CHEMTECH

June 1996

CHEMTECH 1996, 26(6), 24-30.
Copyright © 1996 by the American Chemical Society.

Photocatalyzed oxidation of hydrocarbons in zeolite cages

The technique uses molecular oxygen with visible light and results in intermediate oxidation products with unprecedented high selectivity. Scale-up will be the challenge.

Heinz Frei, Fritz Blatter, Hai Sun

Oxidation of hydrocarbons by molecular oxygen is a key process in the chemical industry. Small, partially oxygenated hydrocarbons are used as building blocks in the manufacture of plastics and synthetic fibers, for example, and many play a role as intermediates in the synthesis of fine chemicals. In large-scale synthesis, the use of O₂ as an oxidant is dictated by economic factors. But reactions that use O₂ as the primary oxidant often produce large amounts of unwanted byproducts. In particular, autooxidation of small alkanes, alkenes, or aromatics is inherently unselective, whether conducted in the gas or liquid phase or catalyzed by transition metals (1­4). One major reason that selectivities are low is that the desired products (such as alcohols or carbonyls) are more easily oxidized by O₂ than the parent hydrocarbon. Overoxidation can be minimized only by keeping conversions low, a serious disadvantage from a chemical processing standpoint. Therefore, a major challenge in this field is to find a reaction path that affords the primary product with high selectivity at high conversion of the hydrocarbon.

There is a strong research effort under way to meet this challenge. For example, one approach focuses on porphyrin analogues of monooxygenase enzymes capable of catalyzing the oxidation of light alkanes with O₂ (5). Whereas most of these systems require stoichiometric reducing agents (6), perhalogenated iron porphyrins afford oxidation of alkanes as small as propane without a reductant (7). Other examples of metalloporphyrin-catalyzed conversions that run without a sacrificial reagent are the UV light-assisted oxidation of cyclohexane (8) and the epoxidation of small olefins by a Ru porphyrin (9). Oxometal functionalities in organometallic complexes and on metal oxide surfaces are being explored for catalytic hydrocarbon oxidation with O₂ (10). There are exciting developments in the use of redox molecular sieves and of heteropolyacid catalysts toward the same goal (10, 11).

We recently discovered a simple method that gives partial oxidation of small alkenes, alkanes, and alkyl-substituted benzenes by O₂ at unprecedented selectivity, even at high conversion of the hydrocarbon. The approach is based on visible light­induced chemistry of hydrocarbon€ O₂ collisional pairs in the cages of large-pore zeolites. Reactions are conducted at ambient temperature in the absence of solvent or photosensitizer. Here we will describe the most interesting reactions established thus far (Figure 1) and define issues that pertain to scale-up of the method.

Click to View Figure 1 (approx 200K)

Concept
Photons can change the electronic properties of molecules in a way that makes them reactive even in the absence of a catalyst or heating. Tight product control most likely will be achieved if the reaction partners are excited with photon energies below those required to dissociate molecules into free atoms and radicals. In this way, indiscriminate attack of these highly reactive fragments is avoided. This typically means use of low-energy visible light instead of short-wavelength UV light.

An ideal way to initiate reaction between two molecules excited below their dissociation limit is to shine light on them while they are held in a state of collision. Molecular cages of a solid matrix offer a natural environment for the formation of collisional pairs or weak complexes at high concentration. In earlier work with rare gas matrix cages at cryogenic temperature (12), we found that highly regio- and stereospecific oxygen transfer can be induced between NO₂ and small alkenes or alkynes with visible light, that is, at photon energies below the dissociation of NO₂ to free oxygen atoms. By contrast, selectivity was completely lost when irradiating NO₂€hydrocarbon pairs with UV light, despite the fact that the chemistry took place at 12 K (12). This finding highlighted the importance of access to low-energy reaction paths with long-wavelength light and motivated us to venture into solid matrix environments that are stable at ambient temperature.

