Contents 1
Zeolites in Catalysis (Stephen H. Brown)
1
2
Sol–Gel Sulfonic Acid Silicas as Catalysts (Adam F. Lee and...
223 downloads
3339 Views
37MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
Contents 1
Zeolites in Catalysis (Stephen H. Brown)
1
2
Sol–Gel Sulfonic Acid Silicas as Catalysts (Adam F. Lee and Karen Wilson)
3
Applications of Environmentally Friendly TiO2 Photocatalysts in Green Chemistry: Environmental Purification and Clean Energy Production Under Solar Light Irradiation 59 (Masaya Matsuoka and Masakazu Anpo)
4
Nanoparticles in Green Catalysis (Mazaahir Kidwai)
5
‘Heterogreeneous Chemistry‘ (Heiko Jacobsen)
6
Single-site Heterogeneous Catalysts via Surface-bound Organometallic and Inorganic Complexes 117 (Christophe Copéret)
7
Sustainable Heterogeneous Acid Catalysis by Heteropoly Acids (Ivan Kozhevnikov)
8
The Kinetics of TiO2-based Solar Cells Sensitized by Metal Complexes (Anthony G. Fitch, Don Walker, and Nathan S. Lewis)
9
Automotive Emission Control: Past, Present and Future (Robert J. Farrauto and Jeffrey Hoke)
10
Heterogeneous Catalysis for Hydrogen Production (Morgan S. Scott and Hicham Idriss)
11
High-Throughput Screening of Catalyst Libraries for Emissions Control (Stephen Cypes, Joel Cizeron, Alfred Hagemeyer, and Anthony Volpe)
12
Catalytic Conversion of High-Moisture Biomass to Synthetic Natural Gas in Supercritical Water 281 (Frédéric Vogel)
37
81
93
153
175
197
223
247
j1
1 Zeolites in Catalysis Stephen H. Brown
1.1 Introduction
Acid catalysis as a modern science is less than 150 years old. From its inception, acid catalysis has been explored as a means of producing fuels, lubes and petrochemicals. Ordinary homogeneous acids, both inorganic and organic, never proved industrially useful at temperatures much above 150 C. The first reports of aluminosilicate solid acid catalysts involved the use of clays after the turn of the century. The inspiration for the first commercial synthetic aluminosilicate catalysts came from work done co precipitating silicon and aluminum salts during WWI by a Sun Oil chemist [1]. The Brønsted acid site in these materials is most often represented as in Scheme 1.1. Useful features of this novel type of acid versus homogeneous liquid acids were their high temperature stability, moderate acidity (roughly equivalent to a 50% sulfuric acid solution), solid and non-corrosive character and regenerability by air oxidation. These features enabled acid catalyzed reactions of chemicals to be contemplated at a greatly extended range of temperatures (up to 600 C) and metallurgies. Scheme 1.1 The Brønsted acid site of an aluminosilicate.
The first embodiments of many modern refining processes including heavy oil cracking, naphtha reforming and light gas oligomerization did not use catalysts [2]. As soon as these thermal processes commercialized, exploration of the use of solid acid catalysts ensued naturally. Because of the key role played in the development of the automotive industry, heavy oil cracking to gasoline provided a focal point for the early development of heterogeneous acid catalysis. Temperatures above 400 C and pressures below 3 atmospheres are thermodynamically favorable for the conversion of heavy oils to light hydrocarbons rich in olefins. Acceptable heavy oil cracking rates are achieved without a
Handbook of Green Chemistry, Volume 2: Heterogeneous Catalysis. Edited by Robert H. Crabtree Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32497-2
j 1 Zeolites in Catalysis
2
catalyst at temperatures above 600 C. This was the basis of the thermal cracking process. Thermal cracking produces high yields of methane and aromatic hydrocarbons. The goal of researchers was to find a catalyst that could crack heavy hydrocarbons selectively to gasoline with only minimal formation of gases with molecular weights of less than 30. Due to thermodynamic constraints, the catalyst had to be effective at a temperature above 400 C. In order to avoid unselective thermal cracking, the catalyst had to be effective below 550 C. The discovery in the early 1920s by Houdry that acid activated clays were active and selective in this temperature window was a breakthrough [2]. In the 1930s and 1940s methods were developed and commercialized to produce high surface area manmade aluminosilicates that were significantly improved catalysts. Examination of the aluminosilicate catalysts led to the understanding that the active site was a Brønsted acid [3]. At the time of the discovery of synthetic zeolites in the early 1950s, only two classes of solid Brønsted acids (solid phosphoric acid and aluminosilicates) were being used commercially to produce commodity fuels or petrochemicals [4]. The commercialization of silica-rich synthetic zeolites in their hydrogen form represented a breakthrough for scientists and organizations interested in the production of fuels, lubes and petrochemicals at temperatures above 200 C. Like amorphous aluminosilicates, zeolite Brønsted acid sites are active and stable up to 600 C. Shortly after Union Carbides discovery of synthetic zeolites in the late 1940s, Mobil Oil researchers in catalytic cracking of heavy oil investigated zeolites as potential catalysts [5]. The zeolite known as faujasite (FAU) was found to be three to five orders of magnitude more active than amorphous aluminosilicates. Unmodified, FAU was too active to be useful. When the activity of FAU was tuned by ion exchange with rare earth cations and/or by reducing aluminum content, it was found to have a dramatically different selectivity to cracked products. Optimized samples of FAU zeolites produced almost 5% less C2-gases and coke and increased gasoline yields by more than 10 wt%. Over the course of the past 50 years, evolving heavy oil cracking catalysts and hardware have been continuously decreasing coke and C2-gas yields while increasing the yield of gasoline. The commercialization of zeolite catalysts for heavy oil cracking unleashed the creative abilities of every organization interested in producing fuels and petrochemicals using acid catalysts between 250 and 600 C. Close to 23 processes have been commercialized (Table 1.1). About two-thirds of the processes had no real precedence using homogeneous acids. The other third involved displacement of homogeneous and amorphous acid catalysts. Introduction of zeolite catalysts for the production of commodities has proceeded at a steady pace. Each commercialization has provided an opportunity for zeolite scientists to find improved catalysts. 1.1.1 The Environmental Benefits of Zeolite-enabled Processes
The petroleum industry has been subject to environmental drivers for many decades [6]. Innovations in technology, some driven by more restrictive regulations,
1.1 Introduction Table 1.1 List of zeolite processes.
Process
Reactor type
Temperature range, C
Toluene þ C9 þ aromatics MSTDP Cumene via transalkylation Ethylbenzene via transalkylation Ethylbenzene Cumene Fluid catalytic cracking (FCC) ZSM-5 in FCC Gasoil hydrocracking Distillate hydrocracking Distillate dewaxing Wax hydrocracking Wax hydroisomerization Gasoline octane enhancement Reformate upgrading Light paraffin isomerization Butene isomerization Xylene isomerization Light paraffin aromatization Methanol to gasoline Methanol to olefins Aromatics feed treating Caprolactam
Fixed Fixed Fixed Fixed Fixed Fixed Fluid Fluid Fixed Fixed Fixed Fixed Fixed Fixed Fixed Fixed Fixed Fixed Moving Fixed Fluid Fixed Fluid
350–450 350–450 150–200 230–260 180–250 100–150 500–550 500–550 350–475 280–350 350–450 290–370 300–350 350–450 450–550 240–300 350–450 400–470 450–550 300–400 400–500 150–250 350–450
have continuously increased the efficiency of refining processes. The trend is to produce fuels having lower concentrations of heteroatoms and polynuclear aromatics (often referred to as clean fuels) that can be burned to carbon dioxide and water with increasingly lower emissions of NOx, SOx and particulate byproducts. For decades, nearly the entire hydrocarbon content of a barrel of oil feeding a refinery or petrochemical complex has been converted to salable products or used for fuel at the manufacturing site. Distillation of crude oil largely splits it into streams with the boiling ranges of the fuels sold to consumers and businesses (gasoline, diesel, fuel oil, etc.). The quantities of the streams produced by distillation rarely match market demand. Processes using zeolite catalysts have reduced the effort required to convert streams that are oversupplied by simple crude oil distillation into undersupplied products. Optimized zeolite catalyzed processes are often high technology operations. Performance can be sensitive to the performance of neighboring units. Operating multiple zeolite-catalyzed processes can provide refiners with an incentive to continuously work to bring the refinery closer to steady state operation. Adoption of these high technology processes and work practices has helped refiners to steadily increase the amount of clean fuel products produced from each barrel of oil, thereby reducing emissions of CO2, NOx, SOx and particulates and increasing energy efficiency.
j3
j 1 Zeolites in Catalysis
4
Zeolite catalysts are remarkably efficient. Each weight unit of zeolite produces between 3000 and 500 000 weight units of fuel or petrochemical product before its lifetime ends and it is removed from catalyst service. As a result, relatively small volumes of spent zeolite catalysts are produced. There are often other uses for spent catalysts, such as an ingredient for cement. In the many cases where reuse is an option, there are little/no catalyst waste disposal costs. Catalytic cracking (also known as fluid catalytic cracking or FCC) is by far the most economically important process in the refining and petrochemicals industry and will be described in some detail to allow the green aspects to be highlighted. World wide, FCC units process almost 20 million bbl/day of feedstock (almost 30% of the crude oil produced) and FCC catalysts generate $1 billion in sales [7]. The remarkable performance of the FCC process is achieved by both optimizing the zeolite catalyst and the reactor design. A schematic of an FCC unit is provided in Figure 1.1. The FCC catalyst spends most of its time in a large, cylindrical regeneration vessel typically 15 meters in diameter and 40 meters tall holding 300 tons of a coarse powder catalyst comprised of a bell-shaped distribution of spheres between 15 and 120 microns in diameter. The vessel is typically held at 15–25 psig and 620 to 700 C. Air is continuously blown up from the bottom of the vessel and is carefully distributed to provide uniform contacting with the solids. When properly engineered, up flowing gases mix with the coarse catalyst powder to form a mixture which behaves like a fluid. The reaction carried out in the regeneration vessel is the combustion of the solid carbonaceous reaction byproducts that accumulate on the catalyst during the cracking reaction. The FCC catalyst enters the isothermal, back-mixed regenerator at the reaction temperature (about 550 C) and is heated to the regenerator temperature by the heat of combustion of the coke. Because of its fluid-like properties in the presence of a flowing gas stream, the catalyst will flow smoothly out of the bottom of the regenerator, up a 2 meter diameter
Figure 1.1 FCC reactor process flow diagram.
1.2 General Process Considerations
pipe (called a riser) where it contacts the heavy oil feedstock and then back into the top of the regeneration vessel. A typical catalyst circulation rate for a unit filled with 300 tons of catalyst would be 3000 tons/hour. An average catalyst particle travels through the riser once every 5 or 6 minutes. Heavy oil feedstock is heated to about 300 C and sprayed into the circulating catalyst (620 to 700 C) at the bottom of the riser. Feed vaporization is accomplished by direct contact with the hot zeolite catalyst. The gaseous product is removed utilizing cyclones at the top of the riser. Feedstock is typically fed into the riser at twice the total catalyst inventory and one fifth the catalyst circulation rate (e.g. catalyst circulation of 3000 tons/h and a feed throughput of 600 tons/h). The total time of feedstock and catalyst contact is several seconds. About 5 wt% of the feedstock (no more, no less) must be converted to the carbonaceous solids (coke) that are required to provide the energy input needed to drive the feedstock vaporization and the endothermic reaction. A typical catalyst particle contains about 1 wt% coke on catalyst upon entering the regenerator. Thirty to fifty percent of a barrel of crude oil boils above the endpoint of gasoline and automotive diesel fuels. The FCC unit converts much of this material into gasoline and diesel fuels with roughly 80 wt% selectivity. Another 5 to 10% of the C4products are easily converted into high quality gasoline in a second step, resulting in an overall selectivity to gasoline and diesel fuels of 85 to 90%. Five wt% of the feed is converted to coke which is used to supply most of the fuel for the unit (regeneration and separations). The remaining ca. 5–10% of the byproducts are mostly low molecular weight gases (<35) and propane. Catalyst, oil feedstock and air are the only significant inputs to the process. The removal rate of spent catalyst is roughly 2 wt% per day (6 tons/day from a 300 ton inventory). The catalyst is often used as an ingredient in cement manufacture. If necessary, the gases produced in the FCC regenerator are treated to meet emissions specifications for NOx, SOx and particulates.
1.2 General Process Considerations
As illustrated by the FCC example, zeolites are important green technologies that are used in processes conducted on a large scale. Zeolite processes with products produced in quantities of <50 000 000 kg/year make a negligible contribution to the overall environmental credits achieved by zeolites. Zeolite processes carried out on a large scale are listed in Table 1.1. Most of these processes produce plastics, lubricants or fuels. The significant production volumes required place many practical constraints on production methods. Commodity materials almost without exception are produced from commodity raw materials (usually fossil fuels) in one to four catalytic steps. Each step takes place in reactors of a size which can be conveniently fabricated, transported and erected and the reactor must be able to run continuously or semicontinuously for >1 year without shut-down. Three reactor types are employed: fixed bed, fluid bed and moving bed. Commodity processes typically produce between 0.5 and 5 product volumes per reactor volume per hour.
j5
j 1 Zeolites in Catalysis
6
1.3 Zeolite Fundamentals
Basic information about the structures and compositions of known zeolites is readily obtained by consulting the International Zeolite Association (IZA) structure atlas [8] (available on the Internet) or the Handbook of Molecular Sieves [9]. Almost all of the zeolite catalysts used in the processes listed in Table 1.1 share a number of basic features. They are silica rich (Si : Al > 5 and <50). They are manufactured (capable of being synthesized in the lab) and they contain 10 or 12 membered ring channel systems. Up-to-date information about zeolite structures is available from the IZA online structure atlas [10]. At the time of writing, the atlas contained a total of 180 known structure types each assigned a unique three letter code. Since an infinite number of zeolite structures are possible, currently available samples are a negligible fraction of total possible structures. 15 of these 180 structure types are readily synthesized in the laboratory with Si : Al > 5 and <50 and with a 10 or 12 membered ring channel system (Table 1.2). FAU, EMT, FER, LTL and MOR were synthesized first at Union Carbide. MEL, MFI, MFS, TON, MTT, MTW, BEA and MWW were first synthesized at Mobil. Many more man-made zeolite frameworks are in the IZA database containing 10 and/or 12 membered ring systems. These materials are not included in Table 1.2 because the type materials are pure silica. Examples include CFI, CON, DON, IFR, ISV, SFE, SFF, STF, STT and VET. Most molecules with less than 7 carbons have critical diameters less than the 5.5 angstrom diameter typical of a 10-ring zeolite pore and can freely diffuse. Many larger molecules also have critical diameters less than 5.5 angstroms. Even molecules the size of 1,3,5 tri-isopropyl benzene can diffuse into 12-ring zeolites
Table 1.2 IZA 10 and/or 12 ring zeolite structures with Si:Al between 5 and 50.
Structure code
Ring size
Diffusional potential
EUO (ZSM-50) MTT (ZSM-23) TON (ZSM-22) NES FER (ZSM-35) MFS (ZSM-57) MWW (MCM-22) MEL (ZSM-11) MFI (ZSM-5) MTW (ZSM-12) LTL MOR BEA EMT FAU
10 10 10 10 by 10 by 10 by 10 by 10 by 10 by 12 12 12 by 12 by 12 by 12 by
1D 1D 1D 2D 2D 2D 2D 3D 3D 1D 1D 2D 3D 3D 3D
10 8 8 10 10 10
8 12 by 12 12 by 12 12 by 12
1.3 Zeolite Fundamentals
with pore diameters exceeding 7 angstroms. This means that the vast majority of molecules present in distilled petroleum fractions can diffuse in and out of zeolites containing 12-membered rings. Larger ring structures, such as 18-ring VPI-5, have a pore diameter of 12 angstroms. These pores are so big that many small molecules fit in side-by-side and ordinary molecules can not be discriminated by molecular size. Once pore sizes have reached >12–20 angstroms, acid sites inside the pores can be conceptualized in the same fashion as acid sites on amorphous aluminosilicates or on zeolite surfaces. Pores so large place few steric constraints on the polymerization of large molecules into larger deactivating oligomeric structures. Structures of the zeolite frameworks listed in Table 1.2 are provided at the IZA website. Although all the structures contain 10 or 12 ring pores, each structure has many unique aspects. Each ring system has its own unique size and shape. Some zeolites, e.g. FAU, have large internal cavities, while others contain only one dimensional cylindrical pores (e.g. LTL). Because there are only a handful of unique structures, it should not be surprising that there is often a large difference in performance when these structures are applied. 1.3.1 Other Properties
At temperatures >200 C zeolite Brønsted acid protons delocalize [11]. At temperatures >550 C dehydroxylation is initiated and Brønsted acid activity is lost [12]. The presence of steam can retard dehydroxylation. Activity loss by dehydroxylation is commonly reversible as the dehydroxylated zeolite can rehydrate at low temperature and resume its original structure. Zeolites exchanged with polyvalent metal ions (typically nitrate salts) become acidic upon thermal dehydration and nitrate decomposition [13–16]. Weak Brønsted acid sites can form by hydroxylation of the metal cation. The mechanism is believed to proceed by association of the cation with a specific framework aluminum accompanied by dissociation of water to form a hydroxyl group attached to the cation (Scheme 1.2). For this reason, the addition of polyvalent cations to zeolites directly impacts the number of Brønsted acid sites and total zeolite pore volume but has only a minor impact on the strength of the remaining Brønsted acid sites. Furthermore, zeolites containing polyvalent cations are considerably more complex because both the metal cations and the protons are mobile and because many metal ions are more active for redox reactions than silicon and aluminum. At reaction temperatures between 250 and 500 C these features generally lead to increased rates of hydrogen transfer reactions and more rapid deactivation explaining the limited use of zeolites exchanged with polyvalent cations. Scheme 1.2 Example of a weak Brønsted acid site in a metalexchanged zeolite.
j7
j 1 Zeolites in Catalysis
8
1.3.2 Number of Acid Sites
Loewensteins rule forbids the formation of AlOAl bonds in zeolite structures [17]. Therefore, the potential number of acid sites equals the number of aluminum atoms in any reference unit of a zeolite crystal. High silica zeolites (Si : Al > 20) can generally be synthesized and converted to the hydrogen form with minimal deviation from the idealized structure. For these materials the number of acid sites determined by analytical techniques agrees well with the number of acid sites derived from a simple analysis of bulk aluminum content. 1.3.3 Acid Strength
All of the catalysts used in the reviewed processes are aluminosilicates. The overall acid strength of a hydrogen form aluminosilicate zeolite depends upon aluminum distribution. Acidity associated with an aluminum tetrahedra is stronger with a smaller number of near aluminum atoms [18–20]. For this reason, zeolites with Si:Al ratios between 1 and 10 can have a variety of acid site strengths. However, careful studies with ZSM-5 demonstrated that acid sites with 0 and 1 next nearest neighbor aluminums are very close in acid site strength [21]. Most of the acid sites in zeolites with Si : Al ratios >10 have only a small number of their sites with more than one aluminum next nearest neighbors leading to uniform acid site strength. The strength of this site has been well characterized by NMR and IR probes of simple sorbates allowing the conclusion to be reached that the acid site strength is similar to that of 70% sulfuric acid [22–24]. Careful studies of model compound reactions uncomplicated by mass transfer limitations or fast secondary reaction provide further support for uniform acid site strength [25–27]. At the present time, aluminosilicate zeolites remain the only class of crystalline solid Brønsted acids to find broad use in the production of commodity chemicals. Although a wide range of materials with alternative framework compositions are known, few commercial uses have been found for these materials. Zeolite frameworks and novel frameworks based on aluminophosphate building blocks were discovered at Union Carbide in the early 1980s [28–30]. When phosphorus sites are substituted with silicon, a Brønsted acid is formed. The acid site in these materials is weaker than an aluminosilicate acid site.
1.4 Reaction Mechanisms 1.4.1 Hydrocarbon Cracking
Academic work in the 1930s and 40s elucidated how AlCl3 – a strong Lewis acid – is converted when dissolved in hydrocarbon fractions to a working catalytic species with
1.4 Reaction Mechanisms
strong Brønsted acidity (Scheme 1.3) [31–33]. The basic mechanistic features of hydrocarbon cracking were well understood by the end of the 1950s and are well explained in many subsequent reviews [34–39]. Any unsaturated molecules (i.e. aromatics, olefins, dienes) in hydrocarbon streams undergo protonation in the presence of a Brønsted acid catalyst. Once protonated, isomerization reactions can proceed. In general, hydride shifts proceed considerably faster than alkyl shifts (Scheme 1.4). Exact relative rates are dependent upon the structure of the hydrocarbon, the catalyst and the conditions and need to be computed or measured on a case by case basis.
Scheme 1.3 Example AlCl3 activation reactions.
Scheme 1.4 Hydride and methyl shifts.
Once protonated, a hydrocarbon molecule is destabilized, existing almost simultaneously as many different carbocations. The energy of most hydrocarbon carbocations are now well understood and can be calculated using algorithms derived from first principles [40]. In most cases it is safe to assume that a representative sample of a specific protonated hydrocarbon exists at any instant with the full pool of its possible cation isomers populated at a distribution at least approaching thermodynamic equilibrium. Because carbocations are stabilized by delocalization and electron donating groups, isomers containing these attributes dominate the instantaneous distribution (Scheme 1.5, for example). The most stable cations, however, can be less reactive and therefore may not be the most important intermediates of the reaction pathway. Scheme 1.5 Sample of a cation stabilized by conjugation and branching.
Acid cracking of cations proceeds most readily via beta scission. Aromatics dealkylation is the least complex as it is dominated by a single class of beta scission reaction (Scheme 1.6). There are many viable cracking pathways for paraffin, olefin and naphthene-derived hydrocarbon cations. Cracking of these species is dominated
j9
j 1 Zeolites in Catalysis
10
by beta scission pathways A, B1, B2 and C (Scheme 1.7) [41]. Relative rates of these reactions along with cracking reactions involving primary cations (pathways D and E, Scheme 1.7) have been determined using model compounds and ZSM-5 catalysts at cracking conditions typical of the industrial processes covered in this review (Tables 1.3 and 1.4) [42]. The combined data from the tables demonstrate that large, branched olefins are readily cracked. Through successive cracking and oligomerization reactions, zeolite catalysts can convert such molecules to a broad distribution of olefins directed by thermodynamic considerations at temperatures below 200 C.
Scheme 1.6 Aromatics cracking reaction.
Scheme 1.7 Hydrocarbon cation cracking types.
Table 1.3 Cracking rate constants of hydrocarbons over HZSM-5 at 510 C.
Carbon #
Rate constant k sec1 olefin
Rate constant k sec1 paraffin
4 5 6 7 8
10 230 1800 5700
0.1 0.3 0.8 1.5 2.2
1.4 Reaction Mechanisms Table 1.4 Relative rates of olefin cracking over HZSM-5 at 510 C.
Olefin feed
Cracking type
Relative rate constant
C6
C D E B C D E B C
40 2 1 120 40 2 1 120 40
C7
C8
Cracking of pure paraffins has been the subject of a great deal of academic interest due to the possibility that paraffin cracking might be initiated by protonation (Scheme 1.8). In fact, under forcing conditions this reaction has been observed [43– 45]. It has a higher activation energy than paraffin cracking involving hydride abstraction (Scheme 1.9). Chain cracking involving hydride abstraction [46] dominates because this pathway is autocatalytic. Furthermore, typical hydrocarbon mixtures almost always contain aromatic and/or olefinic initiators so cracking via paraffin protonation need never be invoked.
Scheme 1.8 Example of paraffin cracking via protonation.
Scheme 1.9 Example of paraffin cracking via hydride transfer.
An unusual reaction that cracks polymethylaromatics to ethylene and propylene, and mono, di and trimethylbenzenes, is known as the Paring reaction [47]. The Paring reaction is unusually important in zeolite catalysis because it provides a mechanism for removing polyalkylbenzenes trapped inside zeolite pores before they can undergo further reactions to form condensed ring polynuclear aromatics that cause permanent deactivation. The mechanistic steps that form ethyl and propyl aromatics from polymethyl aromatics are shown (Scheme 1.10). The reaction can occur via a concerted electrocyclic reaction of migrating double bonds. The reaction is facilitated by the known non-classical structure of protonated polymethylbenzenes [48–51]. Dilute solutions of pentamethylbenzene in FSO3H result in stable solutions of completely protonated aromatic. Irradiation of this cation at 78 C causes nearly quantitative conversion
j11
j 1 Zeolites in Catalysis
12
via ring contraction to a stable pentacoordinate cyclopentyl cation with a non-classical structure (Scheme 1.11). Expansion of this ion back to a 6 ring protonated aromatic occurs upon warming to 30 C.
Scheme 1.10 Formation of iso-propyl-aromatics from polymethyl-aromatics.
Scheme 1.11 Non-classical cation structure from pentamethylbenzene protonation.
1.4.2 Oligomerization and Alkylation
Like cracking, the mechanism was first well described in the 1940s using homogeneous acids [52–56]. Protonation of unsaturated hydrocarbons is governed by the same rules as just discussed for cracking. The adding cation is usually secondary or
1.4 Reaction Mechanisms
tertiary creating branches in the product molecules. Addition is governed by Markovnikovs rule. These features are exemplified by a favored butene trimerization pathway (Scheme 1.12) [57]. Secondary alkyl and hydride shift reactions occurring at competitive rates often result in a very complex product mixture especially at the temperatures between 100 and 200 C which are typically utilized when oligomerizing light olefins over zeolites [58, 59]. Steric constraints impact oligomerization using zeolites. The branchiness and average carbon number of oligomerization products tends to decrease with decreasing dimensionality and pore size. Careful attention to choice of zeolite and conditions allows the production of oligomers with near linear structures (more linear than thermodynamic equilibrium) [60].
Scheme 1.12 Isobutylene trimerization at 0 C with 65–70% H2SO4.
Alkylations of olefins with aromatics or paraffins are reverse cracking reactions which proceed by a similar mechanism to olefin oligomerization. It was first described using homogeneous acid catalysts at near room temperature [61–63]. Aromatics alkylation proceeds by protonation of the olefin followed by addition to the aromatic ring (Scheme 1.13). When an alkylbenzene undergoes further alkylation (i.e. toluene methylation to form xylenes), electron donation by the alkyl substituent activates the ring and causes the addition to be ortho/para selective [64]. Steric constraints play a strong role. While methyl and ethyl groups readily form ortho isomers, ortho di-isopropylbenzene is a minor reaction product and dit-butylbenzene is not formed with acid catalysts [65].
Scheme 1.13 Proposed aromatics alkylation mechanism.
j13
j 1 Zeolites in Catalysis
14
Isoparaffin alkylation is a chain reaction that has a hydride transfer as the rate determining step (Scheme 1.14). Olefin oligomerization has a lower activation energy and proceeds much faster than isoparaffin alkylation at ordinary conditions. Good selectivity to C8 isoparaffins from a mixture of isobutane and n-butenes is achieved by running the reaction in a large excess of isobutane.
Scheme 1.14 Proposed isoparaffin alkylation Mechanism.
1.4.3 Isomerization
Olefins are easily isomerized upon contact with mineral acids. The olefin doublebond shift reaction is one of the fastest acid catalyzed reactions of hydrocarbons (Scheme 1.15). Shifting alkyl groups along the backbone of an olefin also occurs readily but requires more severe conditions since more than simple protonation/ deprotonation reactions are involved. This type of reaction is exemplified by a methyl shift reaction (Scheme 1.16). A third type of olefin isomerization adds or removes a branch [66, 67]. This type of isomerization involves a cyclopropyl alkylcarbonium ion often referred to as a protonated cyclopropane (Scheme 1.17).
Scheme 1.15 Proposed mechanism of olefin double-bond shift reactions.
Scheme 1.16 Proposed mechanism of a methyl shift olefin isomerization reaction.
Scheme 1.17 Proposed mechanism of olefin branching/unbranching isomerization.
1.4 Reaction Mechanisms
At low temperatures and high pressures, oligomerization/polymerization is favored by thermodynamics. Therefore, care must be taken to achieve selective isomerization. Selectively achieving the more demanding skeletal isomerization of olefins usually requires operation at higher temperatures and lower pressures where olefin monomers dominate equilibrium. As olefin molecular weight rises, selective skeletal isomerization becomes increasingly challenging since increasing temperature and decreasing pressure leads to undesired cracking while decreasing temperature and increasing pressure leads to oligomerization. Paraffins isomerize by nearly the same mechanisms as olefins. Paraffin isomerization is of greater industrial interest than the isomerization of unsaturated hydrocarbons because it enabled conversion of C5–C10 n-alkanes of low octane values into branched alkanes with high octane and the conversion of C20 to C100 waxes into lubricant base stocks. It also enabled the production of isobutane from n-butane which was needed to make alkylate. The extensive work aimed at commercializing paraffin isomerization contributed greatly to the fundamental understanding of all acid catalyzed hydrocarbon isomerizations. Paraffin isomerization is more difficult than olefin or aromatic isomerization because it requires hydride abstraction (Scheme 1.19) or dehydrogenation using a noble metal. In the most common method, a noble metal is used to bring the paraffin to the paraffin/olefin equilibrium [68]. At temperatures below 300 C, equilibrium favors the paraffin so only trace olefin is present in the reaction medium. An acid function isomerizes the trace olefin by the mechanisms in Schemes 1.15, 1.16, and 1.17. Polyalkyl aromatics also isomerize by an alkyl shift mechanism. The reaction mechanism for the interconversion of meta and para xylene is shown in Scheme 1.18. Note that in each of Schemes 1.15, 1.16 and 1.18 the intermediate cation structures shown are all secondary. Although these secondary cations are higher energy than the available tertiary cations, they are necessary intermediates in the reaction pathway. The possible tertiary cations are not shown because formation of these favored ions does not lead to the desired transformation.
Scheme 1.18 Proposed mechanism of xylene isomerization.
Scheme 1.19 Proposed chain transfer hydride abstraction step in paraffin isomerization.
1.4.4 Transalkylation of Aromatics
The acid catalyzed reaction, disproportionation of alkylbenzenes to benzene and polyalkylbenzenes, has been investigated thoroughly. Using Friedel–Crafts catalysts
j15
j 1 Zeolites in Catalysis
16
and conditions, tert-alkyl benzenes transalkylate by cracking to olefin and benzene and realkylating [69, 70]. N-Alkylbenzenes transalkylate by a different mechanism. Mechanistic studies of the n-alkyl benzene reaction have been carried out at low temperatures in liquid media containing Friedel-Crafts or superacid catalysts [71–73], showed that the reaction proceeds via a chain mechanism initiated by the formation of benzyl cations and propagated by hydride abstraction. Sec-alkylbenzenes apparently undergo transalkylation by one or a combination of these two mechanisms depending upon the catalyst and conditions. The mechanism of n-alkylbenzene transalkylation depends on the steady state concentration of benzyl cations, which when formed abstract a hydride from a neighboring aromatic or alkylate another aromatic to form a 1,1-diphenylalkane. The mechanism for moving the simplest n-alkyl substituent, a methyl group, from one aromatic ring to another is provided (Scheme 1.21). In the presence of an acid, 1,1-diphenylethane easily cracks back to benzene and styrene (Scheme 1.20). Protonation of styrene forms a benzyl cation, which either abstracts a hydride from another ethylbenzene or alkylates. Although alkylation to 1,1-diphenylethane is much faster than hydride abstraction, repeated cracking back to styrene and benzene (Scheme 1.20) followed by protonation to benzyl cation eventually results in chain transfer of the hydride, completing the transalkylation.
Scheme 1.20 Cracking of diphenylethane to benzene and styrene.
Scheme 1.21 Proposed mechanism of toluene transmethylation.
1.4 Reaction Mechanisms
At temperatures between 0 and 50 C, toluene disproportionation catalyzed by AlCl3 and AlBr3 is inefficient and was estimated to proceed about 107 times more slowly than transethylation [74]. Diphenylmethane lacks a b-CH bond and is more difficult to crack explaining the observed dramatic rate deceleration. Although inefficient with Friedel–Crafts catalysts, the disproportionation of toluene to xylenes is practiced industrially to convert on the order of 10 billion pounds of aromatics per year with high activity zeolite catalysts. In addition to the mechanism already discussed (Scheme 1.21), another acid catalyzed pathway involving dimers has been reported [75]. This mechanism (Scheme 1.22) is consistent with previously reported mechanistic studies on aromatics transmethylation at high temperatures using solid acid catalysts [76–78]. It proceeds via aromatics protonation and is likely to be of increasing importance as the basicity of the aromatic increases. Polyalkylbenzenes, and particularly 1,3,5-trisubstituted polyalkylbenzenes, are orders of magnitude more basic than benzene, toluene and xylenes [79], and are more likely to proceed by this mechanism.
Scheme 1.22 Proposed toluene disproportionation mechanism with a diphenylmethane intermediate.
j17
j 1 Zeolites in Catalysis
18
The Scheme 1.22 mechanism involves the formation of cyclohexadienyl toluenes as intermediates. At low temperatures in the liquid phase, these species would be expected to undergo further alkylations to form oligomers. Consistent with this prediction, benzene solutions of AlBr3–HBr have been reported to be unstable [80]. The use of zeolite catalysts can lead to mixed mechanisms. At high temperatures and moderate pressures unimolecular cracking reactions are favored relative to bimolecular condensations. For example, the vapor phase transalkylation of polyethylbenzenes with ZSM-5 at 150 psig aromatics and 300 C is believed to proceed via cracking/realkylation [81]. On the other hand, the use of zeolites beta, mordenite and faujasite to transalkylate polyisopropylbenzenes in the liquid phase at 300 psig and 200 C is believed to proceed mostly by bimolecular mechanisms [82]. 1.4.5 Hydrogen Transfer or Conjunct Polymerization
In the presence of acid catalysts, olefins can be reacted with isobutane in the liquid phase at <150 C to form paraffins. This reaction is run commercially with very strong homogeneous acids (e.g. H2SO4, HF) at isobutane to olefin ratio in the feed near ten. Zeolites can also be employed, but due to their lower intrinsic acidity exhibit much greater selectivity for olefin oligomerization [83]. Olefins are also converted to unsaturated cyclic hydrocarbons and paraffins by acid catalysts. The rate determining step in this reaction and in isobutane/butene alkylation is believed to involve the transfer of a hydride from a paraffin to a protonated olefin. In the hetergeneous catalysis community, this transformation is commonly referred to as hydrogen transfer. Hydrogen transfer reactions were well described near 1900 when olefins were contacted with homogeneous acids near room temperature. The reaction has been referred to as conjunct polymerisation [84]. When alkenes such as butylenes, 3methyl-1-butene and nonenes are treated with an equal volume of 96% sulfuric acid at 0 C, the hydrocarbons dissolve at first in acid. On standing at 0 C two layers separate. The top layer consists largely of saturated linear and cyclic hydrocarbons. Upon hydrolysis with water, the bottom layer separates into acid and hydrocarbon phases. The hydrocarbon phase is a complex, wide boiling mixture of highly unsaturated cyclic hydrocarbons containing conjugated double bonds and five carbon rings. A general, simplified picture of how cyclopentadienes are produced from olefins is shown in Scheme 1.23 [85]. In the high temperature environment of commercial zeolite catalyzed hydrocarbon processes, byproducts from hydrogen transfer reactions are often observed [86]. Moreover, at >250 C substituted cyclopentadienes are unstable and convert by subsequent isomerization and hydrogen transfer reactions to predominantly 1-ring aromatic compounds. This reaction has recently been directly observed in ZSM5 [87]. Stoichiometric restrictions apply to the hydrogen transfer reaction. For each unit of unsaturation produced, one mole of paraffin is produced. Four moles of paraffins are produced to offset each mole of aromatic formed.
1.4 Reaction Mechanisms
Scheme 1.23 Simplified mechanistic proposal to produce cyclopentadienes from olefins.
Like olefins, alkyl-aromatics undergo hydrogen transfer reactions to form multiring aromatics and light paraffins. Olefin co-feeds or olefins formed in-situ from aromatic cracking reactions are necessary participants in the reaction path. These reactions involve the formation of benzyl cations. When olefins are present, they compete with aromatics to scavenge the reactive benzyl cations. When a benzyl cation alkylates an olefin, an indane can be generally formed. This is illustrated for ethylbenzene and cumene (Schemes 1.24 and 1.25). Compared to isomerization, oligomerization, cracking and alkylation, hydrogen transfer is usually a slow reaction. Nonetheless, it is of critical importance in many zeolite catalyzed processes because catalyst deactivation by coking typically involves hydrogen transfer reactions. In order to achieve acceptable cycle lengths, fixed bed processes must reduce coking reactions to near negligible rates. Hydrogen transfer is favored at high pressures and is nearly inevitable in reactions carried out above 400 C. The bimolecular nature of the reaction explains why the reaction rate can be strongly influenced by the steric constraints imposed by the structure of the pores of a zeolite catalyst. For example, hydrogen transfer has been shown to proceed at higher rates in faujasite than in ZSM-5 [88, 89].
Scheme 1.24 Indanes, isobutylene and EB from cumene.
j19
j 1 Zeolites in Catalysis
20
Scheme 1.25 Indanes, propylene and toluene from ethylbenzene.
In addition to participating in alkylation reactions, the olefin intermediates will also abstract hydrides to form new benzyl or related cations and light paraffin. Because the alkylation products of isobutylene are often difficult to form due to steric hindrance, are easily cracked and monomers are relatively favored by thermodynamics, isobutylene is particularly prone to hydride abstraction to form isobutane. An example reaction path producing isobutane and naphthalene from isobutylene and an indane is shown in Scheme 1.26. Analogous reactions lead to the formation of other light paraffins and multiring aromatics. Indanes are ordinary observed only as minor products because these molecules are much better hydride donors than the feedstock aromatic and tend to undergo the secondary reactions to naphthalenes at faster rates than they are formed from benzyl cations and olefins.
Scheme 1.26 Isobutylene and indane to isobutane and naphthalene.
As just shown, hydrogen transfer reactions involving aromatics lead to the formation of large, polynuclear aromatics. These types of molecules are prone to reacting to form coke faster than they can diffuse out of the pores of zeolites. Control of zeolite structure, process conditions and feedstock composition to minimize or eliminate aromatics hydrogen transfer reactions (at least those leading to deactivating coke) is an area of constant study since increased process cycle lengths generally result in substantial process cost savings.
1.5 Mass Transport and Diffusion
1.5 Mass Transport and Diffusion
Diffusion in pore sizes in excess of 0.5 microns is not significantly different than gas phase diffusion in larger spaces. In pores between 10 and 1000 angstroms, wall effects are important and increase in importance as the pore diameter decreases. For most of this range, however, molecules typically strike other molecules much more frequently than they strike pore walls. Diffusivity in pores of this size is usually well described by the Knudsen flow equations. Below 10 angstroms, molecules strike the pore walls with similar or greater frequency than they strike other molecules. The diffusion of small molecules such as hydrogen, nitrogen, oxygen, CO and methane in large pore (12 ring channel systems) zeolites at temperatures above 200 C can usually be adequately described using the Knudsen flow equations. The reactants and products of interest in this review are molecules with sizes similar to the pore dimensions of the zeolite catalysts. Because of this, minor changes in size, shape and polarity can have a dramatic effect on the diffusion constant of the molecule. For this reason molecular diffusion in zeolites is a very complex topic which has been the subject of a large amount of research. This field has been the subject of many reviews [90–98]. The relative rates of diffusion of different possible reactants and products is usually at least partially responsible when large differences in selectivities are observed for different aluminosilicate catalysts. The relative diffusivities of para xylene, meta xylene and 1,3,5-trimethylbenzene have been measured at 1 atmosphere and 315 C and found to be about 107, 109 and 1012 cm2 per second respectively [99]. Dropping the diffusion constant from 109 to 1012 cm2 per second at these conditions leads to a situation where the intrinsic rate of reaction of 1,3,5-trimethylbenzene is much faster than its rate of diffusion. This circumstance is often described by saying that the reactant is too large to enter the catalyst pores. The ability of zeolites to discriminate between feed and product molecules in this fashion is at the heart of their utility. However, restricted diffusion of desired feed and product molecules is also an inherent problem. It is highly desirable to build a catalytic process where the diffusivity of the target feeds and products is fast enough compared to the intrinsic reaction rate to place the desired reaction under kinetic control (or near to it) rather than mass transport control. It is also highly desirable to design a commodity process to produce at least 0.5 liquid volumes of product per volume of catalyst per hour. Achieving these targets generally requires diffusion coefficients of greater than 1010 cm2 per second. As feed and product diffusion coefficients approach the minimum, increasingly smaller zeolite crystal size is required to achieve the catalyst productivity target. There is a large and growing body of literature reporting the reactions of molecules with highly polar or polarizable functional groups [100]. Increasing molecular weight, increasing polarity and decreasing temperature all decrease molecular diffusion rates. Poor diffusion rates of large, polar molecules constitute a high hurdle for commodity process development. An example of a zeolite-catalyzed process using a large, polar feedstock is the Beckmann rearrangement (Scheme 1.27).
j21
j 1 Zeolites in Catalysis
22
The process utilizes temperatures between 300 and 400 C, pressures below 4 atmospheres and zeolite catalyst with low aluminum content and small crystal size [101, 102]. These conditions minimize the impact of poor inherent diffusivity.
Scheme 1.27 The Beckmann rearrangement.
1.6 Zeolite Shape Selectivity
The interaction of acid catalyzed mechanisms, the unique pore geometry of zeolite catalysts and feed and product diffusion results in zeolite shape selectivity. Shape selectivity enables products to be produced with remarkable kinetic selectivities despite the high temperature conditions and competing acid catalyzed mechanisms to thermodynamically favored products. 1.6.1 Mass Transport Discrimination of Product Molecules
A well-studied example of zeolite shape selectivity is the production of xylenes containing >90% p-xylene from toluene by disproportionation. Toluene disproportionation is the classic example zeolite shape selectivity arising from mass-transport discrimination. This type of shape selectivity arises when there are significant differences in diffusivities among various classes of molecules with respect to the pore channel size/configuration for a given zeolite. Shape selective toluene disproportionation was discovered and commercialized by Mobil and converts low cost distilled toluene or extracted toluene directly into chemical grade benzene and high purity p-xylene [103, 104]. The benzene produced is so pure that no benzene extraction is required. Per pass p-xylene yield is limited by equilibrium to less than 20%. As product p-xylene approaches equilibrium, it slowly isomerizes to meta and ortho xylene. The much more demanding isomerization of xylenes to ethylbenzene proceeds at a faster apparent rate. The ethylbenzene then hydrocracks to ethane and benzene. Briefly, p-xylene is produced selectively by designing the ZSM-5 catalyst to serve as a reactive membrane for internally produced C8 aromatics [105–108]. Toluene diffuses in and out of the ZSM-5 pore system faster than it disproportionates. The rate of disproportionation is independent of crystal size out to significantly larger than 10 microns. Work with small crystal ZSM-5 catalysts at low toluene conversion proved that p-xylene dominates the kinetic distribution of xylenes. Like toluene, p-xylene diffuses in and out of the catalyst faster than it reacts. The ortho and meta
1.6 Zeolite Shape Selectivity
xylenes diffuse 1000–10 000 times more slowly and as a result concentrate inside the catalyst pores. Fast, secondary isomerization reactions convert the ortho and meta byproducts into p-xylene which then rapidly diffuses out. A much slower secondary isomerization reaction converts xylenes to ethylbenzene which also rapidly diffuses out. Excellent quantitative treatments of this remarkable interaction of diffusion and chemical kinetics are available [109]. 1.6.2 Molecular Sieving
Molecular sieving is an extreme boundary case of mass-transport discrimination. This type of shape selectivity arises when classes of molecules in the feedstock are completely excluded from the intracrystalline volume of the zeolite catalyst. Lubricant oil dewaxing is a good example of the industrial application of the molecular sieving properties of acidic zeolite catalysts. The most common lubricants are viscous hydrocarbon liquids with molecular weights ranging from 250 to 650 (C20 to C60). Lubricants can be produced from crude oil fractions in the lube molecular weight range by removing wax (linear and near linear paraffins) and high density aromatics (aromatic molecules with low hydrogen content). Acidic zeolites are highly active for the depolymerization (acid catalyzed cracking) of lubricant molecular weight hydrocarbons at 300–400 C. Commercial manufacturing of lubricants uses special zeolite catalysts that selectively crack linear and near linear paraffins. The structure, acidity and activity of the active sites of wax-selective zeolite cracking catalysts is similar to the active sites of non-selective zeolite catalysts. Molecular sieving is the source of selectivity. Linear and near linear paraffins (waxes) diffuse rapidly in and out of the pores of these catalysts. The diameter of the pores is just the right size to stabilize the adsorption and transport of linear and near-linear paraffins. Paraffins with three or more branches, cycloparaffins and aromatic molecules are all too large to fit into the zeolite catalyst pores. In optimized catalysts transport rates are discontinuous. The target molecules, linear and near linear paraffins, diffuse rapidly (diffusion coefficients near 104 cm2/sec), while the rest of the molecules in the feed are excluded (have diffusion coefficients of <106 cm2/sec). 1.6.3 Molecular Orientation
Reaction pathways within zeolite microchannel systems can be inhibited or blocked if the available space is insufficient to accommodate the steric demands of the required transition states. At the same time, other reaction pathways can be accelerated if the shape of the channel acts as a template that preferentially orients reactants. Selective propylene oligomerization in unidimensional 10-ring zeolites (e.g. IZA structure codes EUO, FER, MFS, MTT, TON, FER, SFF and STF) will be used to illustrate this type of shape selectivity. Oligomerization of propylene at 30 to 60 bar olefin partial pressure and 200–300 C over solid acid catalysts in the absence of steric constraints leads to the preferential
j23
j 1 Zeolites in Catalysis
24
formation of C9 þ products with more than one branch [110]. Ease of protonation increases with olefin molecular weight and degree of substitution. C12 þ multiply branched olefins are preferentially formed. Many of these molecules are unfavored at thermodynamic equilibrium. Isomerization and cracking reactions of C12 þ multiply branched olefins leading to branched lower olefins have good kinetics. Successive dimerization, isomerization and back-cracking leads to an equilibrium carbon number distribution with equilibrium branching. In such a situation, countless oligomerization, isomerization and cracking reactions take place simultaneously steadily bringing the mixture of olefins to equilibrium [111]. The formation and cracking reactions of C12 þ olefins are constrained by zeolite micropore structures. For this reason, every zeolite produces a unique fingerprint of C9–C18 olefins at partial conversion [112]. Unidimensional 10-ring zeolites exhibit unique shape selectivity for propylene oligomerization [113]. The kinetic products from inside the micropores are dominated by monomethyl olefins because the shape of the channel constrains the orientations of the feed molecules. The end to end orientation of molecules in the channels favors dimerization pathways that produce monomethyl olefins. All zeolites preferentially produce 2-methyl-pentene when propylene dimerizes. To further oligomerize inside the pores of monodimensional 10-ring zeolites, methyl-pentenes normally unbranch before condensation with another olefin. Condensation reactions involving monomers reacting with linear oligomers are favored in the pores (Scheme 1.28). Most C12 oligomers are made by condensation of propylene with a linear C9 olefin, rather than by C6 dimerization. Reactions involving monomers are strongly favored by the end-to-end molecular orientation dictated by the cylindrical shape of the pores. Equilibrated linear C6 þ olefins contain <1 wt% alpha olefins. The probability of two oligomers meeting the requirements for condensation is small. The tendency for monomer to participate in most oligomerization reactions produces products predicted using Shultz–Flory kinetics.
Scheme 1.28 Expected kinetic products from propylene condensing with n-hexenes.
2-methyl and 4-methyloctenes are the expected kinetic products of secondary oligomerization reactions of hexenes. During propylene oligomerization inside a unidimensional, 10-ring pore, however, 2-, 3- and 4-methyloctenes are produced at an a ratio of 11 : 14 : 16 (the equilibrium distribution at this temperature and pressure).
1.6 Zeolite Shape Selectivity
This indicates that unimolecular secondary methyl shift reactions proceed at rates much faster than molecular diffusion in and out of the zeolite pores. Dibranched olefins appear among the kinetic products in increasing amounts with increasing molecular weight. These isomers are probably formed by condensations exemplified in Scheme 1.29.
Scheme 1.29 Proposed typical reaction pathway producing a dibranched oligomer in unidimensional 10-ring zeolites.
1.6.4 Transition State Stabilization
Reaction pathways can be accelerated if the shape of the channel acts as a template that stabilizes the ionic transition state. Selective paraffin hydroisomerization will be used to illustrate this type of shape selectivity. Wax hydroisomerization to lubricant base stocks is an important industrial application [114]. Paraffin branching/unbranching involves a strained cyclopropyl carbonium ion (penta-coordinate carbon) transition state. In an ordinary steric environment, this transition state is higher in energy than the three coordinate carbenium ions that typically dominate acid catalyzed reactions. For example, paraffin cracking reactions proceed via ordinary carbenium ions. Amorphous aluminosilicates and large pore zeolite catalysts favor cracking reactions over hydroisomerization reactions, especially once two or more branches are introduced into the paraffin. Molecules like a linear C36 paraffin wax are selectively converted to cracked fragments at >80% wax conversion. Only low yields (40%) of C36 isoparaffins with two or more branches are possible [115]. Unidimensional 10-ring zeolites or silicoaluminophosphates facilitate the formation of the strained cyclopropyl carbonium ion transition state that converts n-paraffins to isoparaffins. The wax molecules are forced by the pore structure into a relatively linear configuration. End to end interactions of molecules dominate. Molecular collisions compress carbenium ions into carbonium ions (Scheme 1.30). The transient population of carbonium ions is increased and the transient population of ordinary carbenium ions is decreased, reversing the relative rates of isomerization and cracking. As a result, good yields of lubricant oils from paraffin wax feedstock can be achieved [116]. Because branched lube molecules are easily hydrocracked, selective wax hydroisomerization has been one of the most challenging commercialized commodity zeolite processes to optimize. Every aspect of zeolite catalyst structure influences isoparaffin yields. The selectivity of any individual zeolite structure depends upon crystal size, shape, acid site concentration and acid site distribution.
j25
j 1 Zeolites in Catalysis
26
Scheme 1.30 Proposed interconversion of carbenium and carbonium ions in the channels of unimolecular 10-ring wax isomerization catalysts.
1.6.5 Organic Reaction Centers
Reaction pathways can be accelerated if cavities in the zeolite trap and activate reactive organic intermediates. The conversion of methanol to hydrocarbons is the only wellunderstood example and will be used to illustrate this highly unusual type of shape selectivity. Methanol lacks a b-carbon H-bond. Consequently, the formation of an olefin and water via acid catalyzed dehydration is prevented, distinguishing methanol from all other industrially important alcohols. In contrast to propanol, which dehydrates readily at 150 C, methanol does not convert to hydrocarbons over 10-ring zeolites below 250 C [117, 118]. Methanol conversion to predominantly olefins has been reported using a wide variety of zeolite and SAPO catalysts and conditions. Conversion of methanol to olefins over zeolites involves C6 þ unsaturated intermediates entrained within the zeolite pore system [119]. Because methanol is a methylating agent, methylation is the first step. Cn olefins are converted to Cn þ 1 olefins [120–125], Cn dienes are converted to Cn þ 1 dienes [126] and Cn aromatics to Cn þ 1 aromatics [127–129] via an ordinary acid catalyzed electrophilic substitution [130]. These methylation reactions occur readily at 200 C, well below the 250 C temperature required to rapidly initiate conversion of a pure methanol feed to hydrocarbons [131]. Once methylated, the trapped organic reaction centers [132–134] undergo isomerization and cracking reactions to produce light olefins. One of the interesting features of methanol conversion chemistry is that when methylation proceeds at rates exceeding cracking, the average molecular weight of the trapped, unsaturated intermediates increase leading to structures with faster cracking rates and more sterically demanding methylation possibilities. For this reason, it is reasonable to expect that the system automatically adjusts to any changes in conditions in a way that accelerates cracking reaction rates and reduces methylation rates until they come into balance. Olefins, aromatics and dienes each have unique methylation/isomerization/ cracking patterns allowing a wide range of possible olefin product distributions. For example, over the most studied catalyst, ZSM-5, reported ethylene selectivity varies from 0 to 32% depending upon the crystal size and shape, aluminum content and distribution and operating conditions (Table 1.5). At temperatures above 400 C and pressures below 0.1 atmospheres, trapped C6 þ olefins are the dominant intermediates. These olefins (in MTO formed from the
1.6 Zeolite Shape Selectivity Table 1.5 Sample kinetic product slates from various zeolites.
Zeolite
ZSM-5
ZSM-5
Erionite
Zeolite T
ZSM-34
REX
Product Ethylene Propylene Butenes C1–C4 paraffins C5þ
0 43 18 1 38
30 22 10 6 32
36 39 9 13 2
46 30 10 10 3
59 24 6 9 2
24 11 8 46 11
MeOH conversion Reference
21 [155]
95 [156]
10 [157]
11 [158]
54 [159]
50 [160]
conversion of methanol) are well known to crack over ZSM-5 with high selectivity to C3 þ olefins [135, 136]. Simultaneous methylation and cracking results in a near equilibrium distribution of C3 þ olefins (Scheme 1.31). The dominance of olefinic intermediates explains the ethylene free product distribution Table 1.5.
Scheme 1.31 Selective conversion of methanol to C3 þ olefins.
A number of researchers in the 1980s linked the selective conversion of methanol to ethylene and propylene over ZSM-5 to trapped aromatic reaction centers [137, 138]. Multiple 13 C labeling studies were conducted [139–144]. The best ethylene plus propylene selectivities are achieved by cofeeding aromatics to ZSM-5 [145], or using 8-ring catalysts such as ZSM-34, erionite, SAPO-18, SAPO-34, SAPO-35 and SAPO56 [146–149]. This body of work has proven that methylation, isomerization and cracking of aromatics leads to a highly non-equilibrium olefinic product dominated by ethylene and propylene (Scheme 1.32). MTO catalyzed by trapped aromatic organic reaction centers proceeds rapidly at temperatures as low as 250 C. The rearrangement of polymethyl to methylethyl and methylpropyl aromatics proceeds at the same absolute rate as aromatics methylation, deethylation and depropylation. Outside of small zeolite cages, the rearrangement of polymethyl to methylethyl and methylpropyl aromatics is known to be slower [150, 151]. Zeolite beta is inactive for hexamethylbenzene cracking at 350–375 C. Even at 550 C only partial conversion is obtained. The light products
j27
j 1 Zeolites in Catalysis
28
Scheme 1.32 Aromatics methylation, isomerization and cracking.
obtained from hexamethylbenzene cracking with zeolite beta show high yields of propane and isobutane, indicating that secondary reactions of the kinetic light olefin products proceed at faster rates than the hexamethylbenzene cracking. These results suggest that the rearrangement of polymethylaromatics to methylethyl and methylpropyl aromatics is accelerated by the cage environment of the many zeolite/ molecular sieve structures shown to have good ethylene selectivity [152]. Trapped organic reaction centers explain the variability in methanol conversion selectivity observed using ZSM-5 catalysts [153]. At most conditions ZSM-5 contains a mixture of olefin, diene and aromatic organic reaction centers. Each type of pool undergoes a different catalytic cycle with a characteristic product selectivity. Changes in catalyst structure, feedstock composition and reaction conditions change the molecular distribution of the organic reaction centers. Process selectivity is an average of the selectivity of the individual active sites. The separate catalytic cycles are linked by secondary hydrogen transfer reactions. Conjunct polymerization reactions of the primary light olefin products convert olefinic organic reaction centers to diene and then aromatic reaction centers (Scheme 1.33) [154].
Scheme 1.33 Proposed conversion of olefins to diene and aromatic organic reaction centers.
References
1.7 Counter Ion Mobility
Zeolite Brønsted acid sites can be partially exchanged with counter ions such as rare earth, magnesium and calcium ions. The resulting catalyst is considerably more complex. The mobility of the cations can result in interesting shape selective effects. The addition of rare earth cations to FAU FCC catalysts will be used to illustrate this unusual type of shape selectivity. Rare earth ions are added to FAU FCCcatalysts in order to increase the relative rate of hydrogen transfer reactions vs. further olefin cracking. A plausible mechanism for the acceleration of hydrogen transfer by rare earth ions in FAU is shown in Scheme 1.34.
Scheme 1.34 Plausible mechanism for rare earth ion acceleration of hydrogen transfer reactions.
1.8 Conclusions
Zeolite catalysts are indispensable tools in the refining and petrochemicals industries. Manipulation of zeolite structures and reaction conditions has enabled the development of processes that direct the conversion of crude oil and natural gas liquids into the ever-changing distribution of commodity fuels and petrochemicals needed in the marketplace. The major inputs to the zeolite based processes are hydrocarbon feedstocks, energy and catalyst. The zeolite catalysts are closely related in structure and composition to beach sand and are much less hazardous than mineral acids. The industry will continue to replace processes using mineral acids with zeolite processes wherever possible. Existing zeolite process technologies will continue to evolve to meet the goals of producing higher yields of fuels and petrochemicals with ever decreasing energy input requirements.
References 1 Williamson, H.F., Andreano, R.L., Daum, A.R. and Klose, G.C. (1963) The Age of Energy, Northwestern Univ. Press, Evanston, IL, pp. 619. 2 Spitz, P.H. (1988) Petrochemicals, The Rise of an Industry, John Wiley and Sons, Inc., New York, pp. 12.
3 Hansford, R.C. (1983) in Heterogeneous Catalysis: Selected American Histories, edited by Davis and Hettinger. ACS Symposium Series 222, Washington, DC. 4 Breck, D.W. (1974) Zeolite Molecular Sieves: Structure, Chemistry and Use, Wiley-Interscience, New York.
j29
j 1 Zeolites in Catalysis
30
5 Plank, C.J. (1983) in Heterogeneous Catalysis: Selected American Histories, edited by Davis and Hettinger. ACS Symposium Series 222, Washington, DC. 6 Bowie, R.R. and Ellison, M.E. (1993) Environmental management and the petroleum industry. APEA Journal, 33 (Pt 1), 395–400. 7 Li, Yongpu. (1999) Discussion on the prospects of entering international market by analysis of FCC technology and catalyst development in China. Shiyou Lianzhi Yu Huagong, 30 (8), 29–33. 8 Meier, W.M., Olson, D.H. and Baerlocher, Ch. (1996) Atlas of Zeolite Structure Types Fourth Revised Edition, Elsevier. 9 Szostak, R. (1992) Handbook of Molecular Sieves, Van Nostrand Reinhold, New York. 10 Meier, W.M., Olson, D.H. and Baerlocher, Ch. (1996) Atlas of Zeolite Structure Types Fourth Revised Edition, Elsevier. 11 Freude, P., Dehme, W., Schmiedel, W. and Staudte, B. (1972) Journal of Catalysis, 26, 191. 12 Wu, E.L., Kuhl, G.H., White, T.E. and Venuto, P.B. (1971) Advances in Chemistry Series, 101, 490. 13 Rabo, J.A. and Poutsma, M.L. (1971) Advances in Chemistry Series, 102, 284. 14 Ward, J.W. (1971) Advances in Chemistry Series, 101, 380. 15 Beaumont, R. and Barthomeuf, D. (1972) Journal of Catalysis, 26, 218. 16 Vedrine, J.C., Abou-Kais, A., Massardier, J. and Dalmai-Imelik, G. (1973) Journal of Catalysis, 29, 120. 17 Loewenstein, W. (1954) The American Mineralogist, 39, 92. 18 Jacobs, P.A. (1982) Catalysis ReviewsScience and Engineering, 24, 415. 19 Barthomeuf, D. (1977) Journal of the Chemical Society. Chemical Communications, 74. 20 McMicol, B.D. and Pott, G.T. (1972) Journal of Catalysis, 25, 223. 21 Lago, R.M., Haag, W.O., Mikovsky, R.J., Olson, D.H., Hellring, S.D., Schmitt, K.D. and Kerr, G.T. (1986) The nature of the catalytic sites in HZSM-5-activity
22
23
24
25
26
27
28 29 30 31 32
33
enhancement. Advances in Chemical Conversions for Mitigating Carbon Dioxide, 28, (New Dev. Zeolite Sci. Technol.), 677–684. Haw, James F. and Xu, Teng. (1998) NMR studies of solid acidity. Advances in Catalysis, 42, 115–180. Haw, James F. (March 29–April 2 1998) In situ NMR and ab initio studies of zeolite acidity, Department Chemistry, Texas A and M University, College Station, TX, USA. Book of Abstracts, 215th ACS National Meeting, Dallas, CATL-043. Nicholas, John B., Haw, James F., Beck, Larry W., Krawietz, Thomas R. and Ferguson, David B. (1995) Density Functional Theoretical and NMR Study of Hammett Bases in Acidic Zeolites. Journal of the American Chemical Society, 117 (49), 12350–12351. Haag, W.O. (1984) Acid catalysis with medium pore zeolites. in Proc Int Zeolite Conf, 6th (eds David Olson and Attilio Bisio) Meeting Date 1983, pp. 466–478. Babitz, S.M., Williams, B.A., Miller, J.T., Snurr, R.Q., Haag, W.O. and Kung, H.H. (1999) Monomolecular cracking of nhexane on Y, MOR and ZSM-5 zeolites. Applied Catalysis A-General 179 (1–2), 71–86. Vinek, Hannelore. and Lercher, Johannes A. (1991) Production and reactions of xylenes over H-ZSM5. Journal of Molecular Catalysis, 64 (1), 23–39. Wilson, S.T. et al. (Jan. 12 1982) U.S. Patent 4,310,440. Wilson, S.T. and Flanigan, E.M. (Jan. 28 1986) U.S. Patent 4,567,029. Lok, B.M. et al. (Feb. 19 1985) U.S. Patent 4,500,651. Nenitzescu, C.D. and Cantuniari, I.P. (1933) Chemische Berichte, 66, 1097. Pines, H. and Wacker, R.C. (1946) Journal of the American Chemical Society, 68, 1642, 595 and 2518. Pines, H., Abraham, B.M. and Ipatieff, V.N. (1948) Journal of the American Chemical Society, 70, 1742.
