Challenges in Oxidation Catalysis - Springer Link

Topics in Catalysis Vol. 24, Nos. 1–4, October 2003 (# 2003)

73

Challenges in oxidation catalysis W. Buijs Applied Organic Chemistry & Catalysis, Faculty of Applied Sciences, TU Delft, Julianalaan 136, 2628 BL Delft, The Netherlands

From an industrial chemical perspective, successful oxidation catalysis poses three challenges: (1) use (di)oxygen, (2) avoid overoxidation of the substrate, and (3) overcome the lack of functionality in the feedstock. From the options discussed, confinement seems to be the most versatile way to meet all the needs of industrial oxidation catalysis. KEY WORDS: DOW-Phenol; acrylonitrile; NaY; cyclohexanone; CoAlPO; adipic acid; hydroxylamine; caprolactam; confinement.

1. Introduction From an industrial chemicals point of view, the purpose of all chemical transformations is to convert simple hydrocarbon feedstocks into compounds with added functionality, usually monomers. After polymerization and further processing, useful materials for the society are obtained. On a molecular level, the added functionality generally means the introduction of oxygen and/or nitrogen. Oxidation is a very important transformation to introduce oxygen in a hydrocarbon, and indirectly, to introduce nitrogen (as ammonia) as well. Major examples are the oxidation of p-xylene to terephthalic acid, the oxidation of ethylene to ethylene oxide, the oxidation of n-butane to maleic anhydride and 1,4-butane diol, and the oxidation of cyclohexane to cyclohexanone, or adipic acid. Nitrogen (as ammonia) can be introduced via ammoxidation processes like the ammoxidation of propene to acrylonitrile, or net substitution processes like the formation of 2,4,6triamino 1,3,5-triazine (melamine) from carbon dioxide and ammonia. An important exception to this rule is the conversion of cyclohexanone into caprolactam by hydroxylamine (instead of ammonia) via cyclohexane oxime and a Beckmann rearrangement. On inspection of these examples, it becomes clear that only the oxidation of methyl aromatics to the corresponding acids are processes with both good conversion and selectivity. Most of the other routes suffer from a rather limited selectivity, even at low conversion. Furthermore, the use of alternative oxidants like nitric acid in the production of adipic acid inherently leads to extra cost to avoid environmental problems. The formation of hydroxylamine from ammonia via oxidation to nitric acid and partial reduction of nitric acid to yield hydroxylamine and NOx is another important example of a process with a limited selectivity To whom correspondence should be addressed. E-mail: [email protected]

and considerable extra cost to avoid environmental damage. The problems in bulk chemical oxidation catalysis can be rephrased as a series of scientific challenges. . The first challenge is to use dioxygen, preferably air instead of any other oxidants. Since dioxygen is a triplet biradical, free radical chain autoxidation (FRCA) dominates both gas–liquid-phase homogeneous and heterogeneous catalysis, even in cases in which the catalytic transformation itself is rather selective. The final result is usually an apparent lack of selectivity in the overall transformation. . The second challenge is to avoid overoxidation because usually the oxygenate is more sensitive to oxidation as the feedstock. The cobalt-catalyzed oxidation of toluene yields only a minor percentage of benzaldehyde amongst the main product, benzoic acid. . The third challenge is to overcome the hurdle that many feedstocks lack functionality, and might be susceptible to several oxidative transformations, still leading to an overall loss of selectivity though the catalytic transformation itself is selective. n-butane, cyclohexane, and benzene are obvious examples.

