Vous êtes sur la page 1sur 72

11

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

11.1

Oxidation processes

11.1.1 Gas phase oxidation processes


Introduction

Selective oxidation processes, in particular those that make use of solid catalysts (heterogeneous oxidation processes), play a fundamental role in the petrochemical industry. About 50% of the principal chemical products and over 80% of monomers are synthesized by means of at least one stage of selective heterogeneous catalytic oxidation. Table 1 contains a list of the main selective oxidation processes for hydrocarbons using solid catalysts, with an indication of the conversion and selectivity values obtained. In many commercial selective oxidation processes there is still scope for a significant margin of improvement in performance. For example, the potential increase in selectivity in the two main processes of selective oxidation (ethylene to ethylene oxide and propylene to acrylonitrile) could result in annual savings in reagent costs of around 800 million of euro. The action of solid catalysts in oxidation processes had already been noted by the beginning of the Nineteenth century, but it was only towards the middle of the Twentieth century that a systematic study of selective oxidation processes using solid catalysts and of their industrial applications was begun. The first processes to be developed industrially were: oxidation and ammonia oxidation (oxidation in the presence of ammonia) of propylene to produce acrolein and acrylonitrile respectively, oxidation of ethylene to ethylene oxide and the oxidation of aromatics to form anhydrides (maleic and phthalic anhydrides). The development of these processes, which was also driven by the growing demand for these types of products, led to the development of fundamental research, with a synergistic effect on both the development of new applications and the improvement of those already on the market.

An example of this is the ammonia oxidation process of propylene using air and ammonia, which quickly replaced the previous process based on the reaction between acetylene and HCN, both because of the lower raw material cost and the reduced safety issues. This made it possible to produce acrylonitrile with a significant reduction in costs, which resulted in a rapid expansion in the market for it between 1960-80. On the other hand, the success of this product stimulated the development of research into the catalysts being used (mixed Bi and Mo based oxides), resulting in their gradual improvement. The first generation catalysts, based on supported Bi9PMo12O52, gave a yield of 55%, which increased to 65% with the development of second generation systems containing iron as the redox element and to about 75% with the development of third generation multi-component catalysts. The current fourth generation catalysts, containing up to 25 elements, allow yields in excess of 80% to be obtained. The development of new catalysts has brought about a comparable evolution in the type of catalytic reactors used, initially fixed bed, then bubbling fluid bed and finally braked fluid bed. In the period from 1990-2005 development and innovation in the sector was instead driven by the growing importance attached to environmental and safety issues. However, in the last decade of that period the introduction of new processes was heavily influenced by the reduction of investment in petrochemicals resulting from the restructuring taking place in businesses throughout the sector. Below is a summary of the principal lines of development during that time (Centi and Perathoner, 2003b). Use of new raw materials and alternative oxidizing agents. There has been an increasingly wider use of alkanes as raw materials, instead of aromatics and alkenes; for example, the synthesis of acrylonitrile from

VOLUME II / REFINING AND PETROCHEMICALS

617

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

Table 1. Principal processes of selective oxidation of hydrocarbons using solid catalysts

and typical results obtained (Arpentinier et al., 2001; Centi et al., 2002)
REAGENT Methane/O2/NH3 CH4 or (CH2)x /O2 Methanol/air Ethylene/O2/acetic acid Ethylene/O2 Ethylene/air or O2/HCl Ethanol/O2 Propylene/air Propylene/air/NH3 Acrolein/air n-butane/air n-butane/air tert-butyl alcohol Isobutene/air Methacrolein/air Benzene/air o-xylene/air Naphthalene/air
* **

Principal product HCN Syngas (CO/H2) Formaldehyde Vinyl acetate Ethylene oxide 1,2-dichloroethane Acetaldehyde Acrolein Acrylonitrile Acrylic acid Maleic anhydride Butenes/butadiene Methacrolein Methacrolein Methacrylic acid Maleic anhydride Phthalic anhydride Phthalic anhydride

Types of catalysts Lattice of Pt-Rh Supported Rh or Ni Ag on a-Al2O3, or Fe-Mo oxides Pd-Cu-K on a-Al2O3 Ag-K-Cl on a-Al2O3 Oxychlorides of Cu-Mg(K) on g-Al2O3 Ag, Cu Bi-Mo-Fe-Co-K supported oxides Bi-Mo-Fe-Co-K supported oxides V-Mo-W oxides V-P oxides Bi-Mo-P oxides Bi-Mo-Fe-Co-K oxides Bi-Mo-Fe-Co-K oxides V-Mo-W oxides V-Mo oxides Oxides of V-P-Cs-Sb on TiO2 Oxides of V-K on SiO2

Conversion* (%) 100 99 97-99 8-12** 13-18** 95 45-50** 92-97 98-100 95 75-80 55-65 99 97 97-99 98 98-100 100

Selectivity* (%) 60-70 90-95 91-98 92 72-76 93-96 94-96 80-88 75-83 90-95 67-72 93-95 85-90 85-90 95-98 75 81-87 84

Conversion of the reagents and selectivity of the products compared with the hydrocarbon In the processes in which the operation includes recycling of the unconverted reagent, the conversion figure is for a single pass

propane instead of propylene and the synthesis of maleic anhydride from n-butane instead of benzene, aimed at reducing costs and/or improving the eco-sustainability of the process. New processes that use alternative oxidants are being researched. An example is the direct synthesis of phenol from benzene (instead of the multi-stage processing of benzene with cumene as an intermediate), using N2O as the oxidizing agent instead of O2. This is in order to reduce the complexity and the risks associated with the process, to avoid the co-production of acetone and to make use of a by-product such as N2O (thereby also reducing its disposal costs). Development of new classes of catalysts and processes. The processes that use solid (heterogeneous)

catalysts are increasingly replacing the homogeneous type, in order to reduce separation costs and the environmental impact and/or to use new raw materials, for example in the direct synthesis of acetic acid from ethane. The processes for oxidative dehydrogenation of alkanes are increasingly more competitive than those for dehydrogenation of alkenes. New processes are also being investigated which will enable the reduction or elimination of the formation of co-products and/or the formation of toxic or dangerous intermediates. An example is the synthesis of methacrylic acid through direct isobutane oxidation, as an alternative to the commercial acetone cyanohydrin process, which uses HCN as a reagent and co-produces ammonium sulphate.

618

ENCYCLOPAEDIA OF HYDROCARBONS

OXIDATION PROCESSES

Conversion of processes based on the use of air into processes based on the feeding of pure oxygen. These processes enable a reduction in polluting emissions; as examples there are the synthesis of formaldehyde from methanol, the epoxidation of ethylene and the oxychlorination of ethylene to 1,2-dichloroethane. Improvement of the productivity of the processes. This is the result of the development of new generation catalysts with improved properties and/or improvements in the engineering of the reactors (for example, the introduction of a monolithic reactor in the synthesis of formaldehyde, or of structured-bed reactors in the synthesis of phthalic anhydride). Moreover, during the period from 2000-05 there was a significant increase of interest in the development of new reactor technologies (such as, for example, membrane reactors), which made it possible to achieve savings in processing even for small-medium scale production (scale-down of the processes; Centi and Perathoner, 2003a). The goal was to decentralize production and reduce its environmental impact, in contrast with the trend typical of the Twentieth century of achieving savings in processing costs through increases in scale and high integration in large petrochemical facilities. This came about due to the high environmental impact and the strong public opposition to the latter approach, as well as due to problems linked to a sluggish market with large fluctuations in demand. Selective catalytic oxidation processes can be divided into three categories. The first relates to oxidation of inorganic molecules (for example, oxidation of ammonia to NO and of H2S by sulphur). The second class relates to synthesis of basic chemical products (for example, ammonia oxidation of methane by HCN or the partial oxidation of methane by syngas; CO/H2 mixtures). Finally, the third category relates to conversion of hydrocarbons by processing in the liquid phase (principally in the homogeneous phase even if there is a growing interest in the use of heterogeneous catalysts) and processing in the gas phase, which is the most commonly used industrially (see again Table 1). It should be pointed out that this last class of processes uses air or O2 as the oxidant (other than the cited process of direct hydroxylation of benzene by phenol with N2O), while in the liquid phase processes, in addition to O2, extensive use is also made of other oxidizing agents such as alkyl peroxides and H2O2 (Centi and Perathoner, 2003b). The different categories of gas phase selective oxidation processes (over solid catalysts) and the related principal industrial reactions are summarized in Table 2 (Arpentinier et al., 2001; Centi et al., 2002). Some important classes of reactions, which are not mentioned in the table, since they are not yet used commercially, include: oxidative dehydrogenation of C2-C5 alkanes to

their corresponding olefins; selective oxidation of alkanes such as synthesis of phthalic anhydride and maleic anhydride from n-pentane, of acrylic acid from propane and of methacrolein or methacrylic acid from isobutane; and ammonia oxidation of propane into acrylonitrile. The catalysts used for these reactions can be classified on the basis of their characteristic reaction mechanisms. Allylic oxidation. For these reactions catalysts based on mixed oxides of transition metals are used. These catalysts are capable of selectively extracting a hydrogen atom by breaking a C H bond in the allyl position and if necessary replacing it with an oxygen atom. Industrial catalysts are generally multi-component (for example, Bi-Mo oxides, used in the synthesis of acrylonitrile from propylene, contain various promoters such as Fe, Cu, W, Te, Sb and K), but typically a principal phase can be identified (Bi-molybdate) which is able to catalyse different reactions, such as: the synthesis of acrolein from propylene, the ammonia oxidation of propylene to acrylonitrile, the dimerization of propylene to cyclohexene and the oxidative dehydrogenation of butenes to butadiene. These reactions are characterized by a common first stage of allylic oxidation (Fig. 1), where the extraction of a hydrogen atom in the allyl position gives rise to a chemisorbed p-allylic complex on the transition metal. The nature of the subsequent stages determines the type of reaction and product that is obtained. Oxidation and ammonia oxidation of a side chain of alkyl aromatics (for example, the oxidation of toluene to benzaldehyde or benzonitrile respectively) in principle follow a similar reaction mechanism, but the interaction of the aromatic ring with the surface is different and therefore different types of catalysts are used, such as vanadium oxides supported on TiO2 or catalysts based on molybdate of Fe-(V P, K). , Nucleophilic oxidation to the C O group (oxidative dehydrogenation of alcohols and oxidation of aldehydes to acids). Although this type of reaction has similarities to the mechanism previously described, there are various types of substrates such as alcohols (methanol) or aldehydes (acrolein or methacrolein) which interact too strongly with the surface of the catalyst when catalysts belonging to the first category are used. In the conversion of methanol into formaldehyde, the catalyst most often used on an industrial level is iron molybdate (which also contains other components in small quantities), while multi-component catalysts, based on Mo-V oxides or heteropolyacids of P-Mo-V are used for , the conversion of aldehydes into their corresponding acids. Electrophilic insertion of an oxygen atom. The catalysts for this category of reaction are highly specific. Examples are the systems based on Ag/a-Al2O3 for the

VOLUME II / REFINING AND PETROCHEMICALS

619

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

H C H C

H H C H Me O Me O H H C H C H C H H Me O Me O Me O

H C H H Me C

H H C O H

Me O Me O anionic vacancy

H O Bi O O Mo O O OH Bi O O Mo O O OH Bi O O

H O Mo O

p-allylic complex

Fig. 1. General outline of the allylic oxidation mechanism and example

of the oxidation of propylene on bismuth molybdate to obtain acrolein.

synthesis of ethylene oxide from ethylene (this catalyst, for example, when applied to the synthesis of propylene oxide from propylene, is not selective) and Fe/ZSM-5 for the hydroxylation of phenol with N2O as the oxidant. Oxidation (or ammonia oxidation) of alkanes. In this case the slow stage is the initial selective activation of the alkane, for example for the concerted extraction of a

hydrogen atom by a surface Lewis site (a transition metal) and of a second hydrogen atom by a base site (oxygen atoms) to give an alkene, which is immediately converted into an oxygenated product through oxidation or allylic ammonia oxidation mechanisms. Catalysts with properties which differ from those of catalysts belonging to the first reaction category are necessary

Table 2. Different classes of gas phase selective oxidation processes (on solid catalysts) and the relative

industrial reactions (Arpentinier et al., 2001; Centi et al., 2002)


Type of reaction Allylic oxidation Examples propylene to acrolein or acrylic acid isobutene to methacrolein or methacrylic acid Synthesis of the acids can be carried out in a single stage from the alkene, but commercially it is preferred to use two stages for the best possible selectivities butenes to butadiene and isopentenes to isoprenes methanol to formaldehyde isobutyric acid to methacrylic acid epoxidation of ethylene to ethylene oxide with O2 direct synthesis of phenol from benzene with N2O synthesis of vinyl acetate from ethylene and acetic acid synthesis of 1,2-dichloroethane from ethylene and HCl in the presence of O2 propylene to acrylonitrile isobutene to methacrylonitrile a-methylstyrene to atroponitrile n-butane to maleic anhydride o-xylene to phthalic anhydride

Oxidative dehydrogenation

Electrophilic insertion of an oxygen atom Acetoxylation Oxychlorination Ammonia oxidation

Synthesis of anhydrides

620

ENCYCLOPAEDIA OF HYDROCARBONS

OXIDATION PROCESSES

because of the weak interaction of the substrate with the surface, and the activation mechanism. For example, catalysts based on vanadyl pyrophosphate are used for the oxidation of n-butane to maleic anhydride, or those based on vanadium antimonates are used for the ammonia oxidation of propane. In the latter case, antimony oxide is active in the ammonia oxidation of propylene, but is not able to activate the propane molecule; the addition of V gives the system the capability of oxidizing the alkane. Wacker-type oxidation mechanism. Vinyl acetate is produced through acetoxylation of ethylene with acetic acid in the presence of oxygen, with catalysts based on supported Pd/Au. Pd supported on V2O5/Al2O3 or V2O5/TiO2 is selective in the gas phase synthesis of acetaldehyde from ethylene or of methylethylketone from 1-butene, with a similar reaction mechanism. Oxychlorination. 1,2-dichloroethane is produced commercially from ethylene, HC1 and O2 on supported copper chloride based catalysts. The mechanism consists of a direct addition of chlorine atoms by the catalyst onto the olefin, rather than in oxidation of hydrochloric acid to molecular chlorine, followed by chlorination of the double bond. Addition of oxygen to the aromatic nucleus, with ring opening. The electrophilic attack of oxygen on hydrocarbon substrates typically leads to the formation of carbon oxides, however in the case of the oxidation of benzene, selective oxidation to maleic anhydride is obtained. This process, which employs catalysts based on mixed vanadium and molybdenum oxides, has been partially replaced by the synthesis by oxidation of n-butane. Similar catalysts are used in the selective oxidation of polyaromatic compounds. Non-classic oxidation mechanisms. Ethylbenzene can be oxidatively dehydrogenated, with high selectivity, to styrene on various catalysts such as oxides and phosphates, but the active phase is constituted by the formation of a thin surface layer of carbon containing the active sites of the reaction. Recently even some types of carbon and carbon nanotubes have shown high selectivity in oxidative dehydrogenation of ethylbenzene to styrene. Another example is the ammoximation of cyclohexanone (to cyclohexanone oxime) over amorphous silica.
Characteristics of gas phase oxidation processes
General aspects

the catalyst, but rather of the structural oxygen of the catalyst (typically mixed oxides, see again Table 1). The O2 oxygen ion removes the hydrogen atoms from the hydrocarbon with the subsequent formation of water or, if inserted into the molecular structure of the reagent, it gives rise to the formation of oxygenated compounds (see again Fig. 1). Instead, the gaseous oxygen intervenes in the reoxidation mechanism of the reduced catalyst, known as the Mars-van Krevelen mechanism (Fig. 2). The oxidation of the catalyst by O2 comes about through the formation of intermediate oxygen species such as O2 and O , which have electrophilic characteristics and tend to make an addition to the

H2O oxidized catalyst O2 M2m reduced catalyst oxide (catalyst) M1n

oxidized product

hydrocarbon

Fig. 2. Mars-van Krevelen mechanism of selective oxidation

of hydrocarbons on oxide based catalysts.

O2 O2 O2 M(n O2 O2
A
1)

O Mn O2 O2 M(n
1)

O2 2 O2 Mn O2 Mn O2 O2

O2

Mn

electrophilic oxygen species (O2 , O , ...) C O2 activation nucleophilic oxygen species (O2 ) C C C H

products with rupture of C C bond and formation of COx products of selective oxidation (e.g. aldehydes)

Although the petrochemical industry uses selective oxidation processes both in the gas phase and in the liquid phase, those in gas phase are more widespread. Typically, oxygen (or air) is used as the oxidizing agent, although the species involved in the selective oxidation reaction are usually made up, not of adsorbed oxygen on

Fig. 3. A, schematic mechanism of the incorporation of

oxygen into oxide based catalysts; B, outline of different types of attack on the hydrocarbon by nucleophilic and electrophilic species of oxygen (Centi et al., 2002).

VOLUME II / REFINING AND PETROCHEMICALS

621

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

unsaturated molecule, breaking the double bond and ending with the formation of carbon oxides; in contrast, the structural oxygen of the catalyst (O2 ) has nucleophilic characteristics (Fig. 3). Hence, in order to be selective, a catalyst must not only possess activation sites for the hydrocarbon and for selective insertion of the oxygen on the substrate, but must also be rapidly reoxidizable. This is so as to prevent the non-selective chemisorbed oxygen species from having a lifetime long enough to allow combustion reactions to take place. This mechanism is generally accepted for the oxidation of alkenes on mixed oxides, but there are doubts about its validity in the case of oxidation of other substrates, such as alkanes. A general characteristic of selective oxidation processes of hydrocarbons is the complexity of the reactions involved. For example, the oxidation of n-butane to maleic anhydride is a reaction which involves 14 electrons, the removal of 8 hydrogen atoms and the insertion of 3 oxygen atoms on the substrate, with the involvement of another 4 oxygen atoms of the catalyst to form 4 molecules of water. Notwithstanding the complexity of the transformation, the reaction takes place without the formation of products with intermediate levels of oxidation; selectivity of between 70 and 85% is achieved, depending on the reaction conditions. Therefore the catalyst, made up of a mixed oxide of V and P with the composition (VO)2P2O7, possesses characteristics such as to avoid both the desorption of the reaction intermediates and their non-selective transformation into carbon oxides. Finally, another characteristic of selective oxidation catalysts is their multi-functionality, which is necessary for the transformation of the hydrocarbon into the final product; in fact, to bring about the complex mechanism described above, it is necessary that the catalyst be capable of actuating different types of transformations on the substrate (Arpentinier et al., 2001). Moreover, it is necessary for the different stages involved in the transformation to have similar rates. Different relative rates could lead to the desorption of intermediate products or an increase in the rate of parallel reactions, with a reduction in the selectivity for the desired product. This is well illustrated in the selective oxidation of n-butane (Fig. 4).
Combined design of the catalyst and the reactor

n-butane COx, H2O concerted abstraction of 2 H atoms isomerization allylic H abstraction

1,4-oxygen insertion allylic dehydrogenation and/or allylic oxygen insertion oxygen insertion maleic anhydride

Fig. 4. Outline of the reaction mechanism in the selective

oxidation of n-butane to maleic anhydride on catalysts based on (VO)2P2O7, showing the multifunctional character of the catalyst.

Optimizing the yield, productivity and selectivity of selective oxidation reactions requires not only a detailed knowledge of the nature of the catalyst and the mechanism of the interaction of the reagents and products with the catalyst itself, but also the optimization of the reactor used. Recently, new reactor solutions (which enable, for example, the separation of the two

stages of interaction of the catalyst with the hydrocarbon and with oxygen) have led to the development of new classes of catalysts. The two aspects, involving the development of the catalyst and the engineering of the reactor, are therefore closely correlated. Industrial reactors used in the petrochemical industry for highly exothermic reactions, such as those for selective oxidation, are typically either of the multi-tubular fixed-bed or fluid-bed type. Nevertheless, there is a growing interest in the development of new reactor solutions, such as for example, the circulating fluid-bed reactor, recently applied by DuPont to the synthesis of maleic anhydride from n-butane. The uncoupling of the two redox reactions, of oxidation of the hydrocarbon by the catalyst and the reoxidation of the latter by oxygen (see again Fig. 2), makes it possible to increase the selectivity towards maleic anhydride compared with the reaction carried out in the simultaneous presence of both the hydrocarbon and oxygen. Other advantages of this type of reactor are isothermicity and a reduction of the risk of explosion. Nevertheless, a limiting factor is its low productivity; in fact it is necessary to circulate large quantities of the catalyst (equal to about 1 kg per g of maleic anhydride produced) between the two reactor vessels, each of which is adapted to one of the reaction stages. Another example of a new reactor configuration, adopted for petrochemical processes, is the monolithic type of reactor; these reactors combine the advantages of the possibility of autothermic conduction of the reaction

622

ENCYCLOPAEDIA OF HYDROCARBONS

OXIDATION PROCESSES

and a reduction in the loss of pressure. The new generations of processes for oxidation of methanol to formaldehyde use a final adiabatic stage (post-reactor), with a catalyst structured in the form of a monolith. Very interesting results have been obtained using reactor with extremely short contact times (on the order of milliseconds, compared with times measured in seconds in conventional reactor), where the catalyst configuration is also of a non-conventional type (for example, in a grid form). Given the high spatial rates used (that is the high ratio between the input rate of the reagents and the quantity of the catalysts) and the type of mechanism involved, it is possible to avoid subsequent oxidations and therefore to obtain high selectivity of the intermediate products (for example, in the oxidative dehydrogenation of alkanes into alkenes). Finally, it is worth remembering the developments in the field of catalytic membrane reactors, which allow the continuous removal of one of the products or the differential addition along the catalytic bed of one of the reagents (for example, oxygen). This makes it possible to maintain the optimum hydrocarbon/O2 ratio along the entire profile, to limit the formation of hot spots and to control the state of oxidation of the catalyst. Nevertheless, one of the current limiting factors is its low productivity, apart from the high cost of the membrane itself. Even conventional fixed-bed reactors can be improved through greater integration of the design of the catalyst and that of the reactor. In the process of the synthesis of phthalic anhydride from o-xylene, specialized catalytic beds are used, that is, containing different layers of catalysts each having a different composition, in order to optimize the axial activity and selectivity profile of the catalyst itself.

A new reactor technique being developed consists of systems in which the flow is periodically reversed; this makes it possible to make the activity and temperature profile in the reactor more uniform, even though there are still some significant problems involving the difficulty of managing non-stationary operations and their potential danger. Also in this case, the design of the catalyst is different from that for operations in stationary conditions.
Use of air and pure oxygen as oxidizing agents

Currently, air is the most widely used reagent in gas phase oxidation processes, but there is a growing interest in the use of pure O2 as a means of increasing the productivity and reducing pollutant emissions and energy consumption. Table 3 illustrates an example of the emissions from the process of oxychlorination of ethylene, where air and oxygen are used as the oxidizing reagents. The significant reduction in the environmental impact of the second type of process can be seen. The following gas phase processes use pure O2, or air enriched with oxygen, as an alternative to air: a) partial oxidation (to syngas) of heavy fractions from the distillation of petroleum; b) oxidation of methanol to formaldehyde (air or enriched air); c) oxidation of ethylene to ethylene oxide (air or oxygen, the latter particularly in new plants); d) oxychlorination of ethylene to 1,2-dichloroethane (air or oxygen, the latter particularly in new plants); e) acetoxylation of ethylene to vinyl acetate (oxygen); f ) oxidation of n-butane to acetic acid (air or oxygen); g) oxidation of ethylene to acetaldehyde (air or oxygen); h) oxidation of acetaldehyde to acetic anhydride (air or oxygen); and i) ammonia oxidation of propylene to acrylonitrile (oxygen enriched air).

Table 3. Composition of the emissions from the process of oxychlorination of ethylene where air and oxygen

are used (Arpentinier et al., 2001). DCE: 1,2-dichloroethane; VCM: vinyl chloride monomer
Process using air Component Content (vol%); flow (m3/h) O2+Ar Ethylene COx (CO2/CO=3-4/1) DCE and chlorinated compounds N2 Waste (m3/h)* *Approximately 300-900 m3 per t of VCM produced 4-8; 400-2,400 0.1-0.8; 10-24 1-3; 100-900 0.1-2.5; 2-5; 15-30; 0.5-1; 25 50 300 10 Process using O2

0.02-0.2; 2-60 remainder 10,000-30,000

remainder 1,000

VOLUME II / REFINING AND PETROCHEMICALS

623

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

Selective oxidation catalysts for hydrocarbons


Characteristics of oxidation catalysts

Oxidation catalysts belong to a wider class of materials having redox or oxidoreductive type characteristics; systems which catalyse reactions of hydrogenation, dehydrogenation, halogenation and dehalogenation also belong to this class. The most important catalysts in the field of petrochemicals for the oxidation of hydrocarbons for processes carried out in the gas phase are listed in Table 1. In addition to these, it is worth mentioning catalysts used for the oxidation of inorganic compounds, such as those employed for the oxidation of SO2 to SO3 (based on supported vanadium oxide), of ammonia to NO (based on Pt/Rh) and of hydrogen chloride to molecular chlorine (based on supported copper chloride). Below is a list of the principal characteristics of oxidation catalysts for gas phase reactions. Presence of a transition metal as the principal active component (V Mo, Cu, Fe, Pd, Pt, Rh, Ag). Often in , these cases, a second element is also present which can be transition or post transition (for example, P, Sb or Bi), which contributes to establishing the reactive characteristics of the catalyst. This effect can be explained by the formation of a mixed oxide (that is, of a specific compound, such as for example Bi2Mo2O9, possibly only on the surface of another oxide, of a solid solution or of an oxide doped with the other element), with reactive characteristics different from those of the single elements, if present in distinct phases. In some cases the element is initially present in a metallic form, but under reaction conditions it can generate the corresponding oxide (or chlorides or oxychlorides). Presence of small quantities of promoter (or doping) elements. The purpose of these elements is to optimize the performance of the principal active elements. The nature of the promoters can vary and they can therefore play different roles in the transformation of the reagents. The active elements and the promoter elements, constitute the active phase, that is the phase directly involved in the transformation of the reagents into products. Presence of a support (usually silica, alumina or titanium oxide). This support in the catalysts formulation can fulfil a variety of tasks. A primary task is that of dispersing the active elements, conferring a larger surface area to the active phase compared with what would have existed in the absence of the support. It is clear, therefore, that the support must have surface area characteristics suitable for the reaction of interest. In selective oxidation, where the selectivity in the formation of the partially oxidized product is heavily dependent on the subsequent reactions to undesired products (for example, carbon oxides, which are

thermodynamically favoured), a support is needed with a surface area which is not too large. In this way the rate of the undesired secondary reactions, which are also dependent on the time needed by the product to diffuse from the active centre into the gas phase, is limited. A further task of the support is that of providing the resistance of the active phase to phenomena which can cause abrasion or disintegration, especially for those applications which involve particular mechanical stresses on the catalyst (for example, in fluidized-bed reactors), in addition to avoiding powdering during the loading of the catalyst into packed fixed-bed reactors. Finally, in some cases the support serves to alter the characteristics of the intrinsic chemical reactivity of the active phase, through the effects of the interaction between the latter and the support itself. This comes about when the support presents functional groups on its surface which can lead to the formation of chemical bonds with the elements of the active phase, or it takes place as a result of particular crystallographic similarities between the surface and the support. These interactive effects can be positive for the reactivity of the catalyst itself, altering its oxidoreductive characteristics or reducing its volatility; the undesirable effects of loss by sublimation of components of the active phase are thus reduced. The most suitable combination of the type and number of active phases (including the promoters) and the type of support is dependent on the characteristics of the reaction and the type of reactors used. In particular, below are listed the factors that have the greatest influence on the formulation and the morphology of the catalyst used for oxidation reactions. Type of chemical transformation involved and mechanism through which it takes place. With increasing complexity of the transformation the composition of the catalyst also becomes the more complex, in terms of the number of elements making up the active phase, or of the structural complexity (the formation of crystalline phases having multi-functional characteristics). For example, catalysts used for oxidation or allylic ammonia oxidation always contain Mo as the principal element for the active phase, while catalysts for the synthesis of anhydrides or of acids almost always contain V . Optimization of the redox characteristics or of the acidity or basicity properties of the catalyst. The promoters (or doping agents) can play a fundamental role in the control of these properties. Promoters with base-type characteristics (alkaline or alkaline earth metal oxides) can reduce the surface acidity of the active phase, with a consequent improvement of selectivity through the suppression of the acid-catalysed reactions (cracking, formation of oligomers of unsaturated compounds). Promoters with acid-type characteristics can reduce the interaction between the active phase and

624

ENCYCLOPAEDIA OF HYDROCARBONS

OXIDATION PROCESSES

intermediates of the reaction which have acid-type characteristics, thus favouring their desorption into gas phase and limiting the contribution of the subsequent undesired reactions. Other promoters can optimize the oxidoreductive properties of the active phase, by a modification of the overall electronic properties of the solid. Reaction scheme. The presence of consecutive reactions (typically, combustion reactions of the desired product, or reactions which lead from the reagent to the desired product through the formation of intermediate products with an increasing state of oxidation) involves the use of a catalyst with characteristics such as to limit (or, alternatively, to favour) the contribution of these reactions. This can be achieved not only by control of the intrinsic activity of the catalyst, but also by a modification of the porosity of the active phase (and therefore of the support, if present). High surface area and porosity values entail effective intra-particle residential times which are much higher that those calculable from the feeding capacity of the reactor, and therefore a significant contribution from the consecutive reactions for a given conversion of the reagent. This can have a considerable influence on the selectivity of the desired product. Reaction heat levels. Highly exothermic reactions involve the need both for fluidized-bed catalytic reactors, which are more efficient in removing heat than multitubular reactors, and for catalysts capable of operating in conditions of high mechanical stress. In these cases fluidizable supports are used, which feature particles with an average diameter of between 50 and 150 mm, resistant to abrasion and with an appropriate density. For medium-low heat levels, tubular or tube-bundle (multi-tubular) reactors can be used. In these cases the catalysts have a characteristic morphology for these applications and are produced in the form of extrusions (or pellets). When possible, supports with a high heat conduction capacity are used, such as SiC, in order to assist the dissipation of the heat from the reaction. Spatial rates in the reactor. High spatial rates in packed catalytic beds can lead to a high loss of pressure, and therefore to the need for heavy compression of the flow upstream of the reactor. It is possible to minimize the loss of pressure by increasing the vacuum level in the catalytic bed, through the use of special structures of the catalyst particles. This can be instrumental in conditioning the performance of the process, as in the case of oxidative dehydrogenation of methanol to formaldehyde. In this case, the use of cylindrical pellets with an axial hole enables the loss of pressure to be reduced and therefore, the linear rate in the reactor to be increased for a given rate of feeding. This involves shorter contact times, better reaction temperature control and reduced catalyst deactivation effects.

