Photocatalytic Hydrogen Evolution from Water Using Fluorene and Dibenzothiophene Sulfone-Conjugated Microporous and Linear Polymers

Three series of conjugated microporous polymers (CMPs) were studied as photocatalysts for hydrogen production from water using a sacrificial hole scavenger. In all cases, dibenzo[b,d]thiophene sulfone polymers outperformed their fluorene analogues. A porous network, SCMP3, showed the highest hydrogen evolution rates of 6076 μmol h−1 g−1 (λ > 295 nm) and 3106 μmol h−1 g−1 (λ > 420 nm), with an external quantum efficiency of 13.2% at 420 nm. S-CMP3 outperforms its linear structural analogue, P35, whereas in other cases, nonporous linear polymers are superior to equivalent porous networks. This suggests that microporosity might be beneficial for sacrificial photocatalytic hydrogen evolution, if suitable linkers are used that do not limit charge transport and the material can be wetted by water as studied here by water sorption and quasi-elastic neutron scattering.


■ INTRODUCTION
The use of hydrogen as an energy carrier has the potential to radically reduce greenhouse gas emissions from hydrocarbon combustion but, at the moment, most hydrogen is produced by methane steam reforming. Photocatalytic water splitting has the potential to produce hydrogen from water with oxygen as the only side product. Inorganic oxide materials such as TiO 2 , WO 3 , and SrTiO 3 have been studied for hydrogen and oxygen production via oxidative or reductive half reactions; based on these half reactions, overall water splitting has been achieved, 1 for example, via Z-schemes 2,3 and heterojunctions. 4 To be scalable, photocatalysts should be based on earth-abundant and nontoxic elements. Recently, organic photocatalysts have become the subject of much research, prompted in large part by a study on carbon nitride published in 2009. 5−8 Other organic material classes such as conjugated polymers, 9−16 conjugated microporous polymers (CMPs), 17−25 covalent triazine-based frameworks (CTFs), 26−34 and covalent organic frameworks 35−39 have also been explored. Copolymerization with a range of comonomers 9,12,40−45 and reducing the size of photocatalyst particles have been found to enhance the activity, 46−48 giving promising results, especially for the hydrogen-evolving half reaction. However, the large structural diversity that is available makes it challenging to formulate simple structure−property relationships.
In particular, we do not yet understand the complex interplay between optical gap, charge mobility, excited state lifetime, and other properties such as particle size, porosity, and surface hydrophilicity. This is compounded by the fact that there are relatively few mechanistic studies for these materials, and it is often unclear whether these polymers act as light absorbers, as catalysts, or as both. Previously, we 12 and others 22−24 have reported nonporous linear polymers with sacrificial hydrogen evolution rates (HERs) that exceed those that we reported for porous CMPs. This suggested that microporosity may not be essential for good photocatalytic activityor perhaps even that porosity and linkers that are used might introduce limitations, perhaps by reducing charge carrier mobilities. Here, we test this by comparing a range of porous network polymers and nonporous linear polymers based on either fluorene or dibenzo [b,d]thiophene units copolymerized with one of six different linkers. We investigate the photocatalytic activity for hydrogen evolution from water under sacrificial conditions and show that porous conjugated polymers can outperform linear analogues significantly, although not for all monomer combinations.

■ RESULTS AND DISCUSSION
All materials were synthesized using Pd(0)-catalyzed Suzuki− Miyaura polycondensation of diboronic acids/diboronic acid ester-functionalized monomers with bromo-functionalized comonomers (Figure 1, see the Supporting Information for synthetic details). 12,17 The materials were insoluble in organic solvents and in water, and they were characterized by Fourier transform infrared spectroscopy (FT-IR) and elemental analysis (see the Supporting Information). Thermogravimetric analysis (TGA) indicated that the polymers were thermally stable in air up to 300°C ( Figure S3).
