Environmentally benign multiphase catalysis with dense phase carbon dioxide

Applied Catalysis B: Environmental 37 (2002) 279–292

Environmentally benign multiphase catalysis with dense phase carbon dioxide Bala Subramaniam∗ , Christopher J. Lyon, Venu Arunajatesan Department of Chemical and Petroleum Engineering, University of Kansas, Lawrence, KS 66045, USA Received 5 October 2001; received in revised form 19 December 2001; accepted 19 December 2001

Abstract Environmental concerns stemming from the use of conventional solvents and from hazardous waste generation have propelled research efforts aimed at developing benign chemical processing techniques that either eliminate or significantly mitigate pollution at the source. This paper provides an overview of heterogeneous and homogeneous catalysis in dense phase CO2 , considered a green solvent. In addition to solvent replacement, the demonstrated advantages of using dense phase CO2 include the enhanced miscibility of reactants, such as O2 and H2 which eliminate interphase transport limitations, and the chemical inertness of CO2 . Further, the physicochemical properties of CO2 -based reaction media can be pressure-tuned to obtain unique fluid properties (e.g. gas-like transport properties, liquid-like solvent power and heat capacities). The advantages of CO2 -based reaction media for optimizing catalyst activity and product selectivity are highlighted for a variety of reactions including alkylation on solid-acid catalysts, hydrogenation on supported noble metal catalysts and a broad range of homogeneous oxidations with transition metal catalysts and dioxygen as an oxidant. Through these examples, the need is emphasized for a systematic approach to research and development of supercritical carbon dioxide based processes, taking into account conventional multiphase reaction engineering principles, catalytic chemistry and phase behavior. © 2002 Elsevier Science B.V. All rights reserved. Keywords: Heterogeneous catalysis; Supercritical carbon dioxide; Kinetics; Isomerization; Hydrogenation; Alkylation; Oxidation; Hydroformylation

1. Introduction Technology Vision 2020 [1], the US chemical industry’s roadmap document for the 21st century, specifies environmentally benign processing as a critical technological challenge. Industrial growth has spawned global environmental concerns and has driven regulations at all levels. The adverse impact of current and future regulations on production costs is expected to increasingly affect the US chemical industry’s worldwide competitiveness [2]. It is gen∗ Corresponding author. Tel.: +1-785-864-2903; fax: +1-785-864-4967. E-mail address: [email protected] (B. Subramaniam).

erally believed that environmental issues cost the chemical industry about 3% of its sales revenue or US$ 10 billion each year [3], clearly driving the industry’s proactive stance in the development of environmentally benign processes. The industry also understands that alterations to processes in the early stages of the chemical production life cycle are the most cost-effective [4–6]. Dense phase carbon dioxide (which includes liquid and supercritical carbon dioxide) as a reaction medium is expected to play a significant role in environmentally benign processing. Potential benefits include replacement of traditional solvents, enhanced product selectivites that minimize waste and inherently safe reactor operation.

0926-3373/02/$ – see front matter © 2002 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 6 - 3 3 7 3 ( 0 2 ) 0 0 0 0 5 - X

280

B. Subramaniam et al. / Applied Catalysis B: Environmental 37 (2002) 279–292

Supercritical phase reactors are gaining acceptance in industry. DuPont has announced a US$ 40 million demonstration plant for supercritical polymerization [7]. A large-scale (1000 t per day) supercritical hydrogenation plant is to be commissioned by Thomas Swan and Company in the UK [8]. Numerous reviews of supercritical phase reactions have been published. During the last decade, supercritical fluids (such as supercritical carbon dioxide and supercritical water) have been increasingly explored for performing a variety of catalytic reactions, such as hydrogenations, alkylations, aminations and oxidations, to mention a few [9–12]. A recent book is devoted to this rapidly emerging field and details several classes of homogeneous and heterogeneous reactions, such as selective oxidations, hydrogenations, hydroformylations, alkylations, polymerizations and Fischer–Tropsch (FT) synthesis in which supercritical reaction media have been exploited [13]. The main differences between these previous reviews and the present one are two-fold. First, this review is limited to environmentally benign catalysis with dense phase CO2 . Secondly, providing examples, this review emphasizes the importance of systematic experimental and theoretical investigations for the rational optimization of near-critical and supercritical phase reactors. While a large body of experimental reports of heterogeneous catalytic reactions in supercritical media has been published, complementary mathematical modeling and analysis of supercritical phase reactors is lacking in the literature. While the same fundamental reaction engineering principles apply to subcritical and supercritical phase reactors, the physical properties of the supercritical reaction mixture are very sensitive to changes in temperature, pressure and composition, and require careful consideration. For example, pressure drop in a fixed-bed reactor under gas or liquid phase operation may be a secondary design consideration, but near the critical point the same pressure drop may lead to significant variations in performance measures, such as product selectivity, catalyst effectiveness factor, or even separation into two phase flow. Such effects are not always intuitively obvious. Modeling the effects of pressure-tuning on the various performance indices could be useful to interpreting experimental results. In other words, complementary mathematical models relating phase equilibrium, fluid physical properties, and product quality

will contribute to a better fundamental understanding of the underlying physicochemical processes, and are essential to the rational analysis, design and optimization of continuous supercritical phase reactors. This review focuses on the numerous reaction engineering benefits of using environmentally benign dense phase carbon dioxide as reaction medium for catalytic reactions, including the in situ decoking of heterogeneous catalysts, enhancing heat and mass transfer, and pressure-tuning the reaction medium to attain optimal operating conditions. Several examples are provided to highlight the research opportunities in and potential applications of this rapidly emerging technological area. With the help of these examples, we emphasize the need for a systematic approach to research and development of carbon dioxide based processes guided by conventional reaction engineering principles, catalyst design and phase behavior.

2. Unique properties of dense phase carbon dioxide While a number of supercritical co-solvents have been investigated (a compiled list of 26 supercritical fluids and their properties is given by Jessop and Leitner [14]), only a few of these have mild critical properties, reasonable cost, and would be considered “green” solvents. Nelson [15] provides an overview of characteristics that make solvents environmentally benign, including toxicological and regulatory issues. Carbon dioxide is often an ideal choice because it has mild critical properties (Tc = 31 ◦ C, Pc = 74 bar), is non-toxic, non-flammable, inexpensive, and unregulated. Supercritical water is also an important green solvent as reviewed elsewhere [16], but its high critical temperature (Tc = 374 ◦ C) is generally unsuitable for many organic synthesis reactions. Fig. 1 shows the typical variations of density and transport properties of carbon dioxide with pressure along a near-critical isotherm (1.02Tc ). It is clear that the physical and transport properties of the fluid can be altered drastically, from gas-like to liquid-like behavior, simply by isothermally varying the pressure around Pc , the critical pressure. For example, at slightly above the critical pressure (1.2Pc ), while the fluid possesses roughly 70% of the liquid density, the diffusivity and viscosity values are significantly

