A.
Behr
* and
G.
Henze
Technische Universität Dortmund Fakultät Bio- und Chemieingenieurwesen, Lehrstuhl für Technische Chemie A, Chemische Prozessentwicklung, Emil-Figgestr. 66, 44227, Dortmund, Germany
First published on 26th October 2010
Carbon dioxide (CO2) is available in almost infinite amounts in our atmosphere and oceans, but its utilisation as feedstock for the chemical industry is often prevented by its thermodynamic stability. Only a few processes based on CO2 as a raw material were realised in a technical scale so far, like the production of urea, methanol or salicylic acid. In the present review mainly catalytic reactions of a lactone platform chemical are described whose production is based on CO2 as a feedstock. The synthesis of this highly functionalised 1 was already invented in the 1970s and was optimised into miniplant‑scale in the meantime. Based on its different functional groups numerous reactions can be carried out starting from this molecule leading to versatile, interesting products: Acids, alcohols or diols as well as aldehydes, amino acids or amines are formed in high yields. Furthermore, esters, silanes or even polymers are obtained using the δ‑lactone as a building block. Thus, by applying efficient catalytic systems which lead to high selectivities a new approach for the utilisation of CO2 as a reasonable feedstock for chemical reactions is described.
Prof. Dr Arno Behr | Prof. Dr Arno Behr is leader of the Chair “Technical Chemisty A” at the Technical University of Dortmund, Germany. He was born in 1952 in Aachen and studied chemistry at the RWTH Aachen where he did his PhD (with award) in the group of Willi Keim in 1979. He did his “Habilitation” at the Institut für Technische Chemie und Petrolchemie at RWTH Aachen with emphasis on carbon dioxide activation by metal complexes, which was published by VCH in 1988. After work at the Henkel Company he moved to Dortmund University in 1996 where he works in carbon dioxide chemistry. |
Dr Guido Henze | Dr Guido Henze was born 1980 in Erwitte, Germany. He studied chemistry from 1999 to 2004 at the Technical University of Dortmund and carried out his diploma thesis on hydroformylation in thermomorphic solvent systems at the Chair of Technical Chemistry A. He did his PhD studies in the group of Prof. Arno Behr and got his doctoral degree (with award) in 2007. His PhD topic was the catalytic functionalisation of the δ-lactone which is formed from butadiene and carbon dioxide. Now he works at BASF SE, Ludwigshafen. |
Furthermore, the public discussion about the impact of CO2 as greenhouse gas strengthens the search for measures for its reduction.4,5 Compared to the enormous amounts of CO2 which are continuously emitted as exhaust gases the demand of the chemical industry for CO2 as a chemical raw material is relatively small. So it has to be doubted that the chemical use of CO2 would have a significantly positive effect on our climate even if the chemical industry would completely substitute oil, natural gas and coal by CO2. Nevertheless, a broader utilisation of CO2 has the potential to preserve fossil resources.
Because of its thermodynamic stability and its high oxidation state, its utilisation has been realised in only few cases: the synthesis of urea, methanol, cyclic carbonates and salicylic acid are the most important processes which could be realised economically in this context. Other processes suffer from the high costs for the activation of CO2.6,7
It seems that catalysis is the best way to overcome these difficulties. Among other palladium-catalysed reactions have gained more and more importance. Inoue et al. and Musco et al. found in the 1970s that a highly functionalised organic intermediate, the δ-lactone 1, can be prepared by homogeneously catalysed telomerisation of butadiene with CO2 using palladium catalysts (Fig. 1).8–10
Fig. 1 Synthesis of the δ-lactone 1. |
In the following decades several investigations were conducted for the explanation of the reaction pathway. Finally the reaction mechanism was investigated by Behr et al. and was described as follows11,12 (Fig. 2): Initially a palladium(0) phosphine complex is formed in situ from a palladium(II) compound and a tertiary phosphine. Through addition of two butadiene molecules a bis-η3-allylkomplex is formed (7) which is balanced with complex 8. Subsequently 8 either reacts to 1,3,7-octatriene (6) or to the allylic carbonate complex 9 by insertion of carbon dioxide into one of the allylic bonds. From 92-ethylidene-6-heptene-5-olide (δ-lactone, 1) is formed by ring closure as well as 2-ethylidene-5-heptene-4-olide (γ-lactone, 2) and its conjugated isomer 2-ethyl-2,4-heptadiene-4-olide (3). As by-products the formation of the aliphatic esters 4 and 5 can be observed.
Fig. 2 Reaction mechanism of the δ-lactone 1 synthesis. |
In the following this reaction was intensively investigated13–30 and was finally optimised and developed to the miniplant scale by Behr et al. So the selectivity could be increased above 95% regarding the δ-lactone by recycling of by-products reaching an overall butadiene conversion of 45%. Additionally all solvent and catalyst loops could be closed yielding a stable process which is ready to be scaled up industrially (Fig. 3).31–33
Fig. 3 Flow scheme of the production process for the δ-lactone 1. |
In the first process step butadiene and carbon dioxide are mixed in a continuously stirred tank reactor (CSTR) together with the selectivity enhancing solvent acetonitrile and the catalyst palladium(II) acetylacetonate/triphenylphosphine. The product mixture is continuously fed into a first thermal separation unit where unreacted carbon dioxide, butadiene and the solvent are removed via the gaseous phase and recycled to the CSTR. Simultaneously, the δ-lactone/catalyst mixture is fed into a second thermal separation unit together with the by-products. In this distillation step the δ-lactone is separated in vacuo as gaseous product while the catalyst is recycled into the CSTR together with the by-products. It was found that the recycling of the by-products lead to an improved selectivity for the δ-lactone due to the chemical equilibrium between the main and by-products. This process was operated continuously over several days.
