Introduction

Methane (CH4) comprises 70–90% of natural gas and is widely recognized as an essential feedstock in the manufacturing of chemicals1,2,3,4. However, due to the intense bond energy (439 kJ mol−1) and the weak polarization of carbon-hydrogen (C–H) bonds in methane, selective conversion of methane remains a significant challenge5,6. Currently, the steam reforming reaction serves as the primary approach for transforming methane into syngas. Unfortunately, this method requires extreme temperatures (>800 °C) and pressures (>30 bar), thereby hindering its application in sustainable development7,8,9. Therefore, it becomes imperative to explore strategies that enable the conversion of methane to value-added products under mild conditions10,11. Recently, photocatalytic oxidation of CH4 to HCHO or CH3OH under room temperature and mild pressure is considered to be one of the “holy grails” in C1 chemistry, but suffers from the challenge to control the reaction pathway and product selectivity12,13.

During the CH4 photo-oxidation, O2 and CH4 undergo initial activation into free radicals with subsequently combining. It is known that the CH4 oxidation process involves multiple free radicals (•CH3, •OOH, and •OH) to produce a wide variety of products, such as HCHO and CH3OH, as well as overoxidized CO2, while the distribution of final products is determined by the O2 activation and oxygen radical species14,15.

$$\bullet {{{\rm{OOH}}}}({{{\rm{aq}}}})+{{{\rm{\bullet }}}}{{{\rm{C}}}}{{{{\rm{H}}}}}_{3}({{{\rm{aq}}}})\to {{{\rm{C}}}}{{{{\rm{H}}}}}_{3}{{{\rm{OOH}}}}({{{\rm{aq}}}})\to {{{\rm{HCHO}}}}({{{\rm{aq}}}})+{{{{\rm{H}}}}}_{2}{{{\rm{O}}}}({{{\rm{l}}}})$$
(1)
$$\bullet {{{\rm{OH}}}}({{{\rm{aq}}}})+\bullet {{{\rm{C}}}}{{{{\rm{H}}}}}_{3}({{{\rm{aq}}}})\to {{{\rm{C}}}}{{{{\rm{H}}}}}_{3}{{{\rm{OH}}}}({{{\rm{aq}}}})$$
(2)

However, random reaction sites on the surface of traditional photocatalysts lead to the generation of series oxygen radicals, and the overly oxidizing nature of •OH frequently results in multi-step oxidation or even over-oxidation of the product, thus influencing selectivity16,17,18. Hence, strategically managing the O2 activation process to produce •OOH is advisable for achieving a highly selective HCHO synthesis19,20. Depending on the competitiveness of the reaction, constructing highly active and selective reaction sites on the surface of randomly positioned catalysts can significantly break the balance between the main reaction and the competing reactions (Supplementary Fig. 1). If the activity gap between the active site and the random site is sufficiently significant, the competitive equilibrium will continuously shift towards the main reaction, ultimately resulting in the near-complete suppression of the competitive reaction. Under this guidance, manipulating the mono-metal site is expected to achieve selectivity in the activation of O2 to •OOH, benefiting from the end-on O2 adsorption on it to minimize O-O bond breaking (Supplementary Fig. 2). Hence, to achieve high efficiency and selectivity for CH4 photo-oxidation, it is crucial to construct single-atom catalysts with well-defined reaction microenvironments to control the types of active free radicals, which put forward higher requirements for the precise design of photocatalysts.

Metal-organic frameworks (MOFs) constructed from secondary building units of metal clusters and organic linkers provide a tunable platform for the locally precise design of photocatalysts21,22,23,24. Metalizing the hydroxyl groups (μ3-OH) on the [Zr6(μ3-O)4(μ3-OH)4] cluster (Zr6) in Zr-MOF is expected to obtain single-atom sites with clear structures, which allow precisely control of O2 activation process25, thereby achieving highly selective CH4 photo-oxidation. Herein, we support well-defined Cu active sites on deprotonated Zr6 connecting nodes in the UIO66-NH2. The pores in MOFs act as microreactors, where O2 and CH4 can be activated on the Zr6 connecting node and organic linkers, respectively. Crucially, the entirely coordinated Zr in the Zr-MOF leads to a deficiency in O2 adsorption sites, making the introduced mono-copper site serve as the primary effective O2 activation site (Supplementary Fig. 3), significantly expediting the formation of •OOH while impeding the production of •OH. By employing meticulously designed in-situ spectroscopy and theoretical simulations, we illustrate that the mono-copper site significantly enhances the transformation of O2 to •OOH, resulting in nearly 100% selectivity towards HCHO in the photo-oxidation of CH4. In addition, since the reaction is confined in the porous microreactor, the steric confinement of the active species and the ultrafast local mass transfer efficiency greatly enhance the reactivity. In the judicious designed reactive microenvironment, an exceptional time yield of 2.75 mmol gcat−1 h−1 for the production of HCHO from CH4 can be achieved.

