield of Cu⁺ in HK-T is apparently higher than that in pristine HKUST-1. The abundant Cu⁺ along with the well-preserved structure, endow the resultant materials with excellent adsorptive desulfurization performance as described below.

3.3. Adsorptive desulfurization performance

Deep desulfurization from transportation fuels is urgent due to the demand for clean energy along with environmental concerns and fuel-cell catalyst poisoning issues. Among various desulfurization technologies, adsorptive desulfurization attracts increasing attention owing to its cost efficiency and flexibility to regenerate. The key to adsorptive desulfurization is to develop efficient adsorbents. Among the most promising candidates, Cu⁺-containing material with the appropriate π-complexation between Cu⁺ active sites and aromatic sulfides attracts attention. The obtained materials were thus applied to capture a typical aromatic sulfide, thiophene, considering the abundant Cu⁺. Their performance was evaluated by utilizing the fixed bed adsorption system at room temperature, as shown in Figs. 4a, b and S13, S14. For pristine HKUST-1, the adsorption capacity is 0.258 mmol/g. After thermal pretreatment, the adsorption capacity of HK-T decreases evidently. In contrast, all Cu⁺-containing samples after VIR exhibit enhanced adsorption capacity. HK-V, with a Cu⁺ content of 1.18 mmol/g, possesses an adsorption capacity of 0.318 mmol/g, which is higher than HKUST-1. To verify the role of reduced metal nodes in adsorptive performance, 1.18 mmol/g Cu²⁺ precursor was introduced to HKUST-1 and reduced to Cu₂O, yielding Cu₂O@HKUS-1 with the same Cu⁺ amount as HK-V. The characteristic peaks of Cu₂O in the XRD pattern of Cu₂O@HKUST-1 (Fig. S15) are more intense than that in HK-V (Fig. 3a), despite the same amount of Cu⁺ content, indicating that majority of Cu⁺ species in HK-V are reduced nodes. Meanwhile, Cu₂O@HKUST-1 can capture 0.283 mmol/g thiophene (Fig. S16), which is higher than pristine HKUST-1 (0.258 mmol/g), but lower than HK-V (0.318 mmol/g). Therefore, both Cu₂O and Cu⁺ nodes are effective in adsorptive desulfurization while the latter performs better. For HK-T(270)-V with Cu⁺ content of 2.13 mmol/g, it can capture 0.558 mmol/g of thiophene. It is noticeable that, in the case of HK-T(275)-V with 2.44 mmol/g Cu⁺, the adsorption capacity reaches 0.809 mmol/g, which is more than 2.5 times higher than HK-V. Although HK-T(280)-V owns the highest Cu⁺ content of 2.85 mmol/g, the adsorption capacity of thiophene is only 0.263 mmol/g. This is because excessive structural degradation results in a significant decline in porosity. These results suggest that the valence of Cu, as well as pore structure, have an impact on the performance of adsorptive desulfurization, and the adsorbent HK-T(275)-V possessing both abundant Cu⁺ and well-preserved pore structure performs best among the materials.

Fig. 4.

Adsorptive desulfurization performance. (a) Breakthrough curves of thiophene with HK-V and HK-T(275)-V; (b) thiophene adsorption capacity of all samples; (c) comparison of thiophene uptake with typical adsorbents; (d) reusability of HK-T(275)-V in thiophene adsorption.

