Natacha
Six
a,
Antonella
Guerriero
b,
David
Landy
c,
Maurizio
Peruzzini
b,
Luca
Gonsalvi
*b,
Frédéric
Hapiot
*a and
Eric
Monflier
a
aUniversité Lille Nord de France, CNRS UMR 8181, Unité de Catalyse et de Chimie du Solide, UCCS UArtois, Faculté Jean Perrin, rue Jean Souvraz, SP18, F-62300 Lens, France. E-mail: hapiot@univ-artois.fr; Fax: +33(0)321791755; Tel: +33(0)321791772
bIstituto di Chimica del Composti Organometallici, Consiglio Nazionale delle Ricerche (ICCOM-CNR), Via Madonna del Piano 10, 50019 Sesto Fiorentino, Firenze, Italy. E-mail: l.gonsalvi@iccom.cnr.it; Fax: +39 055 5225203; Tel: +39 055 5225251
cUniversité Lille Nord de France, UCEIV, ULCO, 145, Avenue Maurice Schumann, MREI 1, F-59140 Dunkerque, France
First published on 11th July 2011
A biphasic catalytic system has been elaborated in which the amphiphilic species concentrations at the aqueous/organic interface could be thermocontrolled by the supramolecular interaction between β-cyclodextrins and the 1-(4-tert-butyl)benzyl-1-azonia-3,5-diaza-7-phosphaadamantyl ligand (1). The system proved efficient in Rh-catalyzed hydroformylations of higher olefins. An increase in the catalytic activity was observed without alteration in either the chemo- or the regio-selectivity. The main advantage of the cyclodextrin/1 couple lies in the rapid decantation of the biphasic system at the end of the reaction. This study represents the first example of thermoregulation of the surface activity of an amphiphilic phosphane by a cyclodextrin.
For instance, we previously reported that the N-benzyl derivative of 1,3,5-triaza-7-phosphaadamantane (=N-Bz-PTA, Scheme 1)28 significantly improved the catalytic activity in a rhodium-catalyzed hydroformylation of terminal alkenes due to its surface-active character.29 However, stable emulsions were often obtained, giving N-Bz-PTA a limited interest.
![]() | ||
Scheme 1 N-Benzyl PTA ammonium salt. |
In this context, a new approach has been developed that solves the emulsion problem. We have imagined that the amphiphilic character of a phosphane could be masked or revealed depending on the temperature by addition in the medium of a water-soluble supramolecular receptor (Scheme 2). More precisely, we anticipated that total or partial inclusion of the hydrophobic part of an amphiphilic phosphane into the hydrophobic cavity of a supramolecular receptor could be thermocontrolled. Knowing that the proportion of phosphane included in the receptor should decrease when increasing the temperature, a higher concentration of an amphiphilic phosphane at the aqueous/organic interface is expected at high temperature with beneficial effects on the catalytic performances. In contrast, decreasing the temperature should lead to a lower proportion of phosphane at the interface resulting in an easier decantation between the aqueous and organic phases.
![]() | ||
Scheme 2 The principle of thermocontrolled catalysis using a receptor/phosphane couple. P: amphiphilic phosphane. Cat.: organometallic catalyst. |
The proof of concept was implemented using native β-cyclodextrin (β-CD) or randomly methylated β-cyclodextrin (RAME-β-CD) as a supramolecular receptor and the PTA-based amphiphilic phosphane 1 capable of interacting with the β-CD cavity (Scheme 3). The consequences of the CD/phosphane interaction have been evaluated in a rhodium-catalyzed hydroformylation of higher olefins.
![]() | ||
Scheme 3 Cyclodextrins and the water-soluble ligand 1. |
We show below that the thermocontrol of the CD/phosphane interaction can advantageously improve both the phase separation and the catalytic activity without affecting either the chemo- or regio-selectivity of the hydroformylation reaction.
