Potassium exchanged zirconium hydrogen phosphate Zr(O₃POK)₂: a heterogeneous basic catalyst for Knoevenagel reaction without solvent

Abstract

A simple work-up to prepare crystalline potassium exchanged zirconium hydrogen phosphate Zr(O₃POK)₂ is proposed. In addition the synthesis of amorphous potassium exchanged zirconium hydrogen phosphate Zr(O₃POK)₂, obtained from an amorphous sample of Zr(O₃POH)₂·H₂O is reported. All these exchanged materials Zr(O₃POK)₂ are powerful catalysts for the Knoevenagel reaction. Furthermore the absence of solvent increases the rate of the reaction. Products can be separated from the catalyst by sublimation thus avoiding the use of solvent in the purification step. The catalyst, which is thus easily recovered, shows no appreciable loss of activity when recycled three times.

---

Green Context

Heterogeneous basic catalysts are of importance in the fine chemicals area, and several have been investigated in reactions such as the Knoevenagel reaction, an important reaction with pronounced solvent dependency. This paper deals with a simplified preparation of one such catalyst, and shows that it is active in the condensation of aromatic aldehydes with C-acids. Of particular interest is the ability to run the reactions without solvent, and to isolate the products by sublimation, so also avoiding the need for solvents in the purification step.

DJM

Introduction

The use of heterogeneous catalysts allows a simplification of the purification step to a simple filtration, separating the catalyst from the reaction media. Heterogeneous basic catalysis is an area of growing interest which has been recently reviewed.¹ The structure of inorganic catalysts may be amorphous (silica, alumina) or organised (Al-MCM, zeolite, clay, etc.).² Even in the latter class of inorganic basic catalysts, their structures are frequently unknown on the atomic scale. This increases the difficulty of characterising the accurate structure of the catalytic centre. Metal–phosphate materials M(O₃POH)_n are a class of compounds that has been widely studied as proton conductors,³ ionic exchangers,⁴ catalysts⁵ or as supports for ionic chromatography.⁶ The protons of these materials may be exchanged to yield potentially basic materials. Furthermore the structure of these exchanged metal–phosphate materials may be determined at the atomic level. The potassium exchanged zirconium hydrogen phosphate Zr(O₃POK)₂ has already been used as a basic catalyst for Michael addition,⁷ Henry reaction,⁸ cyanosilylation of carbonyl compounds⁹ and more recently for the desilylation of phenol silyl ethers.¹⁰ Zr(O₃POK)₂, is a layered material which possesses ionic oxygen pointing towards the interlayer space. This compound may be obtained either from zirconium hydrogen phosphate Zr(O₃POH)₂·H₂O via an exchange with potassium hydroxide⁷ or directly from a solid-phase synthesis at high temperature.¹¹

The Knoevenagel condensation is a useful reaction for the generation of double bonds from a carbonyl compound and an active methylene compound. Many homogeneous catalysts have been used to promote this reaction (triethylamine, pyridine–TiCl₄, secondary amines etc…).¹² More recently the use of heterogeneous basic catalysts have been employed to facilitate the separation of the catalyst from the reaction media. Furthermore in many cases the catalyst may be reused. Metal oxides supported on silica gel, barium hydroxide, zeolites, montmorillonite, magnesium and aluminium oxides, hydrotalcite,¹³ Cs-MCM-41,¹⁴ modified silicagel,¹⁵ and more recently Ga/Al-containing layered double hydroxides¹⁶ are just a few examples of effective materials which may be used for the Knoevenagel condensation. In some cases, the reaction can be achieved under dry conditions thus decreasing both the cost of the synthesis and the amount of waste flow.¹⁷

Here, a simplification of the literature procedure to synthesise crystalline Zr(O₃POK)₂ is reported as well as the synthesis of amorphous samples of Zr(O₃POK)₂. The basic properties of these materials are used to catalyse Knoevenagel condensations.

