Conjugated Microporous Polymer Nanoparticles with enhanced dispersibility and water compatibility for photocatalytic applications†

Here we report the design of well-dispersible conjugated microporous polymer nanoparticles as highly stable heterogeneous organic photocatalysts in aqueous medium. This class of materials combines two different heterogeneous catalytic model systems, (i) nanostructured catalysts and (ii) highly porous materials. By incorporating different electron donor and acceptor moieties, the nanoparticle shape can be controlled ranging from nanospheres via nanorods to nanorings. High photocatalytic activity, stability and reusability of the materials were demonstrated in the reductive activation of molecular oxygen and photooxidation of amines under irradiation of a household energy saving light bulb.


Introduction
Using sunlight as an inexpensive, non-polluting, abundant and clean energy source is undoubtedly a strategy that will aid in reducing our fossil fuel dependence as well as reducing anthropogenic CO 2 emissions. Taking inspiration from Nature, where various chromophores or photocatalysts for converting solar energy into chemical energy have been utilized for millions of years, synthetic chemists have developed a vast number of visible light-driven systems involving photo-redox processes in order to mimic the natural photosynthesis in recent years. [1][2][3][4][5][6] Beside the well-developed transition metal-based complexes, 7,8 metal-free and organic dyes have also been successfully employed as non-metal photo-redox catalysts in the last two decades. [9][10][11] However, a number of disadvantages associated with these homogeneous catalytic systems still affects the long-term usage. For instance, they are either highly expensive, toxic, and with limited availability in nature; or suffer from solvolytic attack in the reaction medium, which can be troublesome for the recycling ability of the catalyst. It is therefore beneficial to develop a new class of organic heterogeneous photocatalyst with high stability and reusability. For this purpose, two different heterogeneous photocatalytic model systems, (i) nanometer-sized system and (ii) porous materials are herein investigated. Nanoparticles (NPs) offer a huge advantage over conventional bulk catalysts due to their high surface-tovolume ratio. 12,13 Porous materials offer an enlarged active interface during the catalytic process due to their large surface area. 14 Recently, conjugated microporous polymers (CMPs), which combine photoactive π-electron backbone and microporous properties, have been employed as stable heterogeneous photocatalysts for organic synthesis under irradiation of visible light. [14][15][16][17][18][19] Recent research activities have demonstrated their ability for singlet oxygen generation, 20 high selective oxidation of organic sulfides, 21 oxidative C-C bond formation, 22 oxidative hydroxylation of arylboronic acids, 23 reductive dehalogenation reaction, 24 visible light-initiated free radical and cationic polymerization, 25,26 and light-induced hydrogen evolution. 27 A transformation of bulk CMPs into nanostructured systems (CMP NPs) offers a combination of the aforementioned advantages in heterogeneous catalytic systems.
Herein, we report the design and synthesis of well-dispersible nanoparticles of conjugated microporous polymers as high efficient metal-free visible light photocatalysts. The CMP NPs combine high porosity with enhanced solution dispersibility, an important feature which allows for a heterogeneous material to behave closer to a homogeneous system. Therefore, this property also opens the possibility for the use of CMP NPs in solution processible methodologies such as spin-coating. Via the incorporation of different electron donor and acceptor building-blocks, CMP NPs were synthesized in a variety of shapes, ranging from nanospheres, nanorods to nanorings. The possible mechanism of non-spherical formation of the CMP NPs was studied by monitoring the polymerization process. High photocatalytic activity of the CMP NPs was demonstrated in the reductive activation of molecular oxygen for degradation of the dye rhodamine B and the photooxidation of N,N,N′,N′-tetramethyl-p-phenylenediamine under irradiation of a household energy saving light bulb.
The design principle and the synthesis of the CMP NPs are illustrated in Figure 1. The CMP NPs backbone structures were designed to demonstrate the formation of electron-hole pairs under visible light irradiation. After the light-induced charge separation of the electron-hole pair, the valence band (VB) and the conduction band (CB) should function as oxidative and reductive  sites of CMP NPs for further redox reactions (Figure 1b).

