Bulk-to-Surface Proton-Coupled Electron Transfer Reactivity of the Metal−Organic Framework MIL-125

Abstract

Stoichiometric proton-coupled electron transfer (PCET) reactions of the metal−organic framework (MOF) MIL-125, Ti₈O₈(OH)₄(bdc)₆ (bdc = terephthalate), are described. In the presence of UV light and 2-propanol, MIL-125 was photoreduced to a maximum of 2(e⁻/H⁺) per Ti₈ node. This stoichiometry was shown by subsequent titration of the photoreduced material with the 2,4,6-tri-tert-butylphenoxyl radical. This reaction occurred by PCET to give the corresponding phenol and the original, oxidized MOF. The high level of charging, and the independence of charging amount with particle size of the MOF samples, shows that the MOF was photocharged throughout the bulk and not only at the surface. NMR studies showed that the product phenol is too large to fit in the pores, so the phenoxyl reaction must have occurred at the surface. Attempts to oxidize photoreduced MIL-125 with pure electron acceptors resulted in multiple products, underscoring the importance of removing e⁻ and H⁺ together. Our results require that the e⁻ and H⁺ stored within the MOF architecture must both be mobile to transfer to the surface for reaction. Analogous studies on the soluble cluster Ti₈O₈(OOC^tBu)₁₆ support the notion that reduction occurs at the Ti₈ MOF nodes and furthermore that this reduction occurs via e⁻/H⁺ (H-atom) equivalents. The soluble cluster also suggests degradation pathways for the MOFs under extended irradiation. The methods described are a facile characterization technique to study redox-active materials and should be broadly applicable to, for example, porous materials like MOFs.

Graphical Abstract:

■INTRODUCTION

Metal−organic frameworks (MOFs) are emerging as promising materials for facilitating redox reactions, including multi-e⁻/multi-H⁺ transformations. Recent studies highlight the ability of the organic linkers or metal ions in or on the MOF nodes to undergo 1−e⁻ oxidations,^1–4 for the MOFs themselves to serve as conductive materials^5–7 or semi-conductors,^8,9 and for the MOFs to facilitate redox reactions that are pertinent to fuel cells and energy.^10–12 In this latter context, the development of MOF photocatalysts that mimic the reactivity of bulk TiO₂ is of timely interest.^13–19

Common photocatalytic reactions such as water splitting or CO₂ reduction are fundamentally proton-coupled electron transfer (PCET) processes. Therefore, tuning the PCET properties of the catalyst is of the utmost importance.^20–22 In this context, several groups have developed MOFs that show PCET behavior resulting in good photo- and electro-chemical activities.^22–24 PCET has been suggested to occur on the linker,²⁴ the node,²⁵ and the linker and node combined^22,23 but has not been studied in well-known photoactive MOFs. In this work, we provide an in-depth analysis of the PCET behavior in the widely used MIL-125.

The Ti-based MOFs MIL-125¹⁶ (Ti₈O₈(OH)₄(bdc)₆) and NH₂-MIL-125¹³ (Ti₈O₈(OH)₄(bdc-NH₂)₆) (bdc-NH₂ = 2- aminoterephthalate) have planar Ti₈ nodes with oxide and hydroxide ligands, linked with terephthalate (bdc) groups. These materials have been shown to be photoactive; for instance, UV irradiation of a slurry of MIL-125 in benzyl alcohol formed benzaldehyde with a color change to blue.¹⁶ It was suggested that this process reduces each MOF node (Ti₈ cluster) by a net H-atom (eq 1). Though the stoichiometry of the reduction was not established, the proposed eq 1 led us to hypothesize that the reduced MOF should be able to facilitate well-defined PCET reactions.

$$\begin{array}{l}\text{T}{\text{i}}_8{\text{O}}_8{\left(\text{OH}\right)}_{\text{4}}{\left(\text{bdc}\right)}_{\text{6}}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\text{+}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\text{0}.5\hspace{0.17em}\hspace{0.17em}\text{PhC}{\text{H}}_2\text{OH}\hspace{0.17em}\stackrel{\text{hv}}{\to }\\ \text{T}{\text{i}}_7\text{T}{\text{i}}^{\text{|||}}{\text{O}}_7\left(\text{OH}\right){\left(\text{OH}\right)}_{\text{4}}{\left(\text{bdc}\right)}_{\text{6}}\hspace{0.17em}\hspace{0.17em}\text{+}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}0\text{.5}\hspace{0.17em}\hspace{0.17em}\text{PhCHO}\end{array}$$ (1)