To further identify the adsorption performance of the resultant materials, a range of adsorbents published in the literature are employed for comparison. As shown in Fig. 4c, the typical adsorbent activated carbon (AC) displays a low adsorption capacity of 0.017 mmol/g [38]. Mesoporous silica and porous metal oxide possess relatively low capacities, such as SBA-15 (0.089 mmol/g) [39] and Al₂O₃ (0.066 mmol/g) [40]. Zeolites exchanged with transition metal ions have relatively high capacities, such as the typical AgY (0.34 mmol/g) [41], CeY (0.43 mmol/g) [41] and NiCeY (0.64 mmol/g) [42]. MOFs with unsaturated metal sites exhibit better affinity, and the typical examples MOF-5, UiO-66, and MIL-101(Cr) can capture 0.23, 0.53, and 0.12 mmol/g thiophene, respectively [30,43]. The adsorption of thiophene commonly benefits from the existence of Cu⁺ sites, and thus a series of Cu⁺-modified materials are prepared. In terms of SBA-15 as well as Al₂O₃, after CuCl loading, the respective uptakes of thiophene are 0.16 and 0.13 mmol/g [40]. For the typical zeolite Cu(I)Y, the uptake of thiophene is 0.548 mmol/g [35]. MIL-101(Cr) after CuCl loading exhibits the uptake of 0.37 mmol/g [44]. A comprehensive comparison can be found in Table S2. The present adsorbent HK-T(275)-V has an uptake of 0.809 mmol/g, which is superior to the most reported adsorbents, including porous carbons, metal oxides, zeolites, mesoporous silicas, and MOFs.

In view of the competition for adsorption sites between aromatics and thiophenic sulfurs in the practical desulfurization, the effect of 10 wt% of tert-butyl benzene on adsorption was tested. The material HK-T(275)-V can still adsorb 0.786 mmol/g of thiophene (Fig. S17). The further calculation reveals only a 3% decrease in the capacity of HK-T(275)-V, implying a very limited effect of aromatics on the adsorption capacity of HK-T(275)-V. In addition, the recyclability of the adsorbent was performed (Fig. 4d). After three cycles, HK-T(275)-V exhibits comparable uptake as the fresh adsorbent, indicating the well-maintained adsorption activity.

In light of the results above, it is clear that the emerging Cu⁺ sites from metal nodes in MOFs result in a significant improvement in adsorptive desulfurization performance. The content of Cu⁺ and the structure of MOFs affect the adsorption capacity. The material HK-T(275)-V exhibits the highest thiophene uptake among the materials studied and is superior to a variety of adsorbents reported so far. These results demonstrate the high potential of our materials in practical adsorptive desulfurization.

4. Conclusion

In summary, this work presents an efficient strategy, loosening metal nodes through facile thermal pretreatment, to adjust the chemical environment of Cu nodes in HKUST-1 and endow them with enhanced reducibility. In subsequent regulation of Cu valence using VIR, the HK-T samples with loosened metal nodes show a high Cu⁺ yield of 69%, which is obviously higher than that of pristine HKUST-1 (19%). Furthermore, such a yield is higher than that of metal nodes including Ce, Cr, Fe, and Cu with target valence in different MOFs by using various methods (6%–30%) reported until now. Thanks to the high Cu⁺ content, the present HK-T-V adsorbents show excellent adsorptive desulfurization performance, and the thiophene uptake is superior to the counterpart HK-V and a variety of reported adsorbents. The present strategy may open up a new avenue to create specific active sites, which are unattainable by direct synthesis, in MOFs through valence regulation for various applications.

Declaration of competing interest

The authors declare that they have no conflicts of interest in this work.

Biographies

Yu-Xia Li is a “Brain Pool” fellow at the Institute for Environmental and Climate Technology, Korea Institute of Energy Technology. She received her Ph.D. degree at Nanjing Tech University in 2019 under the supervision of Prof. Xiao-Qin Liu and Prof. Lin-Bing Sun. Her research interest is the development of high-performance nanomaterials and their applications in adsorption and photocatalysis.

Lin-Bing Sun received his Ph.D. from Nanjing University in 2008. After that, he joined the faculty of Nanjing Tech University, where he is currently a professor at the College of Chemical Engineering. He worked as a postdoctoral research associate at Texas A&M University from 2011 to 2012. His current research interests focus on the fabrication of porous functional materials as well as their applications in adsorption and catalysis.