Zeolites were a natural choice for a nanoporous matrix in which we could conduct light-driven synthesis at almost any temperature (13, 14). Zeolites are natural or synthetic crystalline aluminosilicates that exist in a variety of structures (15). Ideal for our purpose are so-called faujasite zeolites, especially the synthetic analogue of type Y (see Sidebar ZeolityY). The structure consists of a three-dimensional network of molecular-size cages ("supercages") in which collisional pairs of small molecules such as light hydrocarbons and O₂ can be formed in high concentrations. An important characteristic of the alkali and alkaline earth­exchanged forms of zeolite Y is the high electrostatic field inside the supercage. Fields on the order of 1 V/Å have been predicted and confirmed experimentally (see Sidebar Electrostatic fields in zeolites). Aside from these electrostatic fields, the zeolite matrices used in our work are chemically inert. Specifically, zeolites are deliberately kept free of acid sites (Brønsted or Lewis). These catalytic sites play a crucial role in hydrocarbon cracking in the petroleum industry and in synthetic applications in the chemical industry (18). Na⁺- and Ba²⁺-exchanged zeolite Y used in most of our work can readily be prepared free of acid sites (19).

TO SIDEBAR: Zeolite Y

Toluene to benzaldehyde
When we loaded a dehydrated BaY matrix with toluene (5 Torr) and O₂ (1 atm) from the gas phase and exposed it to visible light at room temperature, we found that benzaldehyde is produced without side reaction (20, 21). This was established by in situ monitoring by infrared spectroscopy. Zeolites are transparent to infrared light except for some regions 1200 cm^­1. Product identification, by comparison with spectra of authentic samples and by isotopic labeling (D, ¹⁸O), is unambiguous, because the bands for small organic guest molecules are sharp. Performing the photochemistry at ­70 °C, benzyl hydroperoxide was trapped. Bringing the matrix to ambient temperature resulted in dehydration to benzaldehyde, establishing the hydroperoxide as a reaction intermediate. Benzaldehyde was the sole final oxidation product, even on conversion of as much as one-half of the toluene loaded into the zeolite matrix. We used the visible output of a tungsten lamp or the emission of a continuous-wave dye laser for the photochemistry but did not note any difference in products or yields. Monochromatic laser light could readily show the wavelength dependence of the product yields.

Toluene-to-benzaldehyde oxidation by O₂ without overoxidation or side reaction is unprecedented. Benzaldehyde is an industrial intermediate for the synthesis of agrochemicals, flavors, and fragrances. Co³⁺-catalyzed autooxidation of toluene--used currently in an industrial process for benzaldehyde synthesis--lacks this selectivity, mainly because of overoxidation to benzoic acid (1, 2). As a result, the benzaldehyde product has to be removed from the reaction mixture after very little conversion (only a few percent). In contrast, continued photoreaction of benzaldehyde with O₂ in the zeolite is prevented, because the ionization potential of the aldehyde (9.5 eV) is higher than that of toluene (8.8 eV). Consequently, the benzaldehyde€ O₂ charge­transfer absorption does not absorb visible light; overoxidation, therefore, cannot occur.

The UV-visible spectrum of the toluene and O₂^­ loaded zeolite matrix has a continuous absorption tail extending into the green spectral region (the pellets consist of strongly scattering 1-µm crystallites that require measurement in the diffuse reflectance mode). This band appears only when toluene and O₂ are simultaneously present in the zeolite, hence it originates from the electronic transition of the collisional complex between the two molecules. Of the two conceivable assignments of this band, an O₂-enhanced triplet absorption of the hydrocarbon or a toluene€ O₂ charge­transfer transition, only the latter is consistent with our data (20). Therefore, on absorption of a visible photon, an electron is transferred from the hydrocarbon to the oxygen molecule, resulting in the formation of a hydrocarbon radical cation and an 0₂ (superoxide) ion_o transitions of hydrocarbon€ O₂ contact complexes were discovered in the 1950s in O₂-saturated liquids and high-pressure O₂ gas phase by Evans and by Tsubomura and Mulliken (22, 23). They appear as continuous absorption tails in the UV region. In the presence of the high electrostatic field, the highly polar charge­transfer state is stabilized by 1.5-3.0 eV (35-70 kcal/mol), causing a large red shift of the absorption from the UV into the visible range. It is the effect of the electrostatic field inside the zeolite Y supercage on the hydrocarbon€O₂ charge­transfer absorption that gives us access to the mild oxidation path when we irradiate the matrix with visible light.