Oxidation catalyst mechanisms in gas phase

The principal catalytic oxidation mechanisms are listed in Table 2. The redox type mechanism is the one that is used in the majority of oxidation reactions. It works through a series of successive stages, which include: the adsorption of the reagent (the substrate to be oxidized) on the active centre; the transfer of electrons from the reagent to the active centre and the simultaneous transfer of oxygen ions from this to the reagent (the oxygen is incorporated into the substrate, or alternatively returns in the formation of co-produced water); and finally, the desorption of the product. The same sequence of stages involves the molecule of oxygen for the catalyst reoxidation stage: co-ordination at the metallic centre; transfer of electrons (up to 4 for each oxygen molecule); dissociation of the molecule into two atomic species in ionic form; finally incorporation of the oxygen in its ionic form within the active phase. One or more active centres may be involved for each reagent molecule, depending on the following factors: a) the overall number of electrons transferred and therefore of oxygen ions involved in the oxidoreductive process; b) the ionic and electronic conduction capacity of the solid, and therefore of the surface active phase under reaction conditions; c) the level of cover of the active phase by the adsorbed molecules (reagents and products); d) the surface mobility of the reaction intermediates; and e) the number of active centres close to the one in which the activation of the hydrocarbon took place. On the basis of the redox model, the selectivity of the process, that is the relationship between the quantity of the product formed and the total quantity of reagent transformed, can be traced back to two different situations. First and foremost the selectivity depends on the nature of the oxygen ions present as species adsorbed on the active phase and on the interaction between them and the reagent or the reaction intermediates. As previously stated, the O2 species, incorporated in the lattice of the oxide, is considered to be the selective species, while the O2 and O species have electrophilic characteristics and are considered to be non-selective species. Since the formation of the first species comes about by the intermediate formation of the electrophilic species, it is clear that the transformation rate of each of them and their reactivity with the reaction intermediates obtained through activation of the substrate determine the selectivity of the process (Bielanski and Haber, 1991). Moreover, the selectivity of the oxidation process is traceable to the concentration of O2 species incorporated in the metal oxide lattice, and therefore directly to the average state of oxidation of the catalyst (Grasselli, 2002). A strongly oxidized catalyst has a high density of active centres capable of receiving electrons

VOLUME II / REFINING AND PETROCHEMICALS

625

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

from the substrate and of releasing O2 ions, and therefore it is able to transform that substrate into molecules with a high state of oxidation (for example, into combustion products). In contrast, a catalyst made up of a partially reduced oxide has a modest oxidizing capacity, and therefore is potentially more selective for the partially oxidized products. According to the redox model, the state of oxidation of a metal oxide in a stationary state is dependent on the conditions of the reaction; this implies that the selectivity in its turn is dependent on the operating parameters, such as the composition of the feed (that is the relationship between the substrate to be oxidized and the oxidizing agent) or the reaction temperature. Both models have been experimentally verified for different oxidation reactions, and still remain valid today for the explanation of the selectivity of oxidation processes involving reactions with redox type mechanisms.
Principal industrial processes and relevant applications
Oxidative dehydrogenation of methanol to formaldehyde

Formaldehyde (HCHO) is among the top twenty chemical compounds produced on a world scale, and is used in the synthesis of various resins (ureaformaldehyde, phenol-formaldehyde, and polyacetals) which find applications in the construction, automotive, textile and paper sectors. Methanol can be converted into formaldehyde both by direct oxidative dehydrogenation: CH3OH 0.5O2 HCHO H2O H= 155 kJ/mol

or by dehydrogenation combined with oxidation of the H2 product: CH3OH H2 0.5O2 HCHO H2 H2O H= 84 kJ/mol H= 238 kJ/mol

The two processes differ in their operating conditions and type of catalysts. In the first process low concentrations of methanol are used in the feed, in order to avoid the formation of explosive mixtures and to control the temperature of the reaction. Commercial catalysts are based on iron molybdate, but also contain an excess of molybdenum (Fe2(MoO4)3 MoO3), since the presence of molybdenum oxide is a necessary condition for high selectivity. Typically a ratio of Mo/Fe within the range of 1.5-3.0 is used; occasionally oxides of Co and Cr are added as promoters. The excess of molybdenum is also necessary because the sublimation of the oxide (particularly at the points of greatest overheating) cause the progressive depletion of the Mo in the catalyst and the condensation of MoO3 in the

coldest parts of the reactor. This induces, not just the deactivation of the catalyst, but also a progressive increase in pressure loss. Reaction temperatures are typically within the range of 310-340C, with conversions in excess of 98% and selectivity equal to 92-95%. Multi-tubular fixed-bed reactors are generally used. A recent development has seen the introduction of a final (post-reactor) adiabatic stage. In dehydrogenation combined with partial combustion of H2 (the overall process turns out to be partially exothermic) an understoichiometric oxygen current is fed in, to operate in the upper region at the limit of flammability. Due to the thermodynamic limits of dehydrogenation, it is necessary to operate at higher reaction temperatures than those for oxidative dehydrogenation. In this process use is made of Ag based catalysts supported on alumina with a low surface area, typically in spherical form with a diameter of 1-5 mm. If operating at temperatures in excess of 600C (particularly 680-720C), it is possible to obtain an almost total conversion of the methanol, while at lower temperatures (500-550C) the conversion is less efficient (65-75%) and it is necessary to recycle the methanol which failed to react. Moreover, it is necessary to use short contact times in order to avoid decomposition of the formaldehyde. Selectivity to formaldehyde of 98-99% is obtained, with the formation of the following by-products: dimethylether ((CH3)2O), whose formation is due to the presence of acidic sites in the catalyst; methyl formate (HCOOCH3) obtained through disproportionation of the formaldehyde on basic sites; carbon oxides, derived from both parallel and serial reactions. To limit the formation of carbon oxides, rapid cooling of the reaction products is necessary when they leave the catalyst bed. High selectivity is achieved through the optimization of the acid-base properties of the catalyst, limitation of the oxidation of the formaldehyde to formic acid (a product which decomposes easily) and control of the redox properties of the catalyst. Fig. 5 illustrates the reaction mechanism in the case of direct oxidation of methanol on oxide based catalysts. The methoxy species is the first chemisorbed species that is formed by the contact of methanol with the catalyst; its subsequent transformation depends both on the reaction conditions and on the properties of the catalyst. If the concentration of methanol is high and the rate of the subsequent oxidation of the methoxy species is low, a condensation reaction takes place that leads to dimethylether (typical of the acidic oxides containing non-reducible cations, such as alumina). Oxidation of the methoxy species (extraction of an atom of H and transfer of an electron) leads to co-ordinated formaldehyde, which is in equilibrium with the

626

ENCYCLOPAEDIA OF HYDROCARBONS

OXIDATION PROCESSES

CH3OCH3 H2O CH3OH CH2 OH O CH3OH H O CH2 O CH2O formaldehyde OH

dimethoxymethane OCH3 CH2 OCH3 CH2 2CH3OH O OH O methylformate O formic acid HCOOH H COCH3

[H]

metoxy

[H]
disproportionation

dioxymethylene H2O

H2O CH COx O O oxygen vacancy CH3OH

formate

Fig. 5. Reaction mechanism of the oxidation of methanol on oxide based catalysts (Centi et al., 2002).

dioxymethylene species. The reaction requires a nucleophilic attack by the catalysts structural oxygen. If the Me O bond is of a covalent type, the equilibrium between the dioxymethylene and the co-ordinated formaldehyde is shifted towards the latter species, which in its turn is in equilibrium with the gas phase formaldehyde. Ionic oxides, instead, favour the formation of the dioxymethylene species. This gives rise to the methoxy and formate species by Cannizzaro disproportionation; the formate can also form through direct oxidative dehydrogenation from the dioxymethylene. By reactions with methanol, dioxymethylene forms dimethoxymethane, which can desorb into gas phase. The formate species can also react with methanol to give methyl formate. The formation of this product requires that the desorption rate of the formaldehyde be low and that the oxidation/disproportionation rate be relatively high. Both these reaction pathways contribute to the formation of the formate species. Their relative rates depend on the degree of surface coverage of the dioxymethylene species and on the reaction conditions. Nevertheless, the formate species can also be converted and therefore the selective formation of methyl formate requires that the conversion rate of the formate species be low and that the concentration of methanol be relatively high. A simplified outline of the commercial processes for converting methanol to formaldehyde using iron molybdate and supported Ag as the catalysts is shown in Fig. 6. Oxidative dehydrogenation of methanol on iron molybdate based catalysts is carried out in a cooled multi-tubular reactor (see again Fig. 6 A). Due to the progressive deactivation of the catalyst (the lifespan is

from 1-2 years), it is necessary to gradually increase the temperature of the reactor in order to keep the productivity at a constant level. The formaldehyde yield is about 95-96%. The oldest processes operate with a feed of 6% of methanol in air (a concentration which is less than the lower limit of flammability), but in this case productivity is low, the purity of the formaldehyde is poor because of the formation of formic acid, the catalysts lifespan is limited and it is necessary to operate with high volumes of inert gas. For this reason, about half of the plants have been converted to a feed with reduced levels of oxygen ( 10%), and higher concentrations of methanol (8.2%). On account of the high amount of heat generated, it is necessary to dilute the catalyst and improve the efficiency of the reactors cooling system. Some plants also use steam as a diluent. When the gas leaves the reactor vessel, it undergoes a heat recovery stage, after which it is sent to an absorber column, where water is used as a solvent. The formaldehyde solution, from the bottom of the absorber column, has a concentration of from 50-60%; it is then sent to an ionic exchange column for removal of the formic acid. On the other hand, the gas from the top of the column is recycled; the waste stream keeps the composition in the reactor constant. The process using supported Ag as the catalyst (see again Fig. 6 B) has the advantage, in comparison with the direct oxidation method, of producing a waste stream which can be sent directly for incineration, because it contains H2, methanol and formaldehyde, as well as small quantities of N2 and CO2. The combustion of this flow produces the majority of the steam used in the process. The outline of the process is similar to that discussed previously, however a purer methanol feed is

VOLUME II / REFINING AND PETROCHEMICALS

627

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

recycle

vent H2 O cooling water

using chlorohydrin, which is no longer used, and direct oxidation of ethylene with oxygen over supported Ag based catalysts CH2 CH2 0.5O2 H2C O H CH2 105 kJ/mol

air

methanol
A A

formaldehyde solution

recycle catalyst steam

vent

H2O methanol air H2 O waste


B B

formaldehyde solution

Fig. 6. Simplified outline of the process of the synthesis of

formaldehyde from methanol: A, direct oxidative dehydrogenation with catalysts based on iron molybdate; B, a process using catalysts based on supported Ag (Arpentinier et al., 2001).

required (there must be no iron carbonyls or sulphur compounds which would contaminate the catalyst), and a heat exchanger is required to heat the methanol. Furthermore, the reactor must operate with very short contact times and have a rapid cooling system (cooling times for the discharge flow from the catalyst bed of less than 0.02 s), in order to avoid consecutive reactions of the formaldehyde, and finally two absorber columns in series should be used.
Epoxidation of ethylene to ethylene oxide

Ethylene oxide is an intermediate product that is synthesized on a large scale, being used in the production of ethylene glycol and polyglycols, of ethanolamine and non-ionic detergents, and of esters of ethylene glycol. There are two technologies: a process

CsCl and BaCl2 are used as promoters (the concentration of these doping agents is of 0-10 ppm), since both the alkaline metals and the chloride ions promote the selectivity. In addition to CO2, the by-products are acetaldehyde and formaldehyde, but in concentrations no greater than 0.1%. Although there are still conflicting opinions on the matter, a significant body of data indicates that the active catalysing species is made up of Ag O, with electrophilic characteristics, while oxygen bridging atoms between the silver atoms (AgOAg) have a nucleophilic character and are non-selective, contrary to what happens in the case of selective oxidation on oxides (see again Fig. 3 B). In the past it was held that the selective species in epoxidation consisted of AgO2, which after the insertion of oxygen into the ethylene left an adsorbed O atom and AgO on the surface, which in turn was responsible for the non-selective oxidation of ethylene to CO2 and H2O. On the basis of this model, a maximum selectivity of 85.7% was hypothesized, whereas current commercial processes operate with greater selectivity (at low conversion), in the range of 88-94%. The reaction scheme is of a triangular type, with two parallel reactions of the transformation of ethylene into ethylene oxide and CO2, and a consecutive reaction of ethylene oxide to CO2. The principal factors influencing the reaction kinetics and hence the choice of the optimal reaction conditions are summarized here. The ethylene oxide formation rate increases as the partial pressure of the oxygen increases, while a maximum rate is reached with respect to the concentration of ethylene, as a result of the competition between the ethylene and the ethylene oxide for the same catalytic sites. The ethylene oxide formation rate decreases on increasing the concentration of the chloride ions used as a doping agent (nevertheless, above a certain value traces of chlorinated compounds such as vinyl chloride and dichloroethylene form as by-products) and the concentration of CO2. From a practical point of view, the concentration of ethylene is determined essentially by the limits of flammability. The ratio of the rate of the two parallel reactions of formation of ethylene oxide and combustion, that is the initial selectivity, rises with the concentration of chloride ions, with the content of alkaline metal promoters and with the partial pressure of the ethylene. The processes employed for oxidation of ethylene operate with air or with pure oxygen. The former are still

628

ENCYCLOPAEDIA OF HYDROCARBONS

OXIDATION PROCESSES

Fig. 7. Simplified outline

recycle

of the process of the synthesis of ethylene oxide from ethylene: A, processing with air; B, processing with pure oxygen (Arpentinier et al., 2001).

vent

ethylene oxide

light ends

ethylene air air

heavy ends

steam

recycle

CO2 vent

purge

inert

steam ethylene oxide

light ends ethylene

O2

steam

B
heavy ends

very widespread, but new plants mainly use pure oxygen, which offers advantages in terms of higher yields and productivity, greater selectivity (due to the greater partial pressure of the ethylene in the reactor), lower volumes of waste gas, the ability to choose the

type of diluent gas and lower reactor and equipment costs. On the other hand, the process using pure oxygen incurs greater costs due to its use and the need to separate the CO2 produced. The simplified outline of the two processes is illustrated in Fig. 7. In the process using

VOLUME II / REFINING AND PETROCHEMICALS

629

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

air, the ethylene is oxidized in a series of tubular reactors (two, or three in the bigger plants), in order to increase the conversion and manage the heat from the reaction more effectively (in the first reactor the conversion is around 40%). The outflow from the reactor is cooled and then sent to an absorber column with water, for the recovery of the ethylene oxide. Part of the gas discharged from the absorber column is recycled, while another part is sent to a second reactor where the conversion reaches 80% (95% where there are three reactors in series). Selectivity is about 70% in the first reactor, but is lower in the subsequent ones. The process with pure oxygen (purity 97%) is single stage and uses multi-tubular reactors. The ethylene/O2 ratio is typically 3.0-3.5, while the O2 concentration is kept below 9% to avoid the formation of inflammable mixtures. Conversion through the passage of ethylene is in the range 10-15%, while overall it is greater than 97%. Selectivity typically is greater than 80%. After cooling, the gases leaving the reactor are sent to the ethylene oxide absorber column, while the gases coming out of the column are compressed and recycled. Part of the gas is sent to a column to eliminate CO2 through hot absorption in an aqueous solution of potassium carbonate. A small fraction of the gases (less than 1%) is discharged, to prevent an accumulation of inert gases. In some processes a diluent, such as methane or ethane, is also used. Although CO2 itself can be an effective diluent, it can contaminate the catalyst and therefore it is necessary to keep its concentration level low.
Oxychlorination of ethylene

The oxychlorination (oxidation in the presence of HC1) reaction of ethylene to 1,2-dichloroethane (DCE) CH2 CH2 2HCl 0.5O2 CH2Cl CH2Cl H= 238 kJ/mol

is the basis for the production of the vinyl chloride monomer (CH2 CHCl), used in the production of homo- and copolymers in PVC. Three different processing options are possible, depending on the technology used and the operating conditions: the reactors can be based on fixed or fluidized beds; either air or pure O2 can be used; and a stoichiometric value or an excess of ethylene may be employed. Generally, pure oxygen is used when operating in a fixed-bed reactor, with a large excess of ethylene (compared with the stoichiometric value) where the unconverted ethylene is recycled, or when using stoichiometric ratios, where high conversion levels of ethylene are achieved, but the gas flow is in any case recycled so as to reduce the environmental impact of the process (see again Table 3).

The use of an excess of ethylene, which has better thermal conduction properties than nitrogen, offers advantages such as a more uniform temperature profile in the reactor (which results in better selectivity due to less combustion and reduced formation of secondary products such as trichloroethane), better conversion of the HC1 and a longer catalyst life (thanks to a lower sublimation of the active phase and a reduced formation of charcoal on the catalysts surface). Since the productivity of fixed-bed reactor is limited by the capacity for transferring the heat to the cooling fluid, a gas phase with higher conductivity results in an increase in productivity. The principal problem related to the use of O2 is represented by its higher cost compared with air and the need for more complex systems for operating under safe conditions. Nevertheless, many plants using air have been satisfactorily modified to operate with oxygen. The most widely used reactor is the multi-tubular fixed bed, because of the higher productivity attainable; nevertheless, the higher investment needed (corrosion resistant steel has to be used) leads to a preference for fluidized-bed technology in the construction of new plants. In Fig. 8 a simplified outline is portrayed of an air-fed fixed-bed process. Three multi-tubular reactors in series are used; the catalyst is diluted with graphite, in order to reduce the thermal gradients in the reactor; and a fourth reactor is used for direct chlorination of the residual ethylene with Cl2. The gases are cooled to condense the DCE (1,2-dichloroethane), which is then sent to the purification columns: the first removes the water, the second the light products (ethylene chloride, vinyl chloride, 1,1-dichloroethane and dichloroethylene) and the third the heavy products (trichloroethane, perchloroethane and perchloroethylene). To optimize the local ethylene/O2 ratio and to obtain better control over the temperature in the reactors and hence of the selectivity, ethylene and HC1 are fed into the first reactor, while the air or oxygen feed is distributed over the three reactors in series. This also makes it possible to operate outside the limits of flammability, to reduce the formation of CO2, to have a more homogeneous temperature profile and finally to increase the lifespan of the catalyst. The oxygen and ethylene are fed at slightly higher levels with respect to the stoichiometric quantities, in order to achieve a conversion of the HC1 greater than 99.5%. On the other hand, in processes that operate with oxygen and a large excess of ethylene, the unconverted hydrocarbon is separated and recycled. The catalysts are based on copper chloride supported on alumina and the promoters, made up of alkaline or alkaline-earth metals, play a fundamental role. The reaction temperatures are between 220 and 250C, with

630

ENCYCLOPAEDIA OF HYDROCARBONS

OXIDATION PROCESSES

Cl2

ethylene HCl

NaOH H2O

air water light ends (to purification) vent

1,2-dichloroethane

heavy ends
Fig. 8. Simplified outline of the process of the oxychlorination of ethylene in a fixed-bed reactor (air based process)

(Arpentinier et al., 2001).

pressures up to 5 bar. Selectivity of the DCE product is between 93 and 97%.


Acetoxylation of ethylene

Acetoxylation (oxidation in the presence of acetic acid) in the gas phase of ethylene with acetic acid and O2 is the primary process for producing vinyl acetate (CH3C(O)OCH CH2). Pd supported on silica is used, promoted by gold and alkaline metals (potassium acetate). The gold allows the reduction of the secondary reaction that forms ethyl acetate. Processing takes place with pure O2 at a temperature of around 150C, with pressures in the range of 8-10 bar. In these conditions the reaction takes place in a liquid film, formed through capillary condensation in the pores of the catalyst. The reaction mechanism involves the reduction of the Pd acetate to metallic Pd through reaction with ethylene and reoxidation of the Pd0 by the oxygen. Selectivity levels can reach 98%, although in industrial processes they are typically between 92 and 95%. Conversion per pass is around 10%. After pre-heating, the cool ethylene, mixed with the recycled ethylene and with acetic acid, is sent to the reactor containing the catalyst, previously mixed with O2. The typical composition is as follows: 40-55% ethylene,

10-20% acetic acid and 7-8% O2, as well as inert ingredients (CO2, ethane, Ar, N2 and H2O). After the reaction, the gases are cooled causing condensation of the acetic acid, the water and the majority of the vinyl acetate. The liquid stream is split through azeotropic distillation to recover the acetic acid and the vinyl acetate. Nevertheless additional purification columns are necessary to achieve the required purity of the vinyl acetate, in particular to lower the concentration of ethyl acetate below 150 ppm.The gas phase of the first condensation column is sent to an absorber column into which acetic acid is fed to remove the vinyl acetate, and then to a second column fed with a solution of NaOH to remove the CO2. The gases are then recycled to the main reactor.
Oxidation of propylene to acrolein and oxidation of acrolein to acrylic acid

Oxidation of propylene to acrylic acid (used in the production of acrylic esters) is achieved through the intermediate formation of acrolein: C3H6 O2 CH2 CH2 CHCHO H2O H= 339 kJ/mol

CHCHO 0.5O2 CH2 CHCOOH H2O H= 255 kJ/mol

VOLUME II / REFINING AND PETROCHEMICALS

631

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

By-products of the reaction are carbon oxides, acetic acid, propionic acid, formaldehyde, maleic acid, acetaldehyde and acetone. For the first reaction multi-component catalysts based on bismuth molybdates are used (for example, Mo12BiFe3Co4.5Ni2.5Sn0.5K0.1Ox), while for the second reaction the catalysts are based on molybdenum and vanadium oxides (for example, Mo12V3Cu2.5Fe1.25Mn0.1Mg0.1P0.1Ox). It is possible to carry out the synthesis in a single stage but, because of the strong exothermicity of the reaction the lifespan of the catalyst is reduced. Moreover, the overall selectivity is greater in a two stage process, as it is possible to optimize each independently. In the first stage the selectivity of acrolein is typically greater than 85% and the conversion of propylene exceeds 90%, while in the second stage the selectivity of acrylic acid is greater than 95%, with yields of between 90 and 96%. Multi-tubular fixed-bed reactors are used in series. The first reactor operates with a temperature in the range of 330-400C, and a spatial rate of 1,300-2,600 h 1 (the pressure is 2-2.5 bar), while the second operates at lower temperatures (250-300C), higher spatial rates (1,8003,600 h 1) and lower pressures (due primarily to the loss of pressure in the first reactor). The concentration of propylene entering the first reactor is 5-8% in air. Recirculating gas and/or steam are used as diluents, in order to operate outside the limits of explosion. The use of steam also allows the reactions in homogeneous phase to be reduced, the thermal transfer to be improved and selectivity to be increased, favouring the desorption of acrolein and acrylic acid. Nevertheless, excessive concentrations of steam lower the concentration of the acrylic acid solution. The gases leaving the reactor, after cooling and heat recovery, are sent to an absorber column with water. An inhibitor is added to avoid polymerization of the acrylic acid. The discharge gases are then sent for incineration and in part are recycled, after the elimination of condensable compounds. The solution of acrylic acid is sent to the purification section, which consists of a series of azeotropic distillation columns (with methylethylketone as the third component). In the case of diluted solutions it is possible, as an alternative, to carry out a separation by extraction, using ethyl acetate or aromatic compounds as solvents. The various industrial processes differ in the composition of the catalyst and in the separation section.
Synthesis of methyl methacrylate

presence of methanol) is the most widely used for the synthesis of methyl methacrylate and has many disadvantages, linked to the toxicity of HCN and the co-formation of high quantities of ammonium sulphate, which is generated in the ratio of 2:1 with methyl methacrylate. The alternative process is direct oxidation of isobutene in the gas phase (with subsequent or integrated stages of esterification); however, this process gives yields and selectivity which are too low to be competitive. Alternative methods of synthesis are: a) direct oxidation of isobutane (a process which is still in the research phase), which has the advantage of lower raw material costs and lower environmental impact; b) oxidation of isobutyric aldehyde to isobutyric acid, which is then converted into methacrylic acid by oxidative dehydrogenation (Mitsubishi Kasei/Asahi method); c) oxidation of tert-butyl alcohol to methacrolein, followed by oxidation to methacrylic acid and esterification; and d) hydroformylation of ethylene to propionic aldehyde, which is then condensed with formaldehyde to give methacrolein, which finally is oxidized to methacrylic acid and esterified (BASF process). Among these alternatives, direct synthesis of isobutane is the most interesting; nevertheless, the selectivity obtained and the stability of the catalysts are not sufficient for it to be developed industrially. The average composition of the catalysts for the last of the above applications is as follows: (HmY0.2-1.5) (P1-1.2Mo12-nX0.4-1.5Ox), where Y is the ion of an alkaline metal and X is an element such as V As and Cu; , moreover, various other additives are present. It is necessary to use high concentrations of isobutane and steam (up to 65%) to obtain good selectivity and stability of the catalysts.
Synthesis of acrylonitrile by ammonia oxidation of propylene

Acrylonitrile is produced on a large scale (over 5 million tons per year) by the process of catalysed ammonia oxidation (oxidation in the presence of ammonia) of gas phase propylene: CH2 CH CH3 NH3 1,5O2 CH2 CH CN +3H2O H= 515 kJ/mol Acrylonitrile finds applications in the synthesis of various homo- and copolymers used as fibres, resins and elastomers; it is also an intermediate in the production of adiponitrile and acrylamide. By-products of the reaction are HCN, acetonitrile, N2 (from combustion of the ammonia) and carbon oxides. The reaction is strongly exothermic and to control the temperature most plants use fluid-bed reactors. The commercial catalysts are multi-component, based on bismuth molybdate and supported on silica (for example,

Methyl ester of methacrylic acid, CH2 C(CH3)COOCH3, is used in the production of vinyl polymers. The acetone cyanohydrin process (which consists of a reaction between acetone and HCN, followed by a reaction of acetone cyanohydrin with sulphuric acid and a final hydrolysis of the adduct in the

632

ENCYCLOPAEDIA OF HYDROCARBONS

OXIDATION PROCESSES

(K, Cs)0.1(Ni, Mg, Mn)7.5(Fe, Cr)2.3Bi0.5Mo12Ox/SiO2). It is not necessary to recycle the propylene, since conversion rates are as high as 95%, while maintaining a high selectivity to acrylonitrile (above 80%). The fluidbed reactor contains a high quantity of catalyst, up to 7080 tons, in the form of spherical particles with an average diameter of 40-50 mm, with a view to allowing an efficient fluidization. The purity of the feed must be very high ( 90% for the propylene and 99.5% for the NH3). The ratio of the ammonia/propylene feed is equal to 1.05-1.2, and the ratio of O2/propylene is within the range 10-15. The reaction temperature is between 420 and 450C and the pressure between 1.5 and 3 bar. Pressures above 1 bar have a negative effect on the selectivity to acrylonitrile, but are necessary in order to obtain good fluidization and to increase the productivity. The reagents are fed into the reactor separately, to minimize reactions in the homogeneous phase and prevent the formation of explosive mixtures before reaching the catalytic bed; the composition of the mixture in the reactor is within the limits of flammability, but the presence of the fluid bed inhibits the propagation of radical reactions thereby blocking any flame front. The fluid-bed reactor contains various coils and systems to minimize the formation of slugs and to reduce the phenomena of retro-mixing of the fluid. The top of the reactor has a larger cross section, in order to reduce the velocity of the gas and to diminish the occurrences of pneumatic transmission and elution of the catalyst. Appropriate cyclones allow the recovery of the catalyst particles and their reintroduction into the reactor. The effluents from the reactor are sent to an absorber column with water, while the unconverted ammonia is neutralized with sulphuric acid. The gases leaving this column, containing N2, carbon oxides and small quantities of propylene, are sent for incineration. The acetonitrile/acrylonitrile mixture forms an azeotrope with water that on separation gives rise to an aqueous phase (recycled to the absorber) and an organic phase, rich in acrylonitrile and HCN, that is sent for purification. The aqueous solution of acetonitrile is
Fig. 9. Outline of the synthesis of

concentrated for azeotropic distillation. The acrylonitrile is purified in two columns in series, to recover the hydrogen cyanide and the impurities (acetone, acetaldehyde, propionaldehyde and acrolein). A final stage of vacuum distillation is used to obtain acrylonitrile with purity above 99.4%. Recently a number of studies have been devoted to the use of propane as an alternative reagent to propylene, by virtue of its lower cost. New classes of catalysts (for example those developed by Mitsubishi, composed of quaternary oxides of molybdenum, vanadium, tellurium and niobium, MoV0.3Te0.23Nb0.12Ox) give yields of acrylonitrile above 50%. This value could be sufficient to justify development of a new process starting with propane, although a further increase in yield and stability of the catalyst would be hoped for.
Ammonia oxidation of alkyl aromatics

Numerous aromatic nitriles such as benzonitrile, phthalonitrile, isophthalonitrile, terephthalonitrile and nicotinonitrile find applications in the synthesis of products for fine chemicals. For example, nicotinonitrile can be hydrolysed to the corresponding amine or to nicotinic acid, used for the synthesis of vitamin B. Isophthalonitrile is used in the synthesis of herbicides and fungicides. Phthalonitrile is an intermediate for pigments based on phthalocyanine. A number of catalysts are active in the reaction: vanadium oxide supported on TiO2 (preferably in the anatase crystalline structure), doped with Cs, P and W; multi-component catalysts based on molybdates; vanadium antimonates doped with Bi and Fe; and supported heteropolyacids (PV3Mo12Ox on silica). Apart from maximizing the selectivity to nitrile, it is also important to minimize the oxidizing reaction of the ammonia to N2. Selectivity levels above 90% are generally possible for conversions which are between 50 and 80%, even though the results vary considerably according to the type of substrate. One of the principal problems in the industrial application of the process is the need to carry out successive production runs with different types of alkyl aromatics, as the market demand for these products is

phthalimide by catalytic ammonia oxidation of o-xylene in the gas phase.

O phthalimide NH O COx CN CN

CH3

O2, NH3

CH3 O2, NH3 CN V2Ox /TiO2 tolunitrile phthalonitrile

CH3 V2Ox /TiO2

VOLUME II / REFINING AND PETROCHEMICALS

633

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

insufficient to justify the use of the plant for a single reaction. The aromatic imides can also be obtained by means of catalytic ammonia oxidation (Fig. 9), for example feeding o-xylene and using a catalyst based on supported vanadium oxide on titanium oxide.
Maleic anhydride from n-butane

Maleic anhydride is used as an additive in the synthesis of various polymers, in the synthesis of chemical products for agriculture and of malic and fumaric acids, as well as being an intermediate for the synthesis of g-butyrolactone and of tetrahydrofuran. Today, most plants for production of maleic anhydride use n-butane as the feed; this process has replaced that of benzene, by virtue of the smaller number of by-products, the better atomic efficiency and the lower costs, and the non-toxicity of the reagent. The by-products obtained in the oxidation of n-butane are carbon oxides and acetic acid, whereas numerous by-products are formed by benzene, or when using butenes or butadiene as the raw material. The various commercial processes differ in their reactor technology (fixed-bed, or fluid or circulating-bed reactor), in the percentage of n-butane in the input (less than 2% with fixed-bed reactors, between 2.6 and 5% for processes with fluid-bed reactors, and more than 10% in the DuPont process, with circulating-bed reactors, similar to that used in catalytic cracking) and in the method of recovery and purification of the maleic anhydride (use of organic or aqueous solvents in the recovery and purification method). In all processes the catalyst consists of pyrophosphate of vanadyl (VO)2P2O7, where necessary promoted with doping elements, although the methods of preparation, activation and formation and the dimensions of the particles can vary. The feed composition is directly related to the choice of reactor as shown above. An increase in the concentration of n-butane increases the productivity, but calls for special precautions for dispersal of the reaction heat and to reduce the risks of explosion. The unconverted n-butane is not recycled, but is used in the production of high temperature steam, since the value of this hydrocarbon is close to that of fuels. The conversion rate of n-butane is between 80 and 90% and the yield in maleic anhydride between 55 and 65%. The working temperature is 400-450C. The principal processes are marketed by Denka/Scientific Design, Amoco, BP-UCB, Lonza/Lummus, Mitsubishi Kasei, Mitsui Toatsu, Monsanto and DuPont. Lonzas ALMA process uses a fluid-bed reactor, while the DuPont process uses a circulating-bed reactor. Fig. 10 shows a simplified outline of the two processes. In the ALMA process, after the separation of the catalyst by means of two cyclones in series, the effluents

from the reactor are cooled to 200C, filtered to remove the finest particles and sent to the maleic anhydride recovery section. The latter is based on absorption in cycloaliphatic solvents. The gases are sent for incineration for the production of high temperature steam, while the solution is sent to a stripping unit. The maleic anhydride is then further purified to remove the light products (which are sent for incineration), while the solvent is recycled. The process gives high productivity, production of high temperature steam, reduced quantity of waste, low formation of fumaric acid and heavy products. The production of steam contributes to the economy of the process. The DuPont process is characterized by the use of an innovative reactor for the selective oxidation sector derived from the catalytic cracking, that permits separate contact of the hydrocarbon and oxygen with the catalyst. This brings about a significant increase in the selectivity, but involves the need for higher recirculation of the catalyst between the two reactors (the riser reactor for contact with the hydrocarbon and the fluid-bed reactor for reoxidation of the catalyst). Furthermore it calls for the catalyst to have high mechanical resistance. The DuPont process for synthesis of maleic anhydride is integrated with the downstream hydrogenation section, for the production of tetrahydrofuran. The absorption of the maleic anhydride is, in this case, performed with water.
Phthalic anhydride from o-xylene

Phthalic anhydride is used in the preparation of diesters (plasticizers for PVC), alkylic resins, polyesters and colourings. The original process used naphthalene as the raw material, but today most plants use o-xylene, by virtue of the reduced environmental and safety problems, as well as the greater purity of the product. The principal by-products of o-xylene are o-tolualdehyde and phthalide, small quantities of maleic anhydride, benzoic acid, toluic acid, as well as carbon oxides. The formation of phthalide is a critical aspect of the process, in that, for applications in the polymer sector, the concentration of this compound in phthalic anhydride must be very low. In a multi-tubular reactor, about a third of the catalytic bed enables over 90% conversion of the o-xylene to be obtained, while the remaining two thirds of the bed serves to reduce the concentration of phthalide. The catalysts are based on vanadium oxide supported on TiO2 (in the octahedral crystalline form), with a relatively low surface area (around 10-20 m2/g). K, Cs, Sb, Nb, and P are used as promoters. The latest generation of reactors are loaded with two or three layers of catalyst whose compositions differ from one another (above all in terms of type and quantity of promoter

634

ENCYCLOPAEDIA OF HYDROCARBONS

OXIDATION PROCESSES

vent dry filter maleic anhydride

fresh solvent

n-butane air
A A

steam

vent
H2 recycle tetrahydrofuran

recovery/purification sections. Air is used as an oxidant, although the use of air enriched with oxygen allows higher productivity. The concentration of o-xylene is not greater than 1.2% in air, so as to avoid the formation of flammable mixtures and to achieve good control of the heat of the reaction. Recovery of the phthalic anhydride is carried out by desublimation and then by absorption in oil. The fixed-bed reactor operates at temperatures of between 350-390C and with low spatial rates, while the process in the fluid-bed reactor operates at higher temperatures (450-550C) and with lower contact times. The o-xylene and the air, after preheating, are sent to a multi-tubular reactor operating at about 380C and cooled with molten salts; these salts are cooled externally in a heat exchanger, producing high temperature steam. A yield of phthalic anhydride above 80% is obtained and a conversion of o-xylene greater than 99%. The effluents from the reactor are cooled, producing low pressure vapour, and then sent to an absorber column with water. The solution is first sent to a under vacuum system to decompose the impurities (polymerisation inhibitors are added) and then to a distillation column (also under vacuum) to separate the maleic anhydride and the benzoic and toluic acids from the top. The solution collected at the bottom of the distillation column is sent to a second column, to separate high purity phthalic anhydride ( 99.5% by weight).

reactor

by-products

steam fluid-bed regeneration reactor

References
Arpentinier P. et al. (2001) The technology of catalytic oxidations, Paris, Technip. Bielanski A., Haber J. (1991) Oxygen in catalysis, New York, Marcel Dekker. Centi G., Perathoner S. (2003a) Catalysis and sustainable green chemistry, Catalysis Today, 77, 287-297. Centi G., Perathoner S. (2003b) Selective oxidation. Section E (Industrial processes and relevant engineering issues), in: Horvath I.T. (editor in chief) Encyclopedia of catalysis, New York, John Wiley, 6v. Centi G. et al. (2002) Selective oxidation by heterogeneous catalysis, New York, Kluwer Academic-Plenum. Grasselli R.K. (2002) Fundamentals principles of selective heterogeneous oxidation catalysis, Topics in Catalysis, 21, 79-88.