The resulting δ-lactone 1 is highly functionalised which leads to various reactivities: it has a carboxyl group, an internal carbon-carbon double bond and a further terminal carbon-carbon double bond. Up to now the molecule itself did not find an application. Through further conversion with different bulk chemicals several secondary products of potential industrial relevance have been synthesised mainly catalytically. In this review these reactions of the δ-lactone are summarised (Table 1) demonstrating that CO2 can be efficiently integrated as building block for many organic substances within only few reaction steps.
Reaction | Addition of | Products |
---|---|---|
Hydrogenation | H2 | Carboxylic acids |
Saturated lactones | ||
Diols | ||
Hydroformylation | H2/CO | Aldehydo-carboxylic acids |
Hydroaminomethylation | H2/CO/HNR2 | Amino-lactones |
Amino-carboxylic acids | ||
Hydroamination | HNR2 | Amino-carboxylic acids |
Alcoholysis | ROH | Hydroxy-carboxylic acids |
Alkoxy-carboxylic acids | ||
Hydration | H2O | Hydroxy-carboxylic acids |
Hydrosilylation | HSi(OEt)3 | Silano-carboxylic acids |
Oxidation | H2O2 | Lactone epoxides |
Polymerisation | Dithiols etc. | Polymers |
In industry fatty acids are often used for the production of plasticisers for polymers. Therefore, the acids are reduced to their alcohols which are then reacted with phthalic acid for example. Other plasticisers are based on 2-ethylhexanol which is obtained from condensed and hydrogenated butanal. But since the commercially available product di(2-ethylhexyl)phthalate (DEHP) was found to be one of the more toxic phthalates,38 the interest in alternative compounds has increased. When the δ-lactone is hydrogenated 2-ethylheptanoic acid 11 or 2-ethylheptanol respectively is formed which might be also used for the production of plasticisers. Compared to DEHP, di(isononyl) phthalate (DINP) poses no risk to human reproduction or development. Presumably di(2-ethylheptyl)phthalate is in the same way harmless due to the same number of carbon atoms.
In the heterogeneous catalysed hydrogenation of the δ-lactone (Fig. 4), besides the desired product 2-ethylheptanoic acid (11) the formation of the saturated lactone 12, is observed.39–41
Fig. 4 Heterogeneous catalysed hydrogenation of the δ-lactone 1. |
The highest yield of 2-ethylheptanoic acid 11 in heterogeneous catalysed hydrogenation is achieved using methanol as solvent, however with 28% the yield is rather low. Using non-polar solvents, like n-heptane or THF, the formation of the saturated lactone 12 is enhanced up to selectivities of 95%. A quantitative production of 2-ethylheptanoic acid 11via heterogeneous catalysis cannot be achieved, because hydrogenation of δ-lactone 12 does not lead to ring cleavage.
Under monophasic homogeneous hydrogenation conditions, e.g. with Wilkinson's catalyst ClRh(PPh3)3, the same product distribution as for the heterogeneously catalysed hydrogenation is observed.39 On the contrary the biphasic hydrogenation with a water-soluble in siturhodium triphenylphosphinetrisulfonate (TPPTS) catalyst yields 2-ethylheptanoic acid 11 as the main product (63% after 4 h). Exclusively unsaturated isomeric C9-carboxylic acids are identified as reaction intermediates and neither saturated nor partially hydrogenated lactones are observed (Fig. 5). Obviously under homogeneous catalysed, two-phase reaction conditions the cleavage of the lactone ring is faster than double bond hydrogenation. Saturated lactones are not hydrogenated or cleaved under the same reaction conditions. This can be explained by the lacking coordination of the double bonds to the catalytic active metal atom.
Fig. 5 Homogeneous biphasic hydrogenation of the δ-lactone 1. |
The mechanistic steps are shown in detail in Fig. 5: in the first reaction step the δ-lactone is cleaved under isomerisation to a mixture of three isomeric ethylidene heptanoic acids. The mechanism is explained via a ring opening under formation of a η3-allyl-carboxylate-rhodium complex which is able to insert the hydrogen in two different positions of the molecule. The subsequent total hydrogenation of both C–C-double bonds leads to the final saturated product 11.
The optimisation of the homogeneously catalysed biphasic hydrogenation resulted in a complete conversion of the δ-lactone after a reaction time of 30 to 60 min. The turnover frequency at a reaction temperature of 90 °C was calculated to 4,500 h−1.42,43
However, the reaction rate depends strongly on the temperature: An increase of the temperature of 10 °C from 90 to 100 °C leads to an increase of the TOF of more than +1000 h−1. The highest temperature that can be applied is about 125 °C, because at higher temperatures the catalyst deactivates.