NH₂-MIL-125 can be reduced by visible light in the presence of a sacrificial reductant, which is highly desirable for photocatalysts.¹³ In the presence of triethanolamine (TEOA), visible-light irradiation of NH₂-MIL-125 results in the catalytic reduction of nitrobenzene to aniline,¹⁵ CO₂ to formate,¹³ and protons (aq) to H₂.^14,15,26

Herein, we describe fundamental studies of MIL-125 that address basic mechanistic questions associated with MOF photocatalysis, such as whether the redox chemistry takes place only at the surface of the MOF particles or occurs throughout the pores. Specifically, we develop a solution NMR assay to determine the reduction stoichiometry, the number of e⁻ + H⁺ (H-atoms) transferred per node. We show by judicious choice of substrates that MIL-125 can undergo well-defined PCET reactions. The results show that irradiation causes reduction throughout the MOF, but all of the (e⁻/H⁺) can be removed through PCET reactions that occur only at the surface. This work thus develops our understanding of the redox capabilities in MOFs, which should advance the promising area of using MOF materials to facilitate multi-e⁻/multi-H⁺ transformations.

■ RESULTS AND DISCUSSION

I. PCET Chemistry of the MOF MIL-125.

Irradiation of MIL-125 in neat iPrOH results in a color change from white to dark blue, indicating reduction of the MOF. The EPR spectrum of this material is similar to that reported for photoreduction with benzyl alcohol and exhibits overlapping broad EPR resonances at g = ca. 1.94 (see the SI).¹⁶ Quantification of the radical yield has not been attempted due to the nontrivial nature of the signal. The reduced MOF can be isolated and stored as a solid in the glovebox for days with little decay in the absence of oxygen. Resuspension of the reduced material in C₆D₆ and addition of 2,4,6-tri-tert-butylphenoxyl radical (^tBu₃ArO^•) or the nitroxyl radical 2,2,6,6-tetramethyl-1-piperidinyloxy (TEMPO) resulted in a color change back to white over ∼30 min, indicative of oxidation of the reduced MOF. ¹H NMR spectra of ^tBu3ArO^• reactions showed exclusive formation of ^tBu₃ArOH, indicative of a net PCET reaction (H-atom transfer reaction) (eq 2). No ^tBu₃ArO⁻ was observed, which would have suggested pure electron transfer (ET) from the reduced MOF. Thus, the photoreduced MIL-125 is capable of facilitating PCET.

$$\begin{gathered} \text{T}{\text{i}}_{8-x}\text{T}{\text{i}}_x^{\text{ІІІ}}{\text{O}}_{8-x}{\left(\text{OH}\right)}_{4+x}{\left(\text{bdc}\right)}_6\hspace{0.17em}\hspace{0.17em}+\hspace{0.17em}\hspace{0.17em}{x}^t\text{B}{\text{u}}_3\text{Ar}{\text{O}}^{•}\hspace{0.17em} \\ \to \hspace{0.17em}\hspace{0.17em}\text{T}{\text{i}}_8{\text{O}}_8{\left(\text{OH}\right)}_{\text{4}}{\left(\text{bdc}\right)}_6\hspace{0.17em}\hspace{0.17em}+\hspace{0.17em}\hspace{0.17em}{x}^t\text{B}{\text{u}}_3\text{ArOH} \end{gathered}$$ (2)

To determine the stoichiometry and extent of reduction, a simple solution ¹H NMR method was developed to quantify the e⁻/H⁺ equivalents transferred to MIL-125. This method uses ¹H NMR quantification of a redox probe relative to an internal standard. After photoreduction, the MOF was isolated as a solid in the glovebox. A known mass was resuspended in C₆D₆, and a calibrated C₆D₆ solution of ^tBu₃ArO^• and di-p-tolyl-ether (Ar₂O) was added. The suspension was stirred for at least 3.5 h to ensure complete oxidation of the MOF. ¹H integration of ^tBu₃ArOH vs the Ar₂O standard provided quantification of the reducing equivalents transferred (Figure 1).

Figure 1.

(Top) Scheme showing the photoreduction of MIL-125 and subsequent oxidation. For simplicity, only one Ti8 cluster is shown (the structure of MIL-125 is depicted within the blue circles). For maximum photoreduction, n ≅ 2. (Bottom) 1H NMR spectrum of the subsequent reaction with ^tBu₃ArO^• (C₆D₆). The colored dots identify ^tBu₃ArOH (blue), Ar₂O (internal standard, red), and C₆D₅H, THF, and ⁱPrOH (green).