Results

Synthesis and characterizations

UIO66-NH2 (UION) is synthesized via a solvothermal reaction between ZrCl4 and 2-Aminoterephthalic acid (H2ATA) in a mixture of acetic acid and DMF (1:10 v/v) at 120 °C for 12 h (Fig. 1a)26,27. The UION is then treated with n-BuLi to deprotonate the μ3-OH at its Zr6 node, followed by metallization with CuCl2 in THF at room temperature to obtain UION-Cu(Cl) with node-supported copper chloride. The deprotonation is confirmed by the disappearance of the vibrational bands of μ3-OH (3679 cm−1) in the Fourier transform infrared (FTIR) spectrum (Supplementary Fig. 4). Subsequently, the coordination environment of the mono-copper site is adjusted from -Cu(Cl) to -Cu(OH), which facilitates the initiation of the O2 reduction reaction (Fig. 1b). The XRD patterns of UION, UION-Cu(Cl) and UION-Cu(OH) are similar to that of pristine UIO as well as simulated UION-Cu(OH) MOF, confirming the retention of UIO-type structure constructed from Zr6 nodes and H2ATA linkers (Supplementary Fig. 5). Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) further confirmed that the modification process will not destroy the morphology and crystal form of UION-Cu(OH), and no agglomerated Cu species can be observed (Supplementary Figs. 6–10). Considering the electron beam sensitive characteristics of MOF materials, the low-dose high-angle annular dark-field-scanning transmission electron microscopy (HAADF-STEM) images of samples have been collected to further character the structure information. Figure 1c illustrates the regularly arranged Zr6 clusters in UION-Cu(OH), which were consistent with the simulated results (Fig. 1d, e). Moreover, the high contrast spots of UION-Cu(OH) match well with the high-resolution structure of UION (Supplementary Fig. 11), implying that Cu species are uniformly distributed on Zr6 clusters. Energy dispersive X-ray spectroscopy (EDX) mapping further indicates the uniform distribution of Cu throughout the UION-Cu(OH) particles (Fig. 1f).

Fig. 1: Synthesis and characterization of UION-Cu(OH).
figure 1

a, b Synthetic scheme of the UION-Cu(OH) catalyst (a) and illustration of the μ3-OH site in UION for Cu loading to form mono-copper hydroxyl site (b). c HAADF-STEM image of UION-Cu(OH), inset: FFT of the image with scale bar of 1/nm. d Simulated HAADF-STEM image of UION-Cu(OH). e Crystal structure of UION-Cu(OH) along [100] direction. f STEM-EDXS elemental maps of UION-Cu(OH).

Full size image

The Cu loading content is determined by inductively coupled plasma mass spectrometry (ICP-MS) to be 3.48 Cu per Zr6 node, corresponding to complete metallation of four μ3-OH sites (Supplementary Table 1). Nitrogen adsorption experiments give a Brunauer-Emmett-Teller surface area of 414.9 m2 g−1 for UION-Cu(OH), which is smaller than that of the unmetallated UION sample (763.7 m2 g−1), as expected (Supplementary Fig. 12 and Table 2). The X-ray photoelectron spectroscopy (XPS) of UION-Cu(Cl) and UION-Cu(OH) displays binding energy peaks for Cu 2p3/2 and 2p1/2 along with satellite peaks (Supplementary Figs. 13 and 14), indicating that the oxidation state of Cu is +222. The +2 oxidation state of Cu species is also assigned by Cu LMM spectra (Supplementary Fig. 15), which shows characteristic peaks at 572.5 eV28. Moreover, electron paramagnetic resonance (EPR) of UION-Cu(OH) exhibits two signals at g = 2.003 and g = 2.320, respectively (Fig. 2a). The Lorentzian EPR signal at g = 2.003 can be assigned to the intrinsic signal of MOF, which is also observed on UIO and UION29. The additional hyperfine peak assigned to the hybridized CuII species is observed at g = 2.320, further demonstrating the highly dispersed copper species (CuII)30,31.

Fig. 2: Fine structure of mono-copper sites.
figure 2

a EPR spectra of UIO, UION UION-Cu(Cl) and UION-Cu(OH). b Cu K-edge XANES experimental spectra of UION-Cu(OH), UION-Cu(Cl), CuO and Cu foil. c WT for the k2-weighted EXAFS signal of Cu foil, CuO,UION-Cu(Cl) and UION-Cu(OH). d Fourier-transformed EXAFS spectra of UION-Cu(OH), UION-Cu(Cl), CuO and Cu foil. e Fitting of the EXAFS data of UION-Cu(OH) based on the model obtained from DFT simulation.

Full size image

The local coordination environments of mono-copper site in prepared samples have been probed by X-ray absorption fine structure (XAFS) spectra at the Cu K edge. The rising edge at 8985.5 eV in the X-ray absorption near edge structure (XANES) is assigned to the 1s→4p transition of CuII (Fig. 2b)32. The k2-weighted Fourier transform EXAFS spectra of UION-Cu(Cl) and UION-Cu(OH) exhibit only one peak at around 1.53 Å, with no Cu–Cu (2.23 Å) or Cu-O-Cu (2.54 Å) observed, suggesting that the Cu species are atomically anchored (Fig. 2d). To achieve enhanced resolution in both R and k spaces, an exhaustive wavelet transform (WT) analysis of Cu K-edge EXAFS oscillations has been conducted. As depicted in Fig. 2c, no discernible extended Cu-Cu and Cu-O-Cu features are evident in either UION-Cu(Cl) or UION-Cu(OH), providing additional confirmation of the atomic dispersion of Cu species (Supplementary Fig. 16). The EXAFS spectra are fitted within the 1.0–3.0 Å range based on the density functional theory (DFT) optimized structure model, extracting coordination details for the first shell. The Cu site in UION-Cu(Cl) is stabilized by three oxygens on the Zr6 cluster and connected to a Cl atom with a Cu-Cl bond length of 2.18 Å (Supplementary Fig. 17 and Supplementary Table 3). In UION-Cu(OH), the Cl atom is substituted by OH, resulting in a four-oxygen coordination environment for Cu. This coordination involves three Cu-O bonds with oxygen atoms at a distance of 1.94 Å and one Cu-O bond with an oxygen atom at a distance of 1.97 Å originating from the -OH group (Fig. 2e and Supplementary Table 3). The preceding results clarify the coordination structure of the mono-copper site and lay the groundwork for investigating the reaction mechanism.