Scanning electron microscopy (SEM) images of the CMPs show that the CMPs all consist of larger aggregates and the linear polymers showed morphologies that were similar to the CMPs. Powder X-ray diffraction (PXRD) showed that the CMP materials and the 1,3-benzene-co-fluorene polymer, P32, were amorphous, whereas P3, P7, 12 and P33−P35 were semicrystalline ( Figures S4 and S5). Nitrogen sorption isotherms were measured at 77 K for all materials and apparent Brunauer−Emmett−Teller (SA BET ) surface areas were calculated based on the adsorption isotherms (Figures S6−S12). The fluorene derivatives F-CMP1 and F-CMP2 had low apparent surface areas (62 and 51 m 2 g −1 ) compared to their dibenzo[b,d]thiophene sulfone analogues, S-CMP1 and S-CMP2 (508 and 560 m 2 g −1 ), whereas F-CMP3 showed a larger surface area than S-CMP3 (596 vs 431 m 2 g −1 ). The linear polymers all had low surface areas (<70 m 2 g −1 ), except for the dibenzo[b,d]thiophene sulfone containing polymers P33 and P35, which had modest BET surface areas of 192 and 114 m 2 g −1 , respectively. It is possible that the rigid polymer P33, with its 1,3-substituted linker, is a polymer of intrinsic microporosity as its adsorption isotherm shows a small micropore step at low relative pressures ( Figure S10).
UV/visible absorption spectra were measured for all materials in the solid state (Figures 2a and S21−S30). As expected, the 1,3-linked polymers P32 and P33 have the most blue-shifted absorption onset with optical gaps of 3.20 and 2.99 eV, followed by the 1,3,5-linked polymer networks (3.05 and 2.87 eV for F-CMP1 and S-CMP1, respectively): this is a result of the metalinkages in these polymers that limit the effective conjugation length. 18 The 1,2,4,5-linked polymer networks are red-shifted compared to their 1,3,5-linked analogues, 18 and they have similar optical gaps compared to their linear 1,4-benzene-linked analogues: 2.85 eV for F-CMP2 versus 2.86 eV for P3; 2.70 eV for both S-CMP3 and P7. The 2,2′,7,7′-linked 9,9′-spirobifluorene networks 18,49 and their linear 2,7-fluorene-linked analogues are the most red-shifted materials. Again, the polymer networks show similar optical gaps compared to their linear analogues, with optical gaps of 2.77 eV for F-CMP3 and 2.75 eV for P34. S-CMP3 has a gap of 2.56 eV for F-CMP3 compared to 2.59 eV for P35.
Prediction of difficult-to-measure (opto)electronic properties of polymer networks such as CMPs, for example, their ionization potential (IP) and electron affinity (EA), is challenging because of the complex structure of these materials. Therefore, we focused instead on predicting the relative properties of similarly sized cluster units of the different polymers. Otherwise, we used the same approach as in earlier work 12,40,41,50 based around density functional theory (DFT) calculations with the B3LYP 51,52 density functional and the COSMO 53 dielectric screening model (ε r 80.1, water). These calculations ( Figure 3) predict that switching from fluorene to dibenzo[b,d]thiophene sulfone linker units results in the IP moving to more positive values, increasing the driving force for triethylamine (TEA) oxidation, in line with previous studies. 12,16 However, these calculations also suggest that there should be no significant difference in potentials and driving   force between the CMPs and the corresponding linear polymers. Finally, the torsion angles between adjacent units in the optimized polymer cluster models are ∼30°larger for F-CMP2 and S-CMP2 than for any of the other materials (see Table S9). The photocatalytic activity of the polymers was studied under sacrificial conditions using both broad-spectrum irradiation (λ > 295 nm) and visible light (λ > 420 nm; Table 1). TEA was used as the sacrificial hole scavenger and methanol was added to avoid phase separation between water and the TEA. 12,18 Methanol itself has been shown not to act as a hole scavenger for these photocatalysts. 12 All photocatalysts were tested after purification by Soxhlet extraction, and without any additional metal cocatalysts (e.g., Pt), although significant residual levels of palladium (0.13−1.61 wt %) were found by inductively coupled plasma optical emission spectrometry (see the Supporting Information), which has been shown previously to act as a cocatalyst for CMP photocatalysts, 9,18 linear polymers, 54 CTFs, 29 and carbon nitride. 55 Palladium particles are distributed throughout both porous materials (F-CMP3 and S-CMP3) and the polar nonporous polymer (P35), as evident by transmission electron microscopy (TEM) measurements (Figures S15−S17). By using scanning mode and transmission mode images of the same area, we were also able to distinguish palladium particles on the surface from those within the material (Figures S18−S20). Palladium within CMPs has been previously reported to be accessible to aqueous solutions, 56 and we also found that the palladium content of F-CMP3 and S-CMP3 could be reduced by washing with a palladium scavenger.