References 34 Corma, A. and Orchilles, A.V. (2000) Current views on the mechanism of catalytic cracking. Microporous Mesoporous Mater, 35–36 21–30. 35 Olah, G.A. and Molnar, A. (1995) Hydrocarbon Chemistry, John Wiley and Sons, Inc, New York. 36 Occelli, M.L., Kalwei, M., Wolker, A., Eckert, H., Auroux, A. and Gould, S.A.C. (2000) The use of nuclear magnetic resonance, microcalorimetry and atomic force microscopy to study the aging and regeneration of fluidized cracking catalysts. Journal of Catalysis, 196 (1), 134–148. 37 Chen, N.Y., Garwood, W.E. and Dwyer, F.G. (1996) Shape Selective Catalysis in Industrial Application, Marcel Dekker, N.Y. 38 Al-Khattaf, S. and De Lasa, H. (1999) Activity and Selectivity of Fluidized Catalytic Cracking Catalysts in a Riser Simulator: The Role of Y-Zeolite Crystal Size. Industrial & Engineering Chemistry Research, 38 (4), 1350–1356. 39 Moser, William R., Suer, Murat G., Dardas, Zissis and Ma, Yi Hua (1998) Mechanism of the catalytic cracking of heptane under supercritical fluid conditions Preprints-American Chemical Society, Division of Petroleum Chemistry, 43 (3), 450–453. 40 Nicholas, John B. and Haw, James F. (1998) The Prediction of Persistent Carbenium Ions in Zeolites. Journal of the American Chemical Society, 120 (45), 11804–11805. 41 Weitkamp, J., Jacobs, P.A. and Martens, J.A. (1983) Applied Catalysis, 8, 123. 42 Buchanan, J.S., Santiesteban, J.G. and Haag, W.O. (1996) Mechanistic Considerations in Acid-Catalyzed Cracking of Olefins. Journal of Catalysis, 158, 279–287. 43 Haag, W.O., Dessau, R.M. and Lago, R.M. (1991) Kinetics and mechanism of paraffin cracking with zeolite catalysts. Advances in Chemical Conversions for Mitigating Carbon Dioxide, 60, (Chem. Microporous Cryst.), 255–265.
44 Stefanadis, C., Gates, B.C. and Haag, W.O. (1991) Rates of isobutane cracking catalyzed by HZSM-5: the carbonium ion route. Journal of Molecular Catalysis, 67 (3), 363–367. 45 Boronat, Merce, Viruela, Pedro and Corma, Avelino (1999) Theoretical Study of Bimolecular Reactions between Carbenium Ions and Paraffins: The Proposal of a Common Intermediate for Hydride Transfer, Disproportionation, Dehydrogenation and Alkylation. The Journal of Physical Chemistry. B, 103 (37), 7809–7821. 46 Abbot, J. and Wojciechowski, B.W. (1987) Hydrogen transfer reactions in the catalytic cracking of paraffins. Journal of Catalysis, 107 (2), 451–462. 47 Sullivan, R.F., Egan, C.J., Langlois, G.E. and Seig, R.P. (1961) A New Reaction in the Hydrocracking of Certain Aromatic Hydrocarbons. Journal of the American Chemical Society, 83, 1156. 48 DeVries, L. (1960) The Non-Classical Pentamethylcyclopentadienyl-Carbinyl Cation. Journal of the American Chemical Society, 82, 5242. 49 Winstein, S. and Battiste, M. (1960) Cyclopentadienylmethyl Derivatives as Homoallylic Systems. Journal of the American Chemical Society, 82, 5244. 50 Childs, R.F., Winstein, S. and Sakai, M. (1968) The Observation and Behavior of the Pentamethylcyclopentadienylmethyl Cation. Journal of the American Chemical Society, 90, 7144. 51 Childs, R.F. and Winstein, S. (1968) Ring Opening and Fivefold Degenerate Scrambling in Hexa and Heptamethylbicyclo[3.1.0]hexenyl Cations. Journal of the American Chemical Society, 90, 7146. 52 Schmerling, L. and Ipatieff, V.N. (1950) Advances in Catalysis, 2, 21. 53 Ipatieff, V.N. and Pines, H. (1935) Industrial & Engineering Chemistry, 27, 1364. 54 Whitmore, F.C. and Meunier, P.L. (1941) Journal of the American Chemical Society, 63, 2197.
j31
j 1 Zeolites in Catalysis
32
55 Whitmore, F.C., Ropp, W.S. and Cook, N.C. (1950) Journal of the American Chemical Society, 72, 1507. 56 Pepper, D.C. (1964) Friedel–Crafts and Related Reactions (ed. G.A. Olah), Vol. 2, Wiley Interscience, New York, Chapter 30, pp. 1293. 57 Pines, H. (1981) The Chemistry of Hydrocarbon Conversion, Academic Press, New York, pp. 35. 58 Garwood, W.E. (1983) American Chemical Society, Symposium Series, 218, 383. 59 Tabak, S.A., Krambeck, F.J. and Garwood, W.E. (August 1984) Conversion of Propylene and Butylene over ZSM-5 Catalyst. paper presented at the AIChE National Meeting, Philadelphia. 60 Martens, Johan A., Verrelst, Wim H., Mathys, Georges M., Brown, Stephen H. and Jacobs, Pierre A. (2005) Tailored catalytic propene trimerization over acidic zeolites with tubular pores. Angewandte Chemie, International Edition, 44 (35), 5687–5690. 61 Pines, H. (1981) The Chemistry of Hydrocarbon Conversion, Academic Press, New York. 62 Ipatieff, V.N., Pines, H. and Olberg, R.C. (1948) Journal of the American Chemical Society, 70, 2123. 63 Schriesheim, A. Alkylation of aromatics with alcohols and ethers. In Friedel– Crafts and Related Reactions (G.A. Olah, ed.) Vol. 2, Part I. 477–597. Wiley Interscience, New York. 64 Venuto, P.B. (1994) Organic catalysis over zeolites: a perspective on reaction paths within micropores. Microporous Materials, 2, 297–411. 65 Pines, H. (1981) The Chemistry of Hydrocarbon Conversion, Academic Press, New York. 66 Martens, J.A. and Jacobs, P.A. (1990) Theoretical Aspects of Heterogeneous Catalysis (ed. J.B. Moffat), Van Nostrand, New York, pp. 52. 67 Vogel, P. (1985) Carbocation Chemistry, Elsevier, Amsterdam, pp. 323.
68 Degnan, T.F. and Kennedy, C.R. (1993) AIChE Journal, 39 (4), 607. 69 Burwell, R.L. and Shields, A.D. (1955) Journal of the American Chemical Society, 77, 2766. 70 McCaulay, D.A. and Lien, A.P. (1953) Journal of the American Chemical Society, 77 (75), 2411. 71 Brown, H.C. and Smoot, C.R. (1956) Journal of the American Chemical Society, 78, 2176. 72 Unseren, E. and Wolf, A.P. (1962) The Journal of Organic Chemistry, 27, 1509. 73 Streitweiser, A. and Reif, L.J. (1960 and 86, 1988, 1964) Journal of the American Chemical Society, 82, 5003. 74 Brown, H.C. and Smoot, C.R. (1956) Journal of the American Chemical Society, 78, 2176. 75 Ertl, G., Knozinger, H., Schuth, F. and Weitkamp, J. (April 2008) Handbook of Heterogeneous Catalysis (eds D.L. Stern, J.S. Beck and S.H. Brown), Wiley Press, Aromatics Alkylation and Transalkylation. 76 Guisnet, M., Gnep, N.S. and Morin, S. (2000) Mechanisms of xylene isomerization over acidic solid catalysts. Microporous and Mesoporous Materials, 35–36, 47–59. 77 Sullivan, R.F., Egan, C.J., Langlois, G.E. and Seig, R.P. (1961) A New Reaction in the Hydrocracking of Certain Aromatic Hydrocarbons. Journal of the American Chemical Society, 83, 1156. 78 Xiong, Y., Rodewald, P.G. and Chang, C.D. (1995) Journal of the American Chemical Society, 117, 9427. 79 Kilpatrick, M. and Luborsky, F. (Feb. 1953) Journal of the American Chemical Society, 75, 577 80 Brown, H.C. and Smoot, C.R. (1956) Journal of the American Chemical Society, 78, 2176. 81 Arsenove, N., Haag, W.O. and Karge, H.G. (1997) Advances in Chemical Conversions for Mitigating Carbon Dioxide, 105, 1293–1300. 82 Guisnet, M., Gnep, N.S. and Morin, S. (2000) Mechanisms of xylene
References
83
84
85 86
87
88 89
90
91
92
isomerization over acidic solid catalysts. Microporous and Mesoporous Materials, 35–36, 47–59. Nivarthy, Gautam S., Feller, Andreas, Seshan, K. and Lercher, Johannes A. (2000) The role of hydride transfer in zeolite catalyzed isobutane/butene alkylation. Advances in Chemical Conversions for Mitigating Carbon Dioxide, 130, (International Congress on Catalysis, 2000, Pt. C) 2561–2566. Pines, H. (1981) The Chemistry of Hydrocarbon Conversion, Academic Press, New York. Pines, H. (1972) Intra-Sci Chem Rep, 6 (2), 1–41. Chen, N.Y. and Haag, W.O. (1988) Hydrogen transfer in catalysis on zeolites. Chemistry & Industry (Dekker), 31, (Hydrogen Eff. Catal.), 695–722. Haw, James F., Nicholas, John B., Song, Weiguo, Deng, Feng, Wang, Zhike, Xu, Teng and Heneghan, Catherine S. (2000) Roles for Cyclopentenyl Cations in the Synthesis of Hydrocarbons from Methanol on Zeolite Catalyst HZSM-5. Journal of the American Chemical Society, 122 (19), 4763–4775. Lukyanov, D.B. (1994) Journal of Catalysis, 145, 54. Miller, S.J. (1987) Advances in Chemical Conversions for Mitigating Carbon Dioxide, 38, 187. Fritzsche, S., Haberlandt, R., Jost, S. and Schuring, A. (2000) Modeling diffusion in zeolites by molecular dynamics simulations. Molecular Simulation, 25 (1–2), 27–40. Keil, Frerich J., Krishna, Rajamani and Coppens, Mare-Olivier (2000) Modeling of diffusion in zeolites. Reviews in Chemical Engineering, 16 (2), 71–197. Auerbach, Scott M. (2000) Theory and simulation of jump dynamics, diffusion and phase equilibrium in nanopores, Department of Chemistry and Department of Chemical Engineering, University of Massachusetts, Amherst,
93
94
95
96
97
98
99
100
101
MA, USA. International Reviews in Physical Chemistry, 19 (2), 155–198. Chen, Yudong and Yang, Ralph T. (1997) Multicomponent diffusion in zeolites and multicomponent surface diffusion. Advances in Chemical Conversions for Mitigating Carbon Dioxide, 104, (Equilibria and Dynamics of Gas Adsorption on Heterogeneous Solid Surfaces), 487–518. Utada, Taku (1997) Recent problems concerning reactions and diffusion in zeolites. Kagaku Kogaku, 61 (12), 944–948. Theodorou, Doros N., Snurr, Randall Q. and Bell, Alexis T. (1996) Molecular dynamics and diffusion in microporous materials, Editor(s): Alberti, Giulio; Bein, Thomas. Comprehensive Supramolecular Chemistry, 7, 507–548. Weisz, Paul B. (1995) Department of Bioengineering, University of Pennsylvania, Philadelphia, PA, USA. Molecular Diffusion in Microporous Materials: Formalisms and Mechanisms. Industrial & Engineering Chemistry Research, 34 (8), 2692–2699. Masuda, Takao (2003) Div. of Materials Science and Engineering, Graduate School of Engineering, Hokkaido University, Sapporo, Japan. Peculiar diffusion mechanisms within micropores of zeolite catalysts. Journal of the Japan Petroleum Institute, 46 (5), 281–294. Kaerger, Joerg and Freude, Dieter (2001) Universitat Leipzig, Leipzig, Germany Mass transfer in micro- and mesoporous materials. Chemie Ingenieur Technik, 73 (12), 1517–1526. Chen, N.Y. and Haag, W.O. (1988) Hydrogen Effects in Catalysis (eds Z. Paal and P.G. Menon), Marcel Dekker, New York, pp. 695–722. Venuto, P.B. (1994) Organic Catalysis over zeolites: a perspective on reaction paths within micropores. Microporous Materials, 2, 378, and Tables 2–24. Roseler, J., Heitmann, G. and Holderich, W.F. (1997) Vapor-phase Beckmann rearrangement using B-MFI zeolites.
j33
j 1 Zeolites in Catalysis
34
102
103 104
105
106 107 108
109
110
Advances in Chemical Conversions for Mitigating Carbon Dioxide, 105, (Progress in Zeolite and Microporous Materials, Pt. B), 1173–1180. CODEN: SSCTDM ISSN: 0167–2991. Heitmann, G.P., Dahlhoff, G. and Holderich, W.F. (1999) Catalytically active sites for the Beckmann rearrangement of cyclohexanone oxime to e-caprolactam. Journal of Catalysis, 186 (1), 12–19. Venuto, P.B. (1994) Microporous Materials, 2, 386. Abichandani, J.S., Beck, J.S., Fischer, R.H., Johnson, I.D. and Stern, D.L. (1997) US Patent 5,625,103, assigned to Mobil Oil. Beck, J.S., Dessau, R.M. and Olson, D.H. (1997) US 5,675,047, assigned to Mobil Oil; Beck, J.S. and Stern, D.L. (1997) US Patent 5,625,104, assigned to Mobil Oil; Beck, J.S. and Stern, D.L. (1997) US Patent 5,612,270, assigned to Mobil Oil; Beck, J.S. and Stern, D.L. (1997) US Patent 5,602,066, assigned to Mobil Oil; Beck, J.S., McCullen, S.B. and Olson, D.H. US Patent 5,495,059; Beck, J.S., McCullen, S.B. and Olson, D.H. (1995) US Patent 5,476,823, assigned to Mobil Oil; Beck, J.S., McCullen, S.B., Olson, D.H. (1993) US Patent 5,403,800, assigned to Mobil Oil; Beck, J.S. (1993) US Patent 5,191,137, assigned to Mobil Oil. Verduijn, J.P. and Mortier, W.J. (1999) WO 9928032 A1, assigned to ExxonMobil. Haag, W.O. (1984) Proc Int Zeolite Conf, 6th, 466–478. Olson, D.H. and Haag, W.O. (1984) Mobil Res and Dev Corp, Princeton, NJ, USA. ACS Symposium Series, 248 (Catalytic Materials: Relationship between Structure and Reactivity), 275–307. Weisz, P.B. (1980) Pure and Applied Chemistry, 52, 2091; Olson, D.H. and Haag, W.O. (1984) ACS Symposium Series, 248, 275. Martens, Johan A., Verrelst, Wim H., Mathys, Georges M., Brown, Stephen H. and Jacobs, Pierre A. (2005) Tailored catalytic propene trimerization over acidic
111
112
113
114
115
116 117 118
119
120 121 122 123 124
125
zeolites with tubular pores. Angewandte Chemie, International Edition, 44 (35), 5687–5690. Chen, N.Y., Garwood, W.E. and Dwyer, F.G. (1996) Shape Selective Catalysis in Industrial Application, Marcel Dekker, New York. Ertl, G., Knozinger, H., Schuth, F. and Weitkamp, J. (April 2008) Handbook of Heterogeneous Catalysis, Wiley Press, Olefin Oligomerization. Martens, Johan A., Verrelst, Wim H., Mathys, Georges M., Brown, Stephen H. and Jacobs, Pierre A. (2005) Tailored catalytic propene trimerization over acidic zeolites with tubular pores. Angewandte Chemie, International Edition, 44 (35), 5687–5690. Sapre, A., Hilbert, T., Tabak, S., Cody, I. and Hantzer, S. (May/June 2004) Hydrocarbon Asia, 50. Calemma, V., Peratello, S. and Perego, C. (2000) Applied Catalysis A: General, 190, 207–218. Deldari, H. (2005) Applied Catalysis A: General, 293, 1–10. Kolboe, S. (1986) Acta Chemica Scandinavica A, 40, 711–713. Kolboe, S. (1988) Methane Conversion (eds D.M. Bibby, C.D. Chang, R.F. Howe and S. Yurchak) Elsevier Science Publishers B.V., Amsterdam. Dessau, R.M. (1986) On the H-ZSM-5 Catalyzed Formation of Ethylene from Methanol or Higher Olefins. Journal of Catalysis, 99, 111–116. Haag, W.O. (1983) Proc 6th Int Zeol Conf, Butterworth, London, pp. 466–478. Chang, C.D. (1984) Catalysis Reviews: Science and Engineering, 26, 323. Keading, W.W. and Butter, S.A. (1980) Journal of Catalysis, 61, 155. Dessau, R.M. (1986) Journal of Catalysis, 99, 111. Haag, W.O. Lago, R.M. and Rodewald, P.G. (1982) Journal of Molecular Catalysis, 17, 161. Dessau, R.M. and LaPierre, R.B. (1982) Journal of Catalysis, 78, 136.
References 126 Haw, James F., Nicholas, John B., Song, Weiguo, Deng, Feng, Wang, Zhike, Xu, Teng and Heneghan, Catherine S. (2000) Roles for Cyclopentenyl Cations in the Synthesis of Hydrocarbons from Methanol on Zeolite Catalyst HZSM-5. Journal of the American Chemical Society, 122 (19), 4763–4775. 127 Mirth, G. and Lercher, J.A. (1991) In Situ Spectroscopic Study of the Surface Species during Methylation of Toluene over H-ZSM-5. Journal of Catalysis, 132, 244–322. 128 Lercher, J.A. and Mirth, G. (1991) Journal of Catalysis, 147, 199–206. 129 Dessau, R.M. and LaPierre, R.B. (1982) Journal of Catalysis, 78, 136. 130 Mole, T. (1988) Isotopic and Mechanistic Studies of Methanol Conversion, in Methane Conversion (eds D.M. Bibby, C.D. Chang, R.F. Howe and S. Yurchak) Elsevier Science Publishers B.V., Amsterdam, pp. 145–156. 131 Kolboe, S. (1988) Methane Conversion (eds D.M. Bibby, C.D. Chang, R.F. Howe and S. Yurchak), Elsevier Science Publishers B.V., Amsterdam. 132 Haw, James F., Nicholas, John B., Song, Weiguo, Deng, Feng, Wang, Zhike, Xu, Teng and Heneghan, Catherine S. (2000) Roles for Cyclopentenyl Cations in the Synthesis of Hydrocarbons from Methanol on Zeolite Catalyst HZSM-5. Journal of the American Chemical Society, 122 (19), 4763–4775. 133 Haw, James F. (March 26–30 2000) Synthesis and NMR characterization of organic reaction centers on solid acid catalysts. Book of Abstracts, 219th ACS National Meeting, San Francisco, CA, CATL-069. Publisher: American Chemical Society, Washington, D.C. 134 Song, Weiguo, Haw, James F., Nicholas, John B. and Heneghan, Catherine S. (2000) Methylbenzenes Are the Organic Reaction Centers for Methanol-to-Olefin Catalysis on HSAPO–34. Journal of the American Chemical Society, 122 (43), 10726–10727.
135 Dessau, R.M. (1986) Journal of Catalysis, 99, 111. 136 Buchanan, J.S. Santiesteban, J.G. and Haag, W.O. (1996) Journal of Catalysis 158, 279–287. 137 Mole, T. and Bett, G. (1983) Conversion of Methanol to Hydrocarbons over ZSM-5 Zeolite: An Examination of the Role of Aromatic Hydrocarbons Using 13Carbon and Deuterium Labeled Feeds. Journal of Catalysis, 84, 435–445. 138 Seddon, D., Mole, T. and Whiteside, J.A. (1985) US 4,499,314. 139 Mole, T. and Bett, G. (1983) Conversion of Methanol to Hydrocarbons over ZSM-5 Zeolite: An Examination of the Role of Aromatic Hydrocarbons Using 13Carbon and Deuterium Labeled Feeds. Journal of Catalysis, 84, 435–445. 140 Seddon, D. and Mole, T. (1985) Methanol Conversion to Hydrocarbons with Zeolites and Cocatalysts U.S. 4,499,314. 141 Seddon, D. and Mole, T. (June 10. 1982), WO 82/08166. 142 Isogulyants, G.V., Kovalenko, L.I. and Dubinskii, U.G. (November 1987) Izvestiya Akademii Nauk SSSR, Seriya Khimicheskaya (11), 2443–2448. 143 Kolboe, S. (1995) Studies in Surface Science Catalysis, 98, 176–177. 144 Dessau, R.M. and LaPierre, R.B. (1982. Journal of Catalysis, 78, 136. 145 Kolboe, S., Ronning, O. and Mikkelsen, O. (2000) Microporous and Mesoporous Materials, 40, 95–113. 146 Hall Richard B. et al. (May 12–17 2007) Cage Shape Effects in MTO Catalysis First IDECAT conference on catalysis. 147 Song, Weiguo, Fu, Hui and HawJames F. (2001) Supramolecular Origins of Product Selectivity for Methanol-to-Olefin Catalysis on HSAPO-34. Journal of the American Chemical Society, 123 (20), 4749–4754. 148 Arsted, Bjornar and Kolboe, Stein (2001) The Reactivity of Molecules Trapped within the SAPO–34 Cavities in the Methanol to Hydrocarbons Reaction. Journal of the American Chemical Society.
j35
j 1 Zeolites in Catalysis
36
149 Song, Weiguo, Haw, James F., Nicholas, John B. and Heneghan, Catherine S. (2000) Methylbenzenes Are the Organic Reaction Centers for Methanol-to-Olefin Catalysis on HSAPO-34. Journal of the American Chemical Society, 122 (43), 10726–10727. 150 Sassi, Alain, Wildman, Mark A. and Haw, James F. (2002) Loker Hydrocarbon Research Institute and Department of Chemistry, University of Southern California, Los Angeles, CA, USA. Reactions of Butylbenzene Isomers on Zeolite H Beta: Methanol-to-Olefins Hydrocarbon Pool Chemistry and Secondary Reactions of Olefins. Journal of Physical Chemistry B, 106 (34), 8768–8773. 151 Kolboe, S., Ronning, O. and Mikkelsen, O. (2000) Microporous and Mesoporous Materials, 40, 95–113. 152 Hall, Richard B. et al. (May 12–17) Cage Shape Effects in MTO Catalysis First IDECAT conference on catalysis. 153 Bjørgen, Morten, Svelle, Stian, Joensen, Finn, Nerlov, Jesper, Kolboe, Stein, Bonino, Francesca, Palumbo, Luisa, Bordiga, Silvia and Olsbye, Unni (2007) Conversion of methanol to hydrocarbons
154
155 156
157
158
159
160
over zeolite H-ZSM-5: On the origin of the olefinic species. Journal of Catalysis, 249, 195–207. Pines, H. (1981) The Chemistry of Hydrocarbon Conversion, Academic Press, New York, pp. 42. Dessau, R.M. (1986) Journal of Catalysis, 99, 111. Mole, T. (1983) Conversion of Methanol to Ethylene over ZSM-5 Zeolite. Journal of Catalysis, 84, 423–434. Chang, C.D. Hydrocarbons from Methanol Copyright 1983 by Marcel Dekker Inc. Chang, C.D. Hydrocarbons from Methanol Copyright 1983 by Marcel Dekker Inc. Liu, L., Tobias, R.G., McLaughlin, K. and Anthony, R.G. (1984) Conversion of Methanol to Low Molecular Weight Olefins with Heterogeneous Catalysts. in Catal. Convers. Synth. Gas Alcohols Chem., [Proc. Symp.] (ed. Richard G. Herman), Meeting Date 1983, pp. 323–360. Chang, C.D. Hydrocarbons from Methanol Copyright 1983 by Marcel Dekker Inc.
j37
2 Sol–Gel Sulfonic Acid Silicas as Catalysts Adam F. Lee and Karen Wilson
2.1 Introduction
Catalytic technologies play a key role in the economic development and growth of both the chemical industry and modern society. Ever tightening environmental legislation is constantly driving the fine and speciality chemicals industries to consider alternative processes that avoid the use of hazardous reagents and chemicals. The development of new catalytic routes for the manufacture of fine, speciality and pharmaceutical chemicals is one of the principal solutions for addressing the sustainability and environmental impact of chemical processes. Such routes provide a means to eliminate the use of toxic reagents, reduce waste and promote the implementation of renewable resources. Acid-initiated reactions are widely used in, for example, alkylation, acylation, esterification and isomerization and a major area where new catalyst systems are required. While zeolites are widely used as solid acid catalysts in gas-phase chemistry, their use in liquid-phase organic synthesis is limited by their small pore sizes (<1 nm), which make them unsuitable for reactions involving bulky substrates [1]. The discovery of mesoporous molecular sieves [2], offering pore sizes in the range 2–10 nm, and the subsequent development of a range of templated materials have opened up new avenues for liquid-phase solid acid catalysis. Silica-supported sulfonic acids are a class of solid Brønsted acid catalysts, that generally comprise organo-sulfonic acid groups tethered to silica surfaces. Methodologies to prepare organically modified silica have been widely developed in separation science [3] and the techniques for their preparation are well documented. For example, silica functionalized with sulfonic acid or thiol groups is frequently used for binding metal ions [4, 5]. The application of this chemistry to prepare pure Brønsted sulfonic acid functionalized mesoporous silicas [6–8] has stimulated significant research effort in this area, since these materials are interesting alternatives to commercially available sulfonated polymer resins, such as Amberlyst-15 and Nafion-H (sulfonated polystyrene and perfluorinated sulfonic acid resins, respectively), which suffer from low surface areas and thermal stability [9–11]. This chapter presents an overview of the preparation of mesostructured silica-supported
Handbook of Green Chemistry, Volume 2: Heterogeneous Catalysis. Edited by Robert H. Crabtree Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32497-2
j 2 Sol–Gel Sulfonic Acid Silicas as Catalysts
38
sulfonic acids and their catalytic applications and reviews the approaches taken to tune catalyst performance in organic transformations.
2.2 Preparation of Meso–structured Silica Sulfonic Acid Catalysts 2.2.1 Templating Methods
The discovery of the M41S class of mesoporous materials by Mobil Research and Development opened up a new approach to the synthesis of porous materials, whereby micellar surfactant arrays or supramolecular assemblies were used as templating agents to direct the crystallization of SiO2 frameworks, as illustrated in Figure 2.1. Subsequent calcination to burn out organic templates, resulted in materials with welldefined meso–structured (2–10 nm) pores and high surface areas up to 1000 m2 g1. The preparation of templated materials has recently been reviewed [1, 12–15]. The morphology and stability of porous networks are strongly dependent on the templating conditions (ionic or neutral), silica precursor [e.g. fumed silica, tetraethyl orthosilicate (TEOS), Ludox or sodium silicate] and whether hydrothermal synthesis is employed. The key points pertaining to the synthesis of silica-based sulfonic acids, in particular the effect of template, will be summarized here. 2.2.1.1 Cationic/Anionic Templates Cationic templating routes, employing cetyltrimethylammonium bromide (CTAB) surfactants in conjunction with a base, were employed in the original synthesis of
Figure 2.1 Preparation of micelle-templated silicas.
2.2 Preparation of Meso–structured Silica Sulfonic Acid Catalysts
MCM-type materials. Here the observed electrostatic interaction with the head group is strong and does not favor template extraction and porous materials are normally only obtained after calcination. The original MCM materials had excellent long-range order due to the strong interaction between the silica–micellar arrays, but the silica walls were fairly thin, raising concern over their hydrothermal stability when employed as catalyst supports. The requirement to remove the template by calcination also meant that any organic modification had to be performed by post-grafting methods, which limits the degree of surface functionalization achievable (see Section 2.2.2). However, Dıaz et al. [8] have reported efficient template extraction by HCl–EtOH, opening up possible incorporation of organic functionality during the templating process. Furthermore, thicker walled, more hydrothermally stable MCM-type materials are formed if NH3 is used to catalyze the hydrolysis of TEOS [16]. 2.2.1.2 Neutral Templates Mesoporous materials have also been synthesized employing neutral templates such as long-chain amine or poly(ethylene oxide) (PEO) surfactants. In this instance the silica precursor interacts with the surfactant head group via hydrogen bonding, which is sufficiently weak that template removal (and in some cases recycling) can be readily facilitated by solvent extraction. Hexagonal mesoporous silicas (HMS [17]) are prepared using C8–C18 amine templates and give rise to mesostructured materials with pore diameters of 1.6–2.7 nm. The increased wall thicknesses and stability are improved compared with MCM; however, the pore structures have less long-range order due to the weak hydrogen-bonded interaction of the templated mesostructure. In contrast, PEO templated materials prepared using surfactants such as Pluronic 123 (EO20PO70EO20), Triton X-114 [CH3C(CH3)2CH2C(CH3)2(C6H4)(EO)7OH] and Brij 76 (C18EO10) not only enable larger pore materials (up to 10 nm) to be prepared, but also have the added advantage that they operate via a true liquid crystal templating mechanism (TLCT), enabling the pore structure to be predicted from the liquid crystalline phase [18]. The large pore diameters (up to 6 nm) of the SBA-15 class of materials prepared via the Pluronic 123 template has attracted a great deal of interest for preparing sulfonic acid catalysts. 2.2.2 Organically Functionalized Silica
There are two general methods employed to incorporate organo-functional groups into silica-based materials: post-modification by grafting of organo-silanes on to preformed supports and one pot approaches whereby the organic precursor is incorporated in the sol–gel process as illustrated in Scheme 2.1. The sol–gel preparation is performed in an analogous way to that of micelle templated silicas, with the addition of a small quantity of organo-alkoxysilane [RSi(OEt3)] used to incorporate the organic functionality in the silica framework. Facile template removal is essential, however, if these co-condensation routes are employed, as obviously calcination is not feasible if the organically derivatized surface is to be retained. As a result, most early MCM-based materials were prepared
j39
j 2 Sol–Gel Sulfonic Acid Silicas as Catalysts
40
Scheme 2.1 Sol–gel and grafting routes for the preparation of sulfonic acid silica.
via the grafting route by refluxing preformed mesoporous silica with mercaptopropyltrimethoxysilane [SHC3H6Si(OMe)3] (MPTS). 2.2.2.1 Characterization Having obtained an organically modified silica material via grafting or sol–gel routes, it is of paramount importance that the formation of sulfonic acid moieties can be confirmed. There are numerous techniques that can be utilized to affect this identification, including XPS, Raman and MAS-NMR spectroscopy. The presence of SH groups and their conversion to SO3H can be followed from their characteristic fingerprints in the infrared region. However, DRIFTS analysis can prove inconclusive as the SH stretch at 2700 cm1 is weak in the infrared region and the SiOSi lattice vibrations observed at 1000–1400 cm1 obscure their characteristic ns and nas SO3 vibrational modes at 1040 and 1100 cm1. In contrast, Raman spectroscopy is a powerful alternative to monitoring the thiol to sulfonic acid conversion, since it is more sensitive to the SH stretch and the SiOSi lattice modes are very weak in Raman. Figure 2.2 shows a typical Raman spectrum for asprepared and oxidized mercaptopropyl functionalized silica compared with the corresponding DRIFT spectra. The formation of undesired disulfide species is easily probed by Raman from their characteristic vibrations at 500–550 cm1. X-ray photoelectron spectroscopy (XPS) is also a powerful technique for confirming complete oxidation of surface thiol groups and quantifying the concentration of surface sulfonic acid groups. The binding energy of the S 2p electron is very sensitive to its local chemical environment, ranging from 162.2 eV for S(II) in a thiol environment to 167.7 eV for S(VI) in SO3H (Figure 2.3). Furthermore, XPS can be
2.2 Preparation of Meso–structured Silica Sulfonic Acid Catalysts
Figure 2.2 Comparison of DRIFTS and Raman measurements on thiol and sulfonic acid functionalized silicas.
sensitive down to 0.1 at.% and therefore essential for determining low surface S concentrations. Modern instrumentation, equipped with multi-channel detectors and monochromated X-ray sources of varying excitation energies, offer both improved sensitivity and energy resolution enabling more reliable discrimination of chemical shifts as small as 0.2 eV. Sample charging during photoelectron emission from insulating materials can be problematic; however, focused, tunable charge neutralizers offer effective electrical compensation. Care must still be taken during spectral interpretation to ensure accurate energy referencing of photoemission peaks, with binding energies preferably aligned by reference to at least one of adventitious carbon, the Fermi edge or an internal Au standard.
Figure 2.3 S 2p XP spectra of thiol and sulfonic acid silica.
j41
j 2 Sol–Gel Sulfonic Acid Silicas as Catalysts
42
MAS-NMR can also provide information on the nature of surface organic groups; however, this is an indirect method since oxidation of SH to SO3H must be inferred from the 13 C chemical shifts within the organic tether. Two resonances are observed for grafted mercaptopropyl group at 11.4 and 27.9 ppm, corresponding to the SiCH2 moiety and the two CH2 environments nearest to the SH group. Following oxidation, new resonances peaks are observed at 11.4, 18.2 and 54 ppm, while the peak at 27.9 ppm disappears [19]. Phenyl-based sulfonic acids also exhibit characteristic peaks at 128.1, 135.5 and 148.8 ppm for the ring carbons [20], with the highest chemical shifts being observed for the carbon atoms closest to the SO3H group. The superior thermal stability of silica-based sulfonic acids over polymer resinbased analogues such as Amberlyst-15 and Nafion-H is a major advantage for their catalytic applications. Thermogravimetric analysis (TGA) is a convenient technique to study the thermal stability of organically modified materials and has revealed that silica sulfonic acids are stable to 270 C, at which point they undergo irreversible decomposition evolving SO2 [21]. TGA can also prove useful for probing the OH density of silica supports and to estimate their capacity for grafting, as described in the next section. 2.2.2.2 Grafting Methods The grafting process generally involves reaction of surface silanols (SiOH) with an alkoxysilane, as illustrated in Scheme 2.1. However, some studies also suggest that surface hydration levels may drive reaction with neighboring siloxane bridges (SiOSi). The final loading of organic groups and the efficiency of the tethering process are thus dependent on the nature of the silica surface and its pretreatment. The SiO2 surface is complex and comprises a number of SiOH species (isolated, geminal and vicinal) and siloxane bridges, SiOSi, the precise distribution of which depends on the calcination temperature of the material. These different hydroxyl groups can be readily distinguished by IR spectroscopy (Figure 2.4), with geminal and isolated hydroxyls giving rise to bands at 3750 cm1, while vicinal
Figure 2.4 DRIFT spectra of hydroxyl vibrational modes originating from geminal/isolated and vicinal hydroxyl groups.
2.2 Preparation of Meso–structured Silica Sulfonic Acid Catalysts
hydroxyls give rise to two components corresponding to the hydrogen-bonded OH and the terminal OH at 3560 and 3715 cm1, respectively. The DRIFT spectra in Figure 2.4 compare the hydroxyl region for HMS silica following template removal by ethanol extraction or calcination at 600 C to illustrate the dependence of hydroxyl distribution on silica support preparation. There is a significant contribution from vicinal hydroxyls following template extraction; however, calcination at 600 C results in a surface largely terminated in isolated groups. Vicinal OH groups are known to be unstable towards dehydroxylation and undergo irreversible condensation on heating to form siloxane bridges via 2 SiOH ! SiOSi þ H2O. Thus high-temperature calcination both lowers the overall surface OH density and influences the dominant OH species. This reduced hydroxyl density impacts on the concentration of grafted organic groups, with sulfonic acid loadings of 0.8 and 0.3 mmol g1 achievable on extracted or calcined HMS supports, respectively [22]. Typical OH densities for calcined or extracted MCM-41 silicas are around 2.5 and 3 OH nm2, respectively [23]. The lower surface OH concentrations present in calcined silicas restrict the maximum sulfonic acid loading in grafted MCM-41 to 0.7 mmol g1 [7]. One way to increase the grafting efficiencies over MCM-41 is controlled surface hydration [7], which can increase sulfonic acid concentrations from 0.7 to 1.7 mmol g1. As for HMS materials, grafting on to extracted MCM silicas [24] provides enhanced sulfonic acid loadings of 1.17 mmol g1. In this work, template extraction was achieved using dilute HCl–acetone or HCl–ethanol solutions while successfully retaining the meso–structure. However, while extracted and calcined MCM supports both have similar surface areas (700 m2 g1), the surface area of the extracted material decreases to 550 m2 g1 following the grafting and oxidation of the thiol, raising concerns about the stability of the silica framework if the calcination step is omitted. SBAs provide the highest sulfonic acid loadings of any directly calcined and grafted mesoporous silica at 1 mmol g1 [25]. 2.2.2.3 Direct Preparation Methods Direct preparation or one pot sol–gel routes offer the opportunity to incorporate significantly higher loadings of organic groups than are attainable via grafting techniques. Such materials have largely been prepared using neutral templating methods, with either long-chain alkylamines (HMS type) [21, 26] or PEO surfactants (SBA type) [19], wherein facile template removal by solvent extraction can be achieved. SBA materials with sulfonic acid loadings of 1–2 mmol g1, surface areas of 800 m2 g1 and pore diameters of 6 nm can be routinely prepared. Sulfonic acid loadings of 0.3–4 mmol g1 have been achieved with smaller pore size (0.5–2 nm) HMS templated materials [21]. Although these silicas display a high degree of order, mesopores can only be retained for organosiloxane : tetraethoxysilane (TEOS) molar ratios up to 1 : 4; higher ratios lead to materials containing smaller pores. There are, however, early reports of the successful use of acid extraction methods for the one-pot preparation of MCM sulfonic acids [27], with high surface areas of 574 m2 g1 and S contents of 4.8 mmol g1. More ordered pores are obtained for extracted MCM materials if a mixture of C12 and C16 surfactants is employed [28]. It has been reported
j43
j 2 Sol–Gel Sulfonic Acid Silicas as Catalysts
44
that by employing mixtures of a fluorocarbon surfactant (FC-4) and block copolymer (P123) under strongly acidic conditions, the mesostructure of PEO templated sulfonic acids can be retained even at MPTS : TEOS ratios as high as 1 : 2 [29]. Most sulfonated silicas prepared via post-oxidation of the thiol groups to sulfonic acid show some evidence of incomplete oxidation. An alternative preparation of sulfonic acid silicas involves in situ oxidation of thiol groups facilitated by addition of H2O2HCl during the sol–gel process [19]. Materials prepared by this route exhibit surface areas of 700–800 m2 g1 and larger pore diameters (7 nm) than those prepared via post-oxidation routes. There was also no evidence for unreacted thiol or partially oxidized disulfide groups as frequently observed in post-oxidized materials; hence in situ oxidation routes are deemed to produce a more uniform catalyst material with higher acid site densities. 2.2.3 Acid Strength of Sulfonic Acid Catalysts
The design of heterogeneous catalysts with tunable acidity is the ultimate goal for the catalytic chemist, as this would be hugely beneficial for controlling reaction selectivity in vapor- and liquid-phase catalysis. Various organic transformations require catalysts of different acid strength, for example acetal formation, and hydrolysis reactions generally require medium acidity sites, while electrophilic additions of alcohols or water to alkenes, skeletal rearrangements, esterification and alkylation reactions require strong acid sites. The effect of acid strength is most elegantly demonstrated by the selective isomerization of terpenes such as a-pinene, where the formation of polycyclic or monocyclic compounds require weak or strong acids, respectively, as illustrated in Scheme 2.2. This characteristic can in turn serve as a convenient means of probing the acid strength of solid acid catalysts and has proven useful for determining the strength of a number of solid acid catalysts [30], including sol–gel sulfonic acids [31].
Scheme 2.2 Acid-catalyzed isomerization of a-pinene to polycyclic products in the presence of weak acids and monocyclic products in the presence of strong acids.
2.2 Preparation of Meso–structured Silica Sulfonic Acid Catalysts
There have been a number of approaches to altering the strength of sulfonic acids, including changing the organic linker and positioning groups in specific locations to see if cooperative effects are important. However, the measurement of solid acidity is itself complex, with the most widely adopted methods involving adsorption of a weakly basic probe molecule, such as pyridine, NH3 or triethylphosphine oxide (TEPO) and their associated quantification by DRIFTS, TPD, 31 P MAS-NMR or adsorption calorimetry; the last of which is deemed to be the most reliable measure of both site concentration and acid strength. 2.2.3.1 Phenyl- Versus Propylsulfonic Acids The increased electrophilicity of a phenyl ring compared with an alkyl chain increases the acid strength of organic sulfonic acids, evidenced from the pKa of benzene- and methanesulfonic acids, which are 6.5 and 2.0, respectively [32]. The exchange of propyl for phenyl spacer groups has also been investigated, with the expectation that phenylsulfonic acid silica should exhibit an increased acidity. Phenylsulfonic acid SBA-15 [20] has been prepared via co-condensation of tetraethoxysilane (TEOS) and 2-(4-chlorosulfonylphenyl)ethyltrimethoxysilane (CSPTMS), with the effect of in situ oxidation by H2O2 also investigated according to Scheme 2.3; transformation of chlorosulfonyl moieties into sulfonic groups is mediated by acid-catalyzed hydrolysis alone and does not require H2O2 addition. Base titration reveals acid loadings of 1.3 mmol H þ g1 SiO2, which is in accord with the predicted S loading for quantitative conversion of CSPTMS to sulfonic acid. The corresponding acid strength was assessed by 31 P MAS-NMR of chemisorbed TEPO, whose chemical shift is sensitive to the strength of the acid site at which it adsorbs [33]. Both phenyl- and propylsulfonic acids have acid strengths greater than those of aluminosilicates, with the phenyl material exhibiting the strongest acidity, comparable to that of Amberlyst-15. It is interesting that while H2O2 addition does not affect the extent of oxidation of chlorosulfonyl groups, a higher chemical shift was observed in the 31 P MAS-NMR of peroxide-treated silica, which was interpreted as evidence for a higher acid strength. The increased acid strength of phenylsulfonic acid silica was reflected catalytically in the Fries rearrangement of phenyl acetate, wherein higher conversions were achieved relative to the propylsulfonic acid catalysts.
Scheme 2.3 Sol–gel synthesis of phenylsulfonic acid silica.
j45
j 2 Sol–Gel Sulfonic Acid Silicas as Catalysts
46
A comparison of phenyl- and propyl-SBA-15 in the etherification of benzyl alcohol [34] with 1-hexanol was also reported, which surprisingly indicated that propylsulfonic catalyst was most active. This observation was attributed to the relative ease of etherification reactions, for which propylsulfonic acid is sufficient strong to catalyze. This work also considered whether a stronger interaction of the phenylsulfonic acid with water formed during the esterification process could lower its reactivity. Such hydrophilic interactions are expected to favor aqueous microenvironments surrounding the sulfonic acid groups, hindering access of reactants to the acid sites, an effect likely to be more pronounced for the stronger phenylsulfonic acid. In contrast, during palmitic acid esterification, phenylsulfonic acid SBA-15 [35] exhibited higher activity than the corresponding propyl-based catalyst, although these two materials had different acid site loadings, complicating their direct comparison. Siril et al. [31] used a combination of NH3 calorimetry and a-pinene isomerization to compare the acid strengths of a range of sulfonated resins (Nafion, Amberlyst) and grafted propyl- and phenylsulfonic acid silicas. They found that Nafion catalysts show both the highest DHads(NH3) of –162 kJ mol–1 and corresponding catalytic activities. In contrast, both silica-supported phenyl- and propylsulfonic acids had comparable DHads(NH3) of 126and128 kJ mol1, respectively, withconsiderably lowerturnover frequencies in a-pinene isomerization (11–13 h1 vs 1709 h1). However, the effect of solvents on regulating acid strength cannot be neglected and titration microcalorimetry of SBA- and HMS-supported propylsulfonic acids [36, 37] revealed a much lower neutralization enthalpy with aqueous NaOH of only 52 to 55 kJ mol1. This contrasts with the neutralization enthalpy of 120 kJ mol1 obtained during titration against butylamine in cyclohexane, indicative of a much higher acidity. These solvent effects are similar to those observed during the ionization of strong mineral acids, wherein solvent basicity and dielectric constant can also modify the resultant pKa and could explain the differences in relative reactivity of phenyl- and propylsulfonic acids towards, e.g., non-polar Fries rearrangement versus polar etherification reactions. 2.2.4 Fine Tuning the Catalytic Activity of Sulfonic Acid Silicas
Following the explosion of publications concerned with the synthesis and application of sulfonic acid catalysts [38], two fundamental questions have arisen concerning their application in liquid-phase reactions. First, how do the sulfonic acid head groups interact and are there concentration effects (e.g. lateral interactions) in aprotic solvents as observed in liquid acids, i.e. does the packing density of sulfonic acid groups on the silica surface affect the overall acidity? Second, does the hydrophilic nature of the silica surface influence diffusion and adsorption of hydrophobic organic reactants? Recent investigations have set out to address these issues through the use of designer sulfonic acid precursors which offer control over the spatial location of acid centers and by incorporation of inert organic groups to alter the hydrophilicity of the surface. 2.2.4.1 Cooperative Effects Cooperative effects in acidity require acid sites in close proximity, hence methods to control the spatial distribution of sulfonic acid groups have been investigated to boost
2.2 Preparation of Meso–structured Silica Sulfonic Acid Catalysts
acid strength. A series of organosiloxane precursors were synthesized containing either disulfide or sulfonate ester functionalities [39] to control the location of propyland phenylsulfonic acid groups as illustrated in Scheme 2.4. Propylsulfonic acid groups were generated by cleaving the disulfide bonds with tris(2-carboxyethyl) phosphine (TCEP.HCl) to form the thiol followed by oxidation with H2O2. Phenylsulfonic acid groups were generated from the sulfonate ester by H2SO4-mediated hydrolysis. Catalysts were evaluated in the condensation of phenol and acetone to produce bisphenol A. Arylsulfonic acids consistently displayed higher catalytic activity than their propyl analogues; however, catalysts having spatially organized groups had the highest overall activities and regioselectivities, favoring p,p-bisphenol A formation over the unwanted o,p-bisphenol, as will be discussed in Section 2.3.1.
Scheme 2.4 Routes to prepare propyl- and phenylsulfonic acids with spatially located acid sites.
The per site yield was practically doubled by having spatially located alkylsulfonic acid groups. However, it is unclear whether this enhancement may reflect residual S–H or S–S present in the catalysts and it has been suggested that mixed sulfonic acid–thiol systems can exhibit enhanced activity/selectivity. Indeed, a subsequent study demonstrated that catalysts co-functionalized with thiol and sulfonic acid groups exhibit increased reaction rates and selectivity towards p,p-bisphenol isomer [40]. Shanks [41] also investigated bis(trimethoxysilyl)propyl disulfide as a means to control the location of sulfonic acid groups on the surface of SBA-15, with the acid strength of the resulting solid acid catalysts followed by potentiometric titration of NaOH in MeOH. The pKa of conventionally prepared propyl-, phenyl- and spatially organized propylsulfonic acid silicas were measured in water, with all samples exhibiting increasing acid strength with sulfonic acid loading. Materials prepared with spatially located propylsulfonic acid groups had a pKa of 1.26, which was comparable to the phenylsulfonic acid variant, which in turn was stronger than conventionally prepared propylsulfonic acid materials which had a pKa of 1.4. However, these observations were not reflected in the relative catalytic activities towards esterification, with catalysts with spatially organized and conventionally prepared propylsulfonic acid groups exhibiting comparable activity, with both being
j47
j 2 Sol–Gel Sulfonic Acid Silicas as Catalysts
48
less active than the arylsulfonic acid variant. The absence of beneficial cooperative effects and an associated increase in acid strength may reflect weakened hydrogen bonding interactions between sulfonic acid groups at high reaction temperatures. 2.2.4.2 Effect of Spectator Groups The hydrophilic nature of polar silica surfaces means that it is not ideal for reactions of apolar organic molecules. This can be problematic due to differential diffusion or preferential adsorption of more polar components from mixtures of hydrocarbons. Furthermore, surface hydroxyl groups favor H2O adsorption, which, if formed during reaction such as in esterification or oxidation, can inhibit adsorption of organic substrates at active sites. Surface modification via the incorporation of spectator organic moieties into the oxide surface or dehydroxylation can overcome this polarity issue to increase initial reaction rates in acid-catalyzed transformations of liquid-phase organic molecules [8, 42]. The incorporation of organic spectator groups (e.g. phenyl, methyl or propyl) can be readily achieved during the sol–gel syntheses of SBA-15 [19] and MCM-41 [8] sulfonic acid silicas by simple addition of the respective alkyl or aryl trimethoxysilane during the preparation. Such approaches to facilitating adsorption of hydrophobic can significantly enhance esterification rates. Dıaz [8, 43] investigated methyl and phenyl modified MCM-41 sulfonic acid in the esterification of glycerol with both oleic and lauric acid. The spectator groups were incorporated in a one-pot approach in which the spectator:sulfonic acid loading was varied. TGA revealed that the amount of adsorbed H2O fell with increasing organic content. Catalyst activity in glycerol esterification with oleic acid was highest in methyl-derivatized materials, whereas for reaction with lauric acid both propyl- and methyl-functionalized sulfonic acid silicas exhibited similar activities. These differences were attributed to the different steric bulks of the C12 and C18 fatty acids; however, it was not reported whether the unsaturated nature of oleic acid may also affect its reactivity. SBA-15 propylsulfonic acids modified with methyl, ethyl and phenyl groups were investigated by Mbaraka and Shanks for the esterification of fatty acids [25]. Silicas were derivatized by both one-pot approaches and by grafting spectator groups on to preformed SBA-15 sulfonic acid materials. Whereas the overall fatty acid conversions were comparable, the initial activities depended on the catalyst preparation. Phenylderivatized catalysts exhibited the highest initial rates of the one-pot silicas, comparable to the performance of all the post-modified materials, whose activities were independent of the choice of spectator group. However, the variable SBA-15 textural properties resulting from the differing preparative routes make it difficult to draw firm conclusions about the effect of spectator groups. Surface polarity can also be tuned by incorporating alkyl groups directly into the silica framework as in bridged polysilsesquioxanes, which can be prepared via the cocondensation of 1,4-bis(triethoxysilyl)benzene (BTEB) or 1,2-bis(trimethoxysilyl) ethane (BTME), with TEOS and MPTS in the sol–gel process [44–46]. The increased hydrophobicity derived from framework organic substituents has been shown to increase catalyst activity in acetic acid esterification with ethanol, where comparable activity to Nafion-H is reported [47].
2.3 Application in Organic Transformations
2.3 Application in Organic Transformations
Since the first reported synthesis of mesoporous sulfonic acid silica, its application in a diverse range of acid-catalyzed reactions has been investigated. This section summarizes the key classes of reactions investigated to date. 2.3.1 Condensation and Esterification
Most of the early studies concentrated on condensation and esterification reactions, for example, the reaction of 2-methylfuran with acetone to form 2,2-bis(5-methylfuryl) propane [6] (Scheme 2.5). MCM-SO3H provided the highest conversion and selectivity, which were significantly higher than those for zeolites H-b and H-USY.
Scheme 2.5 2,2-Bis(5-methylfuryl)propane synthesis from 2-methylfuran and acetone.
Other examples of condensation reactions catalyzed by sulfonic acids include Claisen–Schmidt condensation of acetophenone with benzaldehyde (Scheme 2.6) [21]. This reaction is used in the preparation of chalcones, which are valuable pharmaceutical intermediates possessing antibacterial, antifungal, antitumor and antiinflammatory properties, conventionally prepared via environmentally unfriendly routes using HCl, AlCl3 or BF3. Reaction can be performed under mild conditions using silica sulfonic acids, with high selectivity to the chalcone and enhanced conversion (34%) against zeolite HY [48] which offers only 12% conversion at a much higher catalyst:substrate ratio of 10 wt%.
Scheme 2.6 Claisen–Schmidt condensation.
The selective, low-temperature esterification of glycerol with lauric acid to form monoglycerides [7] has also been reported using both grafted and sol–gel sulfonic acid silicas, in which superior activity and monoglyceride yields were observed compared with H-USY. Higher concentrations of sulfonic acid groups did not necessarily result in the highest conversions; indeed, intermediate loadings provided optimum activity, suggesting that accessibility of the active sites is the factor when working with such bulky substrates. Fatty acid esterification with methanol for biodiesel feedstock pretreatment [25] and butanol [21] and cyclohexanol [29] with acetic acid are also reported. Dehydration [49] of D-xylose to furfural, a key derivative in the production of nonpetroleum-based chemical feedstocks, is readily accessible from renewable resources
j49
j 2 Sol–Gel Sulfonic Acid Silicas as Catalysts
50
via the initial acid-catalyzed hydrolysis of pentosan to pentose (i.e. xylose) (Scheme 2.7). MCM-41 with grafted sulfonic acid exhibited higher activity than one-pot sol–gel materials and commercial Amberlyst.
Scheme 2.7 Dehydration of xylose to furfural.
Bisphenol A is another product of significance to the pharmaceutical industry, which is synthesized by the acid-catalyzed condensation of phenol and acetone. This reaction has been studied using MCM propylsulfonic acid [50, 51] and SBA-15-based materials with spatially organized aryl- and propylsulfonic acid groups [39] (Scheme 2.8). Arylsulfonic acid silica materials were one-third less active than the homogeneous, p-toluenesulfonic acid catalyst, but gave approximately the same selectivity to the two structural isomers p,p- to o,p-bisphenol A. The alkylsulfonic acid catalysts were less active than the aryl derivatives, but surprisingly more selective towards p,p- bisphenol A formation.
Scheme 2.8 Bisphenol A synthesis from phenol and acetone.
Etherification of benzyl alcohols [34] such as vanillyl alcohol with hexanol is another reaction of great interest to the fine and fragrance intermediates sectors, which has been explored using SBA-15-based phenyl- and propylsulfonic acids (Scheme 2.9). Both catalysts exhibited superior activity to commercially available sulfonated, resinbased catalysts (Amberlyst-15 and Nafion–silica composite). In contrast to other reactions, the propyl variant exhibited higher activity than the phenyl, which was attributed to strongly bound H2O poisoning acid sites in the latter variant.
Scheme 2.9 Synthesis of 4-hydroxy-3-methoxybenzyl 1-hexyl ether from the etherification of vanillyl alcohol with hexanol.
Vapor-phase, continuous etherification of methanol with isobutanol has also been investigated over SBA-15 sulfonic acids [52], where the main products were methyl
2.3 Application in Organic Transformations
isobutyl ether, dimethyl ether and dehydration products. In this instance, competitive adsorption of the reactants plays a major role in controlling reaction selectivity. At lower reaction pressures, preferential adsorption of isobutanol occurs, favoring isobutene as the initial dominant product. Higher pressures favor methanol adsorption and consequent formation of ether production at the expense of isobutene. Butanol etherification has also been studied over phenyl- and propylsulfonic acid SBA-15 and organic framework-substituted silicas. In this instance, the rate of etherification was superior over the framework-substituted catalysts [53], an observation attributed to the ability of their more hydrophobic surfaces to repel H2O from the acid sites. Similar effects are observed for acetic acid esterification over benzene silica-based propylsulfonic acids [54]. 2.3.2 Electrophilic Aromatic Substitution
More recently, the application of sulfonic acids in more challenging organic transformations, such as Friedel–Crafts alkylation, has been reported. Phenol alkylation with 2-methylpropan-2-ol has been studied over grafted propylsulfonic acid MCM-41 catalysts in the presence of phenyl spectator groups [46]. o-tert-Butylphenol was observed as the primary product; however, positional isomerization to form p-tertbutylphenol was observed after prolonged reaction (Scheme 2.10). The isomerization rate depended on the catalyst composition, with phenyl-modified silicas exhibiting higher ortho selectivities. Alkylation of benzene or toluene with benzyl alcohol [55] was also investigated over SBA-15-, MCM- and HMS-based phenyl- and propylsulfonic acid catalysts. In all cases diphenylmethane and a mixture of phenyl-o- and -p-tolylmethane were observed as reaction products. In general, catalysts containing phenyl sulfonic acid groups made by co-condensation routes exhibited the highest activity and selectivity, with the SBA-based materials giving the highest TOF. Phenol alkylation with cyclohexene has been reported using HMS-based sulfonic acid catalysts, where moderate yields of both cyclohexylphenyl ether and 2- and 4-substituted cyclohexylphenol were observed [56].
Scheme 2.10 Alkylation of phenol with cyclohexene.
Acylation [57] of anisole with acetic anhydride (Scheme 2.11) was studied using phenyl- and propylsulfonic acid-modified SBA-15. All sulfonic acid catalysts gave greater anisole conversions than commercial b-zeolites, with the TON decreasing in the order phenyl > Amberlyst > propyl.
j51
j 2 Sol–Gel Sulfonic Acid Silicas as Catalysts
52
Scheme 2.11 Acylation of anisole with acetic anhydride.
The Fries rearrangement of phenyl acetate to o- and p-hydroxyacetophenones is an important reaction in the production of pharmaceuticals, dyes and agrochemicals and is also the first step of the Hoechst Celanese process for paracetamol synthesis [58]. This is a particularly challenging reaction as it involves an intramolecular rearrangement to form the ortho-substituent, whereas an intermolecular reaction with phenol forms the para-substituent. Cleavage of the acyl group and formation of phenol and ketene intermediates can also occur in competition, lowering product selectivity due to ketene oligomerization. Fries rearrangement has been compared over phenyl- and propylsulfonic acids and commercial Amberlyst [27, 59] (Scheme 2.12). Although conversions using the Amberlyst and phenylsulfonic acids were comparable, the sulfonic acid exhibited higher selectivity towards ortho substitution, consistent with the observed slow isomerization of alkylated phenols [27]. Interestingly, the catalyst lifetime could be extended if dichloromethane was used as a reaction solvent [46].
Scheme 2.12 Fries rearrangement of phenyl acetate.
2.3.3 Miscellaneous Reactions
The synthesis of more complex pharmaceuticals such as Biginelli-type compounds [60], which involve the sulfonic acid-mediated reaction of ethyl acetoacetate with aromatic aldehydes and urea (or thiourea) to produce 3,4-dihydropyrimidinones/
2.4 Conclusions and Future Prospects
thiones, have also been reported (Scheme 2.13). These products are of interest for their antiviral, antibacterial and antihypertensive activity, and also efficacy as calcium channel modulators and 1a-antagonists [61]. Silica-grafted sulfonic acid was found to give >90% isolated yields and could be recycled eight times without a significant loss of yield.
Scheme 2.13 The Biginelli reaction of ethyl acetoacetate with aldehyde and urea (X ¼ S or O) to form 3,4-dihydropyrimidones.
Dehydration reactions such as the pinacolol rearrangement of meso-hydrobenzoin have been studied over MCM-41 sulfonic acid, with high selectivity to diphenylacetaldehyde (DPAA) being observed [46]. The authors suggest that DPPA is favored over 1,2-diphenylethanone (DPE) since the hydride shift is less sterically demanding and pore size constrains the possible transition state conformation (Scheme 2.14).