B. Subramaniam et al. / Applied Catalysis B: Environmental 37 (2002) 279–292

Fig. 1. Variation of physico-chemical properties of carbon dioxide at near-critical conditions. Temperature for µ, ρ, and Cp , 310 K; for D11 , 323 K.

better than in liquid. This behavior could be expected of any fluid or fluid mixture in the vicinity of the critical point. Thus, if a reaction mixture (excluding the solid catalyst) is slightly above its critical temperature ((1.05–1.2)Tc ) at reaction conditions, then by pressure-tuning fluid properties around the critical pressure, desirable combinations of product desorption and pore transport rates-crucial parameters that affect catalyst activity and product selectivity in heterogeneous catalytic reactions can be obtained. To attain reaction mixtures with pressure-tunable properties, the reaction temperature must be in the vicinity of the critical temperature of the reaction mixture ((1.05–1.2)Tc in general). To satisfy this criterion, the Tc of the reaction mixture may have to be suitably modified by adding an inert co-solvent, such as carbon dioxide. For this purpose, reliable knowledge of the critical phase behavior of the reaction mixture including various co-solvent candidates is essential. The operating pressure that provides the optimum fluid properties for maximizing a given performance criterion (catalyst activity, effectiveness or selectivity) is determined by systematic pressure-tuning studies. A useful alternative to supercritical carbon dioxide based reaction mixtures is a CO2 -expanded liquid solvent in which the conventional solvent is significantly replaced (up to 80%) with environmentally benign carbon dioxide. In this case, a conventional solvent such

281

as acetonitrile is volume-expanded several-fold with dense phase carbon dioxide (in which the conventional solvent is miscible). Each CO2 -expandable solvent can, in principle, generate a continuum of media ranging from the neat organic solvent to pure CO2 . Thus, the solvent properties may be varied to accommodate contrasting solubilities simultaneously, like those of oxygen and of homogeneous catalysts based on metallic elements; a large amount of CO2 favors oxygen solubility and polar organic solvents favor metal catalyst solubility. The beneficial physical properties at these limits are elegantly combined in CO2 -expanded reaction mixtures to perform a variety of homogeneous catalytic oxidations of organic compounds with a range of transition metal complexes. Reliable knowledge of phase behavior is essential for the design of experiments, processes and separations involving dense phase CO2 . Street [17] gives detailed descriptions of critical phase boundaries for binary systems. For binary mixtures of Type I classification (as per the Scott–Van Konynenburg classification, see [17]), an appropriate equation of state can usually accurately predict liquid–vapor phase transitions. For multi-component systems or binary systems that exhibit liquid–liquid immiscibility (Type II systems), predictive methods may not be reliable. For more information on experimental determination and prediction of critical phase behavior, the reader is directed elsewhere [18].

3. Attributes of supercritical phase reactors The unique properties of supercritical media have been exploited in catalysis in a variety of ways, such as: (a) enhanced desorption and transport of heavy molecules (such as coke precursors) in mesoporous catalysts alleviating pore-diffusion limitations and improving catalyst effectiveness; (b) in situ removal of primary products, stabilizing primary product selectivity; (c) eliminating oxygen or hydrogen solubility limitations in the liquid phase, thereby eliminating interphase mass transfer resistances in multiphase reaction systems; and (d) enhanced heat capacity ameliorating the problem of parametric sensitivity in exothermic fixed-bed reactors. One or more of these advantages have been demonstrated for several classes of heterogeneous (i.e. solid-catalyzed) reactions such

282

B. Subramaniam et al. / Applied Catalysis B: Environmental 37 (2002) 279–292

Table 1 Some examples of heterogeneous catalysis at sc conditions (see [10] for an exhaustive list) Reaction

Catalyst

T (K)

P (MPa)

Reference

Alkylation 1-Butene and isobutane Naphthalene and isopropyl alcohol Mesitylene and anisole with propene or propan-2-ol

H-USY zeolite, sulfated zirconia LaNaY-73, H-mordenite Alkyl sulfonic acid functionalized polysiloxane

323–413 523 433–573

3.45–15.5 20.0 10–20

[21] [22] [37]

Disproportionation Ethylbenzene to benzene, diethylbenzene

USY zeolite

717

3.6

Esterification Oleic acid and ethanol Glycerol and CO2

Biocatalyst lipozyme IM60 immobilized on duolite Zeolite and ion exchange resin

313 313

13.0 13.0

[40] [41]

Hydroformylation CO + H2 with oct-l-ene CO + H2 with propylene CO + H2 with 1-hexene

Supported Rh catalysts Rh &Rh-Fe/SiO2 or polysiloxane Insoluble Rh complex

353 473 473

12.0 17.0 17.0

[27] [42] [43]

Pd or PtDELOXAN Ni Pd, Pt and other noble metals supported on Deloxan

333–433 393–403 313–593

8.0–16.0 13.8 6.0–14.0

[32] [44] [29,33]

Pd/C Pt/␥-Al2 O3 and cinchonidine as modifier Pd/Al2 O3 Sol–gel RuCl2 X2 , X = Pme2 (CH2 )2 Si(Oet)3 Pt/SiO2

343 313–373 423–493 353 333

13.8 7.0–25.0 12.0–17.5 15.0 12.0

[34] [28] [31] [45] [46]

Co/Al2 O3 ␣-Al2 O3 supported CuI/Cu2 O/MnO2

293–493 413

8.0 14.0

[47] [48]

Pd/Al2 O3 Pt/TiO2 Iron oxide on molybdenum areogels

373 573 473–573

10.0 9.0 9.0

[49] [50] [51]

Hydrogenation Fats and oils Fats and oils Acetophenone, benzaldehyde, 2-butanone oxime, m-cresol, cyclohexanone, cyclohexanole, cyclohexene, furan, isophorone, nitrobenzene, 1-octene, 1-octyne, propionaldehyde Cyclohexene Ethyl pyruvate to (R)-ethyl lactate Double bonds of unsaturated ketone Dimethylamine to dimethylformamide Cinnamaldehyde Oxidation Toluene to benzaldehyde Propylene to propylene glycol with air/H2 O Benzyl alcohol Ethanol and acetaldehyde Methanol to dimetyl ether or methyl formate

as alkylations [19–23], aminations [24], FT synthesis [25,26], hydroformylations [27] hydrogenations [28–34] and selective oxidation [35,36], spanning a wide spectrum of chemical process industries including petrochemicals, fine chemicals, food, agricultural chemicals and pharmaceuticals. These systems are summarized in Table 1. The use of supercritical media in these examples is more than merely as a replacement for organic solvents. In virtually every case, the

[38,39]

supercritical reaction medium represents an enabling tool used to manipulate such factors as catalyst stability, product selectivity, and temperature rise in the reactor. Reactor operation with supercritical media may therefore be viewed as excellent examples of the multifunctional reactor concept. There is a common conception that supercritical reactors are hazardous because they must be operated at extremely high pressures. However, the optimal

B. Subramaniam et al. / Applied Catalysis B: Environmental 37 (2002) 279–292

operating pressure for many reactions is very often in the near-critical region (<150 bar) where the physical and transport properties are easily pressure-tuned. Also, there are a number of factors that make the use of supercritical carbon dioxide in reactors inherently safe, including the replacement of organic solvents with non-flammable CO2 , pressure-tunable heat capacity to avoid reactor runaway, and increased reaction rates to minimize hold-up. In addition, parallel processing (scaling-out as opposed to scaling-up) will lead to smaller vessels that can easily hold these pressures and minimize the risk and consequences of mechanical failures.