Photocatalytic performance

The photocatalytic performance is assessed through methane conversion in a top-irradiation high-pressure batch reactor, in which 5 mg of photocatalysts are suspended in 20 mL of distilled water in a mixture of 9 bar CH4 and 1 bar O2 and subjected to a 5-h irradiation period at 20 °C (Supplementary Fig. 18). Figure 3a present the oxygenates production including HCHO and CH3OH together with overoxidized CO2 over prepared photocatalysts. UIO exhibits trace yields of C1 oxygenates in the reaction, consistent with its limited light absorption and severe charge recombination. UION and UION-Cu(Cl) demonstrate increased production of C1 oxygenates, yet the resultant products lack selectivity. After adjusting the coordination environment of the mono-copper site into -Cu(OH), there is a substantial enhancement in selectivity, with the target product HCHO reaching almost 100%. The highest HCHO yield is attained at 13.76 mmol g−1 over UION-Cu(OH)-3.48, at which point the Cu site is approaching saturation. This outperforms most state-of-the-art photocatalysts operating at similar conditions or even higher pressure (Fig. 3b)6,16,33,34,35,36. The ultra-high selectivity of the product is further confirmed through NMR 1H and 13C spectra, only HOCH2OH (the major species in aqueous solutions of HCHO) can be detected over UION-Cu(OH) (Fig. 3c, d). It is important to note that due to the uncontrolled self-conversion of CH3OOH to HCHO in the product, we quantified HCHO only after CH3OOH was completely converted to enhance accuracy (Fig. 3c, with no CH3OOH in the 1H NMR spectrum). To detect the unstable intermediate species CH3OOH, the reaction solution was stored at low temperature and subjected to NMR testing as soon as possible. The characteristic signal of CH3OOH was observed at 3.68 ppm, providing clear evidence of CH3OOH formation (Supplementary Fig. 19). In addition, when the reaction time is extended to 10 h (Fig. 3e), the yield of HCHO increased linearly, while the selectivity remained almost unchanged. Under visible light irradiation, UION-Cu(OH) maintains its high efficiency in converting CH4 to HCHO, underscoring its excellent catalytic activity (Supplementary Fig. 20).

Fig. 3: Photocatalytic CH4 conversion performance.
figure 3

a Productivity assessments for oxygenated products obtained over UIO, UION, UION-Cu(Cl), and UION-Cu(OH) as the photocatalysts. b Comparison of the catalytic activity for selective oxidation of CH4 to HCHO over UION-Cu(OH) with other photocatalysts6,16,33,34,35,36. c 1H NMR spectra of the photocatalytic products over UION and UION-Cu(OH). d 13C NMR spectra of the photocatalytic products over UION and UION-Cu(OH). e Time course for HCHO selectivity and product yields over UION-Cu(OH). f Photocatalytic activities under different reaction conditions. g GC-MS spectra of the produced HCHO with isotopically labeled H218O, 18O2, and 13CH4. h AQE values at different monochromatic wavelengths along with the diffuse reflectance spectrum of UION-Cu(OH). i Cycling tests over UION-Cu(OH) with five-cycle run.

Full size image

The catalytic performance under varied conditions is investigated with the absence of catalyst, light, CH4, and O2, respectively (Fig. 3f). In all these cases no C1 oxygenates could be detected, thus confirming that the reaction is a photocatalytic CH4 oxidation process driven by UION-Cu(OH). Remarkably, in the absence of solvent (H2O) within the system, CH4 undergoes over-oxidation to CO2. This underscores the significance of the free radical reaction process in the liquid phase as a crucial factor in preventing over-oxidation. Furthermore, the isotope-labeling experiments are conducted to verify that HCHO is indeed produced via photocatalytic CH4 oxidation. As shown in Fig. 3g, the H13CHO and HCH18O peaks prove that the product is derived from CH4 and O2. The control experiment in D2O demonstrates that the solvent does not participate directly in the reaction, it merely serves as a medium for the generation and transfer of free radicals (Supplementary Fig. 21). Notably, a high apparent quantum yield (AQY) of 3.52% at 375 nm has been achieved on UION-Cu(OH), and the calculated AQY at 400, 420, and 450 nm are 1.26%, 0.83%, and 0.22%, respectively, which resembles well with the UV-vis absorption (Fig. 3h).