Scheme 2.14 Pinacolol rearrangement of meso-hydrobenzoin.
2.4 Conclusions and Future Prospects
Mesporous sulfonic acid catalysts offer clean alternatives to H2SO4 in numerous liquid-phase organic transformations. Methods to tailor the electronic properties of the sulfonic acid group through the use of phenyl rather than propyl tethering groups have proven beneficial for increasing the acid strength. Likewise, fine tuning of catalyst surface polarity and hydrophobicity can be achieved through the use of spectator groups or more advanced synthesis of framework-substituted organic silicas. Such modifications can significantly increase catalyst activity by aiding adsorption of non-polar reactants at the acid site or help with the removal of water from the catalyst surface in dehydration reactions. In spite of progress in our understanding and advances in materials synthesis, there are still some fundamental questions about the nature of the catalyst surface which remain unanswered. In particular, a better understanding of how the acid strength of phenyl- and propylsulfonic acid groups varies with solvent and temperature is required. While access to calorimetric techniques to measure acid strength has
j53
j 2 Sol–Gel Sulfonic Acid Silicas as Catalysts
54
increased the accuracy with which acid site distributions are determined, it is clear that simple NH3 adsorption measurements may not provide a true reflection of the acid strength under reaction conditions. Furthermore, it would be desirable for the effect of sulfonic acid loading, acid strength and spectator groups to be compared on materials where the morphology of the carrier remains constant across the series of samples, as in the case for post-grafted materials. There have been many advances in this area, but with porosity and surface area often changing with the loading of functional groups, a unified model for the surface chemistry of sulfonic acid silicas cannot yet be derived from the current literature. If these designer materials are to obtain more widespread application in industrial chemistry, it is critical that more detailed kinetic data are measured on well-defined catalytic systems. In contrast to gas-phase reactions, where partial pressures of reactants can readily be explored, there are no studies of liquid-phase reactions over silica-based sulfonic acid catalysts where the effect of competitive adsorption of reactants at the surface are examined. This is particularly important in the liquid phase since solvent effects may play a key role in aiding diffusion of reactants. The use of molecular modeling should also help with our understanding of reactant activation, with transition states recently identified in vapor-phase etherification processes over sulfonic acid silicas [52]. Molecular simulation studies of adsorption processes at the surface of sulfonic acid catalysts could also provide similar information for liquid-phase processes and assist our understanding of the role of different spectator groups, accessibility of the acid site or cooperative effects. Overall, sulfonic acid silicas are very promising replacements for H2SO4; however, their full potential can only be realized if they are incorporated into continuous reactors, which offer inherently more efficient processing than with conventional batch reactors. Future studies should address issues at the chemistry–chemical engineering interface such as particle forming, stability and optimization of pore networks for efficient reactant/product diffusion and mixing, to aid assimilation of these catalyst systems into modern reactor technology.
Abbreviations
BTEB BTME CSPTMS CTAB DDA DRIFTS EtOH FFA HMS MAS-NMR MCM MPTS
bis(triethoxysilyl)benzene 1,2-bis(trimethoxysilyl)ethane 2-(4-chlorosulfonylphenyl)ethyltrimethoxysilane cetyltrimethylammonium bromide dodecylamine diffuse reflectance Fourier transform infrared spectroscopy ethanol free fatty acid hexagonal mesoporous silica magic angle spinning nuclear magnetic resonance Mobil corporate material mercaptopropyltrimethoxy silane
References
PEO TCEP TEOS TEPO TGA XPS
poly(ethylene oxide) tris(2-carboxyethyl)phosphine tetraethyl orthosilicate [Si(OEt)4] triethylphosphine oxide thermogravimetric analysis X-ray photoelectron spectroscopy
Acknowledgments
Financial support from the EPSRC under grants GR/R39436/01 and EP/E013090/1 and from BP Chemicals is gratefully acknowledged.
References 1 Corma, A. (1997) Chemical Reviews, 97, 2373. 2 Beck, J.S., Vartuli, J.C., Roth, W.J., Leonowicz, M.E., Kresge, C.T., Schmitt, K.D., Chu, C.T.-W., Olsen, D.H., Sheppard, E.W., McCullen, S.B., Higgins, J.B. and Schlenker, J.L. (1992) Journal of the American Chemical Society, 114, 10834. 3 Blitz, J.P. and Little, C.P. (1999) Fundamentals and Applied Aspects of Chemically Modified Surfaces, Royal Society of Chemistry, Cambridge. 4 Ciolino, L.A. and Dorsey, J.G. (1994) Journal of Chromatography. A, 675, 29. 5 Mercier, L. and Pinnavia, T.J. (1988) Environmental Science & Technology, 32, 2749. 6 VanRhijn, W.M., DeVos, D.E., Sels, B.F., Bossaert, W.D. and Jacobs, P.A. (1998) Chemical Communications, 317. 7 Bossaert, W.D., DeVos, D.E., VanRhijn, W.M., Bullen, J., Grobet, P.J. and Jacobs, P.A. (1999) Journal of Catalysis, 182, 156. 8 Dıaz, I., Marquez-Alvarez, C., Mohino, F., Prez-Pariente, K. and Sastre, E. (2000) Journal of Catalysis, 193, 283. 9 Harmer, M.A., Farmeth, W.E. and Sun, Q. (1996) Journal of the American Chemical Society, 118, 7708. 10 Sun, Q., Harmer, M.A. and Farneth, W.E. (1996) Chemical Communications, 1201.
11 Harmer, M.A., Farmeth, W.E. and Sun, Q. (1998) Advanced Materials, 10, 1255. 12 Ying, J.Y., Mehnert, C.P. and Wong, M.S. (1999) Angewandte Chemie-International Edition, 38, 56. 13 Linssen, T., Cassiers, K., Cool, P. and Vansant, E.F. (2003) Advances in Colloid and Interface Science, 103, 121. 14 Davidson, A. (2002) Current Opinion in Colloid and Interface Science, 92. 15 Galarneau, A., Iapichella, J., Bonhomme, K., Di Renzo, F., Kooyman, P., Terasaki, O. and Fajula, F. (2006) Advanced Functional Materials, 16, 1657. 16 Gr€ un, M., Unger, K., Matsumoto, A. and Tsutsumi, K. (1999) Microporous Mesoporous Mater, 27, 207. 17 Tanev, P.T. and Pinnavaia, T.J. (1995) Science, 267, 865. 18 Attard, G.S., Glyde, J.C. and Goltner, C.G. (1995) Nature, 378, 366. 19 Margolese, D., Melero, J.A., Christiansen, S.C., Chmelka, B.F. and Stucky, G.D. (2000) Chemistry of Materials, 12, 2448. 20 Melero, J.A., Stucky, G.D., van Griekena, R. and Morales, G. (2002) Journal of Materials Chemistry, 12, 1664. 21 Wilson, K., Lee, A.F., Macquarrie, D.J. and Clark, J.H. (2002) Applied Catalysis A-General, 228, 127. 22 Wilson, K. unpublished results.
j55
j 2 Sol–Gel Sulfonic Acid Silicas as Catalysts
56
23 Zhao, X.S., Lu, G.Q., Whittaker, A.K., Millar, G.J. and Zhu, H.Y. (1997) The Journal of Physical Chemistry. B, 101, 6525. 24 Boveri, M., Aguilar-Pliego, J., PerezPariente, J. and Sastre, E. (2005) Catal Today, 107–108, 868. 25 Mbaraka, I.K. and Shanks, B.H. (2005) Journal of Catalysis, 229, 365. 26 Dıaz, I., Marquez-Alvarez, C., Mohino, F., Perez-Pariente, J. and Sastre, E. (2001) Microporous Mesoporous Mater, 44–45, 295. 27 Lim, M.H., Blanford, C.F. and Stein, A. (1998) Chemistry of Materials, 10, 467. 28 Dıaz, I., Marquez-Alvarez, C., Mohino, F., Perez-Pariente, J. and Sastre, E. (2001) Applied Catalysis A-General, 205, 19. 29 Feng, Y.-F., Yang, X.-Y., Di, Y., Du, Y.-C., Zhang, Y.-L. and Xiao, F.-S. (2006) The Journal of Physical Chemistry B, 110, 14142. 30 Ecormier, M.A., Wilson, K. and Lee, A.F. (2003) Journal of Catalysis, 215, 57. 31 Siril, P.F., Davison, A.D., Randhawa, J.K. and Brown, D.R. (2007) Journal of Molecular Catalysis A-Chemical, 267, 72. 32 Serjeant, E.P. and Dempsey, B. (eds) (1979) Ionization Constants of Organic Acids in Solution, IUPAC Chemical Data Series No. 23, Pergamon Press, Oxford. 33 Osegovic, J.P. and Drago, R.S. (2000) The Journal of Physical Chemistry B, 104, 147. 34 van Grieken, R., Melero, J.A. and Morales, G. (2006) Journal of Molecular Catalysis A-Chemical, 256, 29. 35 Mbaraka, I.K., Radu, D.R., Lin, V.S.-Y. and Shanks, B.H. (2003) Journal of Catalysis, 219, 329. 36 Koujout, S. and Brown, D.R. (2005) Thermochimica Acta, 434, 158. 37 Koujout, S. and Brown, D.R. (2004) Catalysis Letters, 98, 195. 38 Melero, J.A., van Grieken, R. and Morales, G. (2006) Chemical Reviews, 106, 3790. 39 Dufaud, V. and Davis, M.E. (2003) Journal of the American Chemical Society, 125, 9403.
40 Zeidan, R.K., Dufaud, V. and Davis, M.E. (2006) Journal of Catalysis, 239, 299. 41 Mbaraka, I.K. and Shanks, B.H. (2006) Journal of Catalysis, 244, 78. 42 Wilson, K., Renson, A. and Clark, J.H. (1999) Catalysis Letters, 61, 51. 43 Dìaz, I., Marquez-Alvarez, C., Mohino, F., Prez-Pariente, K. and Sastre, E. (2000) Journal of Catalysis, 193, 295. 44 Yang, J., Yang, Q., Wang, G., Feng, Z. and Liu, J. (2006) Journal of Molecular Catalysis A-Chemical, 256, 122. 45 Yang, Q., Liu, J., Yang, J., Kapoor, M.P., Inagaki, S. and Li, C. (2004) Journal of Catalysis, 228, 265. 46 Rac, B., Hegyes, P., Forgo, P. and Molnar, A. (2006) Applied Catalysis A-General, 299, 193. 47 Yang, Q.H., Kapoor, M.P., Shirokura, N., Onashi, M., Inagaki, S., Kondo, J.N. and Domen, K. (2005) Journal of Materials Chemistry, 15, 666. 48 Climent, M.J., Corma, A., Iborra, S. and Primo, J. (1995) Journal of Catalysis, 151, 60. 49 Dias, A.S., Pillinger, M. and Valente, A.A. (2005) Journal of Catalysis, 229, 414. 50 Das, D., Lee, J.-F. and Cheng, S. (2001) Chemical Communications, 2178. 51 Das, D., Lee, J.-F. and Cheng, S. (2004) Journal of Catalysis, 223, 152. 52 Herman, R.G., Khouri, F.H., Klier, K., Higgins, J.B., Galler, M.R. and Terenna, C.R. (2004) Journal of Catalysis, 228, 347. 53 Sow, B., Hamoudi, S., Zahedi-Niaki, M.H. and Kaliaguine, S. (2005) Microporous and Mesoporous Mater, 79, 129. 54 Yang, Q.H., Kapoor, M.P., Inagaki, S., Shirokura, N., Konodo, J.N. and Domen, K. (2005) Journal of Molecular Catalysis A-Chemical, 230, 85. 55 Rac, B., Molnar, A., Forgo, P., Mohai, M. and Bertóti, I. (2006) Journal of Molecular Catalysis A-Chemical, 244, 46. 56 Yadav, G.D. and Kumar, P. (2005) Applied Catalysis A-General, 286, 61. 57 Melero, J.A., van Grieken, R., Morales, G. and Nuno, V. (2004) Catalysis Communications, 5, 131.
References 58 Davenport, K.G. (1988) EP 251 552, to Celanese Corp. 59 van Grieken, R., Melero, J.A. and Morales, G. (2005) Applied Catalysis A-General, 289, 143.
60 Gupta, R., Paul, S. and Gupta, R. (2007) Journal of Molecular Catalysis A-Chemical, 266, 50. 61 Oliver Kappe, C. (2000) European Journal of Medicinal Chemistry, 35, 1043.
j57
j59
3 Applications of Environmentally Friendly TiO2 Photocatalysts in Green Chemistry: Environmental Purification and Clean Energy Production Under Solar Light Irradiation Masaya Matsuoka and Masakazu Anpo
3.1 Introduction
In recent years, there has been great concern over many serious environmental problems and the lack of natural energy resources, all of which we are facing on a global scale. An increase in world population and industrial development have led to accelerated energy consumption and the unabated release of toxic agents into the air and waterways, leading to such adverse effects as pollution-related diseases and climatic changes such as global warming. It is therefore of the utmost urgency to develop ecologically clean and safe chemical technologies, materials and processes to sustain our present level of population and economic expansion. The development of new catalytic processes which can contribute to environmental protection and new energy production are, therefore, strongly desired. One of the most ideal catalytic processes is the so-called artificial photosynthesis which has the potential to realize safe and clean chemical processes and systems with the use of limitless solar energy. As shown in Figure 3.1a, photosynthesis in plants allows one of the most significant uphill reactions, where the photon energy is converted into chemical energy and stored in the bonds of glucose, accompanied by a large positive change in the Gibbs free energy (DG > 0): hn
H2 O þ CO2 !
1 ðC H O Þ þ O2 6 6 12 6
DG ¼ 502 kJ mol1
ð3:1Þ
Since the first energy crisis in the early 1970s, much research has been devoted to the development of efficient systems that would permit the absorption and conversion of solar light into useful chemical energy resources. One of the most promising of such artificial photosynthetic reactions is the photocatalytic splitting of water to produce H2 and O2 under solar light accompanied by a large positive change in the Gibbs free energy: hn
H2 O ! H2 þ
1 O2 2
DG ¼ 237 kJ mol1
Handbook of Green Chemistry, Volume 2: Heterogeneous Catalysis. Edited by Robert H. Crabtree Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32497-2
ð3:2Þ
j 3 Applications of Environmentally Friendly TiO
60
2
Photocatalysts in Green Chemistry
Figure 3.1 (a) Mechanism of the photosynthetic reaction of green plants; (b) artificial photosynthetic reaction on a visible lightresponsive titanium oxide photocatalyst incorporated within a zeolite (Ti/zeolite).
The photosensitization effect of a TiO2 electrode on the electrolysis of water into H2 and O2 was discovered in 1972 by Fujishima and Honda [1, 2]. The refinement of this uphill reaction is significant not only for the storage of solar energy but also for the clean production of hydrogen, since its consumption is expected to increase
3.2 Principles of Photocatalysis
dramatically in future generations, especially in fuel cell applications [3–6]. In addition to the water decomposition reaction, it has been reported that an artificial photosynthetic reaction system, which produces hydrocarbons and oxygen from carbon dioxide and water under solar light irradiation, can be constructed by utilizing titanium oxide photocatalysts highly dispersed within the framework structure of zeolites, as shown in Figure 3.1b [7–15]. In addition to the uphill reaction, TiO2 photocatalysts also permit a downhill reaction. In a downhill reaction, the photocatalyst induces thermodynamically favored reactions under light irradiation such as the complete oxidation of organic compounds into CO2 and H2O, accompanied by a large negative change in the Gibbs free energy (DG < 0). Downhill reactions induced by solid semiconducting photocatalysts such as TiO2 have been applied in practice for the degradation of toxic organic compounds in air and water [2, 16]. Recently, another notable achievement in inducing artificial photosynthesis has been the development of visible light-responsive TiO2 photocatalysts which can operate not only under UV but also visible light irradiation [3–10, 17]. This chapter introduces the fundamental principles of photocatalysis, the development of such visible light-responsive TiO2 photocatalysts and their applications in environmentally friendly green chemistry and new energy production.
3.2 Principles of Photocatalysis
Semiconducting metal oxides such as TiO2, ZnO and Fe2O3 are known to act as sensitizers for light-induced redox processes due to their unique electronic structure characterized by a filled valence band and an empty conduction band [1–6]. As shown in Figure 3.2, when semiconducting metal oxide absorbs a photon having an energy larger than its bandgap, an electron is promoted from the valence band to the conduction band, leaving a hole. The electrons and holes formed are dissipated within a few nanoseconds by their recombination in the absence of suitable electron and hole scavengers [1–6]. However, if a suitable scavenger or surface defect state is available, recombination is prevented and subsequent redox reactions may occur. The holes in the valence band act as powerful oxidants, while the electrons in the conduction band are good reductants [1–6]. When the TiO2 is irradiated by UV light (l < 380 nm) in water, H þ is q reduced to H2 by the photo-formed electrons, while OH is oxidized to OH radicals by the photo-formed holes to produce O2 through several reaction steps. In this way, TiO2 can decompose water into H2 and O2, allowing the efficient conversion of light energy into chemical energy accompanied by a large positive change in the Gibbs free energy (DG ¼ 237 kJ mol1). It should be noted that the irradiation of vacuum UV light (l < 165 nm) is necessary for the direct photolysis of water molecules into H2 and O2 [1–6]. On the other hand, when TiO2 is irradiated by UV light in the presence of air and reactant molecules such as organic compounds in water, the photo-formed q electrons react with O2 to form O2, while OH is oxidized into OH radicals. The oxygen radicals formed can easily react with the organic compounds, decomposing
j61
j 3 Applications of Environmentally Friendly TiO
62
2
Photocatalysts in Green Chemistry
Figure 3.2 Various reaction systems using a TiO2 photocatalyst under different reaction conditions: (a) production of electrons and holes in TiO2 by absorption of UV light in the presence of oxygen and water (reaction in atmosphere); (b) production of electrons and holes by the absorption of UV light in the presence of water alone without any oxygen.
them into CO2 and H2O accompanied by a large negative change in the Gibbs free energy. In fact, the application of illuminated semiconductors in the remediation of contaminants has been successful for a wide variety of compounds [2–6]. In many cases, complete mineralization of the organic compounds could proceed on the TiO2 photocatalysts. Among the various photocatalysts, TiO2 is the most attractive due to its low cost, availability, high photocatalytic reactivity and chemical stability. However, TiO2 has a large bandgap with an absorption edge in UV regions shorter than 380 nm so that TiO2 semiconductors are able to absorb only 3–4% of the solar light that reaches the Earth. Extensive investigations have been carried out on the synthesis of lightresponsive TiO2 photocatalysts that can extend their absorption into the visible region and the development of such visible light-responsive photocatalysts will be introduced in this chapter.
3.3 Application of Photocatalysts in Green Chemistry: Solar Energy Conversion and Environmental Protection 3.3.1 Water Splitting to Produce Pure Hydrogen as Clean Fuel
Water splitting reactions have been intensively investigated in recent years using various powdered photocatalysts for the production of H2. It has been reported, for
3.3 Application of Photocatalysts in Green Chemistry: Solar Energy Conversion
example, that Pt-loaded TiO2 acts as a photocatalyst for water splitting reactions under UV irradiation. However, a product separation process is also required for the utilization of H2 as fuel since only a gas mixture of H2 and O2 evolves from powdered TiO2 photocatalysts. The reaction yields are also low since the recombination reaction between H2 and O2 to produce H2O occurs on the surface Pt particles which are the additives that promote the water splitting reaction [1, 2]. The development of photocatalytic systems allowing the separate evolution of H2 and O2 from water is, therefore, strongly desired. Recently, an H-type reactor has been constructed and applied to the separate evolution of H2 and O2 from water, where the liquid phases are separated by a TiO2 thin-film device and proton-exchange membrane [3–6]. In order to utilize the abundance of solar light reaching the Earth, visible light-responsive TiO2 thin films were prepared by a one-step magnetron sputtering deposition method in which the substrate temperature was controlled during the sputtering process [8–10]. As shown in Figure 3.3, a visible light-responsive TiO2 thin-film device (Vis-TiO2/Ti/Pt) consisting of a Ti metal foil substrate deposited with the visible light-responsive TiO2 thin film on one side and Pt metal on the other side was constructed by magnetron sputtering deposition. The prepared Vis-TiO2/Ti/Pt device was applied in the separate evolution of H2 and O2 from water. Figure 3.4 shows the time profile of this separate evolution reaction under solar light irradiation from a sunlight gathering system.
Figure 3.3 H-type reactor for the separate evolution of H2 and O2 using a visible light-responsive TiO2 thin-film device (Vis-TiO2/Ti/Pt) under solar light irradiation. Electrolyte: 1.0 M NaOH (right side); 0.5 M H2SO4 (left side).
j63
j 3 Applications of Environmentally Friendly TiO
64
2
Photocatalysts in Green Chemistry
Figure 3.4 Separate evolution of H2 and O2 on a visible light-responsive TiO2 thin-film device (Vis-TiO2/Ti/Pt) irradiated by light beams from a sunlight-gathering system using an H-type reactor.
Sunlight irradiation of Vis-TiO2/Ti/Pt within the H-type reactor successfully led to the stoichiometric evolution of H2 from the Pt side and O2 from the TiO2 side [3–6]. In fact, the solar energy conversion efficiency was found to reach 0.3% for this reaction. Thus, recent advances in the method of preparation of unique photocatalysts have allowed efficient solar energy conversion into useful chemical energy (hydrogen energy), which can be applied in practice for new energy production processes. Photocatalytic H2 evolution reactions from water containing biomass have also attracted much attention. A TiO2 thin-film device immersed in water containing biomass was observed to efficiently produce H2 under visible light irradiation. In this reaction, biomass, i.e. the organic compounds, efficiently scavenge the photo-formed holes while, simultaneously, the photo-formed electrons remain on the visible lightresponsive TiO2 thin film. As a result, the presence of the biomass in water leads to an enhancement of the reduction efficiency of H þ into H2 by photo-formed electrons [3–6]. This reaction system produces CO2 as a result of the oxidation of the biomass. However, this system can be applied for clean H2 production by combining it with an efficient CO2 utilization system such as in a unique agricultural system which can supply concentrated CO2 to green plants to promote their photosynthesis. 3.3.2 Photocatalytic Reduction of CO2 with H2O (Artificial Photosynthesis)
The development of efficient photocatalytic systems which are able to reduce CO2 with H2O into chemically valuable compounds such as CH3OH or CH4 under solar light irradiation is a challenging goal in research on environmentally friendly catalysts. The gas-phase reaction of CO2 and H2O on powdered TiO2 photocatalysts has been found to yield CH4 as the major product, accompanied by the formation of
3.3 Application of Photocatalysts in Green Chemistry: Solar Energy Conversion
CO, C2H6 and C2H4 as minor products [11–15]. The product distributions are affected by the kind of co-catalysts used such as Pt or Cu, e.g. PtTiO2 and CuTiO2 yield CH4 and CH3OH as the major product, respectively. Recently, a highly dispersed tetrahedrally coordinated titanium oxide species incorporated within zeolite frameworks has been found to exhibit high and unique photocatalytic reactivity for the reduction of CO2 with H2O [11–15]. Electrons and holes produced in the conduction and valence bands, respectively, of the semiconducting TiO2 powdered catalysts under UV irradiation were found to play a major role in various photocatalytic reactions. However, as the size of the TiO2 particles is reduced to less than 100 A, the gap between its highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) starts to increase, leading to an enhancement of the reduction ability of the photo-formed electrons in LUMO and also the oxidation ability of the photo-formed holes in HOMO (Figure 3.5) [11]. This size quantization effect led to high and selective photocatalytic reactions completely different from photoelectrochemical reactions occurring on bulk TiO2 powder [11]. This unique photocatalytic activity of small nanoscale TiO2 particles can be ascribed not only to an electronic modification of the TiO2 catalysts but also to the close existence of the photo-formed electron and hole pairs and their balanced contribution to the reactions. As shown in Figure 3.1b, recent advances in the preparation of catalysts have made it possible to disperse the TiO2 species within the framework structure of the zeolite to a
Figure 3.5 Advances in titanium oxide photocatalysts from extended semiconducting TiO2 particles and nanoscale molecular clusters to the isolated titanium oxide species and the changes observed in their electronic states.
j65
j 3 Applications of Environmentally Friendly TiO
66
2
Photocatalysts in Green Chemistry
Figure 3.6 Ordinary photoluminescence spectrum observed for (a) Ti/Y-zeolite catalyst prepared by ion-exchange method; (EX) Its corresponding excitation spectrum. (b), (c) Effect of the addition of CO2 and H2O, respectively, at 77 K. Excitation at 290 nm; emission monitored at 490 nm. Amount of CO2 added: (b) 8.5 and amount of H2 added: (c) 2.9 mmol g1.
molecular level by hydrothermal synthesis. The highly dispersed titanium oxide species formed exhibits unique photocatalytic reactions different from those on bulk TiO2 powders. Various spectroscopic measurements have revealed that the titanium oxide species incorporated within various zeolite frameworks (Ti/zeolites) exist in an isolated four-fold coordination sphere having a TiO bond distance of about 1.83 A [11–15]. These titanium oxide species-containing zeolite catalysts (Ti/zeolites) exhibited a photoluminescence spectrum at around 480–490 nm by excitation at around 220–260 nm (Figure 3.6). The photoluminescence spectrum is attributed to the radiative decay process from the charge-transfer excited to ground state of the highly dispersed titanium oxide species in tetrahedral coordination, as follows [11–15]: hn
ðTi4 þ O2 Þ s
hn0
ðTi3 þ O Þ
ð3:3Þ
The addition of H2O or CO2 molecules to the Ti/zeolite catalysts led to efficient quenching of the photoluminescence and also shortening of the lifetime for the charge-transfer excited state. Such an efficient quenching of the photoluminescence suggests not only that a four-fold coordinated titanium oxide species locates at positions accessible to these small molecules but also that they interact with the titanium oxide species in both their ground and excited states. In fact, UV irradiation of the Ti/zeolite catalysts in the presence of CO2 and H2O led to the photocatalytic reduction of CO2 to form CH3OH and CH4 as major products in addition to CO, O2, C2H4 and C2H6 as minor products at 323 K, while the yields of these photoformed products increased with good linearity with respect to the UV irradiation time [11– 15]. Figure 3.7 shows the relationship between the coordination number of the
3.3 Application of Photocatalysts in Green Chemistry: Solar Energy Conversion
Figure 3.7 Relationship between the coordination numbers and photocatalytic reactivities of the titanium oxides.
titanium oxide species of the Ti/zeolite catalysts as obtained from XAFS analysis and the selectivity for the formation of CH3OH in the photocatalytic reduction of CO2 with H2O on various Ti/zeolite catalysts. A clear dependence of the selectivity for the formation of CH3OH on the coordination numbers of the titanium oxide species can be observed, i.e. the lower the coordination number of the titanium oxide species, the higher is the selectivity for CH3OH formation. Bulk TiO2 semiconducting photocatalysts did not show any reactivity for the formation of CH3OH from CO2 and H2O. From these results, it was proposed that a highly efficient and selective photocatalytic reduction of CO2 to CH3OH by H2O could be achieved using Ti/zeolite catalysts involving a highly dispersed four-fold coordinated titanium oxide species in their framework as the active species [11–15]. 3.3.3 Direct Photocatalytic Decomposition of NO into N2 and O2
The development of efficient photocatalytic systems which can decompose NOx directly into N2 and O2 is strongly desired in order to establish clean and environmentally-friendly deNOx systems for atmospheric purification. It has been reported that the decomposition reaction of NO can proceed photocatalytically on powdered
j67
j 3 Applications of Environmentally Friendly TiO
68
2
Photocatalysts in Green Chemistry
Figure 3.8 View of soundproof highway walls coated with TiO2 photocatalysts for the elimination of NO (Osaka, April 1999).
TiO2 at room temperature [11–14, 18]. In fact, TiO2 photocatalysts have been practically applied for soundproof highway walls to eliminate the NO emitted from automobiles, as shown in Figure 3.8. When the TiO2 photocatalyst is irradiated by UV light in the presence of NO in atmospheric conditions, NO is oxidized into NO2 and then further oxidized into NO3. This NO3 species on the TiO2 surface can be removed as HNO3 by water in the form of, for example, raindrops. In addition to powdered photocatalysts, it has been reported that highly dispersed titanium oxide photocatalysts can decompose NO directly into N2 and O2. UV light irradiation of powdered TiO2 and Ti/Y-zeolite catalysts prepared by ion-exchange (ex-Ti/Y-zeolite) or impregnation (imp-Ti/Y-zeolite) methods in the presence of NO led to the evolution of N2, O2 and N2O in the gas phase at 275 K with different yields and product selectivity [11–14, 18]. The yields of the photo-formed products increased linearly against the UV irradiation time and the reaction ceased immediately when irradiation was discontinued. A comparison of the photocatalytic activity for the Ti/ Y-zeolite catalysts and the widely used bulk TiO2 powdered catalyst was of special interest. Specific photocatalytic reactivities for the Ti/Y-zeolite catalysts, normalized for the unit amount of TiO2 included, were much higher than that for the bulk TiO2 [11–14, 18]. Moreover, the selectivity for the formation of N2 depended strongly on the type of catalyst. The relationship between the coordination number of the titanium oxide species and the selectivity for N2 formation in the photocatalytic decomposition of NO on various types of titanium oxide-based photocatalysts are shown in Figure 3.7. There is a clear dependence of the N2 selectivity on the coordination number of the titanium oxide species [11–14, 18]. From these results,
3.3 Application of Photocatalysts in Green Chemistry: Solar Energy Conversion
it was shown that a highly efficient and selective photocatalytic reduction of NO into N2 and O2 could be achieved with ex-Ti/Y-zeolite incorporating a highly dispersed isolated tetrahedral titanium oxide as the active species. The formation of N2O as the major product was also observed for the bulk TiO2 and imp-Ti/Y-zeolite catalysts, which included the aggregated octahedrally coordinated titanium oxide species. Based on these results, a reaction mechanism for the photocatalytic decomposition of NO on the isolated tetrahedral titanium oxide species could be proposed, as shown in Scheme 3.1. The NO molecule could adsorb on the oxide species as weak ligands to form the reaction precursors. Under UV irradiation, the charge-transfer excited complexes of the oxides, (Ti3 þ O) , were formed. Within their lifetimes, an electron transfer from the trapped electron center, Ti3 þ , into the p-antibonding orbital of NO takes place and, simultaneously, an electron transfer from the p-bonding orbital of another NO into the trapped hole center, O, occurs. These electron transfers led to the direct decomposition of two sets of NO on (Ti3 þ O) into N2 and O2 under UV irradiation in the presence of NO even at 275 K. With the aggregated or bulk TiO2 catalysts, the photo-formed holes and electrons rapidly separate spatially from each other (with large distances between the holes and electrons), thus preventing the simultaneous activation of two NO on the same active sites and resulting in the formation of N2O and NO2 in place of N2 and O2 [11– 14, 18]. Moreover, the decomposed N and O species reacted with NO on different sites to form N2O and NO2, respectively. These results clearly demonstrate that the use of zeolites as supports enabled the anchoring of a titanium oxide species in a highly dispersed state within the zeolite cavities and such tetrahedrally coordinated titanium oxide photocatalysts are promising for applications in removal systems of toxic NOx compounds in the atmosphere.
Scheme 3.1 Reaction scheme of the photocatalytic decomposition of NO into N2 and O2 on the ex-Ti/Y-zeolite catalyst at 275 K.
j69
j 3 Applications of Environmentally Friendly TiO
70
2
Photocatalysts in Green Chemistry
3.3.4 Application to the Purification of Air Polluted with Various Organic Compounds
In recent years, various air-purifying systems using TiO2 photocatalysts have been commercialized. One such system is equipped with a highly active rectangular column-structured TiO2 photocatalyst constructed directly on silica sheets [19]. These rectangular column-structured TiO2 photocatalysts were prepared by a wet or dry process, as shown in Figure 3.9. SEM images of the synthesized TiO2 photocatalysts are shown in Figure 3.9a and b. The rectangular column-structured crystals, with a width of 100–500 nm and a length of 1000–5000 nm, were observed to be anchored perpendicularly to the substrate in a very dense state with stable mechanical strength. TEM images revealed that the TiO2 crystal has a hollow structure which consists of an outer TiO2 shell of high density and an inner region of low density. Moreover, XRD analysis revealed that the TiO2 crystals have an anatase polycrystalline structure. The air-purifying system incorporating the rectangular column-structured titanium oxide photocatalyst is shown in Figure 3.10a. The photocatalytic performance of these purifiers for the complete oxidation of contaminants such as formaldehyde into CO2 and H2O as compared with other TiO2 photocatalytic systems is shown in Figure 3.10b. The efficiency of the air purifiers using activated carbon or absorbents, systems A and B, respectively, were seen to decrease gradually and reach zero as the
Figure 3.9 Synthesis method for rectangular column titanium oxide photocatalysts anchored on a silica sheet and SEM images of rectangular column-structured titanium oxide photocatalysts anchored on a silica sheet: (a) 6000; (b) 15 000.
3.3 Application of Photocatalysts in Green Chemistry: Solar Energy Conversion
Figure 3.10 (a) Air purification systems applying the rectangular column-structured TiO2 photocatalysts; (b) comparison of the capacity for formaldehyde decomposition with other purifiers under different systems.
absorbents and activated carbons became saturated with the various contaminants. In contrast, the air purifier applying the rectangular column-structured TiO2 showed high efficiency in decomposing formaldehyde, the concentration decreasing rapidly to below the guideline limits issued by the Ministry of Health, Labor and Welfare of Japan. These results clearly show that the air-purifying system using these TiO2 photocatalysts can be applied in practice to decompose harmful organic compounds which exist in our living space. 3.3.5 Application to the Purification of Water Polluted with Toxic Compounds Such as Dioxins
In a reaction system in the presence of organic compounds together with water and q air, both the produced O2 and OH formed from the photo-formed electrons and holes, respectively, easily react with the organic compounds, resulting in their complete oxidation into CO2 and H2O [2, 16]. In fact, this strong oxidation ability for the TiO2 photocatalysts has been applied to the purification of water polluted with toxic compounds such as dioxins, which were then completely mineralized into CO2, H2O and HCl under UV irradiation [2, 16]: O2
Cn Hm Oz Cly ! nCO2 þ yHCl þ wH2 O hn
ð3:4Þ
The effective photocatalytic reactivity of these TiO2 photocatalysts has been applied in practice for the purification of underground water polluted with volatile organic compounds (VOCs) such as tetrachloroethylene [2, 16].
j71
j 3 Applications of Environmentally Friendly TiO
72
2
Photocatalysts in Green Chemistry
It has also been reported that visible light-responsive TiO2 photocatalysts can be prepared by applying a metal ion implantation method. By metal ion implantation, metal ions, such as Fe þ , Mn þ and V þ, are accelerated to have a high kinetic energy (150 keV) and are implanted into the bulk of the TiO2. TiO2 catalysts subjected to metal ion implantation were able to absorb and work efficiently as photocatalysts even under visible light irradiation. Significantly, they were found to exhibit reactivity for the liquid-phase degradation of 2-propanol diluted in water at 295 K under visible light (l > 450 nm) irradiation [20]. This advanced preparation method should open the way towards the widespread use of photocatalysts, significantly, for environmental remediation including water purification. 3.3.6 Superhydrophilic Properties of TiO2 Thin Films and Their Application in Self-cleaning Materials
TiO2 thin films have been known to exhibit superhydrophilicity under UV irradiation since the mid-1990s [2]. As shown in Figure 3.11, the contact angle of water droplets on a TiO2 thin film reaches almost zero under UV light and reaches a so-called superhydrophilic state. Under such a superhydrophilic state, water has a tendency to spread perfectly across the TiO2 surface. Moreover, the superhydrophilicity of TiO2 thin films has been applied for practical purposes such as anti-fogging mirrors, since water cannot form droplets on a TiO2 thin film under such a superhydrophilic state, as shown in Figure 3.11. The superhydrophilicity of TiO2 thin films has also been applied in self-cleaning materials such as window glass and architectural tiles since
Figure 3.11 (a) Changes in the contact angle of water droplets under UV irradiation on TiB binary oxide thin films; (b) application of superhydrophilic properties for anti-fogging mirrors in automobiles.
3.4 Development of Visible Light-responsive TiO2 Photocatalysts
the high wettability of the TiO2 thin-film surface can prevent adhesion of oil and dust on the TiO2 surface. TiB binary oxide thin films including ultrafine TiO2 nanoparticles of octahedral coordination and dispersed in the host B2O3 have also been successfully prepared by an ionized cluster beam (ICB) deposition method using multi-ion sources [10, 21]. As shown in Figure 3.11, it was found that Ti–B binary oxide thin films prepared by ICB deposition exhibit higher photoinduced hydrophilic properties than untreated pure TiO2 thin films [10, 21]. These advanced binary oxide thin films can be effectively applied in self-cleaning materials to realize a cleaner living environment.
3.4 Development of Visible Light-responsive TiO2 Photocatalysts 3.4.1 Modification of the Electronic State of TiO2 by Applying an Advanced Metal Ion Implantation Method
TiO2 semiconductors have a relatively large bandgap of 3.2 eV, corresponding to wavelengths shorter than 388 nm. Therefore, TiO2 can make use of only 3–4% of the solar energy that reaches the Earth, as mentioned previously. In order to develop TiO2 photocatalysts that can operate under visible light, a metal ion implantation method was applied to modify the electronic properties of bulk TiO2 photocatalysts by bombarding them with high-energy metal ions [8–10]. Metal ion implantation of the TiO2 with various transition metal ions such as V, Cr, Mn, Fe and Ni was found to lead to a large shift in the absorption band of the catalysts towards the visible region. As can be seen in Figure 3.12, the absorption band of the Cr ion-implanted TiO2 shifts
Figure 3.12 UV–visible absorption spectra of (a) TiO2 and (b)–(d) Cr ion-implanted TiO2 photocatalysts. Amount of implanted Cr ions (mmol g1): (a) 0; (b) 0.22; (c) 0.66; (d) 1.3.
j73
j 3 Applications of Environmentally Friendly TiO
74
2
Photocatalysts in Green Chemistry
smoothly towards visible regions, the extent depending on the amount of metal ions implanted [8–10]. On the other hand, TiO2 catalysts impregnated or chemically doped with Cr ions exhibit a new absorption band at around 420 nm as a shoulder peak due to the formation of an impurity energy level within the bandgap. These results indicate that the method of doping causes the electronic properties of the TiO2 catalyst to be modified in completely different ways, thus confirming that only metal ion-implanted TiO2 catalysts show such shifts in the absorption band toward the visible region. The local environment of the implanted metal ion was investigated by XAFS (XANES and EXAFS) measurements [8–10]. The results show that in the TiO2 catalysts chemically doped with Cr ions by an impregnation method, the ions were present as aggregated Cr oxides having an octahedral coordination similar to that of Cr2O3 and a tetrahedral coordination similar to that of CrO3. On the other hand, in the catalysts physically implanted with Cr ions, the ions were present in a highly dispersed and isolated state in an octahedral coordination, clearly indicating that the Cr ions are incorporated into the lattice positions of the TiO2 catalyst in place of the Ti ions. These findings clearly show that modification of the electronic state of TiO2 catalysts by metal ion implantation is closely associated with the strong and long-distance interaction which arises between the TiO2 and metal ions implanted and not with the formation of impurity energy levels within the bandgap of the catalysts. The photocatalytic activity of Cr ion-implanted TiO2 was examined for the direct decomposition reaction of NO. As shown in Figure 3.13, the unimplanted original or
Figure 3.13 Time profiles of the direct photocatalytic decomposition of NO into N2 and N2O on unimplanted pure TiO2 catalyst and Cr ion-implanted TiO2 catalyst with Cr ions of 6.6 107 mol g1 TiO2 under visible light irradiation (l > 450 nm).
3.4 Development of Visible Light-responsive TiO2 Photocatalysts
chemically doped TiO2 catalysts show no activity for the decomposition reaction of NO under visible light (l > 450 nm) [8–10]. However, visible light irradiation (l > 450 nm) of the Cr ion-implanted TiO2 catalysts led to the decomposition of NO into N2, O2 and N2O with good linearity with respect to the irradiation time. The metal ionimplanted TiO2 catalysts were therefore found to allow the absorption of visible light up to a wavelength of 400–600 nm and to operate effectively as photocatalysts. It is important to emphasize that the photocatalytic reactivity of the metal ionimplanted TiO2 catalysts under UV light (l < 380 nm) retained the same photocatalytic efficiency as the unimplanted original TiO2 catalyst. When metal ions were chemically doped into the TiO2 catalyst, the photocatalytic efficiency decreased dramatically under UV irradiation due to the rapid recombination of the photoformed electrons and holes through the impurity energy levels formed by the doped metal ions within the bandgap of the catalyst. These results clearly suggest that physically implanted metal ions do not work as electron–hole recombination centers but only to modify the electronic properties of the TiO2 catalyst. 3.4.2 Design of Visible Light-responsive Ti/Zeolite Catalysts by Applying an Advanced Metal Ion Implantation Method
Titanium oxide photocatalysts anchored within various zeolites exhibited unique and high photocatalytic activity for various reactions such as the direct decomposition of NO into N2 and O2 and the reduction of CO2 with H2O. However, the isolated tetrahedral titanium oxide species absorbs UV light of wavelengths below 300 nm since the HOMO–LUMO energy gap of this isolated tetrahedral titanium oxide species becomes significantly larger than that of bulk TiO2 due to the size quantization effect. In other words, titanium oxide photocatalysts cannot utilize the abundant solar energy that reaches the Earth, necessitating a UV light source for its use as a photocatalyst. From this viewpoint, photocatalysts which can operate efficiently under both UV and visible light are urgently required for practical and widespread use. A modification of the electronic properties of Ti/zeolite photocatalysts by bombarding them with high-energy metal ions led to the discovery that metal ion implantation with various transition metal ions such as V and Cr, accelerated by high electric fields, can produce a large shift in the absorption band toward visible light regions [8–12]. Figure 3.14 shows the effect of V ion implantation on the diffuse reflectance UV–visible absorption spectra of Ti-containing mesoporous materials, Ti-HMS and Ti-MCM-41. Their absorption spectra at around 200–260 nm can be attributed to the charge-transfer absorption process, involving an electron transfer from the O2 to the Ti4 þ ion of the highly dispersed tetrahedrally coordinated titanium oxide species of these catalysts [8–12]. The spectra shifted smoothly towards the visible region, the extent depending strongly on the number of V ions implanted. These results indicate that the interaction of the implanted V ions with the tetrahedrally coordinated titanium oxide species led to the modification of the electronic properties of the titanium oxide species within the zeolite framework, enabling them to absorb visible light.
j75
j 3 Applications of Environmentally Friendly TiO
76
2
Photocatalysts in Green Chemistry
Figure 3.14 UV–visible absorption spectra of V ion-implanted (a) Ti-HMS and (b) Ti-MCM-41 catalysts. Implanted V ions (from left to right): 0, 0.66, 1.3, 2.0 mmol g1 catalyst).
The photocatalytic activity of the V ion-implanted Ti-HMS and Ti-MCM-41 was also investigated for the decomposition of NO into N2 and O2 under visible light irradiation (l > 420 nm). Visible light irradiation of the V ion-implanted Ti-HMS led to the efficient decomposition of NO into N2 and O2, whereas the unimplanted original Ti-HMS exhibited no activity for the reaction under the same reaction conditions [8–12]. Moreover, no NO decomposition could be confirmed under UV (l < 300 nm) or visible light (l > 420 nm) on the V ion-implanted HMS. These results show that ion implantation is an effective technique for the modification of the electronic properties of titanium oxide photocatalysts, enabling them to absorb and operate under visible light (l > 420 nm) with high efficiency. The local environment of the implanted metal ion was investigated by XANES and EXAFS (XAFS) analyses [8–10]. The V K-edge FT-EXAFS spectra of the Ti-HMS catalyst implanted with V ions show that the nearest neighbors of the V environment are not the same as in vanadium oxide-based catalysts (e.g., V2O5) and suggest the formation of tetrahedral titanium oxides having a VOTi instead of a VOV linkage, as shown in Figure 3.1b [8–12]. These findings show that the formation of the VOTi bridge structures between the isolated tetrahedrally coordinated titanium oxide species and implanted V ions affect the electronic structure of the isolated titanium oxide species, leading to a red shift in the absorption spectra of these catalysts. 3.4.3 Preparation of Visible Light-responsive TiO2 Thin-film Photocatalysts by an RF Magnetron Sputtering Deposition Method
The simple one-step preparation of visible light-responsive TiO2 thin films has been successfully achieved by applying an RF magnetron sputtering (RF-MS) deposition method, as shown in Figure 3.15 [3–6, 8–10]. The system is equipped with a substrate (quartz or Ti foil) center positioned in parallel just above the source material, the calcined TiO2 plate. The calcined TiO2 plate is sputtered by an Ar plasma by inducing
3.4 Development of Visible Light-responsive TiO2 Photocatalysts
Figure 3.15 Schematic diagram of the RF magnetron sputtering (RF-MS) deposition method.
an RF power of 300 W in an Ar atmosphere and the TiO2 thin film is prepared on the quartz or Ti metal foil substrate mounted on a heater. The substrate temperature (TS) was held fixed in the range 473–873 K. Figure 3.16 shows the effect of the TS on the UV–visible transmission spectra of the TiO2 thin films prepared on a quartz substrate with a TiO2 thickness of 1.2 mm. TiO2 thin films prepared at 473 K (UV-TiO2) exhibited no absorption at wavelengths longer than 380 nm, whereas the TiO2 thin films prepared at a TS higher than 673 K (VisTiO2) were yellow and exhibited considerable absorption at wavelengths longer than 380 nm, permitting the absorption of visible light [3–6, 8–10]. Moreover, the onset of the absorption band of Vis-TiO2 prepared at 873 K (Vis-TiO2-873) shifted toward longer wavelength regions of around 600 nm, compared with around 400 nm for UVTiO2 prepared at 473 K. Hence it can be seen that the precise control of the substrate
Figure 3.16 UV–visible absorption (transmittance) spectra of TiO2 thin films prepared on a quartz substrate by the RF-MS deposition method. Preparation temperature: (a) 373; (b) 473; (c) 673; (d) 873; (e) 973 K.
j77
j 3 Applications of Environmentally Friendly TiO
78
2
Photocatalysts in Green Chemistry
temperature allows the development of unique visible light-responsive TiO2 thin-film photocatalysts, while the efficiency of visible light absorption increased with increase in the substrate temperature. It should be noted that the coexistence of O2 with Ar in the sputtering chamber led to the formation of UV light-responsive TiO2 thin films regardless of the substrate temperature (TS > 473 K) during the sputtering deposition process. This suggests that TiO2 deposition under pure Ar gas without any trace of O2 is one of the major factors in the successful preparation of Vis-TiO2 thin-film photocatalysts. To clarify the origin of visible light absorption for the Vis-TiO2 thin film, SIMS and TEM analyses were carried out. The secondary ion intensity due to 18 O for UV-TiO2 exhibited a stoichiometric O : Ti value of 2.00, independent of the depth from the TiO2 surface, whereas in the case of Vis-TiO2, the O : Ti value gradually decreased from the top surface (O : Ti ¼ 2.00 0.01) to the inside bulk (1.93 0.01), showing a distinct contrast to UV-TiO2 [3–6, 8–10]. It could therefore be proposed that the unique declined composition of Vis-TiO2 thin films of an anisotropic structure causes a significant perturbation in the electronic structure of the TiO2, permitting the absorption of visible light. As shown in Figure 3.17, cross-sectional TEM observations revealed that Vis-TiO2 consists of large columnar crystals growing perpendicular to the substrate and the surface of the columnar crystals is covered with a stoichiometric TiO2 phase. Such a stable surface phase of the columnar crystals worked as a passive phase and could protect the slightly reduced TiO2 phases inside the bulk from complete oxidation [3–6, 8–10]. The prepared Vis-TiO2 thin film was found to act as an efficient photocatalyst for the separate evolution of H2 and O2 under solar light irradiation (Figure 3.4) and also for the complete oxidation of organic compounds into CO2 and H2O even under visible light irradiation [8–10].
Figure 3.17 Cross-sectional TEM image of the Vis-TiO2 thin-film photocatalyst prepared on a quartz substrate.
References
3.5 Conclusion
In this chapter, recent advances in research on the photocatalytic reactivity and photoinduced superhydrophilic properties of titanium oxide-based catalysts and their applications in green chemistry, e.g. solar energy conversion and environmental protection, were summarized. Special attention was focused on the application of an advanced metal ion implantation method for the preparation of second-generation TiO2 photocatalysts and highly dispersed titanium oxide photocatalysts, both of which can operate under visible light irradiation. Moreover, a new and cost-efficient RF magnetron sputtering deposition method to produce visible light-responsive TiO2 thin-film photocatalysts was also presented. Detailed characterizations of such unique visible light-responsive TiO2 photocatalysts were carried out along with investigations into the various photocatalytic reactions that could be initiated, significantly for processes related to environmental remediation. It has been demonstrated that advanced physical ion-engineering techniques can provide new approaches to the design of unique titanium oxide photocatalysts which can utilize solar energy on a global scale as the most abundant and safe energy source.
References 1 Fujishima, A. and Honda, K. (1972) Nature, 238, 37–38. 2 Fujishima, A., Rao, T.N. and Tryk, D.A. (2000) Journal of Photochemistry and Photobiology C: Photochemistry Reviews, 1, 1–21. 3 Matsuoka, M., Kitano, M., Takeuchi, M., Anpo, M. and Thomas, J.M. (2005) Topics in Catalysis, 35, 305–310. 4 Kitano, M., Tsujimaru, K. and Anpo, M. (2006) Applied Catalysis A-General, 314, 179–183. 5 Matsuoka, M., Kitano, M., Takeuchi, M., Tsujimaru, K., Anpo, M. and Thomas, J.M. (2007) Catalysis Today, 112, 51–61. 6 Kitano, M., Matsuoka, M., Ueshima, M. and Anpo, M. (2007) Applied Catalysis AGeneral, 325, 1–14. 7 Anpo, M. and Che, M. (2000) Advances in Catalysis, 44, 119–257. 8 Anpo, M. and Takeuchi, M. (2003) Journal of Catalysis, 216, 505–516. 9 Anpo, M. (2004) Bulletin of the Chemical Society of Japan, 77, 1427–1442.
10 Anpo, M., Dohshi, S., Kitano, M., Hu, Y., Takeuchi, M. and Matsuoka, M. (2005) Annual Review of Materials Science, 35, 1–27. 11 Anpo, M. and Thomas, J.M. (2006) Chemical Communications, 31, 3273– 3278. 12 Anpo, M. and Matsuoka, M. (2008) Turning Points in Solid-state, Materials and Surface Science, Royal Society of Chemistry, Cambridge, pp. 492–506. 13 Anpo, M. (ed.) (2000) Photofunctional Zeolites, NOVA Science, New York. 14 Matsuoka, M. and Anpo, M. (2003) Photochemistry and Photobiology C, 3, 225–252. 15 Yamashita, H., Ikeue, K., Takewaki, T. and Anpo, M. (2002) Topics in Catalysis, 18, 95–100. 16 Anpo, M. (2000) Pure and Applied Chemistry, 72, 1265–1270. 17 Takeuchi, M., Anpo, M., Hirao, T., Itoh, N. and Iwamoto, N. (2001) Journal of the Surface Science Society of Japan, 22, 561–567.
j79
j 3 Applications of Environmentally Friendly TiO
80
2
18 Yamashita, H., Ichihashi, Y., Anpo, M., Hashimoto, M., Louis, C. and Che, M. (1996) The Journal of Physical Chemistry, 100, 16041–16044. 19 Kudo, T., Kudo, Y., Ruike, A., Hasegawa, A., Kitano, M. and Anpo, M. (2007) Catalysis Today, 122, 14–19.
Photocatalysts in Green Chemistry 20 Yamashita, H., Harada, M., Misaka, J., Takeuchi, M., Neppolian, B. and Anpo, M. (2003) Catalysis Today, 84, 191–196. 21 Dohshi, S., Takeuchi, M. and Anpo, M. (2003) Catalysis Today, 85, 199–206.
j81
4 Nanoparticles in Green Catalysis Mazaahir Kidwai
4.1 Introduction
A revolution is occurring in science and technology, based on the recently developed ability to measure, manipulate and organize matter on the nanoscale from 1 to hundreds of billionths of a meter (Figure 4.1). Nanoscience has taken scientists around the world by storm. It hopes to revolutionize the world we live in with striking breakthroughs in areas such as materials and manufacturing, electronics, medicine and healthcare, environment and energy, chemicals and pharmaceuticals, biotechnology and agriculture, computation and information technology [1]. In recent years, there has been growing interest in the catalytic properties of transition metal nanoparticles. The high surface area-to-volume ratio of solid-supported metal nanoparticles (1–10 nm in size) is mainly responsible for their catalytic properties, and this can be exploited in many industrially important reactions [2]. This chapter highlights the use of different metal nanoparticles or their compounds in the organic transformations and syntheses of biologically important compounds. It is does not intended to provide a comprehensive review but to serve as a critical overview of the scope and applications of different metal structures such as spherical nanoparticles, nanorods, nanoplates and nanocubes in catalytic phenomena.
4.2 Advanced Catalysis by Gold Nanoparticles
Gold has long been regarded as a poorly active catalyst. A recent theoretical calculation has explained why the smooth surface of Au is noble in the dissociative adsorption of hydrogen [3]. However, when Au deposited as nanoparticles on metal oxides by mean of coprecipitation and deposition–precipitation techniques, it exhibits high catalytic activity for CO oxidation [4, 5]. The extraordinarily high catalytic activity of supported Au catalysts for CO oxidation at room temperature arises from the reaction of CO adsorbed on the steep edge and the corner sites of the
Handbook of Green Chemistry, Volume 2: Heterogeneous Catalysis. Edited by Robert H. Crabtree Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32497-2
j 4 Nanoparticles in Green Catalysis
82
Figure 4.1 Overview of nanoscience.
metallic Au particles with oxygen molecules adsorbed at the perimeter sites on the support surface. Hayashi et al. found that Au supported on TiO2 (Degussa, p-25) could catalyze the epoxidation of propylene in the gas phase containing O2 and H2 (Scheme 4.1) [6, 7]. Propylene oxide is the one of the important bulk chemicals, used for producing polyurethane and polyols.
Scheme 4.1
4.2 Advanced Catalysis by Gold Nanoparticles
In general, the hydrogenation of hydrocarbons is a structure-insensitive reaction over most metal catalysts (Scheme 4.2). A characteristic feature of gold as catalyst is that partial hydrogenation takes place very selectively [8]. In the hydrogenation of a,b-unsaturated aldehydes, selectivity for the hydrogenation of C¼O against that of C¼C has recently been reported to reach 40–50% when Au particles are larger than 2 nm in diameter [9, 10].
Scheme 4.2
Zhu and Angelici recently reported carbon monoxide oxidative amination using gold powder of size 103 nm [11]. The reaction of CO with primary amines in the presence of gold results in the formation of isocyanate, which further reacts with another molecule of amine to form urea (Scheme 4.3). The proposed mechanism suggests that CO absorbed on Au should be susceptible to attack by amines.
Scheme 4.3
Lazar and Angelici reported the oxidative amination of isocyanide to give carbodiimides using gold powder with an approximate size of less than 103 nm and with water as a by-product (Scheme 4.4) [12].
Scheme 4.4
Polymer-stabilized Au clusters dispersed in water act as efficient quasi-homogenous catalysts for the aerobic oxidation of alcohols. A monodisperse Au–poly(N-vinyl2-pyrrolidone) cluster is used for the aerobic oxidation of p-hydroxybenzyl alcohol (Scheme 4.5) [13]. The catalytic activity per unit cluster surface area increases rapidly with decreasing size, which is associated with non-metallic electronic structures.
Scheme 4.5
j83
j 4 Nanoparticles in Green Catalysis
84
Polymer microsphere-stabilized gold metallic colloids have been used for the reduction of 4-nitrophenol to 4-aminophenol with sodium borohydride as reductant at an ambient temperature of 25 C (Scheme 4.6) [14]. Gold nanoparticles are stabilized by active carboxylic acid on the surface of poly(divinylbenzene-co-acrylic acid) microspheres with carboxylic acid-to-gold molar ratios from 20 : 1 to 75 : 1. The microsphere-stabilized gold colloids were recycled by filtration over a G-6 sinteredglass filter after the catalytic reaction and the recovery of the catalyst was proven further by the activity of recycling four times. Preliminary results indicated that the resultant polymer microsphere-stabilized gold nanocolloids were recyclable with high catalytic activity in water medium, which may find further wide application in the field of green chemistry.
Scheme 4.6
The reusable nanosized gold particle-based chiral bisoxazoline catalyst acts like a homogeneous catalyst in the ene reaction between 2-phenylpropene and ethyl glyoxylate in dichloromethane (Scheme 4.7) [15]. A variety of hybrid chiral ligands having spacers of different chain lengths were converted into copper(II) triflate complexes and used in the ene reaction. The catalyst could be reused five times; however this results in a decrease in the yield but the catalyst remains highly enantioselective. The enantioselectivity is up to 86% with fresh catalyst and decreases only to 84% in the fourth run.
Scheme 4.7
Green synthesis of propargylamine via a three-component coupling reaction of aldehyde, alkyne and amine was achieved using recyclable gold nanoparticles in tetrahydrofuran (Scheme 4.8) [16]. The reaction was carried out in an inert atmosphere at 75–80 C. The maximum reaction rate was observed for an average gold particle of diameter of about 20 nm. It is important to stress that the catalyst was
4.3 Nickel Nanoparticles: a Versatile Green Catalyst
recycled and reused for five or seven runs with only a slight drop in the catalytic activity. This drop was induced by agglomeration of Au nanoparticles, which was size dependent. A significant feature of Au nanoparticles in this reaction is that they can be reused without further purification and without using any additives or cofactors.
Scheme 4.8
4.3 Nickel Nanoparticles: a Versatile Green Catalyst
In the field of heterogeneous catalysis, Group VIII metal-based catalysts such as palladium, platinum and nickel are among the most active transition metal catalysts. As the most widely available element of the triad, nickel is an important transition metal catalyst that can be used in many heterogeneous reactions. In recent years, numerous publications have reported the reduction of metal cation, especially Ni2 þ in solution, by hydrazine to produce metallic nanoparticles [17, 18]. Nickel oxide nanoparticles were found to be an effective catalyst in the catalytic activation of CCl bonds in chloroalkanes. The reaction of dichloromethane with amine in the presence of nickel oxide nanoparticles having a small size of 10–20 nm gave a quantitative yield of quaternary ammonium salts, depending on the amine (Scheme 4.9) [19]. Steric and electronic factors of the substituents on the amines played an important role in the catalytic activity. A primary alkyl substituent on the amine gave a quantitative yield whereas no product was obtained with N,N-diisopropylethylamine. It was found that nickel oxide nanoparticles were easily recovered and could be reused at least five times without loss of catalytic activity.
Scheme 4.9
In situ-generated Ni(0) nanoparticles and molecular hydrogen from the system NiCl2–Li–DTBB–ROH is a new mild and simple methodology for the efficient stereoselectivecispartialhydrogenationofalkynes(Scheme4.10).Thismethodologyhasbeen used for the synthesis of cis alkenes, partial hydrogenation of terminal alkynes and reduction of dienes to alkenes. Other conditions allow complete reduction of alkynes and alkenes to the corresponding alkane [20]. The versatility of this reducing system is that it results in complete hydrogenation of both internal and terminal alkynes. This simple protocol is used for the hydrogenation of multiple carbon–carbon bonds, including dienes, to corresponding alkane.
j85
j 4 Nanoparticles in Green Catalysis
86
Scheme 4.10
a,b-Unsaturated carbonyl compounds can be selectively reduced by in situ generated Ni(0) nanoparticles and molecular hydrogen (Scheme 4.11). One of the main advantages of this methodology is that handling of external molecular hydrogen is avoided since it is generated in situ in the reaction flask.