Results and discussion

The synthesis of potassium exchanged zirconium phosphate Zr(O₃POK)₂ according to the literature [from Zr(O₃- POH)₂·H₂O] is quite laborious as it needs several steps. Nevertheless this method is more convenient for organic chemists, compared to the solid-phase synthesis,¹¹ since it does not need special equipment (high temperature oven). The first step, as reported in the literature, is to increase the specific surface area of the parent material Zr(O₃POH)₂·H₂O. Exfoliation with n-butylamine followed by a reprecipitation leads to an increase in the specific surface of the parent material from 0.5 to 17 m² g⁻¹. In order to reduce the number of steps to produce Zr(O₃POK)₂ we have utilized crystalline α-zirconium hydrogen phosphate directly in the exchange process. α-Zirconium hydrogen phosphate Zr(O₃POH)₂·H₂O 1a was obtained according to a modification of the literature procedure¹⁸ (the BET specific surface area of Zr(O₃POH)·H₂O thus obtained was 7.8 m² g⁻¹). In order to produce a large amount of Zr(O₃POK)₂2a the concentration of the KOH and KCl solutions in ultra-pure water was dramatically increased (1 M instead of 0.1 M). Thus 8 g of Zr(O₃POK)₂2a were produced with a reasonable volume of solution. The X-ray powder diffraction pattern of Zr(O₃- POK)₂ thus synthesised is compared (Fig. 1) to the structure obtained by Dörffel and Liebertz¹¹ from a single-crystal X-ray diffraction study. The only notable differences concern the cell parameters which are slightly larger in our study (average 0.18%; see Experimental section). As expected from the crystal structure, a singlet is observed for the ³¹P MAS NMR spectrum of Zr(O₃POK)₂ (Fig. 2) indicating the equivalence of all the phosphorus atoms in the structure and thus demonstrating the efficiency of the exchange. The chemical shift of Zr(O₃POK)₂2a is displaced towards low field (δ_iso = −11.4 ppm) compared to the parent material Zr(O₃POH)₂·H₂O (δ_iso = −18.7 ppm). Previous work describing the high-resolution ³¹P MAS NMR in some antimony phosphates reported that the ³¹P chemical shift (δ_iso) was closely dependent on the P–O bond length.¹⁹ In compounds which possess a non-coordinated oxygen, the shorter the P–O bond is the larger the shielding effect is on the phosphorus. It is of note that in the present example, that the opposite behavior is observed. Indeed, the P–O bond length (the bond pointing to the interlayer space) in compounds 1a and 2a are 1.542 Ų⁵ and 1.508 Å,¹¹ respectively. This behaviour could arise from the three P–O bonds connected to zirconium that are longer in compound 2a (1.548 Å) than in compound 1a (average: 1.515 Å). The excellent thermal stability of Zr(O₃- POK)₂2a, which is stable up to 1200 °C as reported earlier,⁷ was confirmed by TDA analysis which revealed no transition upto 1000 °C except a transition at 150 °C due to the loss of residual water [Fig. 5(c)]. FTIR analysis [Fig. 6(a)] clearly indicates two P–O stretching vibrations: one at 1174 cm⁻¹ arising from the shortest P–O bond (1.508 Å) and one at 976 cm⁻¹ which may be attributed to the three P–O bonds with a length of 1.548 Å. It is worth noting that compound 2a absorbs water from the atmosphere leading to an increase in the broad band at 3450 cm⁻¹ thus decreasing the resolution of the two bands at 974 and 1174 cm⁻¹. The two bands at 552 and 614 cm⁻¹ disappear in the amorphous sample 2c supporting the assumption that these two bands correspond to the vibrations of the network that can be observed only in a well organised structure.

Fig. 1 Calculated (×), observed (—) and difference X-ray profiles for Zr(O₃POK)₂. Reflection marks are shown for the P3̄ space group.

Fig. 2 ³¹P MAS NMR spectra (12 kHz, NS = 4; d1 = 60 s): (a) Zr(O₃POH)₂·H₂O 1a: (b) Zr(O₃POK)₂2a.