Chemistry of Materials
All of the materials in this series acted as photocatalysts for hydrogen evolution, albeit with large variations in activity ( Table 1). As summarized in Figure 1, neither the porous CMPs nor the nonporous linear polymers can be said to be the "best" photocatalysts: for four pairings, the porous materials are superior (S-CMP1, F-CMP1, F-CMP3, and S-CMP3), . Initial amplitudes (A, B 1 , B 2 , and B 3 ) are estimated and iterated along with the lifetimes (τ 1 , τ 2 , and τ 3 ) until a fit is found. The prompt is measured separately and used for deconvolution of the instrument response. f Loaded with 3 wt % Pt from H 2 PtCl 6 .  57 and thus defects may be expected, such as 1,3-linkages. This, taken together with the significantly larger torsion angles of the CMPs compared to their linear analogues, might limit the charge transport along the chain, as observed previously for conjugated materials. 58,59 Related observations have been made by others in photocatalysis experiments where a 1,2,4,5-benzene-linked benzothiadiazole CMP was outperformed by the 1,3,5-benzene-linked analogue, which was itself outperformed by the corresponding 1,4-benzene-linked linear polymer. 22 For the fluorene-linked and 9,9′-spirobifluorenelinked polymers, this trend is reversed: the microporous polymers, F-CMP3 and S-CMP3, have much higher HERs than their nonporous linear analogues. For F-CMP3, the HER under broadband illumination is high (2545.3 μmol g −1 h −1 ), but because of its absorption onset at 447 nm, it shows a lower hydrogen production rate under visible light (538.5 μmol g −1 h −1 ). S-CMP3 has the highest photocatalytic performance of the materials in this study under visible light irradiation: 3106.0 μmol g −1 h −1 , which is much higher than its linear analogue P35 (826.1 μmol g −1 h −1 ). S-CMP3 shows a rate of 6076.0 μmol g −1 h −1 under broadband illumination, again higher than P35 under the same conditions (1176.8 μmol g −1 h −1 ). These HERs are among the highest reported for amorphous materials under sacrificial conditions. In order to rule out the possibility that the difference in activity can be explained by the difference in residual Pd content, we synthesized F-CMP3 and S-CMP3 with varying amounts of the Pd catalyst. This resulted in materials that have different residual amounts of Pd (Tables S1 and S2), but similar optical onsets (Figures S31 and S32). For residual Pd amounts above 0.1 wt %, we see little change in the photocatalytic activity of F-CMP3 and S-CMP3 ( Figure  S53). For the S-CMPs that contain the lowest amount of residual palladium (470 ppm), a drop in performance to 1076.1 μmol g −1 h −1 is observed. We attempted to remove residual Pd from the samples with the lowest amount of Pd by washing the sample with a palladium scavenger. 54,60 This reduced the amounts of residual palladium to 274 ppm for F-CMP3 and to 320 ppm for S-CMP3 and greatly lowered their HERs, to 19.5 and 110.5 μmol g −1 h −1 , respectively. The observation of an increase in activity followed by a plateau region has been reported previously for other CMPs 9 and linear polymers. 54 These results show that residual Pd is involved in the catalytic cycle but also that the differences in activity between the materials cannot be simply explained by differences in the amounts of residual Pd in the samples. When S-CMP3 containing 320 ppm residual palladium was loaded with platinum via in situ photodeposition from H 2 PtCl 6 , an increase in catalytic activity was observed. A loading of 2.1 wt % gave the highest HER, but the dependency on Pt concentration was not very large, with HERs varying between 1762 μmol g −1 h −1 (for 0.6 wt % Pt loading) and 2388 μmol g −1 h −1 (for 2.1 wt % Pt loading). At a higher loading of 7.1 wt % platinum, the rate drops to 1124 μmol g −1 h −1 . Similarly, a sample of S-CMP3 containing 8200 ppm residual palladium showed a lower performance when loaded with additional 5.1 wt % platinum (2210 vs 250 μmol g −1 h −1 ). This reduction in activity with higher metal loadings has been observed previously and it is related to hindered light absorption and enhanced recombination between photogenerated electrons and holes. 55 An external quantum efficiency (EQE) of 7.7% (±0.9%) was measured for platinized S-CMP3 at 420 nm in a water/ MeOH/TEA mixture (the sample contains 320 ppm residual palladium and 2.1 wt % platinum that was photodeposited onto the material). An even higher EQE of 13.2% (±1.4%) was measured for as-synthesized S-CMP3 containing 0.72 wt % palladium at 420 nm in a water/MeOH/TEA mixture. This value is significantly higher than previously reported EQEs for P1 (EQE 420nm = 0.4%), the linear dibenzo[b,d]thiophene sulfone copolymer P7 (EQE 420nm = 7.2%) under the same conditions. 12 This efficiency also exceeds that reported for a platinum-modified tricyano-benzene-centered phenylenevinylene-co-terphenylene polymer network, OB-POP-3, in a triethanolamine mixture (EQE 420nm = 2.0%), 44 and a platinum-modified phenylene-benzothiadiazole-copolymer, B-BT-1,4, in a triethanolamine mixture (EQE 420nm = 4.0%). 22 Wavelength-dependent photocatalytic experiments for S-CMP1 ( Figure S58) and S-CMP3 ( Figure S59) show a decrease of activity with lowered absorptivity of both materials, roughly following their absorption profiles. This shows that the hydrogen production is a photocatalytic process, and it suggests that extension of the light absorption profile into visible and near infrared might further enhance the photocatalytic activity.