Scheme 4.11
The exocyclic conjugated carbon–carbon double bond of (R)-( þ )-pulegone, even though tetrasubstituted, was reduced with a high yield to give a cis–trans (75 : 25) diastereomeric mixture of ( þ )-isomenthone and ()-menthone. Most of the reduction using this protocol was regioselective with diastereomeric mixture and furnished cis isomer in major amount [21]. Even a- and b-ionone were both reduced to the expected product in good yields without any double bond isomerization. Kidwai and co-workers recently reported Ni(0) nanoparticles as a green catalyst for chemoselective reduction of carbonyl compounds using ammonium formate as a hydrogen donor (Scheme 4.12) [22, 23]; 10 mol% Ni nanoparticles in tetrahydrofuran selectively reduced the carbonyl group in the presence of other functional groups, viz. NO2, CN and alkene, to give the corresponding alcohols in excellent yields. This protocol reduces both aromatic and heteroaromatic aldehydes chemoselectively.
Scheme 4.12
NiFe bimetallic nanoparticles [24] were used for the catalytic dechlorination of the highly toxic pentachlorophenol in aqueous solution. The dechlorination is believed to take place on the surface sites of these particles. The dechlorination efficiency was 46%
4.4 Copper Nanoparticles: an Efficient Catalyst
within 30 min under optimal conditions and the use of ultrasonic irradiation enhanced the dechlorination efficiency to 96% for the same period of time.
4.4 Copper Nanoparticles: an Efficient Catalyst
Benzo-fused azoles are an important class of compounds and provide a common heterocyclic scaffold in biologically active and medicinally significant compounds. Benzoxazoles are found in a variety of natural products and are important targets in drug discovery. Benzoxazoles can be prepared by oxidative cyclization of Schiff bases using Cu nanoparticles (Scheme 4.13) [25].
Scheme 4.13
Without any catalytic effect of the Cu nanoparticles, 2-aminophenol condenses with the aldehyde to form the Schiff base, which then undergoes cyclization to form the benzoxazilidine in the presence of K2CO3. The resulting benzoxazilidine then undergoes aromatization via Cu nanoparticle catalysis to give the benzoxazole. Supported metallic Cu nanoparticles [26] were active for the selective dehydrogenation of methanol to produce formaldehyde and hydrogen with 100% H2 selectivity. Cu nanoparticles were impregnated on MoO2, MoO3, ZnO and SiO2 and used for selective dehydrogenation. A combination of Cu2O nanoparticles with P(o-tol)3 shows high catalytic activity for Stille cross-coupling reactions (Scheme 4.14). Li et al. evaluated a series of Cu catalysts and ligands and found Cu2O nanoparticles with P(o-tol)3 to be an effective catalyst for Stille cross-coupling reactions [27]. This catalytic protocol with Cu2O–P(o-tol)3–TBAB (tetrabutylammonium bromide) can be recovered and reused at least three times without any loss of catalytic activity for reactions of aryl iodides and activated aryl bromides. This methodology has good tolerance of aromatic rings.
Scheme 4.14
j87
j 4 Nanoparticles in Green Catalysis
88
In situ-generated Cu nanoparticles can be efficiently used for the Heck reaction in ionic liquids. The Heck reaction of aryl iodides and activated aryl bromides catalyzed by copper bronze in tetrabutylammonium bromide as solvent and tetrabutylammonium acetate was developed by Calo et al. [28]. The copper nanocolloids were derived from reaction of iodobenzene with copper bronze (Scheme 14.15). This catalytic system was recycled 20 times with an average total turnover number of approximately 40. These nanocolloids can be stored for months without loss of activity thanks to the stabilization effect of the TBAB.
Scheme 4.15
Freshly prepared copper nanocolloids were used for the formation of methanol. In a typical procedure, a high-pressure autoclave was charged with copper nanocolloid solution followed by sequential pressurization with synthesis gas (H2CO) and CO2 at roomtemperature [29].Theformationof the methanol was observed at130 Cwith a gas chromatographic detector attached to the high-pressure autoclave. Cu nanoparticles of size 14–17 A were efficiently used for aza-Michael reactions [30] in various N-alkyl- and N-arylpiperazines. Cyanoethylation of aryl- or alkypiperazines was carried out using acrylonitrile at room temperature with 10 mol% of Cu nanoparticles (Scheme 14.16). The reaction is highly chemoselective and only cyanoethylates secondary amines. The reaction studies showed that in the presence of anilines such as p-anisidine, p-toluidine and o-aminophenol, only secondary amines undergo the aza-Michael reaction.
Scheme 4.16
CHsp bond activation of terminal alkynes is of fundamental interest in organic synthesis. Several systems have been developed for CH activation, which mainly include transition metal complexes. Recently, Kidwai et al. reported Cu nanoparticlecatalyzed CH bond activation [31]. Cu nanoparticle-catalyzed A3 coupling via CH bond activation results in the formation of propargylamines (Scheme 4.17). This protocol can be widely used with a variety of secondary amines and aldehyde. Moreover, the nanoparticles can be recycled and reused and avoids the use of co-catalyst and gives the product in quantitative yield.
Scheme 4.17
4.5 Bimetallic Nanoparticles in a Variety of Reactions
4.5 Bimetallic Nanoparticles in a Variety of Reactions
Bimetallic salts or metals have wide application in organic synthesis. The electrode potential of these bimetallic salts can be used to carry out several transformations. Especially palladium catalysts are widely used in organic synthesis. In recent decades, their utilization in a variety of transformations has shown continuous impressive growth, achieving an important place in the arsenal of the practicing organic chemist. Highly dispersed nickel or palladium nanoparticles and silica aerogels were used as catalysts in the Mizoroki–Heck reaction [32]. Different nanocomposite silica aerogels were synthesized using Ni(OAc)2 and Pd(OAc)2 as a metal source. In situ-generated Pd nanoparticles in MCM-41 were used in the catalytic hydrogenation of alkynes in the liquid phase [33]. It was found that Pd particles incorporated in MCM-41 were significantly less active in liquid-phase alkyne hydrogenation than those on the external surface of MCM-41. Despite the difference in catalytic activities, the selectivities of Pd–MCM-41 is very similar and the Z-stereoisomer is selectively formed in the hydrogenation of 3-hexyne. The selectivity is found to be greater than 94%. 1-Pentyne, 1-hexyne and 3-hexyne were also reduced efficiently. In this methodology, hydrogen at 105 Pa was used in the hydrogenation reactor. Ultrasonically generated Pd(0) nanoparticles were found to catalyze the Sonogashira coupling reaction at ambient temperature. Reaction proceeds in acetone or ionic liquid as solvent at room temperature. Ionic liquid solvents such as 1,3-din-butylimidazolium tetrafluoroborate provide excellent chemoselectivity with considerably enhanced reaction rates through the formation of stable and crystalline clusters of zerovalent Pd nanoparticles [34]. A Pd–biscarbene complex (1) is formed by the reaction of an ionic liquid and PdCl2 and this complex can be isolated and characterized; it then undergoes sonochemical conversion to give polydisperse Pd(0) nanoparticles as catalyst for the reaction (Scheme 4.18). The reaction proceeds without copper as co-catalyst and ligand.
Scheme 4.18
j89
j 4 Nanoparticles in Green Catalysis
90
A similar reaction was also reported by Li et al. [35]. using palladium colloids in polyvinylpyrrolidone (PVP) (Scheme 4.19). The reaction takes place in 1% Pd(0) colloids with average diameter less than 7 nm. The catalyst can be recycled eight times without loss of catalytic activity.
Scheme 4.19
Palladium catalysts, however, are expensive and this may limit their utilization in some cases. To overcome this limitation, a number of new catalyst systems have been developed and Cacchi et al. reported carbon aerogel doped with Pd nanoparticles as an efficient catalyst for the hydroxycarbonylation of an aryl iodide [36] to form the corresponding carboxylic acid. High content Pd–carbon aerogels were prepared by sol–gel polymerization [37] of formaldehyde with the potassium salt of 2,4-dihydroxybenzoic acid, followed by K þ exchange with Pd2 þ ions from 0.1 M Pd(OAc)2 solution in acetone and subsequent supercritical drying with CO2. Transmission electron microscopy of the Pd–carbon aerogel showed nanoparticles with a mean particle size 19 4 nm. Hydroxycarbonylation of aryl iodides was carried out in the presence of lithium formate and acetic anhydride as an internal source of carbon monoxide with Pd–carbon aerogel (Scheme 4.20).
Scheme 4.20
Zhou et al. studied bimetallic PtCu nanoparticles as catalysts for the heterogeneous reduction of NO in the gas phase with H2 as the reducing agent [38]. Palladium nanoparticles deposited on polydimethylphosphazene (PDMP) were used for a Hecktype reaction [39]. Pd nanoparticles, obtained by the metal vapor synthesis technique, were deposited on PDMP and showed high catalytic activity in the Heck CC coupling of iodobenzene with methyl acrylate (Scheme 4.21). In the reaction, triethylamine is used as base to shift the equilibrium towards the product as it reacts with the hydrochloric acid formed during the course of reaction.
Scheme 4.21
References
The reaction results in the formation of trans-methyl cinnamate. Pd–PDMP is also used for the alkylative cyclization of 3-ethyl-3-methyl-7-octen-1-yne with iodobenzene (Scheme 4.22). NOE 1 H NMR experiments showed that the resulting product, 1,2-bis (alkylidene)cyclohexanes, possess Z-stereochemistry.
Scheme 4.22
A similar reaction was also carried out with in situ-generated Pd nanoparticles in ionic liquids [40]. Hence the nanoparticle approach has wide application in organic synthesis. Most of the nanoparticle catalysts can be recycled and reused to give quantitative yield even after the fifth or sixth cycle. Their catalytic use, recyclability and atom efficiency put nanoparticles squarely into the field of green chemistry.
References 1 Bell, A.T. Science (2003), 299, 1688. 2 R.A. Van Santen, P.W.N.M. Van Leeuwen, J.A. Moulijn and B.A. Averill (eds), (1999) Catalysis: an Integrated Approach, 2nd edn, Elsevier, Amsterdam, Stud. Surf. Sci. Catal. p. 123. 3 Hammer, B. and Norskov, J.K. (1995) Nature, 376, 238. 4 Haruta, M. (1997) Catalysis Today, 36, 153. 5 Haruta, M., Yamada, N., Kobayashi, T. and Iijima, S. (1989) Journal of Catalysis, 115, 301. 6 Nijhuis, T.A. and Weckhuysen, B.M. (2006) Catalysis Today, 117, 84. 7 Hayashi, T., Tanaka, K. and Haruta, M. (1998) Journal of Catalysis, 178, 566. 8 Jia, J., Haraki, K., Konodo, J.N., Domen, K. and Tamaru, K. (2000) The Journal of Physical Chemistry. B, 104, 11153. 9 Claus, P., Bruckner, A., Mohr, C. and Hofmeister, H. (2000) Journal of the American Chemical Society, 122, 11430. 10 Mohr, C., Hofmeister, H., Lucas, M. and Claus, P. (2000) Chemical Engineering & Technology, 23, 324.
11 Zhu, B. and Angelici, R.J. (2006) Journal of the American Chemical Society, 128, 14460. 12 Lazar, M. and Angelici, R.J. (2006) Journal of the American Chemical Society, 128, 10613. 13 Tsunoyama, H., Sakuari, H. and Tsukuda, T. (2006) Chemical Physics Letters, 429, 528. 14 Liu, W., Yang, X. and Huang, W. (2006) Journal of Colloid and Interface Science, 304, 160. 15 Ono, F., Kanemasa, S. and Tanaka, J. (2005) Tetrahedron Letters, 46, 7623. 16 Kidwai, M., Bansal, V., Kumar, A. and Mozumdar, S. (2007) Greem Chemistry, 9, 742. 17 Ni, X., Su, X., Yang, Z. and Zheng, H. (2003) Journal of Crystal Growth, 252, 612. 18 Shen, J., Hu, Z., Zhang, L., Li, Z. and Chen, Y. (1996) Applied Physics Letters, 15, 715. 19 Park, K.H., Jung, I.G., Chung, Y.K. and Han, J.W. (2007) Advanced Synthesis and Catalysis, 349, 411. 20 Alonso, F., Osante, I. and Yus, M. (2007) Tetrahedron, 63, 93.
j91
j 4 Nanoparticles in Green Catalysis
92
21 Alonso, F., Osante, I. and Yus, M. (2006) Synlett, 3017. 22 Kidwai, M., Bansal, V., Saxena, A., Shankar, R. and Mozumdar, S. (2006) Tetrahedron Letters, 47, 4161. 23 Kidwai, M., Mishra, N.K., Bansal, V., Kumar, A. and Mozumdar, S. (2008) Catal. Commun., 9, 612. 24 Zhang, W., Quan, X., Wang, J., Zhang, Z. and Chen, S. (2006) Chemosphere, 65, 58. 25 Kidwai, M., Bansal, V., Saxena, A., Aerry, S. and Mozumdar, S. (2006) Tetrahedron Letters, 47, 8049. 26 Tada, M., Bal, R., Namba, S. and Iwasawa, Y. (2006) Applied Catalysis A-General, 307, 78. 27 Li, J.H., Tang, B.X., Tao, L.M., Xie, Y.X., Liang, Y. and Zhang, M.B. (2006) The Journal of Organic Chemistry, 71, 7488. 28 Calo, V., Nacci, A., Monopoli, A., Leva, E. and Cioffi, N. (2005) Organic Letters, 7, 617. 29 Vukojevic, S., Trapp, O., Grunwaldt, J.D., Kiener, C. and Schuth, F. (2005) Angewandte Chemie-International Edition, 44, 7978. 30 Verma, A.K., Kumar, R., Chaudhary, P., Saxena, A., Shankar, R., Mozumdar, S. and Chandra, R. (2005) Tetrahedron Letters, 46, 5229. 31 Kidwai, M., Bansal, V., Mishra, N.K. and Kumar, A. (2007) Synlett, 1581. 32 Martinez, S., Manas, M.M., Vallribera, A., Schubert, U., Roig, A. and Molins, E. (1093) New Journal of Chemistry, 2006, 30.
33 Mastalir, A., Rae, B., Kiraly, Z. and Molnar, A. (2007) Journal of Molecular Catalysis AChemical, 264, 170. 34 Gholap, A.R., Venkatesan, K., Pasricha, R., Daniel, T., Lahoti, R.J. and Srinivasan, V.K. (2005) The Journal of Organic Chemistry, 70, 4869. 35 Li, P., Wang, L. and Li, H. (2005) Tetrahedron, 61, 8633. 36 Cacchi, S., Cotet, C.L., Farizi, G., Forte, G., Goggiamani, A., Martin, L., Martinez, S., Molins, E., Moreno-Manas, M., Petrucci, F., Roig, A. and Vallribera, A. (2007) Tetrahedron, 63, 2519. 37 Martinez, S., Vallribera, A., Cotet, C.L., Popovici, M., Martin, L., Roig, A., Moreno-Manas, M. and Molins, E. (2005) New Journal of Chemistry, 29, 1342. 38 Zhou, S., Varughese, B., Eichhorn, B., Jackson, G. and Mcllwrath, K. (2005) Angewandte Chemie-International Edition, 44, 4539. 39 Panziera, N., Pertici, P., Barazzone, L., Caporusso, A.M., Vitulli, G., Salvadori, P., Borsacchi, S., Geppi, M., Veracini, C.A., Marta, G. and Bertinetti, L. (2007) Journal of Catalysis, 246, 351. 40 Fei, Z., Zhao, D., Pieraccini, D., Ang, W.H., Geldbach, T.J., Scopelliti, R., Chiappe, C. and Dyson, P.J. (2007) Organometallics, 26, 1588.
j93
5 Heterogreeneous Chemistry Heiko Jacobsen
5.1 Introduction
The chemical philosophy of Green Chemistry originated in the early nineties and the Pollution Prevention Act that was passed in 1990 in the USA probably played a substantial role in increasing awareness of environmental and sustainability issues. With the advent of the new millennium, the importance of green chemistry has found widespread recognition. The World Summit on Sustainable Development, held in 2002 in Johannesburg, South Africa, provided ample evidence of a growing consensus that the world faces serious challenges to its sustainability. The list of major issues includes concerns regarding energy, resource depletion and the generation and dispersion of toxic substances [1]. This forum was one of the first major events in the 21st century that highlighted the need for global awareness of issues and problems relating to the worlds future. The choice of the Nobel Peace Prize awarded in 2007 [2], 5 years after the Johannesburg Summit, demonstrates that the concerns recognized at the beginning of the new millennium are far from resolved, are ongoing, are of major concern for our future, will clearly influence developments in politics, economics and science and need to become a component of our way of thinking. The prize was awarded to the Intergovernmental Panel on Climate Change and to Albert Arnold Gore Jr for their efforts to build up and disseminate greater knowledge about man-made climate change and to lay the foundations for the measures that are needed to counteract such change. The related issue of Global Warming is perhaps the most challenging, but certainly not the only problem that the world has to come to grips with and solve over the next decade to guarantee its sustainability for long-term survival. In order to achieve the goal of sustainability, Green Chemistry [3–6] is clearly evolving as a quintessential part of the foundation from which efficient and sensible solutions to the challenges at hand are derived. Green Chemistry is characterized by a move away from the command and control approach to environmental protection to a more scientifically based and economically minded approach [7].
Handbook of Green Chemistry, Volume 2: Heterogeneous Catalysis. Edited by Robert H. Crabtree Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32497-2
j 5 Heterogreeneous Chemistry
94
Since Green Chemistry is better viewed as a philosophy rather than a science, it naturally gives rise to a wide spectrum of different interpretations and therefore it behoves us to establish a good understanding of Green Chemistry before we go into more detail. Green Chemistry is an approach to the synthesis, processing and use of chemicals that reduces risks to humans and the environment. Existing chemical procedures are modified and new chemistries are developed that are effective, efficient and more environmentally benign. The benefits to industry and also the environment are all a part of the positive impact that Green Chemistry is having in the chemistry community and in society in general. As we noted above, Green Chemistry reflects a shift away from the historic command and control approach to environmental problems that dealt with issues of waste treatment, waste control and waste clean-up through regulation and towards preventing pollution at its source. Rather than accepting waste generation and disposal as unavoidable, Green Chemistry seeks new technologies that are cleaner and economically competitive. In 2002, Anastas and Kirchhoff summarized the first decade of Green Chemistry and analyzed its origins, its current status and future challenges [8]. They clearly illustrated the fact that Green Chemistry has demonstrated how fundamental scientific methodologies can protect human health and the environment in an economically beneficial manner. They further noted that significant progress is being made in several key research areas, such as catalysis, the design of safer chemicals and environmentally benign solvents and the development of renewable feedstocks. These trends and activities continue to receive major attention within the Green Chemistry community and highlight the potential of the science of chemistry to solve many of the global environmental challenges that the world faces at the beginning of the 21st century. The origins and basis of Green Chemistry chart a course for achieving environmental and economic prosperity inherent in a sustainable world and Anastas and Kirchhoff illustrate how the 12 Principles of Green Chemistry [3, 8] provide helpful guidelines in the efficient realization of the abstract concept and philosophy of Green Chemistry. The 12 Principles of Green Chemistry constitute the foundation of a new way of conducting chemistry and in principle address three major aspects of chemistry, the selection of reactants and products, synthetic methodologies and issues relating to chemical risks and energy requirements, summarized in Figure 5.1. The term Heterogreeneous Chemistry essentially relates to the fact that Green Chemistry has to be recognized as a philosophy, rather than a branch of well-defined science. From an inspection of the 12 Principles of Green Chemistry, it becomes clear that Green Chemistry cannot easily be reduced to just a few scientific principles, but incorporates a wide variety of strategies all devoted to the greater goal of achieving world sustainability. Thus, Green Chemistry is a heterogeneous subject in the purest form of the word: it consists of elements that are not of the same kind or nature, it is not uniform in structure or composition, it is composed of parts of different kinds that have widely dissimilar elements or constituents. The term Heterogreeneous Chemistry was coined with reference to a significant development which utilizes heterogeneous catalysis and follows many of the 12 Principles of Green Chemistry [9]. Heterogreeneous Chemistry also includes
5.1 Introduction
Figure 5.1 The 12 Principles of Green Chemistry. Adapted from [8].
heterogeneous aspects in a pure chemical definition, as in composed of different substances or the same substance in different phases. The most prominent heterogeneous area in chemistry is indeed heterogeneous catalysis, and it has been noted that catalysis is one of the foundational pillars of Green Chemistry [10]. The design and application of new catalysts and catalytic systems are simultaneously achieving the dual goals of environmental protection and economic benefit. Within the framework of the 12 Principles of Green Chemistry, catalysis offers numerous Green Chemistry benefits, including lower energy requirements, catalytic versus stoichiometric amounts of materials, increased selectivity and decreased use of processing and separation agents. Furthermore, catalysis allows the use of less toxic materials. Heterogeneous catalysis, in particular, addresses the goals of Green Chemistry by providing the ease of separation of product and catalyst, thereby eliminating the need for separation through distillation or extraction. In addition, environmentally benign catalysts such as clays and zeolites, may be used to replace the more hazardous catalysts currently in use. Significant progress is being made in several key research areas including environmental catalysis, which is undergoing a transformation from pollution abatement to pollution prevention [11]. The benefits to human health, the environment and the economic goals realized through the use of catalysis in manufacturing and processing are illustrated by focusing on the catalyst design and catalyst applications.
j95
j 5 Heterogreeneous Chemistry
96
This chapter is not intended to present an ultimate and authoritative review and reference work to the many facets of Heterogreeneous Chemistry. Instead, we will elucidate the idea of Heterogreeneous Chemistry and especially catalysis in a few recent exemplarily developments, hoping to disseminate the philosophy of Green Chemistry that continues to be a main incentive for ongoing research activities.
5.2 Heterogreeneous Catalysis
Heterogreeneous Catalysis incorporates many of the 12 Principles of Green Chemistry. To reiterate, it is not so much a breakthrough in one particular scientific approach that is at the heart of Heterogreeneous Chemistry, but rather a change in the conceptual framework dictating strategies how to conduct chemistry. This change and the accompanying increasing awareness manifest themselves in their increasing presence and recognition in the chemical literature of the last decade. The results of a simple topic search in the Web of Science database of ISI Web of Knowledge for publications that contain the topic entries Green Chemistry, Green Chemistry and Catalysis, and also Green Chemistry and Heterogeneous Catalysis, illustrate the growing awareness of the philosophy of Green Chemistry and of research on heterogeneous catalysis that addresses the topic of Green Chemistry (Figure 5.2).
Figure 5.2 Number of publications during the 10 years from 1997 to 2006 that contain topic entries Green Chemistry, Green Chemistry and Catalysis and Green Chemistry and Heterogeneous Catalysis, according to a search in the Web of Science database of ISI Web of Knowledge.
5.2 Heterogreeneous Catalysis
The afore mentioned simplified search is by no means exhaustive, as it does not cover all research activities of the past 10 years that are of significant importance to the field of Green Chemistry. In fact, the contribution by Dumesic and co-workers on catalysts for hydrogen production from biomass-derived hydrocarbons [12], which gave rise to the neologism Heterogreeneous Chemistry, is not captured in this basic search. What the graph and statistics in Figure 5.2 indicate, however, is more than growing awareness of a green philosophy in chemistry – they intimate a link between Green Chemistry and heterogeneous catalysis. We will explore the role of Heterogreeneous Chemistry in a few selective examples that illustrate how and why catalysis is one of the fundamental cornerstones of Green Chemistry. Before we continue, it is helpful to provide a brief definition of one of the key terms that we will encounter in our discussion: biomass. Living and recently dead biological material that can be used as fuel or for industrial production, be it specifically grown for a particular purpose or be it any type of biodegradable waste, is commonly referred to as biomass. Although the definition of biomass is broad and heterogeneous, it excludes organic material which has been transformed by geological processes into substances such as coal or petroleum. What differentiates biomass from other organic material that can be used for the same purposes is the timeline of its production. The time scale of production of biomass is on an equal footing with that of its consumption. This aspect renders biomass a renewable, rather than depletable, feedstock. 5.2.1 An Exemplarily Reaction – Catalysts for Hydrogen Production from Biomass-Derived Hydrocarbons
Concerns about the depletion of fossil fuel reserves and pollution caused by continuously increasing energy demands make hydrogen an attractive alternative energy source. Hydrogen is currently derived form non-renewable natural gas and petroleum [13], but could in principle be generated from renewable sources such as biomass and water. Dumesic and co-workers demonstrated that hydrogen can be produced from sugars and alcohols at temperatures near 500 K in a single-reactor aqueous phase reforming process using a platinum-based catalyst (Pt/Al2O3) [12]. The selectivity for hydrogen production increases greatly when oxygenated hydrocarbons are employed that have a C : O stoichiometry of 1 : 1. This study suggests that catalytic aqueous phase reforming might prove useful for the generation of hydrogen-rich fuel gas from carbohydrates extracted from renewable biomass and biomass waste streams. The transformation of oxygenated hydrocarbons into H2 and CO2 occurs according to the following stoichiometric reaction: Cn On H2n þ 2 þ nH2 O ! ð2n þ 1ÞH2 þ nCO2
ð5:1Þ
The selective generation of hydrogen by this route, however, proves to be difficult, since the products readily react at low temperatures to form alkanes and water: nCO2 þ ð3n þ 1ÞH2 ! Cn H2n þ 2 þ 2nH2 O
ð5:2Þ
j97
j 5 Heterogreeneous Chemistry
98
The steps shown in Scheme 5.1 are proposed to be involved in the formation of hydrogen and alkanes.
Scheme 5.1 Reaction pathways for hydrogen production by reactions of oxygenated hydrocarbons with water. The symbol represents a surface metal site; [M] indicates presence of a metal surface; denotes bond cleavage. Adapted from [12].
The reactant undergoes dehydrogenation steps on the metal surface to yield adsorbed intermediates before the cleavage of CC or CO bonds occurs. For the catalyst employed, it is noted that PtC bonds are more stable than PtO bonds and adsorbed species are probably bonded preferentially to the catalyst surface through PtC bonds. Subsequent cleavage of CC bonds leads to the formation of CO and H2, and the CO reacts with water to form CO2 and H2 by the water gas shift (WGS) reaction (Equation 5.3), a reaction which has been shown to be catalyzed by ceriabased active non-metallic Au and Pt species [14]. CO þ H2 O ! CO2 þ H2
ð5:3Þ
Davda and Dumesic have also shown how this equilibrium can be tuned in such a manner as to produce CO-poor hydrogen [15]. The expansion of gas bubbles formed in the process by the vaporization of water leads to decreasing partial pressures of H2 and CO2, thereby favoring the water gas shift and lowering the CO concentration. This process leads to the production of fuel cell-grade H2 at high pressures. Two reaction channels that influence the selectivity of hydrogen production from hydrocarbons have been identified [12]. A series-selectivity challenge is represented by the hydrogen-consuming reaction, in which either one or both of CO and CO2 react with H2, leading to alkanes and water by methanation and Fischer–Tropsch reactions. In a parallel-selectivity challenge, undesirable alkanes can form on the catalyst surface by cleavage of CO bonds, followed by hydrogenation of the resulting adsorbed species. The above-described scenario is extended by proposing further competing reactions, such as the formation of organic acids by dehydrogenation reactions, which in turn lead to the formation of alkanes from carbon atoms not bonded to oxygen atoms.
5.2 Heterogreeneous Catalysis Table 5.1 Experimental data for reforming of oxygenated hydrocarbons with Pt- and Ni-based catalysts.
SnNi b
Pt/Al2O3a Parameter
Sorbitol
Glycerol
Ethylene glycol
Sorbitol
Glycerol
Ethylene glycol
T (K) p (bar) %H2c %CnH2nþ2c
498 29 66 15
498 29 75 19
498 29 96 4
498 25.8 65 19
498 25.8 81 13
498 25.8 95 4
a
Data compiled from [12]. Data compiled from [19]. c %H2, hydrogen selectivity; %CnH2n þ 2, alkane selectivity; see [12] for a definition of selectivities. b
In addition, the proposed reaction scheme has been investigated in a theoretical study based on self-consistent periodic density functional calculations [16]. In a model reaction, the relative stabilities and reactivities of surface species on Pt(111) derived by subsequent removal of hydrogen atoms from ethanol have been considered. Transition states for CC and CO bond cleavage reactions have been located and the results from these calculations, combined with transition state theory, predict that the rate constant for CC bond cleavage in ethanol will be faster than for CO bond cleavage on Pt(111) at temperatures higher than about 550 K. Further, the calculated value of the rate constant for CC bond cleavage in ethanol is predicted to be much higher than that for CC bond cleavage in ethane on Pt(111). Similarly, the rate of CO bond cleavage in ethanol is predicted to be much higher than for CO bond cleavage in carbon monoxide on Pt(111). These calculations highlight the effectiveness of the platinum catalyst employed, which disfavors elemental reaction steps occurring in reactions competing with hydrogen formation. The experimental results for aqueous phase reforming of sorbitol (C6O6H14), glycerol (C3O3H8) and glycol (C2O2H6) are presented in Table 5.1. Sorbitol can be produced by hydrogenation of glucose [17], a compound that is directly relevant to biomass utilization. Glycerol and glycol, in turn, can be obtained from hydrogenolysis of sorbitol [18]. A variety of other biomass related sources readily provide additional access to glycerol and glycol. The reactions were performed over a 3 wt% Pt catalyst supported on nanofibers of g-alumina. Reactions were carried out under pressure in a tubular reactor at 498 K by continuously feeding an aqueous solution having a 1 wt% feed concentration of organic compound. The data in Table 5.1 indicate that a H2 selectivity as high as 96% was achieved. The corresponding alkane selectivities range from 4 to 19%. The selectivity for production of H2 improves in the order C6O6H14 < C3O3H8 < C2O2H6. The fractions of feed carbon detected in effluent gaseous and liquid streams yield a complete carbon balance, indicating that only negligible amounts of carbon have been deposited on the catalyst. Of further importance is the catalyst performance, which was stable for times on stream of at least 1 week. While the data in Table 5.1 establish that Pt-based catalysts show high activities and good selectivity for the production of hydrogen from biomass-derived alcohols,
j99
j 5 Heterogreeneous Chemistry
100
improvements are necessary to render the process useful. Highly active catalytic materials that satisfy the series and parallel selectivity challenges (Scheme 5.1) at lower materials costs are particularly desirable. In a follow-up study, Dumesic and co-workers introduced a new active system for hydrogen production by aqueous phase reforming of biomass-derived oxygenated hydrocarbons, namely a tin-promoted Raney nickel catalyst (SnNi ) [19]. They found that the addition of tin to nickel decreases the rate of methane formation from CO bond cleavage while maintaining the high rates of CC bond cleavage required for hydrogen formation. Referring back to Table 5.1, results for the same experiments as described above under the same experimental conditions but using an SnNi rather than a Pt/Al2O3 catalyst indicate that the cheap non-precious metal catalyst compares favorably with the expensive platinum-based catalyst. The above-described reactions obey a number of the 12 principles [3, 8] that guide Green Chemistry, such as use of feedstock derived from renewable raw materials, use of efficient and cheap catalysts, avoidance of extensive use of auxiliary materials and prevention of waste. This work also outlines how new catalysts can be expected to provide impetus and lower potential barriers for the implementation of greener industrial processes and technologies. Known chemistry and technology are tailored in a green fashion. This process is accompanied by certain by-products, which, however, do not need to be avoided, but can be maximized and open up another valuable aspect of biomass utilization. This is in accord with the second of the 12 Principles of Green Chemistry, that synthetic methods should be designed to maximize the incorporation of all materials used in the process into the final product. We will explore this aspect in more detail in the next section. 5.2.2 Transportation Fuels from Biomass – Catalytic Processing of Biomass-derived Reactants
Biomass serves as the prototypical green feedstock not only for hydrogen production, but also for transportation fuels. In fact, it has been stated that biomass is the only practical source of renewable liquid fuel [20]. In addition, biofuels generate significantly less greenhouse gas emissions than do fossil fuels and can even be greenhouse gas neutral if efficient methods for biofuels production are developed [21]. A biomass growth and manufacturing scheme relies only on CO2, H2O, light, air and nutrients to produce energy, some of which is fed back into the biomass processing scheme. The three main aspects necessary for a carbohydrate economy are growth of the biomass feedstock, biomass conversion into a fuel and fuel utilization. The format of an integrated biomass production-conversion scheme is illustrated in Figure 5.3. Of the above-mentioned three technologies, the steps of biomass conversion and fuel utilization are processes that can be addressed in the spirit of Green Chemistry and heterogeneous catalysis plays an essential role in designing and employing environmentally friendly and sustainable procedures. The biorefinery concept [21] and the chemical catalytic transformations of biomass-derived feedstocks have been extensively reviewed [22, 23] and hydrolysis, dehydration, isomerization, aldol condensation, reforming, hydrogenation and oxidation have been identified as key reactions involved in the processing of biomass. Here, we shall discuss in detail the
5.2 Heterogreeneous Catalysis
Figure 5.3 Integrated biomass production–conversion system for the sustainable production of transportation fuels. Adapted from [21].
production of renewable alkanes by aqueous phase reforming of biomass-derived oxygenates [24], and identify the green aspects of the technology involved. We have seen in the previous section how it is possible to produce hydrogen from biomass-derived oxygenates, such as glycerol and sorbitol, using a process of aqueous phase reforming (APR). We also noted that the production of hydrogen is accompanied by the formation of light alkanes, primarily methane. Through tuning of the reaction conditions and catalysts, it is possible to tailor the aqueous phase reforming process selectively to produce a clean stream of heavier alkanes consisting primarily of butane. The APR process offers a simple route for the production of renewable fuels from biomass and we will discuss in more detail, and as an illustrative example, how aqueous phase reforming of sorbitol, the sugar alcohol obtained by hydrogenation of glucose, can be adjusted for conversion of sorbitol to heavier alkanes consisting primarily of butane, pentane and hexane [24]. This process incorporates concepts of Green Chemistry, as it utilizes known chemical reactions, such as the WGS reaction and Fischer–Tropsch synthesis in combination with metal catalysts as well as acid catalysts, to conduct chemical reactions according to the 12 Principles of Green Chemistry. Production of alkanes by aqueous phase reforming of sorbitol takes place by a bifunctional reaction pathway that involves first the formation of hydrogen and CO2 on an appropriate metal catalyst [M] and then the dehydration of sorbitol on a solid acid catalyst These initial steps are followed by hydrogenation of the dehydrated reaction intermediates on the metal catalyst. When these steps are balanced properly, the hydrogen produced in the first step is fully consumed by hydrogenation of the dehydrated reaction intermediates, which leads to the overall conversion of sorbitol to alkanes plus CO2 and water. The essential features of the bifunctional reaction pathway for the production of alkanes from sorbitol are depicted in Scheme 5.2.
j101
j 5 Heterogreeneous Chemistry
102
Scheme 5.2 Reaction pathways for the production of alkanes from sorbitol over catalysts with metal [M] and solid acid components. F–T, Fischer–Tropsch; denotes bond cleavage.
Hydrogen is produced on the metal by cleavage of CC bonds followed by the WGS reaction. Dehydrated species are first formed on acid sites, typically solid acids, and then migrate to metal sites [M] where they undergo hydrogenation reactions. Repeated cycling of dehydration and hydrogenation reactions in the presence of hydrogen then leads to formation of heavier alkanes. Formation of lighter alkanes takes place by more rapid cleavage of CC bonds compared with hydrogenation of dehydrated reaction intermediates and the selectivities for production of various alkanes by aqueous phase reforming depend on the relative rates of CC bond cleavage, dehydration and hydrogenation reactions. The selectivities for the production of alkanes can be varied by changing the catalyst composition and the reaction conditions and by modifying the reactor design. In addition, the selectivities can be modified by co-feeding hydrogen with the aqueous sorbitol feed, which then leads to a process in which sorbitol can be converted to alkanes and water without the formation of CO2. Catalysts typically employed are the transition metals Pd or Pt as heterogeneous metal catalysts [M] and silica–alumina composites (SiO2Al2O3) as solid acid catalysts. Solid acids are conventional materials that have wide applications in chemical production and the arsenal of solid acids is constantly expanding, including, for example, carbon materials as strong protonic acids [25]. In Table 5.2 are presented the experimental results for aqueous phase reforming of sorbitol obtained from different catalytic systems. The data demonstrate how aqueous phase reduction of oxygenated hydrocarbons can be adjusted such that it is not hydrogen that evolves as major product of the reformation process [12], but heavier alkanes instead. The H2 selectivity is 4% or lower, the carbon selectivities are largest for heavier alkanes, such as hexane (C6H14), and alkane to total gas-phase carbon selectivities are as high as 98%. Hydrogen, which is needed for the hydrogenation reaction, can be produced in situ by aqueous phase reforming or it can be co-fed into the reactor with the aqueous sorbitol reactant. The alkanes formed are straight-chain compounds with only minor amounts of branched isomers. The selectivities for production of heavier alkanes can be controlled by the choice of the reaction conditions and by co-feeding hydrogen to the reactor.
5.2 Heterogreeneous Catalysis Table 5.2 Experimental data for aqueous phase reforming of sorbitol.a
Catalysts and additives to feed Parameter
Pt–SiAl
Pt–SiAl
Pt–SiAl H2
Pt–SiAl H2
Pd–SiAl H2
T (K) p (bar) % Carbon selectivityb CH4 C2H6 C3H8 C4H10 C5H12 C6H14 % H2 selectivityb % Alkane to total gas phase carbonc
498 52.7
538 60.7
498 29.3
498 34.8
538 58.2
10 10 9 15 19 37 1 53
7 10 11 16 21 35 4 60
0 3 6 11 25 55 — 91
0 4 7 11 24 54 — 91
2 2 4 8 28 56 — 98
a
Data compiled from [23]. See [23] for a definition of selectivities. c The gas-phase carbon consists of alkanes and carbon dioxide. b
This green protocol draws from the wealth of established chemistry and combines well-known reactions such as Fischer–Tropsch processes and the WGS reaction with new applications of old catalyst systems. This reaction qualifies as heterogreen by way of a variety of different aspects, which are combined to achieve the shared goal of creating reactions with environmentally friendly sustainability. Furthermore, this reaction employs the quintessential green solvent water. We will further elaborate on solvents in Green Chemistry, in particular in connection with heterogeneous aspects, at a later point in our discussion. 5.2.3 Diesel Fuels from Biomass – Heterogreeneous Processes for Biodiesel Production
Diesel fuel produced from petroleum is a hydrocarbon mixture, composed of about 75% saturated hydrocarbons and 25% aromatic hydrocarbons. The average chemical formula for common diesel fuel is C12H23, ranging from approximately C10H20 to C15H28. Diesel-powered engines generally have a better fuel economy than equivalent gasoline engines and produce less greenhouse gas pollution. The term biodiesel refers to a diesel-equivalent processed fuel consisting of short-chain alkyl esters of fatty acids, such as methyl or ethyl esters, made by transesterification of vegetable oils or animal fats. Biodiesel fuels can be used alone or blended in with conventional diesel fuel, in unmodified diesel engines. Biodiesel is biodegradable and non-toxic and typically produces about 60% less net-lifecycle carbon dioxide emissions. Having the characteristics of common diesel fuels, sparing the limited resources of fossil fuels and addressing current environmental concerns render biodiesel a key element in the production of energy from biomass according to the 12 Principles of Green Chemistry and a promising alternative fuel to petrodiesel.
j103
j 5 Heterogreeneous Chemistry
104
The advantages of biodiesel as an alternative fuel and the problems involved in its manufacturing have been reviewed [26], with one of the main problems in the production of biodiesel being the identification of a suitable catalyst that is active, selective and stable under the process conditions needed for heterogeneous transesterification. Furthermore, technology development for processing more abundant lignocellulosic biomass for fuels and materials will be critical [23]. Several commercial processes to produce biodiesel as fatty acid methyl esters from vegetable oils have been developed and are available today in recognition of its green characteristics and increasing demand. These processes use homogeneous basic catalysts such as caustic soda or sodium methylate, which lead to waste products after neutralization with mineral acids. A further undesirable effect is that the byproduct glycerol is obtained as contaminated raw product requiring further purification steps. It is highly desirable to have at hand a process for biodiesel production that requires neither catalyst recovery nor aqueous treatment steps and that produces the byproduct glycerol with high purity levels and exempt from any salt contaminants. Heterogeneous catalysis holds the promise of an efficient, green and continuous biodiesel production process and we will describe an exemplarily process developed by Casanave and co-workers in more detail [27]. The transesterfication process that leads to the production of biodiesel from fatty acids is depicted in Scheme 5.3. Groups R1, R2 and R3 typically contain hydrocarbon chains of 15–20 C-atoms. It is important to note that the methanolysis is an equilibrium process. The reaction is promoted by a completely heterogeneous catalyst. This catalyst consists of a mixed oxide of zinc and aluminum, which promotes the transesterification reaction without catalyst loss. The reaction is performed with an excess of methanol and this excess is removed by vaporization and recycled to the process with fresh methanol.
Scheme 5.3 Methanolysis of fatty acids.
A reaction scheme for biodiesel production based on heterogeneous catalysis is presented in Figure 5.4. The catalyst section includes two fixed-bed reactors R1 and R2, fed with vegetable oil and methanol at a given ratio. Excess of methanol is removed after each reactor by partial evaporation. Then, esters and glycerol are separated in a settler or separator. Glycerol outputs are gathered and the residual methanol is removed by evaporation. The purification section of methyl ester output coming from the last separator consists of a finishing methanol vaporization under vacuum followed by a final purification step. The biodiesel produced by this heterogeneous process meets industrial standards and therefore is not only of environmental, but also of economic value. Of the greatest importance, however, is the fact that the quality and the value of the crude glycerol produced by biodiesel plants are significantly improved; glycerol with purity levels of at least 98% can be obtained.
5.2 Heterogreeneous Catalysis
Figure 5.4 Reaction scheme for biodiesel production based on heterogreeneous catalysis.
According to the second of the 12 Principles of Green Chemistry, glycerol from waste glycerol streams that are currently generated as by-products from the production of biodiesel can be converted over platinum-based catalysts into gas mixtures of H2 and CO at temperatures from 498 to 620 K [28]. This gas mixture, known as synthesis gas, serves as a feedstock for a variety of chemical production processes. The temperatures required for the above-mentioned catalysis are lower than those for conventional gasification of biomass, which typically are around 800–1000 K; hence the catalytic conversion of glycerol at low temperatures may allow for economical operation of a small-scale Fischer–Tropsch reactor by producing an undiluted H2CO synthesis gas mixture. The capital cost of a Fischer–Tropsch plant would be greatly reduced by eliminating the need for a biomass gasifier and further scaled down if the need for gas cleaning steps due to the use of pure glycerol were eliminated. More notable incentives for the implementation of Heterogreeneous Chemistry are that the same feedstock can be used in a variety of processes and that its by-products are efficiently incorporated in green synthesis projects. Furthermore, the green concept of recycling is already incorporated in the production process. We have seen a similar strategy in the reforming of biomass-derived oxygenates, where the desired product was obtained in repeated cycles of reforming, dehydration and hydrogenation [24]. Even the principle product obtained from green processing of biomass, biodiesel, can further be fed into other green production lines and might serve as a feedstock for chemical intermediates produced in the chemical industry, such as olefins. Subramanian and Schmidt, for example, have demonstrated how renewable olefins can be obtained from biodiesel by a process of autothermal reforming [29]. Thus, the heterogeneous advantage in the use and production of biodiesel [30] truly is a heterogreeneous advantage in all respects.
j105
j 5 Heterogreeneous Chemistry
106
5.2.4 Other Heterogreeneous Aspects of Catalysis
Although catalysis has long been utilized in order to increase efficiency, yield and selectivity in chemical processes, it has also of late gained recognition for achieving a wide range of green chemistry goals. At the dawn of the 21st century, increasingly demanding environmental legislation, public and corporate pressure and the resulting drive towards clean technology provide an unquestionable impetus for the development of new catalysis and catalytic processes. In compliance with the 12 Principles of Green Chemistry, some salient goals in the search for new catalysis are to increase process selectivity, maximize the use of starting materials, replace stoichiometric reagents with catalysts and facilitate easy separation of the final reaction mixture, including the efficient recovery of the catalyst. The examples presented so far illustrate how catalysis has evolved into one of the foundational pillars of green chemistry [10]. To complete our discussion of heterogreeneous catalysis, we consider various other aspects of catalysis all related to Green Chemistry. Again, note that the following selections are not intended to represent an authoritative and comprehensive review of the subject, but rather to present, since Green Chemistry is still in its infancy, a snapshot of green activities and developments at the end of the first decade of the new millennium. 5.2.4.1 Solid and Solid Acid Catalysts The use of efficient solid catalysts holds great promise in the developments of Green Chemistry [31]. Polymer-supported catalysts, for example, have been widely used and their popularity comes mainly from the fact that product isolation is simplified and that less harsh conditions and higher degrees of selectivity can be attained. As an additional benefit, catalysts based on high surface area inorganic support materials show good thermal stability and have therefore attracted considerable interest as solid catalysts and reagents in liquid-phase organic reactions. Chemically modified mesoporous materials can be prepared to serve as robust catalysts suitable for application in liquidphase processes such as Friedel–Crafts reactions, selective oxidations, nucleophilic substitutions and aromatic brominations, and might form the basis of some new industrial catalysts which will then replace toxic and corrosive traditional reagents [32]. The development and use of mesoporous inorganic support materials as catalysts with chemically bound active centers is emerging as an area of research which seeks to retain the green benefits of heterogenization, enhanced activity and enhanced product selectivity while avoiding the drawbacks of catalyst instability and limited reusability. A recent example of this technology is observable in the heterogeneous aluminasupported ruthenium catalyst designed by Yamaguchi and Mizuno, which is easy to prepare, inexpensive to use, capable of being recycled and efficient in the aerobic oxidation of amines [33]. Further expanding on this technology, the same group have also synthesized an organic–inorganic hybrid support, which allows catalytically active polyoxometalate anions to be immobilized [34]. This truly heterogreeneous catalytic system is effective for liquid-phase oxidation with hydrogen peroxide, such as epoxidation and sulfonation, and is reusable without any loss of catalytic performance.
5.2 Heterogreeneous Catalysis
Solid acids are the most widely studied and commonly used heterogeneous catalysts, often used in large-scale continuous vapor-phase processes such as catalytic cracking and alkane isomerizations. The examples in the previous subsections further illustrate the use of solid acids in heterogreeneous catalysis. Current activities in green research are geared towards the development of solid acid catalysts which are effective in liquid-phase organic reactions such as those employed in many batch-type reactors by fine, specialty and pharmaceutical intermediate chemical manufacturers. A recently reported example of work in this direction is silica sulfuric acid, reported by Dabiri et al., which permits the rapid and green synthesis of 2,5-disubstituted 1,3,4-oxadiazoles under solvent-free conditions [35]. The development of carbon materials as strong protonic acids [25], which illustrates the diverse research activities that all are driven by or related to Green Chemistry, exemplifies the heterogreeneous nature of this subject, as mentioned previously. 5.2.4.2 Recycling Catalysts An important aspect of heterogreeneous catalysis is the recyclability of the catalyst employed. Ideally, the catalyst is not consumed in a green process, nor does it need to be recycled, but this seldom happens in reality. Researchers are therefore looking for new ways to allow easy catalyst recycling. Mizuno and co-workers have designed an easily prepared ruthenium hydroxide catalyst on magnetite, Ru(OH)x/Fe3O4, which allows for simple and straightforward product–catalyst separation [36]. After a particular reaction, the desired separation can be easily achieved with a permanent magnet, allowing more than 99% of the Ru(OH)x/Fe3O4 catalyst to be usually recovered for each reaction. Three kinds of reactions, namely aerobic oxidation of alcohols, aerobic oxidation of amines and reduction of carbonyl compounds to alcohols using 2-propanol as a hydrogen donor, can efficiently be promoted by this magnetic catalyst. A wide variety of substrates, including aromatic, aliphatic and heterocyclic compounds, can be converted to the desired products in high to excellent yields without any additives such as bases and electron transfer mediators. Ru(OH)x/ Fe3O4 is thus an intrinsically heterogeneous catalyst and the recovered catalyst can be reused without appreciable loss of the catalytic performance. In the context of the aforementioned solid acid catalyst and with emphasis on recyclability, Paul and co-workers prepared a covalently anchored sulfonic acid on silica as an efficient and recoverable interphase catalyst [37]. The catalyst is highly stable, completely heterogeneous and recyclable several times. The Biginelli reaction was investigated as a test reaction for the new catalyst and the products were obtained in good to excellent yields, requiring only very simple work-up procedures. The ease of recyclability represents one of the major advantages of heterogeneous versus homogeneous catalysis, but homogeneous catalysis also carries advantages. Homogeneous catalysts that exist in the same phase as reactants and products are usually more selective than heterogeneous catalysts and far less affected by limitations due to slow transport of reactants and products, which has stimulated the development of strategies that facilitate the recycling of homogeneous catalysts [38]. With the concept of ease of catalyst separation in mind, Dioumaev and Bullock designed a tungsten catalyst for the solvent-free hydrosilylation of ketones that retains its activity
j107
j 5 Heterogreeneous Chemistry
108
until essentially all of the liquid substrate is converted to liquid products [39]. The product can then simply be decanted to separate the catalyst that precipitates from the products of the reaction. The incorporation of homogeneous catalysis into Green Chemistry is not a straightforward procedure, but the work of Dioumaev and Bullock illustrates how the limitations of homogeneous catalysis might be circumvented and how homogeneous catalysis might become part of Heterogreeneous Chemistry. 5.2.4.3 One-pot Catalysis Creating one-pot synthetic routes is a challenge already spawning new chemistry, enzymes, materials and mechanistic insight [40]. Through one-pot reactions, chemical products can be produced with less waste and greater economic benefits and still one-pot reactions adhere to the philosophy of Green Chemistry. Synthetic strategies combining the framework of a one-pot synthesis with green catalytic transformations are beginning to evolve, thanks to their inherent green nature. We mention as one example the one-pot synthesis of primary amides from aldoximes or aldehydes in water in the presence of a supported rhodium catalyst, reported by Mizuno and co-workers [41]. 5.2.4.4 Photocatalysis Photocatalysis numbers among the new and emerging green technologies. Since different aspects of the 12 Principles of Green Chemistry can be addressed by the use of photocatalysis, it truly stands as an example of heterogreeneous technology. One of its aspects involves the removal of organic contaminants by way of solar energy, which mainly involves oxidative decomposition of volatile organic compounds (VOCs). Photocatalysis has many advantages over other treatment methods, such as the use of the environmentally friendly oxidant O2 and low reaction temperatures. To date, TiO2 has undoubtedly proven to be the best photocatalyst for the oxidative decomposition of many organic compounds under UV irradiation. Ongoing research activities are aimed towards expanding the electromagnetic spectrum of photocatalysts and Tang et al. have reported a novel photocatalyst, CaBi2O4, which is active in the photocatalytic oxidative decomposition of organic contaminants under visible light irradiation [42]. Another facet of photocatalysis as it gains increasing importance has to do with the development of new energy sources. In view of our limited fossil fuel reserves and the pollution caused by constantly increasing energy demands, hydrogen emerges as an attractive alternative energy source. Thus, visible light hydrogen generation from water [43, 44] holds the promise and potential to become one of the main energy sources of future generations.
5.3 Solvents for Green Catalysis
The examples given under previous headings mostly address green issues dealing with sustainability, renewable feedstocks and renewable energy resources. Setting
5.3 Solvents for Green Catalysis
aside targeted goals of chemical processes, Green Chemistry, in its heterogeneous nature, also tackles the issue of how to conduct chemical reactions in a benign, sustainable and environmentally friendly manner. Solvents are key components of most chemical transformations and have gained center stage in the design of green processes. The current status of green solvents for sustainable synthesis has been reviewed by Sheldon, who addresses many of the problems, challenges and solutions in the field of solvents in chemical synthesis, and in particular organic synthesis [45]. In the context of Green Chemistry, there are several issues which influence the choice of solvent. Ideally, the solvent should be non-toxic and non-hazardous, that is, not flammable or corrosive. Further, the solvent should also be contained, that is, it should not be released to the environment. Commonly used solvents are typically separated from products by evaporation or distillation and most popular solvents are, therefore, highly volatile. Spillage and evaporation inevitably lead to atmospheric pollution, a major environmental issue of global proportions. Moreover, worker exposure to VOCs represents a serious health issue. The ideal green solvent should be modeled to avoid the above-mentioned risks and shortcomings of conventional solvents, while at the same time offering some of the same or similar properties such as polarity or boiling points. Issues surrounding a wide range of volatile and non-volatile solvents have already stimulated the fine chemical and pharmaceutical industries to seek more benign alternatives. Not all issues can be properly addressed by the nature of the solvent alone, but might be solved by the proper conduction of a chemical process. Biphasic reactions hold many green advantages such as ease of separation of products and possible catalysts, and are beginning to take their place in production processes in the chemical industry. We have repeatedly pointed out that catalysis is one of the cornerstones of Green Chemistry implementation. However, depending on the goal at hand, catalysis by itself is not always sufficient to put into practice a green synthetic process. Given that the use of solvents accounts for 50% of the post-treatment greenhouse gas emissions and 60% of the energy used in pharmaceutical processes, consideration must be given to the proper selection of solvents when designing and developing a reaction scheme [46]. Choice of catalyst and choice of solvent are ultimately intrinsically linked when problems in chemistry are solved in the green frame of mind. The term Heterogreeneous Chemistry thus reflects the complexity and multilayered nature of the many problems originating when serious consideration is given to the 12 Principles of Green Chemistry. 5.3.1 Heterogreeneous Solvent Systems
The best solvent is no solvent at all, and, if a solvent is needed, green or potentially green alternatives should be considered. Preferably, such alternative solvents should also allow easy catalyst separation and recycling. Although a number of reactions have been shown to proceed under solvent-free conditions, at present most synthetic reactions still call for the use of solvents. Therefore, alternatives to conventional organic solvents are being actively sought. Currently, one can categorize green and
j109
j 5 Heterogreeneous Chemistry
110
Figure 5.5 Heterogreeneous solvents and solvent systems.
heterogreeneous solvents into four classes, namely water, supercritical carbon dioxide (scCO2), ionic liquids (ILs) and fluorous solvents, as depicted in Figure 5.5. With respect to catalysis, the four solvent systems are further subdivided into those used preferentially for monophasic and those used preferentially for biphasic catalysis. Water and ILs straddle the border between mono- and biphasic applications. Three out of the four solvent classes, scCO2, ILs and fluorous solvents, are currently referred to as non-conventional solvents, which find extensive use in the process of greening organic chemistry. Water represents the most conventional solvent out of the classes of green solvents and, if a prototypical green solvent had to be named, water would represent the best choice. It is cheap, readily available, nontoxic, non-flammable and safe to the environment. Of additional importance to catalysis is the fact that water allows for facile catalyst separation and recycles through a biphasic catalysis mode due to its low miscibility with most organic compounds. We have chosen the example of a success story to illustrate the above-mentioned concept of biphasic catalysis and to stress the important role of water as a quintessential green solvent. The first profitable large-scale application of aqueous catalysts is the oxo process of Ruhrchemie/Rhône-Poulenc, which uses a rhodium catalyst dissolved in water [47]. This reaction is outlined in Scheme 5.4. In this biphasic catalytic reaction, the catalyst resides in the water phase, whereas the product dissolves in the organic phase. This hydroformylation reaction truly represents an environmentally benign technique: highly economic, environmentally sound and pollution reducing. Although water is a conventional solvent, heterogreeneous activities expand its use to non-conventional applications. Aqueous Barbier–Grignard-type reactions described by Li abandon fundamental ideas and principles of organic chemistry and open up new and greener avenues for organic synthesis [48]. This work spawned new developments in the execution of fundamental and conventional reaction steps in organic chemistry, such as CC bond formation, in aqueous media [49]. A recent
5.3 Solvents for Green Catalysis
Scheme 5.4 Ruhrchemie/Rhône-Poulenc process for aqueous biphasic hydroformylation.
application of aqueous organic chemistry, and a step toward a bio-based industry, is the benign catalytic esterification of succinic acid in the presence of water, reported by Clark and co-workers [50]. Supercritical carbon dioxide (scCO2) has been attracting increasing attention as an alternative and green reaction medium in recent years [51]. It is non-toxic, nonflammable, inexpensive, relatively inert towards reactive compounds and readily separable from products upon depressurization. These characteristics of scCO2 are in line with many of the 12 Principles of Green Chemistry. Furthermore, its low viscosity and high diffusivity properties confer advantages on reactions with mass transfer problems. However, scCO2 is apolar and generally only suitable in catalytic processes for compounds of low polarity. The eleventh point of the 12 Principles of Green Chemistry refers to the need for further development of analytical methods. Supercritical fluids such as scCO2 offer opportunities of in situ spectroscopic studies, which provide valuable information useful for rendering a particular chemical process more ecological or more economical [52]. The direct synthesis of propylene oxide from propylene, hydrogen and oxygen provides a recent example of the green use of carbon dioxide as solvent for organic reactions [53]. Ionic liquids (ILs), composed entirely of organic cations and organic or inorganic anions, display physicochemical properties such as low melting point, negligible vapor pressure, low flammability, tunable polarity and miscibility with other organic or inorganic compounds, that are appealing for use in catalysis and separation processes. ILs have low solubility towards low-polarity compounds, such as ethers and alkanes, which makes the application of biphasic catalysis with ILs possible. Even for monophasic catalysis, catalyst-product separation can be readily achieved by extraction of the product, leaving the catalyst in ILs for reuse.
j111
j 5 Heterogreeneous Chemistry
112
In view of the toxic or hazardous properties of many solvents, notably chlorinated hydrocarbons, and in view of serious environmental issues, such as atmospheric emissions and contamination of aqueous effluents, the chemical industry is currently exploring alternative and greener reaction media. The current emphasis on the search for novel reaction media is also fueled by the need for efficient methods of recycling homogeneous catalysts. The use of ionic liquids as novel reaction media may offer a convenient solution to both the solvent emission and the catalyst recycling problems. Applications and advantages of ILs have been reviewed by Sheldon, who emphasizes the importance of ILs in Green Chemistry [54]. Biphasic methodologies heterogenize homogeneous catalysis and represent prime examples of the concept of Heterogreeneous Chemistry. Fluorous solvents are non-conventional solvents primarily used for biphasic catalysis, and are highly fluorinated alkanes, ethers and tertiary amines, with perfluorinated alkanes being the most representative. The term fluorous is analogous to aqueous and the concept of fluorous biphasic catalysis is illustrated by Horvath [55]. Fluorous solvents possess unusual physicochemical properties, such as low dielectric constants, high chemical and thermal stability and low toxicity. They commonly exhibit temperature-dependent miscibility with organic solvents and are immiscible with many common organic solvents at ambient temperature, although they can become miscible at elevated temperatures. This property is of particular relevance in biphasic catalysis since it provides the basis for performing biphasic catalysis or, alternatively, monophasic catalysis at elevated temperatures with biphasic product–catalyst separation at lower temperatures. The four classes of green solvents, as discussed above, span almost the entire range of the solvent spectrum in terms of their chemical and physical properties, which can be further tuned to fulfill the specific demands of a given synthetic task. Although their use is unlikely to solve all the solvent problems faced by industries, these solvents will contribute to the development of green and sustainable synthetic processes by allowing enhanced catalyst activity, selectivity and productivity, new selectivity patterns, reduced or eliminated waste and solvent emissions and ease of operation. The use of green solvents in green catalytic synthesis has been extensively reviewed by Liu and Xiao, who clearly illustrate the critical role that solvents play in greening synthetic chemistry and in particular catalytic organic synthesis [56]. 5.3.2 Solvent-free Heterogreeneous Chemistry
In 2000, Lippard provided a wish list identifying the grand challenges for chemistry that would affect a quiet revolution in chemical practice, on both industrial and laboratory scales [57]. This list includes the art of conducting chemical reactions without solvents. Although much attention has been given to the importance of green solvents, there is an equally desirable need to design reactions that proceed under totally solvent-free conditions. The best solvent is no solvent at all and, in the arena of Green Chemistry, solvent-free reactions give rise to major reduction and
5.4 Conclusion and Outlook
simplification. Green Chemistry as applied to chemical processes can be viewed as a series of reductions, such as reductions in energy, in auxiliaries and in waste. It should always lead to the simplification of the process in terms of the number of chemicals and steps involved. Simply put, removing the solvent factor from a chemical process will likely be the greatest reduction and simplification achievable, in many cases. The quest for solvent-free reactions is gaining momentum: Thomas et al. have developed more environmentally friendly and highly selective solvent-free alternatives for carrying out a number of important chemical conversions, such as selective oxidation of hydrocarbons and aromatics and industrial hydrogenations [58]. The procedures are based on porous heterogeneous catalysts in which the active sites have been atomically engineered. Such solid catalysts operate under solvent-free conditions and usually entail one-step processes. The concept of solvent-free heterogreeneous catalysis can be implemented in different ways. Bose et al., for example, described a general and practical green chemistry route to the Biginelli cyclocondensation reaction under solvent-free conditions [59]. Kotsuki et al. reported that the reaction of epoxides with lithium halides is efficiently promoted on the surface of silica gel in the absence of any solvent to give the corresponding b-halohydrins [60]. Mariani and co-workers used solid– liquid solvent-free phase transfer catalysis and acidic catalysis in dry media to design a solvent-free protocol for the synthesis of cosmetic fatty esters [61]. The necessity to minimize the amount of toxic waste and by-products from chemical processes has led to the development of new, more environmentally friendly synthetic methods in which fewer toxic substances are used, and has become one of key aspects of greening organic synthesis. Since solvents are generally used in large quantities and since many organic solvents are ecologically harmful, their use should be minimized to the greatest degree possible. Solventfree approaches to organic chemistry therefore provide an ideal green solution to the above referenced problems. Exemplary reactions have shown that solvent-free reactions proceed with at least the same yield and selectivity as their solution counterparts [62], catalysis being at the heart of these reactions. Solvent-free chemistry is truly an example of the concept of Heterogreeneous Chemistry.