The amorphous samples of Zr(O₃POK)₂, 2b and 2c, were obtained by an exchange process of Zr(O₃POH)₂·H₂O 1b and 1c in the presence of KCl and KOH aqueous solutions. Compounds 1b and 1c were synthesised by the reaction of ZrOCl₂·8H₂O with H₃PO₄ in water without HF and were obtained, owing to the different conditions of reaction (1b: 14 h at reflux; 1c, 4 h at room temperature), with different specific surface areas (BET), 9 and 226 m² g⁻¹, respectively. The specific surface area (BET) of the potassium exchanged material 2b is 7 m² g⁻¹ while that of compound 2c is dramatically reduced (55.9 m² g⁻¹) compared to the specific surface area of the parent material 1c (226 m² g⁻¹). The FTIR spectra of these exchanged amorphous samples [Fig. 6(c)] clearly indicates the bands corresponding to the P–O stretching (950–1180 cm⁻¹). Figs. 3 and 4 show the ³¹P MAS NMR of the parent materials Zr(O₃POH)₂·H₂O 1b and 1c and materials 2b and 2c resulting from the exchange of the hydrogen atoms by potassium. In agreement with previous work,²⁰ the amorphous samples of Zr(O₃POH)₂·H₂O 1b and 1c show several signals while the crystalline sample simply shows a singlet (Fig. 2). Both the relative intensities of the signals and the specific surface area depend on the advancement of the crystallisation. The chemical shifts of the exchanged compounds 2b and 2c are displaced towards lower field as found for the crystalline sample 2a (Fig. 2) indicating the efficiency of the exchange.

Fig. 3 ³¹P MAS NMR spectra (12 kHz): (a) Zr(O₃POH)₂·H₂O 1b; (b) Zr(O₃POK)₂2b (peaks marked with * are rotation bands).

Fig. 4 ³¹P MAS NMR spectra (12 kHz): (a) Zr(O₃POH)₂·H₂O 1c; (b) Zr(O₃POK)₂2c (peaks marked with * are rotation bands).

Fig. 5 DTA analyses under argon (heating rate: 10 °C min⁻¹): (a) compound 2c; (b) compound 2b; (c) compound 2a.

Fig. 6 FTIR spectra of Zr(O₃POK)₂ in KBr: (a) anhydrous crystalline sample 2a; (b) wet crystalline sample 2a (sample left in an open flask for 1 month); (c) anhydrous amorphous sample 2c (peak marked as * is an artefact due to the presence of CO₂).

Thermal analysis of these two amorphous exchanged materials 2b and 2c (DTA) reveals an endothermic transition from 80 to 250 °C which is due to the loss of residual water [Fig. 5(a) and 5(b)]. Two (740 and 795 °C) and three (545, 701 and 722 °C) exothermic transitions, respectively are observed for compounds 2c [Fig. 5(a)] and 2b [Fig. 5(b)]. Partial hydrogen exchange in compound 2b could explain its thermal behaviour. Indeed, the first transition (545 °C) could be attributed to the condensation of phosphate groups into pyrophosphates as was previously reported for the non-exchanged materials Zr(O₃POH)₂ (transition observed at 550 °C²⁰). The two transitions observed for compounds 2b (701 and 722 °C) and 2c (740 and 795 °C), that have almost the same relative intensity, could arise from the same transformation. These transitions were previously reported for a copper salt intercalated in α-zirconium phosphate.²¹ The different position of these transitions could arise from a partial exchange in compound 2b. Furthermore the absence of a transition around 550 °C for compound 2c indicates a fully exchanged material.

Zr(O₃POK)₂ was first tested as a basic catalyst for the Knoevenagel condensation of piperonal and Meldrum’s acid. According to the kinetic data obtained using the amorphous catalyst 2b (Table 1, entries 1 and 3), THF was the best solvent while the use of ethanol (Table 1, entry 5) inhibited the reaction. The superiority of THF over CH₂Cl₂ may be explained by the fact that Knovenagel condensation produces one mol of water per mol of product. Owing to the non-miscibility of water and dichlorormethane and taking into account the polar character of the catalyst, water produced by this condensation could be readily localised on the catalyst when dichloromethane is used as a solvent. Thus the efficiency of the catalyst could be affected. Nevertheless, when the reaction is carried out without solvent the reaction time is even shorter (Table 1, entries 7 and 9). Noteworthy is that the reaction times and yields are identical whatever the structure (amorphous or crystalline) of the catalyst used for the reaction (Table 1, entries 2 and 4). This is the first example of the use of amorphous potassium exchanged zirconium phosphate as a basic catalyst. In absence of solvent a total conversion is reached after only 14 minutes. Interestingly, no difference (yield, reaction times) was observed by the use of microwave heating or classical heating (isomantle). Nevertheless for the expediancy of the procedure, microwave heating (commercial microwave oven Whirlpool) was chosen. The powder X-ray diffraction of the catalyst after the condensation of piperonal with Meldrum’s acid, and before the elimination of the product, did not reveal any modifications of the d₀₀₁ peak (9.02 Å), indicating that neither the substrate nor the product are present in the interlayer space. This observation indicates that the catalytic process is only located on the surface of the catalyst. Nevertheless the kinetics of the Knoevenagel condensation of piperonal with Meldrum’s acid (experiments carried out in THF, Table 1 entries 2–4) did not show any significant difference between catalysts 2a, 2b and 2c despite their different specific surface areas.