H2

air

aqueous solution of maleic acid n-butane recycle vent

B B

n-butane

Fig. 10. Simplified outline of the process of the synthesis

of maleic anhydride from n-butane: A, the ALMA process; B, the DuPont process.

elements); in this way it is possible to obtain maximum yield and selectivity, by optimizing the activity profile and minimizing hot spots along the reactor. Promoters in the gas phase, such as SO2 also have a positive effect on performance, although they are no longer used today. Various processes exist, among which the principal ones are those developed by Wacker, BASF, Lonza-Alusuisse, Atochem and Nippon Shokubai. The processes differ in the type of reactor (fixed or fluid bed, although the former is used in the majority of cases), in the composition of the catalyst and in the

Fabrizio Cavani
Dipartimento di Chimica Industriale e dei Materiali Universit degli Studi di Bologna Bologna, Italy

Gabriele Centi
Dipartimento di Chimica Industriale e di Ingegneria dei Materiali Universit degli Studi di Messina Messina, Italy

VOLUME II / REFINING AND PETROCHEMICALS

635

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

11.1.2 Oxidation processes in liquid phase with oxygen


Introduction Liquid phase oxidation is widely applied in the chemical industry for the synthesis of intermediates both for the petrochemical industry and for the production of specialty chemicals and pharmaceuticals, as well as for wastewater decontamination. Liquid phase oxidation is chosen rather than heterogeneous gas phase catalysis in the following cases: when the products are thermally unstable (i.e. in the production of hydroperoxides and carboxylic acids, with the exception of b-unsaturated compounds); when the products are so reactive that at high temperature they can be further oxidized (i.e. epoxides, aldehydes and ketones, with the exception of b-unsaturated compounds, ethylene oxide and formaldehyde); and, in fine chemical production where liquid phase oxidation is especially suitable due to the thermal instability and/or reactivity of the reagents themselves (i.e. in the oxidation of polyhydroxy alcohols). In addition to being the largest application of homogeneous catalysis, liquid phase oxidation is more important than heterogeneous gas phase oxidation processes, on the basis of tonnage and variety of products (Prengle and Barona, 1970a; Lyons, 1980; Sheldon and Kochi, 1981). In homogeneous catalysis, the main technological issues also include selectivity, removal of the heat of reaction and safety considerations. Liquid phase oxidations can be subdivided into the following five types of catalytic processes, based on the mechanism involved and the relative catalysts: a) free-radical, chain oxidation (with and without catalyst), with molecular oxygen as the oxidizing agent; b) redox mechanism with Pd or Cu complexes, and molecular oxygen; c) catalytic oxygen transfer, with either alkylhydroperoxide or H2O2 as the oxidizing agent, and with either homogeneous or heterogeneous catalysts; d) oxidative dehydrogenation with molecular oxygen and with supported, transition metalbased catalysts; e) photocatalytic processes. The first class is the most important from an industrial point of view and, therefore, the most important technological aspects of this class of reaction will be examined in detail.

Liquid phase homogeneous metalcatalysed oxidations by molecular oxygen The oxidation of organic substrates: oxidized substrate + substrate + oxygen donor reduced donor has considerable industrial importance and plays fundamental roles in biological processes. Oxo transfer oxygenations of organic substrates (mainly alkanes and alkenes) are utilized in some processes and are the subject of several investigations. Different oxygen donors may be used: molecular oxygen, peracids, alkylhydroperoxides, hydrogen peroxide, iodosylarenes, amine N-oxides, hypochlorite and ammonium persulfate. Different catalysts are employed, such as metalloporphyrins, metallophthalocyanines, soluble transition metal salts with chelating ligands and Schiff base complexes. Polyoxometallates are among the most interesting systems that have been receiving considerable attention in recent years. Oxidation of a substrate with molecular oxygen may involve two electrons (with the involvement of only one oxygen atom), or four electrons (with the involvement of both oxygen atoms). Coordination with the metal modifies the features of oxygen (i.e. its basicity and radical character), making it more susceptible to reaction with organic substrates. A summary of the possible complexes between molecular oxygen and metal ions is presented in Table 1 (Sheldon and Kochi, 1981). In addition to the superoxo and peroxo species, oxo metal species that occur via the transfer of two electrons from the metal to the oxygen atom will be addressed. Hydroperoxo species can also form via hydrolysis of peroxo complexes, or by molecular oxygen insertion in metal hydrides:
M O O H M O OH

M M

O O H O2

M H M O

M OH

OH

The formation of oxo metal species takes place if the metal is able to furnish two electrons; such metals

636

ENCYCLOPAEDIA OF HYDROCARBONS

OXIDATION PROCESSES

Table 1. Summary of the possible species

formed by complexation of oxygen to metal ions


Mn O2
1)

M(n

1)

O 2 (superoxo) M(n M(n M(n


1)

M(n 1) M (n M(n M(n M


1) 2)

O 2 Mn (m-peroxo) (O O2 O)2 Mn

(O

O)2
2)

1) 1)

2 M(n O2

O2 (oxo)
1)

M(n

(m-oxo)

O H2O2 O O2

M(OH) OH

OH (metal peracid)

M(OH) M(n 1)

H2O M (peroxo)

peroxo

include the redox couples Mn /M(n 2) : Co(I)/Co(III), Ir(I)/Ir(III), Pd(0)/Pd(II), Pt(0)/Pt(II). Metals in higher oxidation states (III and IV) can not transfer electrons to molecular oxygen and do not form stable adducts. In this mechanism, the formation of highvalent oxo metal species occurs through the O O bond scission of an intermediate m-peroxo dinuclear complex, with electron transfer from the metal and with spin pairing in the molecular oxygen. A more suitable representation would take into account the two resonance structures: M(IV) O2 M(II) O

liquid phase homogeneous oxidations in mild conditions, the generation of oxo metal species when using molecular oxygen requires the presence of a reducing agent, which is necessary to furnish electrons to the system. In addition, the catalytic action of the oxo metal species in the oxygenation of the C H bond must always compete with very fast autoxidation processes, thus with free radical chain reactions that are largely unselective and indiscriminate. Therefore, the objective is to develop systems that can operate in the absence of other reducing agents and that kinetically compete with radical-initiated autoxidations. Examples of reductants that reduce molecular oxygen to H2O2 are anthraquinol derivatives, NADH and BH 4 . The metal-peroxo complexes M(n 2) O2 can react with water or an acid, leading 2 to the formation of H2O2 or of M OOH species: M(n M(n M(n
2) 2) 2)

2e O2 2 O2 2

O2 M(n 2) O2 (peroxo complex) 2 (n 2) 2H M H2O2 H M(n 2) O OH

Similarly, metal-O2 superoxo adducts (i.e. where only one-electron charge transfer from the metal to molecular oxygen has occurred) can be reduced to metal peroxo species in the presence of a oneelectron reducing agent (sacrificial reductant): M(n 1) O O (peroxo complex) e M(n
1)

O2 2

The most usual representation of this moiety is M O. Different classes of metal-catalysed oxidations by O2 can be distinguished, according to the nature of the complex formed with the oxygen and the metal oxidation states formed (i.e. to the role played by the metal atom in the oxygen activation; Sheldon and Kochi, 1981; Drago and Beer, 1992). Particular attention will be paid to the oxygenations by oxo metals, which are oxidations of relevant importance in the functionality of organic substrates.
Oxo metal species formed by interaction with peroxides and molecular oxygen

The second step is the oxidation of metal complexes to oxo metal species with H2O2, which can proceed as follows: M(II)(H2O) H2O2 M(IV) O 2H2O

The mechanism for the formation of the M O species can be divided into two steps: the reaction for the formation of the peroxide and that of the peroxide with the metal complex to form the oxo metal species. The metal-catalysed oxidation does not occur as a consequence of the activation of molecular oxygen by the metal, since the metal characteristics do not allow the electron transfer to oxygen. The presence of a reductant enables the conversion of molecular oxygen into H2O2 or into an alkylhydroperoxide (which are stronger oxidizing agents than oxygen) to be carried out; the latter may react with the metal complex to yield the high-valent oxo metal species. In fact, in

or through ligand substitution and formation of the dinuclear m-peroxo complex, which then decomposes to the oxo species. The above is the same as the well-known reaction between hydrogen peroxide (or alkylhydroperoxides) with metal ions, with oxidation of the metal ion and reduction of H2O2, catalysed by ions such as Fe2 or Cu : 2M2 2M3 2H H2O2 2M3 2H 2O H2O2 2M2 O2 2H

In this case, however, molecular oxygen is also generated, since the metal ion with higher oxidation state formed by the reaction catalyses the oxidation of H2O2 to O2. This occurs because the metal ion has two stable oxidation states that is M(II)/M(III) and hydrogen peroxide dismutation is preferred rather than the formation of an even higher oxidation state M(IV). The formation of the oxo metal species theoretically may involve either a homolytic or

VOLUME II / REFINING AND PETROCHEMICALS

637

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

heterolytic rupture of the O O bond in the hydroperoxide. The heterolytic process involves a two-electron transfer from the metal (increasing in oxidation state from Mn a M(n 2) ), while the homolytic process would evidently imply oxidation to M(n 1) . In this case, a radical HO (or RO in the case of alkylhydroperoxides) is also generated, which would be responsible for undesired side reactions due to its high intrinsic reactivity, as occurs in metal-initiated autoxidations. The formation of highly valent oxo metal species has been proposed to occur in monooxygenase enzymes, among which the cytochrome P-450 is the most widely studied and characterized. Such enzymes catalyse reactions in which one atom of oxygen is incorporated into an organic substrate, while the second atom of oxygen is reduced and forms water (Hayaishi, 1974): R H O2 2e 2H R OH H2O

[M(n

1)

OH R ]

M(n

1)

ROH

Fe-porphyrin complexes, analogues of cytochrome P-450 (biomimetic systems), form Fe oxo species via preliminary reduction of molecular oxygen by means of a sacrificial co-reductant, such as NaBH4, LiBH4, ascorbic acid, or Pt/H2 (Groves, 1985; Ortiz de Montellano, 1985; Dawson, 1988; Mansuy and Battioni, 1989; Shilov, 1989). The mechanism involves the reduction of molecular oxygen and the formation of a Fe(III)-peroxo complex, which decomposes to (P )Fe(IV)O (P porphyrin), and which, in turn, acts as the active oxidizing agent. After oxidation, the Fe(III) formed is reduced back to Fe(II). The use of H2O2, instead of O2, renders the sacrificial reductant useless, since the hydrogen peroxide directly forms the oxo metal species (in enzymes, this is referred to as the hydrogen peroxide shunt). Oxo metal centres have been claimed to be the active oxidizing site in Fe(TFPP)N3, {Fe(TFPP)}2O and Mn(TFPP)N3 catalysts (TFPP mesotetraphenylporphyrinato) for the oxidation of isobutane to t-butyl alcohol with molecular oxygen (Ellis and Lyons, 1989a e 1989b). It is claimed that the characteristics of the complexes used make the presence of the co-reductant unnecessary. In the reaction of C H bond hydroxylation in alkanes or in other substrates containing C H bonds, different mechanisms are possible (Hill, 1989), depending on the nature of the intermediate species that is formed. The type of reaction pathway is generally a function of the electron acceptor properties of the oxo metal bond. The following mechanisms are possible: A mechanism of H abstraction: Mn O2 R H [M(n 1) OH R ] with the formation of a radical pair, which then evolves to:

Under mild reaction conditions, this mechanism is generally accepted as the operating one in liquid phase alkane hydroxylations by oxo metal species. H abstraction, where an ion pair is formed, [M(n 2) OH R ], which successively evolves to the ROH product and to M(n 2) . Electron transfer with formation of a radical ion pair [M(n 1) ] [R H] . H abstraction, with the formation of an ion pair [(M OH) R ]. Concerted oxygen atom insertion, involving C O bond formation, which resembles a SN 2 transition state, where the M O bond interacts perpendicular to the C H bond. The radical pair [LxM(n 1) OH R ] (L ligand) of the first mechanism can evolve differently (Hill, 1989). Aside from the possibility of a contemporaneous breaking of the M O bond and the formation of the C O bond (with formation of the alcohol product), a possible fate of the radical pair is conversion to an hydroxy-organometallic intermediate: (HO)(Lx)Mn R (Kochi, 1973), which may evolve to the ROH and LxM(n 2) via reductive elimination. Another possible evolution of the radical pair intermediate is conversion into an ion pair by electron transfer: [LxM(n 2) OH R ]. Finally, as pointed out by Hill (1989), formation of the alcohol by hydrolysis of an intermediate ester may also occur, the ester being formed by the attack of a freely diffusing radical at the pendant oxygen. Diffusion of the radicals out of the pair is possible especially in those cases where the radical is particularly resistant towards oxidation (i.e. in primary radicals). Therefore, the fate of the intermediate radical pair is a function of the nature of the complex and the type of the C H bond, which is to be oxyfunctionalized. Theoretically, an organometallic intermediate could be obtained via an electrophilic process involving the contemporaneous formation of the C Mn bond and the O H bond by cleavage of the C H bond in the substrate. However, no evidence exists for such a mechanism in C H oxygenation by oxo metal species. Instead, this mechanism has been claimed in the case of the oxidation of CH2 groups to the corresponding ketones catalysed by the so-called Gif(III) system, and the subsequently developed Gif(IV), GO, GoAgg(I) and GoAgg(III) systems (Barton et al., 1989; Sheu et al., 1990; Barton and Doller, 1991). The ketone can be produced with very high selectivity. These systems are essentially constituted of Fe(II), or Fe(III), pyridine and acetic acid, aside from the substrate to be oxidized, while the oxidizing agent may

638

ENCYCLOPAEDIA OF HYDROCARBONS

OXIDATION PROCESSES

be either a peroxide or molecular oxygen (with a reducing agent). The highlyvalent Fe(V) O species is generated, which inserts into the C H bond (Barton and Doller, 1991): Fe(III) O Fe(V) O RCH2R Fe(III) O Fe(V)(OH) CRR H

the metal-catalysed radical chain reaction, including the well-known Haber-Weiss mechanism of ROOH decomposition, is the following: M(II) ROOH M(III)OH RO (reduction step) M(III)OH ROOH M(II) ROO H2O (oxidation step) and the chain reaction is propagated: RO ROOH ROH 2ROO 2RO O2 ROO RH ROOH ROO R

followed by evolution through several steps to the formation of the ketone and to minor amounts of the secondary alcohol.
Metal peroxo species

As mentioned above, free-radical mechanisms are very important in the metal-catalysed oxidation of organic substrates, and often they are kinetically the most important process in competition with oxo metal transfer reactions. Usually, these processes are indiscriminate and necessarily lead to poor selectivities when the substrate has different sites susceptible to attack. Even so, in some cases, these processes have industrial applications. Reactions of peroxides (i.e. hydrogen peroxide, alkyl hydroperoxides) with organic substrates, catalysed by metal ions, can be divided into two classes: homolytic, one-electron processes, which involve the formation of free radical intermediates, and heterolytic, two-electrons processes, where the metal forms metal hydroperoxide or metal alkylperoxide complexes, able to attack the organic substrate.
Homolytic processes

In the free radical autoxidation process, metal catalysis involves the formation of chain-initiating radicals via reaction with ROOH; therefore, the metal ion is more an initiator than a catalyst. When the metal compound has two oxidation states of comparable stability, the reduction and the oxidation reactions occur concurrently. Cobalt, iron, copper and manganese complexes are generally effective compounds for the homolysis of peroxidic compounds. Correspondingly, Mn(II)/Mn(III) and Fe(II)/Fe(III) are effective catalysts for the decomposition of H2O2, where the metal acts as both reductant and oxidant. Since alkylperoxy radicals are strong oxidants, they oxidize the reduced form of the metal: M(II) ROO M(III) O O R

Reaction between the metal species and the organic substrate generates the free radical R : Mn + R H M(n 1) + R H R H B R BH , B base The formation of R by abstraction of H by the metal, with the formation of a hydride species, can be disregarded due to the fact that RH bonds are generally stronger than M H bonds. The free radical R can then evolve either to the formation of a carbocation (electron transfer process): R Mn R M(n
1)

Therefore, chain propagation can be inhibited by the formation of this metal alkylperoxo compound, which corresponds to an induction period that is observed in metal-catalysed autoxidations, especially in media of low polarity. The induction period can be eliminated by the addition of small amounts of an alkylhydroperoxide to the reaction medium. Termination of the chain reaction occurs through the following reaction: 2ROO R O O O O R O2 R 2C O

or via a ligand transfer mechanism: R Mn X RX Mn

and again by the above-mentioned reaction, at high metal concentration, the prevailing mechanism of termination is M(II) ROO M(III) O O R. The activating effect of some metal ions on the oxidizing activity of H2O2 is well known, as in Fentons reagent: Fe(II) H2O2 Fe(III)OH OH H2O2 H2O HO2 OH

The prevailing mechanism is a function of the nature of the ligand in the metal complex, and both mechanisms can indeed operate at the same time. The reactivity of alkylperoxides and hydroperoxides with metal ions is of the above-mentioned type. Metal ions, such as Co(III), decompose ROOH, generating alkylperoxy radicals, ROO , which initiate autoxidation reactions of organic substrates. A general scheme for

(undesired reaction of unproductive H2O2 decomposition, which lowers the yield to oxidized products) Fe(III)OH HO2 Fe(II) O2 H2O

The OH produced generates R by reaction with

VOLUME II / REFINING AND PETROCHEMICALS

639

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

organic substrates, finally converted to oxidized compounds: RH OH R R Fe3 Fe2 R Fe2 Fe3 R H RH H2O R R products

Table 2. Oxidizability of some organic substances

(Howard, 1972)
Substrate Benzaldehyde 2,3-dimethyl-2-butene Cyclohexene Cumene Ethylbenzene p-xylene Toluene 103kpr 290 3.2 2.3 1.5 0.2 0.005 0.01 2kter

Several examples of oxidation reactions performed by peroxides in the presence of metal ions are described in the book of Sheldon and Kochi (1981). Some of the most important industrial liquid phase oxidation processes proceed via an autoxidation mechanism in the presence of molecular oxygen, such as: oxidation of cyclohexane to adipic acid, oxidation of p-xylene to terephthalic acid, oxidation of toluene to benzoic acid, oxidation of cumene to phenol via intermediate cumylhydroperoxide, and oxidation of linear aldehydes to linear carboxylic acids. Autoxidation processes are indiscriminate and usually low selectivities are achieved. Nevertheless, with molecules that contain only one reactive C H bond, it is possible to obtain high selectivity, especially if low conversions are maintained. In industrial practice, usually an initiator (In) is added, necessary to start the chain reaction and to form ROOH: In2 2In (by thermal decomposition) In RH InH R In the presence of molecular oxygen, the radical species is transformed to a peroxy radical, which then evolves propagating the chain (rate-determining step under usual reaction conditions, at a partial pressure of molecular oxygen higher than 0.13 bar): R O2 RO2 RO2 RH RO2H R

Table 3. C Compound CH3 C3H7 C3H7 i-C4H9 CH2 CH2 C6H5 C6H5 H H(C1) H(C2) H(C3) CH CH H CH2 H H CH2 H

H bond energies
Energy (kcal/mol) 103 99 94 90 105 85 103 85

The reaction is terminated by the coupling of the RO2 radical. Therefore, hydroperoxides are the primary products of the reaction network. Table 2 reports the oxidizability of various organic substrates (Howard, 1972), while Table 3 presents some C H bond energies (Kerr, 1966; Benson and Shaw, 1970). The oxidizability of organic substrates to undergo autoxidation reactions is expressed in [10] (see below) by the ratio between the rate constant relative to the propagation reaction (kpr in Table 2), and the square root of twice the rate constant relative to the termination reaction via the mutual coupling of two alkylperoxy radicals (kter in Table 2). At high substrate conversion, the formation of secondary products (aldehydes and ketones) considerably lowers the selectivity, and thus the conversion is usually maintained lower than 20%. Oxidation is rapid if the bond that will be formed in the rate determining step,

ROO H (bond strength 90 kcal/mol; Benson, 1965), is stronger than the broken bond R H. The overall mechanism in the case of the oxidation of alkylbenzenes under oxygen pressure, in acetic acid, and with Co(III) (CH3COO)3 as the catalyst (and also traces of an activator, to reconvert Co(II) to Co(III), avoiding long induction periods) is shown in Table 4. The last reaction is inhibited by the high concentration of Co2 , which traps the benzylperoxy radical.
Nucleophilic attack on peroxo complexes

Transition metals such as vanadium, molybdenum, tungsten, selenium and titanium in their high oxidation state promote the heterolysis of the O O bond in H2O2 and in alkyl hydroperoxides; these metals have the role of improving the electrophilicity of the peroxide complex. An attack on a coordinated peroxo or alkylperoxo ligand by the substrate occurs, usually with high specificity. The most important example is the industrial production of propylene oxide from propylene and alkyl hydroperoxide, catalysed in both the homogeneous and heterogeneous phase. The above-mentioned metal oxides catalyse the H2O2 reactions through the formation of inorganic

640

ENCYCLOPAEDIA OF HYDROCARBONS

OXIDATION PROCESSES

Table 4. Aerobic oxidation of alkylbenzenes

catalyzed by a CoIII complex


ArCH3 Co3 ArCH 3 ArCH 3 Co2 O 2

ArCH2 H ArCH 2 Co2 (in the absence of O2) ArCH2OAc H

ArCH2 Co3 ArCH 2 AcOH ArCH2 O2

ArCH2O2 (in the presence of O2) ArCH2O2Co2 ArCHO HOCo3 AcOCo3 H2O

ArCH2O2 Co2 ArCH2O2Co2 HOCo3 HOAc

Fig. 2 reports the mechanism proposed by Clerici and Ingallina (1993b) for the epoxidation of alkenes by H2O2 over Ti-silicalite catalysts. A five-membered ring species is the intermediate compound formed by interaction between the Ti sites in silicalite and H2O2, also including the interaction with the solvent ROH. The epoxidation rate was found to be the highest with methanol as the solvent, and lowest with the bulky t-butanol, due to the decreased electrophilicity and increased steric constraints of the cycle (Clerici, 1991; Clerici et al., 1991; Bellussi et al., 1992). An alternative mechanism (Mimoun, 1982) involves the addition of the alkene double bond to form a

O M O R H O

H O

ArCH2O2 ArCH3

ArCH2OOH ArCH2

OH epoxide OR

peracids, which undergo heterolytic cleavage of the O O bond in the presence of nucleophiles;
O O MO3 H2O2 HO M O OH

O OR M epoxide ROH

M O

more generally: M O ROOH HO M O OR


O

In olefin epoxidation: RCH HO CHR HO MO R M O OR RCH(O)CHR HO M OR HO M O OR R OH

O OR O M OR epoxide

R OOH

alkyl hydroperoxides are preferred to H2O2 due to the superior activity. The retarding effect of H2O2 has been attributed to the presence of water. A peculiar feature of this reaction is its stereoselectivity for cis and trans epoxides from the corresponding alkenes. Asymmetric epoxidation of prochiral allylic alcohols can be carried out with optically active diethyltartrate catalysts and t-BuOOH, with more than 90% enantiomeric excess (Finn and Sharpless, 1985). The reason why d 0 transition metals are specific for this reaction has been attributed to the fact that interaction between the oxygen atom and vacant t2 g orbitals of the metal ion provides a low energy route for the formation of the C O bond from the peroxometallocycle adduct (Purcell, 1985). Different mechanisms have been proposed for the oxygen transfer in the epoxidation of olefins. One mechanism involves nucleophilic attack of the double bond on the alkyl hydroperoxide, by one of the different activated complexes, as shown in Fig. 1 (Sheldon, 1973; Chong and Sharpless, 1977).

Fig. 1. Proposed mechanisms for olefin epoxidation,

involving direct oxygen transfer by the alkyl hydroperoxide, in ROH solvent (adapted from Sheldon and Kochi, 1981).

R SiO Ti SiO SiO R alkene SiO SiO SiO O Ti O H O C H C Ti H epoxide O SiO


ROH, H2O2

SiO SiO SiO

O Ti O

H O H

R O H

Fig. 2. Proposed mechanism for olefin epoxidation with

H2O2 catalyzed by Ti-silicalite, in ROH solvent (Clerici and Ingallina, 1993).

VOLUME II / REFINING AND PETROCHEMICALS

641

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

peroxometallocycle, thereafter releasing the epoxide (Fig. 3). Another mechanism involves the oxo metal moiety, which is converted to a metal peroxo (Mimoun et al., 1970) complex the active site.
Metal-centred oxidation of coordinated substrates

O O

R M

O O

R M

O O

R MOR O

Fig. 3. Mechanism for olefin epoxidation involving a

This class of reactions involves the oxidation of a coordinated substrate by a metal ion. Examples include the Pd(II)-catalysed oxidation of olefins (Wacker process), and the synthesis of ketones by the oxidehydrogenation of alcohols. The role of molecular oxygen is to regenerate the high oxidation state of the metal. The substrate is activated towards nucleophilic attack by the metal through the formation of a p-complex with the metal. The overall reaction is an oxidative nucleophilic substitution of hydrogen with two-electron reduction of the metal; therefore, the process is heterolytic. Group VIII metal salts, such as Pt(II), Ir(III), Ru(III) and Rh(III), oxidize olefins in the same way: Pd(II)X2 RCH CH2 H2O RCOCH3 Pd(0) 2 HX

peroxometallocycle intermediate (Mimoun, 1982).

Cl
2 PdCl4

C2H4

H2O

Cl

Pd

OH2

H2O
H

Cl Cl Pd OH2
Cl

Cl Pd OH2 H

Cl Pd OH2 OH

CH2CH2OH CH3CHO Pd0

CH2CH2OH HCl H2O

Nucleophilic attack of water on a Pd(II)-ethylene p complex forms ethanol, which remains coordinated by a Pd C s bond (Fig. 4; Drago and Beer, 1992). The dissociation of a Cl ligand from the complex and the elimination of hydride from the ethanol form a vinyl alcohol p complex with Pd(II). The formation of the aldehyde is obtained by tautomerization of the dissociated vinyl alcohol from the complex and reduction of Pd(II) to Pd(0). Reoxidation of palladium is achieved through a Cu(II) co-catalyst, which is reduced to Cu(I), and then finally reoxidized by molecular oxygen to Cu(II). Kinetics of liquid phase oxidation of hydrocarbons Analysis of the kinetics of liquid phase oxidation can be made according to the nature of the reaction mechanism: reactions that involve the homolytic scission of the C H bond in the organic substrate, thus involving free radicals (autoxidation), where the metal catalyst acts more like an initiator, and reactions that involve the heterolytic scission of the C H bond, where the substrate coordinates to a metal ion catalyst. Autoxidation processes are of fundamental importance in the chemical industry, and a great number of reactions imply this kind of mechanism, as shown in Table 5 (Sheldon and Kochi, 1976; Carr and Santacesaria, 1980; Santacesaria and Pimpinelli, 1986). The process is autocatalytic and exhibits an induction period, an acceleration (controlled by the propagation and branching reactions) and a

Fig. 4. Mechanism of oxidation of ethylene to acetaldehyde

(Wacker process).

deceleration (controlled by termination reactions). The induction period is due to the difficulty in breaking the C H bond and can be overcome by the addition of small quantities of initiators, which readily decompose in the reaction medium. In a typical air-sparged system, in the zone near the sparger, the liquid contains higher concentrations of molecular oxygen and the latter can rapidly scavenge radicals via the reaction: R O2 ROO

Therefore, the overall rate is controlled by the chemical reaction. Under these conditions, the reaction is zero order with respect to molecular oxygen, and the apparent activation energy is typical of reactions under chemical reaction control. The concentration of molecular oxygen in the bubbles progressively decreases, going from the bottom to the top of the reactor, and the mass transfer of molecular oxygen from the gas to the liquid phase can become rate limiting, since the reaction between an alkyl radical and oxygen is extremely fast. Under these conditions, the supply of oxygen can become first order with respect to O2 partial pressure. In this zone, the apparent activation energy falls to low values, 5 kcal/mol. The main reactions involved in the process of homolytic oxidation are summarized in Table 6. When molecular oxygen is present in relatively high

642

ENCYCLOPAEDIA OF HYDROCARBONS

OXIDATION PROCESSES

Table 5. Most important industrial processes involving liquid-phase oxidation of hydrocarbons

through homolytic mechanisms


Reactant n-butane Isobutane Naphtha Waxes (C18-C30) Cyclohexane Cyclohexanol/o Toluene Benzoic acid p-xylene Cumene Acetaldehyde Acetaldehyde Acetic acid t-butylhydroperoxide Acetic acid Fatty acids Cyclohexanol/cyclohexanone Adipic acid Benzoic acid Phenol Terephthalic acid Cumylhydroperoxide Acetic acid Acetic anhydride Product Catalyst Co acetate/acetic acid Co acetate/acetic acid Mn salts/acetic acid Co naphthenate Mn-Cu acetate Mn or Co bromide/acetic Cu-Mg benzoate Mn or Co bromide/acetic Mn acetate/acetic acid CuCo acetate/ethyl

concentrations (for oxygen pressures usually higher than 0.13 atm in the gas phase), the rate determining step is the propagation reaction, kox 104 105 kpr (Table 6). The RO2 species is thus the predominant one and mainly contributes to termination reactions. When the contribution of branching reactions is negligible, because ROOH is not decomposed under the reaction conditions (thus its contribution to the generation of radicals is negligible and it is the main product of reaction), under these conditions the rate of oxygen transformation may be expressed as the balance between oxygen consumption in oxidation and oxygen formation in termination: d[O2] 2 14421 r r [1] ox ter kox [R ][O2] kter[RO2] dt On the other hand, under stationary conditions, the rate of R formation (initiation plus propagation) is equal to the rate of R disappearance (oxidation), provided that termination reactions involving the R species can be neglected. Therefore: [2] kox [R ][O2] kpr[RO2][RH] rin

[5]

d [O2] 14421 dt

rin kpr 1242 2kter

0.5

rin [RH] 1 2

The term relative to the rate of initiation can be neglected, and the final general relationship is obtained for the rate of oxygen (or hydrocarbon) depletion: [6] d[O2] 14421 dt rin kpr 1242 2kter
0.5

[RH]

This expression is independent of oxygen concentration, and is thus generally valid under conditions of high oxygen concentration. A more general expression has been proposed (Twigg, 1962; Franz and Sheldon, 1991), which considers all of the more important steps in the reaction scheme (reactions 3 and 4 in Table 6 for propagation, reactions 7 and 9 for termination, with corresponding kinetic constants kpr3, kpr4, kter7 and kter9): [7] d[O2] 14421 dt
0.5 rin kpr4 kpr3[RH][O2] 12111111111111124341142 0.5 0.5 0.5 0.5 0.5 {kpr4 kter7 [RH] kpr4 kter9 [O2] kter9 kter7 rin }

Under stationary conditions, the rate of initiation equals the rate of termination, and therefore: [3] rin 2kter[RO2]2 rin 14421 2kter
0.5

which gives: [4] [RO2]

By replacing the corresponding terms in the expression for oxygen disappearance, it is possible to obtain:

Two limit situations can be considered. The first regards oxygen concentrations approaching zero. Under these conditions, the rate of reaction is proportional to the oxygen concentration in the liquid phase, since the term at the denominator containing [O2] can be neglected. If, instead, the oxygen concentration is very high, the rate of reaction becomes independent of oxygen concentration and the relationship of [6] is valid, where kpr kpr4 and

VOLUME II / REFINING AND PETROCHEMICALS

643

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

OH radicals, which can lead to very fast reactions:


Table 6. Reaction scheme for autoxidation

of organic substrates
1 In2 2 In 3 R 4 RO2 2 In (initiation) RH O2 RH R InH (initiation)

RO OH

RH RH

ROH R H2O R

RO2 (propagation by oxidation) ROOH R (propagation)