4. A rational approach to the analysis of supercritical phase reactors Fig. 2 shows the conventional approach for multiphase reactor selection and analysis consisting of three different but inter-related decision levels: the solid catalyst phase, reactant and energy injection/removal strategies, and finally the hydrodynamics [52]. According to this approach, one would typically start by designing the catalyst for maximum effectiveness or gas/liquid dispersion strategies (Level I, Fig. 2). Then, aspects related to how the reactants would be injected

283

(pulsed, distributed or separated feed, mixing, in situ product removal, co-current or countercurrent operation) would be addressed (Level II, Fig. 2). Once these are decided, then the type of reactor, ranging from fixed-bed (with no back-mixing) to fluidized-bed (with back-mixing) to dense phase transport reactor, would be addressed (Level III, Fig. 2). If the size of the catalyst particles is small, then a fluidized-bed (or a slurry phase) reactor rather than a packed-bed reactor would be preferable. However, fluid-bed reactor may pose a conflict if back mixing is detrimental to reaction as determined in Level II. The ultimate decision is often iterative and a compromise, dictated by considerations in each one of the three levels. To aid rational decision-making at the three levels, mathematical models are essential. The following examples illustrate how conventional multiphase reaction engineering tools can be effectively employed to develop a fundamental understanding of the underlying physicochemical processes at both the pellet and reactor levels. 4.1. In situ decoking of catalyst pellets with sc reaction media The knowledge of intrinsic reaction kinetics is essential for the rational design of reactors. The ability

Fig. 2. Conventional approach for multiphase reactor selection and analysis (taken from [52]).

284

B. Subramaniam et al. / Applied Catalysis B: Environmental 37 (2002) 279–292

to achieve non-deactivating conditions with sc reaction mixtures renders it possible to investigate the intrinsic kinetics uninfluenced by catalyst deactivation. There exists an optimal condition at which the reaction medium would possess liquid-like densities to solubilize (i.e. desorb) the coke precursors (oligomeric species) and gas-like transport properties to effectively transport the oligomeric species out of the catalyst pores. The existence of such an optimum has been demonstrated experimentally and theoretically modeled employing the Pt/␥-Al2 O3 catalyzed isomerization of 1-hexene to cis- and trans- isomers as the model reaction system [53,54]. Catalyst accessibility or effectiveness is dictated by the magnitude of the Thiele parameter, φ 2 . Higher-than-optimum pressures introduce transport limitations that diminish catalyst effectiveness factor while lower-than-optimum pressure limit desorption of the heavy hydrocarbons, adversely affecting catalyst accessibility and therefore activity. Fig. 3 shows Arrhenius plots for hexene isomerization in the sc region (i.e. non-deactivating conditions) with crushed Pt/␥-Al2 O3 catalyst particles at three different sc pressures (35, 55 and 70 bar). The effective first-order rate constants (ηk) were calculated based on measurements of steady isomerization rates in a fixed-bed reactor at different temperatures ranging from 508 to 583 K. Two regions with different slopes are evident in Fig. 3; a low temperature ‘kinetic control’ region with activation energy of 110 kJ mol−1 and a high temperature ‘pore-diffusion control’ region with

Fig. 3. Arrhenius plots for Pt/␥-Al2 O3 catalyzed hexene isomerization in scCO2 (taken from [54]).

an apparent activation energy of approximately half of the true activation energy for the reaction. Thus, consistent with modeling studies, the pressure-tuning studies clearly establish that there exists an optimum combination of liquid-like densities and gas-like transport properties that thwarts catalyst coking as well as maximizes catalyst effectiveness. It should be noted that the pressures required for maximizing catalyst effectiveness and stabilizing catalyst activity are quite moderate—of the order of tens of bars. Clearly, systematic pressure-tuning studies are essential and must indeed be exploited to establish the optimum combination of density and transport properties (for the supercritical reaction medium), which maximizes the effectiveness factor of catalyst pellets. 4.2. Enhancing primary product selectivity by mitigating pore-diffusion limitations Supercritical phase reactors have also been exploited to improve the selectivity of the primary products. Consider a series reaction network of the type A → B → C. Clearly, pore-diffusion limitations would hinder the removal of B (the desired product) from the catalyst, favoring further reaction to the ultimate product, C. The FT reaction involves the formation of olefins via a sequential chain growth mechanism starting with the methylene group. However, if the transport rates of the intermediate olefins are hindered, they could be hydrogenated to the paraffins. Yokota et al. [25] were the first to report FT investigations with sc reaction media, exploiting the gas-like transport properties and liquid-like heat capacity and solubility characteristics of such media. For FT synthesis on a cobalt catalyst, Yokota et al. report superior activity and 1-olefin selectivity in supercritical n-hexane (P c = 29.7 bar; T c = 507 K) compared to either gas (nitrogen) and/or liquid-phase (n-hexadecane) reaction media. Bukur et al. [55] report that while syngas conversion on a Fe catalyst was essentially the same whether nitrogen or propane is used as the reaction medium, the olefin selectivity was considerably higher when propane was used. Bochniak and Subramaniam [26] performed complementary experimental and modeling studies of isothermal (513 K) pressure-tuning studies of FT synthesis on a supported Fe catalyst with near-critical n-hexane as the reaction medium. The results showed

B. Subramaniam et al. / Applied Catalysis B: Environmental 37 (2002) 279–292

that the overall catalyst effectiveness increased with pressure from 35 to 70 bar. This increase is attributed to the alleviation of pore-diffusion limitations (i.e. maintenance of wider pore channels) as a result of enhanced extraction of heavier hydrocarbons from the catalyst pores by denser sc reaction medium at the higher pressures. Anderson–Schulz–Flory (ASF) plots confirm this hypothesis. At 35 bar, a shoulder develops and persists displaying an increasing slope in the C10 –C15 range, which is similar to trends reported in the literature (see e.g. Donnelly and Satterfield [56]). In contrast, the shoulder gradually disappears with time at 55 and 70 bar, resulting in a steady-state carbon number distribution that conforms to the ASF formalism with a single α value (α = 0.78). Furthermore, the 1-olefin selectivity at a given carbon number also increased with pressure from 35 to 55 bar and was virtually constant (80%) when the pressure is increased to 70 bar. These results are consistent with the conclusion of decreased pore-diffusion limitations at the higher pressures, thus minimizing the readsorption and further reaction of olefins. Clearly, the combination of complementary experimental and modeling investigations contributed to a better fundamental understanding of pressure-tuning effects on catalyst activity and intermediate product (␣-olefin) selectivity.