Table 1 Knoevenagel condensation of piperonal to Meldrum’s acid

Taking into account the fact that reactions carried out without solvent contribute to a reduction in costs and waste flow it is of note that all the following reactions were achieved without solvent. The mixing step of the catalyst with substrates can affect the reproducibility of the results. The first method, which is completely solvent-free, is to mechanically mix for the catalyst with the substrates for several minutes. The second method is less laborious and consists of dissolving the substrates in a minimum of solvent followed by the addition of this solution to the catalyst. Finally the solvent is removed in vacuo leading to the substrates being absorbed on the catalyst. Although the results of these two methods were identical, we chose the second method due to the rapidity of the work up. Selected data corresponding to the condensation of piperonal with different active methylene compounds are reported in Table 2. The mechanism of the Knoevenagel condensation (under basic conditions) is that one proton from the acidic methylene compound is removed by the base to generate a nucleophile that will react with the aldehyde. In view of the results of Table 2, it appears that the catalyst 2a is efficient for methylene compounds which possess a pK_a (determined in water) in the range 4–12. Surprisingly, the use of tetronic acid as the methylenic acid compound (Table 2, entry 6 ; pK_a = 3.76) does not yield the condensation product. Despite the coloration of the catalyst almost all the piperonal is recovered; this coloration might be due to the auto-condensation of tetronic acid. Despite the lack of reactivity of tetronic acid it is worth mentioning that other methods are able to catalyse the aldol type condensation of tetronic acid with aldehyde in a 1∶1²² or 2∶1²³ ratio. On the other hand, catalysts 2a are not sufficiently basic to be efficient when methylenic compounds possessing a pK_a > 13 are used. To illustrate the reusability of the catalyst, catalyst 2a was engaged three times in the condensation of piperonal with Meldrum’s acid (Table 3). After each reaction the product was separated from the catalyst by sublimation. After three cycles the catalyst was still efficient as reported in Table 3.

Table 2 Condensation of piperonal to methylene acid compounds using microwave heating (450 W, 7 × 2 min)

Table 3 Illustration of the reusability of catalyst 2a

Conclusion

The present article deals with the simplification of the synthesis of crystalline potassium exchanged zirconium hydrogen phosphate 2a. Furthermore the synthesis of the amorphous equivalent of compound 2a is described. The basic properties of these materials, which are readily available in large amounts, were used to catalyse the Knoevenagel condensation. Interestingly, both amorphous and crystalline potassium exchanged zirconium hydrogen phosphate are able to catalyse the Knoevenagel reaction. These catalysts are efficient for methylenic compounds possessing pK_a values in the range of 4–12. It is noteworthy the efficiency of these catalysts is increased by the absence of solvent. The yields reported and the reaction times are identical to or better than those previously reported. Furthermore the catalyst is easily recovered and can be reused without significant loss of its efficiency.

Experimental

Catalyst preparation

α-Zr(O₃POH)₂·H₂ O 1a. A modification of the method of slow decomposition of a zirconium fluoro complex was used.²⁴ 8.2 ml of 40% aqueous HF solution (188 mmol) was added to a solution of ZrOCl₂·8H₂O (15.01 g, 46.6 mmol) in 217 ml of ultra-pure water placed in a 1 l round bottomed-flask fitted with an air condenser. 75% aqueous H₃PO₄ (163.3 ml; 1.97 mol) was added to this solution. The homogeneous solution was heated at 90–95 °C for 7 days. The precipitate was recovered by filtration, washed with ultra-pure water (150 ml), acetone (20 ml) and dried at room temperature for 48 h. According to powder X-ray diffraction, the structure of the product is in excellent agreement with literature data.²⁵³¹P MAS NMR (12 KHz; d1 = 60 s; NS = 4): −18.7 ppm (lit.,²⁶ −18.7 ppm); BET surface area 7.8 m² g⁻¹ (lit.,²⁷ 0.5 m² g⁻¹).