The prevailing termination reaction again involves the RO2 species, since the RO and OH radicals (generated by ROOH decomposition) are readily converted. Under these conditions, the rate expression for RH transformation becomes: [9] d[RH] 144231 dt
2 3k pr [RH]2 1441144 1 kter

RO RO2 H2O 5 2ROOH (branching by bimolecular decomposition) RO OH (branching 6 ROOH by unimolecolar hydroperoxide decomposition) 7 2R 8 R non radical products (termination) RO2 non radical products (termination) O2 non radical products (termination)

9 2RO2

which takes into account that RH is consumed through reaction with RO2, RO and OH , and that three molecules of RH are consumed for each molecule of ROOH that has decomposed. This expression represents a limit situation, and hence gives the maximum rate that can be achieved. More generally: [10] d[RH] 144231 dt n kpr 1 4 11341 [RH] f (2kter )0.5

2kter kter9. On the other hand, since the value of kpr4 depends to a large extent on the nature of the organic substrate, when it is very high (e.g. in the case of aldehydes), the first term in the denominator of the general equation is dominant and, therefore, the equation becomes: [8] d[O2] 14421 dt
0.5 0.5 rin kpr3[O2]kter7

The rate of the reaction is thus proportional to the oxygen concentration. The value of oxygen concentration at which the passage from first-order dependence to zero-order dependence occurs is therefore a function of the nature of the organic substrate. For aldehyde oxidation, the reaction depends on oxygen even at relatively high partial pressures, while for cyclohexane oxidation, the reaction depends on oxygen partial pressure up to 0.2-0.3 atm, and for cumene oxidation up to 1.5-1.7 atm (Ladhaboy and Sharma, 1969; Manor and Schmitz, 1984; Andrigo et al., 1992). The ratio [kpr /(2kter )0.5] represents the susceptibility of the substrate towards oxidation (see again Table 2; Sheldon and Kochi, 1976). The propagation rate is affected by the C H bond energy (the homolytic scission of this bond is involved and it is necessary that the ROO H bond, which is going to be formed, is at least energetically equivalent to the C H bond to be broken) and also by the reactivity of alkylperoxo radicals (Howard, 1973; Sheldon and Kochi, 1976; Gates et al., 1979). The latter are more reactive in the presence of an electron-withdrawing substituent. Under conditions at which the hydroperoxide is not stable, decomposition of the latter generates RO and

where n is the number of radicals produced for each ROOH decomposed and f is the fraction of RH consumed via attack by RO2. Once again, the term [kpr /(2kter)0.5], representing the relative importance of propagation and termination reactions, plays an important role in determining the rate of substrate transformation. The presence of a metal ion acting as a catalyst may increase the activity of the autoxidation process. The catalytic action is related to the redox property of the metal ion, that is, to the redox potential for the couple Mn /M(n 1) . One of the most important effects is to favour the decomposition of the hydroperoxides: ROOH ROOH M(n Mn
1)

RO OH Mn (n 1) RO2 H M

Clearly, only those metals having two oxidation states of comparable stability (i.e. cobalt and manganese ions) can perform this catalytic cycle (Sheldon and Kochi, 1976). From the kinetic point of view, [10] can be used to calculate the maximum rate of reaction, assuming that n 1 and f 2/3. In this case, where the catalyst is only involved in the decomposition of ROOH, n/f 1.5. The metal can also interact directly with the substrate: Mn RH M(n R O2 RO2 M(n 1) ROOH
1)

H Mn

R RO OH

Also in this case, the kinetic law obeyed is the same as for the previous case, with n 2 and f 1/3. In all of these cases, therefore, the metal acts as an

644

ENCYCLOPAEDIA OF HYDROCARBONS

OXIDATION PROCESSES

initiator, decreasing the induction period and eventually favouring the oxidation process at lower temperatures. When the bivalent oxidation state of the catalyst is only involved in ROOH decomposition and the trivalent state in the reaction with RH, then n/f 6. Intermediate values of the n/f ratio, between 1.5 and 6, indicate that the catalyst not only takes part in the redox reaction with peroxidic compounds, but also reacts with RH through the trivalent oxidation state. In industrial practice, small amounts of substances highly prone to autoxidation are added to the reaction medium, which will readily give rise to a co-oxidation effect and provide the hydroperoxides for maintaining the chain, thus leading to an increase in the reaction rate. In this way, it is possible to avoid undesired cooxidation on the desired products of partial oxidation (i.e. aldehydes, ketones), which are more easily subjected to autoxidation than the reactant. The kinetic expressions described above are often in good agreement with the experimental limiting rate. However, this is not the case for the oxidation of alkylaromatics catalysed by cobalt acetate, where the reaction is first order upon the hydrocarbon and also depends on the catalyst concentration (Chester et al., 1977). This is due to the different reaction mechanism involved. Regarding oxidation reactions occurring via heterolytic dissociations of the C H bond in the substrate, they are catalysed by V, W, Mo and Ti ions through the transfer of two electrons of the substrate coordinated to the metallic complex. In other cases, such as the Wacker process, the action of two redox couples Pd(II)/Pd(0) and Cu(II)/Cu(I) is necessary for the catalytic cycle. In the mechanism, the rate limiting step is the conversion of the p-complex 2 between the olefin and PdCl 4 into a s complex, the other stages being at equilibrium. The rate expression for ethylene disappearance, which better describes the experimental results, is the following (Moiseev et al., 1974): [11] d[C2H4] 1442231 dt [PdCl2 ][C2H4] 4 1 k 144111441144 [Cl ][H3O ]

Air vs. oxygen in liquid phase oxidations

The use of molecular oxygen in liquid phase oxidations may provide a significant improvement in the reaction kinetics when the overall process is controlled by the reaction and the rate depends on the oxygen partial pressure over a wide interval of oxygen concentration (though not often the case, as described above). Indeed, the oxygen concentration in the bubbles (in addition to the continuous gaseous phase at the top of the reactor) is 100% (neglecting the

vapour pressure of the liquid), against an average 5 mol% in the case of air as the oxidizing agent, thus improving the transfer of molecular oxygen as well as the rate of reaction. As a consequence, similar productivities as with the air-based process can be achieved at significantly lower temperatures and pressures. This may lead to a considerable improvement in selectivity, when the reaction network consists of consecutive reactions of oxidative degradation (usually characterized by higher activation energy than the selective reaction) to the desired product. Lower temperatures also entail less solvent loss. It may also happen that an increase in the oxygen partial pressure, as a consequence of feeding oxygen instead of air, leads to a considerable improvement in the oxygen transfer rate, while the rate of reaction is not much affected by the variation in oxygen concentration. Under these conditions, a passage from a mass-transfer limited process (as is usually the case for the oxidation of liquid hydrocarbons) to a kinetically controlled process can occur, and thus the reaction is transferred from the film surrounding the bubble (at the interface between the gas and the liquid phase) to the continuous liquid phase. This may also change the temperature distribution from a considerable temperature gradient in the film in the former case (where the heat of reaction develops in the thin film) to a lower, more uniform temperature in the latter case. The above may considerably improve the selectivity. The presence of a high oxygen concentration in the reaction zone is also important for favouring the oxidation reactions against the reactions between intermediates, leading to undesired high-molecular weight compounds. A high oxygen concentration in the reaction zone is obtained when the process is kinetically controlled or when, under mass-transfer limiting conditions, the rate of reaction is not very much higher than the rate of diffusion (therefore, concentration gradients in the film are not too pronounced). Higher oxygen concentrations may thus lead to a higher oxygen concentration in the reaction zone. Another advantage in the use of oxygen instead of air is the reduction in total gas throughput, with a consequent reduction in the energy required for gas compression as well as in the amount of vent gas (which is also much more concentrated in the organic contaminants, due to the absence of diluent nitrogen) to be treated, with large gas treatment equipment. Therefore, the nitrogen favours stripping of the solvent and volatile organic compounds from the solution, compounds which have to be removed to conform with the regulations for environmental protection.

VOLUME II / REFINING AND PETROCHEMICALS

645

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

Free-radical chain oxidations


Industrial processes

Free-radical chain oxidation consists in the oxidation of organic substrates through the interaction of molecular oxygen with radicals formed by homolytic cleavage of C H bonds (which, in fact, are referred to as homolytic oxidation reactions; Prengle and Barona, 1970a; Sheldon and Van Doom, 1973, 1981; Lyons, 1980; Emanuel and Gal, 1986). Three types of industrial, radical chain oxidation applications can be distinguished: autoxidation, a spontaneous oxidation (without a catalyst) of alkanes and alkylaromatics to produce hydroperoxides or of aldehydes to produce peracids; catalysed autoxidation, where the catalyst superimposes and/or partially replaces the mechanism of autoxidation, which is involved in the production of acids, alcohols and aldehydes; acid decomposition of hydroperoxides, which serves to obtain alcohols and ketones. The common first step in all of the reaction mechanisms is the introduction of an OOH group into an organic substrate (formation of hydroperoxides or of peracids), which evolves further in the presence of catalysts to more stable products, depending on the type and amount of catalyst, the structure of the organic molecule containing the OOH group, and the reaction conditions and reactor type. A further differentiation of the processes is related to the OOH group decomposition step, which can be carried out either in situ or in a second catalytic step. The free-radical chain oxidation is involved in the industrial production of the following compounds: hydroperoxides, carboxylic and aromatic acids, alcohols and ketones (only for some special applications) and hydrogen peroxide. The main industrial processes can be classified as follows:

Production of hydroperoxides: isobutane to t-butylhydroperoxide (oxidizing agents for epoxidation of olefins); ethylbenzene to ethylbenzene hydroperoxide (oxidizing agents for epoxidation of olefins); cumene to cumylhydroperoxide (and then to phenol and acetone); p-diisopropylbenzene to the corresponding dihydroperoxide (and then to hydroquinone). Production of acids: p-xylene to terephthalic acid; n-butane to acetic acid; toluene to benzoic acid (and then to phenol); higher paraffins to high molecular weight acids and alcohols; cyclohexane to cyclohexanone and cyclohexanol (and then to adipic acid); cyclododecane to cyclododecanol and cyclododecanone (and then to cyclododecan dicarboxylic acid); acetaldehyde to acetic acid; pseudocumene to trimellitic acid; m-xylene to isophthalic acid; 2,6-dimethyl naphthalene to 2,6-naphthendicarboxylic acid; n-butyraldehyde to n-butyric acid; paraffins to fatty acids. Production of ketones, alcohols, anhydrides and quinones: fluorene to fluorenone; acetaldehyde to acetic anhydride; naphthalene to naphthoquinone; anthracene to anthraquinone; cymene (p-methylisopropylbenzene) to cresol; long-chain n-alkanes to secondary alcohols. The operating conditions and performance of the most important industrial processes of autoxidation of organic substrates are summarized in Table 7.
The mechanism of autoxidation and kinetic considerations

The autoxidation of organic substrates occurs through three stages or periods: an induction period, a steady chain-propagation period and a termination period. The reaction mechanism (see above: Kinetics of liquid phase oxidation of hydrocarbons) involves

Table 7. Reaction conditions of industrial liquid-phase, radical-chain oxidation process Substrate n-butane Isobutane Cyclohexane Toluene Ethylbenzene p-xylene Cumene Acetaldehyde Oxidant O2/air O2 air O2/air air air O2/air O2/air T (C) 160-180 110-140 150-170 140-180 120-140 195-205 90-130 60-80 P (atm) 50-60 25-35 8-10 2-10 3-5 15-30 1-8 2-10 Per passage conversion (%) 90 25-40 4-10 20-40 10-20 95 25-35 91-98 Selectivity (%) 50-65 50-60 75-85 91-93 84-87 90-95 90-97 90-95 Catalyst Co, Mn salts no Co salts Co salts no Co/Mn/Br Co, Mn salts Co, Mn salts

646

ENCYCLOPAEDIA OF HYDROCARBONS

OXIDATION PROCESSES

initiation, propagation, branching and termination reactions. Initiation reactions. The initiator reacts with the substrate RH, generating radicals R : In2 2In In RH R InH

or where the organic substrate is itself an initiator: RH R H RH O2 R HO2 2RH O2 2R H2O2 The initiation can be brought about by the introduction of thermally unstable compounds, such as peroxides or azocompounds, generating radicals or metal ions, which autoxidize more readily than RH. The initiation step requires a finite time to develop a sufficient concentration of radicals to support the propagation step. Usually, the decomposition of the hydroperoxide formed with the reaction is the source of radicals (i.e. branching reactions, see below). In this case, the reaction temperature is related to the temperature of decomposition of the hydroperoxide. For this reason, in industrial processes, usually the product is partially recycled to the first reactor in order to furnish the necessary concentration of initiator. Propagation reactions. The radicals generated react with molecular oxygen forming RO2, which in turn reacts with RH, generating other R : R O2 RO2 RO2 RH R ROOH

Branching reactions. These reactions are due to the decomposition of alkylhydroperoxy species to generate RO , ROO , OH radicals. The nature of the products formed in the branching reactions depends on the type of substrate. ROOH RO 2ROOH ROO RO RH R OH RH R ROOH ROH OH RO ROH H2O 2RO H2O H2O

Termination reactions. The radicals generate non-radical products: 2R R2 R RO2 ROOR 2RO2 ROOR O2 RO R ROR The recombination of radicals to terminate the chain reaction occurs more easily with large radicals, whose lifetime is longer than for small radicals. In fact, large radicals are less reactive because they dissipate energy by atomic vibrations. The termination

rate depends on the structure of the radical and also on the oxygen partial pressure. In all industrial processes, at the end of the reactor, the partial pressure of oxygen is maintained practically nil, mainly for safety reasons. Therefore, radicals accumulate and termination occurs by their recombination. The presence of foreign molecules (impurities) that interact with radicals can also be the source for termination reactions. The formulation of kinetic expressions for the autoxidation of organic substrates must not only take account of the reactions occuring in the liquid phase, but also of the transfer of molecular oxygen from the gaseous to the liquid phase (Prengle and Barona, 1970b; Hobbs et al., 1972a; Astarita et al., 1983; Doraiswamy and Sharma, 1984). When the entire process is either mass-transfer limited or reaction limited, more simple rate expressions can be used, which are representative of the rate-determining step. This is usually the case for liquid phase autoxidation processes, where the diffusion of oxygen is very quick for oxygen partial pressures above 0.1-1.0 atm (the limit value depending mostly on reaction conditions), while it becomes rate-controlling when the oxygen partial pressure is lower than these values (this typically occurs in continuous stirred reactors; Uri, 1961; Bamford and Tipper, 1980). However, general rate expressions are necessary when the rate determining-step changes in the same reactor. For instance, in bubble reactors where the oxygen partial pressure progressively diminishes along the reactor, one can move from a reaction-limited region (where the oxygen partial pressure is high) to a region where the process is oxygen-transfer-limited (Hobbs et al., 1972a; Jacobi and Baerns, 1983). In correspondence, one may move from a region where the apparent activation energy is higher than 20 kcal/mol to a region where it is very low, typical of diffusional phenomena (less than 5 kcal/mol). The situation becomes even more complex when taking into account the very complex reaction mechanism, with the formation of many by-products and the participation of several intermediate chemical species (i.e. radical compounds). However, it is usually possible to adopt simplified kinetic models, where the concentrations of radical intermediates are not taken into consideration (i.e. lumped models). This type of approach has been utilized for several different examples of liquid phase catalytic oxidation reactions (Cavalieri dOro et al., 1980; Chen et al., 1985; Krzysztoforski et al., 1986; Morbidelli et al., 1986; Cao et al., 1994). The direct reaction between RH and O2 to yield ROOH is thermally neutral and is characterized by a high activation energy ( 35 kcal/mol). Therefore, it is very slow at temperatures below 150C. The cleavage

VOLUME II / REFINING AND PETROCHEMICALS

647

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

of RH by reaction with RO2 is 104 to 106 times slower than the reaction between R and O2 , and thus the former reaction is usually the rate-determining step. The rate expression then becomes (in a chemically-controlled regime): [12] d[RH] 144231 dt rpr kpr [RO ][RH]

film, and thus [O2]*= P 2* /H = P 2/H (where H is the O O Henry constant for molecular oxygen and P 2 is the O partial pressure of molecular oxygen in the gas phase), the following expression is finally obtained (Jacobi and Baerns, 1983): [19] d[O2] 144231 dt P2 1 kter O 123 133 11211 H akL koxkpr [RH]
1

Under these conditions, the rate expression does not depend on oxygen partial pressure. Since the reaction between the R species and oxygen is very quick, often total molecular oxygen consumption is already achieved in the liquid film. The reaction between O2 and R becomes rate-determining only when the partial pressure of oxygen is very low (below approximately 0.15 atm); the rate of reaction then becomes dependent upon molecular oxygen and is strongly limited by oxygen transfer into the liquid phase: [13] d[O2] 144231 dt r ox kox[R ][O2]

Higher rates of hydroperoxide formation are observed from substrates, which present either more labile C H bonds or form more stable radicals. In the alkane series, the rate is higher for tertiary C H bonds, followed by bonds in secondary and primary carbon atoms. In arylalkanes, the order is the following: tertiary secondary primary benzylic. This order is due more to the instability of the radicals formed than to the lability of the C H bonds (which is the reason for the lower oxidizability of toluene).
The role of catalyst

Therefore, both propagation reactions and oxidation reactions must be taken into account when developing the rate expression. Termination occurs mainly by combination of ROO and R ; the corresponding rate expression is: [14] rter kter[R ][ROO ]

By applying the steady-state approximation to radical compounds ROO and R , the following expression can be obtained (for low oxygen partial pressure): [15] d[O2] 144231 dt koxkpr 1313 [O ][RH] 2 kter

Under these steady-state conditions, the mass-transfer rate of molecular oxygen (OTR): [16] OTR akL([O2]* [O2])

Two types of catalysts are used in the industrial processes of autoxidation. The first type consists of catalysts that superimpose their action on that of the classical autoxidation by favouring the decomposition of OOH groups and by addressing the decomposition profile selectively towards the formation of different, more stable products. These catalysts are either transition element ions or acid/base systems. The second type includes catalysts that are able not only to decompose the OOH group, but also to directly activate the substrate. Therefore, these catalysts also participate in the initiation step. They are Co(III) in acetic acid (unique amongst the transition elements), the MIC (Amoco Mid Century) catalyst based on Co/Mn/Br ions in acetic acid, and Co(III) with a co-oxidant.
Redox catalysts active in the decomposition of hydroperoxide

is equal to the rate of oxygen consumption in the liquid phase, where a is the specific interfacial area, kL is the liquid film mass-transfer coefficient for molecular oxygen and [O2]* is the oxygen concentration at the gas-liquid inter-phase. A general rate expression that takes into account both the chemical and diffusional steps is achieved by combining the two expressions: [17] [18] k kpr 14ox 31 [O ][RH] ak ([O ]* 42 2 L 2 kter d[O2] 144231 dt koxkpr 1313 [O ][RH] 2 kter [O2])

Salts of elements of the first row of the transition series, which present one-electron redox couples such as Co(II)/Co(III), Mn(II)/Mn(III), Fe(II)/Fe(III), Cu(I)/Cu(II) (Co salts are the most commonly used), or metal chelates, which are soluble in the reaction medium, are used to accelerate the rate of decomposition of hydroperoxides through the Haber-Weiss mechanism: M(n Mn
1)

ROOH RO2 H Mn ROOH RO OH M(n 1)

The total reaction is: 2ROOH ROO H2O RO

and by assuming a negligible gradient in the gaseous

In the first reaction, the electron transfer occurs

648

ENCYCLOPAEDIA OF HYDROCARBONS

OXIDATION PROCESSES

between the hydroperoxide and the Co(III) species, through complexation of the hydroperoxide, which weakens the O O bond and facilitates the breakdown of the complex and the formation of the RO2 radical. The second reaction is a reduction by Co(II) with the formation of a RO radical. The second step is very rapid, due to the high oxidation power of hydroperoxides, while the first step has a very slow rate of reaction since hydroperoxides are weakly reducing agents. Therefore, optimal catalysts are those which contain ions with high oxidation power in order to increase the rate of the first step. In fact, Co and Mn ions, which are present in the best catalytic systems, have a very high redox potential: 1.82 eV for Co(III) e Co(II); 1.51 eV for Mn(III) e Mn(II). Only one-electron redox couples are to be used in order to avoid side reactions. The transition element must be used in an organic solvent; in order to increase the solubility of the catalyst, carboxylic acids or naftenic acid must be used as the solvent. In the oxidation of hydrocarbons, the concentration of metal ranges from 1 to 500 ppm, while in the case of aldehyde oxidations, it ranges from 0.001 to 0.1 ppm. Higher concentrations of metal ions have a negative effect, since the catalyst decomposes all of the hydroperoxide and the rate of oxidation decreases because there are not enough radicals for the initiation step. Therefore, the role of the transition element is to decrease the activation energy relative to the decomposition of the hydroperoxide, thus allowing the reaction to occur at lower temperatures and with high enough rates for industrial application. Moreover, by operating at lower temperatures, the selectivity can be better controlled. The type of products obtained in the catalytic decomposition of hydroperoxides depends mainly on their structure, but also on the type of catalyst used and the reaction conditions. For example, t-butylhydroperoxide is decomposed at 45C in the presence of Co(II) octanoate to yield 88% t-butyl alcohol and only 1% acetone (besides molecular oxygen and the dimer of the t-butoxy radical; Hiatt et al., 1968). A similar distribution of products is obtained by decomposition of cumylhydroperoxide, with the formation of the tertiary alcohol (95%) and a product of ketonic cleavage (5%). Under the same conditions, sec-butylhydroperoxide decomposes to 61% sec-butylalcohol and 36% methylethylketone, while n-butylhydroperoxide yields 67% n-butanol and 32% butyraldehyde. For all hydroperoxides, alcohol is the main product of decomposition, but other by-products form depending on the nature of the hydroperoxide. The cleavage of a C H bond in a secondary or primary alkoxy radical to yield the corresponding carbonyl compound is easier than the

cleavage of a C CH3 bond in a tertiary radical (Lyons, 1980). The C C cleavage reaction occurs when methylenic groups are near the carbon atom, as in 3-methylpentane (Lyons, 1980). An important role is also played by the transition metal in directing the decomposition of the hydroperoxide to different products. The decomposition product of hydroperoxides or peracids can be further transformed to different types of compounds, depending on the type of transition metal used.
Acid and basic catalysts active in the decomposition of the hydroperoxide

Acid can also cause the catalytic decomposition of the hydroperoxide. This is the case with phenol and acetone production by acid decomposition of cumylhydroperoxide. For this reason, an emulsion with a base is usually used as the reacting medium in the production of the hydroperoxide, in order to avoid its decomposition due to the occasional presence of acidity.
Catalysts for the direct activation of the organic substrate

Co(III) in acetic acid acts as an initiating species reacting with the hydrocarbon and forming, through a one electron exchange, a radical carbocation, which successively releases a proton forming a radical species. Co(III) is the unique element able to activate a hydrocarbon, especially benzylic hydrogens and C H bonds in aldehydes. Through this mechanism is possible to carry out an autoxidation reaction at lower temperatures, as compared to the classical ones. A second mechanism of activation by Co(III) is indirect and operates through a co-oxidant, thus in the presence of Mn and Br ions. In this type of catalytic system (MIC catalyst), the role of Co(III) is to oxidize manganese; the latter oxidizes Br to Br and is able to abstract H from less reactive C H bonds, such as methyl groups in polymethylated aromatics. The advantages of the MIC catalyst with respect to classical Co-based catalysts include the rate of reaction is higher, and therefore it is possible to use smaller amounts of catalyst or to oxidize groups that present lower reactivity, as well as many aromatic acids, and it precipitates in acetic acid, making the separation of the products easier. Disadvantages include the decarboxylation of acetic acid at high temperature and the problems tied to material corrosion, which make the use of lined apparatus necessary. An alternative synthesis consists in the use of an organic co-oxidant, as in the case of n-butane oxidation to acetic acid, where methylethylketone (obtained as a by-product of the reaction and recycled

VOLUME II / REFINING AND PETROCHEMICALS

649

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

to the oxidation reactor) is the co-oxidant. The latter is first activated by cobalt, and then the radical formed activates n-butane at low temperatures and with high selectivity. Co-based catalysts can compete with the MIC catalyst, provided that higher amounts of catalyst and a co-oxidant (either acetaldehyde or methylethylketone) are used.
Reaction conditions and reactor types

All reactions are carried out in the liquid phase at temperatures between 75 and 200C, and at a pressure ranging from 3 to 70 atm (to keep oxygen dissolved in the liquid phase). The pressure must be high enough to maintain the boiling temperature of the liquid phase higher than the reaction temperature, since high vapour concentrations in the bubble phase decrease the solubility of oxygen. When the substrate is a gas or a low-boiling liquid, high-boiling solvents are used, which in some cases can be the product itself. Acetic acid is often used as the solvent because it is not easily oxidizable, and it has both an optimal boiling temperature (116-118C) and freezing temperature, which makes it easy to handle. It is easily available, low in cost and it dissolves the majority of aromatics at low temperature. In some cases, mixtures of two solvents are used. In the oxidation of m-xylene to isophthalic acid, a mixture of acetic acid and dichlorobenzene is the solvent for the reaction. With this solvent mixture, an increase in selectivity is achieved, ranging from 89 to 91%. The reactors used can be either towers or stirred tanks. The choice essentially depends on the ratedetermining step of the reaction. Often in-series reactors are set up, in order to limit the heat of reaction evolved or to minimize the overall volume necessary for achieving a defined conversion. Sometimes, batch operation is preferred over continuous operation in stirred reactors, since backmixing phenomena can have negative effects on selectivity, favouring undesired overoxidation reactions. On the other hand, a certain extent of backmixing can be necessary to keep the correct concentration of radicals in the reaction medium. A compromize can be reached by stirred reactors in series (Prengle and Barona, 1970b). An advantage arising from the absence of diffusional limitations roots in the non-necessity of stirring inside the reactor in order to improve the gas-liquid interfacial area and hence the mass-transfer rate; therefore, less costly reactors sometimes can be used. Particular attention must be paid in the design of these reactors to achieve efficient gas-liquid contact, realized either through mechanical stirring or by

gas-spraying devices, or through rapid circulation of the reactor content by pumping, as well as efficient heat removal for strongly exothermal reactions, realized either via heat exchangers inside the reactor or by the recirculation of the reacting mixture after external refrigeration and the evaporation of the substrate, the product or the solvent. In all of the processes operating in continuous stirred apparatus or in semi-batch reactors, the oxygen diffusion is the rate-limiting step, due to the concentration of oxygen in the gas phase, which is kept low for safety reasons (at the expense of selectivity). The rate of oxygen diffusion is a function of its partial pressure in the bubble phase and of the interfacial area (i.e. of bubble size). Operation under conditions of mass-transfer resistance may be more favourable, since the rate of reaction is practically unaffected by variations in the temperature (the diffusional constant is characterized by a very low activation energy; Hobbs et al., 1972a). Instead, under conditions of chemical control, fluctuations in temperature can lead to the extinction of the reaction. Complete mass-transfer control, on the other hand, may cause a lack of oxygen in the liquid phase, favouring bimolecular reactions between radicals and the termination of the chain reaction, thus decreasing the process efficiency. In continuous columns, a stable situation consists in having a region where the rate is chemically controlled and another where the process is under mass-transfer control. This ensures stable operation, being that the overall system shows very little dependency on temperature (Hobbs et al., 1972a). It is necessary to avoid coalescence of the bubbles, which can be induced by high, turbulent flow rates and heavy agitation. The gas must be introduced through large-area devices. The reactor surface-to-volume ratio has to be minimized in order to reduce reactions of radical combination at the wall. Safety considerations include good control of heat removal to prevent thermal decomposition of the hydroperoxide and operation at low conversion, so limiting the concentration of hydroperoxide and the extent of consecutive reaction of decomposition. This can be realized by using staged reactors, with intermediate feeding of oxygen. In addition, one must ensure that in the gas phase, the composition is always above the upper flammability limit of the organic vapour by inertizing with nitrogen and with water vapour. Particular care must be taken in the start-up, when no oxygen is consumed along the reactor, and during the shutdown of the reactor.
The control of selectivity

When the hydroperoxide is the final product, the lowering of selectivity is due to its decomposition

650

ENCYCLOPAEDIA OF HYDROCARBONS

OXIDATION PROCESSES

(with formation of alcohols and ketones). Therefore, low conversions and low temperatures allow higher selectivity to be obtained. Usually, when choosing experimental conditions, a compromise between purity of the hydroperoxide and productivity (conversion) must be reached. An addition of bases can increase the selectivity by decreasing the extent of hydroperoxide decomposition, which is catalysed by acids present in the reaction medium. When the hydroperoxide is not the final product but only an intermediate, the presence of the catalyst determines the selectivity. Also in this case, the conversion can be a key factor to control the selectivity for those products which are more easily oxidizable than the substrate. This is the case with the oxidation of cyclohexane to cyclohexanol and cyclohexanone. Higher conversion and selectivity can be reached by the addition of boric acid, which interacts with cyclohexanol, forming an ester and preventing its further oxidation. The ester is finally hydrolysed to recover cyclohexanol. A change from a chemically limited rate to a mass-transfer limited rate, which may occur occasionally due to an increase in the reaction temperature, can lead to significant changes in the distribution of products. For instance, in the oxidation of methylethylketone to acetic acid, this leads to a change in the average oxidation state of the cobalt catalyst, which in the absence of dissolved oxygen is reduced, and to an increase in the formation of CO, methanol and formic acid, due to the following reactions starting from the acetyl radical under molecular oxygen starvation (Hobbs et al., 1972a): CH3 CO CH3CO CH3 OH CH3OH CH3OH 1/2O2 HCOOH

lower entrainment of organic vapours in the waste gas. Air is used for its low cost and its safe handling. Air can be advantageous when the reaction is under chemical control and the partial pressure of oxygen bears no influence on the yields. Further advantages are related to the stirring effect of nitrogen as well as the more favoured dissipation of the heat of reaction. When air is used under pressure, energy can be recuperated in a turbine before venting from pressurized nitrogen. A good example of the importance in choosing an oxidant roots in the processes for the synthesis of hydroperoxides starting from isobutane and from ethylbenzene. The corresponding hydroperoxides are used as oxidizing agents in propylene epoxidation. The main difference between the two oxidation processes is that in the case of isobutene, molecular oxygen is preferred and the process operates at high pressure, while in the oxidation of ethylbenzene, air is the oxidant of choice and the process operates at low pressure. In the latter case, the rate is chemically controlled and, therefore, it is not influenced by the partial pressure of oxygen. In isobutane oxidation, instead, the rate-limiting step is oxygen diffusion. Main industrial processes of liquid phase oxidation In this section, some of the most important industrial processes of liquid phase oxidation of organic substrates are examined, which well represent the chemistry and technology involved in this class of reactions. Moreover, the industrial technology of the synthesis of propylene oxide by oxidation is compared with the processes of propylene epoxidation under study or development, in order to exemplify the current trends and future developments in the field of liquid-phase selective oxidation.
Autoxidation processes

H2O

Molecular oxygen or air as the oxidizing agent

Molecular oxygen, oxygen-enriched air or air are used as the oxidants. The choice is dictated by both safety and economic considerations (Trifir and Cavani, 1994). Pure molecular oxygen can not be used; a small fraction of nitrogen must be present, in a concentration of higher than 3%, in order to keep the gas phase over the reacting liquid always outside the higher explosion limit of the organic vapour. The use of molecular oxygen, or oxygen-enriched air, is advantageous when the reaction is mass-transfer controlled, and thus the rate depends on the partial pressure of oxygen and on efficient mass transfer. Further advantages in the use of oxygen include the non-necessity for separation of large amounts of inert (i.e. nitrogen) before recycling, the use of smaller and less expensive apparatus and

Autoxidation processes have a great importance for the chemical industry. Despite the usual, low selectivity obtained in radical-like processes, industrial applications have widely developed for those reactions involving the use of organic substrates that possess only one site of preferential attack.
The oxidation of cumene to cumylhydroperoxide

The oxidation of cumene to cumylhydroperoxide:


O2 OOH
H H31 kcal/mol 31 kcal/mol

is carried out at 80-120C, either at 6 atm of pressure (Hercules process; Fleming et al., 1976), in an