5. Examples of environmentally benign catalytic processing with dense phase CO2 Dense phase CO2 has been employed in a variety of heterogeneous and homogeneous catalytic processes. In this section, we discuss examples of hydrogenations, alkylation and oxidations. Applications in polymer processing [57,58], pharmaceutical processing [59,60] and biocatalysis [61,62] are reviewed elsewhere and not discussed here. 5.1. Hydrogenation Conventional solid-catalyzed hydrogenation reactions are typically carried out in multiphase reactors involving the sparging of hydrogen through a slurry of the reactant and finely powdered catalyst particles. Typically, these reactions are fast and hence their rates are limited by the low solubility of hydrogen in the

285

solvent. Other drawbacks of the conventional process include large reactor volumes and costs associated with separating the product and the catalyst from the reaction mixture. In the case of solid-catalyzed hydrogenation, solubilizing the reactants (organic substrate and hydrogen) in a single, environmentally benign solvent, such as supercritical carbon dioxide (scCO2 ) eliminates interphase mass transfer limitations. The ensuing enhanced reaction rates reduce hold-up of hazardous reactants making the process inherently safer. Further, product separation from CO2 is achieved by depressurization of the reactor effluent stream. Bertucco and co-workers [63] have studied the hydrogenation of an unsaturated ketone on Pd/alumina using scCO2 as the solvent. They measured the rate of hydrogenation of the allene at various temperatures, pressures, and solvent concentrations. From this, kinetic parameters, such as pre-exponential factor, activation energy, etc. were obtained by fitting the reaction rate to a power law. The authors observed the activation energy determined by this fit (<2 kJ mol−1 ) seemed too low for a reaction. A possible explanation is that the reaction mixture was in the two-phase region due to the addition of the reactants, despite the reaction being studied beyond the critical point of the solvent (CO2 ). This could introduce mass transfer limitations during the reaction. The parameters thus determined led to a poor fit for the axial temperature profile in the fixed-bed reactor [48]. These results emphasize the importance of having an accurate knowledge of the critical phase behavior of the reaction mixture for reliable data analysis. We systematically investigated the Pd/C catalyzed cyclohexene hydrogenation in a scCO2 medium taking into account the phase behavior of the reaction mixture, the temperature control in the reactor during operation and the effect of operating variables on catalyst deactivation [34]. For an equimolar feed of cyclohexene (0.2 gmol h−1 ) and hydrogen in 90 mol% CO2 , the maximum adiabatic temperature rise for complete cyclohexene conversion in gas-like, liquid-like and sc (138 bar and 343 K) carbon dioxide were estimated to be 170, 45, and 47 K, respectively. Indeed, at near-critical experimental conditions, the maximum temperature rise observed was 12 K at 80% cyclohexene conversion. In contrast, for the same catalytic system, axial temperature gradients exceeding 100 K were reported by Hitzler et al. [33], who

286

B. Subramaniam et al. / Applied Catalysis B: Environmental 37 (2002) 279–292

performed the reaction at temperatures of 423 K, which is much above the critical temperature of the reaction mixture (∼323 K). At such temperatures, the heat capacity is more gas-like. This observation highlights the fact that the sc media must be pressure-tuned to possess liquid-like heat capacity at the reaction temperatures to better control the temperature rise in exothermic fixed-bed catalytic reactors. Another important issue during hydrogenation in scCO2 is that of catalyst deactivation. Minder et al. [28] reported the deactivation of modified Pt/Al2 O3 catalyst during the batch hydrogenation of ethyl pyruvate to (R)-ethyl lactate in scCO2 , presumably as a result of Pt poisoning by CO formed from the reverse water-gas shift reaction between CO2 and H2 . The same was not observed when using supercritical ethane (P c = 48 bar; T c = 305.4 K) as the reaction medium. On the other hand, no CO was observed during the hydrogenation of cyclohexene in a flow reactor [34]. Stable catalytic activity was observed as long as the feed peroxides were mitigated to less than 6 ppm [34,64]. However, due to the possibility of the formation of a variety of surface species such as formates, carbonates, and CO [28] in the presence of CO2 + H2 , it is important to understand how reactor operating conditions influence the formation of CO and other possible deactivating species when CO2 is employed as a solvent. 5.2. Solid-acid catalyzed alkylation As reviewed elsewhere [65,66], numerous efforts aimed at developing solid-acid alkylation catalysts and solid-acid-based isobutane-olefin alkylation processes have been reported for more than three decades. However, to-date, none of the solid alkylation catalysts has gained acceptance in industry for one or more of the following drawbacks: rapid catalyst deactivation due to coke formation, unacceptable product quality (i.e. low alkylate fraction) and thermal degradation of catalyst during the regeneration step. Supercritical alkylation, performed with excess isobutane (P c = 36.5 bar; T c = 408 K) above the critical temperature (Tc ) and critical pressure (Pc ) of isobutane, has been reported to slow down deactivation of solid-acid catalysts such as Y-zeolites and sulfated zirconia [19,20,67]. However, at reaction temperatures exceeding 408 K, undesirable side reactions, such as oligomerization

and cracking dominate, resulting in unacceptable product quality. To realize sc reaction mixtures with pressure-tunable properties at lower temperatures (323–373 K) that favor alkylation reactions, Clark and Subramaniam [21] employed carbon dioxide (P c = 71.8 bar; T c = 304 K) as a diluent in the hydrocarbon feed. The C8 alkylate production on USY and sulfated zirconia catalysts attains a nearly steady value after a few hours on stream with scCO2 -based reaction media. However, both the butene conversion (20%) and the alkylate selectivity (around 5%) values were low suggesting the pore sizes and/or acidity in these catalysts are such that it is not possible to effectively balance the coke formation rate with the coke removal rate from the pores. The low alkylate production activity observed on these catalysts could be due to the surface acid sites whose activity is more easily maintained by the sc reaction mixture. To enhance the C8 alkylate selectivity, we employed the three-level decision-making approach (see Fig. 2) to rationally determine catalyst and reactor characteristics. To mitigate pore-diffusion limitations, we decided to use mesoporous catalysts and small catalyst particles <100 ␮m (Level I decision). To promote the pseudo first-order alkylation reaction and suppress the second-order olefin dimerization reaction, we used a stirred reactor to minimize the olefin concentration (Level II). The stirred configuration is also well suited to suspend the small catalyst particles (Level III decision). Based on this approach, we recently demonstrated that enhanced alkylate selectivity can be obtained on a SiO2 -supported Nafion catalyst (Engelhard SAC-13), which has an acidity somewhat similar to sulfuric acid (used in conventional alkylate processing) and has relatively large pores (50–70 Å mean pore diameter) when compared to zeolitic catalysts (5–10 Å pore size). Steady C8 alkylates production activity during experimental runs lasting up to 2 days was demonstrated. Fig. 4 shows the effect of pressure-tuning on butene conversion and C8 alkylates selectivity [23]. The data shown in Fig. 4 are averages of steady values taken between 10 and 25 h on stream. As the reactor pressure is decreased from roughly 170 bar to a more moderate value of 80 bar (which is nearly the critical pressure of the feed), the butene conversion is virtually unaffected; however, the C8 selectivity doubles from 40 to 80% while the C8 alkylate selectivity increases

