Rational approach to guest confinement inside MOF cavities for low-temperature catalysis

Geometric or electronic confinement of guests inside nanoporous hosts promises to deliver unusual catalytic or opto-electronic functionality from existing materials but is challenging to obtain particularly using metastable hosts, such as metal–organic frameworks (MOFs). Reagents (e.g. precursor) may be too large for impregnation and synthesis conditions may also destroy the hosts. Here we use thermodynamic Pourbaix diagrams (favorable redox and pH conditions) to describe a general method for metal-compound guest synthesis by rationally selecting reaction agents and conditions. Specifically we demonstrate a MOF-confined RuO2 catalyst (RuO2@MOF-808-P) with exceptionally high catalytic CO oxidation below 150 °C as compared to the conventionally made SiO2-supported RuO2 (RuO2/SiO2). This can be caused by weaker interactions between CO/O and the MOF-encapsulated RuO2 surface thus avoiding adsorption-induced catalytic surface passivation. We further describe applications of the Pourbaix-enabled guest synthesis (PEGS) strategy with tutorial examples for the general synthesis of arbitrary guests (e.g. metals, oxides, hydroxides, sulfides).

Meanwhile, there are several metal@MOF systems used for CO oxidation tests, such as The most promising method to prepare a ship-in-bottle system is to load metal salts and organometallic precursors into pre-formed open-porous framework via solution-based, gas-phase or mechanical-mixing impregnation followed by either thermal/irradiation decomposition or redox reaction with strong reducing reagents like hydrazine, NaBH 4 or H 2 . [1][2][3][4][5][6][7] Common issue for preparing the ship-in-bottle systems, however, is the poor control in growing the nano-entities within the microporous hosts. In most cases along the postassembly incorporation route, guest moieties are significantly deposited on the outer surface of the porous material. Therefore, the synthesis strategy to form the guests can be far more flexible and versatile than the above-mentioned conventional methods. Meanwhile, a range of guest compounds can be made in various nanoporous hosts. In order to achieve the designed guest@nanoporous-host four key criteria need to be considered: The Pourbaix diagram for Ru-H 2 O system is reconstructed based on the previous efforts reported. [19][20][21] Here, we assume the aqueous concentration of insoluble Ru-containing compound is negligible (effectively 0 M). Since our study is focusing on aqueous systems with pH between 5 and 10, we only consider this pH range to simplify the question. Between pH 5 and pH 10 Figure 3, further below). Since tBMP is hydrophobic, the tBMP inside stays due to the hydrophobic-hydrophilic confinement created by subsequent KRuO 4 (aq) solution impregnation and react with KRuO 4 . As a consequence, hydrous RuO 2 clusters/particles produced are entrapped inside the MOF (Fig. 1c). The product was washed with water and ethanol. It was then dehydrated at ca. 140 °C under nitrogen to achieve the as-

Materials
The following chemicals/items were used as received.

MOF-808-P Preparation
MOF-808-P was synthesized based on the MOF reported by Yaghi et al. 30 The detailed protocol can be find in ref. 25 Figure 6).

tBMP Impregnation and Temperature-Controlled Selective Desorption
Supplementary can be assigned to a number of desorption/decomposition events. tBMP outside the MOF has lower desorption temperature than tBMP inside the MOF due to the stronger interaction when the molecule is trapped inside the nanoporous host. This is consistent with similar system for preparing polymer@MOF systems previously observed. 28,29 Hence, DE (the volatile solvent for tBMP) and tBMP (outside the MOF) can be mostly removed when treated the as prepared tBMP/DE@MOF-808-P at ca. 120 °C. In this way, only tBMP inside the MOF host can remain after treatment, i.e.
tBMP@MOF-808-P. Source data are provided as a Source Data file.