VOLUME II / REFINING AND PETROCHEMICALS

651

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

homogeneous solution containing a sodium carbonate buffer (removed by washing with water before distillation; Hercules process), or at atmospheric pressure (Allied process; Sifniades et al., 1982), without the addition of a base (pH 3-5). In the latter case, however, the stream is washed with caustic water before recycling. Per pass cumene conversion is in the range of 25 to 35%. The reaction mechanism is a typical free-radical chain, with initiation (usually accelerated by decomposition of the hydroperoxide itself), propagation and termination steps. Main by-products include acetophenone, dimethylphenylcarbinol (a, a-dimethylbenzyl-alcohol) and a-methylstyrene. The selectivity to cumylhydroperoxide ranges from 90 to 96%. The reaction steps responsible for the formation of by-products are presented in Table 8 (chain-branching reactions). In the buffered solution (pH 7-8), the organic acids formed (i.e. formic acid), which are responsible for the in situ decomposition of the hydroperoxide with formation of phenol (inhibitor of cumene oxidation), are destroyed. The reaction is carried out in several (typically four) in-series reactors. Air is the preferred oxidizing agent, for both economic and safety considerations. Fresh air is pumped into each reactor. The lower and higher flammability limits for cumene/air mixtures are 0.8 and 8.8%, respectively. The concentration of molecular oxygen at the exit of each reactor is kept lower than 4%, in order to avoid the formation of flammable compositions with cumene (Fleming et al., 1976). The vent gas also contains organic compounds, which are condensed by refrigeration and adsorbed on active charcoal. The rate equation for cumene oxidation in the conditions which are industrially applied can be expressed as follows: 11111 d[O2] 2kin[ROOH] 144231 k [RH] 141111 [20] pr dt kter where kin, kpr and kter are the rate constants for chain initiation, propagation and termination, respectively; RH is cumene and ROOH is the hydroperoxide. The rate of reaction is therefore not dependent on oxygen concentration (Melville and Richards, 1954). However, this expression is only approximate; detailed kinetic studies indeed have demonstrated that the reaction does depend on oxygen up to a partial pressure of approximately 0.3 atm at 130C; at 100C, the rate does not depend on oxygen partial pressure for P 2 O higher than approximately 0.1 atm (Hendry and Amer, 1967; Hattori et al., 1970; Andrigo et al., 1992). Moreover, the rate expression shows that a certain concentration of the hydroperoxide is necessary to

obtain acceptable rates of cumene oxidation. The hydroperoxide contributes to initiation, decomposing into RO and OH species. In fact, long induction periods are necessary when the reaction is started without the presence of the hydroperoxide. Therefore, the hydroperoxide is in part recycled (together with unconverted cumene) to the first reactor, which operates at 8-10 wt% hydroperoxide. The maximum concentration of hydroperoxide in the last reactor is 30 wt%; higher concentrations, in fact, do not lead to further increases in the reaction rate, but rather cause an increase in the side reactions of cumylhydroperoxide decomposition.
The oxidation of cyclohexane to cyclohexanol and cyclohexanone

Cyclohexanol and cyclohexanone (the so-called ol one mixture) are produced by oxidation of cyclohexane:
2 3/2O2 OH H H O H2O

70 kcal/mol 70 kcal/mol

and the products are then used for the synthesis of adipic acid. The most accepted reaction pattern for the transformation of cyclohexane to cyclohexanol/ /cyclohexanone mixtures consists in the oxidation of the cycloalkane to cyclohexylhydroperoxide, followed by its decomposition to the alcohol and to the ketone. Cyclohexanone is also formed by the consecutive reaction of cyclohexanol oxidation. The ketone is then oxidized to adipic acid and to other by-products. The alcohol and the ketone are more reactive than cyclohexane; therefore, they are easily converted to several by-products. High selectivity to alcohol ketone can be reached only at low conversion. The selectivity is around 90% at very low cyclohexane conversion (1-2%), while already at 4-5% conversion, the selectivity drops to 77-85%. Therefore, all processes operate at low perpass cyclohexane conversion, with recycle of unconverted reactant. A detailed kinetic study has shown that the reaction rate is initially zero order in oxygen partial pressure, and then becomes first order with decreasing oxygen partial pressure in a batch reactor (Suresh et al., 1988). The main by-product is adipic acid, whose selectivity reaches high values only at low temperatures (T 100C). The other by-products, such as glutaric acid or the esters formed by condensation between the acids and cyclohexanol are derived from the decarboxylation of adipic acid. Although adipic acid is the final desired product, its formation must be

652

ENCYCLOPAEDIA OF HYDROCARBONS

OXIDATION PROCESSES

avoided at this stage, since it is accompanied by several other by-products. Different processes of oxidation of cyclohexane have been proposed, which yield either a mixture of cyclohexanone and cycloexanol, or adipic acid directly. Only those processes which produce cyclohexanol/cyclohexanone have been commercialized. Common aspects of the various processes include the very low residence times (around 40 min), necessary to limit the consecutive reactions that lead to the formation of by-products, the reaction temperature, which must be high enough (around 150C) to kinetically favour the decomposition of the intermediate hydroperoxide and the pressure, which must be such as to keep the cyclohexane in the liquid phase. A brief description of the different processes follows. Direct synthesis of cyclohexanone and cyclohexanol. This process was developed by Stamicarbon. It operates at 155C. Conversion of cyclohexane is 4%, with an overall selectivity ol one of 77-80%, and a molar ratio ol/one 1. At these conditions, the consecutive reactions upon the desired products are minimized. The catalyst is cobalt naphthenate or octeate, both of which are soluble in cyclohexane (used as the solvent for the reaction). The concentration of the catalyst is very low, ranging from 3 ppm to 300 ppm. The mixture of alcohol and ketone is then oxidized with air in the presence of Cu and Mn based catalysts, with a selectivity to adipic acid of 70%. Process in the presence of boric acid. In order to minimize the formation of by-products, boric acid is added to the reaction medium. The cyclohexanol formed reacts with boric acid, forming an ester and preserving it from the consecutive reactions to cyclohexanone and to by-products. Since the water formed in the reaction of cyclohexanol formation is also a product in the reversible reaction of cyclohexanol esterification, stripping of water from the reaction medium has been found to positively affect the selectivity to the desired product, being that it favours the formation of the borate. The addition of boric acid (H3BO3) allows operation to be carried out at relatively high conversion (12%), with an overall selectivity of 85-87% and a ol/one ratio as high as 10/1. The oxidation reaction occurs in the absence of a catalyst. The ester is then hydrolized in a following step to alcohol and the boric acid is recycled. This process has higher investment and operating costs, due to the recovery and recycle of boric acid. Process in two steps. In a first step, the hydroperoxide is synthesized in the absence of a catalyst, and in a second step, the hydroperoxide is decomposed to the alcohol and ketone in the presence

Table 8. Reactions yielding by-products in the oxidation of cumene to cumylhydroperoxide (R C6H5C(CH3)2 ) RO2 RO2 RO RO RO O2

C6H5C(O)

CH3 CH3 (acetophenone)

CH3 RH CH3 ROOH CH3 O2 CH3OO

CH4 R CH4 ROO

CH3OO RH CH3OOH R

CH3OO ROO CH2O ROH O2 (ROH dimethylphenylcarbinol) CH2O 1 2O2 ROH RO HCOOH CH2 H2O (a-methylstyrene) CH2 R H2O

C6H5C(CH3) RH

C6H5C(CH3)

of a Co catalyst. This process operates at 4-5% conversion, with a selectivity in the range of 82 to 86% and a ratio ol/one 0.4. In the presence of Cr, the ol/one ratio is 0.8. The direct synthesis of adipic acid. In this process, a large excess of cobalt catalyst is used (also with a promoter) to lower the induction time (Tanaka, 1974). The role of the promoter is to start the oxidation of Co(II) to Co(III). The solvent is acetic acid, in which the ketone is soluble. High concentrations of catalyst (feed ratio cyclohexane/catalyst 100) are necessary to obtain around 65% selectivity to adipic acid. Lower catalyst concentrations cause a decrease in selectivity, indicating that the mechanism is a classical autoxidation in the absence of the catalyst.
The oxidation of n-butane to acetic acid

The synthesis of acetic acid by n-butane or naphtha liquid phase oxidation, commercialized by Celanese in 1962, was the predominant process throughout the 1960s. n-C4H10 5/2O2 2CH3COOH H2O
DH 233 kcal/mol

By 1973, approximately 40% of the acetic acid capacity was based on this process. Currently, only a small fraction of acetic acid is produced by liquid phase oxidation, due to the superior economics of the Monsanto methanol carbonylation process. The formation of acetic acid from an alkane is explained by the fact that acetic acid is the most stable product (aside from CO2) and thus tends to accumulate in the reaction medium, which is also the reason why it is used as the solvent for the reaction. The main

VOLUME II / REFINING AND PETROCHEMICALS

653

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

reaction path for the oxidation of n-butane, accounting for 70% of converted paraffin, occurs through the abstraction of the methylenic hydrogen and the formation of the sec-butylperoxy radical; the remaining 30% is oxidized with the formation of the primary peroxy radical. The consumption of these two radicals is not as fast as the H abstraction from n-butane, and they accumulate in the absence of catalyst in the reaction medium. The peroxy radicals react with the Co(II)-based catalyst forming a more reactive secondary or primary alkoxy radical. The main reaction from the sec-butoxy radical is the b scission, which leads to acetaldehyde (precursor of acetic acid) and the ethyl radical and, to a minor extent, the scission leading to propionic acid and formic acid. Alternatively, sec-BuO is transformed to methylethylketone. The n-butoxy radical is not subjected to b scission and is transformed to n-butyric acid. The main reactions occurring are reported in Table 9 (usual passages are omitted). The ethyl radical is the precursor of ethanol. Once formed, it is quickly oxidized to acetaldehyde and acetic acid (Hobbs et al., 1972b). A second route to acetic acid is from methylethylketone, through oxidation at the b position (with respect to the C O group) and the formation of a ketoperoxyradical that produces acetic acid by C C cleavage. Two processes have been proposed: the first uses only cobalt as the catalyst, which operates at 160220C and 50-60 atm, and the second, together with cobalt, uses a co-oxidant (methylethylketone) and operates at milder conditions (120-130C, 30 atm). Methylethylketone is recycled to the first reactor in order to introduce a free-radical initiator able to abstract a hydrogen from the sec-CH bond. Performance varies considerably depending on the reaction conditions and catalyst type either Mn(III) or Co(III); the reaction may also run in the absence of catalyst. Selectivity to acetic acid can be in the range between 50 to 65%; conversion per pass is kept below 30%, to avoid over-oxidation. The air is diluted with residual inert gases up to an oxygen concentration of 8-11%, in order to keep its concentration at the exit below the explosion limits (Prengle and Barona, 1970a).
The oxidation of p-xylene to terephthalic acid or dimethylterephthalate

Table 9. Reactions occurring in n-butane oxidation n-C4H10 secBuO C2H5 sec BuOO sec BuO

C2H5 CH3CHO C2H5O CH3COOH C2H5OH

C2H5OO

CH3CHO 1 2O2 secBuO

CH3 C2H5CHO C2H5COOH C(O) C2H5

C2H5CHO 1 2O2 secBuO n-C4H10 CH3 n-BuOO

n-C3H7COOH

Witten-Dynamit process; Katzschmann, 1966). The first step is the oxidation to p-toluic acid, occurring with the classic free-radical autoxidation mechanism catalysed by Co/Mn naphthenates at 150C and 6 atm pressure, without solvent: H3C C6H4 CH3 1.5O2 H3C C6H4 COOH H2O

The electron-attracting effect of the COOH group prevents the formation of the radical on the second methyl group, and thus prevents oxidation to the diacid. Therefore, the carboxylic group is esterified by reaction with methanol (second step): H3C C6H4 COOH CH3OH H3C C6H4 C(O) OCH3 H2O

Then, the second methyl group can be oxidized (third step): H3C C6H4 C(O) OCH3 1.5O2 HOOC C6H4 C(O) OCH3 H2O The final step is the esterification: C6H4 C(O) OCH3 CH3OH H3CO (O)C C6H4 C(O) OCH3

HOOC

The oxidation/esterification of p-xylene to dimethylterephthalate:


3O2 2CH3OH H3COOC COOCH3

HH 340 340 kcal/mol kcal/mol

is industrially carried out in several steps (Hercules-

The synthesis of p-toluic acid is accompanied by the formation of by-products, such as p-toluicalcohol, p-tolualdehyde (these two compounds are intermediates in the pathway leading to p-toluic acid), toluene, terephthalic dialdehyde, 4-carboxybenzaldehyde and polynuclear products (i.e. p-toluylester of p-toluic acid and p-toluic acid anhydride). 4-carboxybenzaldehyde is highly undesired, since the CHO group can not undergo condensation with ethylene glycol during PET polymerization. The selectivity to by-products (especially p-toluylester of p-toluic acid and p-toluic acid anhydride) increases as the partial pressure of oxygen

654

ENCYCLOPAEDIA OF HYDROCARBONS

OXIDATION PROCESSES

decreases (Jacobi and Baerns, 1983). Therefore, the complete depletion of oxygen in the bulk liquid is undesired for selectivity. The kinetics of liquid-phase oxidation of p-xylene to p-toluic acid has been studied by several authors (Cavalieri dOro et al., 1980; Hronec and Ilavsky, 1982; Jacobi and Baerns, 1983; Hronec et al., 1985; Raghavendrachar and Ramachandran, 1992; Cao et al., 1994). When the reaction is carried out in bubble columns, regions where the process is controlled by different steps can occur. Typically, at high oxygen partial pressure, the process is under chemical control (and the rate does not depend on oxygen partial pressure), while at low oxygen partial pressure (for almost total oxygen consumption), the process becomes mass-transfer controlled. In addition, temperature affects the nature of the rate-determining step; at temperatures below 100C, the reaction is under chemical control and is zero order when either air or oxygen is used. Instead, at 130C, the process is under mass-transfer control when air is used (the apparent activation energy falls from 15 to 4 kcal/mol), while with pure oxygen, the reaction is still under chemical control (Cao et al., 1994; Cincotti et al., 1997). Under chemical control, the rate equation for the Co-catalysed first step of the process (i.e. oxidation of p-xylene to p-toluic acid) can be expressed as follows (Jacobi and Baerns, 1983): [21] d[p-C8H10] 144233121 dt k[ p-C8H10]2

With respect to the hydrocarbon, a second order reaction arises by combining the rate propagation expression (Prengle and Barona, 1970b): [22] d[p-C8H10] 144233121 dt kpr [RO2 ][ RH]

consumed via attack by RO2 . An experimental value of n/f intermediate between the two limit cases (1.5 and 6) was found, indicating that the cobalt catalyst not only was involved in the ROOH decomposition, but also reacted with p-xylene through the Co(III) species (Jacobi and Baerns, 1983). Other models have been developed that also take into account diffusional constraints and that assume a first order dependency of rate upon p-xylene concentration (Hronec and Ilavsky, 1982; Cao et al., 1994). In all cases, the reaction is independent of oxygen partial pressure (in conditions of a reaction-controlled process). An alternative process is one developed by Amoco (and by several other companies), which makes use of a catalyst (Mid-Century MIC catalyst: Co/Mn/Br ions) in acetic acid as the reaction solvent. Several variations of this process exist. The MIC catalyst is able to oxidize the second methyl group of p-toluic acid and, therefore, terephthalic acid is directly obtained as the final product. The acetic acid keeps the intermediates and by-products in solution, but does not dissolve the products. The temperature is around 205C and the pressure is 12-15 atm. The conversion is almost total. The concentration of oxygen in the waste gas is around 4%. The advantage of this process lies in the direct availability of terephthalic acid, which can be further employed for several different transformations. This process is now used to virtually cover the world production of fibre grade terephthalic acid. The Teijin process is an improvement of the original Mid-Century process; with respect to the latter, the conditions are milder (100-130C, 9 atm); a high yield of terephthalic acid (higher than 97%) is achieved by using a cobalt catalyst and a large excess of p-xylene.
Oxidations with oxygen-transfer agents

where R is (CH3)C6H4(CH2) (with the assumption that the corresponding step is the rate-limiting reaction), with the termination reaction expression: [23] d[p-C8H10] 144233121 dt kter [RO2 ]

The reactions of oxidation via catalysed oxygen transfer agents can be represented schematically as follows: S ROY SO RY

(assuming that the termination reaction occurs mainly by bimolecular reaction between RO2 fragments), and by deriving an expression for [RO2 ]. It is also assumed that the concentration of hydroperoxide is quasistationary (Walling, 1969). The kinetic constant k is related to kpr and kter through the following expression: [24]
2 k pr k n 121 2fkter

where n is the number of radicals produced for each ROOH decomposed, and f is the fraction of RH

where S is the substrate and ROY the oxygen transfer agent (hydroperoxides, H2O2 and peracetic acid). The hydroperoxides of isobutane, ethylbenzene and cumene are the best oxygen transfer agents for industrial applications for the following reasons: they are safer to handle and more stable, soluble in non-polar solvents, not acidic, present high selectivity in the reaction of oxygen transfer to the substrate, and the by-products of the reaction are easy to separate. Two main industrial applications can be evidenced: epoxidations of olefins and hydroxylation of aromatic

VOLUME II / REFINING AND PETROCHEMICALS

655

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

compounds. The former process, which is by far the more important, is industrially carried out in either homogeneous or heterogeneous catalysis. The hydroxylation of aromatic compounds, on the other hand, is carried out in heterogeneous catalysis.
The epoxidation of propylene to propylene oxide

All olefins can be epoxidized with a hydroperoxide and a catalyst (see Section 11.1.3), but the most important process, which has been developed at an industrial level, is the synthesis of propylene oxide. Two alternative processes have been developed, one in the homogeneous phase, developed by Halcon and Atlantic Richfield (Oxirane process), with a molybdenum complex as the catalyst (Landau, 1967; Landau et al., 1979), and the other in the heterogeneous phase, with silica-supported titanium oxide, developed by Shell (Wulff, 1975). Alkylhydroperoxides used as oxygen-transfer agents are either t-butylhydroperoxide or ethylbenzene hydroperoxide, due to the fact that the corresponding coproduct is valuable and contributes to the economy of the process. t-butylhydroperoxide is converted to t-butylalcohol, and then to isobutene. Ethylbenzene hydroperoxide is decomposed to phenylcarbinol and the latter is converted to styrene. Cumylhydroperoxide, which could present advantages due its high rate of epoxidation and higher selectivity, is not used as an oxygen-transfer agent due to the non-practical use of the by-product. All processes, both homogeneous and heterogeneous, operate at temperatures between 90 and 120C and at a pressure of 35 atm, in order to keep propylene in the liquid phase. The upper temperature limit is the temperature of hydroperoxide decomposition and the lower limit is dictated by the activity of the catalyst. The catalysts for epoxidation reactions are d 0 metal complexes: Mo(VI), W(VI), V(V) and Ti(IV) for homogeneous epoxidation (Mimoun, 1980; Jrgensen, 1989), and TiO2 supported over SiO2 for heterogeneous epoxidation (Sheldon, 1980). The reasons for the superior activity and selectivity of these metal ions can be summarized as follows: they have a very low oxidation power and, therefore, do not decompose the hydroperoxide homolitically; they are in their high oxidation state and have high Lewis-type acidity, necessary to activate the hydroperoxide towards nucleophilic attack, through withdrawal of electrons from the O O bond. Indeed, V(V) complexes are worse than those of Mo(VI), because they are more active in hydroperoxide decomposition (and thus the selectivity is lower), while Cr(VI) complexes, which have high Lewis acidity, cannot be used due to their high oxidizing power. The mechanism

proposed for homogeneous epoxidation has been described above: Liquidi phase homogeneous metal-catalysed oxidations by molecular oxygen. The industrial process consists of three main parts: synthesis of ethylbenzene hydroperoxide; epoxidation of propylene; dehydratation of phenylcarbinol to styrene. The synthesis of the hydroperoxide is carried out in a bubble reactor with air as the oxidizing agent. After the removal of volatile compounds, the exit solution is sent to a falling film evaporator for concentration. The hydroperoxide-containing solution is then fed together with fresh propylene to the first reactor of three in-series reactors. The final exit solution, after separation of unconverted propylene (recycled to the first reactor), is sent to the purification section.
New technologies of epoxidation under study or development

The epoxidation of propylene can also be carried out catalysed by Ti-silicalite (TS-1), with H2O2 as the oxidizing agent (Neri et al., 1984). The process is not yet commercially applied, though pilot plant operation is operative. The catalyst was developed by Eni researchers and is made of a zeolite isostructural with silicalite-1, where titanium substitutes for silicon in the framework of the zeolite (Taramasso et al., 1983). The solvent for the reaction is an alcohol (methanol or t-butanol). At 50C and atmospheric pressure, the yield of epoxide on an H2O2 basis is around 90%, and the selectivity on a propylene basis is 98% (Sheldon, 1973; Chong and Sharpless, 1977). The main by-products observed for the different olefins are only those derived from the consecutive transformation of the oxirane ring by reaction with methanol (yielding glycol ethers), with water (yielding glycols), or with H2O2 (yielding hydroperoxides). Several Olin patents claim the synthesis of propylene oxide from propylene using molecular oxygen as the oxidizing agent in the presence of molten salts (Meyer and Pennington, 1991; Fullington and Pennington, 1992). Although it has been claimed that the salt acts as a catalyst, it is likely that the oxidation is not catalytic and that the salt probably serves to remove the heat of reaction, allowing better control of the temperature, and thus allowing the very exothermal reaction to run more smoothly and more selectively. Thus the mechanism of reaction is a classical radical chain reaction, and the molten salt may possibly serve to facilitate the initiation step. The feedstock contains propylene at a concentration of around 50%, with a deficiency of oxygen of approximately 10%, and the remainder consists of inert gas (N2 and carbon oxides). The reaction proceeds at 200-300C, 5-20 atm pressure and

656

ENCYCLOPAEDIA OF HYDROCARBONS

OXIDATION PROCESSES

at relatively high residence time (higher than 10 s). Gas reactants may be either bubbled through a stirredtank reactor filled with the molten salt, moved countercurrently in a tower through a downflow of molten salt or reinjected into a loop through which the salt circulates. The molten salt is a mixture of LiNO3, NaNO3 and KNO3. The composition of the mixture determines the temperature of the reaction. Propylene conversion is around 10%, with a selectivity to epoxide in the 55 to 65% range. By-products include carbon oxides (selectivity 22-25%) and acetaldehyde (10-13%); minor amounts of formaldehyde, methanol, acrolein, acetone, allyl alcohol and propylene glycol are also formed. Unreacted propylene is recycled together with part of the by-product gases. The recycle of acetaldehyde improves the yield to propylene oxide, since acetaldehyde is oxidized to peracetic acid, which epoxidizes propylene. Alternatively, acetaldehyde is oxidized to acetic acid in a second reactor in a two-step process. Another technology under development is the in situ generation of H2O2, which finally acts as the oxygen-transfer agent on propylene (Clerici and Ingallina, 1998). This process, proposed by EniTechnologie (Clerici and Ingallina, 1993a), epoxidizes olefins with molecular oxygen in the presence of Ti-silicalite and a redox couple generating in situ H2O2 and is discussed in Section 11.1.3.
Redox processes The oxidation of ethylene to acetaldehyde

reaction between catalyst and ethylene (at 110C and 9-10 atm), with the formation of acetaldehyde; the second step is the reoxidation of the reduced catalyst through contact with air. In the single-step process, ethylene and molecular oxygen are fed into an air-lift tower, where the catalyst-containing aqueous solution is circulated via a separation vessel. After cooling and washing with water of the gaseous stream, the liquid fraction, containing water and crude acetaldehyde, is sent to a twocolumn separation, while the gaseous stream, containing unreacted ethylene and inerts, is recirculated to the reactor. The ethylene per pass conversion is between 20 and 50%; unconverted ethylene is recycled. In the two-step process, following the reaction, an acetaldehyde/water mixture is separated from the solution containing the reduced catalyst in a flash tower. The latter is sent to the reoxidation reactor, where it is put in contact with air. Ethylene and molecular oxygen almost completely react in the two steps, respectively. The crude acetaldehyde/water mixture is first concentrated and then sent to a two-stage distillation section. In this process configuration, the ethylene conversion is up to 90%, which eliminates the need for recycle of the unconverted olefin. In both process configurations, the yield of acetaldehyde is around 95%.

Bibliography
Bailey C.L., Drago R.S. (1987) Utilization of dioxygen for the specific oxidation of organic substrates with cobalt (II) catalysts, Coordination Chemistry Reviews, 79, 321-332. Balci M. (1981) Bicyclic endoperoxides and synthetic applications, Chemical Reviews, 81, 91-108. Benson S.W., Nangia P.S. (1979) Liquid-phase oxidation of isobutane. A reanalysis of the data, International Journal of Chemical Kinetics, 12, 169-181. Betts A.T., Uri N. (1968) Catalyst-inhibitor conversion in autoxidation reactions, Advances in Chemistry Series, 76, 160-181. Burstyn J.N. et al. (1988) Mechanisms of dioxygen activation in metal-containing monooxygenase. Enzimes and model systems, in: Martell A.E., Sawyer D.T. (editors) Oxygen complexes and oxygen activation by transition metals, New York, Plenum Press, 175-187. Choy V.J., OCannor C.J. (1972) Chelating dioxygen compounds of the platinum metals, Coordination Chemistry Reviews, 9, 145-170. Collman P. et al. (1980) Oxigen binding to heme proteins and their synthetic analogs, in: Spiro T.G. (editor) Metal ions in biology, 2: Metal ion activation of dioxygen, New York, John Wiley, 1-72. Dadyburjor D.B. et al. (1979) Selective oxidation of hydrocarbons on composite oxides, Catalysis Reviews. Science and Engineering, 19, 293-350.

Currently, acetaldehyde is mainly obtained through the process developed by Wacker Chemie and Hoechst: C2H4 1/2O2 CH3CHO
DH 58 kcal/mol

The catalyst is made of an aqueous solution of PdCl2 and CuCl2. Acetaldehyde is formed by reaction between ethylene and Pd chloride (i.e. the rate-determining step), which, at the same time, is reduced to metallic Pd. Pd is then reoxidized to PdCl2 by CuCl2, with the formation of CuCl, which is finally reoxidized by molecular oxygen, yielding back CuCl2. (The reaction mechanism was described above: Liquid phase homogeneous metal-catalysed oxidations by molecular oxygen). By-products of the reaction are acetic acid, oxalic acid, crotonaldehyde, chlorinated hydrocarbons and chlorinated acetaldehyde. The process can be carried out either in a single step, with pure oxygen as the oxidizing agent, at 130C and 4 atm, or in a two-step process, where the oxidizing agent of choice is air. In the latter case, the first step consists in the

VOLUME II / REFINING AND PETROCHEMICALS

657

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

Fox M.A., Chanon M. (editors) (1988) Photoinduced electron transfer. Part A: Conceptual basis; Part B: Experimental techniques and medium effects, Amsterdam, Elsevier. Gubelmann M.H., Williams A.F. (1983) The structures and reactivity of dioxygen complexes of the transition metals, in: Structure and bonding, 55: Transition metal complexesStructure spectra, Berlin, Springer, 1-65. Kaufman S., Fisher D.B. (1974) Pterin-requiring aromatic amino acid hydroxilases, in: Hayaishi O. (editor) Molecular mechanisms of oxygen activation, New York, Academic Press, 285-369. Meyer T.J. (1988) Metal oxo complexes and oxygen activation, in: Martell A.E., Sawyer D.T. (edited by) Oxygen complexes and oxygen activation by transition metals, New York, Plenum Press, 33-47. Murray R.W., Wasserman H.E. (edited by) (1979) Organic chemistry, 40: Singlet oxygen, New York, Academic Press. Patai S. (edited by) (1983) The chemistry of peroxide, Chichester, John Wiley. Sheldon R.A. (1991) Fine chemicals by catalytic-oxidation, CHEMTECH, 21, 566-576. Sheldon R.A. (1991) Heterogeneous catalytic oxidation and fine chemicals, in: Guisnet M. et al. (editors) Studies in surface science and catalysis, 59: Heterogeneous catalysis and fine chemicals II. Proceedings of the 2nd international symposium, Poitiers, 2-5 October, Amsterdam, Elsevier, 33-54. Simandi L.I. (editor) (1991) Dioxygen activation and homogeneous catalytic oxidation. Proceedings of the 4th international symposium on dioxygen activation and homogeneous catalytic oxidation, Amsterdam, Elsevier. Stevens B. (1973) Kinetics of photoperoxidation in solution, Accounts of Chemical Research, 6, 90-96. Tsutsui M., Ugo R. (editors) (1977) Fundamental research in homogeneous catalysis, New York, Plenum Press. Vaska L. (1976) Dioxygen-metal complexes. Toward a unified view, Accounts of Chemical Research, 9, 175-183.

References
Andrigo P. et al. (1992) Phenol-acetone process. Cumene oxidation kinetics and industrial plant simulation, Chemical Engineering Science, 47, 2511-2516. Astarita G. et al. (1983) Gas treating with chemical solvents, New York, John Wiley. Bamford C.H., Tipper C.F.H. (edited by) (1980) Comprehensive chemical kinetics, Amsterdam, Elsevier. Barton D.H.R., Doller D. (1991) Selective functionalization of saturated hydrocarbons. Gif and all that, in: Simandi L.I. (edited by) Dioxygen activation and homogeneous catalytic oxidation. Proceedings of the 4th international symposium on dioxygen activation and homogeneous catalytic oxidation, Amsterdam, Elsevier, 1-10. Barton D.H.R. et al. (1989) On the mechanism of the Gif and Gif-Orsay systems for the selective substitution of saturated hydrocarbons, New Journal of Chemistry, 13, 177-182. Bellussi G. et al. (1992) Reactions of titanium silicalite with

protic molecules and hydrogen peroxide, Journal of Catalysis, 133, 220-230. Benson S.W. (1965) Effects of resonance and structure on the thermochemistry of organic peroxy radicals and the kinetics of combustion reactions, Journal of the American Chemical Society, 87, 972-979. Benson S.W., Shaw R. (1970) Thermochemistry of organic peroxides, hydroperoxides, polyoxides, and their radicals, in: Swern D. (edited by) Organic peroxides, New York, John Wiley, 105-139. Cao G. et al. (1994) Kinetics of p-xylene liquid-phase catalytic oxidation, American Institute of Chemical Engineers Journal, 40, 1156-1166. Carr S., Santacesaria E. (1980) Engineering aspects of gas-liquid catalytic reactions, Catalysis Reviews-Science and Engineering, 22, 75-140. Cavalieri dOro P. et al. (1980) On the low temperature oxidation of p-xylene, Oxidation Communications, 1, 153-162. Chen Z. et al. (1985) Gas-liquid interphase mass transfer coefficients in trickle beds, Huagong Xuebao, 3, 339-346. Chester A.W. et al. (1977) Zirconium cocatalysis of the cobaltcatalyzed autoxidation of alkylaromatic hydrocarbons, Journal of Catalysis, 46, 308-319. Chong A.O., Sharpless K.B. (1977) Mechanism of the molybdenum and vanadium catalyzed epoxidation of olefins by alkyl hydroperoxides, Journal of Organic Chemistry, 42, 1587-1590. Cincotti A. et al. (1997) Effect of catalyst concentration and simulation of precipitation processes on liquid-phase catalytic oxidation of p-xylene to terephthalic acid, Chemical Engineering Science, 52, 4205-4213. Clerici M.G. (1991) Oxidation of saturated hydrocarbons with hydrogen peroxide, catalyzed by titanium silicalite, Applied Catalysis, 68, 249-261. Clerici M.G. et al. (1991) Synthesis of propylene oxide from propylene and hydrogen peroxide catalyzed by titanium silicalite, Journal of Catalysis, 129, 159-167. Clerici M.G., Ingallina P. (1993a) European Patent 0549013 to Eniricerche. Clerici M.G., Ingallina P. (1993b) Epoxidation of lower olefins with hydrogen peroxide and titanium silicalite, Journal of Catalysis, 140, 71-83. Clerici M.G., Ingallina P. (1998) Oxidation reactions with in situ generated oxidants. New concepts in selective oxidation over heterogeneous catalysts, Catalysis Today, 41, 351-364. Dawson J.H. (1988) Probing structure-function relations in heme-containing oxygenases and peroxidases, Science, 240, 433-439. Doraiswamy L.K., Sharma M.M. (1984) Heterogeneous reactions. Analysis, examples and reactor design, New York, John Wiley, 2v. Drago R.S., Beer R.H. (1992) A classification scheme for homogeneous metal-catalyzed oxidations by molecular oxygen, Inorganica Chimica Acta, 198, 359-367. Ellis P.E., Lyons J.E. (1989a) Effect of axial azide on the selective, low temperature metalloporphyrin-catalyzed reactions of isobutane with molecular oxygen, Journal of the Chemical Society, Chemical Communications, 16, 1187-1189.

