A One-Step Conversion of Benzene to Phenol with a Palladium Membrane

See allHide authors and affiliations

Science  04 Jan 2002:
Vol. 295, Issue 5552, pp. 105-107
DOI: 10.1126/science.1066527


Existing phenol production processes tend to be energy-consuming and produce unwanted by-products. We report an efficient process using a shell-and-tube reactor, in which a gaseous mixture of benzene and oxygen is fed into a porous alumina tube coated with a palladium thin layer and hydrogen is fed into the shell. Hydrogen dissociated on the palladium layer surface permeates onto the back and reacts with oxygen to give active oxygen species, which attack benzene to produce phenol. This one-step process attained phenol formation selectivities of 80 to 97% at benzene conversions of 2 to 16% below 250°C (phenol yield: 1.5 kilograms per kilogram of catalyst per hour at 150°C).

Phenol is an important commodity chemical in industry: World production exceeded 6.6 megatons in 2000. Industrially, phenol has mainly been produced from benzene via cumene to cumene hydroperoxide (the so-called cumene process), but this three-step process (1) not only has a low phenol yield but is also highly energy-consuming. Furthermore, problems arise in treating its by-products, such as acetone and α-methylstyrene. Recently, direct oxidation of benzene to phenol by nitrous oxide (2–4) has been commercialized, but it is cost-effective only if nitrous oxide can be obtained cheaply as a by-product (5). Here we report an efficient one-step oxidation of benzene to phenol through direct hydroxylation of an aromatic ring in gas phase with oxygen activated by dissociated hydrogen obtained from a palladium membrane.

All direct hydroxylations (6–16) of aromatic nuclei with oxygen and hydrogen that have been reported so far have been done by simultaneously mixing an aromatic compound, oxygen, and hydrogen in liquid phase, using a very complicated system containing a multicomponent catalyst, a solvent, and some additives. Besides the possibility of an explosive gas reaction, these hydroxylations give only very low aromatic alcohol yields of 0.0014 to 0.69% (based on the amount of aromatic hydrocarbon initially used). We developed the direct hydroxylation of aromatic nuclei through a system in which hydrogen and oxygen are separately supplied or in which hydrogen is fed into a mixed gas stream of a substrate and oxygen through a metallic thin layer. This system is quite simple and appears to be practical when compared with other direct hydroxylations (17–25) reported recently.

The membranes were prepared by coating a porous α-alumina tube (NOK Corporation; α-Al2O3, 99.99%; outer diameter, 2.0 mm; inner diameter, 1.6 mm; void fraction, 0.43; average pore size, 0.15 μm) with a palladium thin layer (thickness, 1 μm; length, 100 mm) by means of a metallorganic chemical vapor deposition technique (26), using palladium(II) acetate (reagent grade) as the palladium source. At 300°C, the hydrogen and nitrogen permeation rates of the membranes prepared were 1.0 to 3.0 × 10−3 mol m−2 s−1Pa−0.5 and 0.1 to 1.0 × 10−10 mol m−2 s−1 Pa−1, respectively.

A shell-and-tube reactor was set up with this membrane (Fig. 1) and used for the hydroxylation of aromatics. An aromatic hydrocarbon was fed into the membrane tube together with a mixed gas of oxygen and helium by bubbling it into the hydrocarbon liquid, and a pressurized mixture of hydrogen and helium (about 2 kg/cm2) was flowed into the shell (outside of the tube). In some cases, the feeding of the gas mixtures was carried out in reverse; that is, hydrogen was fed inside of the tube and a hydrocarbon and oxygen were outside. The gaseous mixture coming out of the reactor was analyzed by an online gas chromatograph equipped with a capillary column (diameter, 0.25 mm; length, 25 m; packing reagent, PEG-20 M wide bore).

Figure 1

Apparatus for direct hydroxylation of aromatics and the working principle of the palladium membrane.

This palladium membrane reactor works well under mild conditions (below 250°C) for the direct hydroxylation of an aromatic nucleus. The hydrogen permeation rate of the membrane markedly increased with feeding of a gaseous mixture of oxygen and a substrate, as compared with the rate without such feeding (the intrinsic hydrogen permeation rate) (27). Phenol was produced in a high selectivity over 90% at low benzene conversions (below 3.0%) and over 80% at high conversions (10 to 15%) (Table 1 and Fig. 2). Dihydroxy compounds and quinones were detected in trace amounts only at the high conversions, which suggests that successive oxidations were considerably limited. Under our reaction conditions, hydrogenated compounds such as cyclohexane, cyclohexanol, and cyclohexanone were hardly produced at all. Thus, it is concluded that the hydrogen consumed turns almost completely into water; the molar ratios of water to phenol were 5 to 9 (28). However, when the reactions were prolonged, cyclohexanol and cyclohexanone were detected in small amounts, and the deterioration of the membrane was also observed (29).