Propylene to acrolein and propylene oxide
Autooxidation of small olefins in the conventional phase is completely unselective because of radical chain reactions that result in the formation of oxy radicals. These radicals are very reactive and undergo several competing reactions that result in a multitude of products. Our photochemical method in zeolites offers a new path that affords a high degree of control even for the most difficult reaction: the oxidation of propylene.

Irradiation of propylene and O₂-loaded zeolite BaY at room temperature with green or blue light­induced partial oxidation of the olefin (24). Readily identified products were acrolein, allyl hydroperoxide, and propylene oxide. The hydroperoxide was found to be stable when the zeolite was kept at ­100 °C, so photolysis experiments at this temperature allowed us to find out about the origin of the aldehyde and epoxide. Allyl hydroperoxide was the main product at ­100 °C, and the remaining 13% was propylene oxide. Warming the zeolite after photoaccumulation of the hydroperoxide produced propylene oxide when excess propylene was kept in the matrix; acrolein was made when the olefin was removed before warming. Hence, allyl hydroperoxide is the primary photoproduct, and acrolein originates from thermal dehydration of the hydroperoxide.

Heterolytic thermal rearrangement to the corresponding carbonyl compound under elimination of H₂O occurs in the case of all hydroperoxides with an alpha C­H group. Propylene oxide is produced by O transfer from allyl hydroperoxide to excess olefin. The thermal rearrangement of the hydroperoxide to acrolein exhibits a steep temperature dependence, whereas the epoxidation reaction does not. Therefore, the aldehyde is the preferred final oxidation product of the visible light­induced propylene oxidation at elevated zeolite temperature. For example, when conducting the propylene and O₂ photochemistry at 55 °C, the acrolein-to-propylene oxide ratio is 2/1.

TO SIDEBAR: Electrostatic fields in zeolites

Manipulation of the olefin loading level furnishes an additional means of controlling the aldehyde-to-epoxide branching ratio. Based on diffuse reflectance spectroscopy of the visible propylene€ O₂ charge­transfer absorption and the measured infrared product growth, a rather high reaction quantum yield (defined as the product growth per absorbed photon) of 20% was estimated.

One current industrial process for the oxidation of propylene to acrolein with O₂ uses a bismuth molybdate catalyst (25). No selective method for conversion of propylene to propylene oxide by O₂ via thermal catalysis has been reported to date. Propylene epoxidation, a major industrial process, uses either t-butyl hydroperoxide or phenyl ethyl hydroperoxide as the O donor (1, 2). A recently discovered method based on a Ti-doped zeolite (Ti-silicalite) allows the use of aqueous H₂O₂ for olefin epoxidation (26).

Oxidation of alkanes
Especially interesting is the discovery of small alkane oxidation by this method at high selectivity, even at high conversion of the hydrocarbon.

The first reaction that we explored was oxidation of isobutane. Because its ionization potential is only a few tenths of an electronvolt higher than that of propylene, we speculated that light at blue wavelengths would promote reaction with O₂ in Ba²⁺-exchanged zeolite Y. When isobutane (1.6 Torr) and O₂ (900 Torr) were loaded into a BaY pellet and irradiated with blue light, t-butyl hydroperoxide was produced with 98% selectivity (27). The remaining products were trace amounts of acetone and methanol, the established secondary thermal products of the hydroperoxide. The infrared spectrum before and after photochemical conversion of 60% of the isobutane is shown in Figure 2. No loss of selectivity is noticed, even at this high conversion. All positive bands originate from t-butyl hydroperoxide except the 1686 cm^­1 absorption, which belongs to acetone. The negative bands show the depletion of isobutane. The reaction quantum efficiency is around 15%.