To investigate the stability of the optimized photocatalyst, the cycling test experiment is carried out over UION-Cu(OH) photocatalyst (Fig. 3i). No obvious decrease of oxygenate yield and selectivity can be observed under 25 h reaction (five cycles), demonstrating the good stability of UION-Cu(OH). Meanwhile, the XRD, XPS, TEM, and XAFS comparison of the used UION-Cu(OH) is carried out (Supplementary Figs. 22–25). These characterizations of UION-Cu(OH) after reaction remain almost the same as the fresh samples, confirming the stable topology of catalyst. Nitrogen adsorption experiment has confirmed that the pore structure of the catalyst remained unblocked even after undergoing multiple runs (Supplementary Fig. 26 and Table S2). In addition to its stable structure, the used catalyst does not exhibit significant carbon deposition or other adsorbed species, which ensures the activity of the mono-copper site (Supplementary Fig. 27).

Charge excitation and migration properties

The optical absorption ability of prepared samples is first inspected by UV-vis diffuse reflectance spectroscopy (DRS). As shown in Fig. 4a, the absorption in UIO is induced by O to Zr charge transfer in Zr6 inorganic clusters. However, the introduction of amino functional groups initiates an additional charge transfer from the linker to the Zr6 node, corresponding to strong light absorption in the visible region, indicating superb spatial separation of photogenerated carriers between the linker and node27,37. According to the Tauc-plot curves, the mono-copper sites further induce a subtle red shift in the bandgap (Supplementary Fig. 28), implying that these sites optimize the photoexcitation process. To understand the role of mono-copper sites in photocatalytic reaction, the photoelectric properties and kinetic pathways of photocarriers are systematically investigated. The photoluminescence (PL) spectra of the samples are performed to study the separation of photogenerated charge carriers (Supplementary Fig. 29). An intensive PL emission peak at 400-500 nm is detected over the bare UION samples, which is remarkably decreased after the introduction of mono-copper sites, indicating that Cu species favorably prevents the recombination of charge carriers generated on UION MOFs. The time-resolved PL (TRPL) spectra illustrate the dynamic changes in charge transfer induced by the Cu sites. The UION-Cu(OH) sample exhibits the longest fluorescence decay lifetime, corresponding to its optimal photoinduced charge carrier separation efficiency (Supplementary Fig. 30). Furthermore, the photocurrent density increases after the introduction of mono-copper sites, further confirming the enhancement of photoelectron separation and migration (Supplementary Fig. 31b). Similarly, the charge migration improvement is also demonstrated by electrochemical impedance spectroscopy analysis. Nyquist plots (Supplementary Fig. 31a) show a smaller impedance radius of UION-Cu(OH) compared to the bare UION samples, demonstrating the superior separation and transport of photoinduced charges in UION-Cu(OH), which accounts well for its superior photocatalytic activity38.

Fig. 4: Detection of charge dynamics.
figure 4

a Diffuse reflectance UV-vis spectra of UIO, UION, UION-Cu(Cl), and UION-Cu(OH). b, c Transient absorption contour plot of UION (b) and UION-Cu(OH) (c) under 355 nm pump light. d, e Transient absorption decay kinetics of UION (d) and UION-Cu(OH) (e) probed at 625 nm. f In situ XPS spectra of UION-Cu(OH) under dark and light irradiation conditions. g In situ XANES Cu k-edge spectra of UION-Cu(OH) under dark and light irradiation conditions. h Changes in R space of UION-Cu(OH) under dark and light irradiation conditions. i Fitting of the EXAFS data under lighting conditions.

Full size image

To thoroughly clarify the regulation of transient photoelectron transfer kinetics facilitated by mono-copper sites, femtosecond transient absorption (fs-TA) measurements are conducted. As illustrated in Fig. 4b, c, the fs-TA spectra of bare UION and UION-Cu(OH) are presented in the 390–690 nm region. Evidently, the fs-TA spectra of UION and UION-Cu are primarily characterized by a photoinduced absorption (PA) signal centered at 625 nm, which can be ascribed to the secondary light absorption by excited state electrons. The PA signal intensifies as photogenerated electrons accumulate in the conduction band and subsequently diminishes as a result of recombination between photogenerated electrons and holes (Supplementary Fig. 32). Therefore, by fitting the decay process of the PA signal, the kinetic behavior of photogenerated carrier recombination can be obtained. As shown in Fig. 4d, e, the kinetic plots with typical fitting curves illustrate the decay of both UION and UION-Cu(OH) followed a biexponential model, exhibiting one fast component (τ1) with time constants of 17.87 and 23.10 ps, respectively, as well as a slower component (τ2) with a time constant extending to 183.2 and 400.4 ps. The noteworthy increase in both τ1 (charge-carrier trapping) and τ2 (excitonic recombination) lifetime suggests that the mono-copper species can serve as electron storage sites, which can facilitate the extraction of photoexcited electron and delay the recombination of excited state electrons.