The effect of the hydrogen pressure (10 to 30 bar, TOF +500 h−1) is rather small. The increase of the reaction rate will not compensate the higher costs for a high pressure reaction unit for higher pressure ranges in a potential future plant. The TOF increases by 3000 h−1, while increasing the stirring velocity from 600 to 1200 rpm. This indicates limitations due to mass transport problems, a common subject in two-phase liquid–liquid reactions. Increasing the P/Rh ratio from 5 to 20, the reaction rate doubles. The active catalyst consists of a central rhodium atom modified with several ligands. If the concentration of ligands increases, the number of active species will raise, too. Furthermore, an excess of polar ligands prevents the leaching of rhodium into the organic phase.
The influence of the amount of catalyst was investigated in the range from 250–1500 ppm rhodium referring to the amount of δ-lactone used. The conversion increases with increasing amount of catalyst. The number of active catalyst centers is higher and therefore a greater number of molecules can be converted simultaneously. At 500 ppm Rh a maximum TOF of 8800 h−1 is reached. At higher catalyst concentrations only a small part of the active catalyst species are used for the allylic substitution reaction. Hence, reasonable reaction times are possible with catalyst concentrations of even less than 500 ppm; higher catalyst concentrations are not necessary.
The recycling of the catalyst was realised by phase separation and reuse of the aqueous catalyst phase. Overall 435 g δ-lactone have been converted selectively with only 38 mg rhodium yielding a mixture of the isomeric 2-ethylideneheptenoic acids at 110 °C and 10 bar H2 at a total TON of 7,840.
Although the desired product, 2-ethylheptanoic acid 11, is already obtained by homogeneous biphasic hydrogenation, a quantitative yield was only achieved applying a second heterogeneous hydrogenation step. The δ-lactone ring cleaved homogeneously in a two-phase system followed by double bond hydrogenation with a commercial heterogeneous palladium/charcoal catalyst.38
With methanol and n-heptane two solvents of different polarity were chosen for the heterogeneous catalysed hydrogenation experiments. While methanol is well miscible with water and most organic solvents, n-heptane is almost immiscible with water at room temperature which enables the application of a two-phase system. Quantitative yields of 2-ethylheptanoic acids are achieved with methanol after 3–5 min at 80 °C, with n-heptane as solvent only after 30 min. An explanation for this difference is the better solubility of the ethylideneheptenoic acids in methanol. The presence of 1 wt% water in the mixture of ethylideneheptenoic acids decreases the reaction rate of the heterogeneous hydrogenation considerably. Therefore, it is essential to remove the water before the second hydrogenation step by distillation.
Summarising, a two-step process for the convenient production of 2-ethylheptanoic acid from the δ-lactone 1 is possible (Fig. 6). In the first step a catalytic biphasic hydrogenation of the δ-lactone 1 with the water-soluble rhodium-TPPTS-catalyst is applied. After addition of pentane the phases are separated and the catalyst containing in the aqueous phase is reused. The pentane is distilled off the organic phase subsequently. In a second reaction step the remaining double bonds are hydrogenated using a commercial heterogeneous palladium catalyst in methanol. Then, the methanol is distilled off and after a further distillation the product 2-ethylheptanoic acid 11 is obtained in high purity.
Fig. 6 Process diagram for the synthesis of 2-ethylheptanoic acid 11. |
Subsequently it could be shown that the resulting acid can be hydrogenated to the corresponding 2-ethylheptanol 14, which might be used as plasticiser for polymers like PVC (Fig. 7).44
Fig. 7 General scheme of the hydrogenation chemistry of the δ-lactone 1. |
At first, the reduction of 2-ethylheptanoic acid to 2-ethylheptanol with homogeneous or heterogeneous single metal catalysts was investigated. Several rhodium- and ruthenium-based catalysts and molybdenum hexacarbonyl were studied, but neither a single homogeneous nor heterogeneous catalyst achieved yields higher than 10% of 2-ethylheptanol. However, an increase of 2-ethylheptanol yield could be achieved by addition of a second catalyst precursor. Thus following catalyst combinations were investigated:
Catalyst 1:
• transition-metal carbonyls of group 6 or 7
• Co2(CO)8
• Mo(II)acetate
Catalyst 2:
• homogeneously soluble rhodium compound like [Rh(acac)(CO)2]
• heterogeneously dispersed metals on alumina or charcoal
• ruthenium compounds
With bimetallic catalysts nearly quantitative yields are reached. Comparing the catalytic hydrogenation activities of different carbonyl complexes with [Rh(acac)(CO)2] as second catalyst at 200 °C and 150 bar hydrogen leads to the following sequence with decreasing activity:
[Mo(CO)6] ≫ [Re2(CO)10] > [Mn2(CO)10] > [W(CO)6] > [Cr(CO)6] > [Co2(CO)8] |
Group 8 or 9 precursors in combination with Mo(CO)6 show the following sequence of decreasing activity:
[Rh(acac)(CO)2] > [Rh(II)acetate] > [Ru(acac)3] > [Rh(acac)3] > [Co2(CO)8] > [Rh(cod)Cl]2 > [RhCl3·3H2O] |
In summary, based on the δ-lactone 1, a highly efficient synthesis of 2-ethylheptanol 14, which might find application as plasticiser alcohol, is possible within three hydrogenation steps. With the optimum catalytic systems an overall alcohol yield of 91% can be reached (Fig. 7).