Chemistry of Materials
The photostability of S-CMP3 was evaluated by irradiating a sample for 35 h under λ > 420 nm illumination (Figure 2b). The hydrogen production rate was reduced over the course of the run, but the photocatalytic system remained active throughout. Furthermore, no apparent change in the FT-IR and UV/vis spectra was observed after the experiment ( Figures  S61 and S62).
The lifetime of the excited state was estimated using timecorrelated single-photon counting, with weighted averaged lifetimes ranging from 0.52 to 1.70 ns. Unlike our previous study, 18 we did not observe a clear correlation between the photocatalytic activity and the lifetime of the excited state for this series of materials ( Figure S78). However, when comparing each subset of CMPs, the lifetime of the fluorene CMPs was always lower than the dibenzo[b,d]thiophene sulfone CMP analogues (Figure 2c). For example, 1,3,5benzene-linked F-CMP1 has a lifetime of 0.52 ns compared to 0.98 ns for S-CMP1; likewise, the 1,2,4,5-benzene-linked F-CMP2 has a shorter lifetime of 0.58 ns compared to S-CMP2 (1.07 ns). Only the lifetimes of 9,9′-spirobifluorene-linked F-CMP3 and S-CMP3 are similar, and probably within the error of the measurement (0.68 and 0.72 ns). We note here that the 9,9′-spirobifluorene S-CMP3 and F-CMP3 materials are also more similar in terms of their relative performance under broadband illumination and visible irradiation, as compared to the two other pairs of CMPs (S-CMP1 and F-CMP1 and S-CMP2 and F-CMP2). The linear polymers show comparable trends: for example, the 1,4-benzene-linked dibenzo[b,d]thiophene sulfone copolymer P7 has a longer lifetime (0.87 ns) than the fluorene copolymer P3 (0.52 ns). Particle size may also affect the photocatalytic activity, and we note that the powders for the fluorene materials in the solid state were in general coarser than for the dibenzo[b,d]thiophene sulfone analogues, though a previous study suggests that dispersibility, wettability, and interaction of water and scavenger with the dibenzo[b,d]thiophene sulfone unit play a more important role. 12 Scattering of light can affect the light harvesting significantly. This was studied by means of backscattering experiments (S82), and F-CMP3, S-CMP3, P7, P34, and P35 all show high backscattering values between 4 and 7%.

Chemistry of Materials
Particle size distributions were measured by static lightscattering experiments in water/methanol/TEA mixtures, showing some variation. For each linear polymer/CMP pair, the Sauter mean diameter 40 of the CMP networks was higher. Sauter mean diameters for the networks varied from 11.5 to 4.7 μm, whereas the linear polymers showed ranged between 8.3 μm (P35) and 1.12 μm (P7) ( Table S5). This might indicate that the higher external surface area of smaller particles contributes to their catalytic activity: for instance, P7, easily the most active linear polymer, has by far the largest relative external surface area (5365 m 2 cm −3 ), as compared to the structurally similar P33 (1324 m 2 cm −3 ) and P35 (725 m 2 cm −3 ), and also S-CMP3 (523.8 m 2 cm −3 ). The observation that smaller particles have higher photocatalytic activity is also in line with reports on colloidal polymer photocatalysts. 46,47,61 To investigate the possible effect of swelling, the particle sizes of P34, P35, F-CMP3, and S-CMP3 were also measured in water. The volume mean diameter of the two fluorene polymers decreased upon moving from water to water/TEA/ MeOH, whereas the dibenzo[b,d]thiophene sulfone analogues increased in size when moving from water to water/TEA/ MeOH. This size increase for the network S-CMP3 was significantly larger (factor of 4.4) than for the analogous linear polymer P35 (factor of 1.67).