Figure 2. Infrared difference spectrum on 60% conversion of isobutane and O₂ with blue light in zeolite BaY.

We consider efficient proton transfer from the alkane radical cation to O₂^­

t-Butyl hydroperoxide is a major industrial oxidant for the transformation of small olefins to epoxides. For example, propylene oxide is produced on a large scale by metal-catalyzed oxygen transfer from t-butyl hydroperoxide to propylene (1, 2). The hydroperoxide is generated by direct reaction of isobutane with O₂ at 100­140 °C. Despite a vast research effort, a selectivity of no more than 75% is achieved in the liquid-phase autooxidation at low conversion (8%) under practical conditions (28). Aside from the selectivity of our photochemical method, synthesis in the zeolite offers an opportunity for in situ use of the hydroperoxide, thus avoiding the need for accumulation and storage of this hazardous oxidizer. In a typical experiment, the unreacted oxygen and isobutane were pumped off after photoaccumulation of t-butyl hydroperoxide, and trans-2-butene was loaded into the zeolite. Rapid growth of trans-2,3-epoxy butane and t-butanol was observed in the dark at room temperature (Figure 3). No formation of cis-epoxide was noticed, even on conversion of more than one-third of the trans-butene. Complete stereospecificity was observed when we conducted the epoxidations of cis-2-butene by the same method.

Figure 3. Thermal epoxidation of cis- and trans-2- butene at room temperature in zeolite BaY is stereospecific.

Cyclohexane and O₂ in zeolite NaY irradiated with green or blue light resulted in oxidation of the alkane to cyclohexyl hydroperoxide and cyclohexanone (plus water) as the exclusive products (29). Infrared spectroscopy revealed that the hydroperoxide rearranges at room temperature in the dark to cyclohexanone without any side reaction. Specifically, no cyclohexanol absorption was observed. Complete selectivity persisted, even at 40% conversion of cyclohexane. This result is distinct from all other methods of heterogeneous or homogeneous catalytic oxidation of cyclohexane by O₂ in which cyclohexanol is a major coproduct (10).

The largest industrial process in which partial oxidation of cyclohexane by O₂ plays a role is nylon production (2). Despite extensive improvement in cyclohexane autooxidation, the selectivity of the current liquid-phase technology is limited to 70% (combined cyclohexanone plus cyclohexanol yield) at low conversion (1, 2). The main obstacle in the substantial improvement of the selectivity at high alkane conversion is overoxidation of the alcohol and ketone product. Whereas a mixture of cyclohexanone and cyclohexanol is acceptable for nylon production, cyclohexanone is desirable as the sole oxidation product in a variety of fine chemical syntheses (10, 30).

TO SIDEBAR: Origin of selectivity

Propane was oxidized under irradiation with blue light when several hundred Torr of the alkane and 1 atm of O₂ were loaded in zeolite BaY. The final products were only acetone and H₂O, and isopropyl hydroperoxide was observed as an intermediate. The latter dehydrates spontaneously to acetone at ambient temperature. The ionization potential of ethane is 0.4 eV higher than that of propane, thus ethane could not be expected to react with O₂ in a BaY matrix. However, we did observe photooxidation of ethane with blue light from a tungsten source or a laser in zeolite CaY. Acetaldehyde and H₂O were the exclusive products. No CO₂ was formed. By contrast, all of the thermal catalytic methods yield carbon dioxide as a major byproduct. These completely selective oxidations bring about the possibility of using ethane and propane, constituents of natural gas, as new feedstocks for acetaldehyde and acetone in place of petroleum-derived ethylene and propylene. More generally, the visible light­induced oxidation of isobutane, cyclohexane, propane, and ethane by O₂ constitutes a mild and highly selective method for activation of the C­H bond of these alkanes (6).