For accurate NMR integration, it is essential that neither the internal standard nor the redox probe bind to or be taken up in the MOF, as this would diminish the observed concentration. Addition of MIL-125 or photoreduced MIL-125 to NMR samples containing ^tBu₃ArOH and Ar₂O does not change the relative integration of the two species, indicating that neither was adsorbed (see SI). Given the similarity in size and charge between ^tBu₃ArO^• and ^tBu₃ArOH, it is very likely that the radical oxidant likewise is not taken up into the pores. By contrast, TEMPOH was shown to be taken up by MIL-125, precluding the use of TEMPO^• as a redox probe (see the SI).

Several suspensions of MIL-125 in ⁱPrOH were irradiated for various amounts of time and analyzed by titration with ^tBu₃ArO^•. A plot of H-atom equivalents transferred per Ti₈ node versus photolysis time (Figure 2) shows that increasing the photolysis time increases the extent of reduction. Since EPR spectra indicate that reduction occurs at Ti,¹⁶ we report the number of reducing equivalents relative to the Ti₈ nodes (the moles of Ti₈ nodes was determined from the mass of the MOF and the moles of H-atoms transferred from the NMR method). The increased reduction with time was also evident by eye, as the blue color of the MOF suspension became darker with longer irradiation time. After 5 h, each Ti₈ node is reduced by at least 0.4 H-atoms. At long irradiation times, MIL-125 can be reduced by up to 2 H-atom equivalents per Ti₈ cluster.

Figure 2.

Number of reducing equivalents per Ti₈ node vs time for three different photoreductions of MIL-125 using a 200 W Hg/Xe lamp. Errors for data points are estimated to be ≤±0.1 equiv of e⁻/ H⁺, based on the error of NMR analysis by integration also taking into account error propagation from the balance accuracy and the estimated maximal sample loss. The sample in trial #2 (blue circles) became black by 17.5 h of photolysis and the color persisted after addition of excess ^tBu₃ArO^•, indicating decomposition.

The variability in extent of reduction between the three experiments was attributed to differences in the irradiation conditions, such as distance from lamp, stirring rate, and nature of the suspension, solvent volume, type of quartz vessel, and beam focus (see the SI). In some instances, prolonged irradiation gave fewer H-atom equivalents than anticipated (Figure 2), which is attributed to MOF degradation (vide infra).

To test that the irradiation did not degrade the MOF, two types of experiments were done on various samples. In the first, powder XRD was collected on the photoreduced MOF material. The X-ray diffraction patterns showed no loss in crystallinity of the material upon photoreduction, and no new reflections were observed (see the SI). The second experiment used ¹H NMR analysis of material formed after titration with ^tBu₃ArOH. After removal of the C₆D₆ solvent and resuspension in d₇-DMF, the ¹H NMR spectrum only showed resonances attributed to solvent, Ar₂O, and ^tBu₃ArOH. No terephthalic acid or soluble terephthalate species were detected in the samples in which the titration restored the white color of the oxidized MOF. By contrast, terephthalic acid was present in the solution that corresponds to the blue point at 17.5 h in Figure 2, for which the dark color persisted after oxidation. These observations are indicative of MOF degradation and suggest that extended UV irradiation may degrade MIL-125 under certain conditions.

The measured stoichiometries show that the photo-reductions must be reducing the Ti₈ clusters throughout the MOF particles, not only at the surface. The fraction of Ti₈ nodes that are at the surface was estimated using the roughly elliptical morphologies of the MIL-125 particles observed by SEM, with semiaxes that range from 300 (±100) to 850 (±50) nm (SI). Assuming perfect ellipsoids, the volumetric fraction of the outermost shell — one layer of Ti₈ nodes and linker, 1.9 nm — contains only 1.3% of the Ti₈ clusters. If reduction occurred exclusively in this outer shell, then each surface Ti₈ cluster would have to be reduced by ∼88 H-atoms, or ∼11 H-atoms for each Ti atom, to achieve the observed 1 e⁻/H⁺ per Ti₈ cluster throughout the MOF. Studies with different batches of MIL-125 that differ in size and morphology, ranging from spheres with 85 nm radii to truncated octahedra with radii of 2650 nm, all showed facile reduction to 1 e⁻/H⁺ per bulk Ti₈ cluster. All of these results would require unrealistically high numbers of e⁻/H⁺ per surface Ti (see the SI). Thus, both the Ti₈ nodes at the surface and within the MOF must have been reduced.