This may indicate that S-CMP3 swells in the reaction mixture, thus further increasing the catalytically active surface area in comparison with the other materials. This perhaps shows that to gain a comprehensive insight into the catalytically active surface area in the reaction mixture, other factors, such as swellability and water uptake, need to be considered in addition to dry-state measurements by gas adsorption.
The photocatalytic performance within the polymer series could not be correlated simply with BET surface area measurements ( Figure S73). However, water uptake measurements for F-CMP3 and S-CMP3 show distinct differences: F-CMP3 shows very little water uptake in the relative pressure region of 0.2−0.6 P/P 0 , but uptake is observed in this region for the dibenzo[b,d]thiophene sulfone material S-CMP3 (Figure 4). This means that water can penetrate the pore structure in the more polar S-CMP3 to a greater extent than it does in F-CMP3, which might contribute to the observed difference in photocatalytic activity. At higher relative pressures, both materials exhibit water uptake, and this relates to free water that is adsorbed on the surface and interstitial space of the particles. 62 Very little water uptake was observed for P35, the linear analogue of S-CMP3, which again could contribute to the higher performance of S-CMP3 compared to its nonporous analogue, P35. As discussed above, we found that swellability is evident in S-CMP3 in water/methanol/TEA mixtures. Furthermore, organic molecules such as TEA and methanol affect the wetting of these materials. When Na 2 S was used instead as the sacrificial scavenger, a reduction in activity under visible light was observed to 10.3 μmol g −1 h −1 for F-CMP3, to 216 μmol g −1 h −1 for S-CMP3, and to 68 μmol g −1 h −1 for P35. Compared to the experiments in water/methanol/TEA, the reduction is largest for F-CMP3 (52 times) compared to S-CMP3 (14 times) and P35 (12 times). Similarly, photocurrent measurements ( Figure S60) in a sodium sulfate solution showed that the response for porous and wettable S-CMP3 was greater than its nonporous analogue, P35, as well as its porous, nonpolar analogue, F-CMP3. Taken together, these data suggest that a combination of high surface area and wettability seems to be beneficial for aqueous photocatalysis in water, providing that the provision of these surface properties does not negatively impact other factors, such as the potentials of the charge carriers (IP, EA) that drive the oxidation of TEA and reduction of protons. The latter is not the case here, as predicted by DFT calculations (Figure 3).
To study the differences between F-CMP3 and S-CMP3 further, we performed, for the first time, large-scale, facilitybased neutron-scattering measurements on CMPs. Incoherent quasi-elastic neutron scattering (QENS) allowed us to study the difference in dynamics between dry and hydrated CMPs, as well as to probe trapped water in F-CMP3 and S-CMP3 pore networks, on a 5−50 ps time scale. The large difference in incoherent neutron cross sections between carbons and hydrogens means that the QENS signal is largely dominated by the water signal. However, the use of deuterated water (D 2 O) allows us to reduce the water signal significantly due to the much lower incoherent neutron cross sections of deuterium.
First, we studied the CMPs in D 2 O to extract the QENS signal of the hydrated CMPs. The Q-averaged dynamical structure factor S(E) at 300 K is plotted in Figure S83. Figure  S83a,b show the impact of hydrating F-CMP3 and S-CMP3, respectively. In both cases, the QENS signal broadens as the amount of D 2 O increases and a background develops for a D 2 O content higher than 32 wt % for F-CMP3 and 52 wt % for S-CMP3, respectively. The QENS signal at higher D 2 O content for both samples can be fitted reasonably by   S86). Thus, we assign this background to free water and this onset to a maximum water uptake (Figure 5a,b). Comparing hydrated F-CMP3 and S-CMP3 QENS signals (Figure 5c) points toward faster dynamics of S-CMP3 on the time scale studied. To summarize, S-CMP3 exhibits faster dynamics on 5−50 ps time scale in the fully hydrated states than F-CMP3 and it absorbs more water (52 wt %) than F-CMP3 (32 wt %). These water uptake values calculated from QENS correlate well with the water uptake measurements ( Figure 4). Next, we investigated the water dynamics within the pores of F-CMP3 and S-CMP3. Figure S83c,d shows the impact of the water content on the dynamics of some specific states of water. At high water content, the QENS spectra resemble the free water signal but at a low water content, a clear elastic contribution arising from the CMP itself can be observed.