B. Subramaniam et al. / Applied Catalysis B: Environmental 37 (2002) 279–292

Fig. 4. Pressure-tuning effect on alkylation activity on SiO2 supported Nafion® (368 K, I/O = 5, 0.05 h−1 ) (taken from [23]).

approximately three-fold from 9 to 29% at 80 bar. The decreased C8 selectivity at higher sc pressures is attributed to pore-diffusion limitations that result in the C8 compounds being converted to heavier products (C12 and higher). Catalyst characterization results showed that 80–90% of the initial SAC-13 pore volume, 70–75% of the surface area and 95% of the acid sites were preserved at the end of the runs. Our results clearly suggest that with rational design of catalyst, tailoring parameters, such as acidity and pore structure, it should be possible to further enhance the C8 alkylates selectivity. In addition to isoparaffin/olefin alkylation, Friedel– Crafts alkylation has been demonstrated on Deloxan® (polysiloxane based sulfonic acid catalyst) in scCO2 for anisole and mesitylenem [37], with stable catalyst activity for at least 15 h. Alkylation of naphthalene with alcohols has been demonstrated on a promoted Y zeolite in scCO2 , significantly reducing catalyst deactivation [68]. The potential of using dense phase CO2 in environmentally benign solid-acid catalysts for fine chemical and pharmaceutical syntheses is promising.

287

are eliminated. Furthermore, the tunable liquid-like heat capacity of supercritical fluids may be exploited to control the temperature rise accompanying these highly exothermic reactions and thereby the product selectivity. Dooley and Knopf [69] reported aerobic heterogeneous catalytic oxidation of toluene in scCO2 using a range of alumina-supported oxide (CoO and MoO3 ) and mixed-metal oxide (W, Ni) catalysts as well as on SiO2 /Al2 O3 . Toluene was effectively oxidized to benzaldehyde, benzyl alcohol and cresol isomers over a variety of catalysts. Oakes et al. [71] reported on the diastereoselective oxidation of cysteine derivatives in scCO2 using t-BuOOH and Amberlyst 15 ion exchange resin catalyst. At constant temperature (313 K), the product selectivity was sensitively tuned with pressure. The major isomer selectivity for the oxidation of CysSMe-OMe was nearly 100% at 180 bar but falls off as the pressure is increased further. The fact that no diastereoselectivity is observed in conventional solvents reveals a clear advantage of scCO2 media in inducing setereoselectivity. Recent work on the partial oxidation of alcohols over a Pd/alumina catalyst has exploited the ability to pressure-tune the properties of a supercritical CO2 -based reaction mixture [35]. Fig. 5 shows a maximum for the yield of 2-octanone is achieved during

5.3. Heterogeneous partial oxidation The use of scCO2 in heterogeneously catalyzed partial oxidations has received increased attention during the last decade [50,69,70]. Partial oxidation reactions are sensitive to optimal oxygen concentrations near the catalytic sites. By solubilizing the organic substrate and O2 in scCO2 , the gas/liquid mass transfer interface and accompanying O2 transport limitations

Fig. 5. Dependence of 2-octanone yield on total pressure: (a) (䊉) 120 ◦ C, W/F = 1.02 g h mol−1 , 2.5 mol% O2 ; (b) (䊐) 100 ◦ C, W/F = 2.03 g h mol−1 , 2.5 mol% O2 ; (c) (䉭) 120 ◦ C, W/F = 1.02 g h mol−1 , 5.0 mol% O2 . Other conditions: 5 g of 0.5% Pd/alumina, 5 mol% 2-octanol, rest CO2 (taken from [35]).

288

B. Subramaniam et al. / Applied Catalysis B: Environmental 37 (2002) 279–292

the oxidation of 2-octonol by varying pressure and oxygen composition. Yields as high as 46% for the partial oxidation products were observed; a temperature gradient of 10 K in their fixed-bed reactor was noted, suggesting that higher yields could be achieved with better heat removal. In conjunction with reactor studies, view-cell phase behavior studies proved that there was a small octanol-rich liquid phase as well as a CO2 -rich phase. Consequently, the changing solubility of oxygen in the octanol-rich phase with pressure was noted as a cause for the maximum in yield. This underscores the need for a clear understanding of phase equilibria in rationally evaluating supercritical reactors. Comparing experiments with CO2 and N2 demonstrate that dense phase CO2 facilitates a two–four-fold higher reaction rate; no maximum with pressure was observed in the case of N2 . The lack of catalyst deactivation was also attributed to the use of the supercritical reaction mixture. Baiker and co-workers [72] reported their study of propylene epoxidation over Pd-Pt/TS-1 catalyst with hydrogen peroxide (formed in situ) in a fixed-bed under high pressures. Deactivation of catalysts and changes in product distribution with time were investigated. Initial propylene oxide selectivity was 99% at 3.5% conversion, but catalyst deactivation occurred very rapidly and methyl formate became the predominant product. Through comparison, conclusion was drawn that using carbon dioxide, instead of nitrogen, had a beneficial effect on the formation of propylene oxide. Thermal analysis (TA–MS and TA–FTIR) indicated that catalyst regeneration requires oxidation at elevated temperature; washing with an organic solvent is less efficient.

[73], Leitner and co-workers [74] opened the subject of solubilizing catalysts for homogeneous catalysis in scCO2 showing that perfluoroalkyl side chains increase the solubility of their rhodium phosphine catalysts. Tumas and co-workers [75] studied the catalytic oxidation of alkenes in scCO2 , using oxygen as the oxidant and halogenated porphyrins as catalysts. Cyclohexene was oxidized in the presence of Fe(PFTPP)Cl and Fe(Br8 PFTPP)Cl under rather harsh conditions, 80 ◦ C and total pressure of 345 bar yielding five products: cyclohexene oxide, 2-cyclohexene-1-ol, 2-cyclohexene-1-one, oxabicyclo[4.1.0]heptan-2-one and 4-hydroxy-2-cyclohexene-1-one. Although turnover rate was lower than that in conventional solvent, product distribution appeared influenced by changes of pressure and temperature, suggesting relatively easy tunability of reaction conditions. Hass and Kolis [76] oxidized a number of inactivated alkenes using the same catalyst. Product distribution depended on the alkene chosen as the substrate, and the epoxide product is accompanied by the corresponding 1,2-diol in the presence of water. For alkenes having phenyl substituents, C–C bond cleavage was observed, and the rates of diol and epoxide formation are considerably faster for cis-alkene than for trans-alkene oxidations. Hass and Kolis [77] also designed and synthesized a vanadium(IV) catalyst using a Schiff base ligand for epoxidation of allylic alcohols, using a 1:1 substrate to t-BuOOH mole ratio, 45–50 ◦ C and 213 bar. The authors reported that the yields and diastereoselectivities obtained in the scCO2 medium are comparable to those obtained in traditional organic solvents.