Figure 2

Oxidation of benzene with oxygen and hydrogen catalyzed by a palladium membrane at 200°C. Flow rates: shell (outside), 25 ml/min (H2/He = 5.6/20, volume ratio); tube (inside), 25 ml/min (benzene/O2/He = 0.4/3.8/25, volume ratio). Solid circles, squares, and open circles denote benzene conversions, phenol yields, and phenol selectivities, respectively.

Table 1

Direct hydroxylation of benzene to phenol and toluene to cresol with oxygen and hydrogen catalyzed by a palladium membrane. “Inner” and “outer” mean that a gaseous mixture containing a hydrocarbon was flowed inside or outside of the palladium membrane tube, respectively. Selectivity was based on the amount of benzene or toluene consumed.

View this table:

When the substrate was toluene, a mixture of cresols (o-:m-:p-isomer ratio = 3.2:1.0:5.1) was produced in a high selectivity (>80%) even at a high toluene conversion (>34%). Methyl cyclohexane (hydrogenated compound) was hardly produced at all. Aromatic by-products (dihydroxy compounds) were barely detected (<1%), and methyl group–oxygenated compounds such as benzaldehyde and benzyl alcohol were produced in small amounts (<5% in sum of the two). Direct hydroxylations (12–16) of toluene reported recently, such as those catalyzed by 0.5 weight % (wt %) Pt/1 wt % V2O5/SiO2(12), EuCl3-Zn-CH3COOH (14), and EuX3-TiO(acac)2-Pt oxide/SiO2 (15) (acac, 2,4-penntanedionate ion; X = Cl, ClO4, and CF3SO3 ), had respective ratios (12–15) of ring oxygenation to methyl group oxygenation of 3.7, 2.6, 0.2, 0.6, and 7.0. In the present work, this ratio was more than 8.0.

Generally speaking, the methyl group oxygenation requires abstraction of an H· radical by active oxygen species, whereas ring oxygenation is initiated by the attack of active oxygen species on the ring (15). Active oxygen species that are electrophilic tend to prefer the ring to the methyl group, and oxygen species that have a radical character prefer the methyl group to the ring (15). Thus, the active oxygen species in the present process appear to be relatively more electrophilic.

Among the various active oxygen species (denoted as O* in Fig. 1) considered {for example, O [atomic oxygen at the ground state:3P, oxene (7)], HOO·, HO·, O2–, O, OH, etc.} negatively charged species are unlikely (7) to be the active species because they are nucleophilic but not electrophilic. The former three species can act as the electrophile and be produced as follows. Hydrogen is dissociated in permeating through the palladium membrane (30–32), and the dissociated hydrogen (denoted as H* in Fig. 1) appearing on the surface of the opposite side of the membrane immediately reacts with oxygen to give HOO· and H2O2. Then H2O2 is decomposed to HO·, atomic oxygen, and water. In the direct hydroxylation (8–12) of hydrocarbons with oxygen and hydrogen catalyzed by transition metals, it has been considered so far that, at the first step, oxygen and hydrogen react with the catalysis of the metals to give hydrogen peroxide. We also recognized that this membrane reactor can easily produce hydrogen peroxide under the reaction conditions used here without hydrocarbons, and quite recently it was reported (33) that a similar palladium membrane works well in water as a reactor for the direct production of hydrogen peroxide from oxygen and hydrogen. Oxene, one of the active oxygen species produced by the decomposition of hydrogen peroxide, has been known to easily add to carbon-carbon double bonds, including conjugated ones such as benzene (7, 34). This type of addition is 103 times faster in rate than the hydrogen abstraction from the methyl group (7,35). Thus, it is not unreasonable to consider that oxene is largely responsible for the hydroxylation in this membrane process.

If the HO· radical is the main active species, as has been considered (12), benzyl alcohol and benzaldehyde should be produced more in the hydroxylation of toluene, as occurs with EuX3-TiO(acac)2-Pt oxide/SiO2(X = Cl and ClO4) catalysts (15). However, it is difficult to specify the real active species from the three (HOO·, HO·, and oxene) at present, although it seems likely that the active oxygen species is derived from HOO· and H2O2.

It should be emphasized that this membrane system could be practical, because it is simple in structure; produces phenol in a yield of 1.5 kg per kilogram of catalyst per hour; and has a low probability of causing a detonating gas reaction, because oxygen and hydrogen are not simultaneously mixed.

  • * To whom correspondence should be addressed. E-mail: f-mizukami{at}


View Abstract

Navigate This Article