5.4 Conclusion and Outlook
This brief review has illustrated how catalysis, and especially heterogeneous catalysis, constitutes one of the cornerstones of Green Chemistry. The heterogeneous nature of Green Chemistry is well underlined by the wide range of topics encircled in its domain, concepts ranging from employment of renewable energies and renewable feedstocks to implementation of chemical processes which maintain an ecologically friendly outlook. A broader perspective of the 12 Principles of Green Chemistry also indicates to us that Green Chemistry is more of a chemical philosophy than a chemical science.
j113
j 5 Heterogreeneous Chemistry
114
The chemical industry plays a key role in sustaining the world economy and in so doing underpins future technologies. Poliakoff et al. have explored some of the issues raised by the development of Green Chemistry techniques and have identified potential barriers to their implementation by industry [63]. They ask the question of what might be needed for the industry to embrace efforts to become greener. Economical barriers to the implementation of Green Chemistry necessitate a paradigm shift from the traditional concept of process efficiency, which focuses largely on chemical yield, to one that assigns economic value to eliminating waste at its source and avoiding the use of toxic and hazardous substances [45]. In this context, efforts are being made to make the philosophic concept of Green Chemistry more scientific and to introduce new concepts, such as atom efficiency [64], and measurable parameters, such as the as the E-factor [65], in order to provide a metric for Green Chemistry [66]. New strategies have been developed that allow chemists to assess clearly whether or not and to what extent chemistries and chemical processes can be considered as being green [67]. Aside from economic barriers, certain behavioral barriers exist to the implementation of Green Chemistry. These hurdles relate to the prevalent way of thinking about chemistry, which assesses value and confers weight to ongoing developments in chemistry. The concept of Heterogreeneous Chemistry elevates the 12 Principles of Green Chemistry to center field in Green Chemistry developments. It is holistic in nature and embraces all different green aspects of chemistry. Heterogreeneous Chemistry draws from the wealth of chemical knowledge and approaches old problems with well-established solutions from a different angle. It is an approach that focuses on the essential ideas and motivations behind the 12 Principle of Green Chemistry and it recognizes, for example, that technologies, for which questions of atom economy and E-factor cannot be answered, can still adhere to the 12 Principles of Green Chemistry and be truly heterogreeneous. Heterogreeneous Chemistry also necessitates a paradigm shift and recognizes that the question of the greenness of a chemical process cannot always be answered in terms of measurable parameters, just as the impact and value of a chemical discovery or new chemical procedure cannot necessarily be measured by its novelty appeal (such a paradigm shift is beginning to receive general recognition; the Journal of the American Chemical Society, for example, instructs its authors that titles of manuscripts may not contain the word First or Novel [68]). Re-evaluating chemical beliefs and chemical conceptions will be fundamental if we wish to conduct chemistry in a way that recognizes current challenges and expands views to incorporate inevitable consequences for future generations.
References 1 Anastas, P.T. (2003) Green Chemistry, 5, G29–G34. 2 http://nobelprize.org/nobel_prizes/ peace/laureates/2007/index.html.
3 Anastas, P.T. and Warner, J.C. (1998) Green Chemistry: Theory and Practice, Oxford University Press, Oxford.
References 4 Matlack, A.S. (2001) Introduction to Green Chemistry, Marcel Dekker, New York. 5 Gesellschaft Deutscher Chemiker. (2003) Green Chemistry – Nachhaltigkeit in der Chemie, Wiley-VCH Verlag GmbH, Weinheim. 6 Manahan, S.E. (2005) Green Chemistry and the Ten Commandments of Sustainability, ChemChar Research, Columbia, MO. 7 Anastas, P.T., Bartlett, L.B., Kirchhoff, M.M. and Williamson, T.C. (2000) Catalysis Today, 55, 11–22. 8 Anastas, P.T. and Kirchhoff, M.M. (2002) Accounts of Chemical Research, 35, 686–694. 9 Jacobsen, H. (2004) Angewandte ChemieInternational Edition, 43, 1912–1914. 10 Anastas, P.T., Kirchhoff, M.M. and Williamson, T.C. (2001) Applied Catalysis A-General, 221, 3–13. 11 Farrauto, R.J. and Heck, R.M. (2000) Catalysis Today, 55, 179–187. 12 Cortright, R.D., Davda, R.R. and Dumesic, J.A. (2002) Nature, 447, 964–967. 13 Rostrup-Nielsen, J.R. (2001) Physical Chemistry Chemical Physics, 3, 283–288. 14 Fu, Q., Saltsburg, H. and FlytzaniStephanopoulos, M. (2003) Science, 301, 935–938. 15 Davda, R.R. and Dumesic, J.A. (2003) Angewandte Chemie-International Edition, 42, 4068–4071. 16 Alcala, R., Mavrikakis, M. and Dumesic, J.A. (2003) Journal of Catalysis, 218, 178–190. 17 Gallezot, P., Nicolaus, N., Fleche, G., Fuertes, P. and Perrard, A. (1998) Journal of Catalysis, 180, 51–55. 18 Tronconi, E., Ferlazzo, N., Forzatti, P., Pasquon, I., Casale, B. and Marini, L. (1992) Chemical Engineering Science, 47, 2451–2456. 19 Huber, G.W., Shabaker, J.W. and Dumesic, J.A. (2003) Science, 300, 2075–2077. 20 Bull, T.E. (1999) Science, 285, 1209. 21 Kamm, B. and Kamm, M. (2007) Pure and Applied Chemistry, 79, 1983–1997. 22 Huber, G.W., Iborra, S. and Corma, A. (2006) Chemical Reviews, 106, 4044–4098.
23 Chheda, J.N., Huber, G.W. and Dumesic, J.A. (2007) Angewandte ChemieInternational Edition, 46, 7164–7183. 24 Huber, G.W., Cortright, R.D. and Dumesic, J.A. (2004) Angewandte ChemieInternational Edition, 43, 1549–1551. 25 Hara, M., Yoshida, T., Takagaki, A., Takata, T., Kondo, J.N., Haashi, S. and Domen, K. (2004) Angewandte Chemie-International Edition, 43, 2955–2958. 26 Kiss, A.A., Dimian, A.C. and Rothenberg, G. (2006) Advanced Synthesis and Catalysis, 348, 75–81. 27 Bournay, L., Casanave, D., Delfort, B., Hillion, G. and Chodorge, J.A. (2005) Catalysis Today, 106, 190–192. 28 Soares, R.R., Simonetti, D.A. and Dumesic, J.A. (2006) Angewandte ChemieInternational Edition, 45, 3982–3985. 29 Subramanian, R. and Schmidt, L.D. (2005) Angewandte Chemie-International Edition, 44, 302–305. 30 Kiss, A.A., Omota, F., Dimian, A.C. and Rothenberg, G. (2006) Topics in Catalysis, 40, 141–150. 31 Clark, J.H. and Macquarrie, D.J. (1998) Chemical Communications, 853–860. 32 Clark, J.H., Butterworth, A.J., Tavener, S.J., Teasdale, A.J., Barlow, S.J., Bastock, T.W. and Martin, K. (1997) Journal of Chemical Technology and Biotechnology (Oxford, Oxfordshire: 1986), 68, 367–376. 33 Yamaguchi, K. and Mizuno, N. (2003) Angewandte Chemie-International Edition, 42, 1480–1483. 34 Kasai, J., Nakagawa, Y., Uchida, S., Yamaguchi, K. and Mizuno, N. (2006) Chemistry – A European Journal, 12, 4176–4184. 35 Dabiri, M., Salehi, P., Baghbanzadeh, M., Zolfigol, M.A. and Bahramnejad, M. (2007) Synthetic Communications, 37, 1201–1209. 36 Kotani, M., Koike, T., Yamaguchi, K. and Mizuno, N. (2006) Green Chemistry, 8, 735–741. 37 Gupta, R., Paul, S. and Gupta, R. (2007) Journal of Molecular Catalysis A-Chemical, 266, 50–54.
j115
j 5 Heterogreeneous Chemistry
116
38 Gladysz, J.A. (2001) Pure and Applied Chemistry, 73, 1319–1324. 39 Dioumaev, V.K. and Bullock, R.M. (2003) Nature, 424, 530–532. 40 Broadwater, S.J., Roth, S.L., Price, K.E., Kobaslija, M. and McQuade, D.T. (2005) Organic and Biomolecular Chemistry, 3, 2899–2906. 41 Fujiwara, H., Ogasawara, Y., Yamaguchi, K. and Mizuno, N. (2007) Angewandte Chemie-International Edition, 46, 5202–5205. 42 Tang, J.W., Zou, Z.G. and Ye, J.H. (2004) Angewandte Chemie-International Edition, 43, 4463–4466. 43 Tsuji, I., Kato, H. and Kudo, A. (2005) Angewandte Chemie-International Edition, 44, 3565–3568. 44 Zheng, N., Bu, X.H., Vu, H. and Feng, P.Y. (2005) Angewandte Chemie-International Edition, 44, 5299–5303. 45 Sheldon, R.A. (2005) Green Chemistry, 7, 267–278. 46 Butters, M., Catterick, D., Craig, A., Curzons, A., Dale, D., Gillmore, A., Green, S.P., Marziano, I., Sherlock, J.-P. and White, W. (2006) Chemical Reviews, 106, 3002–3027. 47 Cornils, B. and Wiebus, E. (1996) Recueil des Travaux Chimiques des Pays-Bas, 115, 211–215. 48 Li, C.J. (1996) Tetrahedron, 52, 5643–5668. 49 Li, C.J. (2005) Chemical Reviews, 105, 3095–3165. 50 Budarin, V., Luque, R., Macquarrie, D.J. and Clark, J.H. (2007) Chemistry – A European Journal, 13, 6914–6919. 51 Leitner, W. (2002) Accounts of Chemical Research, 35, 746–756. 52 Grunwaldt, J.-D., Wandeler, R. and Baiker, A. (2003) Catalysis Reviews-Science and Engineering, 45, 1–96.
53 Danciu, T., Beckman, E.J., Hancu, D., Cochran, R.N., Grey, R., Hajnik, G.M. and Jewson, J. (2003) Angewandte ChemieInternational Edition, 42, 1140–1142. 54 Sheldon, R. (2001) Chemical Communications, 2399–2407. 55 Horvath, I.T. (1998) Accounts of Chemical Research, 31, 641–650. 56 Liu, S.F. and Xiao, J.L. (2007) Journal of Molecular Catalysis A-Chemical, 270, 1–43. 57 Lippard, S.J. (2000) Chemical & Engineering News, 78, 64–65. 58 Thomas, J.M., Raja, R., Sankar, G., Johnson, B.F.G. and Lewis, D.W. (2001) Chemistry – A European Journal, 7, 2973–2978. 59 Bose, D.S., Fatima, L. and Mereyala, H.B. (2003) The Journal of Organic Chemistry, 68, 587–590. 60 Kotsuki, H., Shimanouchi, T., Ohshima, R. and Fujiwara, S. (1998) Tetrahedron, 54, 2709–2722. 61 Villa, C., Mariani, E., Loupy, A., Grippo, C., Grossi, G.C. and Bargagna, A. (2003) Green Chemistry, 5, 623–626. 62 Metzger, J.O. (1998) Angewandte ChemieInternational Edition, 37, 2975–2978. 63 Poliakoff, M., Fitzpatrick, J.M., Farren, T.R. and Anastas, P.T. (2002) Science, 297, 807–810. 64 Trost, B.M. (2002) Accounts of Chemical Research, 35, 695–705. 65 Sheldon, R.A. (2007) Green Chemistry, 9, 1273–1283. 66 Curzons, A.D., Constable, D.J.C., Mortimer, D.N. and Cunningham, V.L. (2002) Green Chemistry, 4, 521–527. 67 Constable, D.J.C., Curzons, A.D. and Cunningham, V.L. (2001) Green Chemistry, 3, 1–6. 68 http://pubs.acs.org/paragonplus/ submission/jacsat/jacsat_authguide.pdf.
j117
6 Single-site Heterogeneous Catalysts via Surface-bound Organometallic and Inorganic Complexes Christophe Coperet
6.1 Introduction
Heterogeneous catalysis fully complies with the aims and the rules of Green Chemistry because it associates catalysis – a way to shorten steps and to improve the selectivity and the energy efficiency of reactions (lower temperatures) – and heterogeneous phases, which allow an easier separation of products from the catalyst, greatly simplifying chemical processes (number of separation steps) and avoiding the contamination of products with residual metals. In fact, industry typically prefers to implement these types of processes. However, the rational optimization of these systems from a chemical point of view is more difficult, because of their complexity and the associated lack of molecular understanding of the nature and the environment of their active sites. In contrast, homogeneous catalysts can be developed more rationally through structure–reactivity relationships, which is due to the large body of data obtained in the past 40 years in molecular chemistry. In more recent years, the same molecular approach has been undertaken in heterogeneous catalysis [1–6], and in this chapter the focus will be specifically on the development of single-site heterogeneous catalysts (for generalities, see the next section) and their applications to selective chemical reactions such as hydrogenation and hydrosilylation (Section 6.3), alkene metathesis and other alkene homologation processes (Section 6.4), alkyne metathesis and other alkyne homologation processes (Section 6.5), Lewis acid-catalyzed reactions (Section 6.6), oxidation (Section 6.7) and alkane homologation processes (Section 6.8).
6.2 Generalities
Numerous heterogeneous catalysts are supported on oxides and the active sites are directly bound to their surfaces. A molecular approach to the preparation and development ofthesecatalysts dependslargelyonobtaininga molecular understanding of the structure of the surface species and relies on characterizing them at a molecular
Handbook of Green Chemistry, Volume 2: Heterogeneous Catalysis. Edited by Robert H. Crabtree Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32497-2
j 6 Single-site Heterogeneous Catalysts via Surface-bound Organometallic and Inorganic Complexes
118
level using a combination of tools: chemical reactivity, spectroscopy (IR, Raman, EXAFS–XANES, solid-state NMR, UV–visible) and computational chemistry [5]. The first approach to the preparation of these systems consists in the controlled reaction (grafting) of organometallic or metallo-organic complexes with the functionalities present at the surface of an oxide support, typically hydroxyls (Scheme 6.1a), but also reactive oxygen atoms (Scheme 6.1b). It is therefore critical to have a good understanding of the surface properties and the stability of the oxide supports. For instance, the density of active sites and their environment can be controlled by a thermal treatment of the support under vacuum or a flow of dry inert gas prior to the grafting step. This treatment induces a condensation of adjacent hydroxyl groups, which tunes the density of OH groups (Scheme 6.1c). Prior to this thermal treatment, it is often preferable to perform a calcination of the support to remove adsorbed carbon-containing species. After chemical grafting on to the support, these systems can also be further treated with chemical, thermal and/or photochemical processes (Scheme 6.1d). This typically increases the number of covalent bonds between the metals and the support and can yield isolated surface species, having structures unprecedented in molecular chemistry. This also allows an increase stability and reactivity of active sites, which is translated into Greener chemistry, i.e. enhanced catalytic activities and stabilities as well as the discovery of new reactivities (see below).
Scheme 6.1
In the case of amorphous silica, the OH groups are attached to tetracoordinated Si atoms and the OH density upon thermal treatment varies as follows: ca 2.6, 1.2 and 0.7 OH nm2 at 200, 500 and 700 C, respectively. Above 800 C, the OH density still decreases, but there is also a dramatic decrease in surface area (sintering). The OH groups are mostly isolated and statistically distributed at the surface of silica treated at 700 C [SiO2-(700)] and at lower temperatures the amount of vicinal silanols increases; geminal silanols are proposed to exist only below 200 C (Scheme 6.2a) [7, 8]. The use of
6.3 Hydrogenation and Hydrosilylation
mesoporous silica supports does not dramatically influence the overall trends presented above: the OH density – expressed in OH nm–2 differ only slightly, but the OH concentration in mmol g–1 increases nearly proportionallywith increase in surface area, which allows the metal loading to be increased. Note that for these supports care should be taken because of their lower thermal stability (the mesoporous structure may collapse). In the case of alumina, the situation is more complex: the density is ca 4 OH nm2 fora g-aluminapretreatedat500 C [g-Al2O3-(500)],their distributionis not even [9] and the hydroxyls have very different natures depending on whether they are attached to a tetra-, penta- or hexacoordinated Al atoms and also if they are bound to one (h1), two (h2) or three (h3) Al (Scheme 6.2b) [10–12]. Moreover, the surface of g-Al2O3 also contains different types of Lewis acid sites, which can also react with the molecular precursors during the grafting step (see below) [13]. Each support has its own characteristic and should be studied in detail – typically by the combined use of titration studies, spectroscopy and modeling – prior to investigating grafted species [14].
Scheme 6.2 Surface functionalities on (a) silica and (b, c) alumina [(b) hydroxyls; (c) Lewis acid sites].
6.3 Hydrogenation and Hydrosilylation 6.3.1 Hydrogenation
In contrast to their homogeneous catalyst homologues, single-site heterogeneous hydrogenation catalysts are based on early transition metals, mainly Groups 4–5,
j119
j 6 Single-site Heterogeneous Catalysts via Surface-bound Organometallic and Inorganic Complexes
120
because of their stronger MO bonds, which prevents sintering and formation of metal particles. These systems are prepared by first reaction of perhydrocarbyl organometallic complexes with silica followed by treatment under H2, at 150–200 C. In the case of Group 4 and 5 metals supported on silica, this method yields surface metal hydrides: Ti [TiH/SiO2] [15], Zr [ZrH/SiO2] [16, 17], Hf[HfH/ SiO2] [18] and Ta [TaH/SiO2] [19] (Scheme 6.3). These hydrides can be prepared on a variety of oxides supports and extended to Group 6 metals: [ZrH/Al2O3] [20], [ZrH/SiO2–Al2O3] [21], [TaH/Al2O3] [22], [WH/Al2O3] [22, 23], [WH/SiO2– Al2O3] [24], but their structures have not yet been fully elucidated.
Scheme 6.3 (a) Silica-supported group 4 hydrides; (b) silica-supported tantalum hydrides.
Noteworthily, the silica-supported Ti and Zr hydrides efficiently catalyze the hydrogenation of alkenes and even aromatics (TOF ¼ 3600–360 000 h1 for the hydrogenation of cyclohexene or benzene) [25–28]. Such types of catalysts have also been prepared in situ by directly reacting the molecular precursor with silica in an autoclave prior to adding the reagent and H2, and they display good catalytic performances on various aromatic substrates (Tables 6.1 and 6.2) [29]. Cyclopentadienyl Zr derivatives supported on alumina are also highly active alkene hydrogenation catalysts [30, 31], and the activity of these hydrogenation catalysts depends on both the molecular precursor and the temperature of thermal treatment of alumina (Table 6.3, Entries 1–6) [32]. The best systems are obtained for highly dehydroxylated alumina [Al2O3-(1000)] and this has been associated with the formation of cationic surface species. Therefore, other supports have been investigated in order to generate more electrophilic Zr systems, such as sulfated zirconia [33], sulfated alumina[34] or other sulfated oxide supports (Table 6.3, Entries 7–11) [35]. In these
6.3 Hydrogenation and Hydrosilylation Table 6.1 Hydrogenation of benzene catalyzed by silica-supported
Group 4–6 transition metal complexes. Catalytic system
Conditionsa
Time (min)
Conversion (%)
Ti(CH2SiMe3)4/SiO2 Zr(CH2Ph)4/SiO2 Hf(CH2Ph)4/SiO2 Ta(CH2Ar)5/SiO2 (Ar ¼ 4-MeC6H4) [Nb(m-CSiMe3)(CH2SiMe3)2]2/SiO2
A A A A A B A B B
360 240 600 24h 300 25 720 55 720
39 100 100 100 100 100 100 100 100
[Ta(m-CSiMe3)(CH2SiMe3)2]2/SiO2 [Mo2(CH2SiMe3)6]/SiO2
Conditions: A, 0.056 mmol of M loaded on 5 g of silica treated at 200 C under vacuum, 17.5 g of C6H6 (4000 equiv.), 80–100 bar of H2 at 120 C; B, 0.056 mmol of M loaded on 5 g of silica treated at 200 C under vacuum, 26 mmol of substrate (280 equiv.), 80–100 bar of H2 at 120 C.
a
Table 6.2 Hydrogenation of other aromatics catalyzed by silica supported Group 4–6 transition metal complexes.
Catalytic system
Substrate
Time (min) Conversion (%) Selectivity (%)
Ti(CH2SiMe3)4/SiO2 Zr(CH2Ph)4/SiO2 Hf(CH2Ph)4/SiO2 Ta(CH2Ar)5/SiO2 (Ar ¼ 4-MeC6H4) [Nb(m-CSiMe3)(CH2SiMe3)2]2/SiO2
Naphthalenea Naphthalenea Naphthalenea Naphthalenea Naphthalenea Tolueneb o-Xyleneb m-Xyleneb p-Xyleneb Naphthalenea Tolueneb o-Xyleneb m-Xyleneb p-Xyleneb Naphthalenea
150 30 60 160 20 33 30 20 25 70 64 75 70 65 70
[Ta(m-CSiMe3)(CH2SiMe3)2]2/SiO2
[Mo2(CH2SiMe3)6]/SiO2
100 100 100 100 100 100 100 100 100 100 100 100 100 100 100
55 : 45c 25 : 75c 19 : 81c 20 : 80c 15 : 85c 100 57 : 43c 78 : 22c 57 : 43c 20 : 80c 100 91 : 9c 92 : 8c 82 : 18c 20 : 80c
0.056 mmol of M loaded on 5 g of silica treated at 200 C under vacuum, 2.0 g of naphthalene (280 equiv.), 25 mL of hexane, 100 bar of H2 at 120 C. b 0.056 mmol of M loaded on 5 g of silica treated at 200 C under vacuum, 26 mmol of substrate (280 equiv.), 80–100 bar of H2 at 120 C. c cis : trans ratio. a
cases, the activities have been correlated with the surface Brønsted acidity of the support [35], but the structure of the active sites is not yet understood. This approach can also be applied to actinide complexes, which, when supported on alumina, display much higher activity than the original molecular complexes in
j121
j 6 Single-site Heterogeneous Catalysts via Surface-bound Organometallic and Inorganic Complexes
122
Table 6.3 Activity of cyclopentadienylzirconium derivatives in the hydrogenation of alkenes.
Entry
Molecular precursor
Support
Substrate
Activity (h1)
1 2 3 4 5 6 7 8 9 10 11
Cp ZrMe3 Cp ZrMe3 Cp2ZrMe2 Cp CpZrMe2 Cp2ZrMe2 Cp CpZrMe2 Cp ZrMe3 Cp ZrMe3 Cp ZrMe3 Cp ZrMe3 Cp ZrMe3
Al2O3-(1000) Al2O3-(500) Al2O3-(1000) Al2O3-(1000) Al2O3-(500) Al2O3-(500) ZrO2-sulfated Al2O3-sulfated SnO2-sulfated SnO2-sulfated TiO2-sulfated
Propene Propene Propene Propene Propene Propene 1-Hexene 1-Hexene 1-Hexene 1-Hexene 1-Hexene
3960a 1080a 1080a 720a 360a 216a 970b 360b 10b 5b <3b
Activities measured at 63 C. Activities measured at 25 C.
a b
the hydrogenation of propene at 25 C (Table 6.4, Entries 1–4) [31, 36]. Using silica or alumina dehydroxylated at lower temperatures yields poorer catalysts [37]. Other supports have been used, but highly dehydroxylated alumina remains the best support (Table 6.4, Entries 3 and 5–7) [38]. The initial activity also depends on the alkenes as measured by the hydrogenation at 45 C: cis-2-butene (3960 h1) > trans2-butene (2900 h1) propene (470 h1) isobutene (13 h1). Finally, aromatics can also be efficiently hydrogenated at 90 C and 13 bar of H2 with the following order of reactivity: [Th(allyl)4]/Al2O3-(1000) (1970 h1) [Th (CH2Ar)4]/Al2O3-(1000) (825 h1) > [Cp Th(CH2Ar)3] (765 h1) [39, 40]. The rate of hydrogenation also depends on the aromatic compounds: benzene (6850k) > toluene (4100k) > xylene (1450k) naphthalene (k). Note that the activities observed with [Th(allyl)4]/Al2O3-(1000) are comparable to those obtained for classical heterogeneous catalysts based on supported Rh or Pt particles.
Table 6.4 Activity of cyclopentadienyl actinide derivatives supported or not in the hydrogenation of propene at 25 Ca.
Entry
Molecular precursor
Support
Activity (h1)
1 2 3 4 5 6 7
Cp 2UMe2 Cp 2UMe2 Cp 2ThMe2 Cp 2ThMe2 Cp 2ThMe2 Cp 2ThMe2 Cp 2ThMe2
Al2O3-(1000) None Al2O3-(1000) None SiO2–Al2O3 MgCl2 SiO2–MgO
1080 68 580 0.54 160–230 25–43 0
6.3 Hydrogenation and Hydrosilylation
6.3.2 Hydrosilylation
The silica-supported lanthanide and Group 3 silylamide complexes [(SiO)Ln(N {SiMe3}2)2] (Ln ¼ Y, La, Nd and Sm), prepared by the reaction of the corresponding trisilylamide complexes [Ln(N{SiMe3}2)3] with SiO2-(700), catalyze the hydrosilylation of alkenes (Scheme 6.4). Whereas 1-hexene is transformed mainly into the linear isomer with 90–94% selectivity, styrene gives the branched product in 99% selectively [41]. Overall, the silica-supported systems are only slightly less active than the parent molecular complexes, but the trends of reactivity over a series of lanthanide metals are different: La > Nd > Sm Y for the supported systems vs La Sm > Nd Y for the molecular precursors. This has been associated with the impossibility to generate dimeric species for the supported systems.
Scheme 6.4
Hydrosilylation has also been catalyzed by a well-defined Rh supported complex [(SiO)Rh(COD)L], which is prepared by the reaction of [{(Me3SiO)Rh(COD)}2] with SiO2-(200) (Scheme 6.5) [42]. In this specific case, this catalytic system is noteworthy: the catalyst loading is fairly low (0.01%) and recycling is fairly efficient with possibility of 10–20 cycles without major loss of activity and metal leaching. Moreover, spectroscopic evidence has been obtained for the intermediate silyl hydride complex, expected from the classical Chalk–Harrod mechanism of hydrosilylation. This is again consistent with the greater stability of the supported system.
j123
j 6 Single-site Heterogeneous Catalysts via Surface-bound Organometallic and Inorganic Complexes
124
Scheme 6.5 (a) Preparation of well-defined RhI species; (b) hydrosilylation with [(SiO)RhI(COD)].
6.4 Metathesis and Homologation Processes of Alkenes 6.4.1 Alkene Metathesis 6.4.1.1 Silica-supported Catalysts The preparation of heterogeneous alkene metathesis catalysts by grafting molecular complexes on oxide supports started in the 1970s [43–47], but it has only been recently that the first well-defined silica supported metallocarbene complexes have appeared: they are all prepared by the grafting of well-defined Mo [48–50], W [51] and Re [52–57] alkylidene complexes on silica via cleavage of one MX bond (X ¼ alkyl or amido) by the isolated silanols of a surface of SiO2-(700) yielding the surface complex and HX (Scheme 6.6). By combining reactivity studies, spectroscopy (IR and solid-state NMR and also EXAFS in some cases) and computational studies, it has been possible to determine their structure at a molecular level: they are all isoelectronic d0 tetrahedral syn-complexes, displaying an agostic interaction between the alkylidene proton and the metal center. Their catalytic properties in alkene metathesis are noteworthy (Table 6.5, Entries 1–6), because of: Their greater activities and stabilities than their homogeneous precursors. Their compatibility with functionalized alkenes such as esters without the use of co-catalysts, which is different from the classical heterogeneous alkene metathesis catalysts [58].
6.4 Metathesis and Homologation Processes of Alkenes
Scheme 6.6
The nearly quantitative formation of the cross-metathesis products (initiation step), which is consistent with single-site catalysts. Their improved stability because deactivation pathways via dimerization have been shut down (site isolation) [59–62]. Their high metathesis selectivity for amido systems; the alkyl ones leading to the formation of some amount of 1-butene, especially for the Re systems [63]. The origin of the high reactivity of these silica-supported systems (X)(Y)M(ER) (¼CHR) is due to the asymmetry at the metal center (X6¼Y) [64, 65]. Indeed, alkene metathesis is a four-step reaction pathway involving the coordination of the alkene,
j125
j 6 Single-site Heterogeneous Catalysts via Surface-bound Organometallic and Inorganic Complexes
126
Table 6.5 Activity of well-defined silica supported alkylidene complexes in alkene metathesis.
Propeneb
Ethyl oleate (EO)c
Entry
Catalysta
Initial activityd
Selectivitye
TON
Initial activityd
Time (h)
1 2 3 4 5 6 7 8
Re–R Mo–R W–R Mo–NPh2 Mo–Pyr–Ar Mo–2,5-DiMePyr–Ar Mo–2,5-DiMePyr–CF3 Mo–2,5-DiMePyr–Ad
120 120 8.4 374 362 320 560 780
96.0 99.4 99.6 >99.9 >99.9 >99.9 >99.9 >99.9
6 · 103 22 · 103 6 · 103 138 · 103 62 · 103 101 · 103 135 · 103 275 · 103
1.2 1.2 — 6.4 2.4 3.0 72.0 30.0
24 (30%) 24 — 4 24 (10%) 3 0.5 1
a
For structures, see Scheme 6.6. Experimental conditions: in a flow reactor with a propene flow rate of ca 5000 mol mol1 M min1. c EO/M ¼ 2000, 1.16 M solution of EO in toluene. d The initial activity is defined as the moles of substrate transformed per mole of catalyst per minute. e Selectivity ¼ (Z þ E)-2-butenes/(all butenes). b
the [2 þ 2]-cycloaddition and the corresponding reverse steps (cycloreversion and decoordination), with metallacyclobutane intermediates having trigonal bipyramid (TBP) and square-based pyramid (SP) geometries, the former being on the reaction pathway and the latter being more stable (Scheme 6.7). The first step corresponds to a distortion of the complex from a tetrahedral into a trigonal pyramid structure, in order to generate an empty coordination site and to accommodate the incoming alkenes, and it is favored when the ligand trans to the incoming alkene is a strong s-donor ligand (X ¼ CH2R or NR2), whereas that entering the basal plane of the trigonal pyramid is a weak s-donor ligand (Y ¼ OSi) to avoid the strong competition of the E- and the alkylidene ligands. The second step, the [2 þ 2]-cycloaddition, has a very low activation energy and leads to very stable metallacyclobutane intermediates, which are destabilized by strong s-donor ligands. Therefore, the asymmetric systems (X ¼ CH2R or NR2 and Y ¼ OSi) associates low activation energies and not too stable intermediates, which is optimal for a catalyst.
Scheme 6.7
6.4 Metathesis and Homologation Processes of Alkenes
Moreover, the Mo systems can be further improved by tuning the imido ligands and, for instance, in the case of the 2,5-dimethylpyrrolyl systems there is a clear increase in activity and overall turnover by replacing the 2,6-diisopropylphenylimido by either a 2-CF3-phenyl or an adamantyl imido ligand (Table 6.5, Entries 6–8). The adamantylimido ligand allows high activity (ca 8 mol of propene converted per Mo per second) and overall TON to be reached (230 000) in the metathesis of propene [50], whereas the 2-CF3-phenyl group gives better stability in the metathesis of ethyl oleate. 6.4.1.2 Alumina-supported Catalysts In contrast to the silica-supported systems, obtaining a molecular understanding of the alumina-supported systems is more difficult, due to the complexity of the alumina support (see Section 6.2). In fact, despite extensive studies, Re2O7/Al2O3, a promising catalyst because of its unusual activity at room temperatures [66] and its compatibility with functional groups when activated with organotin reagents [67–69], there is still little information on the nature of its active sites, even though it is clear that the Lewis acidity of alumina is key [70]. This is also true for the oxide-supported methyltrioxorhenium, CH3ReO3 [71–75]. Combining mass balance analysis, in situ IR, EXAFS, solid-state NMR and periodic calculations shows that the Re–C bond is not cleaved upon grafting of CH3ReO3 on g-Al2O3 partially dehydroxylated at 500 C, but that several surface species are formed: (1) the major surface species (1.0 nm2) correspond to CH3ReO3 chemisorbed on aluminum Lewis acid sites through its oxo ligands and (2) the minor surface species (0.15 Re nm2) are formed via the heterolytic cleavage of the CH bond of CH3ReO3 on reactive surface sites, AlSOS [13], yielding AlSCH2ReO3 surface species along with new hydroxyls (Scheme 6.8) [76]. These reactive sites are in fact generated upon thermal treatment of alumina (calcination and partial dehydroxylation typically performed at 500 C) prior to grafting of CH3ReO3 and correspond to Al sites having an empty coordination site (Lewis acid), the reactivity of which depends on the coordination number of Al: (OS)3AlS (AlIII) > (OS)4AlS (AlIV) > (OS)5AlS(AlV) [13]. Of the various possible AlSCH2ReO3 surface species, titration studies in combination with solid-state NMR data and calculations are consistent with the formation of mainly AlVCH2ReO3 or AlVICH2ReO3 surface species.
Scheme 6.8
Noteworthily, in the metathesis of propene, this catalyst deactivates first order in products (very likely ethene) [77], and it has been possible to improve the catalytic stability of the system by changing the adsorption properties of alumina by passivating
j127
j 6 Single-site Heterogeneous Catalysts via Surface-bound Organometallic and Inorganic Complexes
128
the surface with trimethylsilyl group (Scheme 6.9) [78]. Moreover, this change in surface adsorption properties also allows the more selective formation of (Z)-2-butene (Z/E ¼ 3), the kinetic product, even at 10% conversion, whereas the thermodynamic ratio is typically observed (E/Z ¼ 3) in the case of the parent alumina-supported systems.
Scheme 6.9
Finally, the presence of masked carbenic active sites in CH3ReO3/Al2O3 probably explains why this catalyst is compatible with functionalized alkenes, whereas the parent system based on Re2O7/Al2O3 necessitates activators. In fact, as the active sites of Re2O7/Al2O3 require the presence of Lewis acid sites, they are probably poisoned in the presence of functionalized alkenes and it is likely that an activator, typically alkylating agents such as R4Sn, for Re2O7/Al2O3 is necessary to generate active sites, closely related to AlSCH2ReO3. 6.4.2 Other Alkene Homologation Processes
The field of single-site heterogeneous catalyst has probably emerged from the polymer industry, where chemists had been trying to develop well-defined equivalents of the Ziegler–Natta catalysts [1, 2]. In fact, the silica-supported hydrides were discovered by Yermakov et al. [2] within this context and they display good polymerization activities [79–83]. This area of research has already been extensively reviewed and is not directly related to green chemistry and will therefore not been discussed here. This section will be solely devoted to homologation or cyclization processes. 6.4.2.1 Direct Conversion of Ethene into Propene Following the discovery of the alumina-supported tungsten hydrides, an alkane metathesis catalyst (see below) [23, 83], it has been shown that this system transforms ethene directly into propene at 150 C with 95% selectivity (Scheme 6.10) [84]. Kinetic studies have shown that in fact this single-site catalyst performs three reactions successively: step 1, dimerization of ethene into butenes (mainly 1-butene); step 2, isomerization of 1-butene into 2-butene; and step 3, cross-metathesis of 2-butene with ethene yielding propene.
6.5 Metathesis, Dimerization, Trimerization and Other Reactions Involving Alkynes
Scheme 6.10
6.4.2.2 Cyclization of Dienes The silica-supported zirconium hydrides (see Scheme 6.2) also catalyzes carbon–carbon bond-forming reactions, converting dienes into cyclic products (Scheme 6.11) [26]. For instance, 1,5-hexadiene is converted into 2-methylcyclopentene via the successive insertion of the diene into the hydride followed by cyclization and b-H elimination releasing methylenecyclopentane, which is then isomerized into the final product.
Scheme 6.11
6.5 Metathesis, Dimerization, Trimerization and Other Reactions Involving Alkynes 6.5.1 Alkyne Metathesis
Alkyne metathesis is analogous to alkene metathesis, but requires alkylidyne propagating species (Scheme 6.12a) [85]. In fact, the Re-based silica-supported
j129
j 6 Single-site Heterogeneous Catalysts via Surface-bound Organometallic and Inorganic Complexes
130
catalyst presented in Section 6.3, which contains an alkylidyne ligand, also catalyzes the metathesis of 2-pentyne into 2-butyne and 3-hexyne, albeit with low turnovers (Scheme 6.12b) [52]. Recently, Weissman et al. reported that [(SiO)Mo(CEt) (NtBuPh)2], prepared by grafting [Mo(CEt)(NtBuPh)3] on silica, is a highly active catalyst in alkyne metathesis (Scheme 6.12c) [86]. Grafting the dinuclear amido W complex [(Me2N)3WW(NMe2)3] on SiO2-(700) also generates a well-defined system [(SiO)(Me2N)2WW(NMe2)3], but it is only poorly active in alkyne metathesis (Scheme 6.12d). However, upon addition of 5 equiv. of tBuOH, this material catalyzes this reaction very efficiently, converting 50 equiv. of 4-nonyne in 30 min, in comparison with the 20 min required for [(tBuO)3WCtBu] [87].
Scheme 6.12
6.5.2 Dimerization and Trimerization of Alkynes
Dimerization of alkynes yielding dienes can be catalyzed by the silica-supported silylamide lanthanide complexes [(SiO)Ln{N(SiMe3)2}2]. These systems are slightly less active than their homogeneous precursors (Scheme 6.13a) [41, 88]. Noteworthily, the lanthanum and neodymium surface species are more selective for dimerization than oligomerization and, in contrast to the homogeneous system, head-to-head dimerization is favored. Finally, the catalytic trimerization of alkynes into aromatics has been reported using silica-supported zirconium hydrides (Scheme 6.13a) [26].
6.6 Lewis Acid-catalyzed Reactions
Scheme 6.13 (a) Dimerization of phenylacetylene on silicasupported lanthanide and Group 3 amido complexes (selectivity in dimer); (b) trimerization of butylacetylene on silica-supported zirconium hydrides.
6.5.3 Hydroamination of Alkynes
Well-defined PdII surface species, cis- and trans-[(SiO)Pd(X)L2], are prepared by reaction of the corresponding methyl molecular complexes [L2Pd(X)(Me)] {X ¼ Cl, OTf and NO3; L ¼ PMe3, PPh3, dmpe and dppe} with SiO2-(500) (Scheme 6.14). These systems catalyze the cyclic hydroamination of g-aminoalkynes and the best catalyst is trans-[(SiO)Pd(NO3)L2], which can be recycled up to three times using ca 1 mol% of Pd [89, 90].
6.6 Lewis Acid-catalyzed Reactions 6.6.1 Silica-supported Group 4 Metals
Well-defined supported Group 4 alkoxides based on Ti or Zr are readily available by several routes. First, the monosiloxy derivatives are typically prepared with silica partially dehydroxylated above 500 C (Scheme 6.15a). Well-defined mononuclear systems have been obtained by grafting tetrakisalkyl [91–93], amido [94] or siloxide [95] complexes on silica followed by an alcoholysis step for the amido or alkyl surface
j131
j 6 Single-site Heterogeneous Catalysts via Surface-bound Organometallic and Inorganic Complexes
132
Scheme 6.14
complexes. Note that grafting directly alkoxide derivatives generates polynuclear surface species [94]. The preparation of bis-siloxy systems is carried out using the same strategy (Scheme 6.15b), but using a silica dehydroxylated at lower temperatures, i.e. 200–300 C [94, 96], or a mesoporous silica, which generates the bis-siloxy system even when the pre-treatment of silica was 500 C [97, 98]. For tris-siloxy systems, this requires a thermal treatment and can be performed via two strategies (Scheme 6.15c): (1) formation of the hydrides followed by treatment with various reagents such as N2O [17], H2O [93], ROH [91–93], CO2 [17] or acetylacetone [99], and (2) calcination, generating isolated sites (Scheme 6.15d). Their Lewis acid properties have been exploited to catalyze a variety of reactions: the transfer hydrogenation of ketones (Equation 6.1), the transesterification of esters (Equation 6.2) and the selective epoxidation of alkenes by alkyl peroxide or hydrogen peroxide (see Sections 6.7.1 and 6.7.2). ð6:1Þ
ð6:2Þ
6.6 Lewis Acid-catalyzed Reactions
Scheme 6.15
6.6.1.1 Reduction of Ketones Through Hydrogen Transfer Cyclohexanone is reduced selectively to cyclohexanol by 2-propanol in the presence of a catalytic amount of [(SiO)M(OiPr)3] (Meerwein–Ponndorf–Verley reaction) with the order of reactivity Hf (73%) > Zr (50%) Ti (0%), whereas 4-methyl-2-pentanone is not converted with any of the catalysts [91, 93]. Similarly, benzaldehyde is reduced to benzyl alcohol with the same order of reactivity: Hf (90%) > Zr (65%) Ti (0%). Noteworthily, whereas Ti leaches from the silica
j133
j 6 Single-site Heterogeneous Catalysts via Surface-bound Organometallic and Inorganic Complexes
134
support, no leaching has been observed with Zr and Hf. In fact, both catalysts can be recycled even if slight deactivation has been observed. In the case of Zr, it was clearly shown that the Zr(OiPr)4 does not catalyze this reaction and that preparation of the catalyst [(SiO)Zr(OiPr)3] via [(SiO)Zr(CH2tBu)3] followed by treatment with iPrOH provides a more stable catalyst than by direct grafting of Zr(OiPr)4, and this has been associated with the formation of polynuclear species in the latter case. In the case of Hf, it was shown that water had no detrimental effect on the catalytic behavior of [(SiO)Hf(OiPr)3] and in fact [(SiO)Hf(OH)3], prepared by the controlled hydrolysis of [(SiO)Hf(CH2tBu)3], shows similar catalytic performance. 6.6.1.2 Transesterification of Esters The silica-supported complexes [(SiO)Zr(acac)3] and [(SiO)3Zr(acac)] also display high activities in the transesterification of methyl methacrylate, albeit lower than that of the parent molecular system [Zr(acac)4] [99, 100]. Moreover, while leaching is observed for the mono-grafted species, the tris-grafted species is stable under the reaction conditions. In the latter case, it has been shown that the loss of activity is due to the build-up of the corresponding butanolate derivatives [(SiO)3Zr(OBu)], which can be reactivated by the addition of acetylacetone. 6.6.2 Silica-supported Group 3 Metals and Lanthanides
Group 3 and lanthanide metals are well known for their Lewis acid properties and their associated catalytic properties. Therefore, several materials have been prepared by grafting various sources of lanthanide derivatives on mesoporous silica (MCM-41) as the silicic support [101, 102]. First, the reaction of the silyl amide derivatives [Ln{N(SiR2H)2}3(thf)2] with MCM-41 yields [(SiO)xLn{N (SiR2H)2}3–x(thf)y] (Ln ¼ Sc, Y, Nd, La; x ¼ 1 or 2) [103–106], which, upon treatment with alcohol or fodh, gives [(SiO)xLn(OR)3–x(thf)y] or [(SiO)xLn(fod)3–x(thf)y], respectively (Scheme 6.16a). In these cases, note that the surface is also covered by (Me2SiHO) groups, because of the competitive subsequent reaction with surface
Scheme 6.16
6.7 Oxidation
silanols of HN(SiR2H)2, liberated during grafting. In contrast, the reaction of (fod)3Y with MCM-41 gives [(SiO)Y(fod)2] as the sole surface species (Scheme 6.16b) [107]. The surface alkoxide species display good catalytic activities for the transfer hydrogenation reaction (Equation 6.1) [108]. In the case of the fod derivatives, they have been used to catalyze selectively the hetero Diels–Alder reactions of benzaldehyde with trans-1-methoxy-3-trimethylsiloxy-1,3-butadiene yielding the silylenol ether with no sign of formation of the a,b-conjugated ketone resulting from acidcatalyzed side reactions when the catalysts were prepared by the silylamide routes (Scheme 6.17). In contrast, using a catalyst prepared directly from (fod)3Y and MCM-41 gave only the a,b-conjugated ketones, which shows the importance of the partial passivation of the silica surface [107]. Asymmetric Diels–Alder catalysts were prepared by reacting the MCM-41-supported silylamide lanthanide complexes with chiral ligands such binol, menthol and ephedrine. The reactivity of the supported complexes was enhanced compared with their homogeneous precursors, but so far only low enantiomeric and diastereomeric excesses have been obtained [109].
Scheme 6.17
6.7 Oxidation 6.7.1 Single-site Titanium Species
In industry, one of the key epoxidation processes, the production of propene oxide from propene, uses a catalyst based on TiX4 supported on SiO2 and ROOH as primary oxidants (Equation 6.3) [110]. ð6:3Þ After more than 30 years of research, the nature of the active site is still not known, but several studies on the basis of various models (well-defined heterogeneous catalysts prepared via various methods or soluble analogues) agree on the fact that the active sites are probably isolated Ti centers, probably triply bonded to the surface of silica (tripodal system, Scheme 6.18a) [92, 95, 96, 98, 111–115]. This type of active sites probably corresponds to a compromise between an accessible titanium center
j135
j 6 Single-site Heterogeneous Catalysts via Surface-bound Organometallic and Inorganic Complexes
136
and an increase in the electrophilicity of the metal center through the presence of the right number of siloxy substituents. Moreover, the fourth ligand, not coming from the surface, called the capping ligand (Scheme 6.18b), is also critical for combining good catalytic activity, selectivity and stability for these catalysts. Stabilization of the catalyst can also be achieved by using a calixarene scaffold (Scheme 6.18c) [116, 117]. Finally, in studies devoted to the replacement of alkyl peroxides (ROOH) by hydrogen peroxide, it has been shown that the capping ligand and the hydrophobicity of the silica are critical for obtaining both high activity and selectivity (Scheme 6.18d) [118–124]. The ultimate case corresponds to TS-1, where the Ti is surrounded by four siloxy ligands of the crystalline lattice of a zeolite [125], and in fact a system related to TS-1 has been implemented recently in an industrial process for the epoxidation of propene by H2O2 (Equation 6.4) [126].
Scheme 6.18
ð6:4Þ
6.7 Oxidation
6.7.2 Single-site Zirconium Species
Similarly to the case of Ti, it was shown that tripodal isolated zirconium sites are also the best choice for catalyzing the epoxidation of alkenes. For instance, the oxidation of cyclohexene can be carried with H2O2 to give mainly the epoxide (60–70%) along with the diols (5–8%) and the compounds resulting from allylic oxidation (10–35%) [127]. 6.7.3 Single-site Vanadium Species
In several selective alkane oxidation reactions, e.g. the oxidative dehydrogenation of alkanes to alkenes and the selective oxidation of methane, VO4 species have been involved, but it is not clear yet – despite extensive studies – whether or not these species are necessarily isolated under the reaction conditions (small vanadium oxide clusters), even though high dispersion seems to be critical [128–132]. Here, only an example will be discussed, because it relies on the grafting of well-defined mononuclear precursors on silica [133]. The reaction of [{(tBuO)3SiO}3VO] and [(tBuO)3VO] with a mesoporous silica (SBA) generates the corresponding mononuclear species [(SiO)VO{OSi(OtBu)3}2] and [(SiO)VO(OtBu)2], respectively (Scheme 6.19). Calcination of these species yielded oxidation catalysts having isolated VO4 units for loading as high as 0.47 V nm2. These catalysts perform the oxidation of methane to formaldehyde by O2 with higher selectivity (30–40%) and activity (up to 0.48 mol CH4 mol V1 s1) than those prepared by standard techniques based on polyvanadates.
Scheme 6.19
6.7.4 Single-site Tantalum Species
Using the same strategy as for Ti, a well-defined Ta siloxide [(iPrO)2Ta{OSi(OtBuO)3}3] complex was grafted on a mesoporous silica, namely SBA, yielding isolated Ta centers (Scheme 6.20a) [134]. This material displays good activity in the oxidation of cyclohexene by H2O2 (6.7 mol oxidation products mol1 Ta min1), other oxidants such as TBHP being less efficient. However, the selectivity of this catalyst for
j137
j 6 Single-site Heterogeneous Catalysts via Surface-bound Organometallic and Inorganic Complexes
138
cyclohexene oxide is low (36.0%), cyclohexenol (32.7%) and cyclohexenone (31.2%) q being concomitantly formed, probably via a radical-type mechanism (HO ). Calcination of the material as a way to stabilize the active sites (Scheme 6.20b) does not improve the catalyst performances (lower activities and selectivities), but further treatment of this calcined material with various silylylating agents, e.g. RMe2SiNMe2, to obtain capped Ta active sites improved the activity and, more importantly, increased the selectivity in epoxide to 98–99% (Scheme 6.20c) [135]. This study also strongly suggests that the capping agent stays during the catalytic test and that it is essential to obtain high activities; in fact, removing the cap by calcination induces a loss of stability of the system (Scheme 6.20d).
Scheme 6.20
6.7.5 Single-site Group 6 Species
Epoxidation with homogeneous Mo-based catalysts has also been industrially important and strategies to obtain the corresponding heterogeneous catalysts have been investigated. For instance, isolated Mo centers can prepared by reacting [Mo(N) (OtBu)3] on SiO2-(700) or MCM41-(500). During grafting, the tBuO ligand is replaced by a surface siloxy ligand to give [(SiO)Mo(N)(OtBu)2] (Scheme 6.21a) [136]. These systems are highly active and selective in the epoxidation of cyclohexene by tertbutyl hydroperoxide and are much more stable than the parent molecular precursors. However, these systems deactivate through leaching of Mo. Similarly, the molecular complexes [M(¼O)(OSiOR3)4] (M ¼ Mo and W) react with the silanol of a mesoporous silica (SBA-15) to generate the corresponding isolated oxo species [(SiO)M(¼O)(OSiOR3)3] (Scheme 6.21b), which are both highly active epoxidation catalysts [137].
6.7 Oxidation
Scheme 6.21
6.7.6 Single-site Iron Species
Fe-based heterogeneous catalysts have also been identified as interesting oxidation catalysts with the discovery of the direct and selective oxidation of benzene to phenol by N2O in the presence of Fe silicalite (Equation 6.5) [138, 139].
ð6:5Þ
Several studies have therefore investigated ways to obtain single-site Fe species at the surface of silica supports. First, well-defined FeIII surface complex [(SiO)Fe(OSi (OtBu)3)2thf ] have been prepared by grafting [Fe(OSi(OtBu)3)3thf ] on SBA-15 material (Scheme 6.22a) [140]. Using the same approach, a well-defined FeII surface complex [(SiO)Fe(OSi(OtBu)3)] has been prepared by grafting [{Fe(OSi(OtBu)3)2}2] on SBA-15 material (Scheme 6.22b). In both cases, further calcination of this material at 300 C leads to the formation of isolated FeIII species in a tetrahedral environment (Scheme 6.22c). These isolated Fe species catalyze the oxidation of aliphatic and aromatic C–H bonds by H2O2, transforming benzene into phenol (Equation 6.6); toluene into mixtures of cresols (Equation 6.7), benzyl alcohol and benzaldehyde (Equation 6.8); adamantane into the corresponding alcohols and ketones (Equation 6.9); and cyclohexene into cyclohexenol and cyclohexenone with only traces of epoxide (0–1%) (Equation 6.10) [140, 141]. A more selective cyclohexene epoxidation catalyst with H2O2 is obtained when using a well-defined FeII species having N ligands (Scheme 6.22d); however, the epoxide selectivity is still low (24%), the selectivity in
j139
j 6 Single-site Heterogeneous Catalysts via Surface-bound Organometallic and Inorganic Complexes
140
Scheme 6.22
hexenol (32%) and hexenone (34%) still being very high as a result of Fenton-type chemistry [142]. ð6:6Þ
ð6:7Þ
6.8 Alkane Homologation
ð6:8Þ
ð6:9Þ
ð6:10Þ
It is possible, however, to catalyze the selective epoxidation of propene with N2O with silica-supported Fe isolated species (Equation 6.11) [143–145]. The best system is prepared by impregnating silica with FeIICl2 (0.15 wt%) and Rb2SO4 (10/Fe), followed by a calcination step. The combined use of ESR, UV–visible and Raman spectroscopy is consistent with the presence of isolated Fe species, Rb2SO4 being essential to avoid clustering of Fe into Fe2O3 and to guarantee site isolation, which is necessary to obtain good activities and selectivities [145]. ð6:11Þ
6.7.7 Single-site Cobalt Species
The selective oxidation of alkanes and aromatics is still a challenge today and necessitates the development of better catalysts. Using a silica-supported single-site Co catalyst, prepared by grafting [Co(OSi(OtBu)3)2Bipy2] (Bipy ¼ 4,40 -di-tBu-bipyridine) on SBA-15, the air oxidation of ethylbenzenes into the corresponding acetophenones has been achieved with good selectivities (82–100%) and no leaching (Scheme 6.23) [146].
6.8 Alkane Homologation 6.8.1 Alkane Hydrogenolysis
The hydrogenolysis of alkanes typically requires high reaction temperatures, but when silica-supported early transition metal hydrides are used, this catalytic reaction takes place at relatively low temperatures (50–150 C) [15, 18, 147, 148] The product selectivity depends on the metal: hydrogenolysis of propane yields a 1 : 1 mixture of ethane and methane the silica-supported zirconium hydrides, a Group 4 metal, while
j141
j 6 Single-site Heterogeneous Catalysts via Surface-bound Organometallic and Inorganic Complexes
142
Scheme 6.23
methane is the final product with the silica-supported tantalum hydrides, a Group 5 metal (Table 6.6). The difference in selectivity observed for various alkanes is consistent with a different carbon–carbon cleavage mechanism for both metal:balkyl transfer for Group 4 metals [20, 147, 149–151] and a-alkyl transfer for Ta (Scheme 6.24) [152].
Scheme 6.24
The silica-supported zirconium hydride catalyst has also been used to carry out the hydrogenolysis of a wide range of paraffinic materials, including polyalkenes, which can transformed into lower alkanes under mild conditions [149, 153]. Note that the same catalyst can polymerize alkenes and this illustrates nicely the principle of microreversibility. While depolymerization is indeed highly endothermic and necessitates elevated temperatures, the use of hydrogen allows the equilibrium to be displaced via the hydrogenation of the resulting alkenes.
6.8 Alkane Homologation Table 6.6 Hydrogenolysis of alkanes catalyzed by silica supported metal hydrides.
Catalyst
Alkane
Activitya
[ZrH/SiO2] [ZrH/SiO2]
Ethane Propane
0 70
[ZrH/SiO2]
Isobutane
70
[ZrH/SiO2]
Butane
66
[ZrH/SiO2]
Neopentane
66
[TaH/SiO2] [TaH/SiO2]
Ethane Propane
30 12
[TaH/SiO2]
Isobutane
24
[TaH/SiO2]
Butane
18
[TaH/SiO2]
Neopentane
6
Extrapolated selectivity at 0% conversion
Final product selectivity
— CH4 (50%) C2H6 (50%) CH4 (52%) C3H8 (47%)b CH4 (23%) C3H8 (20%)c CH4 (54%) C4H10 (40%)d CH4 (100%) CH4 (55%) C2H6 (45%) CH4 (55%) C2H6 (10%) C3H8 (35%) CH4 (42%) C2H6 (38%) C3H8 (20%) —e
— CH4 (50%) C2H6 (50%) CH4 (67%) C2H6 (33%) CH4 (60%) C2H6 (40%) CH4 (75%) C2H6 (25%) CH4 (100%) CH4 (100%) CH4 (100%)
CH4 (100%)
CH4 (100%)
a
Activity expressed in moles of propane transformed per mole of metal per hour. Ethane (1.5%). c Ethane (54%). d Propane (5%) and ethane (1%). e Since isobutane is hydrogenolyzed faster than neopentane, selectivity at 0% conversion is difficult to measure in a batch reactor. b
6.8.2 Alkane Metathesis
Alkane metathesis transforms a given alkane into its lower and higher homologues (Scheme 6.25a) and it could therefore become an important process in the petrochemical industry. In fact, finding ways to produce higher alkane homologues has been a major focus of research in this area [154–156]. Alkane metathesis was first investigated via a two-step process combining classical heterogeneous dehydrogenation–hydrogenation catalysts and alkene metathesis catalysts, which allows overall for a given alkane to be converted into its lower and higher homologues [157–159]. This process, however, requires higher temperatures (300–400 C) and higher pressures (>10 bar). More recently, in 1997, alkane metathesis was reported using a single-site catalyst at low temperatures and low pressures, typically 150 C and 0.8–1 bar. The original catalyst was a silica-supported tantalum hydrides (Scheme 6.25b and Table 6.7) [160]. Based on structure–reactivity relationships and kinetic studies, it was shown that, in this case also, the reaction takes place via alkene metathesis, but probably involves
j143
j 6 Single-site Heterogeneous Catalysts via Surface-bound Organometallic and Inorganic Complexes
144
Scheme 6.25
alkylidene hydride intermediates [161–163]. This has lead to an important research effort to find new catalyst based on Group 5 and Group 6 hydrides and alkylidene systems. Note that the silica-supported tantalum hydrides is like all hydrides prepared in two steps (see Section 6.3), involving first grafting of [Ta(¼CHtBu) (CH2tBu)3], generating a well-defined alkylidene system [(SiO)Ta(¼CHtBu) (CH2tBu)2] [164, 165], followed by treatment under H2 at 150 C yielding the hydride [TaH/SiO2] (Scheme 6.25b) [19]. In the case of W, whereas [(SiO)W(CtBu) (CH2tBu)2] is a highly active alkene metathesis catalyst [166], it is nearly inactive in alkane metathesis and the corresponding W hydrides ([WH/SiO2] prepared by
6.8 Alkane Homologation Table 6.7 Performances of single-site catalysts in the metathesis of propanea.
Catalystb
Initial activityc at 24 h CH4
[Ta(¼ CHtBu)(CH2tBu)3]/SiO2 [TaH/SiO2] [TaH/Al2O3] [W(CtBu)(CH2tBu)3]/SiO2 [WH/SiO2] [W(CtBu)(CH2tBu)3]/Al2O3 [WH/Al2O3] [W(CtBu)(CH2tBu)3]/SiO2–Al2O3 [WH/SiO2–Al2O3] [Mo(NAr)(¼CHtBu)(CH2tBu)2]/SiO2 [ZrH/SiO2] [ZrH/SiO2–Al2O3]
3.0 3.5 2.5 — — 1.8 8.5 0.7 8.5 1 —e —e
13 10 10 — 6 3 2 2 2 <0.1 — —
Selectivity at 120 hd TON at 120 h
C2H6 C4H10
C5H12
48 46 48 <2 56 65 57 62 58 56 <2 <2
33 (2.2) 37 (5.1) 36 (8.6)
6 (1.4) 35 7 (2.2) 60 9 (5.1) 60
32 24 34 29 32 38
6 (3.6) 7 (3.5) 6 (3.8) 7 (4.2) 7 (3.7) 6 (7.4)
(10.3) (7.1) (7.8) (7.5) (9.0) (13)
8 28 121 29 123 55
150 C and 0.8 bar if propane; propane : catalyst ¼ 500–1000. For structures, see Scheme 6.25. c Activities are expressed in moles of propane transformed per mole of metal per hour. d Selectivity in lower and higher homologues at 120 h; the selectivity in hexanes is complementary to 100% and the number in parentheses correspond to the n : iso ratio. e Note that under supercritical conditions, the homologation of alkanes was observed (see text). a b
treatment with H2 of [(SiO)W(CtBu)(CH2tBu)2]) cannot be selectively prepared because of sintering of the W species on silica (Scheme 6.25c). On the other hand, the corresponding W alkylidyne and hydride complexes can be prepared on alumina [22, 23] and silica–alumina [24], and they display dramatically improved activities. In fact, the alumina or the silica–alumina tungsten hydrides are the best reported catalysts so far, with initial activities and overall turnovers reaching 8.5 mol of propane transformed per mole of W per hour and 120 turnovers. Moreover, these systems display improved selectivities, yielding less methane. Note also that the corresponding alumina-supported Ta hydrides have similar performances to the corresponding silica-supported Ta system, which shows that the key factor is probably site isolation and stabilization of reactive intermediates, rather than an electronic effect related to the support. In addition to these systems having only hydride and perhydrocarbyl ligands, other possible systems to generate alkylidene hydride intermediates have been investigated and in fact the best of these systems is [(SiO)Mo(NAr)(¼CHtBu)(CH2tBu)] (Scheme 6.25d), a highly active alkene metathesis catalyst, which displays activities comparable to the silica-supported tantalum hydrides [167]. Finally, although zirconium hydrides supported on silica or alumina are inactive in alkane metathesis under standard conditions (150 C and ca 1 bar), the formation of lower and higher homologues of propane is observed under supercritical conditions at 200 C, the overall turnover reaching 63 within 48 h [21]. Moreover, in this case, the selectivity in alkane products is completely different from what is observed for Ta-, Mo- or W-based systems: methane (11.5%), ethane (26.7%), butane (27.6% with
j145
j 6 Single-site Heterogeneous Catalysts via Surface-bound Organometallic and Inorganic Complexes
146
iso/n ¼ 3), pentanes (14.6% with iso/n ¼ 0.3), hexanes (18.6% with branched/n ¼ 3.4) and higher alkanes (1%) (for comparison, see Table 6.7). Here the branched isomers of the higher alkane homologues with the exception of pentane are the major products and moreover the higher molecular weight alkanes are obtained (up to decanes). 6.8.3 Alkane Cross-metathesis
Like alkene metathesis, alkane metathesis is a disproportion reaction, which leads to the formation of the lower and higher homologues of a given alkane. Like alkene metathesis, this reaction is not very exothermic and it is possible to carry out the reverse reaction or cross-metathesis reactions. For instance, one molecule of propane is converted into two ethanes at 250 C in the presence of [TaH/SiO2] (Scheme 6.26a). This reaction requires a large excess of methane in order to displace the thermodynamic equilibrium, but shows that methane can be incorporated into higher alkanes and that in principle methane could replace hydrogen in cracking processes [168]. Similarly, [TaH/SiO2] also catalyzes the cross-metathesis of toluene and ethane at 200 C giving ethylbenzene and xylenes in a ca 5:1 ratio along with the self-metathesis products (Scheme 6.26b) [169].