658

ENCYCLOPAEDIA OF HYDROCARBONS

OXIDATION PROCESSES

Ellis P.E., Lyons J.E. (1989b) Effect of fluorination of the meso-phenyl groups on selective tetraphenylporphyrinato metal(III)-catalyzed reactions of isobutane with molecular oxygen, Journal of the Chemical Society, Chemical Communications, 16, 1189-1190. Emanuel N.M., Gal D. (1986) Modelling of oxidation processes, Budapest, Akademiai Kiado. Finn M.G., Sharpless K.B. (1985) On the mechanism of asymmetric epoxidation with titanium-tartrate catalysts, Asymmetric Synthesis, 5, 247-308. Fleming J.B. et al. (1976) Safety in phenol-from-cumene process, Hydrocarbon Processing, 55, 185-190. Franz G., Sheldon R.A. (1991) Oxidation, in: Ullmanns encyclopedia of industrial chemistry, Weinheim, VCH, 1985-1996, 37 v.; v. A18, 261-270. Fullington M.C., Pennington B.T. (1992) WO Patent WO 9209588 to Olin Co. Gates B.C. et al. (1979) Chemistry of catalytic processes, New York, McGraw-Hill. Groves J.T. (1985) Key elements of the chemistry of cytochrome-p-450. The oxygen rebound mechanism, Journal of Chemical Education, 62, 928-931. Hattori K. et al. (1970) Kinetics of liquid phase oxidation of cumene in bubble column, Journal of Chemical Engineering of Japan, 3, 72-78. Hayaishi O. (1974) General properties and biological functions of oxygenases, in: Hayaishi O. (edited by) Molecular mechanisms of oxygen activation, New York, Academic Press, 7. Hendry D.G., Amer J. (1967) Rate constants for oxidation of cumene, Journal of the American Chemical Society, 89, 5433-5438. Hiatt R. et al. (1968) Homolytic decompositions of hydroperoxides. IV: Metal-catalyzed decompositions, Journal of Organic Chemistry, 33, 1430-1435. Hill C.L. (editor) (1989) Activation and functionalization of alkanes, New York, John Wiley, 243. Hobbs C.C. et al. (1972a) Mass-transfer rate-limitation effects in liquid-phase oxidation, Industrial & Engineering Chemistry. Product Research and Development, 11, 220-225. Hobbs C.C. et al. (1972b) Product sequences in liquid-phase oxidation of paraffins, Industrial & Engineering Chemistry. Process Design and Development, 11, 59-68. Howard J.A. (1972) Absolute rate constants for reactions of oxyl radicals, Advances in Free-Radical Chemistry, 4, 49-173. Howard J.A. (1973) Homogeneous liquid-phase autoxidations, in: Kochi J.K. (editor) Free radicals, New York, John Wiley, 2 v.; v. II, 3-62. Hronec M., Ilavsky J. (1982) Oxidation of polyalkylaromatic hydrocarbons. XII: Technological aspects of p-xylene oxidation to terephthalic acid in water, Industrial & Engineering Chemistry. Process Design and Development, 21, 455-460. Hronec M. et al. (1985) Kinetics and mechanism of cobaltcatalyzed oxidation of p-xylene in the presence of water, Industrial & Engineering Chemistry. Process Design and Development, 24, 787-794. Jacobi R., Baerns M. (1983) The effect of oxygen transfer limitation at the gas-liquid interphase. Kinetics and product

distribution of the p-xylene oxidation, Erdoel Kohle Erdgas Petrochemie, 36, 322-326. Jrgensen K.A. (1989) Transition-metal-catalyzed epoxidations, Chemical Reviews, 89, 431-458. Katzschmann E. (1966) Oxidation of alkyl aromatic compounds, Chemie Ingenieur Technik, 38, 1-10. Kerr J.A. (1966) Bond dissociation energies by kinetic methods, Chemical Review, 66, 465-500. Kochi J.K. (1973) Oxidation-reduction reactions of free radicals and metal complexes, in: Kochi J.K. (editor) Free radicals, New York, John Wiley, 2 v.; v. I. Krzysztoforski A. et al. (1986) Industrial contribution to the reaction engineering of cyclohexane oxidation, Industrial and Engineering Chemistry. Process Design and Development, 25, 894-898. Ladhaboy M.E., Sharma M.M. (1969) Absorption of oxygen by n-butyraldehyde, Journal of Applied Chemistry, 19, 267-272. Landau R. et al. (1967) Epoxidation of olefins, in: Proceedings of the 7th World petroleum congress, Mexico City, April, 67-72. Landau R. et al. (1979) Propylene oxide by the co-product processes3, CHEMTECH, 9, 602-607. Lyons J.E. (1980) Up petrochemical value by liquid phase catalytic oxidation, Hydrocarbon Processing, 59, 107-119. Manor Y., Schmitz R.A. (1984) Gradientless reactor for gasliquid reactions, Industrial & Engineering Chemistry Fundamentals, 23, 243-252. Mansuy D., Battioni P. (1989) Catalytically active metalloporphyrin models for cytochrome P-450, in: Ruckpaul K., Rein H. (edited by) Basis and mechanisms of regulation of cytochrome P450, London, Taylor & Francis. Melville H.W., Richards S. (1954) Photochemical autoxidation of isopropylbenzene, Journal of the Chemical Society, 944-952. Meyer J.L., Pennington B.T. (1991) US Patent 4992567 to Olin Co. Mimoun H. (1980) The role of peroxymetalation in selective oxidative processes, Journal of Molecular Catalysis, 1, 1-29. Mimoun H. (1982) Oxygen-transfer from inorganic and organic peroxides to organic substrates. A common mechanism, Angewandte Chemie-International Edition in English, 21, 734-750. Mimoun H. et al. (1970) Epoxidation of olefins with covalent peroxomolybdenum(VI) complexes, Tetrahedron, 26, 37-50. Moiseev I.I. et al. (1974) Kinetics of olefin oxidation by tetrachloropalladate in aqueous solution, Journal of the American Chemical Society, 96, 1003-1007. Morbidelli M. et al. (1986) Gas-liquid autoxidation reactors, Chemical Engineering Science, 41, 2299-2307. Neri C. et al. (1984) European Patent 0100119 to ANIC. Ortiz de Montellano P.R. (edited by) (1985) Cytochrome P450. Structure, mechanism and biochemistry, New York, Plenum Press. Prengle H.W., Barona N. (1970a) Petrochemicals by liquid phase oxidation, Hydrocarbon Processing, 49, 106118.

VOLUME II / REFINING AND PETROCHEMICALS

659

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

Prengle H.W., Barona N. (1970b) Petrochemicals by liquid phase oxidation. 2: Kinetics, mass transfer, and reactor design, Hydrocarbon Processing, 49, 159-175. Purcell K. (1985) Cycloreversion in metal-assisted olefin oxidation by peroxide. Molybdenum(VI) vs. rhodium(III), Organometallic, 4, 509-514. Raghavendrachar P., Ramachandran S. (1992) Liquidphase catalytic oxidation of p-xylene, Industrial & Engineering Chemistry Research, 31, 453-462. Santacesaria E., Pimpinelli N. (1986) Oxidation in the gasliquid phase in industry. I: Kinetics and mechanical aspects, La Chimica e L Industria, 68, 69-79. Sheldon R.A. (1973) Molybdenum-catalyzed epoxidation of olefins with alkyl hydroperoxides. I: Kinetic and product studies, Recueil des Travaux Chimiques des Pays-Bas, 92, 253-266. Sheldon R.A. (1980) Synthetic and mechanistic aspects of metal-catalyzed epoxidations with hydroperoxides, Journal of Molecular Catalysis, 7, 107-126. Sheldon R.A., Doom J.A. van (1973) Metal-catalyzed epoxidation of olefins with organic hydroperoxides. I: Comparison of various metal catalysts, Journal of Catalysis, 31, 427. Sheldon R.A., Kochi J.K. (1976) Metal-catalyzed oxidations of organic compounds in the liquid phase. A mechanistic approach, Advances in Catalysis, 25, 272-413. Sheldon R.A., Kochi J.K. (1981) Metal-catalyzed oxidations of organic compounds, New York, Academic Press. Sheu C. et al. (1990) Activation of dioxygen by bis(2,6dicarboxylatopyridine)iron(II) for the ketonization of methylenic carbons and the dioxygenation of acetylenes, aryl olefins, and catechols. Reaction mimics for dioxygenase

proteins, Journal of the American Chemical Society, 112, 879-881. Shilov A.E. (1989) Historical evolution of homogeneous alkane activation systems, in: Hill C.L. (edited by) Activation and functionalization of alkanes, New York, John Wiley. Sifniades S. et al. (1982) US Patent 4358618 to Allied Corp. Suresh A.K. et al. (1988) Mass transfer and solubility in autocatalytic oxidation of cyclohexane, American Institute of Chemical Engineers Journal, 34, 55-68. Tanaka K. (1974) Adipic acid in one-step, CHEMTECH, 4, 555-559. Taramasso M. et al. (1983) US Patent 4410501 to Snamprogetti. Trifir F., Cavani F. (1994) Selective partial oxidation of hydrocarbons and related oxidations, Catalytica Studies Division, 4193 SO. Twigg G.H. (1962) Liquid-phase oxidation [of organic compounds] by molecular oxygen, Chemistry and Industry, 1, 4-11. Uri N. (1961) Physico-chemical aspects of autoxidation, in: Lundberg W.O. (editor) Autoxidation and antioxidants, New York, John Wiley, 55-106. Walling C. (1969) Limiting rates of hydrocarbon autoxidations, Journal of the American Chemical Society, 91, 7590-7594. Wulff H.P. (1975) US Patent 3923843 to Shell Oil.

Philippe Arpentinier
Air Liquide - Centre de Recherche Claude-Delorme Jouy-en-Josas, France

660

ENCYCLOPAEDIA OF HYDROCARBONS

11.1.3 Oxidation processes with hydrogen peroxide and hydroperoxides


Introduction
Oxidants in industrial chemical processes

Oxidation reactions are of particular importance in petrochemicals because, while the traditional raw materials are hydrocarbons formed solely by carbon and hydrogen, the majority of the derivatives produced from them consist of carbon, hydrogen and oxygen. From both an economic and environmental standpoint, theoretically, the most preferable way by far to oxidize any substrate should be its reaction with molecular oxygen. In practice, however, this reaction presents certain negative aspects. In fact, although oxygen is a formidable oxidant from the thermodynamic point of view, it is not very reactive. This scanty reactivity (enabling all living organisms to remain alive for a reasonable time, instead of burning more or less slowly, forming water and CO2) is due to the fact that the ground state of the O2 molecule is a triplet state (with two unpaired electrons). On the contrary, the ground states of the O2 reduction products (i.e. water or hydrogen peroxide), as in the majority of the stable organic compounds obtained by oxidation, are all singlet states (with paired electrons). As in a chemical reaction the angular moment must be conserved, the reactions of molecules in a triplet state with products in a singlet state are spin-prohibited and can occur only with extremely slow kinetics. The obstacle can be overcome by supplying energy to the system, that is by conducting the reaction at high temperatures and pressures or by operating in conditions that favour the onset of radical type mechanisms (compatible with the triplet state of the molecular oxygen). Both solutions, however, are generally adverse to achieving selective oxidations and hence to a certain part of oxidations of industrial interest. As an alternative to molecular oxygen, the oxidation of an organic substrate S can be obtained by causing it to react with an appropriate oxidant XO, whose basic state is a singlet one: S XO SO X

The reaction shows, though, that together with the desired product SO a second product (coproduct) X is also formed, derived from the reduction of the oxidant. At the end of the reaction, the coproduct has to be recovered, recycled or upgraded, or in the worst case disposed of. Different oxidants form different coproducts (Table 1), however, in every case, the formation of a coproduct increases the costs of the process. In particular, its mere disposal can have an environmental impact, which, in the light of the growing sensitivity to problems of pollution, is often unacceptable. For example, the production of propylene oxide by the chlorohydrin process (still in use) entails that, for every ton of epoxide, about 2.1 t of calcium chloride is coproduced and must be disposed of; furthermore, it may be polluted by chlorinated organic by-products. Therefore, from the environmental standpoint, only a few oxidizing agents (hydrogen peroxide, ozone, nitrous oxide) are acceptable, as they lead to the formation of harmless, easily disposable coproducts (water, oxygen, nitrogen).
Hydrogen peroxide

Of the oxidizing agents presented in Table 1, the most versatile and easiest to use by far is hydrogen peroxide, whose molecule has a non-planar structure, with a C2 symmetry (Fig. 1). It should be stressed that the bond between the two oxygen atoms is relatively weak (51 kcal/mol, to be compared, for example, with 89.5 kcal/mol for the two O H bonds).

Table 1. Different coproducts formed by reduction

of selected oxidants
Oxidant H2O2 N2O O3 Hydroperoxides (ROOH) RCO3H NaClO Coproduct H2O N2 O2 Alcohols (ROH) RCO2H NaCl

VOLUME II / REFINING AND PETROCHEMICALS

661

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

H 9652' 9351' H O 0.97 1.49 O

Fig. 1. The molecule of hydrogen peroxide.

Hydrogen peroxide is commonly produced by the anthraquinone process, originally developed in Germany by IG Farbenindustrie between 1935 and 1945, and used on an industrial scale as of the 1950s. This process is based on the capacity of the alkyl derivatives of anthrahydroquinone to produce hydrogen peroxide under typical autoxidation conditions:
OH R O2 OH O R H2 O OH O OH R O R H2O2

Other organic compounds also possess this property, such as secondary alcohols, hydroquinone derivatives and azobenzenes, which has been exploited for the development of alternative production processes. Mention should be made, in this regard, to the autoxidation of isopropyl alcohol, the Shell process, and of the 1-phenylethanol recently developed by ARCO up to the pilot plant stage. Today the only process truly applied for the production of large volumes is the anthraquinone process. An electrochemical process also exists, which is used for the production of small quantities at local scale. Typically, the anthraquinone process is based on the cyclic hydrogenation and oxidation of a working solution consisting of an alkylanthraquinone and various derivatives thereof, dissolved in a solvent mix. The recovery of the hydrogen peroxide includes an extraction with water, a series of purification treatments of the aqueous solution, removing the organic impurities, and the vacuum distillation of the excess water. The commercial solution thus obtained generally has a concentration of 60-70% in weight. The process is characterized by various critical factors, linked to both the reactivity and the different solubilities of the oxidized and reduced forms of the alkylanthraquinone. In fact, during the hydrogenation

and oxidation cycle, this can undergo various degradation processes with the formation of various by-products, bringing about complex regeneration and purification processes of the working solution (apart from the partial loss of a relatively costly reagent). The solvent should be able to dissolve both the (apolar) anthraquinone and the (polar) anthrahydroquinone, and, in addition, must guarantee immiscibility with water so as to permit the extraction and recovery of the hydrogen peroxide. These problems are tackled, in part, by selecting the proper alkyl substituent of the anthraquinone (generally ethyl, amyl or tert-butyl) and by using a mixture of solvents, chosen from two groups characterized by their different polarity. The first group, suitable for dissolving alkylanthraquinone, includes a number of aromatic compounds (toluene, methylnaphthalene). The second group, suitable for alkylanthrahydroquinone, includes the organic esters of phosphoric acid, diisobutylcarbinol and alkylsubstituted ureas. Typically, the composition of the working solution varies from one producer to another; it is optimized to maximize the solubility of alkylanthraquinone and hence the concentration of the hydrogen peroxide produced, to guarantee high rates of hydrogenation and oxidation, as well as to minimize the formation of degradation by-products. The other characteristics of the plant are generally the result of specific choices and developments. The many problems to be addressed are reflected in the process complex nature and high investment costs. In fact, the anthraquinone process is marketed throughout the world by a relatively small number of companies. The average capacity of the plants is 40,000-50,000 t/y, with numerous examples of outputs of under 20,000 t/y. Plants with a productive capacity of between 70,000 and 110,000 t/y are less common. It is important to note that for certain applications (see below), significantly larger supplies of hydrogen peroxide are required. For all these reasons, considerable efforts have been made in recent years to identify alternative syntheses to hydrogen peroxide, resulting in two main directions. The first one is the reaction of carbon monoxide with oxygen and water to form hydrogen peroxide and carbon dioxide: CO O2 H2O H2O2 CO2

This reaction was reported for the first time in 1979 by Yuri Yermakov (of the Catalysis Institute of Novosibirsk, in Russia). The catalytic system used by Yermakov (based on soluble palladium salts and triphenylphosphine) was rather inefficient, but recently the substitution of the phosphine with nitrogen ligands (phenanthrolines) has led to

662

ENCYCLOPAEDIA OF HYDROCARBONS

OXIDATION PROCESSES

considerable improvement in the reaction performance and has enabled it to be conducted continuously (Bianchi et al., 1999). Of even greater interest is direct synthesis from the elements: H2 O2 H2O2

and (1-phenyl) ethyl hydroperoxide from ethylbenzene:


H3C CH2 O2 H3C CH OOH

This reaction has been known for a long time (the first patents date from the 1960s; Hooper, 1967), but it has not yet been applied industrially due to the problems linked in particular to safety in managing hydrogen-oxygen mixtures. The two gases are made to react in a solvent, typically water or methanol (Paparatto et al., 2003a), in the presence of a palladium base catalyst, possibly modified with the addition of platinum or of other noble metals. At the moment, this direct synthesis seems the most promising candidate to replace the traditional anthraquinone process in the near future.
Hydroperoxides

Alkylhydroperoxides, or more briefly hydroperoxides, formally derive from the substitution of a hydrogen atom of the hydrogen peroxide molecule with an alkyl R group, so that their general formula is R OOH. Hydroperoxides retain the weakness of hydrogen peroxides oxygen-oxygen bond: in fact, the energy of this bond is typically 40-44 kcal/mol, even lower than that of hydrogen peroxide (51 kcal/mol). The energy of the O H bond, instead, is 89-90 kcal/mol, practically coinciding with that of hydrogen peroxide (89.5 kcal/mol). Many hydroperoxides are known primary, secondary or tertiary but few have taken on industrial importance; those produced and used at a large scale are substantially three: two tertiary hydroperoxides, i.e. tert-butyl hydroperoxide and cumyl hydroperoxide, and a secondary one, i.e. (1-phenyl) ethyl hydroperoxide. They are prepared by autoxidation of the corresponding hydrocarbons, a typical radical reaction, and the hydroperoxide group is formed, mostly, at the expense of the C H bonds having less energy (e.g. tertiary or benzylic ones). These two cases are exploited, respectively, in preparing tert-butyl hydroperoxide from isobutane and cumyl hydroperoxide from cumene:
H3C CH H3C H3C CH O2 CH3 H3C CH3 O2 H3C CH3 C OOH

For example, the oxidation of isobutane and of ethylbenzene is commonly conducted in the liquid phase at 120-140C and under moderate pressure (30-40 and about 2 bar, respectively). Analogous conditions (90-130C and 5-10 bar) are adopted also in the oxidation of cumene; however, in this case, the reaction must be conducted in the presence of an emulsified aqueous phase, weakly basic due to the addition of sodium hydroxide or carbonate. Inevitably, in fact, small quantities of acid by-products are formed, mainly formic acid. These acids favour the decomposition of cumyl hydroperoxide giving acetone and phenol; the latter is an excellent inhibitor of many radical reactions and, in particular, its presence is not compatible with autoxidation. In general, the selectivity of all these reactions, between 60 and 95%, is satisfactory and depends mostly on the conversion of the hydrocarbon, which, however, is rather limited, approximately between 10 and 40%. Some uses of hydroperoxides, exploiting two very different aspects of their reactivity, are described below. On the one hand, their oxidizing properties are used, and in particular their capacity to transfer an oxygen atom to olefins, giving rise to the formation of an epoxide and an alcohol (which is the product of hydroperoxide reduction). In this aspect, hydroperoxides (such as hydrogen peroxide) are not very reactive and are therefore used in the presence of appropriate catalysts, generally based on transition metals (in particular, titanium and molybdenum). The use of cumyl hydroperoxide in the traditional cumene process for the production of phenol and acetone is completely different. In this case, the mentioned property of cumyl hydroperoxide to decompose in an acid environment, producing the two final products, is exploited. All hydroperoxides, and in particular those of lower molecular weights, tend to give rise to decomposition reactions, which can be of an explosive character.
Oxidation with hydrogen peroxide

CH3 OOH C CH3

Without any catalyst, hydrogen peroxide is not a particularly reactive oxidizing agent. However, it is employed in a series of non-catalytic uses, for example to bleach paper and wood pulp, in detergency (in the preparation of perborate and sodium percarbonate) and

VOLUME II / REFINING AND PETROCHEMICALS

663

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

in the environmental field (e.g. for eliminating toxic substances in waste water). A completely different use of hydrogen peroxide is made in the Ugine Kuhlmann process for the production of hydrazine, a process in which H2O2 is used in the presence of methyl ethyl ketone (MEK) and of a nitrile or an amide (Goor, 1992):
O C H3C MEK
H2O2, NH3, MEK
3H2O

NH3

NH C

H2O2, NH3, MEK

C2H5

H2O H C 3

C2H5 3H2O

H3C C H5C2 N N
2H2O

C2H5 C CH3

N2H4 + 2MEK

With respect to traditional processes (Raschig, Bayer), which are based on the oxidation of ammonia with chlorine or hypochlorite, the process using hydrogen peroxide avoids the coproduction of very large quantities of sodium chloride. Regarding the far more numerous catalyzed reactions, the first observation of a catalytic effect in oxidation reactions with peroxides dates from the classic work by Henry John Horstman Fenton, who, in 1876, described the capacity of ferrous salts to catalyze the oxidation of tartaric acid by hydrogen peroxide (Fenton, 1876, 1894). In its most classical form, the Fenton reagent consists of a mixture of hydrogen peroxide and ferrous sulphate, but, more generally, the hydrogen peroxide can be activated by any other reducing term of a monoelectronic redox couple such as, for example, copper (I). The question regarding the nature of the active species that comes into play in oxidations with the Fenton reagent (or with systems correlated with this) is still being debated. The traditional view that it is the hydroxyl radical ( OH) has been challenged in recent years and, instead, the intervention of high-valent iron species with oxidation state (IV) or (V) has been claimed (at least for reactions conducted in non-aqueous solvents). In any case, oxidations with the Fenton reagent are not very selective and, with few exceptions that will be discussed below, they are of little use in organic synthesis. They are applied, instead, in the environmental field, for example to purify waters polluted by phenolic compounds. Long after Fentons pioneering studies, in the 1930s, Nicholas Athanasius Milas described the catalytic effect of metals in electronic d0 configuration in oxidation reactions with hydrogen peroxide of organic and inorganic substrates (Milas and Sussman,

1936; Milas, 1937). The Milas reagents are formed by reaction of metallic oxides with hydrogen peroxide in a tert-butanol solution and were often used for oxidizing olefins to the corresponding vicinal diols, a reaction that in most cases leads to the formation of an intermediate epoxide that is successively hydrolyzed to diol by the acid medium. Although Milas used vanadium(V), chromium(VI) and osmium(VIII), it very soon became clear that analogous reactions were possible, often with better results, using molybdenum(VI) or tungsten(VI). However, the doubt remains that some of the reactions observed might have been caused not so much by the hydrogen peroxide as by the tert-butyl hydroperoxide, which, under the conditions adopted, can be formed by reaction of the tert-butanol with the hydrogen peroxide. Meanwhile, in the 1960s and 1970s, the reasons why it is difficult to use oxygen in selective oxidations grew clearer. Thus, many industrial chemists began assessing with interest the oxidizing properties of hydrogen peroxide and other peroxides that are partly reduced forms of molecular oxygen. The already mentioned derivatives of tungsten(VI) and, above all, of molybdenum(VI) were obvious candidates for the role of catalyst and, in fact, they formed the basis of the first industrial success in this sector: the epoxidation of propylene with hydroperoxides (see below). However, the use of hydrogen peroxide remained an aim of great interest as it would have made it possible to decouple the production of propylene oxide (or of other epoxides) from the coproduction of tert-butanol or of styrene. Yet the derivatives of vanadium(V) or of molybdenum(VI) and tungsten(VI), which catalyze the reactions of hydroperoxides extremely well, produced very poor results if used with hydrogen peroxide. To understand the reason for this failure, it must be considered that the catalytically active species is formed by reaction of the metal derivative with the oxidizing agent, whether this is a hydroperoxide or hydrogen peroxide. In the presence of water, however, the latter competes effectively with the oxidizing agent, giving rise to inactive species whose formation eventually inhibits the desired reaction. However, water is always present in oxidations with hydrogen peroxide, both because the latter is normally used in the form of an aqueous solution, and because new water is formed as oxidation takes place and the hydrogen peroxide is consumed. Therefore, strategies were sought that would enable the problem of water to be eliminated or, at least, its consequences to be lessened. Basically three approaches were followed: seeking anhydrous reaction conditions, the application of the phase-transfer technique and the development of

664

ENCYCLOPAEDIA OF HYDROCARBONS

OXIDATION PROCESSES

microporous catalysts in which oxidation occurs in a relatively hydrophobic environment, such as inside the cavities of a zeolite. Maintaining an anhydrous reaction environment was pursued both by using stoichiometric quantities of dehydrating agents, and, in a more practical manner, by getting rid of the water through continuous azeotropic distillation. This latter technology constitutes the basis of the Ugine Kuhlmann process for the epoxidation of propylene, although this has never been done at the industrial scale. More unexpected and promising developments came, however, from the application of the phasetransfer technology and from the discovery of microporous titanium silicates.
Homogeneous catalysts: phase-transfer processes

In the phase-transfer technology, the medium in which oxidation with hydrogen peroxide is conducted is formed by two liquid phases that are mutually immiscible, one aqueous and the other organic, formed by the substrate S possibly dissolved in a suitable solvent. In the case of oxidations with hydrogen peroxide, an anionic metal derivative reacts in the aqueous phase with the hydrogen peroxide, forming a peroxidic complex that will be the catalytically active species in the process. This species, also of an anionic nature, is extracted in the organic phase from a

[ ]
M O

Na

H2O2

[ [

O M O

] ]

Na

H2 O
water organic solvent

[ ]
M O

O M O

SO

Fig. 2. Oxidation under phase-transfer conditions.

lipophilic cation such as a tetraalkylammonium or phosphonium (the phase-transfer catalyst, Q ). In the organic phase, the peroxidic species reacts with the substrate forming the desired product (SO) and regenerating the reduced form of the catalyst. Distribution of this between the organic and the aqueous phases closes the catalytic cycle (Fig. 2). Nevertheless, the first attempts by Charles Starks (then with Continental Oil) to epoxidize simple olefins in the transfer phase using tungstic acid and hydrogen peroxide encountered little success due to the difficulty of limiting, under the reaction conditions, the unproductive decomposition of the hydrogen peroxide into oxygen and water. Only later was it discovered, unexpectedly, that a mixture of tungsten and phosphate (or arsenate) anions in an acid environment is able to catalyze the epoxidation reaction, provided that a tetraalkylammonium or phosphonium salt is present in the system as a phasetransfer catalyst (Venturello et al., 1983; Table 2). The reaction is applicable to a large number of olefins: linear, branched or cyclic, including hardly reactive ones such as terminal olefins or allyl chloride. In all cases, it is carried out under mild conditions (60-90C and ambient pressure), even when very diluted (8-15%) aqueous solutions of hydrogen peroxide are used. The reaction is stereospecific: trans-2-hexene gives only the trans epoxide and cis-2-hexene gives only the cis isomer. The catalytically active species that are formed from tungsten(VI) under phase-transfer conditions and in the presence of phosphate or arsenate anions belong to a new class of peroxidic complexes, which are quaternary ammonium (or phosphonium) salts of the PW4O3 (Fig. 3), or 24 AsW4O3 anions (Venturello et al., 1985). These 24 complexes are the first peroxidic derivatives of a heteropolyacid whose structure has been solved. Epoxidations are conducted batchwise and, under the reaction conditions, most of the catalyst is dissolved

Table 2. Epoxidation of olefins with diluted H2O2 catalyzed by WO2 /PO3 mixtures 4 4 Olefin 1-octene 1-dodecene Allyl chloride Styrene a-methylstyrene Cyclohexene pH 1.6 11.1.3 Clerici fig 02 1.6 2 3 4.5 3 Temperature (C) 70 70 60 40 40 70 Time (min) 45 60 150 180 240 25 H2O2 conv. (%) 98 97 96 93 93 98 Epoxide yield (%) 82 87 80 77 79 88

VOLUME II / REFINING AND PETROCHEMICALS

665

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

3 Fig. 3. The PW4O24

anion.

in the organic phase constituted by the olefin and possibly by a solvent (generally toluene or a chlorinated hydrocarbon). At the end of the reaction (indicated by the more or less complete disappearance of the hydrogen peroxide), the two phases easily separate and the product is recovered with conventional techniques, generally by distillation. Typical yields, based on hydrogen peroxide, vary between 80 and 95%, and any excess olefin is recovered unchanged: generally, in fact, selectivity on the olefin is around 95%. Even sophisticated substrates, of pharmaceutical interest, can be epoxidized with excellent yields. A block diagram of the epoxidation process of a typical olefin is shown in Fig. 4. At the end of the reaction, the catalyst is recovered, with small quantities of by-products, as distillation residues (or by ultrafiltration of the organic phase) and can be regenerated and partly recycled. The epoxidation reaction has been developed on a commercial scale for the production of 1,2-epoxydecane and of isobutyl 3,4-epoxybutyrate. The latter, in particular, has been produced on the scale of 100 t/y by epoxidation of isobutyl vinylacetate
olefin

and is a key intermediate in the synthesis of the nootropic drug oxiracetam:


O cat. C O H2O2
H2O

O O HO N O C O

CH2CONH2

Although epoxidation is the most interesting of the reactions promoted by these new catalysts, they are able to use hydrogen peroxide efficiently for numerous other transformations that are also of significant interest, including oxidations of primary alcohols and of aldehydes to form carboxylic acids, of secondary alcohols to give ketones or of sulphides to give sulphoxides (Ricci, 1996). Furthermore, an accurate choice of reaction conditions makes it possible to obtain, in addition to epoxides, a whole series of other olefin derivatives, often difficult to prepare by other methods. Thus it is possible to transform olefins into

Fig. 4. Epoxidation of

olefins with hydrogen peroxide under phase-transfer conditions.

olefin catalyst epoxidation phase separation distillation epoxide

H2O2

H2O

residue

666

ENCYCLOPAEDIA OF HYDROCARBONS

OXIDATION PROCESSES

olefins

epoxides

1,2-diols

1,2-hydroxyketones

carboxylic acids

1,2-diketones

Fig. 5. Products obtainable from oxidation of olefins

in phase transfer.

vicinal diols or into 1,2-hydroxyketones. Both of these classes of compounds can be oxidized, in their turn, to give 1,2-diketones. Lastly, diols or even the original olefins can be oxidatively cleaved with the formation of carboxylic acids (Fig. 5). Recently, papers have been published which seem to suggest that, under appropriate conditions (e.g. in the complete absence of any chlorinated species), the performances of tungsten and phosphorus based peroxide catalysts can be equalled by tungstic acid alone or, perhaps, by polymeric species produced by this in the reaction environment. However, peroxidic phosphotungstates retain the merit of having made possible the first commercial applications of oxidations with hydrogen peroxide catalyzed by d0 metals and, more generally, that of having rekindled interest in their study, then further revitalized by the discovery of the new rhenium(VII) based catalysts, especially CH3ReO3 (Herrmann et al., 1991).
Heterogeneous catalysts: micro- and mesoporous titanium-silicates

As already mentioned, another successful approach in the field of oxidation with hydrogen peroxide is that
Fig. 6. Formation and

which followed the discovery of micro- and mesoporous titanium-silicates. In particular, certain titanium-zeolites are to be included among the most promising oxidation catalysts. Their activity and selectivity, often very high, and their intrinsic stability against oxidative degradation make them ideal catalysts for industrial applications. These titaniumzeolites are part of the more extensive family of mixed oxides containing a transition metal at high dispersion, whose structure can vary from microporous crystalline (redox zeolites) to wholly amorphous, with an intermediate situation represented by amorphous metal silicates, with an orderly system of pores (MCM-41, MCM-48, HMS). Among these titanium-silicates, the first, in order of time and importance, is titanium-silicalite-1 (TS-1), discovered in the late 1970s (Taramasso et al., 1983). This is a microporous crystalline titanium-silicate, with an MFI structure (Baerlocher et al., 2001) which derives formally from silicalite-1 (S-1), wholly siliceous, by the substitution of part of the silicon with titanium, up to a maximum of around 2.5%. It is obtained by hydrothermal synthesis, from the progressive condensation of SiO4 and TiO4 tetrahedra around the template ion (i-C3H7)4N (Fig. 6). Its porosity is due to a three-dimensional system of interconnected channels, having an average diameter of about 0.55 nm. This is an important parameter, as it sets an insuperable limit to the diffusion of reagents in the active titanium sites and thus to the feasibility of the oxidation process. Materials with an analogous composition, but of different structure, have been prepared by incorporating titanium in the S-2 (titanium-silicalite-2, TS-2), BEA (titanium-b, Ti-b), MOR (titanium-mordenite, Ti-MOR), and MWW (Ti-MCM-22) structures, and others of less interest for catalysis. Mesoporous titanium-silicates (Ti-MCM-41, Ti-MCM-48) are also known. Apart from titanium, other metals with redox characteristics have been incorporated into the

growth of the crystal lattice of TS-1.

Si O

Si O O Si

Ti O O Si

Si O O Si

Si H2O2 Si

Si

SiO4 TiO4
Si

Si Si OO O H H O O O O O Si Si Si Si Si O O

Si

Ti

VOLUME II / REFINING AND PETROCHEMICALS

667

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

Table 3. Some zeolites structurally substituted with transition metals.