RuO 2 Formation Inside the MOF-808-P
The as-prepared tBMP@MOF-808-P was collected and reweighed. An excess amount of KRuO 4 aqueous solution (20 mM) was then added to tBMP@MOF-808-P. Hydrous RuO 2 forms inside the MOF by mixing tBMP@MOF-808-P with the KRuO 4 solution. Since tBMP is immiscible with the aqueous solution, tBMP will be trapped in the MOF during the reaction. Meanwhile, the partially filled MOF host uptakes the KRuO 4 solution and accommodates the tBMP-KRuO 4 redox reaction within in it.
During the reaction, KRuO 4 reduces to RuO 2 while tBMP is oxidized to its oxidizing derivatives similar to the oxidation of BHT. 26,27 The liquid chromatography-mass spectrometry (LC-MS) analysis confirms the presence of ketone derivatives. We kept the reaction for ca. 4 h. The pH of the system was kept within the range of 5-10. In the case of tBMP:MOF-808-P=2:20, the pH value was measured to be ca. 8.5 and ca. 6 before and after the reaction. The as-synthesized hydrous RuO 2 @MOF-808-P was collected by filtration (the filtrate remains yellow indicating some KRuO 4 left after the reaction) and washed with excess amount of ethanol followed by water. It was then dried at ca. 140 °C 31 for ca. 2 h to become as-synthesized RuO 2 @MOF-808-P. After the synthesis, the white MOF-808-P turns to almost black RuO 2 @MOF-808-P. Meanwhile, we verified that the MOF-808-P by itself is not reacting with KRuO 4 , as the MOF-808-P remains white color and no color change in the KRuO 4 solution upon mixing. We confirmed that the MOF-808-P is stable throughout the sample preparation based, as there is no significant change in PXRD patterns (Supplementary Figure 6, further below). The as-synthesized RuO 2 @MOF-808-P is stable in air and can be stored in ambient condition. area decrease for as-synthesized RuO 2 @MOF-808-P (from N 2 adsorption measurements) is observed when more guest (i.e. RuO 2 ) is incorporated (from ICP-OES). As a consequence, the measured surface area decreases and the measurement pore volume also decreases which is shown in (d) and (e).
More specifically, the RuO 2 forms mostly inside the secondary pores due to their larger size. This explains the decrease of the available volume of the secondary pores to the adsorbing gas and therefore a drastic drop of the total pore volume of the material. This observation is also supported by S13 HR-TEM images of the RuO 2 particles (Supplementary Figure 12). We also noticed that the primary cavity volume also decreases by half due to the partial RuO 2 occupation. The error bars in (a) and (c) represent the standard errors for Ru weight fraction in RuO 2 @MOF-808-P from ICP-OES or surface area from N 2 adsorption measurements. The parentheses in samples' labels in (b) and (e) represent the Ru-element weight fraction measured by ICP-OES. Source data are provided as a Source Data file.
Samples were heated from room temperature up to 900 ºC at a rate adjusted based on the mass loss per unit change in temperature (i.e. high-resolution mode) in Ar.

Nitrogen adsorption measurements:
The samples in Supplementary

RuO 2 @MOF-808-P Characterizations
Supplementary Figure 6. PXRD patterns for simulated MOF-808 based on Ref. 25 , as-synthesized MOF-808-P, dried MOF-808-P and as-synthesized RuO 2 @MOF-808-P. The MOF's structure is mostly preserved after RuO 2 incorporation. No peak for RuO 2 crystal is shown indicating that the RuO 2 particle is very small (< 3 nm) if RuO 2 were there. 15 The PXRD intensity is rescaled for better visualization. Source data are provided as a Source Data file. and as-synthesized RuO 2 @MOF-808-P obtained by ex situ XAS. The apparent Ru-O pair can be S18 identified in RuO 2 @MOF-808-P but not Ru-Ru pair as in metallic Ru (i.e. Ru foil). A marginal peak shift can be observed in RuO 2 @MOF-808-P compared with reference RuO 2 . This may be caused by the presence of C (from the organic ligand of the MOF) in proximity to Ru. 35  RuO 2 @MOF-808-P. The MOF structure is more defective after RuO 2 inclusion (i.e. post-synthetic modification), as PXRD peaks (long-range ordering features) can be hardly observed above 40° (2 theta) for as-prepared RuO 2 @MOF-808-P (in red) as compared to dried MOF (in black). Note that the variation of intensity in peak below 5° (2 theta) can be largely influenced by the presence of guest compounds. 36 Source data are provided as a Source Data file. samples were washed by Milli-Q water thoroughly at room temperature to remove K.