Scheme 6.26
References 1 Ballard, D.G.H. (1973) Advances in Catalysis, 23, 263–325. 2 Yermakov, Y.I., Kuznetsov, B.N. and Zakharov, V.A. (1981) Studies in Surface Science and Catalysis, 8, 1. 3 Burwell, R.L., Jr. (1984) Journal of Catalysis, 86, 301–314. 4 Marks, T.J. (1992) Accounts of Chemical Research, 25, 57–65.
5 Coperet, C., Chabanas, M., Petroff SaintArroman, R. and Basset, J.-M. (2003) Angewandte Chemie-International Edition, 42, 156–181. 6 Thomas, J.M., Raja, R. and Lewis, D.W. (2005) Angewandte Chemie-International Edition, 44,6456–6482; Tada, M., Iwasawa, Y. (2007) Coordination Chemistry Reviews, 251, 2702–2716.
References 7 Tao, T. and Maciel, G.E. (2000) Journal of the American Chemical Society, 122, 3118–3126. 8 Morrow, B.A. (1990) Studies in Surface Science and Catalysis, 57, A161–A224. 9 Metivier, R., Leray, I., Roy-Auberger, M., Zanier-Szydlowski, N. and Valeur, B. (2002) New Journal of Chemistry, 26, 411–415. 10 Digne, M., Sautet, P., Raybaud, P., Euzen, P. and Toulhoat, H. (2002) Journal of Catalysis, 211, 1–5. 11 Digne, M., Sautet, P., Raybaud, P., Toulhoat, H. and Artacho, E. (2002) The Journal of Physical Chemistry. B, 106, 5155–5162. 12 Digne, M., Sautet, P., Raybaud, P., Euzen, P. and Toulhoat, H. (2004) Journal of Catalysis, 226, 54–68. 13 Joubert, J., Salameh, A., Krakoviack, V., Delbecq, F., Sautet, P., Coperet, C. and Basset, J.M. (2006) The Journal of Physical Chemistry. B, 110, 23944–23950. 14 Zecchina, A., Scarano, D., Bordiga, S., Spoto, G. and Lamberti, C. (2001) Advances in Catalysis, 46, 265–397. 15 Rosier, C., Niccolai, G.P. and Basset, J.-M. (1997) Journal of the American Chemical Society, 119, 12408–12409. 16 Corker, J., Lefebvre, F., Lecuyer, C., Dufaud, V., Quignard, F., Choplin, A., Evans, J. and Basset, J.-M. (1996) Science, 271, 966–969. 17 Rataboul, F., Baudouin, A., Thieuleux, C., Veyre, L., Coperet, C., Thivolle-Cazat, J., Basset, J.-M., Lesage, A. and Emsley, L. (2004) Journal of the American Chemical Society, 126, 12541–12550. 18 d Ornelas, L., Reyes, S., Quignard, F., Choplin, A. and Basset, J.M. (1993) Chemistry Letters, 1931–1934. 19 Vidal, V., Theolier, A., Thivolle-Cazat, J., Basset, J.-M. and Corker, J. (1996) Journal of the American Chemical Society, 118, 4595–4602. 20 Joubert, J., Delbecq, F., Thieuleux, C., Taoufik, M., Blanc, F., Coperet, C., Thivolle-Cazat, J., Basset, J.-M. and Sautet, P. (2007) Organometallics, 26, 3329–3335.
21 Thieuleux, C., Maraval, A., Veyre, L., Coperet, C., Soulivong, D., Basset, J.-M. and Sunley, G.J. (2007) Angewandte Chemie-International Edition, 46, 2288–2290. 22 Taoufik, M., Le Roux, E., Thivolle-Cazat, J., Coperet, C., Basset, J.-M., Maunders, B. and Sunley, G.J. (2006) Topics in Catalysis, 40, 65–70. 23 Le Roux, E., Taoufik, M., Coperet, C., de Mallmann, A., Thivolle-Cazat, J., Basset, J.-M., Maunders, B.M. and Sunley, G.J. (2005) Angewandte Chemie-International Edition, 44, 6755–6758. 24 Le Roux, E., Taoufik, M., Baudouin, A., Coperet, C., Thivolle-Cazat, J., Basset, J.M., Maunders, B.M. and Sunley, G.J. (2007) Advanced Synthesis and Catalysis, 349, 231–237. 25 Schwartz, J. and Ward, M.D. (1980) Journal of Molecular Catalysis, 8, 465–469. 26 King, S.A. and Schwartz, J. (1991) Inorganic Chemistry, 30, 3771–3774. 27 Yermakov, Y.I., Ryndin, Y.A., Alekseev, O.S., Kochubei, D.I., Shmachkov, V.A. and Gergert, N.I. (1989) Journal of Molecular Catalysis, 49, 121–132. 28 Zakharov, V.A. and Ryndin, Y.A. (1989) Journal of Molecular Catalysis, 56, 183–193. 29 Profilet, R.D., Rothwell, A.P. and Rothwell, I.P. (1993) Journal of the Chemical Society. Chemical Communications, 42–44. 30 Burwell, R.L., Jr. and Marks, T.J. (1985) Chemistry & Industry, 22, 207–224. 31 Marks, T.J. (1992) Accounts of Chemical Research, 25, 57–65. 32 Dahmen, K.H., Hedden, D., Burwell, R.L., Jr. and Marks, T.J. (1988) Langmuir, 4, 1212–1214. 33 Ahn, H., Nicholas, C.P. and Marks, T.J. (2002) Organometallics, 21, 1788–1806. 34 Nicholas, C.P., Ahn, H. and Marks, T.J. (2003) Journal of the American Chemical Society, 125, 4325–4331. 35 Nicholas, C.P. and Marks, T.J. (2004) Langmuir, 20, 9456–9462.
j147
j 6 Single-site Heterogeneous Catalysts via Surface-bound Organometallic and Inorganic Complexes
148
36 Bowman, R.G., Nakamura, R., Fagan, P.J., Burwell, R.L., Jr. and Marks, T.J. (1981) Journal of the Chemical Society. Chemical Communications, 257–258. 37 He, M.Y., Xiong, G., Toscano, P.J., Burwell, R.L., Jr. and Marks, T.J. (1985) Journal of the American Chemical Society, 107, 641–652. 38 Gillespie, R.D., Burwell, R.L., Jr. and Marks, T.J. (1990) Langmuir, 6, 1465–1477. 39 Eisen, M.S. and Marks, T.J. (1992) Journal of the American Chemical Society, 114, 10358–10368. 40 Eisen, M.S. and Marks, T.J. (1994) Journal of Molecular Catalysis, 86, 23–50. 41 Gauvin, R.M., Delevoye, L., Hassan, R.A., Keldenich, J. and Mortreux, A. (2007) Inorganic Chemistry, 46, 1062–1070. 42 Marciniec, B., Szubert, K., Potrzebowski, M.J., Kownacki, I. and Leszczak, K. (2008) Angewandte Chemie-International Edition, 47, 541–544. 43 Mowat, W., Smith, J. and Whan, D.A. (1974) Journal of the Chemical Society. Chemical Communications, 34–35. 44 Smith, J., Mowat, W., Whan, D.A. and Ebsworth, E.A.V. (1974) Journal of the Chemical Society-Dalton Transactions, 1742–1746. 45 Weiss, K. and Loessel, G. (1989) Angewandte Chemie-International Edition, 28, 62. 46 Buffon, R., Leconte, M., Choplin, A. and Basset, J.M. (1993) Journal of the Chemical Society. Chemical Communications, 361–362. 47 Buffon, R., Leconte, M., Choplin, A. and Basset, J.-M. (1994) Journal of the Chemical Society-Dalton Transactions, 1723–1729. 48 Blanc, F., Coperet, C., Thivolle-Cazat, J., Basset, J.-M., Lesage, A., Emsley, L., Sinha, A. and Schrock, R.R. (2006) Angewandte Chemie-International Edition, 45, 1216–1220. 49 Blanc, F., Thivolle-Cazat, J., Basset, J.-M., Coperet, C., Hock, A.S., Tonzetich, Z.J. and Schrock, R.R. (2007) Journal of the American Chemical Society, 129, 1044–1045.
50 Blanc, F., Berthoud, R., Salameh, A., Basset, J.-M., Coperet, C., Singh, R. and Schrock, R.R. (2007) Journal of the American Chemical Society, 129, 8434–8435. 51 Rhers, B., Salameh, A., Baudouin, A., Quadrelli, E.A., Taoufik, M., Coperet, C., Lefebvre, F., Basset, J.-M., SolansMonfort, X., Eisenstein, O., Lukens, W.W., Lopez, L.P.H., Sinha, A. and Schrock, R.R. (2006) Organometallics, 25, 3554–3557. 52 Chabanas, M., Baudouin, A., Coperet, C. and Basset, J.-M. (2001) Journal of the American Chemical Society,123, 2062–2063. 53 Lesage, A., Emsley, L., Chabanas, M., Coperet, C. and Basset, J.-M. (2002) Angewandte Chemie-International Edition, 41, 4535–4538. 54 Chabanas, M., Baudouin, A., Coperet, C., Basset, J.-M., Lukens, W., Lesage, A., Hediger, S. and Emsley, L. (2003) Journal of the American Chemical Society, 125, 492–504. 55 Coperet, C. (2004) New Journal of Chemistry, 28, 1–10. 56 Solans-Monfort, X., Clot, E., Coperet, C. and Eisenstein, O. (2005) Organometallics, 24, 1586–1597. 57 Solans-Monfort, X., Filhol, J.-S., Coperet, C. and Eisenstein, O. (2006) New Journal of Chemistry, 30, 842–850. 58 Ivin, K.J. and Mol, I.C. (1996) Olefin Metathesis and Metathesis Polymerization, 2nd edn, Academic Press, San Diego, CA. 59 Robbins, J., Bazan, G.C., Murdzek, J.S., ORegan, M.B. and Schrock, R.R. (1991) Organometallics, 10, 2902–2907. 60 Lopez, L.P.H. and Schrock, R.R. (2004) Journal of the American Chemical Society, 126, 9526–9527. 61 Sinha, A., Lopez, L.P.H., Schrock, R.R., Hock, A.S. and Mueller, P. (2006) Organometallics, 25, 1412–1423. 62 Lopez, L.P.H., Schrock, R.R. and Mueller, P. (2006) Organometallics, 25, 1978–1986. 63 Coperet, C. (2007) Dalton Transactions, 5498–5504. 64 Solans-Monfort, X., Clot, E., Coperet, C. and Eisenstein, O. (2005) Journal of the American Chemical Society, 127, 14015–14025.
References 65 Poater, A., Solans-Monfort, X., Clot, E., Coperet, C. and Eisenstein, O. (2007) Journal of the American Chemical Society, 129, 8207–8216. 66 Mol, J.C. (1999) Catalysis Today, 51, 289–299. 67 Verkuijlen, E., Kapteijn, F., Mol, J.C. and Boelhouwer, C. (1977) Journal of the Chemical Society. Chemical Communications, 198–199. 68 Mol, J.C. (1994) Journal of Molecular Catalysis, 90, 185–200. 69 Mol, J.C. (2004) Topics in Catalysis, 27, 97–104. 70 Salameh, A., Coperet, C., Basset, J.-M., Bohm, V.P.W. and Roper, M. (2007) Advanced Synthesis and Catalysis, 349, 238–242. 71 Herrmann, W.A., Wagner, W., Flessner, U.N., Vokhardt, U. and Komber, H. (1991) Angewandte Chemie-International Edition, 30, 1636–1638. 72 Buffon, R., Choplin, A., Leconte, M., Basset, J.M., Touroude, R. and Herrmann, W.A. (1992) Journal of Molecular Catalysis, 72, L7–L10. 73 Malek, A. and Ozin, G. (1995) Advanced Materials, 7, 160–163. 74 Rost, A.M.J., Schneider, H., Zoller, J.P., Herrmann, W.A. and Kuehn, F.E. (2005) Journal of Organometallic Chemistry, 690, 4712–4718. 75 Moses, A.W., Ramsahye, N.A., Raab, C., Leifeste, H.D., Chattopadhyay, S., Chmelka, B.F., Eckert, J. and Scott, S.L. (2006) Organometallics, 25, 2157–2165. 76 Salameh, A., Joubert, J., Baudouin, A., Lukens, W., Delbecq, F., Sautet, P., Basset, J.M. and Coperet, C. (2007) Angewandte Chemie-International Edition, 46, 3870–3873. 77 Salameh, A., Baudouin, A., Soulivong, D., Boehm, V., Roeper, M., Basset, J.-M. and Coperet, C. (2008) Journal of Catalysis, 253, 180–190. 78 Salameh, A., Baudouin, A., Basset, J.-M. and Coperet, C. (2008) Angewandte ChemieInternational Edition, 48, 2117–2120.
79 Tullock, C.W., Tebbe, F.N., Mulhaupt, R., Ovenall, D.W., Setterquist, R.A. and Ittel, S.D. (1989) Journal of Polymer Science Part A, 27, 3063–3081. 80 Collette, J.W., Tullock, C.W., MacDonald, R.N., Buck, W.H., Su, A.C.L., Harrell, J.R., Mulhaupt, R. and Anderson, B.C. (1989) Macromolecules, 22, 3851–3858. 81 Tullock, C.W., Mulhaupt, R. and Ittel, S.D. (1989) Makromolekulare Chemie Rapid Communication, 10, 19–23. 82 Ittel, S.D. (1990) Journal of Macromolecular Science A, 27, 1133–1146. 83 Joubert, J., Delbecq, F., Sautet, P., Le Roux, E., Taoufik, M., Thieuleux, C., Blanc, F., Coperet, C., Thivolle-Cazat, J. and Basset, J.-M. (2006) Journal of the American Chemical Society, 128, 9157–9169. 84 Taoufik, M., Le Roux, E., Thivolle-Cazat, J. and Basset, J.-M. (2007) Angewandte Chemie-International Edition, 46, 7202–7205. 85 Zhang, W. and Moore, J.S. (2007) Advanced Synthesis and Catalysis, 349, 93–120. 86 Weissman, H., Plunkett, K.N. and Moore, J.S. (2006) Angewandte ChemieInternational Edition, 45, 585–588. 87 Gauvin, R.M., Coutelier, O., Berrier, E., Mortreux, A., Delevoye, L., Paul, J.-F., Mamede, A.-S. and Payen, E. (2007) Dalton Transactions, 3127–3130. 88 Gauvin, R.M. and Mortreux, A. (2005) Chemical Communications, 1146–1148. 89 Richmond, M.K., Scott, S.L. and Alper, H. (2001) Journal of the American Chemical Society, 123, 10521–10525. 90 Richmond, M.K., Scott, S.L., Yap, G.P.A. and Alper, H. (2002) Organometallics, 21, 3395–3400. 91 Leyrit, P., McGill, C., Quignard, F. and Choplin, A. (1996) Journal of Molecular Catalysis A-Chemical, 112, 395–400. 92 Holmes, S.A., Quignard, F., Choplin, A., Teissier, R. and Kervennal, J. (1998) Journal of Catalysis, 176, 173–181. 93 Quignard, F., Graziani, O. and Choplin, A. (1999) Applied Catalysis A-General, 182, 29–40.
j149
j 6 Single-site Heterogeneous Catalysts via Surface-bound Organometallic and Inorganic Complexes
150
94 Bouh, A.O., Rice, G.L. and Scott, S.L. (1999) Journal of the American Chemical Society, 121, 7201–7210. 95 Jarupatrakorn, J. and Tilley, T.D. (2002) Journal of the American Chemical Society, 124, 8380–8388. 96 Crocker, M., Herold, R.H.M., Roosenbrand, B.G., Emeis, K.A. and Wilson, A. (1998) Colloids and Surfaces A, 139, 351–368. 97 Wang,X.-X.,Veyre,L.,Lefebvre,F.,Patarin, J. and Basset, J.-M. (2003) Microporous Mesoporous Materials, 66, 169–179. 98 Bini, F., Rosier, C., Saint-Arroman, R.P., Neumann, E., Dablemont, C., de Mallmann, A., Lefebvre, F., Niccolai, G.P., Basset, J.-M., Crocker, M. and Buijink, J.K. (2006) Organometallics, 25, 3743–3760. 99 Capdevielle, V. PhD Thesis, Universite Lyon 1, (1996). 100 Ferret, N., Dufaud, V., Salinier, V. and Basset, J. (1997) French Patent 2 747 675 (Elf Atochem). 101 Anwander, R. (2001) Chemistry of Materials, 13, 4419–4438. 102 Anwander, R. (2003) Nanostructured Materials, 15–56. 103 Anwander, R. and Roesky, R. (1997) Journal of the Chemical Society-Dalton Transactions, 137–138. 104 Anwander, R., Runte, O., Eppinger, J., Gerstberger, G., Herdtweck, E. and Spiegler, M. (1998) Journal of the Chemical Society-Dalton Transactions, 847–858. 105 Gerstberger, G., Palm, C. and Anwander, R. (1999) Chemistry – A European Journal, 5, 997–1005. 106 Anwander, R., Nagl, I., Zapilko, C. and Widenmeyer, M. (2003) Tetrahedron, 59, 10567–10574. 107 Gerstberger, G., Palm, C. and Anwander, R. (1999) Chemistry-A European Journal, 5, 997–1005. 108 Anwander, R. and Palm, C. (1998) Studies in Surface Science and Catalysis, 117, 413–420. 109 Gerstberger, G. and Anwander, R. (2001) Microporous Mesoporous Materials, 44–45, 303–310.
110 Sheldon, R.A. and Van Vliet, M.C.A. (2001) In Fine Chemicals Through Heterogeneous Catalysis, (eds R.A. Sheldon and H. van Bekkum), Wiley-VCH Verlag GmbH, pp. 473–490. 111 Blasco, T., Corma, A., Navarro, M.T. and Perez Pariente, J. (1995) Journal of Catalysis, 156, 65–74. 112 Maschmeyer, T., Rey, F., Sankar, G. and Thomas, J.M. (1995) Nature, 378, 159–162. 113 Maschmeyer, T., Klunduk, M.C., Martin, C.M., Shephard, D.S., Thomas, J.M. and Johnson, B.F.G. (1997) Chemical Communications, 1847–1848. 114 Crocker, M., Herold, R.H.M., Orpen, A.G., Overgaag, M.T.A. (1999) Journal of the Chemical Society-Dalton Transactions, 3791–3804. 115 Rosier, C. (1999) PhD Thesis, UCBL. 116 Notestein, J.M., Iglesia, E. and Katz, A. (2004) Journal of the American Chemical Society, 126, 16478–16486. 117 Notestein, J.M., Solovyov, A., Andrini, L.R., Requejo, F.G., Katz, A. and Iglesia, E. (2007) Journal of the American Chemical Society, 129, 15585–15595. 118 Holmes, S.A., Quignard, F., Choplin, A., Teissier, R. and Kervennal, J. (1998) Journal of Catalysis, 176, 182–191. 119 Corma, A., Domine, M., Gaona, J.A., Jorda, J.L., Navarro, M.T., Rey, F., PerezPariente, J., Tsuji, J., McCulloch, B. and Nemeth, L.T. (1998) Chemical Communications, 2211–2212. 120 Tatsumi, T., Koyano, K.A. and Igarashi, N. (1998) Chemical Communications, 325–326. 121 DAmore, M.B. and Schwarz, S. (1999) Chemical Communications, 121–122. 122 Bu, J. and Rhee, H.-K. (2000) Catalysis Letters, 66, 245–249. 123 Corma, A., Lorda, J.L., Navarro, M.T., Perez-Pariente, J., Rey, F. and Tsuji, J. (2000) Studies in Surface Science and Catalysis, 129, 169–178. 124 Brutchey, R.L., Ruddy, D.A., Andersen, L.K. and Tilley, T.D. (2005) Langmuir, 21, 9576–9583.
References 125 Notari, B. (1993) Catalysis Today, 18, 163–172. 126 Tullo, A.H., (2007) Chemical & Engineering News, 95 (9), 25–26. 127 Quignard, F., Choplin, A. and Teissier, R. (1997) Journal of Molecular Catalysis AChemical, 120, L27–L31. 128 Albonetti, S., Cavani, F. and Trifiro, F. (1996) Catalysis Reviews: Science and Engineering, 38, 413–438. 129 Gao, X. and Wachs, I.E. (2002) Topics in Catalysis, 18, 243–250. 130 Kustrowski, P., Segura, Y., Chmielarz, L., Surman, J., Dziembaj, R., Cool, P. and Vansant, E.F. (2006) Catalysis Today, 114, 307–313. 131 Fu, H., Liu, Z.-P., Li, Z.-H., Wang, W.-N. and Fan, K.-N. (2006) Journal of the American Chemical Society, 128, 11114–11123. 132 Hess, C. (2007) Journal of Catalysis, 248, 120–123. 133 Hess, C., Drake, I.J., Hoefelmeyer, J.D., Tilley, T.D. and Bell, A.T. (2005) Catalysis Letters, 105, 1–8. 134 Brutchey, R.L., Lugmair, C.G., Schebaum, L.O. and Tilley, T.D. (2005) Journal of Catalysis, 229, 72–81. 135 Ruddy, D.A. and Tilley, T.D. (2007) Chemical Communications, 3350–3352. 136 Yang, Q., Coperet, C., Li, C. and Basset, J.M. (2003) New Journal of Chemistry, 27, 319–323. 137 Jarupatrakorn, J., Coles, M.P. and Tilley, T.D. (2005) Chemistry of Materials, 17, 1818–1828. 138 Panov, G.I., Kharitonov, A.S. and Sobolev, V.I. (1993) Applied Catalysis A-General, 98, 1–20. 139 Pirutko, L.V., Chernyavsky, V.S., Uriarte, A.K. and Panov, G.I. (2002) Applied Catalysis A-General, 227, 143–157. 140 Nozaki, C., Lugmair, C.G., Bell, A.T. and Tilley, T.D. (2002) Journal of the American Chemical Society, 124, 13194–13203. 141 Holland, A.W., Li, G., Shahin, A.M., Long, G.J., Bell, A.T. and Tilley, T.D. (2005) Journal of Catalysis, 235, 150–163.
142 Roukoss, C., Fiddy, S., de Mallmann, A., Rendón, N., Basset, J.-M., Kuntz, E. and Coperet, C. (2007) Dalton Transactions, 5546–5548. 143 Wang, X., Zhang, Q., Yang, S. and Wang, Y. (2005) The Journal of Physical Chemistry. B, 109, 23500–23508. 144 Wang, Y., Yang, W., Yang, L., Wang, X. and Zhang, Q. (2006) Catalysis Today, 117, 156–162. 145 Moens, B., De Winne, H., Corthals, S., Poelman, H., De Gryse, R., Meynen, V., Cool, P., Sels, B.F. and Jacobs, P.A. (2007) Journal of Catalysis, 247, 86–100. 146 Brutchey, R.L., Drake, I.J., Bell, A.T. and Tilley, T.D. (2005) Chemical Communications, 3736–3738. 147 Lecuyer, C., Quignard, F., Choplin, A., Olivier, D. and Basset, J.M. (1991) Angewandte Chemie-International Edition, 30, 1660–1661. 148 Chabanas, M., Vidal, V., Coperet, C., Thivolle-Cazat, J. and Basset, J.-M. (2000) Angewandte Chemie-International Edition, 39, 1962–1965. 149 Thieuleux, C., Coperet, C., Dufaud, V., Marangelli, C., Kuntz, E. and Basset, J.M. (2004) Journal of Molecular Catalysis AChemical, 213, 47–57. 150 Mortensen, J.J. and Parrinello, M. (2000) The Journal of Physical Chemistry. B, 104, 2901–2907. 151 Besedin, D.V., Ustynyuk, L.Y., Ustynyuk, Y.A. and Lunin, V.V. (2005) Topics in Catalysis, 32, 47–60. 152 Mikhailov, M.N. and Kustov, L.M. (2005) Russian Chemical Bulletin, 54, 300–311. 153 Dufaud, V. and Basset, J.-M. (1998) Angewandte Chemie-International Edition, 37, 806–810. 154 Crabtree, R.H. (2001) Journal of the Chemical Society-Dalton Transactions, 2437–2450. 155 Labinger, J.A. and Bercaw, J.E. (2002) Nature, 417, 507–514. 156 Olah, G.A. and Molnar, A., (2003) Hydrocarbon Chemistry, 2nd edn., John Wiley & Sons, New York.
j151
j 6 Single-site Heterogeneous Catalysts via Surface-bound Organometallic and Inorganic Complexes
152
157 Heckelsberg, L.F. and Banks, R.L. (1969) US Patent 3 445 541 Phillips Petroleum. 158 Hughes, T.R. (1971) US Patent 2 075 370 Chevron Research. 159 Burnett, R.L. and Hughes, T.R. (1973) Journal of Catalysis, 31, 55–64. 160 Vidal, V., Theolier, A., Thivolle-Cazat, J. and Basset, J.-M. (1997) Science, 276, 99–102. 161 Coperet, C., Maury, O., Thivolle-Cazat, J. and Basset, J.-M. (2001) Angewandte ChemieInternational Edition, 40, 2331–2334. 162 Le Roux, E., Chabanas, M., Baudouin, A., de Mallmann, A., Coperet, C., Quadrelli, E.A., Thivolle-Cazat, J., Basset, J.-M., Lukens, W., Lesage, A., Emsley, L. and Sunley, G.J. (2004) Journal of the American Chemical Society, 126, 13391–13399. 163 Basset, J.M., Coperet, C., Lefort, L., Maunders, B.M., Maury, O., Le Roux, E., Saggio, G., Soignier, S., Soulivong, D., Sunley, G.J., Taoufik, M. and ThivolleCazat, J. (2005) Journal of the American Chemical Society, 127, 8604–8605.
164 Dufaud, V., Niccolai, G.P., Thivolle-Cazat, J. and Basset, J.-M. (1995) Journal of the American Chemical Society, 117, 4288–4294. 165 Chabanas, M., Quadrelli, E.A., Fenet, B., Coperet, C., Thivolle-Cazat, J., Basset, J.-M., Lesage, A. and Emsley, L. (2001) Angewandte Chemie-International Edition, 40, 4493–4496. 166 Le Roux, E., Taoufik, M., Chabanas, M., Alcor, D., Baudouin, A., Coperet, C., Thivolle-Cazat, J., Basset, J.-M., Lesage, A., Hediger, S. and Emsley, L. (2005) Organometallics, 24, 4274–4279. 167 Blanc,F., Coperet,C., Thivolle-Cazat,J. and Basset, J.-M. (2006) Angewandte ChemieInternational Edition, 45, 6201–6203. 168 Soulivong, D., Coperet, C., ThivolleCazat, J., Basset, J.-M., Maunders, B.M., Pardy, R.B.A. and Sunley, G.J. (2004) Angewandte Chemie-International Edition, 43, 5366–5369. 169 Taoufik, M., Schwab, E., Schultz, M., Vanoppen, D., Walter, M., Thivolle-Cazat, J. and Basset, J.-M. (2004) Chemical Communications, 1434–1435.
j153
7 Sustainable Heterogeneous Acid Catalysis by Heteropoly Acids Ivan Kozhevnikov
7.1 Introduction
Catalysis by heteropoly acids (HPAs) is at the forefront of fundamental and applied catalysis [1–8]. Systematic mechanistic studies of HPA catalysis at the molecular level have led to a series of large-scale industrial applications of these catalysts in organic synthesis [1, 4]. HPAs possess unique physicochemical properties, with their structural mobility and multifunctionality being the most important for catalysis [1–4]. Unlike metal oxides and zeolites, HPAs have discrete and mobile ionic structure. On the one hand, HPAs possess very strong Brønsted acidity, and on the other, appropriate redox properties. Both the acid and redox properties can be tuned by varying the chemical composition of HPA. Consequently, acid catalysis and selective oxidation are the major areas of catalytic applications of HPAs. There are many structural types of heteropoly compounds (polyoxometalates), which have been reviewed in detail elsewhere [9, 10]. The majority of catalytic applications use the most stable and easily available Keggin HPAs, especially for acid catalysis. The Keggin HPAs comprise heteropoly anions of the formula [XM12O40]n (a-isomer), where X is the heteroatom (P5þ , Si4þ , etc.) and M the addendum atom (Mo6þ , W6þ , etc.). The structure of the heteropoly anion is composed of a central tetrahedron XO4 surrounded by 12 edge- and corner-sharing metal–oxygen octahedra MO6. Figure 7.1 shows the Keggin structure in polyhedral, ball-and-stick and space-filling representations. Most typical Keggin HPAs such as H3PW12O40, H4SiW12O40 and H3PMo12O40 are commercially available. HPA catalysts are currently used in several industrial processes (Table 7.1) [4]. The top two processes in the table are heterogeneously catalyzed selective oxidations in the gas phase – the oxidation of methacrolein to methacrylic acid and ethene to acetic acid. All other processes are acid-catalyzed reactions. These include homogeneous liquid-phase hydration of alkenes, propene and butenes, to yield alcohols, and biphasic polymerization of tetrahydrofuran to poly(tetramethylene glycol). More recently, synthesis of ethyl acetate by direct addition of acetic acid to ethene in the gas
Handbook of Green Chemistry, Volume 2: Heterogeneous Catalysis. Edited by Robert H. Crabtree Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32497-2
j 7 Sustainable Heterogeneous Acid Catalysis by Heteropoly Acids
154
Figure 7.1 Structure of Keggin heteropoly anion [a-XM12O40]n in polyhedral (a), ball-and-stick (b)and space-filling (c)representations [9]. Table 7.1 Industrial processes catalyzed by heteropoly acids [4].
Reaction
Catalyst
Typea
Start
CH2 ==CðCH3 ÞCHO þ O2 ! CH2 ==CðCH3 ÞCOOH CH2 ==CH2 þ O2 ! CH3 COOH CH2 ==CHCH3 þ H2 O ! CH3 CHðOHÞCH3 CH2 ==CðCH3 Þ2 þ H2 O ! ðCH3 Þ3 COH CH3 CH==CHCH3 þ H2 O ! CH3 CHðOHÞCH2 CH3 nTHF þ H2 O ! HO-½-ðCH2 Þ4 -O-n-H CH2 ==CH2 þ CH3 COOH ! CH3 CH2 O2 CCH3
Mo–V–P–HPA Pd–H4SiW12O40/SiO2 H4SiW12O40 H3PMo12O40 H3PMo12O40 H3PW12O40 H4SiW12O40/SiO2
het het hom hom hom bip het
1982 1997 1972 1984 1989 1985 2001
a
hom, homogeneous; het, heterogeneous; bip, biphasic.
phase over HPA catalyst has been commercialized by Showa Denko in Japan [6] and BP in the UK (the Avada process) [11]. Heterogeneous acid catalysis by HPAs has attracted much interest because of its potential for great economic rewards and green benefits [1–8]. Table 7.2 shows the broad scope of reactions catalyzed by heteropoly acids in heterogeneous gas–solid and liquid–solid systems. These reactions are placed in the order of decreasing Table 7.2 Reactions catalyzed by heteropoly acids in heterogeneous gas–solid and liquid–solid systems [4].
Isomerization and cracking of alkanes Conversion of MeOH to alkenes Nitration of benzene Alkylation of alkanes Oligomerization of alkenes Friedel–Crafts and related reactions Esterification and transesterification Hydration of alkenes Dehydration of alcohols Hydrolysis Addition: isobuteneþ MeOH ! TBE; alkeneþ AcOH ! alkyl acetate Diels–Alder reaction
7.1 Introduction
Figure 7.2 TGA for H3PW12O40 hydrate [4].
catalyst acid strength required for the reaction to occur, ranging from highly demanding alkane isomerization and cracking to very mild additions to C¼C double bond and the Diels–Alder reaction. HPAs possess stronger (Brønsted) acidity than conventional solid acid catalysts such as acidic oxides and zeolites. The acid strength of Keggin HPAs decreases in the order: H3PW12O40 > H4SiW12O40 > H3PMo12O40 > H4SiMo12O40 [1, 3, 4]. The acid sites in HPA are more uniform and easier to control than those in other solid acid catalysts. Being stronger acids, HPAs are generally more active catalysts than the conventional solid acid catalysts, which allows efficient operation under milder conditions. A serious problem to HPA catalysts is their low thermal stability, hence limited reaction temperature and, especially, difficulty of regeneration of solid HPA catalysts (decoking) [4, 8]. The thermal stability of Keggin HPAs, defined as the temperature at which all acidic protons are lost, decreases in the order H3PW12O40 (465 C) > H4SiW12O40 (445 C) > H3PMo12O40 (375 C) > H4SiMo12O40 (350 C), the strongest acid H3PW12O40 being the most stable [4]. Figure 7.2 shows the TGA profile for H3PW12O40 hydrate [4]. Three main peaks can be observed: (1) a peak at a temperature below 100 C, corresponding to the loss of physisorbed water (a variable amount depending on the number of hydration waters in the sample); (2) a peak in the temperature range 100–280 C centered at about 200 C, accounted for by the loss of ca 6 H2O molecules per Keggin unit, corresponding to the dehydration of a relatively stable hexahydrate H3PW12O406H2O, in which the waters are hydrogen bonded to the acidic protons to form the [H2OH þ OH2] ions; and (3) a peak in the range 370–600 C centered at 450–470 C, which is due to the loss of 1.5 H2O molecules, corresponding to the loss of all acidic protons and the beginning of decomposition of the Keggin structure. For tungsten HPAs, the last loss is practically irreversible, which causes irreversible loss of catalytic activity. The decomposition is complete at about 610 C to form P2O5 and WO3, which exhibits an exotherm in DTA and DSC [3, 4]. The course of thermal decomposition of H3PW12O40 is shown in Scheme 7.1 [4].
j155
j 7 Sustainable Heterogeneous Acid Catalysis by Heteropoly Acids
156
Scheme 7.1 Thermal decomposition of H3PW12O40 hydrate.
Various aspects of catalyst deactivation and regeneration are well covered in the literature [12, 13]. Coke formation is the most frequent cause of catalyst deactivation in heterogeneous acid-catalyzed organic reactions [13–16]. Much research has been carried out into coke formation on the catalysts for petrochemical processes such as catalytic cracking, reforming and hydrotreatment. The most studied catalysts include amorphous silica–alumina, zeolites and acidic alumina, and also those doped with metals such as palladium, platinum and nickel [13–16]. Catalyst regeneration (decoking) is usually carried our by coke combustion at 450–550 C [13–16]. For the oxide and zeolite catalysts possessing sufficient thermal stability, combustion of coke is an effective method to recover catalyst activity. Solid HPA catalysts in organic reactions suffer from deactivation by coking, similar to the conventional solid acid catalysts. However, little information is available about coke formation on HPA catalysts. The problem is that the standard catalyst regeneration by coke combustion is not applicable to HPA catalysts due to their low thermal stability. This makes coking the most serious problem for heterogeneous acid catalysis by HPAs [4, 8]. Other possible causes of HPA deactivation, such as poisoning, aggregation, dehydration and decomposition of HPA, could also play a role but not as crucial as coking, at least at the moderate reaction temperatures that are typical of acid catalysis by HPAs (100–300 C). The question is how to overcome the problem of coking and make heterogeneous acid catalysis by HPA sustainable. Several directions that may be instrumental to achieve this goal will be discussed here, namely developing HPA catalysts possessing high thermal stability, modification of HPA catalysts to enhance coke combustion, inhibition of coke formation on HPA catalysts during operation, reactions in supercritical fluids and cascade reactions using multifunctional HPA catalysis [8].
7.2 Development of HPA Catalysts Possessing High Thermal Stability
In recent years, there has been considerable activity in this direction, focusing mainly on oxide composites comprising tungsten(VI) polyoxometalates and niobium(V), zirconium(IV) and titanium(IV) oxides as an oxide matrix [17–24]. These composites are usually prepared by wet chemical synthesis, followed by calcination at 500–750 C, i.e. at temperatures considerably higher than the temperature of HPA decomposition. The materials thus made contain HPA precursors or HPA decomposition products, possessing Brønsted and Lewis acid sites of moderate strength. These materials have been found active in a range of Friedel–Crafts reactions, often
7.3 Modification of HPA Catalysts to Enhance Coke Combustion
with good catalyst recycling. However, their activity is considerably lower than that of the standard Keggin HPA catalysts. For example, the H3PO4–WO3–Nb2O5 (9 : 55 : 36 wt%) composite with a surface area of 58 m2 g1 has been prepared by interaction of (NH4)10W12O41, Nb(V) oxalate and H3PO4 in aqueous solution, followed by evaporation and calcination at 500 C [17]. It has been tested in the alkylation of anisole by benzyl alcohol (Scheme 7.2) to yield 94% of the alkylation product. The catalyst is reported to be recyclable many times without loss of its activity. However, it is less active than the H3PW12O40/Nb2O5 catalyst prepared by the usual impregnation of niobium(V) oxide with H3PW12O40, but the latter is not recyclable.
Scheme 7.2
Another composite has been obtained by impregnation of 15% H4SiW12O40 on zirconium(IV) oxide and calcination at 700 C [19]. From Raman spectra, it contains ZrO2-anchored monooxotungstate and possesses Brønsted and Lewis acid sites. This solid acid material is active in the acylation of veratrole by benzoic anhydride (Scheme 7.3), with no leaching and good catalyst recycling after regeneration by coke combustion at 500 C. However, this catalyst is less active than HY zeolite per gram of catalyst, whereas the standard HPA catalysts are usually much more active than the HY. Therefore, the composite materials based on W(VI) polyoxometalates and Nb(V), Zr(IV) and Ti(IV) oxides possess relatively weak acid sites and as solid acid catalysts have practically no advantage over acidic zeolites. Work should be continued to obtain HPA materials possessing stronger acid sites.
Scheme 7.3
7.3 Modification of HPA Catalysts to Enhance Coke Combustion
Doping of solid acid catalysts with platinum group metals (PGM) such as palladium and platinum to enhance catalyst regeneration by coke combustion is well docu-
j157
j 7 Sustainable Heterogeneous Acid Catalysis by Heteropoly Acids
158
mented [13, 14]. Typical examples are zeolite and alumina doped with Pd and Pt employed as the catalysts for alkane isomerization and cracking [13, 14]. It is suggested that the role of metal sites is twofold: on the one hand, they catalyze coke combustion, and on the other, they change the nature of coke depositing on the catalyst to make it more aliphatic and hence more easily combustible [14]. 7.3.1 Propene Oligomerization
PGM doping has been found to be effective for enhancing the regeneration of solid HPA catalysts for propene oligomerization [25, 26]. The effect of Pd doping on coke combustion can be seen from the TGA/TPO for the coked catalyst 20% H3PW12O40/ SiO2 (Figure 7.3) [25, 26]. The catalyst has been coked by propene in a fixed-bed flow reactor at 200 C. In the absence of Pd, coke burns at about 500 C. In the case of Pd-doped catalyst, this temperature decreases; the higher the Pd loading, the lower is the temperature of coke combustion. With 2% Pd doping, coke burns at 350 C, which is well below the decomposition temperature for H3PW12O40. Using XPS and 31 P MAS NMR, it has been shown that coking does not affect the structure of H3PW12O40 [26]. From XPS, the oxidation state of tungsten is 6þ in both the fresh and coked catalyst. The 31 P chemical shift is the same for the as-made, Pddoped and coked catalysts (about 15 ppm versus 85% H3PO4, as expected for H3PW12O40), which indicates that the Keggin structure remains intact. As shown by 13 C CP/MAS NMR spectroscopy of the coked H3PW12O40 catalysts, Pd doping does affect the nature of coke depositing on the catalyst [25]. On the undoped catalyst, both aliphatic (soft) and polyaromatic (hard) coke are formed. In contrast, the Pd-doped catalyst builds only aliphatic coke, which will burn off more easily. From these results, the effect of palladium in HPA catalysts appears to be the same as in alumina or
Figure 7.3 TGA/TPO in air for Pd-doped 20% H3PW12O40/SiO2 coked by propene in a fixed-bed flow reactor at 200 C: (a) no Pd doping; (b) 1.6% Pd; (c) 2.0% Pd; (d) 2.5% Pd [25, 26].
7.3 Modification of HPA Catalysts to Enhance Coke Combustion
Figure 7.4 Performance of fresh and regenerated 2.5% Pd/20% H3PW12O40/SiO2 catalysts in gas-phase propene oligomerization (fixed-bed flow reactor, 200 C, 2% C3H6 in N2, GHSV ¼ 6000 h1; in situ regeneration by air calcination at 350 C/2 h followed by reduction with H2 at 225 C/2 h) [26].
zeolite, i.e. Pd catalyzes the combustion of coke and inhibits the formation of hard polyaromatic coke, which is more difficult to burn off [25, 26]. Palladium doping has been found to be effective in enhancing in situ regeneration of silica-supported H3PW12O40 catalyst for the gas-phase oligomerization of propene [26]. The reaction has been carried out in a fixed-bed flow reactor, yielding C12–C18 oligomers as major products. The undoped and Pd-doped H3PW12O40 catalysts both have high initial activity, but suffer from fast deactivation due to coking. The Pd-doped catalyst can be regenerated by combustion of coke at 350 C to regain its activity fully (Figure 7.4), as expected from the TPO studies. In contrast, the undoped catalyst fails to recover its activity under such conditions. 7.3.2 Friedel–Crafts Acylation
Doping with palladium and platinum has also proved effective for regeneration and recycling of HPA catalysts for Friedel–Crafts acylation in liquid-phase batch processes [27–30]. This is illustrated by studies of Fries rearrangement of phenyl acetate, yielding acylated phenols (Scheme 7.4) [29, 30]. HPAs are very efficient solid acid catalysts for this reaction, much more active than H2SO4 and acidic zeolites. The bulk acidic salt Cs2.5H0.5PW12O40 (CsPW), which is insoluble hence easily recyclable, is an especially good catalyst for this reaction. However, CsPW is deactivated by carbonaceous deposit and requires regeneration. The TGA/TPO analysis of the coked CsPW after its use for the Fries rearrangement of PhOAc shows that coke combustion is complete at about 550 C (Figure 7.5). This temperature, however, is too high for the catalyst to retain its integrity. In the case of Pd-doped CsPW (2 wt% Pd), coke is already gone at 350 C, indicating that catalyst regeneration may be possible at this temperature [27].
j159
j 7 Sustainable Heterogeneous Acid Catalysis by Heteropoly Acids
160
Scheme 7.4
Figure 7.5 TGA/TPO in air for coked catalysts CsPW and 2.1%Pd/ CsPW after use for Fries reaction of PhOAc (in nitrobenzene, 130 C, 2 h) [27].
Platinum doping has been found to be even more effective, enhancing coke combustion at a Pt loading as low as 0.3 wt% [28]. The TGA/TPO (Figure 7.6) shows two combustion peaks at a lower and higher temperature, which can be attributed to soft aliphatic coke and hard polyaromatic coke, respectively. PGM doping has been demonstrated to allow sustainable regeneration of solid HPA catalysts for Fries reaction [27, 28]. Figure 7.7 shows excellent recycling of the 0.3% Pt/CsPW catalyst in the Fries rearrangement of PhOAc. After each run, the catalyst has been separated and regenerated by air calcination at 350 C, followed by steaming at 200 C to restore the acid sites. As evidenced by FTIR spectroscopy, the Keggin structure of the CsPW remains unchanged after multiple catalyst regeneration and reuse [28]. It should be noted, however, that PGM doping could initiate side reactions, thus impairing the selectivity. Although no such effect has been observed in Friedel–Crafts acylation [27, 28], it may be the case in other reactions; therefore, care must be taken regarding the possible effect of PGM on reaction selectivity.
7.4 Inhibition of Coke Formation on HPA Catalysts
Figure 7.6 TGA/TPO in air for coked catalysts after use for Fries reaction of PhOAc (in nitrobenzene, 130 C, 2 h): (a) CsPW; (b) 0.3% Pt/CsPW; (c) 1% Pt/CsPW [28].
Figure 7.7 Catalyst reuse in Fries rearrangement of PhOAc: conversion and total acylation selectivity in successive runs [0.3% Pt/CsPW (2.3 wt%), in nitrobenzene, 130 C, 2 h] [28].
7.4 Inhibition of Coke Formation on HPA Catalysts
Catalyst regeneration is an expensive procedure. Obviously, it would be preferable to prevent the catalyst from coking in the first place to avoid its regeneration. Coke inhibition on HPA catalysts has been studied using propene oligomerization as a model reaction [25, 26]. The reaction occurs via the carbenium ion mechanism yielding propene oligomers and coke (Scheme 7.5). The oligomers may be considered as coke precursors. Addition of nucleophilic molecules, such as water, methanol
j161
j 7 Sustainable Heterogeneous Acid Catalysis by Heteropoly Acids
162
Table 7.3 Effect of additives (7 vol.%) to propene flow on coke formation on 40% H3PW12O40/SiO2 at 150 C [26].
Additive
Time on-stream (h)
Amount of coke (%)
None H2O Methanol Acetic acid
3.0 3.0 3.0 3.0
3.6 0.5 1.7 2.6
and acetic acid, has been found to affect greatly the reaction selectivity by reacting with carbenium ion intermediates to yield oxygenates at the expense of the oligomers and coke (Scheme 7.5). Water has been found to be the most effective coke inhibitor, with the amount of coke decreased sevenfold compared with the background (Table 7.3) [26].
Scheme 7.5 Acid-catalyzed oligomerization of propene.
On an industrial scale, addition of water to the reactor feed has been proved effective to prolong catalyst lifetime in BPs Avada process for the synthesis of ethyl acetate [11, 31]. In this process, ethyl acetate is produced by interaction of acetic acid with ethene in the gas phase in the presence of tungstosilicic acid supported on silica as the catalyst (Scheme 7.6).
Scheme 7.6
In 2001, this process was commercialized on a scale of 220 000 t yr1 at Hull in the UK. This is the largest ethyl acetate plant in the world. Figure 7.8 shows a schematic flowchart for the Avada process. Ethene and acetic acid are fed through evaporator to the adiabatic fixed-bed reactor containing the catalyst H4SiW12O40/SiO2. In order to achieve sustained catalyst performance, steam (3–8 mol%) is added to the reactor. The addition of water leads to reversible formation of ethanol and diethyl ether as byproducts, which are recycled back to the reactor. The product ethyl acetate is obtained in several separation and purification steps. Table 7.4 compares the performance of H4SiW12O40/SiO2 with that of other solid acid catalysts, illustrating that the HPA is a highly active catalyst for the synthesis of ethyl acetate [31]. Addition of water is essential for the stable performance of HPA catalyst. Without water, the catalyst deactivates
7.5 Reactions in Supercritical Fluids
Figure 7.8 Flowchart for BPs Avada process [11].
Table 7.4 Solid acid catalysts for the synthesis of ethyl acetate from ethene and acetic acid [31].
Catalyst
C2H4:AcOH (mol:mol)
Temperature ( C)
Pressure (bar)
Contact time (s)
H2O in feed (mol%)
STY (g L1 h1)
H-montmorillonite XE386 resin Nafion-H H-Zeolite Y H4SiW12O40/SiO2
5:1 5:1 5:1 5:1 12 : 1
200 155 170 200 180
50 50 50 50 10
4 4 4 4 2
0 0 0 0 6
144 120 102 2 380
quickly due to extensive formation of coke. The effect of water on the catalyst in this process is probably manifold. In addition to coke inhibition, water stabilizes HPA by preventing dehydration. This simple remedy has allowed BP to achieve an economically viable lifetime of their HPA catalyst [11, 31]. It should be pointed out, however, that the addition of water is not a universal cure. Although effective in the watertolerant ethyl acetate process, it is unlikely to work in other reactions that are incompatible with water such as Friedel–Crafts acylation and alkane isomerization.
7.5 Reactions in Supercritical Fluids
Heterogeneous catalysis in supercritical fluids offers considerable benefits (for a review, see [32]). The use of supercritical fluids can greatly intensify mass and heat transfer, thus enhancing the reaction rate and selectivity and also product separation.
j163
j 7 Sustainable Heterogeneous Acid Catalysis by Heteropoly Acids
164
On top of that, supercritical methodology can facilitate catalyst regeneration and increase catalyst lifetime. Supercritical fluids possess unique solvent properties which have long been utilized in separation technologies such as extraction and chromatography and are now gaining increasing interest for application in catalytic synthesis. Supercritical fluids are miscible with gases and can dissolve solids and liquids. Usually, the supercritical methodology is applied in the region near the critical point, (1.0–1.2) Tc and (1–2) Pc, where Tc and Pc are the critical temperature and pressure of the fluid, respectively. In this region, densities are close to or above the critical density of the fluid and the dissolution power of the fluid is at its maximum. Supercritical fluids exhibit considerably higher solubilities than the corresponding gases for heavy organic compounds which may deactivate catalysts and promote coking. Changing the process conditions from gas phase to dense supercritical medium can suppress this deactivation. Furthermore, enhanced diffusivity in a supercritical system can accelerate the transfer of coke precursors from the catalyst surface hence reduce the amount of coke formed. It has been reported that the lifetime of solid HPA catalysts can be significantly longer in supercritical systems than in conventional gas or liquid systems and regeneration of HPA catalysts deactivated by coking can be accomplished in supercritical systems by extracting the carbonaceous deposits from the catalyst surface [33–36]. The isomerization of n-butane has been studied in supercritical n-butane in a fixed-bed flow reactor using 20% H3PW12O40/TiO2, 20% H4SiW12O40/ TiO2 (260 C, 110 bar), sulfated zirconia (215 C, 61 bar) and H-mordenite (300 C, 138 bar) as the catalysts [34, 35]. Gas-phase isomerization on these catalysts suffers from rapid deactivation due to catalyst coking. In contrast, the supercritical system shows stable activity without catalyst deactivation, for more than 5 h on-stream in the case of HPA/TiO2. The catalysts coked in the gas-phase isomerization can be regenerated in the supercritical system at an n-butane density close to its critical value to regain almost fully their initial activity. The isomerization on HPA/TiO2 and sulfated zirconia in supercritical n-butane provides 80% selectivity to isobutane at 20–25% conversion. H-mordenite gives <40% selectivity at 25% conversion, with n-butane cracking dominating. The alkylation of isobutane with butenes competing with butene oligomerization has been studied in supercritical and conventional gas–liquid systems over a range of solid acid catalysts such as 20% H3PW12O40/TiO2, 20% H4SiW12O40/TiO2, sulfated zirconia, 15% WO3/TiO2 and 0.5% Pt/g-Al2O3-Cl [36]. In the supercritical system (140–165 C, 40–45 bar), both reactions are fast, with a steady catalytic activity. In contrast, in the liquid phase and particularly in the gas phase the reactions occur more slowly and rapid catalyst deactivation has been observed. The alkylation of cresols with tert-butanol over H3PW12O40 supported on MCM-41 mesoporous silicate in supercritical CO2 has been found to exhibit less catalyst deactivation compared with other reaction media such as hexane as solvent and without solvent. The catalyst is recyclable without a significant loss of catalytic activity and retained mesoporous structure after three recycles [37]. It should be noted, however, that the benefits gained from the supercritical methodology must be carefully weighed against the higher costs of supercritical
7.6 Cascade Reactions Using Multifunctional HPA Catalysts
process technology. Typically reactions at supercritical conditions require high pressures and the potential danger of such conditions should never be ignored [32].
7.6 Cascade Reactions Using Multifunctional HPA Catalysts
Development of cascade (tandem) processes without intermediate separation steps using multifunctional catalysts is an important strategy to carry out sustainable organic synthesis with high atom and energy efficiency [38, 39]. Multifunctional catalysts contain two or more catalytic functions (acid, base, metal, etc.) acting synergistically to carry out a multistep cascade reaction. Figure 7.9 illustrates the difference between a conventional step-by-step process with recovery after each conversion step and a cascade (one-pot) process without intermediate recovery steps. In addition to reduction of the number of separation steps, cascade processing can overcome thermodynamic limitations by combining thermodynamically unfavorable reaction steps with favorable steps, driving the cascade process forward [38, 39]. HPAs are inherently multifunctional compounds [1–4]. Their acid and redox properties can be tuned by varying the HPA composition. Solid HPAs allow for considerable alteration of their texture and can be modified to introduce another chemical function, e.g. metal function [1–4]. Solid Keggin HPAs such as H3PW12O40 doped with Pd and Pt have been reported as bifunctional catalysts for alkane isomerization [40, 41]. There is evidence that a combination of HPA acid catalysis and redox catalysis in heterogeneous cascade processes can lead to efficient processing, less sensitive to deactivation by coking compared with conventional HPA catalysis [42]. In the following sections, this approach is illustrated on several recent examples including the synthesis of methyl isobutyl ketone (MIBK), conversion of glycerol to propanediol and synthesis of ()-menthol from (þ )-citronellal using multifunctional HPA catalysts.
Figure 7.9 Conventional step-by-step process with recovery after each conversion step (a) and cascade process without intermediate recovery steps (b) [39].
j165
j 7 Sustainable Heterogeneous Acid Catalysis by Heteropoly Acids
166
7.6.1 Synthesis of MIBK
MIBK is one of the most widely used aliphatic ketones and is mainly employed as a solvent for paint and extraction [43]. Diisobutyl ketone (DIBK), a byproduct in the synthesis of MIBK, is a good solvent for a variety of natural and synthetic resins. Traditionally, MIBK is manufactured via a three-step process involving base-catalyzed aldol condensation of acetone to diacetone alcohol (DA), acid-catalyzed dehydration of DA to mesityl oxide (MO) and metal-catalyzed hydrogenation of MO to MIBK [44]. One-step synthesis of MIBK in the liquid and gas phase has been developed. This process uses bifunctional acid/redox catalysts comprising acidic resins, zeolites or zirconium phosphate with addition of platinum group metals, usually palladium ([42] and references therein). The proposed mechanism of the one-step synthesis of MIBK on a Pd/{Hþ } bifunctional catalyst is shown in Scheme 7.7 [45]. MIBK forms in three steps: acid-catalyzed condensation of acetone to DA, acid-catalyzed dehydration of DA to MO and selective hydrogenation of the C¼C bond in MO on Pd sites to yield MIBK. The first step is limited by an unfavorable equilibrium; the other two are thermodynamically favorable, making the overall process exothermic (DH ¼ 117 kJ mol1) [44]. DIBK forms in subsequent reactions of MIBK.
Scheme 7.7 Synthesis of MIBK and DIBK over bifunctional Pd/acid catalyst.
The Pd-doped acidic heteropoly salt Cs2.5H0.5PW12O40 (CsPW) has been found to be a very efficient bifunctional catalyst for the one-step conversion of acetone to MIBK in the gas and liquid phase [42]. CsPW is well known as a water-insoluble strong Brønsted acid and a versatile solid acid catalyst possessing considerable thermal stability (500 C) [1, 4]. In the liquid-phase batch process, Pd/CsPWcatalyst gives 92% MIBK selectivity and 95% total MIBK þ DIBK at 140–160 C and a very low H2 pressure of only 5–7 bar (usually 20 bar or higher) [42]. Doping the Pd/CsPW catalyst with Cu leads to a further increase in selectivity up to 95% MIBK and 98% MIBK þ DIBK. The Cu–Pd/CsPW catalyst could be reused after washing it with acetone, albeit with reduced activity. In the gas-phase process in a fixed-bed flow reactor, the Pd/CsPW catalyst gives 83% MIBK selectivity, with 91% total selectivity to MIBK þ DIBK [42]. The reaction occurs at remarkable low temperatures of 80–100 C (usually above 140 C, e.g. with Pd/zeolite catalyst). This may be explained by the strong acidity of CsPW enhancing
7.6 Cascade Reactions Using Multifunctional HPA Catalysts
Figure 7.10 Gas-phase acetone conversion and product selectivities versus time on-stream (fixed-bed flow reactor, 0.20 g 0.5%Pd/CsPW catalyst, 7.5 mL min1 H2 flow, [acetone] : [H2] ¼ 2:1, 100 C) [42].
the first two steps of the process (Scheme 7.7). The reaction clearly requires both acid and redox catalysis, which is provided by CsPW and Pd, respectively. When CsPW is used alone in the absence of palladium, MIBK is not observed, with MO being the major product. In the gas-phase reaction, Pd/CsPW exhibits very good durability. Catalyst deactivation is not observed after 25 h of continuous operation (Figure 7.10). Only a small amount of coke (1%) is formed in the reaction, which does not affect the catalyst performance. Such minor coking may be due to the efficient removal of coke precursors by in situ hydrogenation on Pd. It should be noted that in the absence of H2 the catalyst loses its activity in about 4 h, with MO being the main reaction product. Therefore, in this heterogeneous cascade process, combination of HPA acid catalysis with Pd-catalyzed hydrogenation is effective to protect the catalyst from deactivation by coking. 7.6.2 Hydrogenolysis of Glycerol to Propanediol
Production of marketable chemicals via catalytic transformation of renewable bioresources is a global challenge [46]. Glycerol is one of the top 12 building block chemicals that can be derived from plant sources [46]. In addition, the development of biodiesel production by transesterification of vegetable oils makes large amounts of glycerol available as a reaction by-product, about 10 wt% of the biodiesel produced [47]. The availability and low price of glycerol make it a promising feedstock for producing a wide range of value-added chemicals. Synthesis of propanediols, 1,2PDO and 1,3-PDO, from glycerol has attracted significant interest [48–53]. 1,2-PDO is an important commodity chemical which finds use as an antifreeze, aircraft deicer and lubricant. 1,3-PDO is copolymerized with terephthalic acid to produce polyesters, which are used for manufacturing carpet and textile fibers exhibiting strong chemical and light resistance [48].
j167
j 7 Sustainable Heterogeneous Acid Catalysis by Heteropoly Acids
168
1,2-PDO and 1,3-PDO are currently produced from petroleum derivatives by chemical catalytic routes: 1,2-PDO from propylene oxide and 1,3-PDO from ethylene oxide or acrolein [44]. These diols can be produced by an alternative route involving hydrogenolysis of glycerol. A number of patents and papers have disclosed the hydrogenolysis of glycerol in the presence of homogeneous and heterogeneous catalysts ([48–53] and references therein). This reaction has been carried out in liquidphase batch systems at 110–260 C and an H2 pressure up to 300 bar, yielding 1,2- and 1,3-PDO together with 1- and 2-propanol, ethylene glycol, ethanol and methane as byproducts. More efficient hydrogenolysis procedures that have been reported recently employ CuO–ZnO (200 C, 42 bar, 12 h) [48, 52], copper chromite (200 C, 14 bar, 24 h) [49] and 5%Ru/C þ Amberlyst-15 (120 C, 80 bar, 10 h) [50, 51] as heterogeneous catalysts. These procedures, however, have serious drawbacks such as high temperature and pressure of the process, low selectivity to propanediols and low catalyst activity, hence low glycerol conversion and long reaction times. The hydrogenolysis of glycerol to 1,2-PDO and 1,3-PDO is suggested to proceed via dehydration of glycerol to acetol and 3-hydroxypropanal by acid catalysis followed by catalytic hydrogenation (Scheme 7.8), which has been supported by the observation of acetol amongst the reaction products and its selective hydrogenation to 1,2PDO [49–51]. Alternatively, dehydrogenation of glycerol to glyceraldehyde followed by dehydration to 2-hydroxyacrolein and hydrogenation to yield 1,2-PDO (Scheme 7.9) has been suggested [53, 54]. These mechanisms imply that bifunctional acid/hydrogenation catalysis should be an effective course to carry out the glycerol hydrogenolysis in a one-pot system. Recently, Miyazawa and co-workers [50, 51] used a mixture of 5%Ru/C as a hydrogenation catalyst and Amberlyst-15 acid resin as an acid catalyst for glycerol hydrogenolysis in one-pot system. However, the catalyst performance is rather poor: 1,2-POD together with 1,3-POD were obtained with 55 and 5% selectivity, respectively, at 13% glycerol conversion. A serious drawback to Amberlyst-15 is its decomposition above 120 C [50]. In this system, Rh/C also exhibits some catalytic activity, whereas Pd/C and Pt/C are almost inactive.