2. Concepts In the vast amount of work on oxidation catalysis several new concepts can be distinguished, which will be highlighted here with some examples. The concepts are as follows: (1) Shift to non-free radical pathways like . Metal ion (pair) oxidation, . The use of 1 O2 instead of 3 O2 . (2) Heterogeneous gas-phase oxidation (3) Confinement: . Photocatalyzed oxidation of hydrocarbons in zeolite cages with visible light. 1022-5528/03/1000–0073/0 # 2003 Plenum Publishing Corporation

74

W. Buijs/Challenges in oxidation catalysis

. Direct oxidation of n-hexane to adipic acid in CoAlPO-18. . Oxidation of NH3 to NH2 OH and its transformation into caprolactam.

limited use to gain selectivity, apart from the additional cost of producing and transporting it. Less known are real metal ion (pair) oxidations like

2.1. Non-free radical pathways: metal ion (pair) oxidation The role of transition metals is not limited to several steps of the FRCA process. They can also play a role in the activation of dioxygen to a selective oxygenating species. High valent metal-oxo species, capable of performing selective oxygenation are well known; however, these processes usually suffer from some side reactions, which initiate free radical autoxidation, thus spoiling the catalytic transformations. Maybe the most beautiful way to illustrate this phenomena is the decomposition of hydrogen peroxide by Fe(II)-complexes: classical Fenton chemistry, yielding epoxides and a series of other oxidation products. The formation of an epoxide is an indication of a very selective oxygen activation and transfer process by the metal ion, while the series of other oxidation products are reminiscent of the previously mentioned FRCA process. Ensing and Baerends beautifully described the contradictory behavior of this catalytic process recently in a series of molecular modeling studies [1–3]. Scheme 1 shows the overall reactions and scheme 2 shows how these two reactions were connected.

Scheme 1. The formation of an unselective OH radical and a selective iron(IV)oxo species from Fe(II) and hydrogen peroxide.

Scheme 2. The formation of the penta aqua iron(IV)oxo species from Fe(II) and hydrogen peroxide via dihydroxo tetra aqua iron(IV).

Hydrogen peroxide coordinated to the penta aqua iron(II) ion was converted to a formally dihydroxo tetra aqua iron(IV) intermediate and water. In a second step, hydrogen atom transfer leads to the penta aqua iron(IV) oxo species. In an aqueous environment, the hydrogen atom transfer can also occur with solvent water molecules, thus leading to unselective OH radicals in solution. This example also illustrates why the use of hydrogen peroxide as alternative oxidant for dioxygen might be of

Scheme 3. Reaction equations of the DOW-Phenol reaction.

the DOW-phenol process. Scheme 3 gives an overview of the main steps involved. It was shown not only that the stoichiometry of the reactions requires two Cu(II) ions but also that in the mechanism the binuclear CuðIIÞ2 benzoate4 unit most likely plays a decisive role. Inner sphere ligand-to-metal electron transfer leads eventually to the formation of the benzoyloxy cation, which reacts with a benzoate in an electrophilic aromatic substitution process to form the reactive intermediate salicylic benzoate. After decarboxylation and hydrolysis, the final product, phenol, is formed with almost complete selectivity under appropriate conditions [4]. The role of the dioxygen in this process is limited to the reoxidation of Cu(I). The paddlewheel structure of M(II) carboxylates is very common, and is not limited to Cu(II). However, an extended investigation to explore the potential of Cu(II) carboxylates showed that generally no selective transformation can be observed, despite the fact that the initial step, ligand-to-metal electron transfer is most likely the same. Unselective one-electron transfer processes are observed frequently, and even in the case of a net two-electron transfer process, the resulting carbenium ion can undergo several reaction steps leading to alkenes, ; -unsaturated acids, and cyclic compounds amongst the desired ester, or alcohol [5]. Figure 1 shows the result of a DFT calculation on the binuclear CuðIIÞ2 acetate4 ðH2 OÞ2 complex. There is generally excellent agreement in angles and distances of the calculated and the crystal structure. The position of both HOMO and LUMO support the view of ligandto-metal electron transfer as initial step.