Those most studied for catalytic applications are in bold type


Name TS-1 TS-2, TS-3 TS-12 Ti-FER Ti-MWW Ti-b Ti-MOR Ti-SSZ-33 Ti-UTD-1 Ti-MCM-41 Ti-MCM-48 Structure MFI MFI/MEL MTW FER MWW BEA MOR CIT-1 DON Metals Ti, Fe, V, Zr, Cr Ti, Fe, V Ti, Fe Ti, V Ti Ti, Fe, Sn, Cr Ti Ti Ti, V Ti, V Sn, Mn , Ti, V Cr , Size of pores () 5.65.3 5.35.4 5.95.5 5.44.2 5.44.0 6.47.6 6.57.0 6.47.0 ca. 10 20-100 20-100 Dimensionality 3 3 1 2 2 3 2 3 1 1 3

structures of many zeolites. Table 3 offers a selection of the materials described in scientific and patent literature. The latter form a relatively small group compared with the number of materials proposed, the majority of which, however, under reaction conditions, are not sufficiently stable and in some cases are actually lacking in any catalytic activity.
Physico-chemical characterization of mixed oxides

Reactivity tests, too, can help in the characterization process. The hydroxylation of phenol, for example, is very sensitive to the presence of extraframework species and has been proposed as a supplementary test in this regard (see below).
Physico-chemical properties and catalytic properties of titanium-zeolites

A crucial factor for catalytic action is the location of the metal in the zeolitic structure. It is thus essential to establish whether it is inserted in the crystal lattice in atomic dispersion or is deposited on the surface in the form of discrete particles of oxide. It is equally important to be able to exclude the presence of occluded amorphous phases and to determine the morphology of the crystals (dimension and shape). Before any catalytic study, the physico-chemical characterization of the potential catalyst must therefore be accurately established. Unfortunately no single technique exists that is both simple and at the same time able to provide clear answers (Boccuti et al., 1988; Millini and Perego, 1996). Generally, it is necessary to resort to the combined use of different techniques to obtain sufficient certainty as to the quality of the catalyst. For titanium-silicalite and the other titanium-zeolites, the most commonly used techniques are X-Ray Diffraction (XRD), UltraViolet absorption (UV), FTIR and Raman spectroscopy, X-ray absorption spectroscopy (EXAFS and XANES) and electron microscopy (SEM, TEM).

Catalytic properties depend on many factors, some of which (position of the titanium, morphology of the crystals) have already been mentioned. In addition to these are the geometry and the surface properties of the pores, which impart to the titanium-zeolites the properties of molecular sieves (i.e. the capacity to discriminate the compounds to be adsorbed on the basis of their dimensions and physico-chemical properties). The relations between surface properties and catalytic performance may be summed up as follows: Only the titanium present in the structure is catalytically useful. Titanium in an external position is actually harmful due to its capacity to catalyze parasite reactions, such as the decomposition of the hydrogen peroxide and radical oxidation pathways. An analogous criterion applies for zeolites containing metals other than titanium. The titanium present in the structure confers to the silicalite-1 very different catalytic properties from those conferred by vanadium, by iron and by any other transition metal (see below). The dimensions and geometry of the pores give

668

ENCYCLOPAEDIA OF HYDROCARBONS

OXIDATION PROCESSES

Fig. 7. Chemical and steric properties

of active sites of TS-1.

Si O O

Si O Ti OH2 OH2 OH O H Si

Si H2O O O Si

Si O Ti O Si H2O/H2O2

Si O O

Si O Ti OH2 OH2 OOH O H Si

Si

Si

rise to what is termed shape selectivity (i.e. the capacity of the material to discriminate the reagents and products on the basis of their shape and size; see below). Selective adsorption phenomena, governed by the nature of the surface, concentrate specific components of the reaction mixture in the channels, in the vicinity of the catalytic sites, decisively influencing the course of the reaction (activity, selectivity, deactivation). The capacity of TS-1 to adsorb selectively non-polar compounds, even in the presence of water and other polar and protic substances, explains the apparent difference of the catalytic properties of titanium in TS-1 and in soluble alkoxides. The latter are more or less inactive in an aqueous solution of hydrogen peroxide, that is under the conditions of maximum activity of TS-1. The phenomena that take place on the catalytic site are closely connected, on the one hand, with the tendency of framework titanium to expand its sphere of coordination from tetrahedral to octahedral, chemiadsorbing polar molecules, and, on the other hand, with the poor resistance to hydrolysis of a Ti OSi unit (Fig. 7; Bellussi et al., 1992; Clerici and Ingallina, 1993). In accordance with this, the water and the alcohols are coordinated on the titanium and reversibly hydrolyse a Ti OSi bond, producing Ti OH and Si OH species. The hydrogen peroxide behaves in like manner, producing the Ti OOH species, which is at the origin of the oxidizing properties of the TS-1/H2O2 system. The structures illustrated in Fig. 8 have been proposed for the Ti OOH species (for the sake of clarity, the other ligands have been omitted; for these, see again Fig. 7). The structure represented in A is that which has the best chances of existing under the conditions suitable for catalysis (i.e. in the presence of water and/or of an alcohol). TS-1 catalyzes the oxidation of many organic functions with hydrogen peroxide, and in particular the epoxidation of olefins and the hydroxylation of paraffins and aromatic compounds, as well as the oxidation of alcohols, ethers and various sulphur and nitrogen compounds. On the contrary, it is not active in

R O Ti O H O H Ti O O H

Fig. 8. Structure of catalytically active species

of TS-1.

R O Ti O H O H

R O Ti O H O H

Fig. 9. Mechanism of epoxidation catalyzed by TS-1.

oxidation processes with organic hydroperoxides. The other metal-zeolites are active for a more limited spectrum of reactions. Those with large pores and mesoporous materials, however, permit the use also of organic hydroperoxides as oxidants. TS-1, but also TS-2, Ti-b, Ti,Al-b and Ti-MWW catalyze the epoxidation of olefins, in a diluted solution of hydrogen peroxide ( 10%) and at temperatures below 80C. TS-1 already shows a good level of activity even below 0C. In fact, epoxidation is the preferred reaction when other reactive groups are present in the molecule. In this regard, chemoselectivity can be extremely high. As opposed to what has been observed in the case of soluble titanium catalysts, the solvents that favour high reaction rates are protic and polar: methanol (TS-1, TS-2, Ti-b, Ti,Al-b), acetone (TS-1) and acetonitrile. In a moderate quantity, water has a minor effect on yields and selectivity. On the other hand, its presence in the reaction environment cannot be avoided, being added with the oxidant and produced by the same reaction. The decomposition of hydrogen

VOLUME II / REFINING AND PETROCHEMICALS

669

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

TS-1:
1.0 4.7 6.0 16.2

Ti,Al-:
1 1.1 R 1 1.6 R1 R 20-24 1.6

CH3CO3H:

AcO 0.04

27

240

Fig. 10. Relative rate of epoxidation of some substituted olefins with various oxidants.

peroxide is generally slow with the catalysts indicated or even negligible with TS-1. Yields and selectivity are generally high, and values close to 100% can be reached with TS-1, in accordance with the heterolytic mechanism (Fig. 9; Clerici and Ingallina, 1993; Neurock and Manzer, 1996). By-products of the reaction are generally products of the hydrolysis of epoxide. Their formation may be reduced, neutralizing the residual acidity of the catalyst, with the addition of small amounts of bases in the reaction environment. All zeolitic catalysts, not only oxidation ones, possess shape selectivity: only olefins capable of being diffused inside the pores can be oxidized with high reaction rates. This explains the different reactivity of cyclohexane on TS-1 (very poor) and on Ti,Al-b (good), as its molecular size is very close to the pore diameter of the former and significantly smaller than that of the latter. More generally, shape selectivity completely reverses the order of reactivity foreseen on the basis of the electronic properties of the double bond. Fig. 10 compares olefins with a different steric size in epoxidation with TS-1/H2O2, Ti,Al-b/H2O2 and with peracetic acid, which is not significantly affected by steric restrictions (the numbers associated with olefins indicate their relative reactivity). The steric effects of the substituents on the double bond clearly prevail over the electron effects with heterogeneous catalysts, whereas the opposite is the case for peracetic acid. While the activity, the selectivity and the mild reaction conditions of the TS-1/H2O2 system make it a preferable alternative to conventional oxidants, the steric restrictions imposed by its microporous nature set evident limits to the range of its potential applications. Ti-b and the other titanium-zeolites with large pores provide a partial solution to the problem. The synthesis of mesoporous catalysts (Ti-MCM-41, Ti-MCM-48 and suchlike) could be an answer to the need. However, the results obtained with hydrogen peroxide are not very encouraging. The conclusions of these studies may be summed up as follows:

The rate of reaction diminishes rapidly in the order: TS-1 Ti-b Ti-MCM-41, becoming more or less negligible on the mesoporous catalyst. Furthermore, Ti-b, Ti,Al-b and Ti-MCM-41 show, to an increasing extent, a structural instability in aqueous hydrogen peroxide. The drawback is common to other metal-zeolites, in particular those containing vanadium, and takes the form of a progressive collapse of the crystal lattice, with the release of the metal in solution. The mesoporous materials available have a substantially hydrophilic surface. This circumstance favours the adsorption of water to the detriment of the olefin or other non-polar reagent, thereby lowering the reaction rate. Unfortunately, water cannot be avoided in using hydrogen peroxide as the oxidant. A possible alternative is provided by the use of tert-butyl hydroperoxide (TBHP), for which the activity increases in inverse order: Ti-b Ti-MCM-41 while TS-1 is practically inactive for steric reasons. The above conclusions, therefore, do not imply any decreasing activity of titanium sites with increasing pore size, but rather a decrease in the olefin adsorption capacity. It is the density of the surface Si OH groups and thus the hydrophilic nature of the pores that increases in the TS-1 Ti-b Ti-MCM-41 series, in parallel favouring the adsorption of water at the expense of the olefin (or some other non-polar reagent). The instability, too, is to some extent a consequence of the increase of the Si OH groups, which may be regarded as defective sites of the siliceous matrix and the starting point of the hydrolytic phenomena that cause the decay of the matrix. TS-1 catalyzes the hydroxylation of paraffins, at temperatures below 100C, in an aqueous or methanolic solution of hydrogen peroxide (Huybrechts et al., 1990; Clerici, 1991):
TS-1 H2O2 TS-1 H2O2 TS-1 H2O2

OH

OH

It may seem surprising that commonly inert compounds such as paraffins, even in a diluted solution, can be oxidized preferentially with respect to the solvent methanol. The reason for this is probably due to the hydrophobic nature of the catalyst, able to selectively adsorb the paraffin in the vicinity of the active sites in accordance with the criteria already stated for the epoxidation of olefins. The same reason

670

ENCYCLOPAEDIA OF HYDROCARBONS

OXIDATION PROCESSES

likely explains the poor activity, or the lack of activity, of the large-pore titanium-zeolites and of mesoporous catalysts, for which the order of reactivity, already seen for olefins, now appears far more accentuated. In competition with hydroxylation, the decomposition of the hydrogen peroxide also takes place, generally to a notable extent. Only secondary carbons undergo oxidation, the methyl groups being inert, which produces a mix of secondary alcohols and ketones (formed by consecutive oxidation). The vanadium-silicalites, instead, are active also for the hydroxylation of the primary carbons, forming a mix of primary and secondary alcohols and products of successive oxidation. As far as the aromatic compounds are concerned, TS-1 and Ti-MOR catalyze their hydroxylation to the corresponding phenols (Romano et al., 1990):
TS-1 TS-1

OH OH

OH
H2O2 H2O2

HO

HO OH

OH

The reaction follows the rules of electrophilic attack: the reactivity of phenol and toluene is high, while that of nitrobenzene, chlorobenzene and benzoic acid is negligible. In alkylbenzenes, the oxidation of the alkyl groups competes with that of the aromatic nucleus, except for the methyl groups, which are almost completely inert. In this case as well, the behaviour of the vanadium-silicalites is different, as they preferentially catalyze the hydroxylation of the alkyl side chains, including methyl groups. The hydroxylation of paraffinic and aromatic compounds is regulated by severe steric restrictions imposed by the size of the pores. The kinetics of the hydroxylation of toluene is already comparable with that of benzene, in spite of the fact that it is more nucleophilic. A second consequence of the steric effects is that the decomposition of hydrogen peroxide can become the predominant reaction when the substrates are bulky molecules. The mechanisms of hydroxylation reactions have not been studied in depth. However, there is strong evidence of their homolytic nature, causing them to differ from epoxidation, which is heterolytic. A mechanism able to explain both the hydroxylation and

the competition of the decomposition of hydrogen peroxide has recently been proposed (Clerici, 2001). The titanium-zeolites resemble other catalytic systems in the oxidation of primary and secondary alcohols, except for the effects of steric restrictions (Maspero and Romano, 1994). The products are aldehydes and the corresponding ketones, to which, in the case of the primary alcohols, the carboxylic acids formed by consecutive oxidation can also be added (Fig. 11). The primary alcohols are not as easy to oxidize as the secondary ones. It is worth noting the scanty reactivity of methyl alcohol, which makes it suitable as a solvent for the other oxidations. Ketones are generally stable under the oxidation conditions of the secondary alcohols. They can, however, be oxidized to lactones and the corresponding esters, selecting the appropriate catalysts. Sn-b proves to be the most selective catalyst in this context, with values that can reach as high as 98% (Corma et al., 2001). Contrary to the case of the titanium-zeolites, if double bonds are present, they do not undergo epoxidation, as exemplified by the oxidation of dihydrocarvone (Fig. 12). The reason for this must be sought in the different mechanism, which for Sn-b is based on the Lewis acid properties of the active site, instead of redox properties as in the case of Ti-b. In the oxidation of amines, various products can be obtained, according to the nature of the amine and of the reaction conditions:
NH2 NHOH
ArNHOH

NO N N

ArNHOH

RCH2NH2

RCH

NOH

O RCHO

From aniline, when hydrogen peroxide is scarce, the final product obtained is azoxybenzene, while with an excess of oxidant, the main product is nitrobenzene (not shown). Oximes, alkyl-hydroxylamines and other products are formed in the oxidation of the primary and secondary aliphatic amines. Also in the case of the oxidation of amines, there is a different selectivity on the various metal sites. In the oxidation of aniline, for example, the vanadium-silicalites form mainly nitrobenzene even when hydrogen peroxide is scarce.

Fig. 11. Mechanism of alcohol oxidation catalyzed by TS-1.

RCH2 O Ti O H O H

H O Ti O

H C O

R H H OH2 Ti O H RCHO

VOLUME II / REFINING AND PETROCHEMICALS

671

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

O O H2O2

Sn O

O O O

O R O Ti O H O H O OH

O OH

Fig. 12. Different oxidation mechanism of Sn-b and Ti-b catalysts.

The thioethers are easily oxidized to the corresponding sulphones and sulphoxides: [12] R S R R SO R R SO2 R

The reaction, initially hardly considered due to the ease with which the thioethers oxidize, may take on importance as a means of removing sulphur from motor fuels. It is clear that in this application the most suitable catalysts are those with a mesoporous structure, and consequently the preferred oxidant is tert-butyl hydroperoxide. Epoxidation of olefins
Industrial production of epoxides

Among the products that can be obtained by oxidation of olefins, epoxides have always received special attention due to their high reactivity, which Chlorohydrin process makes them extremely useful (and used) intermediates Introduced about a century ago, the chlorohydrin in organic synthesis. process is still widely applied. Propylene is reacted As often happens, the simplest epoxide, ethylene with chlorine in an aqueous solution, producing a oxide, is produced with a method (i.e. direct oxidation of mixture of two chlorohydrins, from which propylene ethylene with oxygen in the gas phase) which differs oxide is obtained by treatment with lime or soda from that used for all the other ones. The higher epoxides, (global yield around 89%): ula 9 instead, are prepared under milder conditions and with Cl OH more complex processes. In fact, at the relatively high temperatures required by the use of molecular oxygen, in CH3 CH CH2 H2O Cl2 n CH3 CH CH2 addition to double-bond, also allylic C H are oxidized OH Cl OH Cl with the consequence of obtaining a broad spectrum of CH CH HCl (1(1 n)n) CH3 CH CH2 2 HCl CH3 products. For this reason, the higher epoxides, when not Cl OH OH Cl obtained through oxidation of the corresponding olefins with percarboxylic acids (a troublesome reaction, also n CH3 CH CH2 (1 n) CH3 CH CH2 NaOH from the standpoint of safety), are generally prepared by CH CH CH NaCl 2 2 CH3 3 CH CH2 2 NaCl HHOO oxidation of olefins with chlorine (through the OO corresponding chlorohydrins) or with hydroperoxides. In

the future, epoxidation with hydrogen peroxide, now in the development stage, could be added to these processes. The manufacture of propylene oxide, which, among the higher epoxides, is by far the most important commercially, will be examined below. The production capacity for propylene oxide (PO) in the year 2000 globally amounted to 5.7106 t, with an annual demand of around 4.7106 t. It is estimated that about 7% of the propylene produced is used for the production of this derivative. Propylene oxide is a basic intermediate of the chemical industry, used for producing a long series of commercial products. For this purpose, it must first be transformed into a number of its derivatives: polyether polyols (65%), propylene glycols (20%), and other intermediates such as glycolethers, alkanolamines, 1.4-butanediol, allyl alcohol and propylene carbonate. Polyether polyols, produced by the reaction of propylene oxide with polyhydric alcohols, are used in the production of polyurethane foams, of flexible type (applied, for example, in the automobile industry for seats, or for household or furnishing purposes for mattresses and carpets) or of rigid type (used, for example, for heat insulation). Propylene glycol (1.2-dihydroxypropane) is used in the production of polyester resins and, due to its biocompatibility, in the formulation of cosmetic, pharmaceutical and food products. For the same reason, dipropylene glycol and other short chain oligomers are used to produce antifreeze fluids, hydraulic fluids, cutting oils and lubricants. As opposed to analogous products derived form ethylene oxide, in fact, they are biodegradable and harmless to living organisms. Mention should also be made of certain uses of alkanolamines (detergents, anticorrosion products), of polyethers (antifoam agents, surface-active agents) and of esters (solvents for inks and paints).

672

ENCYCLOPAEDIA OF HYDROCARBONS

OXIDATION PROCESSES

The chlorohydrin process is burdened by a series of problems, with environmental implications, such as the formation of inorganic and organic chlorinated coproducts and byproducts. In fact, a large part of the chlorine used eventually forms a stoichiometric quantity of hydrogen chloride, which is neutralized a process taking place simultaneously with the formation of epoxide using lime or soda. A considerable quantity of organic chlorinated by-products is also formed, which uselessly consume propylene and complicate downstream operations. The effluent, consisting of a diluted solution of sodium or calcium chloride and chlorinated organic impurities, must be properly treated and then disposed of. A further problem arises from the risks of corrosion and hence from the need to use resistant materials. In spite of this, the chlorohydrin process is economically viable, able to satisfy approximately one-half of the world market of propylene oxide.
Epoxidation with hydroperoxides

CH3 CH3 C H CH3

CH3 C OOH CH3

CH3 C OH

CH3 CH3 CH3 C OOH CH3

CH3

CH3

CH

CH2 CH3

CH3 CH3 CH O CH2 CH3

OH

CH3

The processes in which an organic hydroperoxide acts as the oxidizing agent were originally developed by Halcon, ARCO and Shell and were applied in the industry as of the 1970s. These processes produce approximately the other half of the propylene oxide marketed in the world. They are liquid-phase oxidation processes, in which the catalysts may be soluble (Halcon/ARCO) or supported (Shell). The oxidants are TBHP (tert-butyl hydroperoxide) or EBHP (ethylbenzene hydroperoxide or (1-phenyl)ethyl hydroperoxide), obtained by oxidation with air of isobutane or ethylbenzene, respectively. The alcohols coproduced, tert-butanol (TBA) or 1-phenylethanol, are used for the production of MTBE (methyl tert-butyl ether) and styrene, respectively. The commercial advantage of propylene oxide produced in this way depends largely on their value. A third process also exists, introduced recently by Sumitomo, in which the oxidant is cumyl hydroperoxide and the alcohol coproduced is recycled instead of being placed on the market.
Epoxidation processes with TBHP

Two versions of the process exist, as used by Lyondell (formerly by ARCO) and by Huntsman (formerly by Texaco). The main reactions are the same:

In the ARCO process, six stages are distinguished: oxidation of the isobutane, epoxidation of the propylene, separation of the products, purification of PO, purification of TBA, preparation and recovery of the catalyst. In the first stage, isobutane is oxidized in the liquid phase with oxygen, at around 140C under pressure, producing a mixture of tert-butyl hydroperoxide, tert-butanol and various by-products, such as acetone and carboxylic acids. The reaction is run in recycled tert-butanol and is a typical autoxidation requiring no catalyst, although the presence of initiators speeds up the initial phases. For conversion values close to 35%, the tert-butyl hydroperoxide and tert-butanol selectivities are 53% and 40%, respectively. The stream from the autoxidation reactor is subjected to fractionation to remove the isobutane, diluted with recycled tert-butanol to bring the concentration of TBHP to 40% and, after adding the molybdenum catalyst solution, fed to the epoxidation reactor. A large excess of propylene is supplied to maximize yields and diminish reaction times. The reaction is conducted at about 120C and under pressure, with up to 98% selectivity with respect to propylene. Among the by-products are methyl formate (particularly unwelcome since its boiling point is close to that of epoxide), carbonyl compounds, carboxylic acids and propylene glycol. The successive phases include the separation of the products from the excess propylene (to be recycled), the recovery of the molybdenum and the purification of the propylene oxide and of the tert-butanol. The process is characterized by the production of PO and tert-butanol in a ratio of 1:2.4. In a second version of the process, originally developed by Texaco, epoxidation is carried out in two successive stages at 110C and 135C. This second version differs from the preceding one also in the methods of preparing the catalyst and the separation and purification of the products, as well as in the lower PO/tert-butanol ratio.

VOLUME II / REFINING AND PETROCHEMICALS

673

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

Epoxidation processes with EBHP

Oxidation process with cumyl hydroperoxide

Two processes exist in which the oxidant is (1-phenyl)ethyl hydroperoxide, differing in type of catalyst:
CH3 C H CH3 C H CH3 CH3 CH O CH2 C H OH OOH CH3 CH CH2 H CH3 C H OOH CH3 C H OH

The process developed by Sumitomo, based on cumyl hydroperoxide as the oxidant, is characterized by the fact that the co-product, cumyl alcohol, is dehydrated and hydrogenated to cumene and then recycled. The hydroperoxide route, in this particular case, avoids having to place another product on the market. Furthermore, it is characterized by its temperature and pressure conditions, significantly milder than in other hydroperoxide processes. So far it has been applied only in a single plant in Japan, which has been in operation since 2003 (200,000 t/y).
Epoxidation with hydrogen peroxide catalyzed by TS-1

In the ARCO (now Lyondell) process, a soluble compound of molybdenum is used, whereas in the process developed by Shell, a silica supported titanium(IV), Ti/SiO2, is used. The ratio of the products PO/styrene, instead, is similar and close to 1:2.2. In the first stage of both processes, the hydroperoxide solution is prepared by oxidation of ethylbenzene with air at about 150C and a pressure of about 3.5 bar. For safety reasons, conversion is limited to values of less than 10%. Selectivity is 80-85% with respect to hydroperoxide, the remainder consisting mostly of 1-phenylethanol and acetophenone. The EBHP concentration required for the epoxidation reaction (17-19%) is reached by removing the excess of ethylbenzene by distillation. Epoxidation is conducted at 100-115C. In the Shell process, the heterogeneous catalyst Ti/SiO2 makes it possible to adopt a packed fixed-bed reactor. A notable feature of this catalyst is its stability regarding the release of soluble species in solution. On the contrary, under analogous reaction conditions, supported molybdenum, tungsten or vanadium catalysts release soluble species. The regeneration of the deactivated catalyst Ti/SiO2 takes place by oxidizing the organic deposits with air. What characterizes the processes with EBHP is the stage dedicated to the production of monomeric styrene. The mixture of 1-phenylethanol and acetophenone, recovered from the effluent of the epoxidation reactor, is dissolved in triphenylmethane and subjected to dehydration at a high temperature in the presence of an acid catalyst. After the separation of the styrene, the residue containing acetophenone is conveyed to a hydrogenation reactor, after which the additional 1-phenylethanol rejoins the stream sent for dehydration. Selectivity exceeds 98%. From the standpoint of overall performance, in both the ARCO and the Shell process the yields of epoxide are nearly 92% and those of styrene are 94%.

This process differs from the preceding ones in its use of hydrogen peroxide as the oxidant, made possible by adopting titanium-silicalite as the catalyst (Clerici et al., 1991). The coproduct in this process is merely water, which makes it unnecessary to market or recycle it, and offers indisputable environmental advantages due to the absence of chlorinated by-products. Epoxidation is conducted in aqueous methanol at a temperature below 60C and pressures slightly above 1 atm. The reaction rate is high, with a TOF (Turn Over Frequency) of 1-2 s 1 (measured in the laboratory at 40C). Its selectivity with respect to hydrogen peroxide is nearly quantitative. Hydrogen peroxide consumption, due to methanol oxidation and to decomposition into water and oxygen, is negligible. The by-products are propylene glycol and its two monomethyl ethers, produced by the attack of methanol on the epoxide ring. The selectivity, generally high, can be made nearly quantitative ( 98%), by keeping the acidity of the environment under control with the addition of small quantities of bases (in the order of some ppm). The reaction takes place by contacting, in a slurry reactor, a stream of propylene with a suspension of TS-1 in methanol/water, into which an aqueous solution of hydrogen peroxide is fed (Romano, 2001). The high activity of the catalyst in a diluted solution makes it possible to operate with oxidant concentrations of even less than 10%. The effluent is subjected to distillation to recover the unreacted propylene, the propylene oxide and the methanol. The residual aqueous solution, after possible recovery of the propylene glycol and the monomethyl ethers, is conveyed to an ordinary biological plant for final treatment. Proper control of the operating conditions enables the phenomena of catalyst deactivation, caused by the depositing of heavy organic by-products, to be minimized. These by-products can be removed by washing with solvent at temperatures higher than 100C or, simply, by combustion with air. No

674

ENCYCLOPAEDIA OF HYDROCARBONS

OXIDATION PROCESSES

CH3 CH H2C CH2 OH H 3C C CH3 OOH

Ti (i-PrO)4 D - (-) - DET

O H CH2OH H H (R)-glycidol H3C

CH3 C CH3 OH

H HO D - (-) - DET HO C C COOC2H5 H COOC2H5

Fig. 13. Enantioselective epoxidation of allyl alcohol.

phenomena of deactivation due to the leakage of titanium from the crystal lattice have been observed. Compared with the chlorohydrin and hydroperoxide processes, the new technology is characterized by its low environmental impact, a simpler process layout and reduced investment costs. For example, there is five times less aqueous waste and it is suitable for normal biological treatment. A prototype plant of about 6 t/d capacity was set up by EniChem in 2001.
Enantioselective epoxidations

Some rather peculiar reactions, enantioselective epoxidations, will now be addressed, which have found various industrial applications, although mostly in the production of small quantities of compounds having extremely high value added, typically used by the pharmaceutical industry. The first procedure for enantioselective epoxidation with high enantiomeric excesses was published in 1980 by Barry Sharpless (then at Stanford University), which consists in the reaction of primary allyl alcohols with a hydroperoxide (e.g. tert-butyl hydroperoxide or cumyl hydroperoxide), promoted by stoichiometric quantities of an alkoxide of titanium(IV) and of enantiomerically pure diethyl tartrate (Katsuki and Sharpless, 1980). The active species is formed in situ from the titanium that simultaneously binds the enantiomerically pure tartrate, the hydroperoxide and the substrate (the latter through the OH group). The synthetic versatility of the allyl alcohols caused the reaction to rapidly become a synthetic instrument of primary importance. The road towards industrial applications was further eased when, in 1986, Sharpless discovered that, in the presence of molecular sieves, it was possible to run the oxidations using catalytic quantities (5-10% in mols) of the titanium-tartrate complex.

The epoxidation of allyl alcohol was one of the first enantioselective epoxidation processes developed at industrial scale (by ARCO; Fig. 13). Using the tartaric esters of the one form or the other, both the enantiomers of glycidol can be obtained with enantiomeric excess (ee) between 91 and 95%. The isolation of the product is greatly facilitated by transforming in situ the glycidol (a labile, water soluble product) into the corresponding m-nitrobenzenesulphonate, which, after crystallization, is recovered with ee 99%. More or less at the same time as Sharplesss enantioselective epoxidation, another efficient procedure was published, using hydrogen peroxide as the oxidizing agent. This was the Juli-Colonna epoxidation, which uses polyamino-acids as chiral catalysts, typically polyalanine (Juli et al., 1982). The enantiomeric excesses are very large (up to 96%), but limited to the epoxidation of a number of a, b-unsaturated ketones. New and important contributions followed. In 1988, it was again Sharpless who introduced asymmetric dihydroxylation (which actually produces diols and not epoxides), catalyzed by osmium tetroxide and chiral amines; then, in 1991, a protocol was defined for asymmetric epoxidation (catalyzed by manganese complexes with chiral Schiff bases), which is applied to a large number of olefins and no longer only to allyl alcohols. In both cases, however, the oxidants used are no longer peroxides, but are amine oxides, K3[Fe(CN)6] or sodium hypochlorite. Enantiomerically pure epoxides, prepared by the one procedure or the other, are used in the synthesis of various pharmaceutical products (e.g. b-blocking or antiviral drugs) and of products for agriculture (pheromones).

VOLUME II / REFINING AND PETROCHEMICALS

675

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

Oxidation of aromatics
Industrial production of phenol

the oxidant and iron containing ZSM 5 zeolites as catalysts:


ormula 13 N2O
Fe-zeolite

Phenol is one of the main chemical intermediates with increasing world production, amounting to about 8106 t in 2004. Above all, it is used in the production of intermediates for polycarbonates and epoxy resins, phenolic resins (used, for example, in adhesives) and caprolactam, the raw material of Nylon 6. At present, nearly all the global phenol production is based upon the cumene process, which is also the main way of industrially producing acetone: for every 10 t of phenol, in fact, about 6 t of acetone are coproduced. The cumene process foresees three stages: reaction of benzene with propylene (alkylation), which gives isopropylbenzene (cumene); oxidation of the cumene to the corresponding hydroperoxide; acid decomposition of the hydroperoxide, which produces phenol and acetone:
H3C CH H3C H3C H3C
O2 cat.

OH N2

CH

CH3
O2

Oxidation, run at 350C, enables high rates of conversion of benzene (27%) to be achieved with 98% selectivity with respect to phenol (Panov, 2000). Despite this, the use of N2O can hardly be applied on a large scale; in fact, its supply is not adequate for phenol synthesis and its production for this purpose (e.g. by pyrolysis of ammonium nitrate or by selective oxidation of ammonia) would be too costly. Thus, the only plausible scenario for the application of the Solutia process is that of foreseeing its implementation side-by-side plants for adipic acid (an important intermediate in the nylon cycle; Bellussi and Perego, 2000). N2O co-produced in the latter, instead of being eliminated, could be used for the production of phenol, practically at zero cost. A viable alternative to N2O is hydrogen peroxide:
OH H2O2
cat.