The above studies indicate photoreduced MIL-125 can transfer net H-atom equivalents (e⁻/H⁺) to ^tBu₃ArO^•. It is also possible to remove some of the electrons with the simple outer-sphere oxidant decamethylferrocenium hexafluorophos-phate ([FeCp*₂⁺]PF₆⁻). Treatment of reduced MIL-125 with [FeCp*₂⁺]PF₆⁻ resulted in some lightening of the blue color, and ¹H NMR analysis showed the presence of both FeCp*₂⁺/ FeCp*₂, indicative of oxidation of the MOF solid. (FeCp*₂ does not fit in the pores of MIL-125 or photoreduced MIL-125; see the SI). As FeCp*₂⁺ and FeCp*₂ undergo fast electron exchange on the NMR time scale, and a single resonance for the two species is observed. The position of this resonance indicates the mole fractions (χ) of the two species (eq 3).²⁷

$${\delta }_{\text{obs}}\hspace{0.17em}\hspace{0.17em}=\hspace{0.17em}\hspace{0.17em}{\chi }_{\text{FeCp*2}}\hspace{0.17em}{\delta }_{\text{FeCp*2}}\hspace{0.17em}\hspace{0.17em}+\hspace{0.17em}\hspace{0.17em}{\chi }_{\text{FeCp*}{2}^{\text{+}}}{\delta }_{\text{FeCp*}{2}^{\text{+}}}$$ 3

Identical samples of reduced MIL-125 transferred fewer reducing equivalents to FeCp*₂⁺ (electron transfer, ET) than to ^tBu₃ArO^• (PCET/H-atom transfer). In one comparison, FeCp*₂⁺ removed 0.36 e⁻ versus 0.83 H-atom equivalents per Ti₈ node by ^tBu₃ArO ^•. Further, the resulting solution NMR spectrum of MIL-125 that has been treated with FeCp*₂ had several new, unidentifiable resonances, indicative of side reactions. Attempts to use other oxidants, including anilinium radicals, likewise did not give clean electron transfer chemistry. Thus, PCET seems to be more facile and reversible chemistry for the MOF than simple ET. The advantage of chemical reactions that move electrons and protons together is that charge balance is maintained. A principle of solid-state chemistry is that each unit cell must be electrically neutral, and that should apply to MOF materials as well. Removal of just electrons from the MOF leaves an unfavorable positive charge, which likely makes the remaining protons more acidic and leads to decay of the MOF.

The PCET reaction with ^tBu₃ArO^• must occur at the surface because ^tBu₃ArOH is too large to enter the MOF pores (see above). Since complete reoxidation is observed, the e⁻/H⁺ equivalents must be able to migrate through the MOF to the surface. Thus, the material allows for e⁻/H⁺ hopping. Proton^28–31 and electron ^5–7 conductivity in MOFs have already been described; in this study, the two occur together here to maintain charge balance (see above). The proton migration likely occurs via trapped alcohol in the MOF pores and through mobility of the H⁺ from the clusters’ bridging hydroxides, a process very similar to that observed in Zr-based MOFs.^32,33 The H⁺ ion accompanying the e⁻ will likely bind to a bridging oxygen on the MIL-125 cluster to form a hydroxide. A similar mechanism has been shown for COK-69, another photoreducible Ti-MOF.²⁵ Similarly, the PCET reduction of bulk TiO₂ and other oxides is thought to form hydroxides.³⁴ Broad EPR resonances at rt for reduced MIL-125 have previously been ascribed to e⁻ hopping between different Ti centers.¹⁶ The e⁻/H⁺ hopping mechanism between different Ti₈ nodes is not clear. The kinetics of this process are likely complicated and multiexponential due to the size dispersion of the MIL-125 crystals. Monitoring the time course of the formation of ^tBu3ArOH from reaction of photoreduced MIL- 125 with ^tBu₃ArO^• over time is consistent with multi-exponential kinetics and shows that the reaction is complete within 30−60 min (SI).

II. PCET Chemistry of the Soluble Cluster Ti₈O₈(OOC^tBu)₁₆.

Parallel studies using a known, well-defined molecular cluster, Ti₈O₈(OOC^tBu)₁₆ (Ti₈),³⁵ provide additional insight into the nature of the photoreduced state and possible degradation pathways. Cluster Ti₈ serves as a model for the nodes of MIL-125, with a similar ring of eight Ti atoms that are connected via bridging oxo ligands in a relatively flat arrangement (Figure 3). The ligand arrangement differs from that in MIL-125 in that the MOF has four fewer bridging carboxylates decorating the outside of the ring and an additional four bridging hydroxo ligands in the interior of the ring. The empirical cluster formula in the MOF is Ti₈O₈(OH)₄(OOCR)₁₂ (Figure 1) vs the molecular Ti₈O₈(OOC^tBu)₁₆. We chose the Ti₈ cluster with alkyl carboxylates rather than the related benzoate that is more analogous to MIL-125 because of the limited solubility and uninformative ¹H NMR spectrum of Ti₈O₈(OOCPh)₁₆. Ti₈ is readily identified by its two sharp ¹H NMR resonances at 1.28 and 1.37 ppm (C₆D₆) that correspond to ^tBu groups that are axial and equatorial with respect to the planar Ti₈ ring.