Scheme 7.8 Hydrogenolysis of glycerol to 1,2-PDO and 1,3-PDO.
Scheme 7.9 Hydrogenolysis of glycerol to 1,2-PDO via glyceraldehyde.
7.6 Cascade Reactions Using Multifunctional HPA Catalysts Table 7.5 Hydrogenolysis of glycerol over bifunctional catalysts [55].a
Selectivity (%) Catalyst (4 wt%)
Conversion (%)
Temperature ( C)
1,2-PDO
1,3-PDO
Acetol
1-PO
EG
2-PO
Cs/PW Ru/CsPW Ru/CsPW Rh/CsPW
180 150 180 180
<1 21 23 6.3
0 95.8 73.6 65.4
0 0 0 7.1
Trace 0 5.7 0
0 4.2 4.3 27.5
0 0 11.9 0
0 0 4.5 0
a
Reaction conditions: 5 bar H2 pressure, 2 wt% glycerol aqueous solution, 1 h.
Ruthenium-doped (5 wt%) CsPW has been reported to be an active bifunctional catalyst for the one-pot hydrogenolysis of glycerol to 1,2-PDO in the liquid phase, providing 96% selectivity to 1,2-PDO at 21% glycerol conversion at 150 C and an unprecedented low hydrogen pressure of 5 bar (Table 7.5) [55]. Rhodium catalyst, 5% Rh/CsPW, although less active, shows considerable selectivity to 1,3-PDO (7.1%), with 1,2-PDO being the main product (65%). Acetol, 1- and 2-propanol (1-PO and 2-PO), ethylene glycol (EG) and methane are found amongst the by-products, similarly to the previous reports [48–53]. Both catalyst functionalities, metal hydrogenation (Ru) and acidity (CsPW), are essential for the efficiency of Ru/CsPW catalyst, acting synergistically in glycerol hydrogenolysis to yield 1,2-PDO. Without ruthenium, CsPW is not active. In the absence of CsPW, ruthenium (e.g. Ru/C) exhibits some activity in 1,2-PDO formation [50, 51]. Supporting ruthenium on the CsPW greatly enhances its activity in this reaction. Catalyst deactivation has been noted, which is probably caused by reduction of CsPW. Catalyst coking has not been observed in this reaction. As regards the reaction mechanism, the absence of acetol amongst the products when the reaction is carried out with CsPW without Ru present, although it forms in the presence of Ru/CsPW (Table 7.5), is inconsistent with the mechanism of 1,2-PDO formation presented in Scheme 7.8. The alternative mechanism shown in Scheme 7.9 is, therefore, more likely. The question is how acetol forms in the presence of Ru/CsPW. Scheme 7.10 shows the proposed mechanism of glycerol hydrogenolysis over Ru/CsPW, detailing the formation of 1,2-PDO [55]. In this mechanism, 1,2-PDO forms from 2-hydroxyacrolein by two routes, including hydrogenation of either the C¼C or C¼O bond in 2-hydroxyacrolein over Ru sites. The C¼C hydrogenation gives 2-hydroxypropanal. Acetol is formed by the C¼O hydrogenation followed by enol–ketone rearrangement. Both the acetol and 3-hydroxypropanal intermediates are further hydrogenated to yield 1,2-PDO. The hydrogenolysis of glycerol to 1,2-PDO over Ru/C þ Amberlyst ¼ 15 (120 C, 80 bar H2 pressure, 20% glycerol) has been suggested to proceed via the mechanism shown in Scheme 7.8 [50, 51], although only traces of acetol (<0.01% yield) have been observed. It is conceivable that this reaction also occurs via Scheme 7.10 as in our case and the negligible acetol yield may be due to the much higher H2 pressure in this system. In contrast, with copper chromite catalyst (200 C, 14 bar H2, 80% glycerol), 1,2-PDO is likely to form via Scheme 7.8 because in this case
j169
j 7 Sustainable Heterogeneous Acid Catalysis by Heteropoly Acids
170
acetol has been obtained with a high yield by interaction of glycerol with copper chromite in the absence of H2 and further selectively hydrogenated to 1,2-PDO [49].
Scheme 7.10 Proposed mechanism for glycerol hydrogenolysis over Ru/CsPW [55].
7.6.3 Synthesis of Menthol from Citronellal
Menthol is important ingredient of various cosmetic, pharmaceutical and other specialty products, hence its synthesis is of considerable interest [56, 57]. Amongst eight isomers of menthol, only ()-menthol is the desired product, which possesses a characteristic odor and produces a physiological cooling effect. One of the major synthetic routes to ()-menthol is the Takasago process, involving acid-catalyzed cyclization of (þ )-citronellal to ()-isopulegol in the presence of aqueous ZnBr2 followed by Ni-catalyzed hydrogenation (Scheme 7.11) [57]. In addition to ()-isopulegol, the cyclization may produce three stereoisomers together with other byproducts. The present industrial practice of citronellal cyclization requires large quantities of ZnBr2 as a homogeneous catalyst, which causes environmental problems due to the formation of large amounts of waste upon catalyst separation. The use of heterogeneous catalysis for this reaction, therefore, would be a much cleaner option.
Scheme 7.11 Synthesis of menthol from citronellal.
Recently, one-pot synthesis of menthol from citronellal, with simultaneous cyclization and hydrogenation over a bifunctional acid/hydrogenation catalyst, has attracted interest. Catalysts such as Ru–ZnBr2/SiO2 [58], Cu/SiO2 [59] and Ir/H-Beta zeolite [60] have been used. Although simpler, the one-pot syntheses are lengthy procedures or require a high catalyst:substrate ratio, often with a relatively low ()-menthol yield. Thus with 3%Ir/H-Beta, a 75% yield of ()-menthol in a 30 h
7.6 Cascade Reactions Using Multifunctional HPA Catalysts
reaction at 80 C has been obtained [60]. The 10%Ru–ZnBr2/SiO2 catalyst gives 85% ()-menthol yield, with less than 1 g of substrate converted per gram of catalyst [58]. In the one-pot synthesis via Scheme 7.11, the cyclization must be fast to avoid hydrogenation of citronellal. The reported procedures [58–60], however, have used catalysts with relatively weak acid sites. As a result, the cyclization is slow and only moderately active hydrogenation catalysts can be used to prevent hydrogenation of citronellal. Therefore, for process intensification, it would be beneficial to apply a bifunctional catalyst with stronger acid sites in addition to more active metal sites such as palladium. Silica-supported heteropoly acid H3PW12O40 has been reported as an efficient solid acid catalyst for the cyclization of (þ )-citronellal to ()-isopulegol [61]. More recently, silica-supported H3PW12O40 doped with 5 wt% palladium has been reported as an active catalyst for the one-pot transformation of (þ )-citronellal to menthol via acidcatalyzed cyclization followed by Pd-catalyzed hydrogenation, with a 92% yield of menthol at 100% citronellal conversion and 85% stereoselectivity for the desired ()menthol [62]. The reaction occurs in cyclohexane at 70 C and 35 bar H2 pressure. This result is on the level with or better than those reported so far [58–60]. The time course for menthol synthesis over Pd–H3PW12O40/SiO2 is shown in Figure 7.11. As expected, the yield of isopulegol intermediate passes through a maximum and the formation of menthol goes through an induction period. It is important that no products of citronellal hydrogenation have been found. This indicates that in this system citronellal cyclization occurs much faster than the hydrogenation of isopulegol. The reaction appears to be truly heterogeneous. No HPA leaching from the catalyst into cyclohexane solution has been observed. The catalyst could be recycled several times, albeit with gradually decreasing activity.
Figure 7.11 Conversion of (þ )-citronellal to menthol [85% ()menthol selectivity] on 5% Pd–20%PW/SiO2 catalyst (0.15 wt%) in cyclohexane at 70 C and 35 bar H2 pressure [62].
j171
j 7 Sustainable Heterogeneous Acid Catalysis by Heteropoly Acids
172
7.7 Conclusion
Heterogeneous acid catalysis by heteropoly acids offers substantial economic and environmental benefits. However, the relatively low thermal stability of HPAs is a serious problem to HPA catalysis, making regeneration (decoking) of solid HPA catalysts difficult. Several approaches can be instrumental in overcoming deactivation of HPA catalysts to achieve sustainable catalyst performance. One of these could be the development of new HPA materials possessing high thermal stability. Recent studies have provided some new solid acid catalysts, for example oxide composites comprising tungsten(VI) polyoxometalates and niobium(V), zirconium(IV) or titanium(IV) oxides, exhibiting good regeneration and reuse. However, these catalysts have relatively weak acid sites hence a lower catalytic activity compared with the standard HPA catalysts. Work should be continued to obtain HPA materials possessing stronger acid sites. Another approach is the modification of HPA catalysts by platinum group metals to enhance coke combustion. This method has proved effective for in situ and ex situ catalyst regeneration by combustion of coke without destroying the structure of HPA. For acid-catalyzed processes that tolerate the presence of water, addition of water to the reactor feed can effectively inhibit coke formation and prolong the catalyst lifetime (BPs Avada process). The life of an HPA catalyst can also be longer when the reaction is carried out in a supercritical system. However, the high cost of supercritical process technology should be taken into account. Finally, HPAs can be used with high efficiency and better stability towards deactivation within heterogeneous multifunctional catalysts for cascade processes involving acid and redox catalysis.
References 1 Okuhara, T., Mizuno, N. and Misono, M. (1996) Advances in Catalysis, 41, 113. 2 (a) Hill, C.L. (ed.) (1998) Polyoxometalates Chemical Reviews, 98; (b) Kozhevnikov, I.V. (1998) Chemical Reviews, 98, 171;Mizuno, N. and Misono, M. (1998) Chemical Reviews, 98, 199. 3 (a) Moffat, J.B. (2001) Metal–Oxygen Clusters. The Surface and Catalytic Properties of Heteropoly Oxometalates, New York; Kluwer. (b) Izumi, Y., Urabe, K. and Onaka, M. (1992) Zeolite, Clay and Heteropoly Acid in Organic Reactions, Kodansha/VCH, Tokyo. 4 Kozhevnikov, I.V. (2002) Catalysts for Fine Chemicals, Vol. 2. Catalysis by Polyoxometalates, John Wiley & Sons, Ltd, Chichester.
5 Neumann, R. (1998) Progress in Inorganic Chemistry, 47, 317. 6 Misono, M. (2005) Catalysis Today, 100, 95. 7 (a) Yadav, G.D. (2005) Catalysis Surveys from Asia, 9, 117. (b) Timofeeva, M.N. (2003) Applied Catalysis A-General, 256, 19. 8 Kozhevnikov, I.V. (2007) Journal of Molecular Catalysis A-Chemical, 262, 86. 9 Pope, M.T. (1983) Heteropoly and Isopoly Oxometalates, Springer, Berlin. 10 Pope, M.T. and Muller, A. (eds) (2001) Polyoxometalate Chemistry from Topology via Self-Assembly to Aplications, Kluwer, Dordrecht. 11 Dobson, I.D. (2003) Green Chemistry, 5, G78.
References 12 Twigg, M.V. (ed.) (1989) Catalyst Handbook, 2nd edn, Wolfe, London. 13 Furimsky, E. and Massoth, F.E. (1993) Catalysis Today, 17, 537. 14 Barbier, J. (1987) Studies in Surface Science and Catalysis, 34, 1. 15 Guisnet, M. and Magnoux, P. (1994) Studies in Surface Science and Catalysis, 88, 53. 16 Karge, H.G. (2001) Studies in Surface Science and Catalysis, 137, 707. 17 Okumura, K., Yamashita, K., Hirano, M. and Niwa, M. (2005) Journal of Catalysis, 234, 300. 18 Okumura, K., Yamashita, K., Yamada, K. and Niwa, M. (2007) Journal of Catalysis, 245, 75. 19 Devassy, B.M. and Halligudi, S.B. (2005) Journal of Catalysis, 236, 313. 20 Devassy, B.M., Lefebvre, F. and Halligudi, S.B. (2005) Journal of Catalysis, 231, 1. 21 Devassy, B.M. and Halligudi, S.B. (2006) Journal of Molecular Catalysis A-Chemical, 253, 8. 22 Sawant, D.P., Vinu, A., Lefebvre, F. and Halligudi, S.B. (2007) Journal of Molecular Catalysis A-Chemical, 262, 98. 23 Kumbar, S.M., Shanbhag, G.V., Lefebvre, F. and Halligudi, S.B. (2006) Journal of Molecular Catalysis A-Chemical, 256, 324. 24 Bhatt, N. and Patel, A. (2007) Journal of Molecular Catalysis A-Chemical, 264, 214. 25 Siddiqui, M.R.H., Holmes, S., He, H., Smith, W., Coker, E.N., Atkins, M.P. and Kozhevnikov, I.V. (2000) Catalysis Letters, 66, 53. 26 Kozhevnikov, I.V., Holmes, S. and Siddiqui, M.R.H. (2001) Applied Catalysis A-General, 214, 47. 27 Kozhevnikova, E.F., Rafiee, E. and Kozhevnikov, I.V. (2004) Applied Catalysis A-General, 260, 25. 28 Musawir, M., Kozhevnikova, E.F. and Kozhevnikov, I.V. (2007) Journal of Molecular Catalysis A-Chemical, 262, 93. 29 Kozhevnikova, E.F., Derouane, E.G. and Kozhevnikov, I.V. (2002) Chemical Communications, 1178. 30 Kozhevnikov, I.V. (2003) Applied Catalysis A-General, 256, 3.
31 Howard, M.J., Sunley, G.J., Poole, A.D., Watt, R.J. and Sharma, B.K. (1999) Studies in Surface Science and Catalysis, 121, 61. 32 Baiker, A. (1999) Chemical Reviews, 99, 453. 33 Gayraud, P.Y., Stewart, I.H., DerouaneAbd Hamid, S.B., Essayem, N., Derouane, E.G. and Vedrine, J.C. (2000) Catalysis Today, 63, 223. 34 Bogdan, V.I., Klimenko, T.A., Kustov, L.M. and Kazansky, V.B. (2004) Applied Catalysis A-General, 267, 175. 35 Bogdan, V.I., Klimenko, T.A., Kustov, L.M. and Kazansky, V.B. (2004) Kinetics and Catalysis, 45, 890. 36 Bogdan, V.I. and Kazansky, V.B. (2005) Kinetics and Catalysis, 46, 834. 37 Kamalakar, G., Komura, K. and Sugi, Y. (2006) Applied Catalysis A-General, 310, 155. 38 Wasilke, J.-C., Obrey, S.J., Baker, R.T. and Bazan, G.C. (2005) Chemical Reviews, 105, 1001. 39 Bruggink, A., Schoevart, R. and Kieboom, T. (2003) Organic Process Research & Development, 7, 622. 40 Baba, T., Hasada, Y., Nomura, M., Ohno, Y. and Ono, Y. (1996) Journal of Molecular Catalysis A-Chemical, 114, 247. 41 Okuhara, T. (2003) Applied Catalysis AGeneral, 256, 213. 42 Hetterley, R.D., Kozhevnikova, E.F. and Kozhevnikov, I.V. (2006) Chemical Communications, 782. 43 Hoelderich, W.F. (1988) Studies in Surface Science and Catalysis, 41, 83. 44 Weissermel, K. and Arpe, H.-J., (1997) Industrial Organic Chemistry, 3rd edn, Wiley-VCH, Weinheim, Germany. 45 Melo, L., Magnoux, P., Giannetto, G., Alvarez, F. and Guisnet, M. (1997) Journal of Molecular Catalysis A-Chemical, 124, 155. 46 Ragauskas, A.J., Williams, C.K., Davison, B.H., Britovsek, G., Cairney, J., Eckert, C.A., Frederick, W.J., Hallett, J.P., Leak, D.J., Liotta, C.L., Mielenz, J.R., Murphy, R., Templer, R. and Tschaplinski, T. (2006) Science, 311, 484. 47 Strivastava, A. and Prasad, R. (2000) Renewable and Sustainable Energy Reviews, 4, 111.
j173
j 7 Sustainable Heterogeneous Acid Catalysis by Heteropoly Acids
174
48 Chaminand, J., Djakovitch, L., Gallezot, P., Pinel, C. and Rosier, C. (2004) Green Chemistry, 6, 359. 49 Dasari, M.A., Kiatsimkul, P.P., Sutterlin, W.R. and Suppes, G.J. (2005) Applied Catalysis A-General, 281, 225. 50 Miyazawa, T., Kusunoki, Y., Kuminori, K. and Tomishige, K. (2006) Journal of Catalysis, 240, 213. 51 Miyazawa, T., Koso, S., Kuminori, K. and Tomishige, K. (2007) Applied Catalysis AGeneral, 318, 244. 52 Wang, S. and Liu, H. (2007) Catalysis Letters, 117, 62. 53 Maris, E.P. and Davis, R.J. (2007) Journal of Catalysis, 249, 328. 54 Montassier, C., Menezo, J.C., Hoang, L.C., Renaud, C. and Barbier, J. (1991) Journal of Molecular Catalysis, 70, 99. 55 Alhanash, A., Kozhevnikova, E.F. and Kozhevnikov, I.V., (2008) Catalysis Letters, 120, 307. 56 Gerhertz, W. (ed.) (1988) Ullmanns Encyclopedia of Industrial Chemistry, VCH Verlag GmbH, Weinheim.
57 Pybus, D.H. and Sell, C.S. (1999) The Chemistry of Fragrances, Royal Society of Chemistry, Cambridge. 58 Milone, C., Gangemi, C., Neri, G., Pistone, A. and Galvagno, S. (2000) Applied Catalysis A-General, 199, 239. 59 Ravasio, N., Poli, N., Psaro, R., Saba, M. and Zaccheria, F. (2000) Topics in Catalysis, 13, 195. 60 Iosif, F., Coman, S., P^arvulescu, V., Grange, P., Delsarte, S., De Vos, D. and Jacobs, P. (2004) Chemical Communications, 1278. 61 da Silva, K.A., Robles-Dutenhefner, P.A., Sousa, E.M.B., Kozhevnikova, E.F., Kozhevnikov, I.V. and Gusevskaya, E.V. (2004) Catalysis Communications, 5, 425. 62 da Silva Rocha, K.A., Robles-Dutenhefner, P.A., Sousa, E.M.B., Kozhevnikova, E.F., Kozhevnikov, I.V. and Gusevskaya, E.V. (2007) Applied Catalysis A-General, 317, 171.
j175
8 The Kinetics of TiO2-based Solar Cells Sensitized by Metal Complexes Anthony G. Fitch, Don Walker, and Nathan S. Lewis
8.1 Introduction
World annual energy demand is currently 0.5 ZJ (1 ZJ ¼ 1 1021 J) with a mean consumption rate of 15 TW [1]. Energy use has risen on average by 2.4% per year for the past 25 years and projections are that world energy use will increase by 50% by 2030 [1]. The world now obtains >80% of its energy from fossil fuels. With the prospect of significant climate change and other environmental impacts due to the buildup of CO2 in the atmosphere, the need for a significant source of carbon neutral energy has become apparent. Currently, few renewable energy options contribute significantly to our energy needs; however, the sun provides 1.2 105 TW of power to the Earth and in 1 h more than 0.4 ZJ of solar energy strikes the Earth [2]. Hence in a little more than 1 h enough solar energy falls on the Earth to provide the yearly energy needs of the planet. We would only need to convert 0.01% of the incoming solar radiation to supply the worlds energy needs, using, for example, single-crystal photovoltaic (PV) technology, which can have solar conversion efficiencies as high as 40% [3]. Therefore, one may wonder what the energy crisis is and why we are still using fossil fuels. The answers, of course, lie in the economics of solar conversion. Currently, the cost of electricity from high-efficiency PVs is about five times more expensive than electricity from fossil fuels and PVenergy is even more expensive than fossil fuels, on an energy basis. In fact, PVs produce energy that costs 20–50 times that produced from fossil fuels [2]. The problem with photovoltaic devices is that the cost of materials to make the cell is high. If one considers silicon, the raw materials are inexpensive, since Si is the eighth most common element. Unfortunately, the solar light absorptivity of silicon is low and thus a thick piece (100 mm) of silicon is needed to absorb most of the incoming solar radiation. For planar solar cells, this means that the electron–hole pair produced by the absorption of a solar photon must survive long enough for the electron or hole to transverse the thick slab of Si and reach the semiconductor junction where electron–hole separation can be achieved. Therefore, the silicon used in solar cells must be of very high purity and well passivated, so that the rate of electron–hole recombination is slow and the excitation energy is not lost.
Handbook of Green Chemistry, Volume 2: Heterogeneous Catalysis. Edited by Robert H. Crabtree Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32497-2
j 8 The Kinetics of TiO -based Solar Cells Sensitized by Metal Complexes
176
2
For single-crystal photovoltaics, high-purity materials are therefore needed to obtain high efficiencies, resulting in a requirement for expensive processing techniques. New solar energy-conversion methods are therefore needed to compete with the cost of energy from fossil fuels. Liquid junction, dye-sensitized solar cells (DSSCs) constructed from relatively impure and unprocessed TiO2 are one type of solar cell that may circumvent the requirement for production and use of expensive materials in the construction of solar energy-conversion devices.
8.2 History
Dye sensitization has been known for over a century, starting with the field of photography. In 1874, the first example of dye sensitization, sensitizers were mixed with silver halide crystals to extend the photoresponse of silver halide from 460 nm, due to its 2.7–3.2 eV bandgap, to longer wavelengths [4]. Shortly thereafter, the same silver halide semiconductor was used to create the first dye-sensitized photoelectrode, using erythrosine dye [5]. However, little progress was made in the next 100 years in exploiting dye sensitization to produce usable energy. In 1968–69, the wide bandgap semiconductor ZnO was sensitized by Tributsch and Gerisher and exhibited a photocurrent response [6, 7]. It was later shown that a cell made from ZnO sensitized with Rose Bengal, with hydroquinone used as a reducing agent, yielded a photocurrent with an internal quantum efficiency close to unity [8]. Even with the high internal quantum efficiency, the overall energy-conversion efficiency of these devices was low, less than 1%. The low efficiency was a result of the use of singlecrystal materials. The use of a planar, single-crystal electrode only allows a single monolayer of sensitizer to be in close contact with the semiconductor. This mandates that only a monolayer can absorb light and effectively inject charge into the semiconductor. At the wavelength of maximum absorbance, even the use of a sensitizer with an absorptivity of 105M1 cm1 will result in <5% absorbance in the cell. A monolayer of dye on a planar surface is thus not sufficient to absorb a significant amount of the solar radiation. In 1991, ORegan and Gratzel developed a new type of DSSC [9]. The single-crystal electrodes used previously were replaced with a colloidal TiO2 film that consisted of nanometer-sized particles that had been sintered to form a nanoporous film on a transparent, conducting electrode. Also, the dye was anchored to the TiO2 surface through covalent chemical bonds and not merely physisorbed on the surface of the semiconductor. The colloidal film produced a semiconductor with a surface area roughly 1000 times larger than that of the planar electrode. Thus, a monolayer of dye absorbed on this electrode increased the light absorbency of the film by that same ratio. This approach resulted in an overall solar energy-conversion efficiency of 7%. This event sparked a large research effort aimed at understanding and optimizing nanoporous liquid junction DSSCs. However, despite almost 20 years of work, the current efficiencies of the DSSC are at most 50% better than those of the original cells and the overall design of the cell has changed relatively little.
8.3 DSSC Design
8.3 DSSC Design
The basic design of the DSSC is shown in Figure 8.1. The cell consists of a working or back electrode and a counter electrode immersed in a solution that contains a redox mediator species and various electrolytes. The working electrode is composed of a base layer of a transparent, conductive oxide (TCO) on glass. The commonly used TCO is a fluorine-doped tin oxide (FTO), but other TCOs such as indium-doped tin oxide (ITO) and aluminum-doped zinc oxide have been used [10–14]. The working electrode also has an active layer formed from a deposited semiconductor film. The most widely used semiconductor film is nanoparticulate TiO2, due to its performance, stability, ease of preparation and low cost. The TiO2 film is made from a suspension of nanoparticles and can be deposited by numerous methods such as screen printing, a doctor-blade technique, or spin coating. The films are then heated to moderately high temperatures (500 C) to sinter the TiO2 particles, leaving a porous network of nanoparticles. This type of electrode allows enough surface area that a monolayer of dye on the surface of the particles will absorb a significant amount of the incident solar radiation. The sensitizers are usually covalently attached to the surface through carboxylate linkages. The most commonly studied sensitizers are Ru polypyridyl dyes, due to the broad metal-to-ligand charge-transfer (MLCT) bands and the unmatched performance of these dyes in DSSCs to date. The working electrode is placed in a solution that contains a redox shuttle, that moves electrons between the working and counter electrodes. The best performing and most widely used redox shuttle is I3/I. Small cations are added to the solution to improve the quantum yield of the system. The small cations have been shown to affect the energetics of the DSSC and the rates of the different processes [15–17]. Lithium is the most commonly used cation. Other additives, such as tert-butylpyridine, pyridine and 4-guanidinobutyric acid, are also used to improve the efficiency [18–21]. A Pt counter electrode is used to complete the cell, due to the fast kinetics of I3/I reduction at the platinum surface. With this design, efficiencies as high as 11% have been achieved by optimization of all of the rates of the DSSC [22].
Figure 8.1 Illustration of a DSSC.
j177
j 8 The Kinetics of TiO -based Solar Cells Sensitized by Metal Complexes
178
2
8.4 Function of the DSSC
The performance of the DSSC depends on the control of all of the electron transfer rates and optimization of the cell potential. The various rates are outlined in Figure 8.2. The first step in producing a photocurrent is dye excitation, k1, which produces an excited state of the dye that can inject an electron into the TiO2 film, k2. For the excited state to inject an electron, the reduction potential of the excited dye needs to be more negative than the potential of the bottom of the conduction band of the TiO2. Injection takes place on a femtosecond to picosecond time scale [23–28]. Many dyes have been shown to have very high quantum yields for injection of electrons into TiO2. Once in the conduction band, the electron diffuses, k4, through the TiO2 network and is collected at the back contact. The movement of charge through the semiconductor film is viewed as electron hopping between shallow trap sites in the film. After being collected, the electrons pass through the external circuit, perform work and arrive at the counter electrode. At the counter electrode, the electron reduces the redox mediator, which diffuses through the solution and ultimately reduces the oxidized dye, k6. To produce cells that exhibit high currents, the numerous deleterious rates that short-circuit the cell need to be minimized. Three competitions dominate such optimization: (1) injection of electrons, k2, competes with excited-state relaxation, k–1; (2) regeneration of the oxidized dye by the redox mediator, k6, competes with recombination between the electron and the oxidized dye, k3; and (3) electron collection by the back electrode, k4, competes with reduction of the oxidized solution redox mediator by electrons in the film, k5. For charge injection, the injection rate must be fast with respect to excited-state relaxation back to the ground state, k1, and excited-state quenching by solution species. Once the electron is in the conduction band of TiO2, a loss in current can occur by recombination of the electron with the oxidized dye, k3, or by recombination with the redox couple in solution, k5. With the high surface area of the semiconductor film and its intimate contact with the solution, it is not surprising that this competition plays a key role in determining the efficiency
Figure 8.2 Rate diagram for the kinetics processes involved of a DSSC.
8.5 Performance of a DSSC
of the cell. Recombination of the electron with the oxidized dye is prevented by fast regeneration of the dye by the solution redox mediator. Recombination to the redox mediator is controlled by selecting a redox couple with the appropriate kinetics at the TiO2 electrode. The reason for the high efficiency will be further explored in the following sections, along with a detailed discussion of all of the different processes.
8.5 Performance of a DSSC
The performance of the DSSC is determined by measuring, in the light and dark, the current density response of the cell to an externally applied bias. This behavior is known as a current density, J, vs voltage, V, curve. A typical current density–voltage plot is presented in Figure 8.3. When the TiO2 electrode is negative with respect to the counter electrode, the TiO2 is said to be under negative bias. In the dark, with the cell at short-circuit (i.e. an applied bias of zero), the system comes to equilibrium and the Fermi level of the TiO2 equilibrates with the electrochemical potential of the solution, as defined by the solution redox couple. Under such conditions, no net current flows. When a small negative voltage is applied, the Fermi level of the working electrode shifts to more negative potentials. Little or no current flows, due to the large overpotential needed to reduce I3 at the TiO2 electrode. As more negative potentials are applied to the cell, current will begin to flow and the current increases rapidly in magnitude with increasingly negative applied potentials. The resulting current is defined as negative in sign, with electrons flowing from the back electrode to the solution. In the light, when the cell is at short-circuit, a positive current is observed due to the photogenerated electrons injected into the TiO2. This current density is denoted as the short-circuit current density, Jsc. Initially, little change in the photocurrent occurs with applied negative potentials, but when the potential becomes negative enough to reduce I3 rapidly, the net current drops to zero. The potential at which zero current is observed determines the open-circuit voltage, Voc, of the solar cell.
Figure 8.3 Current density versus voltage plot for a DSSC in the dark and under illumination.
j179
j 8 The Kinetics of TiO -based Solar Cells Sensitized by Metal Complexes
180
2
The power produced by the cell at any applied voltage is given by the product of the cell voltage and the observed current density. The maximum efficiency of the cell is calculated from the position of the current density-voltage curve that yields the maximum power, divided by the power of the incoming light. The efficiency is given by h¼
ðJ Cell V app Þmax J sc V OC f f ¼ P in P in
ð8:1Þ
where the fill factor, ff, is defined as ff ¼
ð J Cell V app Þmax J sc V OC
The fill factor indicates the shape of the J–V curve, has a value between 1 and 0 and is the ratio of the shaded region in Figure 8.3 to the rectangle defined by Jsc and Voc. The overall device performance of a DSSC is similar to that of traditional solid-state photovoltaics, even though there are significant differences in the kinetics that underlie the operation of the two devices. Some other useful measures of the DSSC performance are the external and internal quantum yield and the light-harvesting efficiency. The light-harvesting efficiency, LHE, indicates the extent of light absorption of a dye-coated electrode and is calculated as LHE ¼ ð110ad Þ 100
ð8:2Þ
where a is the absorption coefficient of the dye-coated TiO2 electrode and d is the film thickness. To achieve high efficiencies, the LHE must be high for a large portion of the solar spectrum. The external quantum yield, Fext, is given by equation 3: Fext ¼ ðLHE=100ÞFinj hcoll
ð8:3Þ
where Finj is the injection quantum yield and hcoll is the charge-collection quantum yield. The quantity Fext is related to the incident photon-to-current efficiency, IPCE. Because Fext is dependent on the LHE, the charge-collection and injection efficiency, it is difficult to determine the cause of a low Fext solely from measurement of the current. The internal quantum yield, Fint, is given by Fint ¼ Finj hcoll
ð8:4Þ
and is independent of the LHE of the dye. The value of Fint measures the quantum yield of conversion of absorbed photons into electrons. The value of Fint provides one measure of whether the kinetics of the DSSC are optimized.
8.6 Kinetics Processes
The current–voltage properties and the kinetics processes of solid-state photovoltaics are well understood. However, the kinetics of the DSSC are significantly more
8.7 Charge Injection
complicated and have still not been fully elucidated. In the remaining sections, we will examine the kinetics of charge injection, k2, dye regeneration, k6, and recombination of electrons in TiO2 with the oxidized form of the dye, k3. The other processes are also important to the device performance, but will not be discussed here. In studying the kinetics of the various processes, pump–probe laser techniques are often used to measure the rates of the various reactions. For recombination and regeneration, only nanosecond time scales are required, but for charge-injection studies, picosecond and femtosecond lasers are required.
8.7 Charge Injection
For efficient injection, the sensitizer must be in intimate contact with the TiO2 surface, to provide good coupling between the two phases. For most molecular sensitizers, this is accomplished by a covalent linkage to the TiO2 surface. Simple attachment to the surface does not suffice to assure good electronic coupling to the TiO2. Linkers such as carboxylates, phosphonates, sulfonates and many others have been used successfully [14, 29–31]. Different linkers affect the energetics of the dye and the interaction between the excited state and conduction band of TiO2, thus having a significant effect on the rate of injection. Even with the same type of linker, the coupling can be drastically reduced by the insertion of one unsaturated carbon between the linker and the dye. For example, in one system of concern, a change of one carbon resulted in a >200-fold decrease in the injection rate [32]. Also, the position of the linker can affect the efficiency of the DSSC. For bpy (2,20 -bipyridyl) ligands, relocation of the carboxylate groups from the 4,40 to the 5,50 positions resulted in a significant decrease in Fext [33]. Thus the linker does not merely secure the dye to TiO2, but must also provide good electronic coupling of the excited state to TiO2. Carboxylates are by far the most widely used linkers and enable near unity injection yields for a wide variety of dyes. However, due to the instability of the ester bond, a more stable bond is desirable. The linker also has a significant effect on the recombination rate, as discussed later. Even with good coupling, the process of injection takes a certain amount of time. The excited state must live long enough to facilitate collection of the carriers. For metal complexes, the lifetime is extended by having a lower lying triplet state with a lifetime of nanoseconds or longer. The triplet state quickly converts the excited electron from a spin-allowed to a spin-forbidden deactivation process, resulting in extended lifetimes. It would be ideal not to have this lower triplet, but the lifetime of most singlet excited states in transition metal complexes is too short to allow for high values of Fint. For example, when Fe(dcb)2(CN)2, where dcb is 4,40 -dicarboxylic acid 2,20 -bipyridine, was used, there is no lower lying triplet and the overall efficiency of the DSSC is significantly reduced [34, 35]. This behavior illustrates that some energy is needed loss was extend the excited-state lifetime and allow for charge separation. Injection rates have been shown to consist of a fast femtosecond component, assigned to singlet injection, and a slower picosecond rate, assigned to triplet
j181
j 8 The Kinetics of TiO -based Solar Cells Sensitized by Metal Complexes
182
2
Figure 8.4 Potential energy diagram for singlet and triplet injection into TiO2.
injection [24, 25, 27, 36–39] (Figure 8.4). In this scheme, the dye is excited and promoted to the singlet state, k1. Once in the singlet state, the dye can either inject into TiO2, ks, or perform intersystem crossing to the triplet state, kisc. Once in the triplet state, the electron can inject into TiO2, kt, or relax to the ground state, k1. These rates need to be optimized to achieve high injection yields. To maximize the efficiency of the DSSC, the dye levels need to be tuned to the lowest energy that provides sufficient driving force while still maintaining near unity injection yields. It has been demonstrated that the injection rates depend on ELUMOECB (where LUMO is the lower unoccupied molecular orbital and CB is the conduction band). Asbury et al. have studied the driving-force dependence of a series of Ru dyes. Ru(dcb)2(SCN)2, Ru(dcb)2(CN)2 and Ru(dcb)32 þ were studied, with the ground state, the thermally unrelaxed singlet metal-to-ligand charge transfer ð1 MLCT Þ and the triplet MLCT ð3 MLCTÞ potentials given in Table 8.1 [24]. Even though these potentials are for the free dye and not the adsorbed dye, Asbury et al. claim that these values should still provide good guideline for trends in the system of interest. The injection rates for these dyes are also presented in Table 8.1. As the ELUMOECB difference decreased, the injection rates decreased. Also, a wavelength dependence was found on the injection rate, with the shorter wavelengths creating higher Franck–Condon states and resulting in faster injection. However, not all studies have observed this phenomenon [27]. The conduction band was also shifted by varying the pH and showed the same rate dependence on ELUMOECB. When the
Table 8.1 Dye potential dependence on the fast and slow injection components.a 1
MLCT (V)
3
Dye
Ground state (V)
MLCT (V)
Singlet injection factorb
t fast (fs)
t slow (ps)
Ru(dcb)2(SCN)2 Ru(dcb)2(CN)2 Ru(dcb)3
0.65 1.08 1.40
2.45 2.02 1.70
0.82 0.68 0.48
0.73 0.55 0.43
<100 <100 <100
20 87 333
a b
All potentials are versus SCE. The singlet injection factor is the amount injected from the unrelaxed singlet state [24].
8.7 Charge Injection
pH was increased, the conduction band shifted to more negative potentials, resulting in a decrease in the injection rate. Hence the levels of the conduction band, singlet and triplet states greatly affect the rate, but the mode of electron transfer cannot be determined from these data. This driving-force dependence can be explained by either decreasing the free energy of electron transfer or by decreasing the number of accessible energy states in the conduction band as explained by non-adiabatic electron-transfer theory. Other evidence, such as the extremely fast injection rate and the large effect that one –CH2– spacer has on the injection rate, suggests an adiabatic process, because a much smaller effect is predicted from non-adiabatic electron-transfer theory. For the organic alizarin dye, strong evidence has been obtained that the injection occurs adiabatically into a surface-state atom and the electron then diffuses into the conduction band [40]. It seems likely, for some systems, that both adiabatic and non adiabatic processes occur. Injection was also studied by Tachibana et al., by application of an external bias to fill up some of the empty electronic states in the TiO2 [41]. These workers observed only a small dependence of the injection current on the applied potential, even at potentials near the open-circuit potential of the DSSC. At an applied potential of 0.7 V, the injection rate was only retarded by a factor of 25 relative to its value at short-circuit. This decrease is not large enough to prevent high injection yields from being obtained with Ru(dcb)2(SCN)2 as the dye. To minimize the energy lost due to the driving force needed to support efficient intersystem crossing and electron injection, new dyes need to be developed. The Stokes shift of 0.9 eV in Ru(dcb)2(SCN)2 is rather large and hence leaves room for optimization. Thus, the splitting between the singlet and triplet levels can, in principle, be reduced. Furthermore, the triplet state of Ru(dcb)2(SCN)2 injects from a level higher than is needed for efficient injection. The Ru(dcb)2(CN)2 has a LUMO energy 200 meV lower than that of Ru(dcb)2(SCN)2 and yet Ru(dcb)2(CN)2 still provides near unity injection. For 2,9,16,23-tetra(carbonylmethylaminocarboxyl) phthalocyaninatozinc(II), ZnPcGly, possibly only 100 meV is needed for efficient injection [42]. These observations suggest that the energy splitting between the singlet and triplet excited states can be decreased and the energy difference between the triplet state and the conduction band can also be decreased. Hence the optimal transition of the dye can be red shifted for better spectral overlap and higher currents from the DSSC could be obtained. Another type of dye that has promise in increasing the DSSC performance is the direct-injection dye [43–45]. Unlike the dyes discussed above, the direct-injection dyes do not go through an excited-state level associated with the dye. Instead, direct injection occurs by a dye to semiconductor transition that results from the binding of the dye to the TiO2 surface. This charge-transfer band moves electrons from the dye directly to the TiO2. Examples of this type of sensitizer are catechol- and hexacyanometal complexes [43, 46]. This type of transition could theoretically eliminate the energy loss due to the singlet–triplet splitting and consequently could minimize the driving force needed to produce rapid injection from the excited state of the dye. However, analysis of the energy of the charge-transfer band maximum suggests that
j183
j 8 The Kinetics of TiO -based Solar Cells Sensitized by Metal Complexes
184
2
the injection does not go into the lowest energy conduction band states, but rather into states significantly above the bandgap in energy [46, 47]. Further, no highly efficient DSSC has yet been developed from such systems, because of poor electroncollection efficiencies in the cells studied to date.
8.8 Recombination to the Dye
The process of recombination of electrons in the TiO2 back to the oxidized dye is greatly influenced by the interaction of the dye with the TiO2 surface. These processes are much like the electron injection process, but the kinetics are significantly different from those for injection. For efficient charge collection, the rate of recombination, k5, must be slower than the rate of dye regeneration, k6. With the dye in intimate contact with the TiO2 surface and the coupling sufficient to produce injection rates on the order of femtoseconds, if regeneration takes microseconds, it is surprising that all the injected electrons do not quickly recombine with the oxidized dye. However, near unity collection efficiencies, that only allow electron transfer into the semiconductor and not back to the solution or oxidized dye, have been achieved. In single-crystal semiconductor–liquid junctions, this type of charge-selective interfacial kinetics is accomplished by an electric field that is generated at the interface, due to a difference in the initial chemical potentials of the solution and semiconductor. With the nanocrystalline films used in DSSCs, this type of behavior is not observed, due to the small size of the nanoparticles and their inability to support large electric fields. The reasons for the slow kinetics have not yet been fully elucidated. There have been many studies of the recombination kinetics, and the process cannot be explained by simple first-order kinetics [48–53]. After electron injection, the electron quickly begins to diffuse through the TiO2 network, and the oxidized dye charge is free to move about on the surface [54]. Hence, in the simplest case, the kinetics would be expected to be second order. However, second-order kinetics do not explain the observed behavior either. Two second-order rate constants, and also stretched exponentials, have been required to produce accurate fits and have been justified because of the random electron transport in the nanocrystalline network. The rate of recombination is then explained by an electron and a hole diffusing to the recombination site, kd, and then undergoing electron transfer, as in Equations 8.5 and 8.6: kd
TiO2 ðeÞ þ Dye þ $ TiO2 ðeÞ=Dye þ
ð8:5Þ
ket
ð8:6Þ
TiO2 ðeÞ=Dye þ ! TiO2 þ Dye The expression for ket is given by
ket
" # 4p2 1 lsc ðDG 0 þ lÞ2 2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi jHAB j ¼ 1 exp 2 h 4lkB T 4plkB T dsc 3 ð6=pÞ3
ð8:7Þ
8.8 Recombination to the Dye
where h is Plancks constant, l is the reorganization energy, kB is Boltzmanns constant, T is the temperature, HAB is the electronic-coupling matrix element, lsc is the effective coupling length between the oxidized dye and the electrode, dsc is the density of atoms that contribute to the density of states in the band of concern, and DG0 is the standard interfacial free energy change of the reaction. From these processes, three models can explain the slow kinetics observed for recombination to the oxidized dye. One explanation is slow diffusion of the electron and hole together prior to recombination. The other reasons derive from a slow electron-transfer rate, which can be explained by recombination lying in the Marcus inverted region or by poor coupling between the oxidized dye and the electron in the TiO2. Unfortunately, the exact reasons have not yet been determined, but all three options have been explored and will be discussed below. The process of electron diffusion has been studied and in some systems is claimed to be the reason for slow recombination kinetics [51, 55–57]. Nanocrystalline TiO2 has a high density of trap states, most likely due to oxygen deficiencies or to surfaceadsorbed species. Electron diffusion is believed to occur through these traps. These states are mid-gap traps, and electrons thus have a waiting time, dependent on the energy of the trap, in each trap, until the electron moves to the next trap state. There are no long-range forces and electrons thus move merely by diffusion. To explain the slow recombination kinetics, the process of diffusion to recombination sites would need to be the rate-limiting step. When modeling the electron diffusion and comparing it with the recombination kinetics, reasonable fits were obtained. This mechanism thus explained the possible need for stretched exponentials to fit the data [51]. A rapid acceleration in the recombination rate has been observed as the applied potential is made more negative [49, 58]. This trend is not expected if the electrons are coming from the conduction band, but can be explained by the filling of deep trap states, resulting in faster diffusion of electrons. The diffusion model does explain the observed behavior of recombination, but slow kinetics of the electron transfer can not be ruled out as the rate-limiting step. For a slow ket, one possible reason is that recombination lies in the inverted region of the Marcus free-energy curve. Evidence for recombination being in the Marcus inverted region was obtained by studying a series of Ru and Os polypyridyl dyes [50]. The five dyes used were Ru(dcb)2(SCN)2, Ru(dcb)2(CN)2, Os(dcb)2(SCN)2, Os (dcb)2(CN)2 and Os(dcb)2(Cl)2. The ground-state potential of these dyes varied from 0.24 to 1.08 V versus the saturated calomel electrode (SCE). The data were fit to two second-order rate constants and the value used was the weighted average of the two. The results of the study are shown in Figure 8.5, and it is clearly evident that the rate decreased with increases in driving force. The coupling between the dye and TiO2 was also estimated from the temperature dependence of the electron-transfer rate, and changes in the coupling did not explain the observed trend. With recombination in the Marcus inverted region, the large difference in rates between injection and recombination would be expected. An other explanation for a slow rate constant is poor coupling between the dye cation and the electron in the TiO2. In one study, the following dyes were used: Ru (dcb)2(SCN)2, Ru(dcb)2(CN)2, and Ru(dcb)2Cl2, Ru(dcb)2DCT, where DCT is diethyl-
j185
j 8 The Kinetics of TiO -based Solar Cells Sensitized by Metal Complexes
186
2
Figure 8.5 Dependence of the recombination rate constant k3 on the ground-state M(III)/M(II) reduction potential of the complexes Ru(dcb)2(SCN)2, Ru(dcb)2(CN)2, Os(dcb)2(SCN)2, Os(dcb)2(CN)2 and Os(dcb)2(Cl)2. Also shown is the estimated driving force DG0 for the charge recombination process [50].
dithiocarbamate [53]. The recombination kinetics were fitted to stretched exponentials and ket was assumed to be 1/t50%. The free-energy dependence of the recombination rate is plotted in Figure 8.6. Because the rate is insensitive to the dye ground-state potential, the authors claimed that recombination lies near the peak of the Marcus free-energy curve. The slow recombination rates arise from an increased charge-transfer distance, resulting in reduced coupling, HAB. From molecular orbital
Figure 8.6 Plot of the logarithm of 1/t50% for charge recombination versus the ground-state oxidation potential of each dye E (A þ/A) (lower x-axis) for nanocrystalline TiO2 films sensitized by the dyes Ru(dcb)2Cl2 (1), Ru(dcb)2DCT (2), Ru(dcb)2(CN)2 (3) and Ru(dcb)2(SCN)2 (4). The
upper x-axis shows the corresponding estimated free energy driving force, DG , for the charge recombination process. The smooth line shows the free energy dependence of 1/t50% expected from Equation 8.7, assuming ket / 1/t50%, with l ¼ 0.8 eV and constant HAB. Adapted from [53].
8.9 Regeneration
Figure 8.7 Plots of the logarithm of 1/t50% versus the spatial separation r of the dye cation from the TiO2 surface for dyes Ru(dcb)2Cl2 (1), Ru(dcb)2DCT (2), Ru(dcb)2(CN)2 (3), Ru(dcb)2(SCN)2 (4) [53].
modeling, the separation of the highest occupied molecular orbital (HOMO) and the electrode surface was obtained and the effect on the recombination rate can be seen in Figure 8.7. The HAB dependence on the spatial separation, r, is given by H2AB ¼ H 20 ebr
ð8:8Þ
where b is a function of the barrier height. With the excited state of the dye residing on the bipyridine ligands that are directly attached to the TiO2, electron injection proceeds rapidly. This is in contrast to the situation for recombination, in which the HOMO of the dye cation resides on the metal center and on the unbound ligands. The difference in charge-transfer distance between injection and recombination thus results in a large difference in rates. In contrast, a study using extended conjugated linker groups revealed essentially no difference in the recombination rate with, and without the extended linker [59]. Unfortunately, the reason for slow recombination kinetics has not been determined, but slow recombination is clearly a requirement for a highly efficient solar cell. Both the injection and recombination processes depend on the interaction between the dye and the TiO2 surface. When designing new dyes, it is difficult to predict exactly how this interaction will affect all of the rates. The large number and variety, of dyes that work in DSSCs suggests that the rate-determining steps depend mainly on the nature of the TiO2. However, there are only a few examples of dyes with near-unity collection efficiencies, which means that there is at least some dependence on the properties of the dye.
8.9 Regeneration
Regeneration is the last process in which the dye is directly involved. Regeneration requires a redox couple that has the appropriate kinetics to regenerate the dye, while serving to shuttle holes between the two electrodes. The redox couple also must not
j187
j 8 The Kinetics of TiO -based Solar Cells Sensitized by Metal Complexes
188
2
allow recombination between electrons in the TiO2 and the acceptor in solution. This last requirement produces a limited number of suitable redox couples, with the I3/I performing the best to date. The kinetics of I3/I are complex and sluggish at theTiO2 surface, allowing for high charge collection yields [56, 60]. If one-electron redox couples with rapid interfacial charge-transfer rates are used, the injected electrons quickly recombine through the TiO2 or FTO interface into the solution and little or no net current is obtained [61]. With I3/I being essentially the only redox couple to give high overall cell efficiencies, a maximum Voc for a TiO2 DSSC is consequently imposed and only dyes with ground-state potentials more positive than that of I3/I can then be used. The other requirement of the redox couple is to efficiently regenerate the dye. Even with its complex kinetics, iodide is able to perform this function well for a large number of dyes. The regeneration is expected to proceed according to one of two different pathways: Dye þ þ I ! Dye þ I
.
ð8:9Þ
I þ I ! I 2
ð8:10Þ
Dye þ þ 2I ! Dye þ I 2
ð8:11Þ
2I 2 ! I3 þ I
ð8:12Þ
.
The first pathway is represented by Equations 8.9 and 8.10, in which the regeneration step would be first order in iodide concentration. In the second pathway, Equation 8.11, the regeneration step involves a complex of iodide and would show a second-order dependence on iodide. The transient absorbance at 800 nm measured for different iodide concentrations can be seen in Figure 8.8. In the absence of iodide, the absorbance at 800 nm arises from the oxidized dye. Upon the addition of iodide, the main feature of the oxidized dye begins to decay at a faster rate, but on the longer time scales a new feature is observed. This feature is attributed to the formation of the I2 intermediate. For the first-order process, Equation 8.9, the I2 intermediate is formed by Equation 8.10. Photoinjected electrons also exhibit an absorbance in the red/near-infrared region, but their extinction coefficient is expected to be relatively small compared with that of the oxidized dye or of I2 [58]. The I2 intermediate then decays by Equation 8.12. Other groups have also looked at the bleach of the dye and have measured the rate of the return of the ground state. From these studies, a wide range of regeneration rates have been measured, with rate constants ranging from 105 to 1010M1 [33, 50, 58, 62]. Some of these studies claim that the pathway is believed to be first order with respect to iodide whereas other workers see evidence of second-order kinetics. There is also evidence for the iodide complexing with the dye, in which case high regeneration rates should be observed [63–65]. The regeneration rate also shows a strong dependence on the type and concentration, of cations [17]. In Figure 8.9, the bleach of the dye is monitored in the absence of iodide and with 0.1 M iodide, as a function of the presence of a series of different cations. The size of the cations greatly affected the rate of regeneration, with small
8.9 Regeneration
Figure 8.8 Transient absorption data collected at 800 nm for a Ru(dcb)2(NCS)2-sensitized TiO2 film measured as a function of iodide concentration. The electrolytes consisted of LiClO4 and LiI in propylene carbonate, with the
salt concentration adjusted to maintain a constant Li þ concentration (0.1 M) and iodide concentrations as indicated in the figure. Also shown is an expansion of data collected at early times for I– concentrations of 0 and 100 mM [58].
cations causing a much faster oxidation of I at the TiO2 surface. Different cations can cause a shift in the band edges, with small cations shifting the bands to more positive potentials [66]. However, this is unlikely to directly affect the regeneration rate. The regeneration rate does not depend on applied potential [58], and it is unlikely that the electrode potential would have a direct effect on the regeneration rate. The change in regeneration rate is thus ascribed to a change in the dye potential, due to the electric field caused by the charges at the TiO2 surface, or due to a different regeneration
Figure 8.9 Time course of the transient absorbance changes obtained upon nanosecond pulsed laser excitation (l ¼ 490 nm, 5 ns FWHM pulse duration, 1 mJ/pulse) of Ru(II) (dcb)2(NCS)2 dye adsorbed on mesoporous
TiO2 films. Bleaching signals were measured at l ¼ 520 nm in dry propylene carbonate, without electrolyte (a) and in the presence of 0.1 M TBAI (b), 0.1 M LiI (c) and 0.05 M MgI2 (d). The inset displays trace (b) on a longer time scale [17].
j189
j 8 The Kinetics of TiO -based Solar Cells Sensitized by Metal Complexes
190
2
Figure 8.10 Reciprocal half-life (1/t1/2) of the Ru(II)(dcb)2(NCS)2 dye ground-state absorbance recovery (l ¼ 520 nm) following electron injection into TiO2. The concentration of iodide [I] ¼ 0.1 M in anhydrous propylene carbonate was kept constant, while the concentration of Li þ cations was varied. For (Li þ ) 0.1 M, the iodide counterions were TBA þ and Li þ . At higher Li þ concentrations, Li þ was added, in the form of dry LiClO4 [17].
pathway, with iodide complexing on the surface. In Figure 8.10, the regeneration rate is measured for a changing Li þ :TBA þ ratio. A drastic change was observed when the Li þ concentration was >0.01 M. At that point, electrophoretic measurements indicated that the surface switched from net negatively charged to net positively charged. To improve the efficiency of the regeneration step, it would be beneficial to use different redox couples. There are currently a few redox systems that provide for reasonable efficiencies, with all these redox systems being either kinetically slow oneelectron transfers or complex two-electron reactions [67–72]. The most efficient alternative is a cobalt complex that has been able to achieve DSSC energy-conversion efficiencies as high as 5.2% [72]. Other examples are ionic liquids of SCN and SeCN, which are both two-electron couples [68]. There has also been a large amount of research on polymers and ionic liquids, to convert the device to an all-solid-state form [73–77]. In these systems, recombination is still a significant problem. Use of fast one-electron redox couples could be achieved through surface passivation to prevent recombination. There is one successful example of this approach, in which an insulating polymer layer was electropolymerized on the exposed FTO and a longchain hydrocarbon was attached to the exposed TiO2 areas between dye molecules [61]. With this approach, the authors were able to use the ferrocenium/ferrocene redox couple and still obtain significant device efficiencies. The use of different redox couples could lead to fast regeneration rates, increasing the number and variety of dyes that could provide efficiencies that match, or exceed, those of the Ru polypyridyl dyes.
8.10 Conclusion
The dye kinetics in the DSSC have been shown to be complex, and the reasons for the high efficiencies have been explored but can not be fully explained. The dye is directly
8.10 Conclusion
involved in the injection, recombination and regeneration processes. To achieve the maximum performance of the DSSC, the dye levels must be properly tuned to allow efficient injection and regeneration with a minimum amount of driving force. Injection proceeds from the thermally excited singlet states and from the relaxed triplet states on the femtosecond and picosecond time scales, respectively. The regeneration process occurs in microseconds, before recombination of the conduction-band electrons with the oxidized dye. All of these kinetics are of a complex nature, requiring second-order kinetics and even stretched exponentials to explain the observed behavior. Because of the complex kinetics, the fast injection and regeneration processes still require further investigation to determine which of the theories accurately predicts the properties that greatly influence these rates. When designing new dyes for the DSSC, the electronic absorption spectrum is the key factor in determining whether the new dye can perform at a higher efficiency. With the processes discussed above allowing high collection yields, the limit to Jsc is set by the absorption spectrum of the dye. The solar spectrum is displayed in Figure 8.11, along with the absorption spectrum of Ru(dcb)2(SCN)2. To increase the value of Jsc, the absorbance must be shifted to longer wavelengths while still maintaining strong absorbance in the blue region. A good example of this is the black dye Ru(tct)(SCN)3, where tct is 4,40 ,400 -carboxylic acid-2,20 :60 ,200 -terpyridine, which has a red-shifted absorbance and shows a higher efficiency in the longer wavelengths than Ru(dcb)2(SCN)2 does (Figure 8.12) [78]. To have a dye outperform the Ru(tct)(SCN)3 or Ru(dcb)2(SCN)2 systems in the TiO2, I3/I cell, the absorbance spectrum must have a better spectral overlap and must maintain the favorable kinetics. Dyes such as porphyrins, phthalocyanines, corroles, quantum dots, organic dyes and many others have been used in DSSCs, but only a few exhibit efficiencies close to those of the systems described above [30, 79–85]. Antenna systems have also been used to increase the spectral overlap and improve the efficiency, with minimal results [10, 86, 87]. If these devices are ever to be produced on a large scale, the dyes need to be made from relatively earth-abundant metals, and research in this area is
Figure 8.11 Spectral overlap of Ru(dcb)2(SCN)2 on TiO2 (left) with the solar spectrum (right).
j191
j 8 The Kinetics of TiO -based Solar Cells Sensitized by Metal Complexes
192
2
Figure 8.12 Photocurrent action spectrum obtained with the black dye [RuL0 (SCN)3] attached to nanocrystalline TiO2 films. The incident photon to current conversion efficiency is plotted as a function of the wavelength of the exciting light. Photocurrent action spectra for bare TiO2 and TiO2 doped with Ru(dcb)2(SCN)2 have been included for comparison [78].
thus key to advancing DSSC use in energy production and thereby reducing further the cost of solar energy-conversion systems.
Acknowledgment
We acknowledge support from the Department of Energy, Basic Energy Sciences, that made the preparation of this chapter possible.
References 1 Energy Information Administration (2007) International Energy Outlook 2007, Report DOE/EIA-0484(2007). 2 Lewis, N.S. and Nocera, D. G. (2006) Proceedings of the National Academy of Sciences of the United States of America, 103, 15729–15735. 3 King, R.R., Law, D.C., Edmondson, K.M., Fetzer, C.M., Kinsey, G.S., Yoon,
H., Sherif, R.A. and Karam, N.H. (2007) Applied Physics Letters, 90, 183516. 4 West, W. (1974) Photographic Science and Engineering, 18, 35. 5 Moser, J. (1887) Monatshefte fur Chemie, 8, 373. 6 Tributsch, H. (1968) PhD Thesis, Technical University, Munich.