2.2. Non-free radical pathways: singlet oxygen ð1 O2 Þ Normal triplet dioxygen can be efficiently activated by a photosensitizer to yield singlet (di)oxygen [6]. There are also numerous ways to produce singlet dioxygen in a chemical reaction, but their relevance seems to be limited to fine chemistry. Alkenes, containing an allylic hydrogen atom, react with 1 O2 to the analogous allylic


75

Figure 1. HOMO and LUMO of CuðIIÞ2 acetate4 ðH2 OÞ2 ðB3LYP=LACVP Þ respectively indicating ligand-to-metal electron transfer on thermal decomposition.

hydro peroxides in the so-called ene-reaction. The reaction is outlined in figure 2. Alkenes with more than one distinct allylic hydrogen usually yield all possible hydroperoxides. Apart from this limitation, there are some other disadvantages: . The primary product is an allylic hydroperoxide, and selective decomposition to one single oxygenated product is not guaranteed, especially not under the reaction conditions. . The photosensitizer is usually destroyed quite fast under irradiation conditions. Recently, remarkable progress was reported with respect to the observed regiochemistry of the reaction [7]. It was found that the reaction of 1,2-dimethyl cyclohexene with dioxygen in thionin-exchanged NaY under irradiation ð> 420 nmÞ yielded 90% of the allylic hydroperoxide with the exocyclic double bond. However, in fact, this is another good example of the concept of confinement, to be discussed later.

3. Heterogeneous gas-phase oxidation The elegant idea behind this concept is that heterogeneous gas-phase oxidation avoids a bulk liquid phase, and thus avoids bulk liquid-phase FRCA with its inherent lack of selectivity. The success of the heterogeneous gas-phase concept is probably best demonstrated in the propene ammoxidation process, leading to acrylonitrile, and C4 oxidation, leading to maleic anhydride. H

H O

O

O

O

Figure 2. The ene-reaction of singlet oxygen with an alkene.

However, similar processes as in the bulk liquid phase might still occur on the surface of the heterogeneous catalysts. The previously mentioned DOW Phenol reaction is a typical example. The heterogeneous gasphase oxidation of benzoic acid to phenol with a Cu catalyst yields similar by-products (biphenyl ether, phenolic oligomers) as heavies on the surface, apart from the product phenol and some excess of the byproduct benzene. The mechanism is thought to be similar as in the liquid phase.

4. Confinement . Photocatalyzed oxidation of hydrocarbons in zeolite cages with visible light. . Direct oxidation of n-hexane to adipic acid in CoAlPO-18. . Oxidation of NH3 to NH2 OH. At first sight, the concept of confinement might look just as a variation on the concept of heterogeneous (gasphase) oxidation, but the concept reaches much further and deeper, as will be demonstrated in the following examples.

4.1. Photocatalyzed oxidation of hydrocarbons in zeolite cages with visible light H. Frei et al., began to publish on photocatalyzed oxidation of hydrocarbons in zeolite cages in the early nineties. Reference [8] gives a brief but impressive overview of the achievements. Simple alkenes like propene, and 2-butenes, aromatics like toluene and ethyl benzene, and alkanes ranging from methane to cyclohexane are converted with 100% selectivity to the corresponding aldehydes and ketones with conversions ranging from 20–75%. Figure 3 shows an outline of the proposed mechanism. Dioxygen and cyclohexane are activated by light-induced electron transfer from cyclo-

76


OH O +. λ < 520 nm

.

O2 - .

H+

HO2 .

O2, NaY

OH O

O NaY + H2O

Figure 3. Overview mechanism of photo-catalyzed oxidation of cyclohexane according to H. Frei [9].

hexane to oxygen. Next, the resulting radical cation can easily transfer a proton to O2 to yield an alkyl and an OOH radical. Recombination of the two radicals leads to the hydroperoxide, and subsequent elimination of H2 O gives the desired cyclohexanone. The radicals in the reaction pathway described here are essentially the same as produced in conventional liquid- or gas-phase autoxidation. So, the confinement should play an essential role in the course of the reaction; otherwise the remarkable selectivity cannot be understood. A first effect of the zeolite is the presence of a substantial electrostatic field around the extra-framework alkali, or alkaline-earth metal ion. This electrostatic field stabilizes the highly polar charge-transfer state, and shifts the frequency of the light from the UV to the visible part of the spectrum. After the formation of the radical pair, their coupling to the hydroperoxide in a restricted environment is the second effect of the zeolite. Thirdly, the basic properties of the zeolite enable selective decomposition of the hydroperoxide to the ketone and water. Finally, both overoxidation of the ketone, and a possible oxidation of a second methylene group, are avoided mainly by strong binding of the carbonyl compound to the alkali or alkaline-earth metal ion, thus blocking the catalytic center. Despite these positive findings, a lot of problems still have to be solved before this piece of science can be transformed into reliable technology. Apart from chemical engineering problems in the application of photocatalysis, there are also several chemical issues. As indicated before, similar reactive intermediates and transient species seem to be formed as in liquid or heterogeneous gas-phase oxidation. Thus, like in the normal cases, the current selective zeolitic system might be very susceptible to leakage to a free radical pathway, thus spoiling the catalysis. Acid and radical sites in the