The isomeric diols can be selectively obtained by formation of two different precursors:43
• Heterogeneous hydrogenation of the δ-lactone gives a diastereomeric mixture of the 2-ethylheptan-5-olides 12.47
• The γ-lactone 2 can be produced by isomerisation of the δ-lactone 1 with palladium/tricyclohexylphosphine,15 which is subsequently hydrogenated with standard heterogeneous hydrogenation catalysts to 2-ethylheptan-4-olide 13.
For the cleavage of these lactones to their corresponding diols bimetallic catalyst systems consisting of [Rh(acac)CO)2] and [Mo(CO)6] were used which were known as highly active for the hydrogenation of acids to alcohols as already shown in section 2.1.
The reduction of 2-ethylheptan-5-olide 12 with hydrogen in the presence of a bimetallic Rh/Mo catalyst leads to 2-ethyl-1,5-heptanediol 15 with 2-ethylheptanol and 6-methyl-3-octanol as byproducts from a consecutive hydrogenation. The reaction was optimised with respect to temperature, hydrogen pressure, catalyst concentration, and rhodium/molybdenum ratio. With 97% the highest yield was reached at 190 °C and 150 bar H2 after 2 h. Longer reaction times lead to product decomposition and mono-alcohol formation. Decreasing the hydrogen pressure from 150 to 100 bar influences the yield slightly whereas at 50 bar only 38% diol 15 are formed. The ratio of the two transition metal precursors strongly influences the catalyst activity: Lowering the rhodium amount leads to a great loss of activity, whereas the influence of a lower molybdenum amount is smaller. The optimum Rh/Mo ratio is 1. Presumably, the main catalytic reaction occurs at the rhodium central atom, which is activated by the molybdenum compound.
The reduction of 2-ethylheptan-4-olide 13 to 2-ethyl-1,4-heptanediol 16 was performed analogously to the reduction of 2-ethylheptan-5-olide 12. The desired main product is 2-ethylheptane-1,4-diol, whereas the by-products are 2-ethylheptanol and 6-methyl-4-octanol. Due to the higher thermodynamic stability of five-membered rings compared to six-membered rings, the yield of diols by hydrogenation of 2-ethylheptan-4-olide is, with 2%, comparably low. At longer reaction times, the conversion increases, however causing the formation of by-products like alcohols and hydrocarbons. Hence, more thermal energy is necessary to open the lactone ring, which simultaneously causes decomposition of the product. For that reason the yield of the target product (15%) does not increase either at higher temperatures or at longer reaction times.
The hydroformylation of the δ-lactone was carried out by Brehme using a phosphite modified rhodium catalyst (Fig. 8).40 As a ligand BIPHEphos was used, which was prepared according to a protocol by Cuny and Buchwald.49 The hydroformylation was carried out in tetrahydrofuran as solvent and reaction parameters like pressure, time, temperature and Rh/P ratio were optimised aiming on economical feasible values and high selectivities to the linear aldehyde 17. It was found that under relatively mild reaction conditions only the terminal carbon-carbon double bond shows activity for the reaction. In summary, the linear aldehyde 17 can be formed at 90 °C under 5 bar of syngas (1:1) within 90 min with a yield of 95%. The catalyst concentration was 0.1 mol% regarding the δ-lactone with a Rh:P ratio of 1:2.42
Fig. 8 Hydroformylation of the δ-lactone 1 to aldehydes 17 (n) and 18 (iso). |
Based on an isomeric mixture of 2-ethylideneheptenoic acids which are obtained by biphasic homogeneous hydrogenation (Fig. 5) additional hydroformylation experiments are described.42 By application of the same Rh/BIPHEphos catalyst as it was used for hydroformylation of the δ-lactone 2-ethylidene-8-oxooctanoic acid 19 can be synthesised in 89% yield. In this reaction, the internal C–C double bond is first isomerised to the terminal chain position and then hydroformylated to the corresponding aldehyde with high selectivity to the linear isomer 19. Subsequent oxidation of the aldehyde group with chromic acid followed by hydrogenation of the remaining C–C-double bond in the diacid 20 leads to the completely saturated 2-ethyl-α,ω-octanedioic acid 21 (Fig. 9).
Fig. 9 Isomerising hydroformylation of the hydrogenation products of the δ-lactone 1. |
Besides the hydroamination (section 5) the hydroaminomethylation is one of the most promising methods for substitution of some of the classical processes in the future. The reaction was originally developed by Reppe at BASF51 and can be described as a domino or one pot reaction. In the first step an alkene is hydroformylated with syngas using a rhodium catalyst yielding the corresponding aldehydes. This reaction is followed by a condensation of the aldehydes with an amine followed by a terminal hydrogenation of the enamine or imine formed (Fig. 10). In this way it is possible to synthesise amines with high selectivities with water as the only by-product.
