Thiophosphate-Based Covalent Organic Framework (COF) or Porous Organic Polymer (POP)?

Abstract

There are few examples of covalent organic frameworks (COFs) based on phosphorus as the building element, probably because the structure of most phosphorus derivatives is pyramidal, which may prevent the stacking expected for classical 2-dimensional COFs. In addition, they are generally associated with linear difunctional derivatives. In this paper is reported the original association of a trifunctional 3-D compound with a trifunctional 2-D compound in an attempt to get a new COF. The condensation reaction between a thiophosphate derivative bearing three aldehydes and the trihydrazinotriazine has been carried out with the aim of obtaining either a COF or simply a porous organic polymer (POP), consisting in both cases of associated macrocycles, affording a new covalent triazine framework (CTF). The material resulting from this condensation has been characterized by multinuclear MAS NMR (³¹P, ¹H, and ¹³C), IR, and thermogravimetric analysis (TGA). All these data confirmed the condensation reactions. However, BET (Brunauer–Emmett–Teller) measurements indicated that the porosity of this material is low. Trapping dyes in solution, as a model of pollutants, by the insoluble porous material 3 has been attempted.

1. Introduction

Covalent organic frameworks (COFs) [1] have generated widespread interest for more than 15 years, leading to many advanced applications, as emphasized in several recent reviews [2,3,4,5,6]. The general aim is to synthesize associated macrocyclic structures suitable for stacking in successive layers. Such arrangements are so perfect in some cases that they lead to single crystals [7,8]. Covalent organic frameworks are a part of porous organic polymers (POPs). They are based on organic elements, using classical organic reactions such as Schiff-base chemistry for their synthesis [9]. In order to favor the stacking, large flat units such as perylenes, coronenes, porphyrins, or phthalocyanines, to name a few, have been used [10]. On the contrary, there are relatively few examples of COFs based on phosphorus as the building element, probably because the structure of most phosphorus derivatives is pyramidal, 3-dimensional, which may prevent the stacking expected for classical 2-dimensional COFs. However, a recent review has gathered several COFs based on phosphines and phosphine complexes and their catalytic properties [11]. Other examples of COFs based on phosphorus derivatives can be cited; for instance, chiral BINOL-phosphoric acid was reacted with Cu(II)-porphyrin modules to produce a multifunctional chiral covalent framework suitable for the catalytic asymmetric benzylation of aldehydes [12]. A series of borophosphonates was shown to self-condense into cubic units and polycubane COFs of valency 8, displaying unusual rod-within-layer arrangements [13]. Hexachlorocyclotriphosphazene (N₃P₃Cl₆) was reacted with either hydroquinone or phloroglucinol to afford COFs with excellent acid and radiation stability and suitable for uranium adsorption [14]. In another report, N₃P₃Cl₆ was reacted with hydroxyphenylboronic acid to afford COF derivatives with efficient flame retarding effects and toxic smoke suppression [15].

However, in many cases using phosphorus derivatives, the obtained material is not crystalline and has a variable porosity. In this case, the material is still considered a porous organic polymer (POP) but not a COF. A recent review presented the synthesis and properties of phosphorus-containing porous organic polymers, with a focus on their catalytic properties [16]. A table in this review displays the very large scale of BET (Brunauer–Emmett–Teller) values of surface areas measured for phosphorus POPs, from 24 m²/g [17] to greater than 1600 m²/g [18]. Of course, other POPs not including phosphorus derivatives are also known, particularly those based on triazine, affording covalent triazine frameworks (CTFs), which have also been reviewed [19]. A table in this review displays very large values for the BET surfaces, larger than in the case of phosphorus derivatives, from 90 m²/g [20] to 4000 m²/g [21]. To the best of our knowledge, there is only a single report combining phosphorus and triazine derivatives, in which PCl₃ is reacted with cyanuric acid (N₃C₃OH₃) in the presence of NEt₃ as a base to afford a POP suitable for the complexation of copper and used as an efficient catalyst in C-O bond formation reactions [22]. However, no data on the BET surface area were provided, and copper was deposited on the surface of the material, not inside, which presumably indicates the very low porosity of this material.