The charge transfer from UION to mono-copper sites is proved by in situ XPS spectra (Fig. 4f). Under light illumination, the XPS characteristic peak of Cu moves toward the low binding energy direction accompanied by the weakening of the CuII satellite peak, confirming the role of mono-copper site as an electron acceptor in the photocatalytic processes by UION-Cu(OH). In situ EPR spectra of the UION and UION-Cu(OH) are performed to further explore the details of the photoexcitation and charge transfer mechanisms. Under illumination, the EPR signal of UION and UION-Cu(OH) at g = 2.003 gradually enhanced, suggesting the generation of photoexcited electrons (Supplementary Fig 33)39. By contrast for the UION-Cu(OH) sample, the CuII EPR signal weakened under the light irradiation, indicating that the content of CuII species decreased, as the CuI and Cu0 are EPR silent, proving that photogenerated electrons will move and gather to Cu sites (Supplementary Fig 34)17. The XAFS test under illumination reveals a consistent trend. As illustrated in Fig. 4g, the Cu k-edge XANES spectra of UION-Cu(OH) under illumination shift towards a lower energy direction, suggesting a decrease in the valence state of the Cu site, confirming the transfer of photogenerated electrons to the mono-copper sites. It is worth noting that the peak intensity in the R-space decreases under illumination (Fig. 4h), suggesting a reduction in the coordination number of the mono-copper site, which may originate from the detachment of -OH groups during the photoactivation process. Hence, further fitting of the data under illumination was conducted (Fig. 4i). The results indicate that the Cu species undergo a transition to tri-oxygen coordination after illumination, exposing the activated mono-copper sites. In summary, the mono-copper site accumulates photogenerated electrons and eliminates coordinated hydroxyl groups, creating an electron-rich exposed atomic active site and serving as a platform for selective methane oxidation.

Mechanistic investigations

The core issue regarding the reaction mechanism is the generation of radicals, which involves the band structure of the photocatalyst and the surface reaction sites. We first determined the bandgap and band positions of the UION-Cu(OH) photocatalyst using UV-VIS DRS and XPS valence band spectra (Supplementary Fig. 35). Given that the valence band and conduction band positions of the UION-Cu(OH) photocatalyst are 2.18 eV and −0.65 eV, respectively, the generated photoinduced carriers can drive the activation of CH4 and O2 but cannot convert H2O into •OH6,16,40. Therefore, the activation process of O2 at the reaction sites determines the final product selectivity. Subsequently, a series of in situ characterizations, including in situ XAFS, in situ EPR, in situ TRPL, and in situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS), have been employed to elucidate the evolution of the electronic states at the reaction sites and the surface species during the photocatalytic process.

As shown in Fig. 5a, a shift of the Cu K edge to higher energy levels can be observed as O2 is introduced under dark condition, indicating the increase in the oxidation state of the Cu species, presumably due to the adsorption of O2 on the mono-copper site. Upon commencement of illumination, a reversal in the Cu K-edge absorption energy change was noted, suggesting a reduction in the oxidation state of Cu, potentially attributed to the transfer of photogenerated electrons to mono-copper site and the photoreduction of adsorbed O2. As CH4 is additionally introduced into the reaction system, a discernible shift in the Cu K-edge absorption energy towards higher values is observed, possibly because the consumption of active oxygen species by CH4 thus facilitates the adsorption of O2 onto mono-copper sites. To further elucidate the local coordination structure evolution of the mono-copper active sites during the reaction, the Cu K-edge EXAFS spectra are investigated (Fig. 5b, c and Supplementary Fig. 36). Following the introduction of O2, the coordination bond length of the Cu site expands, suggesting that the adsorption of O2 reduces the charge density of the Cu site and weakens the coordination bond. With the introduction of light and CH4, the bond length gradually recovered, confirming the light-driven reaction between CH4 and reactive oxygen species.

Fig. 5: Investigation of the active sites and reaction mechanism for the photo-oxidation of CH4 to HCHO.
figure 5

a, b In situ XANES Cu k-edge spectra of UION-Cu(OH) photocatalyst treated successively with dark, O2 + dark, O2 + light, and O2 + CH4 + light condition (a) and the corresponding R space spectra (b). c In situ time-resolved Cu K-edge XANES spectra of the UION-Cu(OH) photocatalyst treated successively with dark, O2 + dark, O2 + light, and O2 + CH4 + light condition and the corresponding R space spectra. d, e In situ DRIFTS spectra for CH4 photo-oxidation over UION-Cu(OH) (d) and UION (e). f EPR spectra of •OH radicals trapped by DMPO over UION, UION-Cu(Cl), and UION-Cu(OH) under light irradiation. g EPR spectra of •CH3, •OOH and •OH radicals trapped by DMPO over UION-Cu(OH) under light irradiation. h, i DFT calculation of HOMO (h) and LUMO (i) in UION-Cu, blue and yellow areas represent different spin states. j Reaction profiles for activation of O2 to •OOH or •OH obtained from DFT calculations. k The absolute advantage of single-atom sites in the O2 activation process results in a remarkably selective photo-oxidation of CH4 to HCHO. l Schematic process of charge transfer and photocatalytic CH4 oxidation reaction over UION-Cu(OH).

Full size image

In situ EPR measurements have also been conducted to unveil the charge transfer behavior during the photocatalytic O2 activation over UION-Cu(OH) (Supplementary Fig. 37). The CuII EPR peak displays an evident decrease under light irradiation, and then these EPR signals are partially recovered after the introduction of O2, clearly revealing that the photogenerated electrons first transfer to the mono-copper site and then migrate to O2, giving rise to its subsequent reduction41. Moreover, the excited-sate lifetimes of pure UION-Cu(OH) and those of UION-Cu(OH) in the presence of O2, CH4 or O2 + CH4 are investigated by TRPL measurements (Supplementary Fig. 38). A decrease in the excited-state lifetime of UION-Cu(OH) has been observed with either O2 or CH4, suggesting the involvement of both activation of O2 and CH4 in the reaction mechanism. Notably, the presence of O2 + CH4 decreased the excited-state lifetime further, indicating a rapid reaction between the activated species of O2 and CH442.