Fig. 10 General reaction pathway of the hydroaminomethylation of alkenes. |
In recent years several tailor-made catalysts for hydroaminomethylation were developed leading to high selectivities and activities. A comprehensive survey was written by Eilbracht et al.52
Based on the results of the rhodium catalysed hydroformylation of the δ-lactone its behaviour in hydroaminomethylation with morpholine as model substrate was investigated by Henze (Fig. 11).53 The regioselectivity of the reaction is determined in the first step. With the catalyst system Rh/BIPHEphos high selectivities of 95% for the linear intermediate aldehyde can be obtained. This is directly condensed with morpholine followed by hydrogenation of the enamine yielding the linear tertiary amine 22. This amine can further react within two reaction routes to the same hydroxy amino acid 25:
(a) Through hydroformylation of the internal carbon-carbon double bond the aldehydic amine 23 is formed. Due to steric hindrance and lack of amine (δ-lactone : morpholine = 1:1) this aldehyde does not condensate with morpholine, but gets hydrogenated to the corresponding alcohol 24. In a last reaction step the lactone ring is cleaved viahydrogenation leading to a hydroxy amino acid 25.
(b) Hydrogenation of the lactone ring of 22 leads to the formation of an unsaturated amino acid 26. The carbon-carbon double bond of this compound is then hydroformylated to the corresponding aldehyde 27 which is subsequently hydrogenated to the alcohol 25.
Fig. 11 Hydroaminomethylation of the δ-lactone 1 and consecutive reactions. |
The reaction conditions were chosen based on investigations of Beller.54 A 1:1 mixture of methanol and toluene served as solvent for the reaction. The reaction temperature was 125 °C with a syngas pressure of 40 bar (7 bar CO and 33 bar H2). Rhodium together with iridium was used in concentrations below 0.1 mol% regarding the δ-lactone in combination with BIPHEphos as ligand. The reaction can be stopped after 2 h with a quantitative conversion of the δ-lactone leading to a 55:45 mixture of the unsaturated amine 22 and the aldehydic amine 23. Longer reaction times lead to the formation of the hydrogenation products describe above. Other ligands such as triphenylphosphine or DPEphos led to the additional formation of branched products due to lower n-selectivity.
Mainly two reaction routes are known for the hydroamination. With monoalkenes classical 1,2-additions are reported inter- as well as intramolecular leading to branched amines in accordance with Markownikow's rule. On the other hand conjugated dienes are often subjected to an 1,4-addition (Fig. 12).
Fig. 12 1,2- or 1,4-hydroamination. |
However, the electronic properties of amines complicate this reaction. Nevertheless many hydroaminations are described in literature using mostly homogeneous catalysts for activation. In this context almost all elements of the periodical table have been mentioned as catalyst. In contrast to other homogeneous catalysed reactions, no “universal” catalyst has been found for the hydroamination like rhodium complexes for hydroformylation or palladium for telomerisation reactions. Depending on the nature of the substrates completely different metal complexes are used. The preferred transition metals are palladium, iridium, nickel, platinum, rhodium and lanthanum in combination with mostly bidentate phosphorous based ligands. With an annual production of 2000 t, the Takasago process for the production of (−)-menthol from myrcene represents the largest technical realisation of a hydroamination.56
A derivatisation of the δ-lactoneviahydroamination is described by Behr et al.53,57 Selective 1,4-addition, like it is found with dienes, of the model amine morpholine to the α-double bond of the δ-lactone 1 allowed the synthesis of the ζ-amino acid 28 in a single step (Fig. 13).
Fig. 13 Hydroamination of the δ-lactone 1 with morpholine. |
Starting from known catalyst systems in a high throughput investigation of numerous catalyst systems it was found that several metals like aluminium, iron, zirconium, platinum, palladium, rhodium or iridium together with the bisphosphine ligand DPEphos (see Fig. 13) show high activity for this reaction. Further reaction parameters like the reaction temperature or time were optimised aiming on a high selectivity and optimum space time yield. The best catalytic systems are summarised in Table 2. Thus the reaction time could be reduced below 1 h yielding 95% of amino acid 28 (TOF = 846 h−1) with relatively cheap and non-toxic catalysts like aluminium/DPEphos which may be the most profitable in a technical process.
Catalyst | T/°C | TOF/h−1 |
---|---|---|
a 1 eq. morpholine. b 2 eq. morpholine. | ||
Pt(cod)Cl2 | 120 | 1192 |
Al(OTf)3 | 100 | 846 |
Zr(OTf)4 | 100 | 590 |
Pt(cod)Cl2 | 100 | 374 |
La(OTf)3 | 100 | 363 |
Ir(acac)(CO)2b | 100 | 177 |
Ce(OTf)3 | 100 | 159 |
Ir(acac)(CO)2 | 120 | 152 |
Co(acac)2 | 120 | 137 |
Fe(acac)2 | 100 | 131 |
Ir(acac)(CO)2a | 100 | 77 |
Co(acac)2 | 100 | 26 |
Ir(acac)(CO)2 | 100 | 24 |
Dependent on the kind of the catalytically active metal two different mechanisms have been proposed for the formation of amino acid 28. It is suggested that metal complexes which react as “soft” Lewis acids induce a 1,4-addition of the amine to the δ-lactone while metal complexes with a “hard” Lewis acidic character affect a ring opening through coordination on the carbonyl carbon. Both reaction pathways lead to the same product.