In the present paper, we used, for the first time, a thiophosphate trialdehyde with trihydrazinotriazine as the second component to obtain either new COF or POP derivatives composed of associated macrocycles. We have previously shown that thiophosphite dialdehydes are perfectly suitable for forming macrocycles with diamines [23]. The association of two compounds with a C₃ symmetry should give hexagonal structures, with each component constituting one edge of the hexagon in alternance. Thus, three equivalents of each component are necessary to build the associated macrocycles [1]. If both compounds with the C₃ symmetry have a 2-D structure, the sheets formed by associated macrocycles stack easily in organized layers, as they have a 2-D structure. However, if one of the compounds with a C₃ symmetry has a 3-D structure, the resulting sheet of associated macrocycles is also 3-D, making it more difficult to stack into an organized layer. The aim of this work was to synthesize these challenging materials and study their characteristics and potential properties, particularly for trapping pollutants.

2. Materials and Methods

Solid state NMR experiments were recorded on a Bruker Avance 400 III HD spectrometer (Bruker Biospin, Ettlingen, Germany) operating at magnetic fields of 9.4 T. Samples were packed into 3.2 mm zirconia rotors. ¹H MAS were performed with the DEPTH pulse sequence and a recycle delay of 3 s. ¹³C and ³¹P CP MAS spectra were recorded with a recycle delay of 1.5 s and contact times of 2 ms and 1 ms, respectively. ³¹P MAS single pulse experiments were conducted with a recycle delay of 10 s. Chemical shifts were referenced to TMS and 85% H₃PO₄ for ¹H, ¹³C, and ³¹P as internal standards. TGA and DSC analyses were carried out on ATG/DSC 3+ Mettler-Toledo (Mettler-Toledo, Greifensee, Switzerland). BET experiments under N₂ were carried out on Micromeritics ASAP 2020 (Micromeritics, Norcross, GA, USA). IR spectra were recorded on Perkin Elmer Frontier FT-IR/FIR spectrometer (Perkin-Elmer, Waltham, MA USA). UV-Vis spectra were recorded on Perkin Elmer Lambda 950 UV/VIS/NIR spectrometer (Perkin-Elmer, Waltham, MA USA). Elemental analyses were carried out with Perkin Elmer 2400 Series II Flash Combustion Analyzer (Perkin-Elmer, Waltham, MA USA).

2.1. Synthesis of Precursors 1 and 2

Compound 1 (4,4′,4′′-[phosphinothioylidynetris (oxy)]tris[benzaldehyde]) was synthesized by reacting P(S)Cl₃ with 3 equiv. of 4-hydroxybenzaldehyde in the presence of cesium carbonate instead of NaH, contrary to what was previously reported [24]. ³¹P MAS NMR of compound 1: 52.7 ppm (¹H MAS, ¹³C CP MAS, ³¹P CP MAS and ³¹P MAS NMR spectra in Figure S1, Figure S2, Figure S3 and Figure S4, respectively). Elemental analyses for C₂₁H₁₅O₆PS: Required: C, 59.16; H, 3.55. Found: C, 59.49; H, 3.46.

Compound 2 (trihydrazinotriazine) was obtained by reacting hydrazinium hydroxide with trichlorotriazine, as previously published [25]. Elemental analyses for C₃H₉N₉: Required: C, 21.05; H, 5.30. Found: C, 20.31; H, 4.81.

2.2. Synthesis of Material 3

In a dry closed Pyrex tube under Argon were added 27.3 mg (0.064 mmol) of compound 1, 10.9 mg (0.064 mmol) of trihydrazinotriazine 2, 1.0 mL of 1,4-dioxane, and 1.0 mL of mesitylene. Compound 2 in the resulting mixture was dispersed under ultrasound. Then, 100 µL of 3 N acetic acid was added under stirring and the mixture was heated for 72 h at 120 °C in a sealed tube. The reaction medium was filtered, and the precipitate was washed with a solution of 1,4-dioxane/mesitylene (1:1) (2 × 10 mL), then with water (2 × 10 mL). The precipitate was then dried in vacuo for one week at 50 °C to afford material 3 as a white non-crystalline powder with a 90% yield (Scheme 1). ³¹P MAS NMR of material 3: 49.0 ppm (¹H MAS, ¹³C CP MAS, ³¹P CP MAS and ³¹P MAS NMR spectra in Figure S5, Figure S6, Figure S7 and Figure S8, respectively). Elemental analyses for C₂₄H₁₈N₉O₃PS: Required: C, 53.04; H, 3.34. Found: C, 45.03; H, 4.38.