Surface-tension measurements revealed the surface active character of 1. Indeed, a regular decrease in the interfacial tension γ was observed when increasing the concentration (Fig. 1), indicative of the adsorption of 1 at the air/water interface. Its behaviour was similar to that of N-Bz-PTA. No critical micellar concentration (cmc) could be detected in the studied concentration range. The insolubility of 1 beyond 25 mM at room temperature prevented getting information on its surface activity at higher concentration. Indeed, solid particles were observed in water above this concentration. However, the impact of 1 on γ at low concentrations clearly illustrated its ability to adsorb at the aqueous/air interface. Interestingly, addition of RAME-β-CD (1 equiv.) led to a higher water solubility of 1 (53 mM). Moreover, addition of RAME-β-CD onto an aqueous solution of 1 (50 mM) resulted in an increase in γ to an intermediate value (57.2 mN m−1) between those of pure 1 (54.6 mN m−1) and pure RAME-β-CD (58.0 mN m−1). The effect was even more marked with addition of excess native β-CD onto an aqueous solution of 1 (20 mM) as the surface tension value of pure native β-CD was recovered (72 mN m−1). These observations demonstrated that RAME-β-CD and the native β-CD can conceal the amphiphilic character of 1. Note that no effect on the surface tension was noticed when RAME-β-CD was added onto an aqueous solution of N-Bz-PTA, thus highlighting the essential role of the tert-Bu group. The decrease in the surface activity of 1 in the presence of β-CD or RAME-β-CD was attributed to the formation of inclusion complexes between 1 and CD.
![]() | ||
Fig. 1 Surface tension of N-Bz-PTA and 1 in water at 22 °C. |
The stoichiometry of these inclusion complexes has been determined by NMR at room temperature and 80 °C (temperature of the catalytic experiments) using the method of continuous variation (Job plot).32 For each complex and regardless of the temperature, the symmetrical curves (host/(host + guest) = 0.5) were indicative of a 1:
1 stoichiometry. As an example, the Job plot derived from the tert-butyl protons variation for native β-CD/1 mixtures is depicted in Fig. 2. The association constant was determined by isothermal titration calorimetry (ITC, see Experimental) at 80 °C, the lowest temperature used in catalysis.
![]() | ||
Fig. 2 Continuous variation plot (Job plot) derived from chemical shift variations of the tert-Bu signal on 1H NMR spectra of β-CD/1 mixtures in D2O at 25 °C. Δδ(1) = δtBu obs. − δtBu init.. The total concentration of species was 10 mM. |
While no association constants could be measured between RAME-β-CD and N-Bz-PTA at 80 °C, the RAME-β-CD/1 and native β-CD/1 interactions remained strong (Kass = 5440 and 4930 M−1, respectively). Note that the RAME-β-CD/1 association constant determination at 100 °C and 120 °C (highest temperature used in catalysis) could not be performed for experimental reasons (ITC measurements require an aqueous phase under atmospheric pressure). 2D T-ROESY NMR experiments confirmed the recognition process between the CD “host” and the phosphane “guest”. Actually, at room temperature and 80 °C, correlations were detected in the 2D T-ROESY spectra of equimolar mixtures of RAME-β-CD or native β-CD and ligand 1. Cross-peaks between the tert-butyl protons and the H-3, H-5 and H-6 CD protons were indicative of a deep inclusion of the tert-butyl group in the CD cavity. As an example, the 2D T-ROESY spectrum of a native β-CD/1 equimolar mixture is depicted in Fig. 3.
![]() | ||
Fig. 3 Partial 2D T-ROESY NMR spectrum (tBu protons) of an equimolar mixture of native β-CD and 1 at room temperature in D2O (mixing time = 300 ms, species concentration = 10 mM). |
The direction of inclusion of 1 (from the primary or secondary face of the CD) could be precisely determined as dipolar contacts could be detected between the aromatic protons and the H-3 CD protons, indicative of an inclusion from the secondary face (Fig. 4). From the T-ROESY experiment, it follows that the PTA group protruded beyond the CD cavities as no contact was detected between PTA protons and CD inner protons. The tert-Bu group appears to be essential for the recognition process to take place. Actually, no contact could be detected between native β-CD or RAME-β-CD and N-Bz-PTA. One can then easily understand why addition of these CDs did not affect the surface tension of an aqueous solution of N-Bz-PTA.