In additional investigations on the hydroamination of the δ-lactone the reactivity of other nitrogen containing substrates than morpholine is described.53 The following nitrogen compounds lead to reaction products with the δ-lactone: ammonia and aqueous ammonia solutions, ethylene diamine, ethanol amine, n-propylamine, piperidine and pyrrolidine (Table 3). It was shown that depending on the basicity of the nitrogen compound the reaction takes place catalysed or even non-catalysed (Fig. 14). Basic nitrogen compounds like ammonia or aqueous ammonia solutions, ethylene diamine, ethanol amine and n-propylamine formed products with the δ-lactone without any additional catalyst.
• Gaseous ammonia dissolved in toluene leads non-catalytically to 100% conversion of the δ-lactone. Two reaction products are observed: a ζ-amino acidvia1,4-addition of NH3 and an amide resulting from a nucleophilic ring opening of the δ-lactone. The usage of catalysts has no influence on the product distribution due to the high reaction rate.
• An aqueous ammonia solution on the other hand favours the formation of the amide compared to the ζ-amino acid. Without a catalyst 15% conversion of the δ-lactone are achieved. The usage of Fe(acac)2 or Pt(cod)Cl2 increases the lactone conversion up to 93 and 89%, respectively, yielding preferentially the amide and the ζ-amino acid as coproduct.
• Ethylenediamine leads similar to ammonia to a 100% conversion of the δ-lactone without any catalyst. The corresponding ζ-amino acid is the main product.
• Ethanolamine converts 25% of the δ-lactone under the cited reaction conditions. Similar to ethylenediamine and ammonia a ζ-amino acid is formed as main product. Pt(cod)Cl2, Ce(OTf)3 and La(OTf)3 are the best catalyst precursors to increase the lactone conversion up to 90% and above.
• The usage of n-propylamine as nitrogen compound forms an amino acid with the δ-lactonevia ring opening nucleophilic addition. Without any catalyst a conversion of 32% was observed. Pt(cod)Cl2, Ce(OTf)3 and Bi(OTf)3 are the best catalysts to increase the lactone conversion above 90%.
• Without catalysis pyrrolidine and piperidine only show conversion of the δ-lactone below 5%. Rh(cod)BF4/DPEphos is the best catalyst system leading to 90 and 62%, respectively, conversion of the δ-lactone to the corresponding ζ-amino acid.
Fig. 14 Reactivity of nitrogen compounds in δ-lactone 1hydroamination. |
Amine | Main product | Catalyst | X1 [%] |
---|---|---|---|
0.4 mol%, DPEphos (M : P 1:32), cδ-lactone = 0.16 mol L−1, δ-lactone : amine 1:1, 5 bar Ar, t = 24 h, T = 100 °C, toluene. | |||
Ammonia | none | 100 | |
NH4OH-solution | none | 15 | |
Ce(OTf)3 | 61 | ||
Pt(cod)Cl2 | 89 | ||
Fe(acac)2 | 93 | ||
Ethylenediamine | none | 100 | |
Ethanolamine | none | 25 | |
cpRe(CO)3 | 66 | ||
Ce(OTf)3 | 92 | ||
Ir(acac)2(CO)2 | 69 | ||
Pt(cod)Cl2 | 99 | ||
Fe(acac)2 | 81 | ||
Co(acac)2 | 58 | ||
La(OTf)3 | 90 | ||
n-Propylamine | none | 32 | |
Ce(OTf)3 | 94 | ||
Ir(acac)2(CO)2 | 41 | ||
Pt(cod)Cl2 | 97 | ||
Fe(acac)2 | 46 | ||
Co(acac)2 | 52 | ||
Al(OTf)3 | 56 | ||
Sn(OTf)2 | 65 | ||
Bi(OTf)3 | 91 | ||
La(OTf)3 | 61 | ||
Piperidine | none | <5 | |
Ir(acac)(CO)2 | 70 | ||
Rh(cod)BF4 | 62 | ||
Pyrrolidine | none | <5 | |
Ir(acac)(CO)2 | 22 | ||
Rh(cod)BF4 | 90 |
In summary, hydroamination of the δ-lactone is – like it was already known from other alkenes – highly substrate specific. Unlike other homogeneous catalysed reactions, like hydroformylation, hydroamination has no universally active catalyst metal which generally enables the reaction. Even small variations of the amine compound changes the reactivity as well as the product distribution in the reaction with the same alkene.
Furthermore, the unsaturated ζ-amino acid 28 was hydrogenated with the catalyst Rh(cod)BF4 at 40 bar H2 yielding the two single unsaturated acids and the completely saturated ζ-amino acid shown in Fig. 15.