Results and discussion
The CMP NPs were obtained via palladium-catalysed Suzuki-Miyaura and Sonogashira-Hagihara cross-coupling polycondensation reaction in an oil-in-water miniemulsion ( Figure  1a) 28,29 . For each polymerization reaction, the same amount of starting compounds and surfactant were used to achieve similar reaction conditions within the miniemulsion droplets. The detailed reaction conditions along with detailed characterization of the polymers are described in the experimental section and in the Supplementary Information (SI). The building-blocks, i.e. the A 3 -type cross-linkers and B 2 -type co-monomers used for the different CMP NPs and their physical properties are listed in Table 1. The obtained CMP NPs showed a high stability as dispersion in organic solvents, and despite their hydrophobic nature, also in water ( Figure S1 in Supplementary Information).
Solid state 13   The Brunauer-Emmett-Teller (BET) surface area of the CMP NPs varied from 18 m 2 g -1 to 512 m 2 g -1 (Table 1). The low surface area of B-FL 3 -a and B-FL 3 -b can be due to the dihexylfluorene unit, which most likely blocked the micropores with the alkyl side chains, leading to almost impossible diffusion of nitrogen gas molecules during the gas sorption analysis. The biphenyl (BPh) containing B-BPh 3 -b showed the highest surface area of 512 m 2 g -1 , which is lower than the bulk CMP reported. 30 The same effect also applies to B-BT 3 -b with a surface area of about 200 m 2 g -1 , which is lower in comparison to the bulk CMP from our previous work. 18 The overall trend of the CMP NPs obtained via Sonogashira coupling showed that they have higher surface areas than the ones prepared via Suzuki-coupling. This can be explained because the preparation of the polymers via Sonogashira coupling results in the formation of triple bonds, which act as rigid bridges between the building blocks, therefore, adding an extra space in the polymer backbone resulted in higher surface area. The FL-containing CMP NPs showed broad pore size distribution with size larger than 2 nm. This could be caused by the alkyl side chains of the fluorene unit, which blocked the micropores and required larger space inside the porous polymers as reported before. 24 A previous study of Cooper et al. about the formation mechanism of CMPs via Sonogashira coupling showed that the high surface area was mainly obtained at the last phase of the reaction, after the fusion of small porous fragments. 31 Here, the lower surface area of the CMP NPs could be partially due to the lack of small fragments inside the droplets during the reaction.
In Figure 2, SEM and TEM images of the CMP NPs are displayed. As it can be seen, a variety of nanoparticle shapes was obtained. As expected, the FL-containing B-FL 3 -a and B-FL 3 -b appeared as round spheres, exhibiting an average size ranging from 60 nm to 100 nm. B-BPh 3 -a was obtained as irregular shape particles of ca. 80 nm, whereas B-BPh 3 -b appeared as spherical NPs with size ranging between 100 -150 nm, despite its great rigid polymer backbone structure. The overall trend here, similar to the porous properties, is that the CMP NPs obtained via Sonogashira-coupling are larger in comparison to the ones obtained via Suzuki-coupling. This can be due to a more rigid and extended network structure of the CMP NPs containing triple bonds than the ones without them. Significantly, the BT containing CMP NPs were absolutely nonspherical, B-BT 3 -a showed a rod-like shape with a length of ca. 200 nm and a width of ca. 20 nm. Most surprisingly, B-BT 3 -b appeared as nanometer-sized rings with a diameter of ca. 200 nm and a width of 20 nm (Figure 2).
To investigate the ring formation mechanism of B-BT 3 -b, the polymerization process was monitored with defined reaction time intervals. As displayed in Figure 3, the UV/vis spectra of the asmade nanoparticle samples were taken, indicating their development upon the completion of the polymerization process. The graph shows that the main absorption band of B-BT 3 -b with a maximum peak at ca. 430 nm appeared directly after the miniemulsion preparation using ultrasonication. Figure 3a -3f display the SEM images of B-BT 3 -b particles after the corresponding reaction time. Directly after the ultrasonication, small particles of ca. 20 nm were already formed (Figure 3a). These could be the preformed oligomers, which subsequently formed the ring-like particles inside the droplets of the miniemulsion within the first minutes ( Figure 3b). After 5 minutes, the ring-shaped particles were clearly formed (Figure 3c), which did not differentiate largely from the ones obtained after 10, 30 or 60 minutes (Figure 3d -3f) and the final product after the completion of the polymerization ( Figure  2). The monitoring experiment indicates a possible, but still not totally clear, ring formation mechanism of B-BT 3 -b. An average droplet size of ca. 250 nm determined by dynamic light scattering (DLS) suggests that the final ring formation of B-BT 3 -b might have first taken place at the interface (Figure S12 in SI), followed by an aggregation of the oligomers formed within the first minutes after the ultrasonication. According to the very rigid structure caused by the BT-based polymer backbone, a possible contraction of aggregates might have occurred, which led only to the formation of ring-shaped CMP NPs instead of capsule-like particles. After changing from an ionic surfactant (sodium n-dodecylsulfate) to a nonionic one (Lutensol® AT50) in the reaction, a ring-shaped CMP NPs were still formed ( Figure S13 in SI). This result suggests that the formation mechanism is not depending on the type of surfactant used. And no strong evidence led to the assumption that the nanorings could likely be collapsed capsules in dry state.