CH2 OOH C
H2SO4

H2O

OH

O C H3C CH3

The first reaction, the alkylation of benzene, is catalyzed by acids: traditionally aluminium trichloride or supported phosphoric acid. Recently, however, innovative processes have been developed that foresee the substitution of these acids with appropriate zeolites which have a very reduced environmental impact (Bellussi and Perego, 2000). Also thanks to these improvements, the cumene process is fully satisfactory in many aspects. Nevertheless, any increase in the productive capacity of phenol implies the need to find commercial outlets for a corresponding quantity of acetone. A first route to avoid such a need consists in recycling the acetone, transforming it back into propylene or, more briefly, into isopropyl alcohol, which is also able to alkylate the benzene (Girotti et al., 2003). The processes of direct oxidation of benzene into phenol provide a more effective and definitive solution, which would completely eliminate the coproduction of acetone. The first of these processes is known as the Solutia process, from the name of the American company that recently finalized it in collaboration with the Boreskov Institute of Catalysis of Novosibirsk (Russia), using nitrous oxide, N2O, as

The main problem encountered in the direct oxidation of benzene to phenol (not only with hydrogen peroxide) is the inadequate kinetic control of the reaction: phenol, in fact, is oxidized more readily than benzene and, therefore, rather than being accumulated, is transformed into a whole series of further oxidized products (catechol, hydroquinone, benzoquinone, etc.), up to the complete degradation of its cyclic structure or the formation of polymeric tars. Indeed, the ease with which phenol is oxidized with hydrogen peroxide is exploited in the production of mixtures of catechol and hydroquinone (see below). Therefore, to obtain phenol selectively by the oxidation of benzene, strategies must be found that enable the reactions of overoxidation to be slowed down, allowing the desired product to be accumulated. The first step in this direction was taken by George Olah who, working with extremely concentrated hydrogen peroxide (98%) in a superacid environment (FSO3H-SbF5 1:1) at 78C, obtained phenol with a yield of 54% based on both hydrogen peroxide and on benzene (Olah and Ohnishi, 1978). Although the prohibitive reaction conditions meant that this work can only be of historical value, the idea that in a superacid medium phenol could be protonated (and hence deactivated towards subsequent oxidations) is most definitely noteworthy. A second approach towards reducing the rate at which phenol is oxidized arises from the consideration

676

ENCYCLOPAEDIA OF HYDROCARBONS

OXIDATION PROCESSES

that many consecutive oxidation problems are effectively handled and resolved by biological systems, which succeed in segregating the catalyst and the reaction product in different environments. In this way, the product substantially no longer has access to the catalytic site and consequently can no longer be transformed, and is accumulated without any difficulty. A typical example is constituted by several enzymes of the oxygenase class, in which the active sites are buried into deep hydrophobic pockets easily accessible to lipophilic substrates, whereas the more hydrophilic products, once released, no longer return there. This important feature of biological systems can be reproduced, to some extent, by adopting a reaction medium consisting of two mutually immiscible liquid phases, one aqueous and the other one organic; the first one contains the catalyst and the second is very efficient in extracting the phenol produced, conveying it away from the catalyst and thereby minimizing further oxidations. The best results have been obtained using acetonitrile. In fact, in the presence of benzene, a heterogeneous system is formed by water and acetonitrile with two phases, one mainly aqueous and the other mainly organic. The concentration of the benzene in the mainly aqueous phase is more than quadrupled. Simultaneously, a large part (85%) of the phenol formed is extracted in the organic phase and is thus protected against the overoxidation reactions. The catalytic system consists of an acid (e.g. trifluoroacetic or sulphuric), iron sulphate and a ligand able to modulate its activity and its selectivity: 2-methylpyrazine-4-carboxylic acid N-oxide:
O N HOOC N CH3

The most recent developments in the oxidation of benzene to phenol with hydrogen peroxide regard the use of a particular solvent (sulpholane) in combination with TS-1. Oxidizing benzene under these conditions in the most commonly used solvents, the selectivity for phenol decreases very rapidly with the increase in conversion and, typically, falls below 50% already with benzene conversions of around 3%. This drop is strongly reduced when sulpholane is used as the solvent. In this case, the selectivity for phenol remains higher than 80% even with benzene conversions of 8%. The by-products are catechol (7%), hydroquinone (4%), 1,4-benzoquinone (1%) and tars (5%; Balducci et al., 2003). Probably sulpholane forms a complex with phenol, by means of a hydrogen bond, once this has been desorbed by the catalyst:
O O H O S

With this system it has been possible to convert 8.4% of the benzene, obtaining phenol with a selectivity of 97% with respect to benzene and of 88% with respect to hydrogen peroxide (Bianchi et al., 2000).
Fig. 14. Direct

The complex is definitely larger in size than phenol and so, although it is relatively labile, it is in any case able to delay, to some extent, a new entry of the phenol into the cavities of the zeolite. In this way the sulpholane modifies, by reducing it, the coefficient of distribution of the phenol between the volume of the intrazeolite cavities and the external solvent, thereby delaying its subsequent oxidation. A further improvement in the performance of the catalytic system was obtained by subjecting the titanium-silicalite to a post-synthesis treatment with aqueous solutions of NH4HF2 and hydrogen peroxide, at 80C. The treatment removes part of the titanium and also substantially modifies the environment of part of the titanium not removed. The new catalyst thus obtained (TS-1B) enables a 94% selectivity to phenol to be reached, with benzene conversions of around 9% (Balducci et al., 2003). The selectivity of the whole process could be further improved by converting the by-products into phenol by treating them with hydrogen, in water, over

oxidation of benzene to phenol with hydrogen peroxide catalyzed by TS-1B. HDO, hydrodeoxigenation.

H2O

phenol H2O benzene

NaOH

H2SO4

H2

oxidation

distillation

sulpholane purification

diphenols recovery phenol

HDO

H2O2

sulpholane/benzene

benzene make-up

H2O/ Na2SO4

tars

VOLUME II / REFINING AND PETROCHEMICALS

677

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

commercial catalysts based on nickel and molybdenum oxides (Bianchi et al., 2004). A simplified block diagram of the process is shown in Fig. 14. The oxidation of benzene is carried out continuously, between 95 and 110C and at 6 bar, over a fixed bed containing TS-1B. At the end, the unconverted benzene, the water and the phenol are removed through distillation, while the sulpholane used as a solvent is purified by extracting the by-products (hydroquinone, catechol and tars) with aqueous soda and then recycled. The process permits the complete conversion of the hydrogen peroxide and conversion of more than 15% of the benzene, with a selectivity to phenol higher than 97% based on benzene and 71% on hydrogen peroxide. Conversion of the benzene obtained in direct oxidation with hydrogen peroxide, although low, is fully comparable with that of the traditional process, in which the overall conversion after the two stages does not exceed 8.5% and, more often, is around 6%. While it is still too early to know whether the results described will lead to a new process for the industrial production of phenol without the coproduction of acetone, undoubtedly it lays the foundations for it.
Oxidation of phenol to hydroquinone and catechol

resorcinol continues to be an exception, as it is still obtained by the alkaline fusion of m-benzendisulphonic acid or by the oxidation of m-diisopropylbenzene.
Direct hydroxylation of phenol with hydrogen peroxide

The direct hydroxylation of phenol with hydrogen peroxide was introduced during the 1970-1980s and was a major development from the standpoint of environmental safeguard. The first commercial processes were developed by Brichima and Rhone-Poulenc. Two products are obtained, catechol and hydroquinone, in a ratio that depends both on the process and on the working conditions. The main by-products are water and tars, the latter being possibly reused for the plants energy requirements. In the Brichima process, no longer in operation for about two decades, the hydroxylation of the phenol took place with a radical-type mechanism:
H2O2 Fe2 OH .OH OH .OH OH OH Fe3

H OH OH OH

In 2002, the production capacity of hydroquinone and of catechol in the developed countries amounted to about 50,000 and 32,000 t/y, respectively. The production of the former is mainly driven by the demand of the photographic industry. That of catechol has undergone an appreciable development since the 1970s, with the introduction on the market of synthetic vanillin and other artificial aromas. In addition, both are used for the production of various antioxidants and inhibitors of polymerization. In the past, diphenol synthesis processes foresaw the transformation of pre-existing functional groups, through a succession of stoichiometric reactions. The demand for hydroquinone for the budding photographic industry at the beginning of the Twentieth century, for example, was first satisfied by reducing with sulphur dioxide the benzoquinone produced by aniline oxidation with chromic acid. Although the environmental compatibility of the process was subsequently improved, by substituting chromic acid with manganese dioxide, the quantity of by-products continued to be far greater than that of hydroquinone. It is likely that the aniline oxidation process is still being applied in some developing countries. Also an alternative method, the peroxidation of p-diisopropylbenzene, is based on a series of stoichiometric reactions. It was not until the 1970-1980s that some catalytic processes using hydrogen peroxide as the oxidant were developed and became part of industrial practice. The production of

H OH

H2O2

.OH H2O

This was a typical chain reaction in which iron(II) and cobalt(II) played the role of radical initiators, generating a hydroxyl radical. The propagation reaction, in which the radical species responsible for attacking the aromatic nucleus was also regenerated, consisted mainly in the oxidation of the cyclohexadienylic intermediate by the hydrogen peroxide. In this phase, the intervention of the metal ions, present at the level of ppm, was overall regarded as negligible (Maggioni and Minisci, 1977). In the Brichima process, the chief product was catechol, obtained in a ratio of about 2.0-2.3 to hydroquinone. To minimize reactions of consecutive oxidations, which produce tars, it was necessary to use a large excess of phenol with respect to the oxidant, limiting its conversion to less than 25%. This entailed having to separate considerable quantities of phenol by distillation from the tars and to recycle them. Selectivity with respect to hydrogen peroxide, the more costly of the two reagents, slightly exceeded 60%, no longer sufficient with the passing of time to guarantee an economic advantage to the process as compared with others. In the Rhone-Poulenc process (Varagnat, 1976), the mechanism is heterolytic and the catalyst is a mixture of a strong mineral acid (H2SO4, HClO4) with phosphoric acid. The latter acts as a sequestering agent of metal

678

ENCYCLOPAEDIA OF HYDROCARBONS

OXIDATION PROCESSES

Table 4. Yields of various hydroxylation processes of phenol with hydrogen peroxide Catalyst TS-1 Co2+, Fe2+ (radical) Acid (H+)

Ortho/Para
0.5-1.3 2.0-2.3 1.2-1.5

Conversion of phenol (%) 30 9 5

ON

YIELD BASED H2O2 (%) 82 66 85-90

YIELD BASED ON PHENOL (%) 92 79 90

impurities that may be present, thereby preventing the establishment of undesired radical chains:
H3O OH H2O2 H O O H OH OH
OH OH HO H33O

H2O H3O2 OH H H OH

H2O

To favour selectivity on phenol and to diminish the formation of tars, the process operates with a large excess of phenol greater than in the Brichima process , so that conversion is limited to some 5%. Selectivity with respect to hydrogen peroxide is instead higher, with values of over 85%. The hydroquinone/catechol ratio can vary between 1.2 and 1.5, and is regulated by the choice of the acid and by the possible addition of additives, giving the process a certain flexibility to meet market demands.
Direct hydroxylation of phenol with hydrogen peroxide catalyzed by TS-1

Hydroxylation of phenol was the first commercial application of TS-1, accomplished in 1986 (Romano et al., 1990). This is a continuous process of slurry type; EniChem set up a production unit with a capacity of 10,000 t/y on this basis. The strength of this process lies in its yields, appreciably higher than those of other direct hydroxylation processes (Table 4). TS-1, in fact, makes it possible to operate at lower phenol/oxidant ratios and therefore with higher phenol conversion, without selectivity suffering. The process is thus characterized by more efficient use of a relatively costly reagent such as hydrogen peroxide, and by lower costs of phenol separation and recycling. In addition to being industrially viable, the development of the process had the merit of attracting interest in the structural and catalytic properties of TS-1, with two results: other zeolites were discovered able to catalyze oxidations (see again Table 3), and a great

deal of information was acquired on the hydroxylation of phenol, although this was not always reliable due to the uncertainties that sometimes existed regarding the purity of the catalyst used. The most reliable information includes: Solvent: water, acetone-water, methanol. Temperature: 80-100C. H2O2/phenol (molar): 0.25-0.35. Conversion of phenol (H2O2): 20-30% (100%). Selectivity of phenol (H2O2): 90-95% (80-90%). Catechol/hydroquinone ratio: 0.5-1.3. The choice of solvent strongly influences the performance of the process, first and foremost the yields. Generally the acetone-water mixture is preferred, but high yields and selectivities have been reported also with the use of methanol. The catechol/hydroquinone ratio is the other parameter influenced by the choice of solvent. This can vary between 0.5 and 1.3, reaching the highest values in acetone and the lowest ones in methanol. High selectivity depends partly on the choice of optimal operating conditions, but even more so on the purity of the TS-1 (i.e. the absence of titanium in extraframework position). The presence of amorphous titanium-silicates or of titanium dioxide, in the form of either amorphous nanoparticles or crystalline anatase, increases the incidence of the decomposition of hydrogen peroxide and of tar formation. Other zeolitic catalysts have been studied as possible substitutes of titanium-silicalite, but with disappointing results. Ti-b has less activity and its selectivity on hydrogen peroxide is comparable to and likewise insufficient as that of radical catalysts. Ti-MOR seems to possess good activity and selectivity characteristics, but to date has not been used in industrial production. Ammoximation
Industrial production of nylon

Polyamide fibres, commonly called nylon, are one of the earliest and most important typologies of

VOLUME II / REFINING AND PETROCHEMICALS

679

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

adipic acid O HO C O H2N hexamethylenediamine NH2 C OH O H N N H Nylon 6,6

NH3 NH3 oxidation O2

preparation of ammonium carbonate N2O3 (NH4)2CO3 CO2

NH3

CO2

[C
O

[n

preparation of ammonium nitrite (NH4)NO2

H O NH e-caprolactam

SO2

N O Nylon 6

[n

NH3

preparation of hydroxylamine sulphate

NH2OH . H2SO4 cyclohexanone preparation of cyclohexanone oxime oxime (NH4)2SO4 2.8 kg/kg

Fig. 15. Various types of nylon and their synthesis.

NH3

synthetic fibres (Petrini et al., 1996; Bellussi and Perego, 2000). Nylon 6,6, synthesized by William H. Carothers at the DuPont company in the 1930s, was the first to be placed on the market. It is produced from adipic acid and hexamethylenediamine. A different type of nylon, called Nylon 6, developed in Germany by I.G. Farben and marketed under the name of Perlon, is obtained by ring opening polymerization of e-caprolactam (Fig. 15). e-Caprolactam, the world productive capacity of which amounts to about 4106 t/y, is obtained industrially mainly from benzene, through a complex series of transformations. The benzene can be hydrogenated to cyclohexane which can be oxidized to cyclohexanone; alternatively, cyclohexanone can be obtained by the hydrogenation of phenol, which is also produced from benzene. In its turn, cyclohexanone is transformed into its oxime and the latter into e-caprolactam:
OH OH O N O NH

oleum NH3

preparation of caprolactam

(NH4)2SO4 1.6 kg/kg

e-caprolactam
Fig. 16. Raschig process for production

of e-caprolactam.

20 O NH2OH.H2SO4 2NH3 (NH4)2SO4 OH N H2SO4 2NH3 O NH N

OH

H2O

(NH4)2SO4

The conventional technology used in these last two stages of production raises the problem of a considerable coproduction of a salt, ammonium sulphate, which is today of extremely low commercial value. Moreover, the process is penalized by the complexity of the cycle connected with the inorganic raw materials used and with the synthesis of hydroxylamine, as well as by the problems linked with the emission of nitrogen oxides (NOx) and sulphur oxides (SOx).

In the Raschig process for the synthesis of e-caprolactam (Fig. 16), hydroxylamine sulphate is produced from NH3, CO2 and SO2 through a complex series of operations; these include the synthesis of NOx (obtained by combustion of ammonia in air) and of ammonium carbonate (obtained by the reaction of ammonia and CO2), their combination to form ammonium nitrite and the subsequent reduction of this with SO2. In the synthesis of the oxime, about 2.8 kg of ammonium sulphate is coproduced for every 1 kg of product. In the following conversion step of oxime to caprolactam, by means of the Beckmann rearrangement conducted in the presence of oleum,

680

ENCYCLOPAEDIA OF HYDROCARBONS

OXIDATION PROCESSES

another 1.6 kg is coproduced, making a total of 4.4 kg of ammonium sulphate per 1 kg of caprolactam. This process has undergone constant improvement to reduce the coproduction of ammonium sulphate, but with only partial results, or involving considerable complexity in the operative cycle. Significant examples of this trend are the BASF and DSM processes, which exploit the reducing action of hydrogen in the presence of palladium catalysts for the production of hydroxylamine. The rearrangement step of the oxime into caprolactam has also formed the object of intense studies aimed at cutting out the use of oleum and the coproduction of sulphate. Recently, Sumitomo has industrialized a vapour-phase catalytic process using a solid catalyst of a zeolitic nature instead of oleum, thus enabling the coproduction of salt to be completely avoided. Other forms of synthesis have tackled the problem of the production of oxime, or of caprolactam directly, in a completely different way. For example, in the Toray process, the photonitrosation of the cyclohexane to oxime takes place; the SNIA process foresees the oxidation of toluene to benzoic acid followed by hydrogenation to hexahydrobenzoic acid, and the conversion of the latter to caprolactam with nitrosyl sulphuric acid. These processes have had limited commercial success. Other alternative routes based on butadiene, via oxidative carbonylation (DSM) or hydrocyanation (BASF/DuPont), are at present in the final stage of development.
vent to treatment catalyst make-up H2O solvent make-up

Ammoximation of cyclohexanone

In this panorama of unresolved problems, the ammoximation process represents a radical innovation (Roffia et al., 1989; Petrini et al., 1996). The term ammoximation indicates the production of oxime directly from ammonia. In fact, the process is based on a catalytic reaction between cyclohexanone, ammonia and hydrogen peroxide, and completely eliminates, on the one hand, the problems linked with the production and use of hydroxylamine and, on the other, the coproduction of sulphates:
OH O NH3 H2O2 cat. TS-1 N 2H2O

NH3 H2O2 cyclohexanone

spent catalyst
Fig. 17. EniChem process for production

aqueous oxime to purification

of cyclohexanone oxime.

The catalyst in the process is TS-1. The cyclohexanone ammoximation reaction with ammonia and hydrogen peroxide (or oxygen) had already been described earlier, but scanty yields have been obtained. Only the use of TS-1 as the catalyst, patented for the first time in the 1980s by Paolo Roffia and co-workers, enabled a breakthrough to be obtained, allowing EniChem to develop the process on an industrial basis. In this process (see schematic diagram in Fig. 17), the reaction is carried out continuously in the liquid phase in a stirred reactor, in the presence of the catalyst dispersed in slurry in the reaction medium at a concentration of 2-3% in weight. The reaction is typically conducted between 80C and 90C, at a slight overpressure, supplying cyclohexanone, NH3 and aqueous H2O2 in molar ratio 1.0:2.0:1.1 in the reaction solvent consisting of water and tert-butanol, with a residence time of around 1.5 h. The reaction product is separated from the catalyst, through filters soaked in the reactor, and conveyed to a rectifying column for recovery of the unconverted excess ammonia and solvent (water/tert-butanol azeotrope), which are recycled in the reaction. The aqueous solution of crude oxime obtained from the bottom of the column is conveyed to the purification unit for recovery of the cyclohexanone oxime with the required purity, to be used in the subsequent rearrangement to caprolactam. During the process, the catalyst must undergo periodic purging and make-up operations since, in the presence of ammonia, the siliceous structure of TS-1 slowly dissolves, with loss of weight of the catalyst accompanied by migration of the titanium to the outside surface of the solid and loss of catalytic activity. Operating under these conditions, the conversion of the cyclohexanone is nearly complete (99.9%) and the selectivity to the oxime based on cyclohexanone is

VOLUME II / REFINING AND PETROCHEMICALS

681

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

Fig. 18. Mechanism

of ammoximation reaction catalyzed by TS-1.

H2O(NH3) NH2OH O O Ti O O H H O O OH2(NH3) OH Ti OH2(NH3) O Ti O O O H2O

H2O(NH3) O O OH2(NH3) OH Ti NH2OH O OH2(NH3) OOH Ti NH3(H2O) O O O

O O

H2O2

higher than 98%; the yield of oxime based on H2O2 supplied is about 94%. The main inorganic by-products, derived from the oxidation of the ammonia or the decomposition of the H2O2, are N2, N2O, O2, nitrites and ammonium nitrates. The main organic by-products stemming from the cyclohexanone include azine, cyclohexenylcyclohexanone (the product of aldol condensation of the cyclohexanone), nitrocyclohexane and cyclohexenonoxime (the last two derived from consecutive reactions on the oxime). These by-products derive from competitive reactions in which the active sites of titanium present in TS-1 are not involved. Basically, two reaction mechanisms have been proposed. The first one foresees the formation of imine as an intermediate, by reaction of the cyclohexanone with the ammonia:
OH O
NH3

able to catalyze the production of hydroxylamine from NH3 and H2O2 with good yields, and bulky ketones unable to diffuse in the pores of the catalyst, such as 4-tert-butylcyclohexanone, are transformed, with good yields, into the corresponding oximes. More in detail, the mechanism proposed for the formation of hydroxylamine on the basis of spectroscopic studies foresees that, in an aqueous ammoniacal medium, the tetrahedrally coordinated titanium atoms present in the structure of TS-1 are able to coordinate up to two other ligands (H2O or NH3), assuming an octahedral coordination. Through the action of hydrogen peroxide, a species would then be formed characterized by the simultaneous presence of ammonia and hydroperoxide as ligands, which would then lead to the formation of hydroxylamine with the regeneration of the catalytic centre (Fig. 18).
Salt-free production of caprolactam

NH
H2O2 cat. TS-1

The second one foresees the formation of hydroxylamine by the action of the hydrogen peroxide on the ammonia:
O

OH N

NH3

H2 O2 cat. TS-1

NH2OH

The latter appears more likely (Zecchina et al., 1992) since TS-1, in the absence of cyclohexanone, is

The development of the ammoximation process has simplified the very complex part of the caprolactam production technology linked to the preparation of derivatives of hydroxylamine and to the synthesis of cyclohexanone oxime, completely avoiding, at this stage, the coproduction of ammonium sulphate and the emission of SOx and NOx. In turn, the catalytic vapour-phase process, developed in Japan by Sumitomo, makes it possible to avoid the coproduction of ammonium sulphate in the rearrangement stage of oxime to caprolactam (Ichihashi and Sato, 2001). The process is run with the oxime dissolved in methanol, at 300-400C and at approximately 1 atm, in a fluidized bed reactor that

682

ENCYCLOPAEDIA OF HYDROCARBONS

OXIDATION PROCESSES

permits frequent regeneration of the catalyst (necessary due to its rapid deactivation caused by the depositing of organic tars). The catalyst is a zeolite of MFI structure with a very low aluminium (and other hetero-elements) content, basically a silicalite-1. Its use, associated with that of alcohol as a component of the reaction medium, has enabled complete conversions of oxime and selectivities to caprolactam of more than 95% to be achieved. These performances, although slightly inferior to those of the conventional rearrangement conducted with oleum, in which the yield exceeds 99%, have been reckoned suitable for industrial development due to the absence of the coproduction of any salts. The combined process bringing together the ammoximation and catalytic transposition technologies results in a totally salt-free production of caprolactam. This process was industrialized for the first time in Japan by Sumitomo, which, on the basis of an agreement with EniChem, set up a plant with a capacity of 60,000 t/y. The combined process, apart from allowing salt-free production, eliminates all gaseous emissions of NOx and SOx and results in a significant reduction in investment and operating costs. This new technology will certainly play a primary role in caprolactam production processes based on cyclohexanone. Conclusions and perspectives The prospects of using hydrogen peroxide, even for large-scale productions, seem encouraging. An ammoximation plant has come into production in Japan, one or more pilot plants for the epoxidation of propylene are operating in various chemical companies, and considerable research efforts are being made in many industrial laboratories. Nevertheless, it cannot be denied that obstacles exist, not to be underestimated, which are inherent in the current method of producing hydrogen peroxide and which hinder its diffusion in petrochemistry. The anthraquinone process, in fact, is based on a complex technology that requires considerable investments. Moreover, it is anticipated that a propylene oxide plant, based on the TS-1/H2O2 technology, will require a hydrogen peroxide unit of a size never experimented. Despite such considerations, however, some chemical companies have recently announced the setting up of large-scale propylene oxide plants that are based on this technology. The need to exceed the limits of the anthraquinone process gave fresh impetus to the studies on the direct synthesis of hydrogen peroxide, as previously mentioned. At the same time, it led to exploring other alternatives, such as process integration and the in situ

generation of hydrogen peroxide, made possible by the molecular sieve properties of TS-1. On this basis, the direct use of the working solution of the anthraquinone process for the epoxidation of propylene has been studied (Clerici and Ingallina, 1996):
OH R O2 OH O
TS-1

O HHOO 22 OO

In the process, formally, the oxidant is molecular oxygen, with all the advantages thereof with respect to the use of hydrogen peroxide. In effect, however, the latter is produced in the reaction environment by the oxidation in situ of the alkylanthrahydroquinone. Successively, it diffuses to the active sites within the pores, where it is immediately consumed in the epoxidation reaction, with the mechanism already illustrated in Fig. 9. The feasibility of the process is based on the impossibility of the alkylanthraquinone and the other components of the working solution to diffuse inside the pores, where they could undergo oxidative degradation processes or interfere with the redox mechanisms. Their molecular dimensions are, in fact, larger than the opening of the catalysts pores. However, it should be emphasized that the scheme envisaged does not enable the complexity of the anthraquinone process to be side-stepped, but just to eliminate the stage relative to the separation, purification and concentration of the hydrogen peroxide. A variant foresees that the two processes, the production of hydrogen peroxide and epoxidation, remain substantially separate, but that the epoxidation solvent (aqueous methanol) is used also for extracting hydrogen peroxide from the working solution (Clerici and Ingallina, 1996). The principle foreseeing the use of aqueous methanol as the solvent for the direct synthesis of hydrogen peroxide is quite similar; it is targeted on obtaining a solution that can be supplied to the epoxidation reactor without further treatments, except for the removal of any additives (Paparatto et al., 2003b). In general, this method (like the others mentioned above) can also be used for other TS-1-catalysed oxidations, and not only in the case of propylene epoxidation. Then there is a second category of studies, to which mention has already been made, on the possible use of a mixture of hydrogen and oxygen directly in the oxidation reactor. For this purpose, it is necessary to finalize a bifunctional catalyst that possesses both the catalytic centres necessary for synthesis of the hydrogen peroxide, and the active centres for the

VOLUME II / REFINING AND PETROCHEMICALS

683

SYNTHESIS OF INTERMEDIATES FOR THE PETROCHEMICAL INDUSTRY

oxidation reaction. TS-1, on which metallic palladium has been supported, is an adequate catalyst (Clerici and Bellussi, 1993; Meiers et al., 1998). An interesting application of mixed oxides regards the deep desulphuration of motor fuels. The large size of the sulphur molecules, in this case, entails the use of mesoporous mixed oxides, instead of zeolites, which are microporous. Consequently, for the reasons already discussed, the oxidant cannot be hydrogen peroxide, but rather an organic hydroperoxide, a solution that, moreover, is best fitted for the actual set-up of the refinery. In principle, oxidants such as tert-butyl hydroperoxide, tert-amyl hydroperoxide, (1-phenyl)ethyl hydroperoxide or other such hydroperoxides are obtainable by oxidation with air of streams present in the refinery. Patents filed by various oil companies show the interest taken in this method, as an alternative to the more traditional process of deep hydrotreatment.

References
Baerlocher C. et al. (2001) Atlas of zeolite framework types, Amsterdam, Elsevier. Balducci L. et al. (2003) Direct oxidation of benzene to phenol with hydrogen peroxide over a modified titanium silicalite, Angewandte Chemie. International Edition, 42, 4937-4940. Bellussi G., Perego C. (2000) Industrial catalytic aspects of the synthesis of monomers for nylon production, Cattech, 4, 4-16. Bellussi G. et al. (1992) Reactions of titanium silicate with protic molecules and hydrogen peroxide, Journal of Catalysis, 133, 220-230. Bianchi D. et al. (1999) Biphasic synthesis of hydrogen peroxide from carbon monoxide, water, and oxygen catalyzed by palladium complexes with bidentate nitrogen ligands, Angewandte Chemie. International Edition, 38, 706-708. Bianchi D. et al. (2000) A novel iron-based catalyst for the biphasic oxidation of benzene to phenol with hydrogen peroxide, Angewandte Chemie. International Edition, 39, 4321-4323. Bianchi D. et al. (2004) European Patent 1424320 to Polimeri Europa. Boccuti M.R. et al. (1988) Spectroscopic characterization of silicalite and titanium silicalite, in: Structures and reactivity of surfaces. Proceedings of a European conference, Trieste, September 13-16, Amsterdam, Elsevier, 133-144. Clerici M.G. (1991) Oxidation of saturated hydrocarbons with hydrogen peroxide, catalysed by titanium silicalite, Applied Catalysis, 68, 249-261. Clerici M.G. (2001) The role of the solvent inTS-1 chemistry: active or passive? An earlier study revisited, Topics in Catalysis, 15, 257-263. Clerici M.G., Bellussi G. (1993) US Patent 5235111 to Eniricerche-Snamprogetti. Clerici M.G., Ingallina P. (1993) Epoxidation of lower olefins with hydrogen peroxide and titanium silicalite, Journal of Catalysis, 140, 71-83.

Clerici M.G., Ingallina P. (1996) Clean oxidation technologies. New prospects in the epoxidation of the olefins, in: Anastas P.T., Williamson T.C. (editors) Green chemistry. Designing chemistry for the environment, Washington (D.C.), American Chemical Society, 59-68. Clerici M.G. et al. (1991) Synthesis of propylene oxide from propylene and hydrogen peroxide catalyzed by titanium silicalite, Journal of Catalysis, 129, 159-167. Corma A. et al. (2001) Sn-zeolite beta as a heterogeneous chemoselective catalyst for Baeyer-Villiger oxidations, Nature, 412, 423-425. Fenton H.J.H. (1876) On a new reaction of tartaric acid, Chemical News, 33, 190. Fenton H.J.H. (1894) Oxidation of tartaric acid in presence of iron, Journal of Chemical Society, 65, 899-910. Girotti G. et al. (2003) Alkylation of benzene with isopropanol on -zeolite. Influence of physical state and water concentration on catalyst performances, Journal of Molecular Catalysis A: Chemical, 204-205, 571-579. Goor G. (1992) Hydrogen peroxide: manufacture and industrial use for production of organic chemicals, in: Strukul G. (edited by) Catalytic oxidations with hydrogen peroxide as oxidant, Dordrecht, Kluwer Academic Publishers, 13-43. Herrmann W.A. et al. (1991) Methyltrioxorhenium as catalyst for olefin oxidation, Angewandte Chemie. International Edition in English, 30, 1638-1641. Hooper G.W. (1967) US Patent 3336112 to ICI. Huybrechts D.R.C. et al. (1990) Oxyfunctionalization of alkanes with hydrogen peroxide on titanium silicalite, Nature, 345, 240-242. Ichihashi H., Sato H. (2001) The development of new heterogeneous catalytic processes for the production of -caprolactam, Applied Catalysis. A: General, 221, 359-366. Juli S. et al. (1982) Synthetic enzymes. 2: Catalytic asymmetric epoxidation by means of polyamino-acids in a triphase system, Journal of Chemical Society. Perkin Transactions I, 11, 1317-1324. Katsuki T., Sharpless K.B. (1980) The first practical method for asymmetric epoxidation, Journal of the American Chemical Society, 102, 5974-5976. Maggioni P., Minisci F. (1977) Catechol and hydroquinone from catalytic hydroxylation of phenol by hydrogen peroxide, La Chimica e lIndustria, 59, 239-242. Maspero F., Romano U. (1994) Oxidation of alcohols with H2O2 catalyzed by titanium silicalite-1, Journal of Catalysis, 146, 476-482. Meiers R. et al. (1998) Synthesis of propylene oxide from propylene, oxygen, and hydrogen catalyzed by palladiumplatinum-containing titanium silicalite, Journal of Catalysis, 176, 376-386. Milas N.A. (1937) The hydroxylation of unsaturated substances. III: The use of vanadium pentoxide and chromium trioxide as catalysts of hydroxylation, Journal of the American Chemical Society, 59, 2342-2344. Milas N.A., Sussman S. (1936) The hydroxylation of the double bond, Journal of the American Chemical Society, 58, 1302-1305. Millini R., Perego G. (1996) Structural evidence for isomorphous substitution in microporous materials and methods of characterization. The case of titanium silicalite, Gazzetta Chimica Italiana, 126, 133-140.

684

ENCYCLOPAEDIA OF HYDROCARBONS

OXIDATION PROCESSES

Neurock M., Manzer L.E. (1996) Theoretical insights on the mechanism of alkene epoxidation by H2O2 with titanium silicalite, Chemical Communications, 1133-1134. Olah G.A., Ohnishi R. (1978) Oxyfunctionalization of hydrocarbons. 8: Electrophilic hydroxylation of benzene, alkylbenzenes, and halobenzenes with hydrogen peroxide, Journal of Organic Chemistry, 43, 865-867. Panov G.I. (2000) Advances in oxidation catalysis. Oxidation of benzene to phenol by nitrous oxide, Cattech, 4, 18-32. Paparatto G. et al. (2003a) European Patent 1307399 to EniEnichem. Paparatto G. et al. (2003b) US Patent 6541648 to Polimeri Europa. Petrini G. et al. (1996) Caprolactam via ammoximation, in: Anastas P.T., Williamson T.C. (editors) Green chemistry. Designing chemistry for the environment, Washington (D.C.), American Chemical Society, 33-48. Ricci M. (1996) Electrophilic activation of hydrogen peroxide. Some recent applications in organic synthesis, in: Proceedings of the Seminars in organic synthesis. XXI Summer School A. Corbella 17-21 June, Milano, Societ Chimica Italiana, 247-266. Roffia P. et al. (1989) Cyclohexanone ammoximation. A break through in the 6-caprolactam production process, in: New developments in selective oxidation. Proceedings of an international symposium, Rimini (Italy), 18-22 September, 43-52. Romano U. (2001) Ossido di propilene. Nuova tecnologia produttiva, La Chimica e lIndustria, 83, 30-31.

Romano U. et al. (1990) Selective oxidation with ti-silicalite, La Chimica e lIndustria, 72, 610-616. Taramasso M. et al. (1983) US Patent 4410501 to Snamprogetti. Varagnat J. (1976) Hydroquinone and pyrocatechol production by direct oxidation of phenol, Industrial & Engineering Chemistry. Product Research and Development, 15, 212-215. Venturello C. et al. (1983) A new, effective catalytic system for epoxidation of olefins by hydrogen peroxide under phasetransfer conditions, Journal of Organic Chemistry, 48, 3831-3833. Venturello C. et al. (1985) A new peroxotungsten heteropoly anion with special oxidizing properties. Synthesis and structure of tetrahexylammonium tetra(diperoxotungsto)phosphate(3), Journal of Molecular Catalysis, 32, 107-110. Zecchina G. et al. (1992) Ammoximation of cyclohexanone on titanium silicalite. Investigation of the reaction mechanism, in: Proceedings of the 10 th International congress on catalysis, Budapest, 19-24 July, 719-728.

Mario G. Clerici
EniTecnologie San Donato Milanese, Milano, Italy

Marco Ricci Franco Rivetti


Polimeri Europa Novara, Italy

VOLUME II / REFINING AND PETROCHEMICALS

685

Vous aimerez peut-être aussi