Fig. 15 Hydrogenation of the unsaturated ζ-amino acid 28. |
Because of the ability of the δ-lactone 1 to undergo carbonylation reactions (see sections 3 and 4) its reactivity under hydrocarbalkoxylation conditions was investigated using different alcohols.53 Applying a catalyst concentration of 0.4 mol-% Pd(OAc)2, a triphenylphosphine to palladium ratio of 4:1, a tenfold excess of p-toluenesulfonic acid (TSA) as promotor, 35 bar carbon monoxide at 75 °C for two hours with the pure alcohol as solvent high δ-lactone conversions >95% were achieved with the primary alcohols methanol, ethanol, 1-propanol and 1-butanol. The lactone conversion with secondary alcohols like 2-propanol and 2-butanol is with 22 and 5% drastically lower. Tertiary alcohols like tert.-butanol show no reaction.
However, the resulting products were not formed through hydrocarbalkoxylation of the terminal C–C double bond of the δ-lactone, but by alcoholysis of the lactone ring. The product spectrum of the reaction with methanol is given in Fig. 16 showing the hydroxy ester 29 and the two methoxy esters 30 and 31. A similar product mix was already previously observed under acidic and basic conditions in methanolic solution.59 Thus, the acidic promotor TSA is responsible for the ring cleavage while the palladium catalyst stays inactive. The analogous products were observed through reaction with the other alcohols mentioned above.
Fig. 16 Methanolysis of the δ-lactone 1. |
Obviously, alkoholysis is much faster than hydrocarbalkoxylation. But an increase of the reaction time did not lead to a carbonylation of any C–C double bond of the cleaved δ-lactone. The assumption that the C–C double bonds of the cleaved δ-lactone are no more active for any carbonylation reaction was disproved by successful hydroformylation of the isolated cleavage product 30 yielding the aldehyde 32 (Fig. 17).
Fig. 17 Hydroformylation of the methanolysis product 30. |
Through hydration of the δ-lactone 1 a hydroxyl group is introduced into the molecule. Analogous to the reaction mechanism of hydroamination, no 1,2-addition of water to the C–C double bond occurred at 100 °C and with Amberlyst® for 24 h.53 Instead a ring opening was observed yielding four products with 2-ethylidene-5-hydroxy-hept-6-enoic acid 33 as the main component, which was already known from the methanolysis experiments.59 The conversion of the δ-lactone 1 is 98% and the products can be easily separated because of their immiscibility with water. The resulting products remain unsaturated, even after increasing the reaction time to 72 h and the temperature to 100 °C (Fig. 18).
Fig. 18 Hydration of the δ-lactone 1. |
Similar to the reactions with nitrogen compounds or alcohols it could be shown that also in hydration the δ-lactone 1 tends to ring opening leading to functionalised and unsaturated aliphatic products.
Many hydrosilylation reactions are carried out using homogeneous transition metal catalysts, especially platinum compounds such as H2PtCl6 in 2-propanol solution (Speier's catalyst) or the Pt(0)-divinyltetramethyldisiloxane complex (Karstedt catalyst). The exact mechanism of this reaction is not completely evidenced, but two accepted mechanisms were proposed by Chalk and Harrod.60
In contrast to the addition reactions described above the hydrogen of the silane is not protic, but has a hydridic character. According to Markownikow's rule a 1,2-addition to the terminal double bond of the δ-lactone was expected. A reaction with the internal double bond is generally possible, but not favoured due to the higher degree of substitution.
We investigated the hydrosilylation of the δ-lactone with triethoxysilane. As the main product the aliphatic silane 34 was formed under mild reaction conditions (Fig. 19).53
Fig. 19 Hydrosilylation of the δ-lactone 1. |
Comparable to hydroamination, alcoholysis and hydration, hydrosilylation also leads to ring opening instead of a 1,2-addition yielding a product with a similar substitution pattern notwithstanding the different character of the silane as substrate.
The following reaction mechanism was proposed as explanation for the product formation (Fig. 20): at the beginning of the catalytic cycle the PtCl62− anion is reduced to Pt(II) by elimination of two chloride ions analogous to the Speier mechanism. Excess silane as well a the δ-lactone itself are the proposed reduction agents. Afterwards, one equivalent of silane adds oxidatively to the to Pt(II) forming a Pt(IV) complex. This reactive species initiates the catalytic cycle. At first (step I) the terminal carbon-carbon double bond of the δ-lactone coordinates at the Pt(IV) forming a π-complex. Subsequently, the double bond inserts into the platinum-silicone bond and thus a σ-complex is formed (step II). The following β-elimination (step III) of platinum leads to a metal carboxylate with a double bond in β-position to the silicone. After reductive elimination of the intermediate and repeated oxidative addition of a silane (step IV) the catalytic active species is re-obtained and the cycle re-starts. In the last step (step V) the carboxylic acid is esterified with the excess of silane forming the product 34.
Fig. 20 Proposed reaction mechanism for the hydrosilylation of the δ-lactone 1. |
It is conspicuous that a product is formed which resembles to those of hydroamination, alcoholysis and hydration in spite of the generally different reactivity of the silane compared to amines, alcohols or water. Obviously the lactone tends to ring opening via1,4-addition in presence of a substrate with a labile hydrogen – independent of the electronic character of the hydrogen.