The UV/Vis absorption spectra of the CMP NPs are displayed in Figure 4 and all of them show broad absorption band. The donorbased CMP NPs, B-FL 3 -a and B-FL 3 -b, B-BPh 3 -a and B-BPh 3 -b absorb mainly in the blue light range. After introducing a strong electron acceptor, BT unit into the polymer backbone, a large bathochromic shift up to 100 nm occurred, indicating a narrower band gap of the polymers. According to our recent studies on CMPs with various donor-acceptor combinations, the introduction of strong acceptors lowers mostly the LUMO level and therefore the redox-potential of the polymer, which could favor catalytic reactions initiated by light. 32,33 The UV/vis absorption spectra also show that the triple   (Table 1), the reason behind it can be due to a more ordered structure obtained via efficient π-π stacking during the CMP synthesis, which led to a better conjugation throughout the CMP networks, and therefore, lowering the energy levels.
Taking advantage of the excellent dispersibility of the hydrophobic CMP NPs in water, we first explored the photocatalytic activity of the CMP NPs in the reductive activation of molecular oxygen, in which the photon energy was transferred from the photogenerated electron of the electron-hole pair onto oxygen. The obtained activated oxygen species was then subsequently used for the degradation of a water-soluble organic dye, rhodamine B (RhB) (Figure 5a). It is worth to mention that a 23 W household energy saving light bulb was used as light source for the experiments ( Figure S15 in Supporting Information). B-BT 3 -b showed the highest photodegradation rate of RhB. This result is in accordance to the lowest optical band gap from B-BT 3 -b with 1.76 eV. After 25 min over 80% of RhB was already degraded, while its counterpart without triple bond, B-BT 3 -a (band gap: 1.96 eV) only degraded about 50% of RhB after the same period. Interestingly, B-BPh 3 -b showed, despite its larger band gap (2.36 eV), also a slightly higher photocatalytic efficiency in the first minutes in comparison to B-BT 3a. This could be attributed to its high BET surface area (512 m 2 g -1 ), which could offer a large reactive interface during the catalytic process. B-BT 3 -a initially showed a low degradation rate of RhB in the first minutes, however after 20 min it showed an accelerated degradation process of RhB, which might indicate a different reaction order in comparison to its triple-bond containing counterpart B-BT 3 -b. Other CMP NPs only showed very low photodegradation rates of RhB. At the same time, it is worth mentioning that B-BT 3 -b also showed higher photocatalytic efficiency in comparison to its bulk form, which indicates the enhanced photocatalytic activity of B-BT 3 -b in the nanostructured form ( Figure S16 in SI). The efficiency of the CMP NPs, especially of B-BT 3 -b is comparable to porous carbon nitride (C 3 N 4 ), a state-of-art non-metal based visible light photocatalyst under similar reaction conditions. 34 In addition, the CMP NPs showed high stability during the photodegradation reaction. Five extra repeating experiments using B-BT 3 -b as photocatalyst were performed. RhB was degraded almost in a quantitative manner after every cycle, indicating the high stability and reusability of the CMP NPs in water. As shown in Figure S17 in Supporting Information, the absorption band of B-BT 3b barely changed after the photodegradation reaction, whilst the UV/vis absorption spectrum of RhB almost disappeared completely. It is important to mention that almost no degradation of RhB occurs under light irradiation without using the CMP NPs as photocatalyst.
In the photodegradation of organic dyes in water, it is well known that reactive species such as superoxide (˙O 2ˉ) , singlet oxygen ( 1 O 2 ), ˙OH and the photogenerated hole (h⁺) play important roles in the photocatalytic process. 18,34,39 In order to better understand whether those species are involved in the RhB degradation using CMP NPs, we conducted a series of control experiments ( Figure 6) using B-BT 3 -b as catalyst with the following radical scavengers: sodium azide (NaN₃) as 1 O 2 scavenger, benzoquinone (BQ) for as ˙O 2ˉ scavenger, isopropanol (IP) as ˙OH scavenger and ammonium oxalate (AO) as h⁺ scavenger. As shown in Figure 6, the addition of NaN₃ and the absence of oxygen in the reaction led to an expressive decrease in the photodegradation process of RhB. These results suggest that under visible light irradiation the B-BT 3 -b is able to induce the formation of radical oxygen species, which are responsible for the degradation of RhB, with the main role played by 1 O 2. In our previous study, it was shown that the bulk-made CMP similar to B-BT 3 -b was able to generate singlet oxygen species ( 1 O 2 ) under visible light irradiation. Therefore, as an additional control experiment, we further confirmed the role of 1 O 2 in the photocatalytic system by using the well-studied furoic acid as 1 O 2 trap. 33, 35-39 5-Hydroxy-2(5H)-furanone was obtained as the oxidized product in a quantitative manner using B-BT 3 -b as photocatalyst ( Figure S18 in Supporting Information).
A significant decrease in the photocatalytic reaction was also observed by adding AO as hole scavenger (Figure 6), indicating the vital role of the photogenerated hole for the direct oxidation of RhB in a photocatalytic system. 41 We then further evaluated the electron transfer between the photogenerated hole inside the CMP NPs and substrate compounds, i. e. the photooxidation reaction. We examined the transformation of N,N,N′,N′-tetramethyl-pphenylenediamine (TMPD) into its cationic radical form (TMPD •+ ) under an oxidative process using the CMP NPs as photocatalysts. The oxidized form TMPD •+ can be determined by a color change from colorless to blue, and thus make the intermolecular oxidative electron transfer reaction from TMPD to the CMP NPs visible. 40 As displayed in Figure 5b, after the same reaction time under light irradiation, the absorption spectra of the dispersion containing B-BT 3 -b and TMPD exhibited the highest intensity. B-BT 3 -a showed slightly higher photooxidative ability than other CMP NPs, but significantly lower than B-BT 3 -b. Similar to the photodegradation experiments of RhB, the absence of any color changes in the TMPD solution occurred without using the CMP NPs as photoactive oxidizing agent under the same light irradiation condition. It can be observed that B-BT 3 -b can generate more and longer-living electron-hole pairs efficiently, which can further catalyze reactions driven either under reductive or oxidative processes.