Furthermore, in situ DRIFTS measurements have been carried out to investigate the adsorption state of the reactants and reaction intermediates during the reaction process. With the extension of the reaction time, the signals of the reaction intermediate *CH3 (1356 cm−1) and product *HCHO (1738 cm−1) steadily rise43,44, and no distinctive peaks related to CH3OH are detected, which is consistent with the ultra-high HCHO selectivity of UION-Cu(OH) (Fig. 5d). However, a hydroxyl signal possibly attributed to HCHO is observed on UION, corresponding to poor selectivity in the absence of a single copper site (Fig. 5e)45. Clearly, both UION and UION-Cu(OH) exhibit distinct CH4 adsorption and *CH3 intermediate signals46. Given that the linker can enrich photogenerated holes and CH4 (Supplementary Fig. 39), it can reasonably be speculated that CH4 undergoes oxidation to form *CH3 active species through the action of photogenerated holes on the linker. Interestingly, a distinct O2 end-on adsorption signal (1080 cm−1) is observed on UION-Cu(OH), but this phenomenon is not found on UION16, suggesting that the difference in product selectivity may arise from a change in the O2 adsorption state (Supplementary Fig. 40). As distinct modes of O2 adsorption can influence its reduction process, consequently yielding different types of free radicals to affect the selectivity (Supplementary Fig. 41)16,25,47, EPR testing has been conducted to confirm the existence of free radicals during the reaction48,49. The DMPO-OOH signal displays twice the intensity on UION-Cu(OH) compared to UION, implying that the formation of •OOH is more favorable over the mono-copper site (Supplementary Fig. 42). Contrastingly, the DMPO-OH signal of UION-Cu(OH) almost disappeared after the introduction of mono-copper sites (Fig. 5f). Based on the above in situ analysis, it is determined that O2 exhibits a preferential activation to form •OOH at mono-copper sites. The copper site has an absolute advantage in competing for O2, and the rapid O2 conversion causes near-total inhibition of the reaction at random sites, thus promoting the significant tilt of the competitive equilibrium and resulting in highly selective •OOH generation (Fig. 5k). Therefore, the predominant active species are •OOH and •CH3 in the UION-Cu(OH)-mediated photocatalytic CH4 oxidation process (Fig. 5g), resulting in the ultra-high selectivity for the production of HCHO.

Drawing from the aforementioned experimental results, the reaction pathway for the highly selective photocatalytic oxidation of methane to formaldehyde on the engineered catalyst can be unequivocally determined. Initially, the UION-Cu(OH) catalyst is excited when exposed to illumination, leading to the migration of photogenerated electrons from the linker to the mono-copper site. Following that, methane is adsorbed by the amino group and subsequently undergoes oxidation and dehydrogenation facilitated by photogenerated holes, leading to the formation of •CH3 (Supplementary Fig. 43a). Simultaneously, the mono-copper site is activated under light irradiation, leading to the removal of the original -OH and providing O2 adsorption sites. Notably, due to the end-on adsorption of O2 on the mono-copper site, a highly selective reduction occurs, leading to the formation of •OOH (Supplementary Fig. 43b).

Then, DFT calculations have been conducted to substantiate both the experimental findings and the derived reaction mechanism. In both UION and UION-Cu, the highest occupied molecular orbital (HOMO) is situated on the linker (Fig. 5h and Supplementary Fig. 44a), affirming that the amino group-containing linker serves as the primary excitation region for photoelectrons50. However, in contrast to the uniform distribution of the lowest unoccupied molecular orbital (LUMO) across the UION framework (Supplementary Fig. 44b), the LUMO level on UION-Cu is predominantly concentrated on the mono-copper site (Fig. 5i), which strongly indicates that the mono-copper sites function as electron-rich region (Supplementary Fig. 45). Partial density of states calculations provide a more intuitive demonstration of this conclusion, as they reveal that Cu provides the lowest empty orbital just above the Fermi level (Supplementary Fig. 46). To gain further insights into the reaction mechanism at the molecular level, the catalytic processes for methane and oxygen molecule activation were computed. The obtained energy distribution diagrams indicate that the linker with amino groups can facilitate the adsorption of CH4, with an adsorption energy of −0.22 eV. More importantly, the activation barrier for methane activation to •CH3 on this linker is significantly reduced, being only 0.93 eV, thereby substantially enhancing the activation of CH4 (Supplementary Fig. 47). On the other hand, the exposed Cu sites serve as open platforms for O2 adsorption and activation (Supplementary Fig. 48). Calculations indicate that the Cu site significantly promotes the reduction process of O2 to •OOH. Although the protonation of *O2 is exothermic on both UION-Cu and UION, the protonation step from O2 to •OOH on UION-Cu is more thermodynamically and kinetically favorable than on UION, with the Gibbs activation free energies of −1.11 eV and −0.48 eV, respectively (Fig. 5j). Moreover, the •OOH generated at the Cu site needs to overcome a barrier of 0.8 eV to be further reduced to •OH, which directly contributes to the high selectivity of •OOH in the O2 reduction process. As a result, the generated •CH3 and •OOH swiftly react within the reaction microcavity of the MOF, resulting in the high-selective production of HCHO (Fig. 5l).