Chemistry of Materials
It was not possible to fit the QENS signals by using a combination of the hydrated CMP signal (extracted from the previous measurements with D 2 O) and the pure water signal. We therefore fitted these QENS data by adding a third contribution (modeled by a delta function and two Lorentzians convoluted by the resolution of the instrument) that we assign to the dynamics of water in the pores. Below 32 wt % H 2 O for F-CMP3 and 52 wt % H 2 O for S-CMP3, respectively, the QENS signal can be modeled by a combination of the hydrated CMP signal and the signal from the water in the pores, as shown in Figures S85 and S87. For higher H 2 O contents, the QENS signal is modeled reasonably by a combination of the hydrated CMP signal, the free H 2 O signal, and the signal from the water in the pores. The fraction of all the contributions is shown in Figure 5d,e. The free H 2 O contribution lies below the expected water contribution, as calculated using the mass ratio and the difference in the incoherent cross section between the CMP molecules and H 2 O. By adding the third contribution to the free H 2 O contribution, we obtain a total H 2 O contribution (red squares in Figure 5d,e) that is in reasonable agreement with the expected H 2 O contribution, confirming our assignment of the extra contribution to the dynamics of water in the pore networks of the CMPs. Figure 5f shows that there is an increase in the elastic contribution and a narrowing of the QENS signal for the water in the pores in comparison with free water. This indicates that the dynamics of the water in the pores is frustrated on the accessible 5−50 ps time scale. This effect is more pronounced in the case of F-CMP3.
To our knowledge, this is the first time that QENS has been applied to probe local dynamics of CMPs. In the accessible 5− 50 ps instrumental time scale, we have shown that the dynamics of the F-CMP3 and S-CMP3 in a dry state are similar, but the dynamics of S-CMP3 in a fully hydrated state are faster than for F-CMP3. This correlates with S-CMP3 adsorbing more water than F-CMP3. We also show that the water dynamics is frustrated up to the maximum water uptake due to pore confinement; this frustration effect is more pronounced in the case of F-CMP3. The lower water uptake of F-CMP3 in comparison to S-CMP3, combined with more frustrated water dynamics, points toward reduced mass transport in the F-CMP3 network that may contribute to its reduced photocatalytic activity.

■ CONCLUSIONS
The introduction of 9,9′-spirobifluorene linkers in S-CMP3 results in a microporous material with a HER that is more than three times higher than the linear nonporous analogue, P35, under visible irradiation, and more than five times higher under broadband illumination. These rates are among the highest observed for organic polymers. By contrast, porous analogues of copolymers P3 and P7 show reduced photocatalytic activities. In both cases, the observed optical gaps and light absorption profiles for the porous and nonporous analogues are relatively similar (Figure 2a). This seems to originate in the highly twisted nature of the CMPs because of the linker used, which may cause a reduction in charge mobility in the lowdensity porous materials, which might in turn counterbalance or even negate any mass-transfer benefits that arise from  CMPs may yet have advantages as components for overall water splitting schemes, and it is a CMP, S-CMP3, that shows the highest performance in this study. Although no global correlation between the excited state lifetime and the HER was found, it seems that the dibenzo[b,d]thiophene sulfone units prolong exciton separation, and this may contribute toward increased activity over fluorene analogues, alongside other factors such as red-shifted absorption onsets and enhanced wettability of the material particles and its pores. We have also demonstrated the relevance of QENS measurements for studying the structure−property relationship of CMPs in the context of photocatalysis. A forthcoming work will be dedicated to a detailed study of the Q-dependence of the QENS spectra. This is expected to provide deeper insights into the mass transport processes in these CMP networks.

Chemistry of Materials
■ ASSOCIATED CONTENT

* S Supporting Information
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.chemmater.8b02833. Full synthetic details, additional synthetic experiments and analysis for all materials, TGA, FT-IR, PXRD, gas sorption, UV/vis and photoluminescence spectra, SEM, TEM, energy-dispersive X-ray data, computational data, further photocatalysis data, time-correlated singlephoton counting data, selective laser sintering, and QENS data (PDF)