Amorphous zirconium phosphate Zr(O₃POH)₂·H₂O 1b. The literature procedure²⁸ was modified as follows: a solution of ZrOCl₂·8H₂O (10 g, 31 mmol) in water (55 ml) was added to a solution of 85% aqueous H₃PO₄ (7.15 g; 62 mmol) in 30 ml of water. The solution was refluxed for 14 h, the solid filtered off, and washed with ultra-pure water (50 ml), acetone (20 ml) and dried at 100 °C for 5 h to yield a white solid (9.3 g; 100%). The specific surface area (BET) of the resulting compound was 9.0 m² g⁻¹.

Amorphous zirconium phosphate Zr(O₃POH)₂·H₂O 1c. The literature procedure²⁵ was modified as follows: a solution of ZrOCl₂·8H₂O (10 g, 31 mmol) in water (55 ml) was added to a solution of 85% aqueous H₃PO₄ (7.15 g; 62 mmol) in water (30 ml). The solution was stirred at room temperature for 4 h, the solid filtered off, and washed with ultra-pure water (50 ml) and acetone (20 ml) and then dried at 100 °C for 10 h to yield a white solid (9.3 g; 100%). The specific surface area (BET) of the resulting compound was 226.0 m² g⁻¹.

Crystalline potassium exchanged zirconium phosphate Zr(O₃POK)₂ 2a. A modification of the method described by Costantino et al.⁷ was used: 50 ml of KOH (1 M in ultra pure water) and 50 ml of KCl (1 M in ultra pure water) was added to 6.9 g of crystalline α-Zr(O₃POH)₂·H₂O 1a (21.6 mmol). After 24 h of slow stirring KCl (7 ml; 1 M) and KOH (7 ml; 1 M) were added. The heterogeneous solution was stirred for a further 6 days. The precipitate was filtered off, washed with potassium hydrogenphosphate (20 ml; 1 M in ultra pure water) and dried at room temperature for 2 days. The product was dried further at 190 °C for 24 h to yield 7.75 g of white solid (100%). The specific surface area (BET) of the product was 11.0 m² g⁻¹. ³¹P MAS NMR (12 kHz ; d1= 60s ; NS = 4) δ_iso = −11.3 ppm; powder X-ray diffraction (see Fig. 1): trigonal, space group P3̄ a = 5.188(9) [lit.,¹¹a = b = 5.176(1)], c = 9.022(2) [lit.,¹¹c = 9.012(8)], number of reflections: 162; GOF = 1.7, R_p = 0.067; R_Wp = 0.093; R_I = 0.08.

Amorphous potassium exchanged zirconium phosphate Zr(O₃POK)₂ 2b and 2c. The same procedure described for compound 2a was used. The starting compounds were 1b and 1c, respectively. The specific surface area of the exchanged materials 2b and 2c were 7.0 and 55.9 m² g⁻¹, respectively.

Knoevenagel condensations

In a typical experiment 200 mg of compound 2a were added to a solution of piperonal (75 mg; 5 mmol) and Meldrum’s acid (72 mg ; 5 mmol) in dichloromethane (2 ml) placed in a 100 ml round bottomed flask. The solvent was removed under reduced pressure. The round bottomed flask was placed in a microwave oven (Whirlpool, M430, Jet 900) and irradiated (450 W) for a total of 14 min and in 2 min intervals. After cooling, the crude product was purified by sublimation (10 mTorr, isomantle at a temperature between 180 and 220 °C) to give 5-(3,4-methylenedioxyphenyl)methylene-2,2-dimethyl[1,3]dioxane-4,6-dione (0.14 g; 100%). mp 179 °C (lit.,²⁹ 178.5–179.5 °C). The results of other experiments are summarised in Table 2.

Acknowledgements

We thank Valérie Montouillout and Christian Fernandez (Laboratoire de Catalyse et Spectroscopie, L.C.S. UMR CNRS 6506) for assistance in the solid-state NMR experiments.