2.3. Attempted Trapping of Dyes

A solution of 10 mg of azulene, pyrene, or fluorenone was prepared in 200 mL of 1,4-dioxane; this solution was then diluted 1/100th. To 5 mL of the diluted solution containing 0.0025 mg of dyes was added 10 mg of material 3 (insoluble). The resulting suspension containing 0.025% of dye relative to material 3 was left under ultrasonic conditions for 10 min, then soaked for 24 h at room temperature. The suspension was filtered with filters of 0.45 μM porosity followed by 0.2 μM porosity to eliminate material 3. UV-Vis spectra of this solution and the dilute solution of dyes were compared. Analogous experiments were carried out with Orange G in water. A 10 mg mass of Orange G was diluted in 200 mL of deionized water. Then, 4 mL of this solution was diluted with water to 10 mL. To 5 mL of the diluted solution containing 0.1 mg of Orange G was added 10 mg of material 3 to afford a suspension containing 1% of dye relative to material 3. This suspension was treated as previously. Another attempt was made with pyrene in water (10 mg in 200 mL), then this solution was diluted 1/100th, and all the experiments were conducted as for dioxane.

3. Results

The 3-dimensional and trifunctional compound 1 was reacted with the trifunctional but 2-dimensional compound 2, which was previously used to synthesize COFs [25,26]. The condensation reaction was carried out at 120 °C in 1,4-dioxane/mesitylene (1:1) and acetic acid 3N in a sealed tube. These harsh conditions are necessary to ensure the solubility of both reagents at the beginning of the reaction—we verified the stability of each reagent individually under these conditions. Elimination of water in this reaction resulted in the formation of a white precipitate of material 3 (Scheme 1). It was characterized first by multinuclear solid-state MAS (magic angle spinning) NMR. Figure 1 displays the comparison between the MAS ³¹P spectra of thiophosphate 1 and material 3. The resonance of the P=S group occurs in the same range of chemical shifts, at δ = 52.7 ppm for 1 and δ = 49.0 ppm (broad band) for 3.

Material 3 was also characterized by ¹H and ¹³C MAS NMR. The disappearance of the signal corresponding to the aldehydes (δ = 190.9 ppm) is observed in ¹³C MAS NMR spectra of material 3 compared to 1. A new signal is also observed at 164.9 ppm, corresponding to the carbon in the C=N bond (Figure 2). A large decrease in the aldehyde signal at 1660 cm⁻¹ is also observed in the IR spectra of material 3 compared to that of compound 1, whereas most of the other main bands in the IR spectrum of compound 1 are also found in material 3 (Figure 3).

The thermal stability of material 3 was determined in air by thermogravimetric analysis (TGA) (Figure 4). A first loss of about 10% is observed up to 200 °C, and about 80% of the mass is retained up to 300 °C. After this temperature, a large decrease in the mass, to almost zero, is observed between 300 and 700 °C.

BET (Brunauer–Emmett–Teller) measurements were carried out on material 3 to determine the capacity of the physical adsorption of N₂, i.e., the specific surface area of porous materials. The value obtained was 3.8542 m²/g (Figure 5).