5.4. Homogeneous catalytic oxidation

5.5. Homogeneous catalytic oxidation in CO2 -expanded solvents

Despite the well-known advantages of using scCO2 media, such as non-flammability, resistance to oxidation, pressure-tunable heat capacity, and high miscibility with oxidants (particularly dioxygen), relatively few explorations have been reported on homogeneous catalytic oxidations in scCO2 and closely related media until the past 10 years [72–77]. The majority of these studies can be grouped into alkene, alkane, and alcohol oxidations. Whereas the first studies of homogeneous solutions of transition metal compounds in scCO2 may be attributed to Wai and co-workers

Although enhancement has been observed in selectivities and conversions, significant limitations were also found in using scCO2 including (1) relatively low reaction rates compared to neat organic solvents, (2) high process pressure (of the order of hundreds of bars), and (3) the limited number of transition metal catalysts that are sufficiently soluble in CO2 without substantial structure modification. Our recent study [78] has shown that, by keeping substantial amount of conventional solvent (20–35% mole fraction)

B. Subramaniam et al. / Applied Catalysis B: Environmental 37 (2002) 279–292

in dense CO2 media, a completely homogeneous CO2 -expanded reaction mixture can be created without sacrificing desirable properties, such as dielectric constant, solvation and coordination properties. The total pressure is reduced to ∼50–90 bar, and the solubility of terminal oxidant dioxygen is 1–2 orders of magnitude higher in the CO2 -expanded media than in the neat organic solvent. Significant improvements were observed in reaction rate, product selectivity, and operational safety. We defined and determined the relationships between CO2 -expansion of a conventional solvent such as CH3 CN and catalyst solubility in the expanded phase. For a given catalyst/solvent system, the expansion level at which the catalyst precipitates from solution (termed as the homogeneous expansion limit) increases with the rise in temperature and the drop in catalyst concentration. This phenomenon can be exploited to separate the catalyst from the reaction medium by adding more CO2 and reducing the temperature in the reactor. This concept has been used in other separation schemes [79,80]. The measured O2 mole fraction in CO2 -expanded CH3 CN (V/V0 = 2) is about 2 orders of magnitude higher than in neat CH3 CN (5 × 10−4 mole fraction at 25 ◦ C and 1 bar) [81], and is of the same order of magnitude as observed in liquid CO2 [82]. We studied the homogeneous catalytic O2 -oxidation of 2,6-di-t-butylphenol, DTBP, by CO (Salen∗ ) in scCO2 , CO2 -expanded organic solvents (V/V0 = 2), and neat organic solvents under cornmon conditions. At catalyst:substrate:O2 mole ratio = 1:40:80, the turnover frequency (TOF), defined as the moles of DTBP converted per mole of catalyst per hour, was 1–2 orders of magnitude greater in the CO2 -expanded CH3 CN (P = 60–90 bar) than in scCO2 (P = 207 bar). The observed selectivity towards DTBQ (80–88%) was comparable in scCO2 and in CO2 -expanded CH3 CN with no detection of CH3 CN oxidation in either case. Similar results were obtained during cyclohexene oxidation with dioxygen using iron porphyrin catalyst [78]. In addition, an optimum in the TOF was observed by varying the CO2 :solvent ratio in the expanded phase reaction medium. These results showed that CO2 -expanded solvents complement scCO2 as reaction media by broadening the range of conventional catalyst + solvent

289

combinations with which homogeneous oxidations by O2 can be performed. Additionally, there are also dramatic safety benefits—the total pressure is of the order of 50–100 bar (much lower than typical pressures for scCO2 ), and O2 can be used as terminal oxidant with no danger of producing an explosive mixture. 5.6. Heterogeneous hydroformylation Hydroformylation involves the addition of carbon monoxide and hydrogen across a C==C double bond to produce aldehydes. The catalysts employed are of the form Hx My (CO)z Ln ; the two transition metals (M) utilized are rhodium and cobalt, and the most commonly utilized ligands (Ln ) are phosphines (PR3 where R = C6 H5 or n-C4 H9 ). To overcome the catalyst separation problem, the Rh-based catalysts have been immobilized on solid catalyst supports. However, the intrinsic hydroformylation activity of the immobilized catalyst was found to be inferior relative to its homogeneous counterpart and the catalyst was also found to be leached gradually from the support by the reaction mixture. Recently, scCO2 has been demonstrated as a reaction medium for the continuous hydroformylation of oct-1-ene on silica-supported Rh catalysts at 80 ◦ C and 12.0 MPa [27]. The stated advantages include elimination of fluid phase mass transfer resistances and the extended (up to 30 h) production of nonanal. Sellin and Cole-Hamilton [43] took advantage of the insolubility Rh complexes in scCO2 by using these catalysts as heterogeneous catalysts; this enabled the easy separation of the catalyst from the reaction mixture. In their study of the hydroformylation of hex-1-ene to heptanal, they observed that the selectivity toward the preferred linear aldehyde was higher in scCO2 than in toluene. Abraham and co-workers [42] used silica and polysiloxane supported Rh and Rh-Fe catalysts to study the effect of hydrophobic and CO2 -philic supports for the hydroformylation of propylene. The catalyst on a CO2 -philic support yielded gave the highest initial activity, which was attributed to the changes in the local environment at the catalyst surface.