![]() | ||
Fig. 4 Partial 2D T-ROESY NMR spectrum (aromatic protons) of an equimolar mixture of native β-CD and 1 at room temperature in D2O (mixing time = 300 ms, species concentration = 10 mM). |
Knowing that 1 has a strong affinity for the RAME-β-CD and native β-CD cavities, we then carried out catalytic experiments to evaluate the performances of 1 in hydroformylation of hydrophobic higher olefins. The reactions were performed at various temperatures under 50 bar CO/H2 for 6 h. The results are gathered in Table 1. At 80 °C without CD, 1 appeared to be an effective ligand as the conversion was close to that obtained with N-Bz-PTA in hydroformylation of 1-decene (26 vs. 19%, respectively, entries 5 and 1). Note that 1 was much more effective than the TPPTS ligand for which a very low 3% conversion was measured under the same experimental conditions.29
Entry | Higher olefin | Phosphane | Cyclodextrin | T/°C | Conv.b (%) | Chemoselectivityc | l/b Ratiod |
---|---|---|---|---|---|---|---|
a Experimental conditions: Rh(acac)(CO)2 (4.07 × 10−2 mmol), water-soluble ligand (0.21 mmol), cyclodextrin (0.48 mmol), H2O (11.5 mL), 1-alkene (20.35 mmol), 1500 rpm, CO/H2 (1/1): 50 bar, 6 h. b Calculated with respect to the starting olefin. c (mol. of aldehydes)/(mol. of converted olefins) × 100. The side products were mainly isomeric olefins. d Ratio of the linear to branched aldehyde product. e Recycling of the aqueous phase of entry 9. f Recycling of the aqueous phase of entry 10. g This kinetic profile of this experiment is given in ESI.† | |||||||
1 | 1-Decene | N-Bz-PTA | — | 80 | 19 | 98 | 1.9 |
2 | 1-Decene | N-Bz-PTA | RAME-β-CD | 80 | 24 | 98 | 1.9 |
3 | 1-Decene | N-Bz-PTA | — | 100 | 77 | 99 | 1.8 |
4 | 1-Decene | N-Bz-PTA | RAME-β-CD | 100 | 86 | 96 | 1.8 |
5g | 1-Decene | 1 | — | 80 | 26 | 99 | 2.1 |
6g | 1-Decene | 1 | RAME-β-CD | 80 | 10 | 97 | 1.8 |
7 | 1-Decene | 1 | Native β-CD | 80 | 9 | 97 | 1.9 |
8g | 1-Decene | 1 | — | 100 | 56 | 99 | 1.7 |
9g | 1-Decene | 1 | RAME-β-CD | 100 | 72 | 99 | 2.0 |
10e | 1-Decene | 1 | RAME-β-CD | 100 | 71 | 99 | 1.9 |
11f | 1-Decene | 1 | RAME-β-CD | 100 | 66 | 97 | 1.9 |
12 | 1-Decene | 1 | Native β-CD | 100 | 67 | 95 | 1.8 |
13 | 1-Decene | 1 | — | 120 | 94 | 99 | 1.9 |
14 | 1-Decene | 1 | RAME-β-CD | 120 | 98 | 99 | 1.8 |
15 | 1-Dodecene | 1 | — | 80 | 39 | 96 | 1.5 |
16 | 1-Dodecene | 1 | RAME-β-CD | 80 | 11 | 99 | 1.7 |
17 | 1-Dodecene | 1 | Native β-CD | 80 | 8 | 98 | 1.7 |
18 | 1-Dodecene | 1 | — | 100 | 49 | 98 | 1.5 |
19 | 1-Dodecene | 1 | RAME-β-CD | 100 | 75 | 99 | 1.7 |
20 | 1-Dodecene | 1 | Native β-CD | 100 | 71 | 98 | 1.6 |
21 | 1-Tetradecene | 1 | — | 80 | 53 | 93 | 1.7 |
22 | 1-Tetradecene | 1 | RAME-β-CD | 80 | 14 | 95 | 1.6 |
23 | 1-Tetradecene | 1 | Native β-CD | 80 | 11 | 98 | 2.1 |
24 | 1-Tetradecene | 1 | — | 100 | 57 | 97 | 1.5 |
25 | 1-Tetradecene | 1 | RAME-β-CD | 100 | 88 | 97 | 1.9 |
26 | 1-Tetradecene | 1 | Native β-CD | 100 | 66 | 98 | 1.7 |
Increasing the temperature to 100 or 120 °C logically led to an increase in the conversion (entries 8 and 13). Also note that, the conversions being of the same order of magnitude whatever the substrate, the catalytic system is truly an interfacial process. Very different conversions would have been observed for a reaction in the bulk water phase due to the difference in higher olefins solubility. Astonishingly, at 80 °C, addition of excess RAME-β-CD led to a drop in catalytic activity (from 26 to 10% conv., entries 5 and 6). Similar results were obtained with 1-dodecene (from 39 to 11% conv., entries 15 and 16) and 1-tetradecene (from 53 to 14% conv., entries 21 and 22). Conversely, at 100 and 120 °C, addition of excess RAME-β-CD resulted in a marked increase in catalytic activity. At 100 °C, the conversion increased from 56 to 72% for 1-decene (entries 8 and 9), from 49 to 75% for 1-dodecene (entries 18 and 19) and from 57 to 88% for 1-tetradecene (entries 24 and 25). Thus, RAME-β-CD impacted upon the conversion either positively or negatively depending on the reaction temperature. In contrast, catalytic tests relative to the CD non-interacting ligand N-Bz-PTA showed a different behaviour. Indeed, an increase in the conversion was observed upon addition of RAME-β-CD whatever the temperature (entries 1–4).