zeolite and a high temperature are obvious dangers. Furthermore, under basic conditions in the zeolite, condensation reactions of the carbonyl compound can occur easily. Finally, the same strong physisorption that avoids overoxidation also seriously hinders product desorption and, thus, effective catalysis.

4.2. Direct oxidation of n-hexane to adipic acid in CoAlPO-18 In 2000, J.M. Thomas et al. published about the successful design of a molecular sieve catalyst for the aerial oxidation of n-hexane to adipic acid [10]. FRCA in the liquid phase would lead to excess C2 and C3 oxyfunctionalization instead of the observed double C1 selectivity. The actual catalyst is a highly Co-substituted AlPO-18. The importance of the work lies in a series of aspects of the catalyst, which are all critical for eventual success. In the active catalyst, a rather high fraction (10%) of the Al(III) framework ions is replaced by Co(III). This leads to a situation in which two Co(III) ions in tetrahedral coordination are likely to be found at the opposite vertices of the chabazitic cage, separated by a distance of ca. 7.6 A˚, similar to the distance of two methyl groups in n-hexane. At a lower fraction of incorporated Co (4%), virtually neither adipic acid nor other 1,6-disubstituted intermediates like 1,6-hexanediol and 1,6-hexanedialdehyde are produced, albeit that the activity for C1 -oxyfunctionalization was left intact. The size and shape of the pore are also of great importance. While CoAlPO-34, which also has chabazitic cages like CoAlPO-18, shows a similar catalytic behavior, CoAlPO-36, which has a much larger pore aperture, behaves quite differently. Figures 4 and 5 show the main features of the CoAlPO-18 catalyst.

77


CoIII Al

7.6Å

P O

Al

P

O

Co

C

H

Figure 4. The interior of the chabazitic cage of CoAlPO-18, showing tetrahedral coordination of a framework-incorporated Co(III) (left) and the size of the aperture compared to the size of a methyl end-group of n-hexane [11].

Also, in this case many problems have to be solved before application comes into sight. One of the first remarks that should be made is that n-hexane is not a common feedstock in industrial chemistry; thus its direct application for adipic acid does not seem very likely. However, the beautiful scientific concept, in principle, can be expanded toward other alkanes that are available feedstocks for the industry: C3 (propane) and C4 (butane, butenes). In principle, optimization might lead to higher yields of desired ; !-disubstituted products (diols, dials, and diacids). More serious problems are the low rate of the reaction, probably related to dioxygen mass-transfer limitation and pore size of the catalyst. If the latter would be the case, there is a serious problem because the pore size is critical for selectivity. Also, desorption of the product might be nontrivial, especially in the case of diacids in which leaching is an always present threat. Finally, the (largescale) synthesis of these catalysts poses a serious challenge for any catalyst supplier.

4.3. Oxidation of NH3 to NH2 OH and its transformation into caprolactam The concept of confinement was recently also beautifully demonstrated by a publication of J.M. Thomas et al. [12]. It was shown that NH3 could be converted to NH2 OH in the presence of an oxidizing agent (O2 , H2 O2 , and TBHP) and subsequently into the oxime of cyclohexanone and caprolactam with a true bifunctional M1 ðIIÞM2 ðIIIÞAlPO-36. The function of the redox active center M2 is to catalyze the oxidation of NH3 , and the function of the (Brønsted) acid site at M1 is to catalyze the conversion of the oxime of cyclohexanone into caprolactam. Experiments with substituted AlPO-36 and AlPO-18, provided evidence that the mechanism is oxidation of NH3 to NH2 OH, followed by the formation of the oxime, rather than formation of the imine, followed by