For the epoxidation of alkenes a number of catalytic and non-catalytic system are described in literature. However, in the epoxidation of the δ-lactone 1 neither the Jacobsen catalyst, based on Mn(III),61 nor the well-known lanthanoid catalysts62 were successful. Only methyltrioxorhenium (MTO) with H2O2 as oxidation agent and pyridine as auxiliary led to a 9% yield of a mono-epoxide mixture of the δ-lactone mixture (Fig. 21). Diepoxidised products were not detected.
Fig. 21 Epoxidation of the δ-lactone 1 with methyltrioxorhenium (MTO) as catalyst. |
Also the non-catalytic epoxidation of the δ-lactone with m-CPBA (meta-chloro perbenzoic acid) was investigated. At 80 °C and with a fourfold excess of m-CPBA a 9% yield of the same epoxides is achieved.
In general it can be summarised that the δ-lactone is surprisingly stable against well known and usually highly active oxidation systems.
The copolymerisation of the δ-lactone with dithiols such as 2,2′-(ethylenedioxy)diethanethiol (a) or 1,3-bis(3-mercaptopropyl)-1,1,3,3-tetramethyldisiloxane (b) leads to high-molecular colourless, transparent polymers in yields of up to 80%. The polymer chain contains the intact lactone structure. Benzildimethylketal (BDMK) serves as photo initiator for this reaction together with the light of a medium-pressure mercury lamp (Fig. 22).
Fig. 22 Linear polyaddition of the δ-lactone 1 to dithiols. |
Applying a linker such as pentaerythriol-tetrakis(3-mercaptopropionate) (Fig. 23) together with the δ-lactone, the dithiols 2,2′-(ethylenedioxy)diethanethiol or 1,3-bis(3-mercaptopropyl)-1,1,3,3-tetramethyldisiloxane and the diene co-monomers 1,1,3,3-tetramethyl-1,3-divinyldisiloxane or diallyl succinate (Fig. 24) lead to complete conversion of the δ-lactone yielding highly networked polymers. The reactions were carried out in substance and the resulting products were polymer films or bulk polymers of different viscosity and elasticity.
Fig. 23 Pentaerythriol-tetrakis(3-mercaptopropionate) as linker for network polymerisation of the δ-lactone 1. |
Fig. 24 1,1,3,3-Tetramethyl-1,3-divinyldisiloxane and diallyl succinate as co-monomers for network polymerisation of the δ-lactone 1. |
Additional trials for radically initiated homopolymerisation of the δ-lactone did not lead to polymers. The identified starting species are too stable for polymerisation (Fig. 25). In case of the terminal carbon-carbon double bond an allylic radical could be identified.
Fig. 25 Radicals of the δ-lactone 1. |
It can be summarised that the conversion of the δ-lactone, and thus the recovery of CO2, by various catalysed reactions has widely been investigated:
• Viahydrogenation, several branched aliphatic C7-acids, -alcohols or diols can be produced within two to three catalysed reaction steps. These products might find application as softener alcohols or starting compounds for polyurethanes or polyesters.
• Under mild hydroformylation conditions the δ-lactone forms its corresponding monoaldehydes. When the conditions are adjusted for a hydroaminomethylation by increasing the hydrogen partial pressure and adding a protic nitrogen compound, the expected amines can be obtained, but additional products are formed which result from consecutive reactions like hydroformylation of the internal C–C double bond and hydrogenating ring opening. Both C–C double bonds with their different reactivity as well as the cleavable lactone functionality result in several side reactions which inevitably leads to a broad product mixture.
• Other carbonylation reactions like the hydroesterification do not lead to an insertion of carbon monoxide in the C–C double bonds of the δ-lactone. Instead a ring cleavage of the lactonevia alkoholysis can be observed.
• Polymerisation reactions as well as oxidations only lead to limited success. The high stability of the δ-lactone 1 against oxidative conditions as well as the stability of the lactone radicals inhibit these types of reactions.
• Numerous catalytic conversions of the δ-lactone 1 with substrates of the type H–Y are described where H can be of protic or hydridic character. These substrates H–Y are commercially available chemicals like water, alcohols, ammonia, amines or silanes. These functionalisations lead to esters, diols, hydroxy acids, amines or amino acids. In most cases it could be shown that the δ-lactone does not react via1,2-addition. Instead a different reaction scheme can be identified (Fig. 26): an attack of H–Y type substrates on the terminal C–C double bond leads to a 1,4-addition which is associated with a ring cleavage of the lactone (path A). An addition of the H–Y substrates can also take place at the acid- and base sensitive C–O single bond of the lactone which also leads to a ring opening (path B). Finally, an attack of a nucleophile can occur at the carbonyl carbon which again leads to an opening of the δ-lactone to an aliphatic product (path C). It is dependent on the substrate and on the catalyst which reaction pathway is favoured. Thus, for many cases the reactivity of the δ-lactone is predictable.
Fig. 26 General reactivity of the δ-lactone 1 with substrates of the type H–Y. |
Through the conversion of the carbon dioxide-containing δ-lactone 1 with different basic chemicals it is thus possible to enable the usage of CO2 as feedstock for various, highly functionalised products with potential industrial interest.
This journal is © The Royal Society of Chemistry 2011 |