Figure 3.

¹H NMR spectra (C₆D₆, 500 MHz): (top) Ti₈ with Ar₂O standard; (middle): after 15 min of irradiation with excess benzhydrol; (bottom) after addition of ^tBu₃ArO^•. Peaks marked with purple # correspond to Ti₈, while orange # correspond to the speculative H atom reduced cluster.

Irradiation of a C₆D₆ solution of Ti₈ (0.95 mM), benzhydrol (Ph₂CHOH, 45 mM), and Ar₂O internal standard for 15 min resulted in darkening of the solution to blue. The ¹H NMR spectrum of this reaction revealed the presence of two broad new resonances at 1.41 and 1.48 ppm, along with the presence of unreacted Ti₈ (Figure 3). Integration showed 80% of starting Ti₈ accounted for, and of this material, ∼43% has been converted to the new species. Treatment of this solution with 1 equiv of ^tBu₃ArO^• caused an immediate color change from blue to pale gray. The ¹H NMR spectrum of this solution shows the single set of ^tBu resonances of Ti₈ (85% mass balance from initial mass) and resonances ascribed to ^tBu₃ArOH (64% yield by initial mass). These results suggest that the new resonances correspond to a photoreduced cluster. Its reactivity with ^tBu₃ArO^• suggests that the new cluster has been reduced by 1e⁻ and 1H⁺.

The observation of only two ^tBu resonances for the reduced cluster indicates that this cluster retains the effective 8-fold symmetry of the ring on the NMR time scale. This suggests that the proton is moving around the Ti₈ ring rapidly on the NMR time scale. Such rapid proton movement is consistent with the proton movement during oxidation of the reduced MIL-125.

The discrepancies in mass balance in the Ti₈ chemistry indicate that partial cluster degradation accompanies photo-reduction and that some of the NMR silent Ti^III species can reduce ^tBu₃ArO^• (see the SI for UV−vis studies). Increasing the irradiation time gave rise to several additional resonances in the corresponding ¹H NMR spectrum, indicative of multiple reduction products. Similar results are noted when pivalic acid is added as a sacrificial reductant. Resonances that correspond to pivalic acid, isobutane, and isobutene implicate the intermediacy of ^tBu^•, which disproportionates to the hydro-carbon products. Thus, the carboxylate ligands of Ti₈ can serve as a sacrificial reductant (photoreduction also occurs in the absence of an added alcohol). A similar process has been reported for pivalate on single-crystal TiO₂ surfaces.³⁶ The reduced Ti₈ cluster reacts much more rapidly with ^tBu₃ArO^• than the MOF does (≤ seconds vs hours). This likely reflects the requirement that e⁻ and H⁺ diffuse to the surface of the MOF to react with ^tBu₃ArO^•.

■ CONCLUSIONS

In summary, the proton-coupled electron transfer (PCET) chemistry of MIL-125 has been developed. Irradiation of MIL-125 with ⁱPrOH as a sacrificial reductant gives a maximum of two (e⁻/H⁺) per Ti₈ cluster node. Subsequent oxidation by ^tBu₃ArO^• occurs by PCET to give ^tBu₃ArOH. The extent of reduction of the MOF shows that the reducing equivalents must be stored within the MOF and not only on the outer surface. Since ^tBu₃ArO^• is too large to enter the MOF, the PCET oxidation must, however, occur at the surface, with the e⁻ and H⁺ diffusing from the bulk to the surface. The coupling of e⁻ and H⁺ in the MOF likely results from the requirement for charge balance, that it is energetically unfavorable for charges to build up inside the MOF. This study thus presents an unusually detailed look at PCET involving a MOF material. Though there are currently very few well-defined examples, we believe that PCET will prove to be a very common type of redox reactivity in MOFs and other porous materials. This work may therefore be relevant to a number of emerging MOF applications, such as CO₂ reduction and water splitting. Our results also show the utility of the simple solution NMR methods developed for quantification of e⁻/H⁺ equivalents transferred to the MOF. These methods should be widely applicable to other redox-active MOFs and heterogeneous materials, showing that they can store redox equivalents and that reactivity can be limited to the surface.