C H

Figure 5. The interior of the chabazitic cage of CoAlPO-18, showing nhexane between two framework-substituted Co(III) ions at high loading of Co [11].

oxidation of the imine to the oxime. Figure 6 shows the most important features of the catalysts. Direct oxidation of NH3 to NH2 OH is rare but very attractive. Once again the versatility of the catalytic concept of confinement has been demonstrated. Still there is a lot to be done. The most important issue to resolve is the yield based on nitrogen. Until now, it is unclear whether the oxidation of NH3 to NH2 OH is selective or not. Among the possible side products, N2 is the most harmless, but N2 O, and NOx cannot be excluded yet.

5. Conclusions From the various briefly discussed examples, it has become clear that oxidation catalysis is largely complicated by the FRCA mechanism, usually operative in the bulk liquid phase. Attempts to utilize alternative catalytic pathways or chemical engineering concepts are useful in some particular cases but do not offer more general solutions. However, the concept of confinement seems to be a clear exception. Confinement has been successfully explored in photocatalyzed oxidation of hydrocarbons, frameworksubstituted transition metal–catalyzed oxidation of hydrocarbons, and even framework-substituted transition metal–catalyzed oxidation of NH3 . In all cases, normal dioxygen was used, apparent overoxidation was avoided, and remarkable regioselectivity was obtained, clearly because of the confined nature of the catalyst. A bit of care is still necessary. While the relatively few cases, using an alternative catalytic route, or chemical engineering concept, represent real large-scale industrial processes, the present status of confined catalysis is merely academic. However, this is also an encouraging message because approximately more than 40 years of

78


Figure 6. Graph of M1 ðIIÞM2 ðIIIÞAlPO-36 and M1 ðIIÞM2 ðIIIÞAlPO-36, with Co(III) as redox center, and Mg(II) ions to induce Brønsted acid large sites. The pore of AlPO-18 is too small (3.8 A˚) to gain access to cyclohexanone, oxime, or even TBHP. AlPO-36 has a pore size ð6:5 7:5 AÞ enough to enable bifunctional catalysis, leading to caprolactam [11]

optimization of these current industrial processes lay behind us, so the concept of confinement can be regarded as still very young. There is now the obvious challenge to the chemical industry to turn these beautiful scientific findings into real sustainable processes.

References [1] B. Ensing, F. Buda, P.E. Blo¨chl and E.J. Baerends, Angew. Chem., Int. Ed. 40 (2001) 2893. [2] B. Ensing, F. Buda, P.E. Blo¨chl and E.J. Baerends, J. Phys. Chem. Chem. Phys. 4 (2002) 3619.

[3] B. Ensing and E.J. Baerends, J. Phys. Chem. A 106 (2002) 7902. [4] W. Buijs, J. Mol. Catal. A 146 (1999) 237. [5] F. Agterberg, Copper-Catalyzed Oxidation of Carboxylic Acids (Thesis Leiden University, 1996). [6] R.W. Wasserman and R.W. Murray, Singlet Oxygen (Academic Press, New York, 1979). [7] J. Shailaja, J. Sivaguru, R.J. Robbins, V. Ramamurthy, R.B. Sunoj and J. Chandrasankar, Tetrahedron 56 (2000) 6927. [8] F. Blatter, H. Sun, S. Vasenkow and H. Frei, Catal. Today 41 (1998) 297. [9] H. Sun, F. Blatter and H. Frei, J. Am. Chem. Soc. 118 (1996) 6873. [10] R. Raja, G. Sankar and J.M. Thomas, Angew. Chem., Int. Ed. Engl. 39 (2000) 2313. [11] J.M. Thomas and R. Raja, Chem. Commun. (2001) 675. [12] R. Raja, G. Sankar and J.M. Thomas, J. Am. Chem. Soc. 123 (2001) 8153.