General Characterization Methods in This Section
In situ XAS: In-situ XAS measurements were carried out at the BL14W1 beamline of SSRF.
The spectra were recorded in transmission mode. Self-supporting pellets were prepared from

In situ diffuse reflectance infrared Fourier transform spectroscopy (in-situ DRIFTS):
In situ DRIFTS spectra were recorded on a BRUKER TENSOR 27 spectrometer equipped with S22 a diffuse reflectance accessory (the Praying Mantis) and a reaction chamber (operation temperature from -150 °C to 600 °C). The powder sample was loaded into a sample cup. The sample temperature was controlled by a heater and measured by two thermocouples. One of them was placed in the sample cup; the other one was immobilized on the sampling stage.
The flow rate passing through the reaction chamber was controlled by the mass flow controllers. The DRIFT spectra were recorded using a spectral resolution of 4 cm -1 and accumulating 32 scans. Before the DRIFTS acquisition, the samples were pre-treated in 20 vol% O 2 with 80 vol% Ar at 150 °C for 10 min (RuO 2 @MOF-808-P) or 250 °C for 1 h (RuO 2 /SiO 2 ) and cooled down to room temperature in Ar.
For temperature-dependent CO desorption characterization, 5 vol% CO with 95 vol% He was used. The sample was exposed to 5% CO at room temperature first and then decreased to -50 °C by liquid nitrogen and kept for 2 h (RuO 2 @MOF-808-P) or 1 h (RuO 2 /SiO 2 ) until no change was observed in the real-time spectra (i.e. CO adsorption in equilibrium). Then the gas flow was switched to Ar gas at room temperature and increased the sample temperature to the targeted one. After each targeted temperature was reached for 10 min, the corresponding DRIFTS spectra were collected.
Under reaction conditions, the O 2 -activated samples were exposed to the reaction gas and then switched to He gas. After cooling down to -50 o C in He, the treated samples were S23 exposed to CO pulses consisting of 5 vol% CO balanced with He. All gas follow rate was set to 30 ml•min -1 . The CO concentration was measured using a thermal conductivity detector.

CO Oxidation Tests
The catalysts were loaded into a fixed-bed micro-reactor. Before catalytic activity, the RuO 2 @MOF-808-P and Ru/SiO 2 catalysts were exposed to O 2 (O 2 -activated) or Ar (Aractivated) gas with a flow rate of 30 ml•min -1 and treated at 150 o C for 10 min (to form activated RuO 2 @MOF-808-P) and 250 o C for 1 h (to form activated RuO 2 /SiO 2 ), respectively.
After cooling down to room temperature in Ar gas (30 ml•min -1 ), the gas stream was switched to a reaction gas (1 vol% CO, 20 vol% O 2 , 1vol% N 2 , and balanced with He) with a specific weight hourly space velocity (WHSV). The WHSV in Fig. 4a  heated at a high temperature. After the water treated, the catalysts were exposed to Ar at 120 o C for 60 min. The activity test was carried out from 30 o C to 100 o C with a heating rate of 0.5 o C•min -1 . The gas products were analyzed by an on-line gas chromatography (Agilent GC 6890) equipped with a packed column PQ200 and a TCD. Before the products analysis, the moisture was condensed by ice.

Supporting Results
Supplementary Figure 17.
In situ x-ray absorption spectroscopy results for RuO 2 /SiO 2 (in blue) and RuO 2 @MOF-808-P (in red) before and after the CO adsorption at 30 °C (Ru foil and RuO 2 as reference samples). XANES spectra show that both RuO 2 /SiO 2 and RuO 2 @MOF-808-P are partially reduced upon CO exposure, which is revealed by the change of slope in near-edge region (highlighted in grey). We speculate that surface oxygen atoms in RuO 2 are reacted and replaced by CO at 30 o C.
The results implied that the Ru-O interaction is weakened by confining RuO 2 in MOF. Source data are provided as a Source Data file.
Supplementary Figure 18. In situ DRIFTS spectrum for MOF-808-P treated in the reaction gas and then in Ar at 30 °C. The treatment condition is the same as those mentioned in Fig. 3. There is no peak in the IR spectrum since the MOF-808-P does not adsorb CO under this condition. Source data are provided as a Source Data file. •h -1 . The catalytic results indicated that, for impregnation method, the CO oxidation performance for RuO 2 /SiO 2 with RuCl 3 is better than that for RuO 2 /SiO 2 with KRuO 4 . Therefore, this excludes the precursor contribution (i.e. KRuO 4 ) to the superior performance of RuO 2 @MOF-808-P. Source data are provided as a Source Data file. indicating the formation of carbonates can be noticed. 43 Source data are provided as a Source Data file.