Conclusions
In conclusion, we report a new platform of well-dispersible CMP NPs as a combination of nanostructure and highly porous and heterogeneous photocatalytic system. The variety of shapes of the NPs can be controlled by introducing different electron donor and acceptor as building blocks into the polymer backbone, leading to shape variations ranging from nanospheres, nanorods to nanorings. High photocatalytic activity of the CMP NPs was demonstrated in the reductive activation of molecular oxygen for degradation of rhodamine B and the photooxidation of N,N,N′,N′-tetramethyl-pphenylenediamine under visible light irradiation. It was shown that the transformation of the CMPs into nanoparticles could enhance the photocatalytic activity due to the higher dispersity and larger active surface content of the material. Additionally, it is important to note that a household energy saving light bulb was used as a light source, which provides an economically low-cost solution for the photocatalytic process in broader application fields.

Photodegradation of RhB
A mixture of CMP NPs dispersed in water (1 mg mL -1 ) and RhB solution (10 mg L -1 ) were stirred in the dark until reaching the adsorption/desorption equilibrium (60 min) between the nanoparticles and the solution. After that, the mixture was illuminated with a 23 W household energy saving light bulb (Osram) during different time intervals in air. Changes in RhB concentration were analyzed using UV-vis spectroscopy. A control experiment with the reaction mixture under N 2 , therefore, without oxygen in the reaction, was also performed. For the repeating experiments using B-BT 3 -b as photocatalyst, the nanoparticles were separated via centrifugation after each photocatalytic reaction, and added to another RhB solution with the same concentration. The radical scavenger experiments were performed using ammonium oxalate (0.142 g mL -1 ), p-benzoquinone (0.108 g mL -1 ), sodium azide (65 mg mL -1 ) and isopropanol (99%) (60 μL mL -1 ). Each of the radical scavengers was added to a mixture of B-BT 3 -b (1 mg mL -1 ) and RhB solution (10 mg L -1 ). After reaching the adsorption/desorption equilibrium the mixture was illuminated with a 23 W household energy saving light bulb (Osram) for 30 min and changes in RhB concentration were analyzed using UV-vis spectroscopy.

Photooxidation of TMPD
A 10 mL glass vial containing Ar-saturated acetonitrile solutions (2 mL) of TMPD (32.85 mg mL -1 ) with CMP NPs (4 mg mL -1 ) was irradiated by a 23 W household energy saving light bulb for 5 h. After the irradiation, the CMP NPs were completely removed by centrifugation and filtration. The absorbance changes in UV-visible absorption spectra of the resulted solutions before and after irradiation were used to estimate the relative photocatalytic efficiency between the different CMP NPs.

Supplementary Information
Additional experimental details and characterization data such as 1 H NMR and solid state 13 C CP/MAS NMR, extra synthetic procedures, TGA, UV/vis, fluorescence spectroscopy and photocatalytic procedures, additional SEM images, N 2 sorption isotherms and pore size distribution of the polymers are shown. This material is available free of charge via the Internet at http://pubs.acs.org.