6. Concluding remarks The several examples provided in this article demonstrate that the pressure-tunable density and

290

B. Subramaniam et al. / Applied Catalysis B: Environmental 37 (2002) 279–292

transport properties of supercritical CO2 -based reaction media may be exploited in heterogeneous catalysis in a variety of ways including: (a) enhanced desorption and transport of heavy molecules (such as coke precursors) in mesoporous catalysts alleviating pore-diffusion limitations and improving catalyst effectiveness; (b) in situ removal of primary products stabilizing primary product selectivity; (c) eliminating O2 or H2 solubility limitations in the liquid phase, and hence interphase mass transfer resistances, in multiphase reaction systems; (d) enhanced heat capacity ameliorating the problem of parametric sensitivity in exothermic fixed-bed reactors. One or more of these advantages have been demonstrated for several classes of reactions, such as alkylations, hydrogenations, oxidations and hydroformylations, spanning a wide spectrum of chemical process industries. In virtually every case, the sc reaction medium represents an enabling tool used to manipulate such factors such as catalyst stability, product selectivity and temperature rise in the reactor. In the field of homogeneous catalytic oxidation, the use of scCO2 has certain advantages over conventional solvents (environmentally benign, complete O2 miscibility, resistance to oxidation). However, a drawback is the high pressures (of the order of hundreds of bars) required to ensure adequate solubility of many transition metal catalysts in CO2 . Although fluorocarbon ligands are known to enhance the solubility of transition metal complexes in scCO2 they are expensive. By employing CO2 -expanded solvents as reaction media, such properties as the dielectric constant may be readily varied with mixture composition while maintaining solubility of the substrate, oxidant and catalyst in the expanded phase. There are a number of traditional solvents, such as CH2 Cl2 , CH3 CN and DMF that can be expanded with dense CO2 . The CO2 -expanded solvent mixtures represent a continuum of reaction media with different physicochemical properties that may be exploited for optimizing conversion and selectivity in catalytic oxidation systems. Further, the low process pressures and the ability to separate and recycle the catalyst and solvent make homogenous catalytic oxidation in CO2 -expanded solvents environmentally friendlier than using neat solvents and potentially more economical compared to scCO2 -based oxidation. The outlook for industrial applications of supercritical reaction media in multiphase catalysis is promis-

ing. Economic factors that dictate the viability of supercritical processes include the capital costs associated with the high-pressure equipment and operating costs associated with the compression of the supercritical media for recycling purposes. Many examples discussed in this paper show that the optimum pressure lies closer to or below the critical pressure of the reaction mixture. Clearly, the critical pressure of the solvent medium and the extent of dilution of the reactants with the solvent medium are important parameters that would influence the process economics. The announcements of a development plant for producing fluoropolymers in supercritical carbon dioxide and of a supercritical phase hydrogenation plant are strong indicators that industries are seriously considering sc processes for chemical processes. The challenge will be in demonstrating processes that simultaneously display the following attributes: selective, environmentally benign, stable and economical. Conventional reaction engineering theory provides us the approaches and tools for the rational analysis and development of sc phase reactors.

Acknowledgements This work was supported in part by the National Science Foundation (CHE-9815321; CTS-9816969) and the Environmental Protection Agency (R827034-01-0). References [1] Technology Vision 2020, The US Chemical Industry, ACS, AIChE, CMA, CCR, SOCM, December 1996. [2] Office of Technology Policy, Chemtech, August 1996, p. 46. [3] C.J. Pereira, Chem. Eng. Sci. 54 (1999) 1959. [4] D.T. Allen, K.S. Rosselet, Pollution Prevention for Chemical Processes, Wiley, New York, 1997. [5] P.T. Anastas, T.C. Williamson, Green Chemistry: Frontiers in Benign Chemical Syntheses and Processes, Oxford University Press, Oxford, 1998, p. 1. [6] J.H. Clark, Green Chem. 1 (1999) 1. [7] M. McCoy, Chem. Eng. News 14 (June) (1999) 11. [8] M. Freemantle, Chem. Eng. News 28 (May) (2001) 30. [9] P.E. Savage, S. Gopalan, T.L. Mizan, C.J. Martino, E.E. Brock, AIChE J. 41 (1995) 1723. [10] A. Baiker, Chem. Rev. 99 (1999) 453. [11] B. Subramaniam, Appl. Catal. A. 212 (2001) 199. [12] K.W. Hutchenson, in: Y.P. Sun (Ed.), Supercritical Fluid Technology in Materials Science and Engineering, Marcel Dekker, New York, in press.

B. Subramaniam et al. / Applied Catalysis B: Environmental 37 (2002) 279–292 [13] P.G. Jessop, W. Leitner (Eds.), Chemical Synthesis Using Supercritical Fluids, Wiley, New York, 1999, p. 213. [14] P.G. Jessop, W. Leitner, in: P.G. Jessop, W. Leitner (Eds.), Chemical Synthesis Using Supercritical Fluids, Wiley, New York, 1999, p. 6. [15] W.M. Nelson, in: P.T. Anastas, L.G. Heine, T.C. Williamson (Eds.), Green Chemical Syntheses and Processes, American Chemical Society, Washington, DC, 2001, p. 313. [16] P.E. Savage, Chem. Rev. 99 (1999) 603. [17] W.B. Street, in: M.E. Paulaitis, J.M.L. Penninger, R.D. Gray, P. Davidson (Eds.), Chemical Engineering at Supercritical Conditions, Ann Arbor Science, Ann Arbor, MI, 1983, p. 3. [18] M.A. McHugh, V.J. Krukonis, Supercritical Fluid Extraction, Butterworth-Heinemann, Boston, 1994. [19] L. Fan, I. Nakamura, S. Ishida, K. Fujimoto, Ind. Eng. Chem. Res. 36 (1997) 1458. [20] G. Funamoto, S. Tamura, K. Segawa, K. Wan, M.E. Davis, Res. Chem. Intermed. 24 (1998) 449. [21] M.C. Clark, B. Subramaniam, Ind. Eng. Chem. Res. 37 (1998) 1243. [22] R. Gläser, J. Weitkamp, in: M.M.J. Treacy (Ed.), in: Proceedings of the 12th International Zeolite Conference, Vol. 2, Warendale, PA, Materials Research Society, 1999, p. 1447. [23] C. Lyon, B. Subramaniam, C. Pereira, in: J.J. Spivey, B. Davis (Eds.), Catalyst Deactivation 9: Studies in Surface Science and Catalysis, Vol. 139, Elsevier, Amsterdam, 2001, p. 221. [24] A. Fischer, M. Maciejewski, T. Bürgi, T. Mallat, A. Baiker, J. Catal. 183 (1999) 373. [25] K. Yokota, Y. Hanakata, K. Fujimoto, Fuel 70 (1991) 989. [26] D.J. Bochniak, B. Subramaniam, AIChE J. 44 (1998) 1889. [27] N.J. Meehan, A.J. Sandee, J.N.H. Reek, P.C.J. Kamer, P.W.N.M. van Leeuwen, M. Poliakoff, Chem. Commun. (2000) 1497. [28] B. Minder, T. Mallat, K.H. Pickel, K. Steiner, A. Baiker, Catal. Lett. 34 (1995) 1. [29] M. Härröd, P. Møller, High pressure chemical engineering, in: P.R. von Rohr, C. Trepp (Eds.), Proc. Tech. Proc. 12, Elsevier, Amsterdam, 1996, p. 43. [30] M.G. Hitzler, M. Poliakoff, Chem. Commun. (1997)1667. [31] A. Bertucco, P. Canu, L. Devetta, A. Zwahlen, Ind. Eng. Chem. Res. 36 (1997) 2626. [32] T. Tacke, S. Wieland, P. Panster, High Pressure Chemical Engineering in: P.R. von Rohr, C. Trepp (Eds.), Proc. Tech. Proc. 12, Elsevier, Amsterdam, 1996, p. 17. [33] M.G. Hitzler, F.R. Smail, S.K. Ross, M. Poliakoff, Org. Process Res. Dev. 2 (1998) 137. [34] V. Arunajatesan, B. Subramaniam, K.W. Hutchenson, F.E. Herkes, Chem. Eng. Sci. 56 (2001) 1363. [35] G. Jenzer, M.S. Schneider, R. Wandeler, T. Mallat, A. Baiker, J. Catal. 199 (2001) 141. [36] B. Kerler, A. Martin, A. Jans, M. Baerns, Appl. Catal. A. 220 (2001) 243. [37] M.G. Hitzler, F.R. Smail, S.K. Ross, M. Poliakoff, Chem. Commun. (1998) 359. [38] G. Manos, H. Hofmann, Chem. Eng. Technol. 14 (1991) 73. [39] F. Niu, G. Kolb, H. Hofmann, Chem. Eng. Technol. 18 (1995) 278.