References

1. J. M. Fraile, J. I. Garcia and J. A. Mayoral, Catal. Today, 2000, 57, 3.
2. A. Corma, Chem. Rev., 1997, 97, 2373.
3. G. Alberti, M. Casciola, S. Cavalaglio and R. Vivani, Solid State Ionics, 1999, 125, 91.
4. A. Clearfield, Chem. Rev., 1988, 88, 125.
5. A. Clearfield and D.S. Thakur, Appl. Catal., 1986, 26, 1.
6. L. Maya, J. Inorg. Nucl. Chem., 1981, 43, 400.
7. U. Costantino, F. Marmottini, M. Curini and O. Rosati, Catal. Lett., 1993, 22, 333.
8. U. Costantino, M. Curini, F. Marmottini, O. Rosati and E. Pisani, Chem. Lett., 1994, 2215.
9. M. Curini, F. Epifano, M.C. Marcotullio, O. Rosati and M. Rossi, Synlett, 1999, 315.
10. M. Curini, F. Epifano, M.C. Marcotullio, O. Rosati, M. Rossi and A. Tsadjout, Synth. Commun., 2000, 30, 3181.
11. M. Dörffel and J. Liebertz, Z. Kristallogr., 1990, 193, 155.
12. L. F. Tietze and U. Beifuss, in Comprehensive Organic Synthesis, ed. B. M. Trost and I. Fleming, Pergamon Press, Oxford, 1991, vol. 2, p. 341..
13. A. Corma, S. Iborra, J. Primo and F. Rey, Appl. Catal., 1994, 114, 215.
14. K. R. Kloestra and H. Van Bekkum, J. Chem. Soc., Chem. Commun., 1995, 1005.
15. P. M. Price, J. H. Clark and D. J. Macquarrie, J. Chem. Soc., Dalton Trans., 2000, 101.
16. I. Rousselot, C. Taviot-Guého and J. P. Besse, Int. J. Inorg. Mater., 1999, 165.
17. D. Villemin, Chem. Ind., 1983, 478.
18. G. Alberti and E. Torracca, J. Inorg. Nucl. Chem., 1968, 30, 317.
19. F. Taulelle, C. Sanchez, J. Livage, A. Lachgard and Y. Piffard, J. Phys. Chem. Solids, 1988, 49, 299.
20. H. Benhamza, P. Barboux, A. Bouhaouss, F.-A. Josien and J. Livage, J. Mater. Chem., 1991, 1, 681.
21. C. Ferragia, A. La Ginestra, M. A. Massucci, P. Cafarelli and R. Di Rocco, J. Therm. Anal., 1994, 41, 1469.
22. H. Zimmer, W. W. Hillstrom, J. C. Schmidt, P. D. Seemuth and R. Vögeli, J. Org. Chem., 1978, 43, 1541; D. Villemin and B. Labiad, Synth. Commun., 1990, 20, 3207.
23. L. Wolff and C. Schwabe, Justus Liebigs Ann. Chem., 1896, 291, 231; N. Olthoff, R. Huttenrauch and H. Brauniger, Z. Chem., 1970, 10, 341.
24. G. Alberti and E. Torracca, J. Inorg. Nucl. Chem., 1968, 30, 317.
25. A. Clearfield and G. D. Smith, Inorg. Chem., 1969, 8, 431.
26. N. J. Clayden, J. Chem. Soc., Dalton Trans., 1987, 1877.
27. U. Costantino, F. Marmottini, M. Curini and O. Rosati, Catal. Lett., 1993, 22, 333.
28. A. Clearfield, A. Oskarsson and C. Oskarsson, Ion Exch. Membr., 1972, 1, 91.
29. G. A. Kraus and M. E. Krolski, J. Org. Chem., 1986, 51, 3347.
30. H. F. Ebel, CH-Acidität, (Houben-Weyl), Georg Thieme Verlag, Stuttgart, 1970, vol. 13/1, p. 31..
31. B. B. Corson and R. W. Staughton, J. Am. Chem. Soc., 1928, 50, 2829.
32. Kauffmann, Chem. Ber., 1919, 1435.
33. D. Villemin and B. Labiad, Synth. Commun., 1990, 3333.

---