S28
Supplementary Figure 23. PXRD patterns for RuO 2 @MOF-808-P after treatments/tests labeled. The structure is mostly preserved after these treatments/tests. The PXRD experimental setup is the same as those mentioned in SI section 2.6. Source data are provided as a Source Data file.
Supplementary Figure 24. CO oxidation tests for RuO 2 @MOF-808-P which was tested after the standard O 2 -activation mentioned in this work (stage 1) and tested again after being treated with 10 vol% water vapor at 100 °C for 60 minutes (stage 2). Catalysts mass: 30 mg, WHSV = 400 L•g Ru -1 •h -1 .
There is no decrease in the catalytic activity after the water treatment at high temperature. The results imply that the RuO 2 @MOF-808-P catalysts have a high water tolerance. Source data are provided as a Source Data file. As a further demonstration, we use the Pourbaix diagrams constructed with using Materials Project 52-54 to predict the potential PEGS conditions for Pt and Pd inside a MOF (Supplementary Figure 26). Briefly, since no stable Pt 2+ is seen on the Pourbaix diagram for Pt (Supplementary Figure 26, left), it would be very difficult to use the Pt 2+ for host (e.g. MOF) impregnation. Extra stabilization with ligands would be required in the Pt precursor. This explains why Pt precursors such as [Pt(NH 3 ) 4 ]Cl 2 rather than PtCl 2 is used for preparing Pt@MOF via solution-based synthesis, where extra NH 3 is involved to stabilize the Pt(II) salt. Project. [52][53][54] In contrast, according to the Pourbaix diagram for Pd (Supplementary Figure 26, right) Pd 2+ is more ready to be used as the mobile precursors to impregnate MOF under low pH (stable Pd 2+ phase shown at low pH even at fairly high Pd 2+ concentration, 10 -2 mol•kg -1 ≈ 10 mM). This Pourbaix diagram further rationalize the preparation of some published examples of Pd@MOF 1,2,55 with Pd(II) salts, such as Pd(NO 3 ) 2 and Pd(II) acetylacetonate. As an experimental verification, we first stabilized 0.106 g Pd(NO 3 ) 2 •H 2 O in 20 ml HNO 3 (aq, 0.1 M). Although Pd 2+ can be easily reduced to Pd 0 if the pH were unaltered, the ΔE reduction can be more than 0.5 V if the pH becomes significantly higher. Since the pH after the reaction is very likely to be higher than the acidic Pd 2+ solution (i.e. precursor solution), we used NaBH 4 as reducing agent (with standard reduction potential of -1.24 V versus SHE) which can sufficiently reduce the Pd 2+ to Pd 0 regardless the pH change. To prepared the NaBH 4 (aq) solution, 0.15 g NaBH 4 (excess amount) were dissolved by 280 ml Milli-Q water which has the pH value of ca. 8. Note that unlike tBMP for RuO 2 , NaBH 4 has no capability (e.g. hydrophobic-hydrophilic interaction and temperature-controlled elective desorption) to control the Pd loading position. Since the redox reaction would take place in aqueous condition with pH value slightly less than 8 (due to excess NaBH 4 ), we chose the MOF-808-P as the host.

S31
To load Pd in MOF-808-P (i.e. to form Pd@MOF-808-P), we first impregnated Pd(NO 3 ) 2 solution in the dried MOF-808-P. The Pd(NO 3 ) 2 (aq)@MOF-808-P was then reacted with the prepared NaBH 4 solution at room temperature for 10 min. Black suspension was observed upon the reaction indicating the formation of metallic Pd 0 . The product was collected by centrifugation and washed with water and ethanol. It was then dried in vacuum oven at room temperature for 24 h.
Since there is no control about the Pd loading position for the as-prepared Pd@MOF-808-P, metallic Pd 0 forms both inside the MOF and on its outer surface as revealed in Supplementary Figure 27. Some Pd particles can agglomerate on the outer surface of the MOF (without MOF pore confinement). The presence large Pd particles are also confirmed by a peak at ca. 40° (for Pd 0 ) in PXRD pattern in Supplementary Figure 28 for Pd@MOF-808-P. Meanwhile, the PXRD patterns also verify the preserve of the MOF's structure throughout the synthesis. Hence, our PEGS strategy works for preparing Pd@MOF-808-P. Figure 27. A DF-STEM image for Pd@MOF-808-P and its corresponding STEM-EDS mappings for Zr and Pd. Raw images are provided as a Source Data file.