Challenges for scale-up
Typical rates of the visible light­induced hydrocarbon oxidations in zeolites are of the order of 0.1 mmol/h if light from a lamp with 100 W of power in the blue-green region is focused on a 1-cm² zeolite pellet of 20-µm thickness. This corresponds to 0.1 mol/cm³ h, a reasonable space­time yield for an industrial process (10).

Scale-up of our micromolar-quantity experiments hinges on progress in three areas. One is reduction of the scattering of visible light by the zeolite matrix to ensure that the light reaches all reactants. An ideal solution would be transparent membranes like the ones already available for pentasil-type zeolites (e.g., ZSM-5) (35). We envision stacks of such transparent membranes with free gas flow in between them to facilitate adsorption of reactants and desorption of products (here nature provides a model--the stacked leaves of a tree). Alternatively, one could think of a photochemical fluidized-bed reactor in which the visible light penetrates into a cloud of zeolite particles that would then be collected for product desorption.

A second challenge is obtaining acceptable desorption rates of products from the zeolite matrix. Although the release of small oxygenated products from zeolite Y by polar organic solvents is routine, a solvent-free method is clearly preferable. Desorption of the oxygenated hydrocarbon products, which are rather polar, is easier with lower alkali (alkaline earth) ion concentrations. On the other hand, these cations are responsible for the red shift of the hydrocarbon€ O₂ charge­transfer absorption. Hence, a balance between these two effects has to be found. One possibility is to use a continuous-flow process, perhaps at a modestly elevated temperature, to facilitate product desorption, with the reactant O₂ as the carrier gas.

The third challenge is the photochemical reactor design. This problem is not new, but the parameters for the visible light­driven oxidations may be considerably different from previously considered designs (36). Here only conventional visible lamps are required, and there is no need for photosensitizers. With the recent development of highly efficient visible light sources, there will be a clear economic advantage over traditional designs that use UV arc lamps.

The future
The reactions of small hydrocarbons with molecular oxygen demonstrate the potential of visible light­induced chemistry for commercially important syntheses. The high selectivity of the mild photochemical routes may be especially attractive for the manufacture of fine chemicals. Many of these transformations are characterized by regio- and stereochemical complexity that has thus far prevented the use of O₂ as an oxidant. This photochemical method opens up possibilities for the use of O² in partial oxidation reactions now being conducted with far more expensive and/or polluting oxidants.

More generally, the visible light­induced reaction of hydrocarbon€ O₂ collisional pairs in zeolites encourages the exploration of photochemistry in nanocage materials for other synthetic transformations. It might be possible to activate CO² by a bimolecular photochemical reaction in a zeolite with a very high electrostatic field, for example. Discovery of new chemistry, or much milder reaction paths for established reactions, need not be limited to cation-exchanged zeolites. Microporous materials with nonionic frameworks may be just as interesting. For example, the cages of zeolites with all silicon frameworks or aluminum phosphate molecular sieves may offer ideal environments for selective bimolecular photochemistry. The distinct hydrophilic and hydrophobic properties of these materials furnish means to influence adsorption and desorption rates of reactants and products.

The pace of progress with new systems will depend on a more detailed mechanistic understanding of already established reactions. Through better insight by in situ monitoring techniques, such as nanosecond Fourier transform infrared spectroscopy, we will learn about the factors that influence yields and selectivity and find ways to enhance them.

Acknowledgment
This work was supported by the Director, Office of Energy Research, Office of Basic Energy Sciences, Chemical Sciences Division, U.S. Department of Energy, under contract no. DE-AC03-76SF00098.

REFERENCES

ACS Publications Division Page ACS Home Page