Discussion

In summary, we have constructed a model of highly reactive mono-copper sites on the secondary structural units of Zr6-MOF to precisely tailor the selectivity and reactivity of CH4 photo-oxidation. The mono-copper sites can efficiently convert O2 into •OOH, maintaining an absolute advantage in competition with low-active random sites, consequently suppressing the formation of •OH. The •OOH and •CH3 formed in the micro-reaction chamber provided by the MOF combine rapidly to promote reaction efficiency while achieving high HCHO selectivity. Fundamentally, substrate adsorption and radical generation during CH4 photo-oxidation have been observed through a series of in situ spectroscopy and assigned to the oxidation-reduction active zones integrated in MOF pores. The operando investigations elucidated the highly selective •OOH generation caused by end-on O2 adsorption at the mono-copper site, thus inducing a highly single CH4 oxidation pathway. As an outcome, the UION-Cu(OH) exhibited near 100% product selectivity for CH4 photo-oxidation to HCHO with a rate of 2.75 mmol gcat−1 h−1. Our study offers guidance for the selective oxidation of CH4 to C1 compounds facilitated by single-atom catalysts and underscores the potential of using MOFs as innovative supports for the design of metal-centered catalysts.

Methods

Chemicals

The following chemicals were purchased and used as-received without further purification. Zirconium tetrachloride (ZrCl4, ≥99.9%, Sigma Aldrich), 2-aminoterephthalic acid (H2ATA, ≥99%, Sigma Aldrich), 1,4-benzenedicarboxylic acid (≥98%, Sigma Aldrich), copper(II) chloride (CuCl2, ≥99%, Sigma Aldrich), Formic acid (HCOOH, reagent grade, CS Pharm Chemical), n-Butyllithium (n-BuLi, 2.5 M solution in hexanes, Sigma Aldrich), sodium triethylborohydride (NaEt3BH, 0.1 M in tetrahydrofuran solution, Sigma Aldrich), dimethylformamide (DMF, HPLC grade, Sigma Aldrich), tetrahydrofuran (THF, HPLC grade, Sigma Aldrich).

Catalyst preparation

Synthesis of UION

UIO-66-NH2 (abbreviated as UION) is synthesized via a solvothermal reaction of H2ATA and ZrCl4 in DMF and water. H2ATA (0.2 mmol, 36 mg) is dissolved in 1.2 mL DMF containing 0.1 mL of HCOOH in a 15 mL glass vial and kept on stirring for 15 min. In a separate glass vial, ZrCl4 (0.2 mmol, 47 mg) is dissolved in 0.2 mL water and transferred to the vial containing a solution of H2ATA. The mixture is kept on heating for 12 h at 120 °C. After cooling to room temperature, the light yellow solid is collected via centrifugation, which is washed with DMF, acetone, and ethanol several times. The as-synthesized UiO-66-NH2 is immersed in the acetone to exchange DMF, followed by drying at 80 °C under vacuum for 12 h. For comparison, UiO-66 (abbreviated as UIO) is synthesized by using 1,4-benzenedicarboxylic acid instead of 2-aminoterephthalic acid, whereas other parameters remain the same.

Synthesis of UION-Cu(Cl)

n-BuLi (50 μL, 2.5 M in hexanes) is added to the slurry of UION (20 mg) in 1 mL THF in a 5 mL glass vial, and the mixture is kept for 1 h at room temperature inside the glovebox. The solid is washed with THF several times to remove the excess n-BuLi. A THF solution of CuCl2 (1 mg/mL) is added to the vial, and the mixture is kept overnight at room temperature. The UION-Cu(Cl) is centrifuged out of the suspension, and followed by washing with THF several times. The sample, UION-Cu(Cl), is further dried at 80 °C under vacuum for 12 h.

Synthesis of UION-Cu(OH)

A weighted amount of UION-Cu(Cl) (20 mg) is added into an anhydrous THF solution (50 mL) of the reducing agent sodium triethylborohydride (NaEt3BH, 0.1 M, 1 mL). The mixture is stirred for 1 h at room temperature, and the obtained suspension of UION-CuH is separated via centrifugation followed by washing with THF several times. UION-CuH is taken outside of the glovebox and kept in water for 30 min to afford UION-Cu(OH). The obtained sample is dried under vacuum at 80 °C for 12 h. The ratio of Cu to Zr in the sample is determined by ICP-MS.

Characterizations

Powder XRD are characterized by a powder X-ray diffraction instrument (Bruker D8 Advanced A25 diffractometer) with a Cu Kα target (λ = 1.54056 Å) at 40 kV and 40 mA. FTIR spectra are obtained on a Nicolet iS10 FTIR spectrometer. SEM is performed on an FEI Teneo VS SEM. TEM images are obtained on a Titan ST microscope from Thermo Fisher Scientific. HAADF-STEM and elemental mapping are performed on a Titan Themis Z microscope. XPS measurements are performed on a Thermo Scientific K-Alpha spectrometer with a monochromatic Al Kα X-ray source. EPR measurements are obtained at room temperature using a Bruker EMX-10/12 EPR spectrometer operated in the X-band frequency. The XAFS spectra (Mn K-edge) are collected at the beamline 1W1B of the Beijing Synchrotron Radiation Facility (BSRF, Beijing) in a fluorescence mode at room temperature. The optical absorption properties of the samples are determined using the diffuse reflection method on a UV-visible light near-infrared spectrometer (Lambda 950). Steady-state PL spectra are recorded on a Carry Eclipse fluorescence spectrometer. N2 adsorption isotherms were operated using a Micromeritics ASAP 2420 at 77 K. CH4 adsorption isotherms were operated using a Micromeritics ASAP 2420 at 273 K. TRPL decay curves are obtained on FLS980 fluorescence spectrometers. In situ DRIFT spectra measurements are performed on a Nicolet 6700 Harrick spectrometer with an MCT detector.