291

[40] R. Goddard, J. Bosley, B. Al-Duri, J. Chem. Technol. Biotechnol. 75 (2000) 715. [41] C. Vieville, J.W. Yoo, S. Pelet, Z. Mouloungui, Catal. Lett. 56 (1999) 245. [42] G. Snyder, A. Tadd, M.A. Abraham, Ind. Eng. Chem. Res. 40 (2001) 5317. [43] M.F. Sellin, D.J. Cole-Hamilton, Dalton Trans. (2000) 1681. [44] J.W. King, R.L. Holliday, G.R. List, J.M. Snyder, J. Am. Oil Chem. Soc. 78 (2001) 107. [45] O. Kröcher, R.A. Köppel, M. Fröba, A. Baiker, J. Catal. 178 (1998) 284. [46] M. Shirai, T. Tanaka, M. Arai, J. Mol. Catal. A: Chem. 168 (2001) 99. [47] K.M. Dooley, F.C. Knopf, Ind. Eng. Chem. Res. 26 (1987) 1910. [48] M. Gaffney, J.A. Sofranko, American Chemical Society Division of Petroleum Chemistry, Vol. 37, 1992, p. 1273 (preprints). [49] G. Jenzer, T. Mallat, A. Baiker, Catal. Lett. 73 (2001) 5. [50] L.B. Zhou, A. Akgerman, Ind. Eng. Chem. Res. 34 (1995) 1588. [51] C.-T. Wang, R.J. Willey, J. Non-Cryst. Solids 225 (1998) 173. [52] R. Krishna, S.T. Sie, Chem. Eng. Sci. 49 (1994) 4029. [53] S. Baptist-Nguyen, B. Subramaniam, AIChE J. 38 (1992) 1027. [54] M.C. Clark, B. Subramaniam, AIChE J. 45 (1999) 1559. [55] D.B. Bukur, X. Lang, A. Akgerman, Z. Feng, Ind. Eng. Chem. Res. 36 (1997) 2580. [56] T. Donnelly, C.N. Satterfield, Appl. Catal. 56 (1989) 231. [57] J.L. Kendall, D.A. Canelas, J.L. Young, J.M. DeSimone, Chem. Rev. 99 (1999) 543. [58] E. Kiran, in: E. Kiran, P.G. Debenedetti, C.J. Peters (Eds.), Supercritical Fluids: Fundamentals and Applications, Kluwer Academic Publishers, Dordrecht, 2000, p. 10. [59] B. Subramaniam, S. Saim, R. Rajewski, V.J. Stella, in: P.T. Anastas, L.G. Heine, T.C. Williamson (Eds.), Green Engineering, American Chemical Society, Washington, DC, 2001, p. 96. [60] F.S. Mandel, J.D. Wang, M.A. McHugh, Poly. Mat. Sci. Eng. 84 (2001) 39. [61] A.J. Mesiano, E.J. Beckman, A.J. Russell, Chem. Rev. 99 (1999) 623. [62] O. Aaltonen, in: P.G. Jessop, W. Leitner (Eds.), Chemical Synthesis Using Supercritical Fluids, Wiley, New York, 1999, p. 414. [63] L. Devetta, P. Canu, A. Bertucco, K. Steiner, Chem. Eng. Sci. 52 (1997) 4163. [64] M.C. Clark, B. Subramaniam, Chem. Eng. Sci. 51 (1996) 2369. [65] A. Corma, A. Martinez, Catal. Rev. Sci. Eng. 35 (1993) 483. [66] J. Weitkamp, Y. Traa, Catal. Today 49 (1999) 193. [67] A.S. Chellappa, R.C. Miller, W.C. Thompson, Appl. Catal. A 209 (2001) 359. [68] R. Gläser, J. Weitkamp, in: Proceedings of the 5th International Symposium on Supercritical Fluids, Atlanta, GA, 8–12 April 2000. [69] K.M. Dooley, F.C. Knopf, Ind. Eng. Chem. Res. 26 (1987) 1910.

292

B. Subramaniam et al. / Applied Catalysis B: Environmental 37 (2002) 279–292

[70] G.J. Suppes, R.N. Occhiogrosso, M.A. McHugh, Ind. Eng. Chem. Res. 28 (1989) 1152. [71] R.S. Oakes, A.A. Clifford, K.D. Bartle, M.T. Pett, C.M. Rayner, Chem. Commun. (1999) 247. [72] G. Jenzer, T. Mallat, M. Maciejewski, F. Eigenmann, A. Baiker, Appl. Catal. A: Gen. 208 (2001) 125. [73] Y.-H. Lin, R.D. Brauer, K.E. Laintz, C.M. Wai, Anal. Chem. 65 (1992) 2594. [74] S. Kainz, D. Koch, W. Baumann, W. Leitner, Angew. Chem. Int. Ed. Engl. 36 (1997) 1628. [75] E.R. Birnbaum, R.M. LeLacheur, A.C. Horton, W. Tumas, Chem. Commun. (1998) 1015. [76] G.R. Hass, J.W. Kolis, Tetrahedron Lett. 39 (1998) 5923.

[77] G.R. Hass, J.W. Kolis, Organometallics 17 (1998) 4454. [78] G. Musie, M. Wei, B. Subramaniam, D.H. Busch, Coord. Chem. Rev. 219–221 (2001) 789. [79] C.A. Eckert, D. Busch, J.S. Brown, C.L. Loitta, Ind. Eng. Chem. Res. 39 (2000) 4615. [80] M. Wei, G. Musie, D.H. Busch, B. Subramaniam, J. Am. Chem. Soc., in press. [81] J.H. Hildebrand, J.M. Prausnitz, R.L. Scott (Eds.), Regular and Related Solutions: the Solubility of Gases, Liquids, and Solids, New York, Cincinnati, Toronto, London, Melbourne, 1970. [82] R. Battino et al. (Eds.), Oxygen and Ozone, Solubility Data Series, Vol. 7, Pergamon Press, Oxford, 1981.