4. Discussion

The reaction between two components with a C3 symmetry should afford a material with a hexagonal structure [1], as shown in Figure 1. Reacting one equivalent of compound 1 with one equivalent of compound 2 induced the elimination of water, creating hydrazone bonds to produce material 3 as an insoluble amorphous white powder, as shown by polarized light microscopy. Furthermore, the broadening of the ³¹P signal in MAS NMR (Figure 1) indicates that a rather amorphous structure for material 3 is likely. Indeed, in solid state NMR, the line width is often related to structural heterogeneity (crystalline vs. amorphous structure). Thus, material 3 cannot be considered a COF but probably a POP. ³¹P MAS NMR data, which give almost identical signals for the starting compound 1 (δ ³¹P = 52.7 ppm) and material 3 (δ ³¹P = 49.0 ppm), indicate that the P=S groups and the P-O-Ar groups are preserved during the synthesis of material 3, despite the use of harsh reaction conditions. The reactions occurred far from these groups, i.e., on the aldehydes. Indeed, comparison between the ¹³C MAS NMR spectra of compound 1 and material 3 (Figure 2) displays two important differences: the disappearance of the signal corresponding to the aldehydes (δ ¹³C = 190.9 ppm) and the appearance of a signal at 166.4 ppm, which corresponds to the creation of the hydrazone bonds. Both chemical shifts are typical for such linkages. The disappearance of the aldehydes is also shown using IR as the disappearance of the band at 1660 cm⁻¹ (Figure 3), together with the appearance of a new band at ca 1630 cm⁻¹, corresponding to the creation of the hydrazone bonds. Taken together, these data confirm the obtention of the expected material 3. However, elemental analyses of material 3 display an important deficit in the carbon content (45.03% instead of 53.04%), together with an increase in the hydrogen content (4.38% instead of 3.34%). These differences cannot be due to entrapped solvents or reactants, such as mesitylene, dioxane, and even carboxylic acid, which would have increased the carbon content. Additionally, water, introduced with carboxylic acid and generated by the condensation reaction, can lower the carbon content and increase the hydrogen content. Considering the trapping of three water molecules per macrocycle, corresponding to the quantity of water generated by the condensation reactions, the relative weight of water is about 9%. This value is very close to the first mass loss (ca 10%) observed up to 200 °C in TGA experiments (Figure 4). The main mass loss occurs above 350 °C for TGA organic compounds in air. It should be noted that dendrimers also containing (S)P-O-C₆H₄-CH=N-N linkages, display an analogous thermal behavior in air, even if 20% of the mass is still retained at 700 °C [27], contrary to material 3, for which only about 5% of the mass is retained at this temperature.

An important characteristic of COFs and POPs is their porous properties. For this purpose, BET experiments were carried out to determine the specific surface area (Figure 5). Deceptively and surprisingly, the measured surface area of material 3, obtained by adsorption of N₂ at 77.175 K, was very low (3.8542 m²/g). This is due to the difficult or impossible stacking of organized layers induced by the 3-D structure of the phosphorus moieties, as emphasized in the Introduction. Despite this result indicating the quasi absence of pores, we tried to determine whether material 3 could be used for trapping dyes in solution. Indeed, the trapping of dyes can be monitored easily by UV-Vis measurements and can be used, for instance, as a model for trapping water pollutants [28]. We chose four relatively small dyes—three hydrophobic (azulene, pyrene, and fluorenone) and one hydrophilic (Orange G) (Figure 6). The concentrations of the dyes were chosen to have an intensity of between 0.8 and 1 in the UV-Vis spectra. Attempts were first made with azulene in solution in 1,4-dioxane and material 3 (insoluble). However, no difference was detected in the UV-Vis spectrum of azulene with or without material 3. The same experiment was carried out with pyrene and fluorenone using the same solvent, and the same results were obtained in both cases. Another attempt was made with Orange G in water, in which material 3 is also insoluble. Again, no difference was observed using UV-Vis with or without material 3.

To diversify the experiments in water, hydrophobic dyes were considered as other models of pollutants. However, according to their providers, the solubility of these hydrophobic dyes in water is very low: 0.001 g/L for fluorenone [29], 0.015 g/L for azulene [30], and 0.1 g/L for pyrene [31]. An attempt was made using the most soluble compound, pyrene; however, as previously observed for dioxane, it was not possible to ascertain the trapping by material 3.

5. Conclusions

In conclusion, using numerous techniques, we have, for the first time, synthesized and characterized a new type of material that has a thiophosphate as branching points. In fact, with all these results in hand, it is clear that the expected condensation reaction occurred; however, given its non-crystallinity and the low porosity of material 3, its structure should be considered more of a POP or a highly cross-linked polymer than a COF. The absence of crystallinity is probably due to the 3-dimensional structure of compound 1, which is more favorable to a 3-dimensional polymer, than to a stacked 2-dimensional structure of each sheet that is necessary to obtain a COF. Material 3 is a new example of a covalent triazine framework (CTF) and a hyper-crosslinked polymer (HCP).