Femtosecond transient absorption measurement

The fs-TA measurements are conducted using a commercial fs-TA system51. A fundamental 800 nm pulse generated by a Coherent Astrella regenerative amplifier served as the pump source. This pulse is used to pump an optical parametric amplifier (Coherent, OperA Solo), producing a frequency-tunable pump beam spanning the visible light region. The pump beam, with a wavelength of 350 nm, is then focused onto the sample. A white-light continuum probe beam is generated by focusing a weaker portion of the fundamental 800 nm beam onto a sapphire window. The sample is positioned at the overlap of the pump beam and the white-light continuum probe beam. All measurements are conducted with samples in water solution using 1 mm quartz cuvettes.

In situ XAFS measurements

The in situ XAFS measurements are collected at the beamline 1W1B of the (BSRF, Beijing) in a fluorescence mode at room temperature. Placing the sample and vacuum-degassed deionized water into the reaction vessel ensures uniform dispersion. Continuously introduce argon gas into the reaction vessel as a protective atmosphere. Collect data separately under dark and illuminated conditions to investigate the activation of mono-copper sites under light exposure. For in-situ testing of methane oxidation, reaction conditions are introduced step by step to study the reaction process. Firstly, subject the sample to thorough illumination in an argon atmosphere to preliminarily activate mono-copper sites. Subsequently, in the absence of light, collect spectroscopic information as the initial state of the reaction. Introduce O2 into the reaction vessel, stabilize, and collect spectroscopic data to obtain material information during O2 adsorption. Resume illumination to gather information during O2 activation. Finally, introduce CH4 to complete the methane oxidation reaction.

In situ DRIFTS measurements

The in situ DRIFTS measurements were conducted on a Nicolet 6700 IR spectrophotometer (Thermo Scientific) equipped with a Harrick Praying Mantis DRIFTS gas cell. After loading the catalyst and assembling the reaction cell, He gas (20 mL/min) was introduced to purge the system for 30 min to remove air. The background was collected after purging. Subsequently, the catalyst was irradiated using a xenon lamp light source, and infrared spectra collection was initiated. Each spectrum was scanned 32 times at a resolution of 4 cm−1. After collecting spectra for 10 min, a mixed gas of CH4 and O2 (CH4/O2 = 1/1, total flow rate of 20 mL/min) was introduced, and data collection continued for 50 min. Throughout the experiment, the system was temperature-controlled at 20 °C using a recirculating water cooling system to prevent potential hazards.

TRPL measurements

The TRPL spectra were conducted on a FLS980 fluorescence spectrometer. Specifically, the catalyst was dispersed in deionized water and sonicated to form a uniform dispersion (1 mg/mL). Then, 1 mL of the dispersion was placed into a cuvette, excited at 380 nm, and decay transients were recorded at an emission wavelength of 610 nm. For TRPL testing with different atmospheres, the corresponding gas was bubbled through the dispersion until saturation before the test, and the gas was continuously purged over the liquid surface during the test. All other testing conditions remained the same.

EPR measurements

The EPR measurements were carried out at X-band on a Bruker EMX-10/12 EPR spectrometer. For the solid EPR tests, the MOF sample was prepared by loading 25 mg into a J. Young quartz tube (outer diameter, 4 mm; inner diameter, 2.8 mm). The gas in the tube was replaced with Ar, and measurements were taken, including EPR spectra under dark conditions and after 2 min and 5 min of illumination. The EPR spectra under an oxygen atmosphere were measured by replacing Ar with O2. Additionally, the radical signals were tested in a dispersion system. In these experiments, DMPO was used as the spin-trapping reagent. As a nitrone spin trap, DMPO is widely employed to capture short-lived radicals by forming more stable radicals, thereby generating a significant EPR signal. Specifically, mix 50 µL of the catalyst dispersion (1 mg/mL) with 10 µL of DMPO solution (10%) in the dark. Introduce the reaction gas (O2 or CH4) into the system, and then collect in-situ EPR spectra under visible light. For the detection of •OOH, methanol was used as the solvent for the system, while deionized water was used as the solvent for •OH and •CH3. All reagents and test tubes were deoxygenated with Ar gas before use.

Computational details

The first-principles calculation of spin polarization based on DFT implemented by CP2K/Quickstep package52 was performed with the plane wave cutoff of 350 Ry. The Gaussian basis set consisting of a double-ζ with one set of polarization functions (DZVP)53 was used to optimize structures. The Perdew-Burke-Ernzerhof exchange-correlation functional54 with the approach of Grimme (DFT-D3)55 was adopted. Due to the large size of the model, single gamma point grid sampling was used. The analysis of the excited state was finished via the Multiwfn code56.