The explanation of the above results lies in the strength of the RAME-β-CD/1 interaction depending on the temperature. At 80 °C, the surface active character of 1 and its adsorption ability at the aqueous/organic interface were masked due to inclusion of the hydrophobic part of 1 in the RAME-β-CD cavity (Kass = 5440 M−1).
Hence, most of the catalytically active species were stuck in aqueous solution. Only a few of these catalytic species were present at the aqueous/organic interface resulting in an activity decrease. In contrast, at 100 or 120 °C, the markedly higher activities were indicative of a change in the behaviour of the catalytic system components. Though the RAME-β-CD/1 association constant determination could not be performed at 100 °C (see above), it is well known that the association constant values decrease with the temperature.33 Hence, at 100 °C, we postulate that the concentrations of the “free” ligand (non-included 1) increased at the aqueous/organic interface. The surface activity of 1 being revealed, an increase in the catalytic activity at the aqueous/organic interface was then observed. Additionally, the RAME-β-CD/1 dissociation also contributed to increase the “free” RAME-β-CD concentration. More “free” RAME-β-CDs were then available to promote the mass transfer by inclusion of higher olefins in their cavity,23 thus participating in the catalytic activity enhancement (Scheme 2).
The implication of both the ligand surface activity and RAME-β-CD properties has been indirectly confirmed when replacing RAME-β-CD by the native β-CD. At 80 °C, the native β-CD/1 association constant (Kass = 4930 M−1) was similar to that of RAME-β-CD. Hence, as described above for RAME-β-CD, the native β-CD could act as an inhibitor of the ligand surface activity. However, the native β-CD could not be considered as a mass transfer promoter as it did not adsorb at the aqueous/organic interface.34 Consequently, both these features could be translated in terms of catalytic activity. While only a slight difference in conversion was noticed between the native β-CD and RAME-β-CD at 80 °C (11 vs. 14% conv. for 1-tetradecene for example), a significant variation was measured at 100 °C (66 vs. 88% conv.). The 22% conv. gap between both CDs at 100 °C was relative to the mass transfer contribution of RAME-β-CD.
Besides catalytic aspects, the interest of the RAME-β-CD/1 couple lies in the easy recovery of both the aqueous and the organic phases after 6 h reaction time. While an emulsion was observed for the RAME-β-CD/N-Bz-PTA couple due to the impossibility to conceal the surface activity of N-Bz-PTA by the formation of an inclusion complex, a clear phase separation was obtained for 1. More precisely, the decantation was about hours with N-Bz-PTA and about seconds with 1. In fact, decreasing the temperature at the end of the reaction led to inclusion of 1 in the RAME-β-CD cavity (Scheme 4). The subsequent loss in surface activity resulted in a rapid decantation. Thus, by supramolecularly controlling the surface activity, the products and the rhodium catalyst can be easily recovered in two distinct phases. Indeed, the aqueous phase has been recycled twice in hydroformylation of 1-decene in the presence of RAME-β-CD with no significant loss either in conversion or selectivities (entries 10 and 11).
![]() | ||
Scheme 4 CD-based thermocontrolled hydroformylation of higher olefins. |
The chemoselectivity of the reaction, for its part, was independent of the length of the terminal alkene. Actually, increasing the length of the hydrocarbon chain from ten to fourteen carbons did not significantly modify the aldehydes proportions. A similar trend was observed for the regioselectivity as no significant variation in the l/b ratio was observed whatever the olefin. Interestingly, note that the chemo- and regio-selectivities were independent of the presence or absence of RAME-β-CD suggesting that the catalytically active species proved to be stable under these conditions. This observation is opposite to what was previously observed with TPPTS.35
Footnote |
† Electronic supplementary information (ESI) available: Job plots and 2D T-ROESY spectra of β-CD/1 and RAME-β-CD/1 inclusion complexes at various temperatures. See DOI: 10.1039/c1cy00156f |
This journal is © The Royal Society of Chemistry 2011 |