References 7 Tributsch, H. and Gerischer, H. (1969) Berichte Der Bunsen-Gesellschaft-Physical Chemistry Chemical Physics, 73, 251. 8 Nakao, M., Itoh, K., Watanabe, T. and Honda, K. (1985) Berichte Der BunsenGesellschaft-Physical Chemistry Chemical Physics, 89, 134–138. 9 ORegan, B. and Gratzel, M. (1991) Nature, 353, 737–740. 10 Koehorst, R.B.M., Boschloo, G.K., Savenije, T.J., Goossens, A. and Schaafsma, T.J. (2000) The Journal of Physical Chemistry B, 104, 2371–2377. 11 Mane, R.S., Lee, W.J., Pathan, H.M. and Han, S.H. (2005) The Journal of Physical Chemistry B, 109, 24254–24259. 12 Martinson, A.B.F., Elam, J.W., Hupp, J.T. and Pellin, M.J. (2007) Nano Letters, 7, 2183–2187. 13 Cameron, P.J. and Peter, L.M. (2003) The Journal of Physical Chemistry B, 107, 14394–14400. 14 Nazeeruddin, M.K., Kay, A., Rodicio, I., Humphry-Baker, R., Muller, E., Liska, P., Vlachopoulos, N. and Gratzel, M. (1993) Journal of the American Chemical Society, 115, 6382–6390. 15 Olson, C.L. (2006) The Journal of Physical Chemistry B, 110, 9619–9626. 16 Watson, D.F. and Meyer, G.J. (2004) Coordination Chemistry Reviews, 248, 1391–1406. 17 Pelet, S., Moser, J.E. and Gratzel, M. (2000) The Journal of Physical Chemistry B, 104, 1791–1795. 18 Boschloo, G., Haggman, L. and Hagfeldt, A. (2006) The Journal of Physical Chemistry B, 110, 13144–13150. 19 Yin, X., Zhao, H., Chen, L., Tan, W., Zhang, J., Weng, Y., Shuai, Z., Xiao, X., Zhou, X., Li, X. and Lin, Y. (2007) Surface and Interface Analysis, 39, 809–816. 20 Zhang, Z.P., Zakeeruddin, S.M., ORegan, B.C., Humphry-Baker, R. and Gratzel, M. (2005) The Journal of Physical Chemistry B, 109, 21818–21824. 21 Hinsch, A., Kroon, J.M., Kern, R., Uhlendorf, I., Holzbock, J., Meyer, A. and
22 23
24
25
26
27
28
29
30
31 32
33
34 35
Ferber, J. (2001) Progress in Photovoltaics, 9, 425–438. Gratzel, M. (2005) Inorganic Chemistry, 44, 6841–6851. Pelet, S., Gratzel, M. and Moser, J.E. (2003) The Journal of Physical Chemistry B, 107, 3215–3224. Asbury, J.B., Anderson, N.A., Hao, E.C., Ai, X. and Lian, T.Q. (2003) The Journal of Physical Chemistry B, 107, 7376–7386. Asbury, J.B., Wang, Y.Q., Hao, E.C., Ghosh, H.N. and Lian, T.Q. (2001) Research on Chemical Intermediates, 27, 393–406. Bauer, C., Boschloo, G., Mukhtar, E. and Hagfeldt, A. (2002) The Journal of Physical Chemistry B, 106, 12693–12704. Durrant, J.R., Tachibana, Y., Mercer, I., Moser, J.E., Gratzel, M. and Klug, D.R. (1999) International Journal of Research in Physical Chemistry & Chemical Physics (Munich), 212, 93–98. Furube, A., Katoh, R., Hara, K., Sato, T., Murata, S., Arakawa, H. and Tachiya, M. (2005) The Journal of Physical Chemistry B, 109, 16406–16414. Zakeeruddin, S.M., Nazeeruddin, M.K., Pechy, P., Rotzinger, F.P., HumphryBaker, R., Kalyanasundaram, K., Gratzel, M., Shklover, V. and Haibach, T. (1997) Inorganic Chemistry, 36, 5937–5946. Walker, D., Chappel, S., Mahammed, A., Brunschwig, B.S., Winkler, J.R., Gray, H.B., Zaban, A. and Gross, Z. (2006) Journal of Porphyrins and Phthalocyanines, 10, 1259–1262. Galoppini, E. (2004) Coordination Chemistry Reviews, 248, 1283–1297. Asbury, J.B., Hao, E.C., Wang, Y.Q. and Lian, T.Q. (2000) The Journal of Physical Chemistry B, 104, 11957–11964. Heimer, T.A., Heilweil, E.J., Bignozzi, C.A. and Meyer, G.J. (2000) Journal of Physical Chemistry A, 104, 4256–4262. Ferrere, S. (2000) Chemistry of Materials, 12, 1083–1089. Ferrere, S. and Gregg, B.A. (1998) Journal of the American Chemical Society, 120, 843–844.
j193
j 8 The Kinetics of TiO -based Solar Cells Sensitized by Metal Complexes
194
2
36 Asbury, J.B., Hao, E., Wang, Y.Q., Ghosh, H.N. and Lian, T.Q. (2001) The Journal of Physical Chemistry B, 105, 4545–4557. 37 Cherepy, N.J., Smestad, G.P., Gratzel, M. and Zhang, J.Z. (1997) The Journal of Physical Chemistry B, 101, 9342–9351. 38 Heimer, T.A. and Heilweil, E.J. (1997) The Journal of Physical Chemistry B, 101, 10990–10993. 39 Ellingson, R.J., Asbury, J.B., Ferrere, S., Ghosh, H.N., Sprague, J.R., Lian, T.Q. and Nozik, A.J. (1998) The Journal of Physical Chemistry B, 102, 6455–6458. 40 Duncan, W.R., Stier, W.M. and Prezhdo, O.V. (2005) Journal of the American Chemical Society, 127, 7941–7951. 41 Tachibana, Y., Haque, S.A., Mercer, I.P., Moser, J.E., Klug, D.R. and Durrant, J.R. (2001) The Journal of Physical Chemistry B, 105, 7424–7431. 42 Liu, G.M., Jaegermann, W., He, J.J., Sundstrom, V. and Sun, L.C. (2002) The Journal of Physical Chemistry B, 106, 5814–5819. 43 Liu, Y., Dadap, J.I., Zimdars, D. and Eisenthal, K.B. (1999) The Journal of Physical Chemistry B, 103, 2480–2486. 44 Rajh, T., Chen, L.X., Lukas, K., Liu, T., Thurnauer, M.C. and Tiede, D.M. (2002) The Journal of Physical Chemistry B, 106, 10543–10552. 45 Weng, Y.X., Wang, Y.Q., Asbury, J.B., Ghosh, H.N. and Lian, T.Q. (2000) The Journal of Physical Chemistry B, 104, 93–104. 46 Harris, J.A., Trotter, K. and Brunschwig, B.S. (2007) The Journal of Physical Chemistry B, 111, 6695–6702. 47 Khoudiakov, M., Parise, A.R. and Brunschwig, B. S. (2003) Journal of the American Chemical Society, 125, 4637–4642. 48 Haque, S.A., Tachibana, Y., Klug, D.R. and Durrant, J. R. (1998) The Journal of Physical Chemistry B, 102, 1745–1749. 49 Haque, S.A., Tachibana, Y., Willis, R.L., Moser, J.E., Gratzel, M., Klug, D.R. and Durrant, J. R. (2000) The Journal of Physical Chemistry B, 104, 538–547.
50 Kuciauskas, D., Freund, M.S., Gray, H.B., Winkler, J.R. and Lewis, N. S. (2001) The Journal of Physical Chemistry B, 105, 392–403. 51 Nelson, J., Haque, S.A., Klug, D.R. and Durrant, J. R. (2001) Physical Review BCondensed Matter, 63, 205321. 52 Hoertz, P.G., Carlisle, R.A., Meyer, G.J., Wang, D., Piotrowiak, P. and Galoppini, E. (2003) Nano Letters, 3, 325–330. 53 Clifford, J.N., Palomares, E., Nazeeruddin, M.K., Gratzel, M., Nelson, J., Li, X., Long, N.J. and Durrant, J. R. (2004) Journal of the American Chemical Society, 126, 5225–5233. 54 Higgins, G.T., Bergeron, B.V., Hasselmann, G.M., Farzad, F. and Meyer, G. J. (2006) The Journal of Physical Chemistry B, 110 2598–2605. 55 Green, A.N.M., Palomares, E., Haque, S.A., Kroon, J.M. and Durrant, J.R. (2005) The Journal of Physical Chemistry B, 109, 12525–12533. 56 Kopidakis, N., Benkstein, K.D., van de Lagemaat, J. and Frank, A.J. (2003) The Journal of Physical Chemistry B, 107, 11307–11315. 57 Barzykin, A.V. and Tachiya, M. (2002) The Journal of Physical Chemistry B, 106, 4356–4363. 58 Montanari, I., Nelson, J. and Durrant, J.R. (2002) The Journal of Physical Chemistry B, 106, 12203–12210. 59 Galoppini, E., Guo, W., Zhang, W., Hoertz, P.G., Qu, P. and Meyer, G.J. (2002) Journal of the American Chemical Society, 124, 7801–7811. 60 Huang, S.Y., Schlichthorl, G., Nozik, A.J., Gratzel, M. and Frank, A.J. (1997) The Journal of Physical Chemistry B, 101, 2576–2582. 61 Gregg, B.A., Pichot, F., Ferrere, S. and Fields, C.L. (2001) The Journal of Physical Chemistry B, 105, 1422–1429. 62 Nasr, C., Hotchandani, S. and Kamat, P.V. (1998) The Journal of Physical Chemistry B, 102, 4944–4951. 63 Greijer, H., Lindgren, J. and Hagfeldt, A. (2001) The Journal of Physical Chemistry B, 105, 6314–6320.
References 64 Marton, A., Clark, C.C., Srinivasan, R., Freundlich, R.E., Sarjeant, A.A.N. and Meyer, G.J. (2006) Inorganic Chemistry, 45, 362–369. 65 Clifford, J.N., Palomares, E., Nazeeruddin, M.K., Gratzel, M. and Durrant, J.R. (2007) The Journal of Physical Chemistry C, 111, 6561–6567. 66 Redmond, G. and Fitzmaurice, D. (1993) The Journal of Physical Chemistry, 97, 1426–1430. 67 Hattori, S., Wada, Y., Yanagida, S. and Fukuzumi, S. (2005) Journal of the American Chemical Society, 127, 9648–9654. 68 Bergeron, B.V., Marton, A., Oskam, G. and Meyer, G.J. (2005) The Journal of Physical Chemistry B, 109, 937–943. 69 Cameron, P.J., Peter, L.M., Zakeeruddin, S.M. and Gratzel, M. (2004) Coordination Chemistry Reviews, 248, 1447–1453. 70 Nusbaumer, H., Zakeeruddin, S.M., Moser, J.E. and Gratzel, M. (2003) Chemistry - A European Journal, 9, 3756–3763. 71 Sapp, S.A., Elliott, C.M., Contado, C., Caramori, S. and Bignozzi, C.A. (2002) Journal of the American Chemical Society, 124, 11215–11222. 72 Nusbaumer, H., Moser, J.E., Zakeeruddin, S.M., Nazeeruddin, M.K. and Gratzel, M. (2001) The Journal of Physical Chemistry B, 105, 10461–10464. 73 Matsumoto, M., Wada, Y., Kitamura, T., Shigaki, K., Inoue, T., Ikeda, M. and Yanagida, S. (2001) Bulletin of the Chemical Society of Japan, 74, 387–393. 74 Kubo, W., Kambe, S., Nakade, S., Kitamura, T., Hanabusa, K., Wada, Y. and Yanagida, S. (2003) The Journal of Physical Chemistry B, 107, 4374–4381. 75 Wang, P., Zakeeruddin, S.M., Gratzel, M., Kantlehner, W., Mezger, J., Stoyanov, E.V. and Scherr, O. (2004) Applied Physics A–Materials Science & Processing, 79, 73–77. 76 Wang, P., Wenger, B., Humphry-Baker, R., Moser, J.E., Teuscher, J., Kantlehner, W.,
77
78
79
80
81
82
83
84
85
86
87
Mezger, J., Stoyanov, E.V., Zakeeruddin, S.M. and Gratzel, M. (2005) Journal of the American Chemical Society, 127, 6850–6856. Mazille, F., Fei, Z.F., Kuang, D.B., Zhao, D.B., Zakeeruddin, S.M., Gratzel, M. and Dyson, P.J. (2006) Inorganic Chemistry, 45, 1585–1590. Nazeeruddin, M.K., Pechy, P., Renouard, T., Zakeeruddin, S.M., Humphry-Baker, R., Comte, P., Liska, P., Cevey, L., Costa, E., Shklover, V., Spiccia, L., Deacon, G.B., Bignozzi, C.A. and Gratzel, M. (2001) Journal of the American Chemical Society, 123, 1613–1624. Campbell, W.M., Burrell, A.K., Officer, D.L. and Jolley, K.W. (2004) Coordination Chemistry Reviews, 248, 1363–1379. He, J.J., Benko, G., Korodi, F., Polivka, T., Lomoth, R., Akermark, B., Sun, L.C., Hagfeldt, A. and Sundstrom, V. (2002) Journal of the American Chemical Society, 124, 4922–4932. Islam, A., Sugihara, I., Hara, I., Singh, L.P., Katoh, R., Yanagida, M., Takahashi, Y., Murata, S. and Arakawa, H. (2000) New Journal of Chemistry, 24, 343–345. Hara, K., Horiguchi, T., Kinoshita, T., Sayama, K., Sugihara, H. and Arakawa, H. (2000) Solar Energy Materials, 64, 115–134. Hara, K., Sayama, K., Ohga, Y., Shinpo, A., Suga, S. and Arakawa, H. (2001) Chemical Communications, 569–570. Plass, R., Pelet, S., Krueger, J., Gratzel, M. and Bach, U. (2002) The Journal of Physical Chemistry B, 106, 7578–7580. Horiuchi, T., Miura, H., Sumioka, K. and Uchida, S. (2004) Journal of the American Chemical Society, 126, 12218–12219. Hattori, S., Ohkubo, K., Urano, Y., Sunahara, H., Nagano, T., Wada, Y., Tkachenko, N.V., Lemmetyinen, H. and Fukuzumi, S. (2005) The Journal of Physical Chemistry B, 109, 15368–15375. Krebs, F.C. and Biancardo, M. (2006) Solar Energy Materials, 90, 142–165.
j195
j197
9 Automotive Emission Control: Past, Present and Future Robert J. Farrauto and Jeffrey Hoke
9.1 Introduction
It has been over 30 years since the first catalytic converters were installed on gasolinefueled passenger cars in response to the Clean Air Act passed by the US Congress in 1970. This was done in order to address the growing pollution problems in most major US cities caused by automobile emissions. In order to facilitate the development of catalytic converters and standardize emissions testing, a federal test procedure (FTP) simulating the driving habits of its citizens was established [1]. It included measuring emissions in cold and hot start modes and at low and high speeds with accelerations and decelerations. Limits were placed on the emissions of carbon monoxide (CO), hydrocarbons (HC) and nitrogen oxides (NOx), and warrantees for catalyst lifetime (e.g. 50 000 miles) were required. The FTPemissions test was very demanding, with low and high flow conditions and rapid changes in temperature simulating high speeds and accelerations. Over the years, this test has been modified and the standards for allowable emissions decreased with greater severity. In order to meet these increasingly stringent requirements, new technologies have been developed with much success. By definition, catalysts accelerate the rate at which reactants are converted to desirable products. In the case of emissions catalysts, the reactants are CO, HC and NOx while the products are CO2, H2O and N2. The emissions catalyst allows the reactions of unwanted pollutants to occur at temperatures typical of the automobile exhaust and considerably lower than in its absence (i.e. thermal reactions only). A more complete description of catalyst fundamentals is available [2, 3]. The first catalytic converters installed in the exhausts of the internal combustion (IC) engines operated with excess air and abated CO and unburned hydrocarbon (CxHy or HC) emissions. Nitrogen oxides were not treated, but rather were reduced using combustion control to meet the regulations. The catalyst operated as a passive device installed in the exhaust of the vehicle independent of the engine and its controls. At the time of the Clean Air Act, the poisoning affects of CO on humans was known. It was also known that most hydrocarbons and the nitrogen oxides emitted from the IC engine participated in photochemical reactions with sunlight generating
Handbook of Green Chemistry, Volume 2: Heterogeneous Catalysis. Edited by Robert H. Crabtree Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32497-2
j 9 Automotive Emission Control: Past, Present and Future
198
an irritating gas referred to as smog (smoke and fog). We now know that smog contains ozone as the primary unhealthy component. Given the toxicity of smog, it was necessary to reduce both HC and nitrogen oxides (NO, NO2 and N2O). Regulations required a 90% reduction in both CO and HC and a 50% reduction in NOx for 50 000 miles of driving relative to a 1970 uncontrolled vehicle. Methane was excluded from the HC emission regulations since it is relatively non-reactive in photochemical reactions compared with other saturated, unsaturated and aromatic molecules. Its contribution to greenhouse gases was not considered at the time. One key factor that had to be considered for the successful functioning of a catalytic converter was the presence of lead-containing compounds in pre-1975 gasoline. Lead was added to gasoline to increase octane ratings, but because it was a known poison to people and catalysts, the regulations specified its removal. With the removal of lead from gasoline, achieving the appropriate octane levels for stable engine operation was accomplished primarily by increasing oxygenates in the fuel and altering the process for reforming petroleum naphtha from which gasoline is derived. Since 1975, vehicle emission regulations all over the world have become more stringent, thereby requiring both advances in catalyst technology and improved fuel quality. In 1980, the development of the oxygen sensor and the three-way catalyst (TWC) allowed the simultaneous conversion of CO, HC and NOx. The catalyst advanced from a passive (oxidation only) to active (TWC) device dictating the operation of the engine via electronic feedback control strategies. In the early 1990s, European and Asian countries mandated lead removal from fuels and adapted automotive emission standards that required catalytic converters. Also, emission regulations were enacted in the USA, Asia and Europe for diesel engines. Simultaneously controlling the particulate and gaseous emissions from diesel engines is a very complicated process that has stimulated advanced catalyst, filtration and engine technologies. There is some hope that the homogeneous charge combustion ignition (HCCI) engine will simplify emission control. As the world is becoming sensitized to the environmental consequences of the continued use of fossil fuels, we are seeing the development of new technologies that hold great promise in addressing our fuel security while preserving the environment. The emergence of a hydrogen economy and H2 and O2 fuel cells will bring about a quantum change in the way we manage transportation and minimize the negative impact of fossil fuels. Catalysis will play a key role.
9.2 The First Oxidation Catalysts (1975–80) 9.2.1 Pollution Abatement Reactions for Gasoline-Fueled Engines
Destruction of CO and hydrocarbon pollutants is achieved by catalytic oxidation with excess air as follows: catalyst
CO þ 1=2 O2 ! CO2
ð9:1Þ
catalyst
ð9:2Þ
Cx Hy þ ð1 þ y=4ÞO2 ! xCO2 þ y=2H2 O
9.2 The First Oxidation Catalysts (1975–80)
In contrast, destruction of NOx is achieved by catalytic reduction with CO and hydrocarbons as described in the following generic equations and which will be described in more detail later: catalyst
ð9:3Þ
catalyst
ð9:4Þ
CO þ NO=NO2 ! N2 þ CO2 HC þ NO=NO2 ! N2 þ CO2 þ H2 O
Both the NOx and the excess O2 compete for CO and HC reducing agents in the complex engine exhaust. In actuality, since the O2 wins the selectivity competition and consumes almost all of the CO and HC, none is available to reduce the NOx. In other words, no catalyst is sufficiently selective to permit both the oxidation of CO and HCs and the reduction of NOx in the presence of excess O2. One early solution proposed to meet the initial NOx emission requirements was to operate the engine fuel rich (sub-stoichiometric in air) in order to provide sufficient reducing agents (i.e. hydrocarbon fuel components) to reduce the NOx to N2. Extra air was then added to the inlet of an oxidation catalyst downstream to convert the remaining CO and HC. Loss of fuel economy, the necessity for a pump for air injection and the absence of a suitable reduction catalyst eliminated this approach from consideration. In response, the automobile companies introduced exhaust gas recirculation to reduce the NOx to acceptable levels and meet the regulations. A small percentage of the exhaust gas, rich in N2, H2O and CO2, is recycled back into the combustion chamber, thereby reducing both the flame temperature (<1300 C) and the resulting NOx formed from the thermal reactions between N2 and O2. 9.2.2 Catalyst Materials
Early on, a major question was which catalyst materials would be suitable to drive the oxidation reactions and meet the regulations. At the time, platinum group metals (PGMs), especially platinum (Pt) and palladium (Pd), were known to be excellent oxidation catalysts but were rare, from countries with political unrest (South Africa and the former Soviet Union) and consequently expensive. The transition metals and their oxides such as copper (Cu), chromium (Cr), nickel (Ni), cobalt (Co) and manganese (Mn) were also known to be catalytic but were considerably less active than the PGMs. These metals are often referred to as base metals. Research was intensive as various metals, metal oxides and their combinations were investigated. The dominant performance issues were catalytic activity for CO and HC oxidation, resistance to poisons such as oxides of sulfur and hydrothermal stability under a broad range of driving conditions. Laboratory and engine aging equipment simulating actual driving cycles was implemented in catalyst, chemical and automobile laboratories throughout the USA, Europe and Japan. There was a strong imperative to utilize transition metals and metal oxides as the catalysts for emission control, given their availability and low price relative to the precious metals. In particular, the oxides of Cu and Cr were good candidates for
j199
j 9 Automotive Emission Control: Past, Present and Future
200
catalyzing the oxidation of both CO and HC. In laboratory experiments using propane and propylene as reasonable model compounds for exhaust HCs, a mixture of CuO and CuCr2O4 showed very good initial activity, but once exposed to the maximum temperature expected in the exhaust (i.e. 800 C), the structure converted to Cu2Cr2O4 with lower activity (as measured by an increase in T50, the temperature necessary for 50% conversion). The decrease in activity was accompanied by a change in physical characteristics such as a reduction in BETsurface area and growth in XRD crystal size due to sintering [4]. 800 C=air
CuO þ CuCr2 O4 ! Cu2 Cr2 O4 þ 1=2 O2 T50 ( C)
CO ¼ 205 HC ¼ 290
ð9:5Þ
230 315
BET surface area (m2 g1)
¼18
3
XRD crystallite size (mm)
¼0.1
10
In order to disperse and enhance the number of catalytic sites available to the reactants and to stabilize the catalytic components against sintering, a high surface area g-Al2O3 material was introduced as a carrier for the Cu and Cr oxides. g-Al2O3 is composed of pores in the 2–50 nm range which are large enough for the gaseous reactants to penetrate and undergo surface reaction with the catalytic components dispersed on its internal surface. Although the initial activity of the supported Cu/Cr catalyst was good, thermal treatment resulted in a loss of activity due to the formation of CuAl2O4 which had considerably less activity that the fresh catalyst. 800 C=air
CuO=g-Al2 O3 ! CuAl2 O4 low activity
high activity
ð9:6Þ
In addition to thermal treatments, catalysts were also exposed to oxides of sulfur (e.g. SO2), since gasoline contained up to 0.1% (1000 ppm) of sulfur in the mid-1970s. Typically, catalyst performance declined after exposure to sulfur oxides due to the formation of inactive sulfates. Equation 9.7 shows the formation of catalytically inactive CuSO4 when CuO is exposed to SO2/SO3. Cobalt-containing catalysts also suffered from the same type of chemical reaction (Equation 9.8). 500 C=air
CuO þ SO2 =SO3 ! CuSO4 inactive
active 500 C= air
Co2 O3 þ SO2 =SO3 ! Co2 ðSO4 Þ3 active
ð9:7Þ
ð9:8Þ
inactive
In contrast, CuCr2O4 had been shown to adsorb SO2/SO3 (Equation 9.9) forming a chemisorbed layer rather than a compound of sulfate. Although this reduces the
9.2 The First Oxidation Catalysts (1975–80)
catalytic activity, it allows for regeneration by a simple water wash as shown in (Equation 9.10) [5]. 500 C=air
CuCr2 O4 þ SO2 =SO3 ! CuCr2 O4 . . . SO2 =SO3
H2 O=25 C
CuCr2 O4 . . . SO2 =SO3 ! CuCr2 O4 þ SO2 ðSO3 Þ in H2 O low activity
ð9:9Þ
poor activity
high activity
ð9:10Þ
moderate activity
It was clear in 1973 that the lack of thermal stability and the vulnerability to oxides of sulfur eliminated base metals from further consideration. In contrast, precious metals deposited on carriers were found to function after such treatments and were the only viable catalyst materials that could be used [1]. Studies attempting to replace precious metals continue, but none have been commercially successful. 9.2.3 Carriers
Particulates of the high surface area g-Al2O3, used to carry or support the catalytic components existed for many years in the chemical and petroleum industries in the form of spheres, tablets and extrudates. They were manufactured in large volumes with geometries designed for the process of interest and were relatively inexpensive. Reactor designs to maximize performance of specific geometries were routinely developed. Hence it was natural to expect that these particulate catalyst carriers would be used in the catalytic converter. However, a new extruded ceramic structure referred to as a monolith or honeycomb made of the mineral cordierite (2MgO5SiO22Al2O3) and consisting of multiple parallel channels was also being developed. The walls of the monolith had low porosity but were suitable for coating a thin layer of high activity catalyst (e.g. Pt and Pd) deposited on a high surface area carrier such as g-Al2O3. This coating is generally referred to as the catalyzed washcoat. As exhaust flows through the monolith, the CO and HC pollutants and excess oxygen diffuse from the bulk gas to the washcoat surface, through the porous network of the washcoat pores, and react on the active catalytic sites to form carbon dioxide and water. Specific honeycomb structures were designed to have low thermal expansion so as to resist cracking (thermal shock) during the rapid temperature transients expected during the various driving cycles imposed by the Federal Test Procedure and also under normal consumer driving. Both pelleted and monolithic structures were used in the early years. However, attrition of the particulates eventually limited their useful life and monoliths became the universally accepted catalyst support. Monolith compositions and designs have improved with respect to the increasingly stringent regulations and driving demands (i.e. high temperature stability), but the basic structure of the multi-channel honeycombs remains today. Cell or channel densities (often defined as cells per square inch or cpsi) have increased from 200–300 to over 900 cpsi today, with correspond-
j201
j 9 Automotive Emission Control: Past, Present and Future
202
Figure 9.1 Monolith and oxidation catalyst washcoated in a single channel.
ingly thinner walls, lighter weights and increased geometric surface area. Metal monoliths also have been introduced and afford high cell densities with thinner walls than ceramics. This results in lower pressure drop and lower mass for faster light-off times (the time required for the catalyst to become sufficiently hot to initiate conversion). Figure 9.1 shows a cartoon of a multi-channel monolith and an ideal channel with an oxidation catalyst as a washcoat.
9.3 Three-Way Catalysis (1980–present) 9.3.1 Three-Way Catalysis
By 1979, it was necessary to address the emissions of nitrogen oxides since engine approaches such as EGR were no longer sufficient. Rhodium was discovered to be a good NO/NO2 reduction catalyst (Equation 9.11), whereas Pt and Pd were best for oxidation reactions (Equations 9.1 and 9.2). Reaction 9.11 is not balanced given the variation in reactants species. catalyst
HC þ CO þ NO=NO2 ! N2 þ H2 O þ CO2
ð9:11Þ
It was also known that near the stoichiometric air-to-fuel weight ratio (about 14.6) entering the engine, all three pollutants could be simultaneously converted [6]. The
9.3 Three-Way Catalysis (1980–present)
optimum conversion occurs at the stoichiometric point where l ¼ 1, l being defined as the actual air-to-fuel ratio entering the engine divided by the air-to-fuel ratio at the stoichiometric point. Therefore, by definition, l ¼ 1 at the stoichiometric point. Under fuel-rich conditions (i.e. a deficiency of air with l < 1), the conversion of NOx is high since CO and HC are not fully converted and hence are available to reduce the NOx to N2. Steam reforming (Equation 9.12) and water gas shift (Equation 9.13) reactions also lead to H2 generation on the fuel-rich side of the profile. The hydrogen generated also contributes to NOx reduction (Equation 9.14). The example below is given for ethane steam reforming, but it should be understood that many other reactions are thermodynamically favorable, so the exhaust is more complicated than indicated here. For this reason, the reaction equation is not chemically balanced. C2 H6 þ H2 O ! 4H2 þ CO ðand varying amounts of CO2 Þ
ð9:12Þ
CO þ H2 O ! H2 þ CO2
ð9:13Þ
H2 þ NO=NO2 ! N2 þ H2 O
ð9:14Þ
9.3.2 Oxygen or Lambda Sensor
The question still remained of how to control the engine operation within the narrow air-to-fuel ratio required for three-way conversion. This challenge was met by Bosch with the development of the oxygen (or lambda) sensor. This solid electrolytic system, comprising Pt as anode and cathode with an oxygen ion conducting ceramic composed of zirconia stabilized with yttrium, detects the O2 content in the exhaust immediately before the inlet to the catalyst. When the concentration deviates from the proper amount, an electronic signal is generated and sent to an electronic control unit. This in turn feeds another signal to the fuel delivery system in order to increase or decrease the amount of fuel entering the engine. As a result, the propulsion of the vehicle became a closed loop system in which the proper operation of the catalyst controlled the operation of the engine. Although the catalyst was a passive device with no communication to the engine before 1979, the introduction of the TWC and O2 sensor made it the heart of the engine control system. 9.3.3 Oxygen Storage Component
Due to non-ideal response times of the air–fuel adjustments, the exhaust composition generally oscillates slightly rich (l < 1) and slightly lean (l > 1) around the stoichiometric point (l ¼ 1). To minimize the loss of conversion during these 0.5–1 s deviations, an oxygen storage component (OSC) is incorporated into the TWC. Compounds of CeO2/ZrO2 are most commonly used because oxygen can be easily released for CO and HC oxidation when the exhaust is deficient in oxygen (l < 1) and
j203
j 9 Automotive Emission Control: Past, Present and Future
204
Figure 9.2 Simultaneous three way catalyst (TWC) conversion of CO, HC and NO pollutants by controlling the air-to-fuel ratio.
re-absorbed when the exhaust has excess oxygen (l > 1). Equations (9.15) and (9.16) show only the Ce-containing oxides for simplicity. CO þ 2CeO2 ! Ce2 O3 þ CO2
ð9:15Þ
Ce2 O3 þ 1=2 O2 ! 2CeO2
ð9:16Þ
The conversion of CO, HC and NO for a modern TWC catalyst versus the air-to-fuel ratio is shown in Figure 9.2. The optimum conversion for all three pollutants is at the stoichiometric point of l ¼ 1. 9.3.4 Further Improvements in TWC
Catalyst technology that successfully met standards in the 1980s was not adequate in the 1990s with increasingly stringent worldwide governmental regulations. New catalyst formulations, monolith designs, engine control strategies and fuel quality improvements (i.e. sulfur reductions) were implemented. A great concern in meeting the regulations was the cold start issue. When a vehicle is at rest for a long time, the engine and the exhaust system containing the catalyst are cold. When the engine is started, the catalyst temperature is too low for light-off (i.e. too low to convert CO, HC or NOx). During this relatively short time of 60–120 s before the catalyst heats up, the hydrocarbon emissions are sufficiently high to fail the Federal Test Procedure. To address this issue, a close-coupled catalyst was developed. It is positioned close to the cylinder exhaust manifold where it heats up within 8–10 s to initiate catalytic oxidation of the hydrocarbons. This close-coupled catalyst solved the cold start problem and helped meet the Federal Standards for hydrocarbon emissions. However, the approach required the development of new catalyst technologies that could withstand temperatures in excess of 1000 C. This in turn required enhanced stabilization of the carrier, precious metals, oxygen storage components and ceramic and metal monoliths. New catalytic washcoat materials were developed to meet the severe temperature extremes of over 1000 C [1].
9.3 Three-Way Catalysis (1980–present)
Figure 9.3 (a) Multiple washcoat layer architecture on monolith; (b) zone washcoats on monolith.
Multilayered catalyst washcoats, containing varying amounts of Pt, Rh and sometimes Pd, also helped to minimize the negative interactions between certain catalytic components and stabilizers that caused deactivation (Figure 9.3a). Washcoat properties were varied to filter inorganic poisons such as the phosphorus and zinc typically found in lubricating oils that are subsequently deposited on the catalyst after combustion. Zoned washcoat layers (Figure 9.3b) were also developed that allow variation of catalyst compositions axially down the monolith channel. At the inlet of the monolith a higher catalyst loading may be needed to insure rapid light-off where the reaction kinetics are favored by increased metal loadings. Correspondingly, less metal may be deposited in the downstream portion of the channel since the reactions here are generally mass transfer controlled and not very sensitive to metal loading. Sulfur levels in fuels have now been reduced to about 30 ppm in the US, a factor of at least 10–30 lower than in the mid-1970s. Monoliths have enhanced thermal and mechanical stability, higher cell densities and decreased mass for fast light-off relative to earlier designs. Combining these catalyst and substrate advances with additional engine, computer and sensing advancements, catalytic converters in the 21st century have a lifetime exceeding 150 000 miles. The converter is housed in a metal canister welded into the exhaust system in the under-floor position (below the drivers seat). It can withstand mechanical and vibration abuse over a wide variety of driving conditions. A cartoon of a close-coupled, feedback controlled exhaust system is shown in Figure 9.4. The gasoline catalytic converter is a heroic accomplishment for improving air quality. A more comprehensive resource is available [1].
j205
j 9 Automotive Emission Control: Past, Present and Future
206
Figure 9.4 Close-coupled catalyst with under-floor TWC and closed-loop feedback engine control.
9.4 Diesel Catalysis 9.4.1 Controlling Diesel Emissions
Due to significant benefits in fuel economy, diesel vehicles have enjoyed a surge in popularity in recent years, particularly in Europe where approximately 50% of new cars sold are diesel. Although improved fuel efficiency is the key aim, improvements in the drivability of diesel vehicles and also a reduction in their tailpipe emissions have also helped to improve the image of diesels and spur further growth. Unlike gasoline emissions that are mainly gaseous in nature, diesel emissions from passenger vehicles, buses and trucks contain solid, liquid and gaseous components. Furthermore, because diesel engines operate with a large excess of air, the three-way conversion strategy of controlling the air-to-fuel ratio near the stoichiometric point is not consistent with their operation. Based on the combustion characteristics of lean burn compression ignition diesel engines, the focus of emission regulations is on two key tailpipe pollutants – particulate matter (PM) and NOx. These are generally produced in high quantities during the combustion of diesel fuel. Since diesel fuel is less volatile than gasoline and it is injected directly into the combustion chamber as the piston nears top dead center (TDC), combustion initiates before the fuel–air mixture has had sufficient time to vaporize and mix completely. As a result, the combustion of the diesel fuel usually begins at the outer edges of the fuel spray and then progresses towards the center. This diffusion combustion phenomenon results in the production of large amounts of NOx at the outer edges of the spray where the temperature and oxygen concentration are high and also high amounts of soot in the interior of the spray where the temperature and oxygen concentration are low. As a result, both soot and NOx are produced simultaneously in large amounts from diesel engines. This is summarized in Figure 9.5, which plots local flame temperature versus equivalence
9.4 Diesel Catalysis
Figure 9.5 Engine map of flame temperature versus equivalence ratio of diesel emissions. Reprinted with permission from SAE International.
ratio (fuel:air ratio actual divided by fuel:air ratio at stoichiometric) in the piston and the regions where soot and NOx are typically generated [7]. NOx forms in the region of localized high temperature and low equivalence ratio (excess air) at the outer edge of plume whereas soot forms in the region of localized low temperature and high equivalence ratio (deficiency of air) at the center of the plume. Due to the adverse health effects associated with diesel soot and the ozoneforming potential of NOx, both are a major focus for emissions regulations (for a review of worldwide diesel emission regulations, see [8]). Although CO and hydrocarbons (HC) are also regulated, the amounts produced by diesel engines are generally low enough not to be a major obstacle for meeting emission regulations. This is particularly true for heavy duty diesel applications. 9.4.2 Diesel Emissions
The solids emitted from diesel engines are essentially dry soot (carbon-rich particles). The liquids are primarily unburned diesel fuel and lubricating oils (commonly referred to as soluble organic fraction or SOF) and to some extent sulfates originating from the combustion of the sulfur compounds present in the diesel fuel. The combination of solid and liquid pollutants is referred to as particulates or total particulate matter (TPM). Note that H2SO4 derived from the combustion of sulfur compounds in diesel fuel is included since it is a liquid under the collection conditions for TPM. The gaseous pollutants are CO, HC and NOx. .
Total particulate matter: –dry soot –liquids (oil, fuel), called soluble organic fraction or SOF –H2SO4.
.
Gases: –CO, HC, NOx.
j207
j 9 Automotive Emission Control: Past, Present and Future
208
Generally, an inverse relationship exists between NOx and particulate emissions, and this is called the NOx–particulate trade-off. When NOx emissions are high (e.g. at high combustion temperature), particulate emissions are low. However, when NOx emissions are low (e.g. at lower combustion temperature), the particulate emissions are high. In 1998, the US truck emission standards were such that less than 0.15 g bhp1 h1 of particulates and less than 4 g bhp1 h1 of NOx were permitted. Europe adopted the same particulate standards in 2000, but required a lower limit for NOx (2.6 g bhp1 h1). The units of grams per brake horsepower per hour (g bhp1 h1) are used for certification of heavy- and medium-duty trucks whereas the units for passenger vehicles are typically represented as grams per mile or grams per kilometer. Further restrictions in allowable particulate and NOx emissions have been adopted with the US 2010 standards being the most stringent. Particulate emissions will be limited to 0.01 g bhp1 h1 and NOx to 0.2 g bhp1 h1. As standards have become very stringent over the past 10 years [8], the demands on diesel manufacturers and catalyst companies also have become greater. 9.4.3 Diesel Oxidation Catalysts (DOCs): the Past
In 1994, diesel oxidation catalysts for diesel trucks were introduced into the US market and to a limited extent in Asia and Europe. Their primary function was to convert the liquid portion of the diesel particulates and a small amount of CO and HC to carbon dioxide and water without generating significant amounts of sulfate. Successfully converting these pollutants without the generation of sulfates would be sufficient to meet emission standards until about 2000. As discussed previously, precious metal catalysts were found to catalyze successfully the reductions of pollutants in stoichiometric gasoline engines. Consequently, the use of precious metals was also investigated intensely for controlling diesel emissions. However, because the level of sulfur in diesel fuel was about 500 ppm (0.05 wt%) in 1994, the precious metal catalysts were found to generate too much sulfate (sulfuric acid). A surprising discovery occurred in the early 1990s that cerium oxide alone was sufficiently active for converting the liquid portion of the TPM to meet the emission standards in the USA for trucks without the need for precious metals [9, 10]: SOFðoil þfuelÞ þ O2 ! CO2 þ H2 O
ð9:17Þ
Figure 9.6a shows that the conversion of SOF was independent of the Pt loading at very low Pt levels while the CO and SO2 activity was highly sensitive. As a result, a dual-function catalyst was developed whereby SOF was catalytically oxidized by ceria and CO was oxidized by small amounts of Pt. Less than 0.005% of Pt was added to the finished catalyst to reduce the gaseous CO. However, this level of Pt was too low to convert the SO2 to SO3 and form sulfate. Hence the truck regulations in the USA could be achieved with a highly active cerium oxide containing catalyst with a trace of Pt for CO control.
9.4 Diesel Catalysis
Figure 9.6 (a) Conversion of SOF and CO as a function of Pt levels; (b) effect of zeolite addition to the diesel oxidation catalyst.
This cerium-based technology was then modified for European diesel passenger vehicles [11] with the addition of a relatively small amount of zeolite plus additional precious metal to meet European requirements for gaseous CO and HC reductions. The purpose of the zeolite was to adsorb gaseous HC (mostly unburned diesel fuel) during cold start conditions and subsequently release them to the catalyst once it was sufficiently hot to initiate catalytic oxidation on the Pt. Data are shown in Figure 9.6b. These technologies were satisfactory for the standards of the time, but in 2003 new requirements for dry soot reduction required an additional solution. This led to the introduction of the wall-flow or diesel particulate filter (DPF). A DPF is a honeycomb structure with alternating or adjacent channels plugged at opposite ends [12]. This requires exhaust entering channels from one end of the monolith to pass through the channel wall and exit the opposite end of the monolith via the adjacent channel (Figure 9.7). Soot that is entrained in the exhaust stream is trapped in the wall while the gaseous components pass unrestricted. Periodically (e.g. every 1000 km of driving), the filter is heated to a temperature high enough (ca. 550 C) to combust the soot and regenerate the filter. For vehicles containing both a DOC and DPF, regeneration is accomplished by injecting diesel fuel upstream of the DOC where it is oxidized to create the exotherm required to increase the DPF temperature. Alternatively, for vehicles without a DOC, fuel is injected into the cylinders during the exhaust stroke to promote combustion in the exhaust manifold and thereby raise the exhaust and DPF temperature. For the light duty market, diesel particulate filters are usually manufactured out of silicon carbide or aluminum titanate. Silicon carbide is an ideal material for this
j209
j 9 Automotive Emission Control: Past, Present and Future
210
Figure 9.7 Capture of particulates through a wall-flow filter.
application since it has a high melting temperature, high heat capacity and high thermal conductivity. All of these characteristics enable SiC to survive the harsh environment encountered during filter regeneration without cracking or melting. However, SiC is prohibitively expensive for heavy duty applications since large filters are required. In this application, cordierite is the filter material of choice, although extreme care must be taken to include adequate engine controls to avoid thermal shock and filter damage during regeneration. A DPF may also contain a catalyst coating (i.e. a catalyzed soot filter or CSF) to assist with the combustion of soot and to oxidize CO generated during the soot regeneration process. In general, DPFs are required to meet the current US 2007 and future US 2010 medium- and heavy-duty diesel regulations. The same is true for the US EPA Tier 2 light-duty vehicle regulations that will be phased in by 2009. Likewise, filters will be required to meet the upcoming Euro 5 (2009) and Euro 6 (2014) regulations in Europe. However, due to the persistent political debate concerning unhealthy levels of ambient particulates in many European cities, several European diesel vehicle OEMs have voluntarily implemented DPFs for controlling soot emissions before the Euro 5 mandate arrives.
9.5 Diesel Emission Control: the Future 9.5.1 Catalytic Solutions for the Existing Diesel IC Engine
The US and European standards for 2010 will require the reduction of all three phases of diesel emissions (i.e. solids, liquids and gases). In particular, reduction of nitrogen oxides will offer considerable challenges. Although particulate matter after-treatment from diesel engines is fairly well advanced and has reached commercial implementation with the use of the wall-flow filter, NOx after-treatment is still under intense development. Two of the most promising technologies for controlling NOx at the tailpipe are selective catalytic reduction (SCR) and lean NOx traps (LNTs). Both utilize catalytic processes to eliminate NOx by chemically reducing it to nitrogen.
9.5 Diesel Emission Control: the Future
SCR relies on the reduction of NOx by ammonia over either a vanadia (V2O5) or metal-exchanged zeolite based catalyst [13, 14]. The major desired reactions are catalyst
4NH3 þ 4NO þ O2 ! 4N2 þ 6H2 O catalyst
4NH3 þ 2NO2 þ O2 ! 3N2 þ 6H2 O
ð9:18Þ ð9:19Þ
Although commonly used in stationary source applications for over 30 years where large supplies of ammonia are readily available [1], SCR is relatively new for vehicle applications. Since handling gaseous ammonia is not practical in automobiles or trucks, an ammonia surrogate such as urea is utilized to generate ammonia in situ in the vehicle exhaust. Typically, a solution of urea and water is injected into the exhaust stream before the SCR catalyst and, in the presence of water vapor and high temperature, is hydrolyzed to ammonia, which can thus participate in the SCR NOx reduction reactions. Although SCR technology has been successfully demonstrated for mobile applications and is in fact used on some heavy-duty diesel trucks in Europe, it relies on a extensive supply infrastructure for the distribution of urea. Development of such an infrastructure has been slow in Europe and has met significant resistance in the US. In addition, an ammonia cleanup catalyst (often referred to as an AMOX catalyst) may be required to remove any ammonia slip that may pass through the SCR catalyst unconverted. Clearly, integration of the SCR catalyst within the exhaust system may be a challenge. Depending on the specific application, an oxidation catalyst (DOC), a catalyzed soot filter (CSF), an SCR catalyst and an ammonia destruction catalyst (AMOX) may all be required to meet the combined CO, HC, total particulate matter (TPM) and NOx regulations. The relative location of each component within the system is the subject of intense development for vehicle OEMs. In addition, sophisticated engine controls are required to ensure proper operation of all components within the system. Figure 9.8 shows the catalytic unit operations anticipated for meeting 2010 diesel emission standards using an SCR system. The diesel oxidation catalyst serves to oxidize gaseous and liquid pollutants and also oxidize injected diesel fuel for soot filter (DPF) regeneration. The DPF
Figure 9.8 Proposed simplified diesel exhaust after-treatment system (2010).
j211
j 9 Automotive Emission Control: Past, Present and Future
212
Figure 9.9 Chemistry of NOx reduction in an alkaline trap.
(or CSF) traps the liquid and solid particulates, the SCR catalyst removes NOx and the ammonia decomposition catalyst destroys excess ammonia escaping the SCR. The second promising technology for controlling NOx at the tailpipe is lean NOx trapping. In Figure 9.9, the mechanism of a LNT is shown. The technology utilizes a Pt- and Rh-based TWC catalyst in combination with an NO2 trapping agent (e.g. an alkaline earth metal compound such as a Ba compound) [15]. During the normal lean operation mode of the diesel engine, NO is oxidized to NO2 over the Pt catalyst and the NO2 thus formed is adsorbed by the alkaline earth metal trapping compound incorporated within the catalyst washcoat. Periodically (e.g. every 60–120 s), the trap is regenerated by introducing a rich pulse of CO and hydrocarbons into the exhaust stream by switching the engine operating mode to stoichiometric or rich for 1–2 s. This rich pulse provides the necessary chemical reductant to convert the adsorbed nitrate to nitrogen over the Rh catalyst component via Equation 9.11. Although LNT technology has been successfully demonstrated in vehicle applications, the primary disadvantages are that high Pt levels are required to maintain sufficient catalyst durability and a significant fuel penalty (ca. 3–5%) results from the periodic trap regeneration. In addition, the sulfur oxides derived from the fuel-borne sulfur form alkaline compounds that are much more stable than the corresponding nitrates and are not removed during the stoichiometric or rich operation mode. Therefore, the trap becomes progressively poisoned by sulfur. Complicated engine control strategies are being developed by engine manufacturers to desulfate the poisoned trap by operating the engine at a high temperature (>550 C) and with a rich stoichiometric air-to-fuel ratio for a short period of time to remove the adsorbed sulfur oxides. In addition, the air-to-fuel ratio must be carefully controlled to avoid the formation of H2S during excessively rich conditions. LNT technology has the capability of removing up to 90% of the NOx in the exhaust. Having lower sulfur fuels available will favor high NOx
9.5 Diesel Emission Control: the Future
conversion levels and also reduce the requirements for desulfation. Improvements in catalyst and engine control technologies hold promise for the future. A recent publication summarizes the worldwide standards and candidate NOxcontrol catalyst technologies under consideration [16]. 9.5.2 The Homogeneous Charge Compression Ignition Engine (HCCI) and Advanced Engine Technology
An alternative approach to controlling NOx emissions from diesel engines is HCCI. HCCI is an acronym for Homogeneous Charge Compression Ignition and it is an alternative engine combustion technology that reduces NOx and particulates at the source rather than treating them in the tailpipe. Although HCCI technology has been known for many years [17], it has recently experienced renewed interest due to its benefit for emissions reduction (i.e. lower PM and NOx) [18, 19]. The key feature of HCCI is that the fuel–air charge is thoroughly mixed prior to ignition by compression. As such, HCCI combines the compression ignition feature of diesel engines with the homogeneous (vaporized) fuel characteristic of gasoline spark ignited (SI) engines. Although HCCI is valid for any fuel (e.g. gasoline or diesel), diesel HCCI has received the most interest due to its benefits for emission control. In particular, HCCI combustion has been demonstrated to reduce significantly both the particulate and NOx emissions produced during combustion. The improved fuel–air mixing resulting from the homogeneous charge eliminates the diffusion combustion characteristic of standard diesel engines and thus dramatically reduces particulate or soot formation. Better mixing of the fuel and air allows for better contact of oxygen with the fuel and thus less particulate make. In the process, the maximum combustion temperature is also reduced, which results in dramatically lower amounts of NO formation. This is illustrated in Figure 9.10, which plots the local flame temperature versus equivalence ratio (fuel:air ratio actual divided by fuel:air ratio at stoichiometric) in the piston and the regions where soot and NOx are generated.
Figure 9.10 Engine map of flame temperature verses equivalence ratio for the HCCI engine. Reprinted with permission from SAE.
j213
j 9 Automotive Emission Control: Past, Present and Future
214
In the case of HCCI, the combustion temperature is lowered out of the region for NOx formation and the equivalence ratio is lowered out of the range of soot formation (i.e. the air:fuel ratio is increased). For sustainable development of diesel engines in the future, lower NOx and particulates is the real potential of HCCI and, in fact, the use of particulate and NOx control after-treatment may not be required. Because diesel fuel is much less volatile than gasoline, ignition delay is the key for achieving the correct combustion conditions for HCCI. Ignition delay is the time difference between fuel injection and fuel ignition. In standard diesel engines, the injection of fuel and the start of combustion overlap in time. In other words, combustion begins before injection has stopped. As stated previously, this leads to diffusion combustion (i.e. burning from the outside of the fuel spray to the center) and the formation of high levels of particulates and NOx. However, in the case of HCCI, fuel injection stops before combustion begins and, in general, the longer the delay time, the better is the mixing. In the ideal case, the diesel fuel has vaporized and complete mixing has occurred. Unlike standard diesel combustion, no discernible flame propagation occurs; rather, combustion occurs simultaneously throughout the cylinder. This leads to a very high combustion rate and a very fast pressure rise in the cylinder. The heat release is over within a few crank angle degrees. From a practical consideration, this rapid pressure release creates excessive noise and it can make the combustion process very hard to control, particularly at high engine loads. As a result, HCCI is currently practical only for low-load diesel operation such as in light-duty diesel vehicles [19, 20]. In addition, although NOx and particulate emissions are reduced substantially, carbon monoxide and hydrocarbon emissions are increased. As a result, the focus of exhaust remediation is shifted from controlling NOx and particulates to controlling CO and hydrocarbons. Depending on their tailpipe concentrations, catalytic reduction of CO and hydrocarbons may be a challenge. Due to the difficulty in achieving complete mixing of the diesel fuel and air charge and because of the practicality of controlling HCCI combustion under high loads, commercialization of the technology has not yet been realized. However, engine manufacturers are continually working to optimize the combustion process of their current diesel engines to achieve the same NOx-reducing benefit of HCCI. If a stable and viable engine technology can be developed, then use of SCR and LNT technology could be avoided in order to meet NOx emission regulations. Over the past several years, numerous advances with new combustion strategies or improved hardware have been realized. For example, the use of EGR (exhaust gas recirculation), optimized fuel injection timing, high-pressure fuel injectors, multi-hole fuel injectors and variable valve actuation have been shown to help reduce engine out NOx formation (for a review on recent HCCI engine advances, see [21]). Although particulate matter formation may increase in some cases (e.g. with higher levels of EGR), this is not considered a major problem since the increased particulate emissions can be handled easily with a standard wall flow particulate filter. However, like true HCCI, one common characteristic of these advanced combustion techniques is the formation of higher carbon monoxide and hydrocarbon levels in the exhaust relative to a conventional diesel engine. In these cases, the focus to meet emission regulations then becomes the development of oxidation catalysts capable of
9.5 Diesel Emission Control: the Future
achieving high levels of CO and hydrocarbon conversion under the low-temperature conditions of diesel exhaust. Although regulations for NOx and particulate emissions from diesel vehicles have become significantly more stringent in recent years, those for HC and CO have largely remained unchanged. Nevertheless, meeting these regulations with new HCCI or advanced combustion technologies will become more difficult due to the increased emissions of CO and hydrocarbon from these engines. For light duty vehicles, it is likely that CO and HC emissions will increase two to four times over the levels from current engines. As a result, end of useful life efficiency requirements (e.g. after 160 000 km for Euro 6 light duty standards or 120 000 miles for US Tier 2) will realistically be in the region of 90%. This is a significant change from current engines, where only 50–60% conversion is typically required. In addition, as CO and hydrocarbon concentrations increase, catalytic performance generally decreases. This is especially true when considering the oxidation of CO over Pt, which in fact shows a very strong tendency for poisoning by CO (i.e. as CO concentration increases, Pt catalyst performance decreases via the CO self-inhibition effect). Therefore, new catalyst technologies will need to be developed to meet these new requirements. The impact of the increased CO and HC emissions on catalyst performance is illustrated in Figure 9.11, which plots CO and HC conversion for two aged catalyst formulations tested both on current Euro 4 emission standard and on developmental advanced combustion engines. The first formulation is comprised only of Pt as the active catalyst ingredient whereas the second is comprised of Pt and Pd. In the case of the current Euro 4 diesel engine, the Pt catalyst performs significantly better than the Pt/ Pd catalyst (Figure 9.11) after high-temperature post-injection aging on a vehicle (25 h, 700 C). However, on the advanced combustion engine, which has four times higher CO emissions and three times higher HC emissions relative to a conventional
Figure 9.11 Catalytic control of CO and HC emissions from standard and HCCI engines using a Pt and Pt/Pd catalyst.
j215
j 9 Automotive Emission Control: Past, Present and Future
216
diesel engine, the performance of the two catalysts is very similar after oven aging (650 C, 10% steam, 5 h). Clearly, the quantity of CO and HC has an important effect on catalyst performance. Although high CO and HC emissions push the conversion requirements for oxidation catalysts to higher levels, a side benefit is the higher heat release within the catalyst. As emission levels increase, the heat released by these emissions during combustion over the catalyst also increases. In other words, the higher the emissions, the greater is the exotherm that is generated over the catalyst once it has become active. This increases the temperature of the catalyst, which further enhances the catalyst performance. The extent of the exotherm is illustrated for the Pt/Pd catalyst described previously and tested on both advanced combustion and standard diesel combustion engines. Figure 9.12 shows the inlet and outlet temperatures of the Pt/ Pd catalyst driven on the European certification drive cycle (MVEG) for the advanced combustion engine. Once the catalyst has become active after approximately 120 s, the outlet temperature of the catalyst is significantly higher than the inlet temperature due to the large exotherm generated by combusting the high concentration of CO and HC emissions. However, when the same catalyst is tested on the standard Euro 4 diesel engine where the CO and hydrocarbon emissions are much lower, the outlet temperature is lower than the inlet temperature (Figure 9.13). Little if any exotherm is created within the catalyst and the reaction temperature therefore is limited by the exhaust temperature. In the advanced combustion case, the high emission concentrations can generate a large exotherm in the catalyst, which helps to sustain and drive the oxidation reaction. Since little CO and hydrocarbon breakthrough will occur once the catalyst becomes active, then the overall performance of the catalyst in meeting the emission regulations is determined by the amount of emissions that remain unconverted during the first 1–2 min of engine operation when the catalyst has not yet become active (cold-start phase).
Figure 9.12 DOC inlet and outlet temperatures on diesel advanced combustion engine.
9.6 Fuel Cells and the Hydrogen Economy for Transportation Applications: the Future
Figure 9.13 DOC inlet and outlet temperatures on standard diesel engine.
Clearly, HCCI and related advanced combustion engine technologies hold promise for reducing NOx and PM emissions from diesel engines. Whether these technologies can reach commercialization or not will depend primarily on further improvements to combustion control and subsequent drivability of the vehicle (e.g. higher load operation with less noise). Additionally, development of new catalyst technologies will be required in order to control the higher emissions of CO and hydrocarbons that will be produced from these new engines. Ultimately, the cost of the necessary engine and catalyst advancements in relation to developing SCR and LNT NOx-reducing technologies will determine the market viability of HCCI.
9.6 Fuel Cells and the Hydrogen Economy for Transportation Applications: the Future 9.6.1 The Fuel Cell
The H2–O2 fuel cell (Figure 9.14) operates by the electrocatalytic oxidation of H2 and reduction of O2 to form H2O, electricity and heat. The fuel cell reaction will run indefinitely provided that H2 and O2 are supplied and the electrocatalysts retain their activity. Both anode and cathode reactions are catalyzed by Pt on carbon, however, their respective compositions are different. Pt=C
H2 ! 2H þ þ 2e Pt=C
O2 þ 4H þ þ 4e ! 2H2 O
E ¼ 0:00 V
ð9:20Þ
E ¼ 1:23 V
ð9:21Þ
Adding the two half-cell reactions gives Pt=C
Net reaction : H2 þ 1=2 O2 ! H2 O
E cell ¼ 1:23 V
ð9:22Þ
j217
j 9 Automotive Emission Control: Past, Present and Future
218
Figure 9.14 A single cell of the PEM fuel cell.
The net voltage of the cell (Ecell) is related to standard state cell voltage (E cell) and the partial pressures of the reactants and products through the Nernst equation: 1
E cell ¼ E cell þ RT=2FlnðP H2 Þ=P H2 O þ RT=2FlnðP O2 Þ2
ð9:23Þ
The anode and cathode compartments are separated by a solid polymer membrane composed of a fluorocarbon with a substituted sulfonic acid group to conduct Hþ . The membrane is permeable to the Hþ ions generated at the anode that migrate to the cathode, where they combine with the O2 and form H2O. The membrane is impermeable to H2 and O2 and prevents mixing [22]. 9.6.2 Fuel Cells for Transportation
The IC engine was developed in Germany in the late 19th century and mass production began in the early 20th century. Although continuously improved over the years, the IC engine is still the engine used today. In 2003, Japanese car makers (Toyota and Honda) introduced hybrid vehicles combining an IC engine with a battery for the power train. In 2005, demonstration vehicles were introduced that selectively combined a fuel cell and a battery. This was the first indication of the feasibility of replacing the IC engine with a new power source which operates on H2 and air with
9.6 Fuel Cells and the Hydrogen Economy for Transportation Applications: the Future
high efficiency and virtually no pollutants. Today, all major car companies have advanced fuel cell programs, commercial targets varying between 2011 and 2020. Critical issues being addressed are reduction of the expensive platinum content in the fuel cell itself, enhanced life of the solid polymer membrane and safe storage of hydrogen on-board. Notwithstanding the actual vehicle issues is the necessity to develop a hydrogen infrastructure. 9.6.3 The Hydrogen Service Station
The fundamental question is where the hydrogen will come from for the hydrogen economy. Hydrogen is the most abundant element on Earth but it is bound to H2O. Clearly, we are looking at renewable fuels and eventually natural sources of energy such as solar and wind to provide the energy to decompose water to generate the necessary hydrogen. In the interim, we can consider natural gas as a source since it is 25% hydrogen compared with less than 10% for most liquid fuels, maintaining a lower greenhouse gas footprint. Possibly combinations of ethanol and gasoline (E85) now present in many service stations in the USA can also be considered a source of hydrogen via the steam reforming process. Furthermore, an elaborate infrastructure exists for natural gas in the form of pipelines and services to many homes and buildings around the world. So the approach has been to convert catalytically the natural gas to hydrogen of fuel cell quality [23]. Such a system is conceptually shown in Figure 9.15. It converts pipeline natural gas to hydrogen of fuel cell quality. Part of
Figure 9.15 Hydrogen filling station on the hydrogen highway using catalyst technology for reforming natural gas.
j219
j 9 Automotive Emission Control: Past, Present and Future
220
the hydrogen is used in a fuel cell providing electricity to the service station, with the majority used to provide hydrogen for fuel cell vehicles.
9.7 Conclusions
The use of catalysis for emission control from pollutants generated from gasolinefueled internal combustion engines has been a remarkable demonstration of the application of heterogeneous catalysis for environmental control in the consumer market. Its success has stimulated the development of advanced technologies, including the modern three-way catalyst capable of simultaneously converting CO, HC and NOx, where the catalyst dictates the operational mode of the engine via the oxygen sensor and the feedback control loop. The technology has been now been modified and extended to controlling the complicated emissions from a diesel engine including CO, HC, NOx and particulates. As emission standards become more and more stringent, other technologies such as the homogeneous charged compression ignition engine have become a potential advancement in particulate emission control. Finally, the convergence of alternative energy and pollution free power generation, via the hydrogen economy, and the fuel cell, promises freedom from petroleum-based fuels and elimination of greenhouse gas emissions. This is clearly a viable goal which will require advances in catalysis and catalytic engineering which are close at hand.
References 1 Heck, R. and Farrauto, R. (2002) Catalytic Air Pollution Control: Commercial Technology, John Wiley & Sons, Inc, Hoboken, NJ. 2 Bartholomew, C. and Farrauto, R.J. (2005) Fundamentals of Industrial Catalytic Processes, 2nd edn John Wiley & Sons, Inc, Hoboken. NJ, Chapters 1–5. 3 Farrauto, R.J. (2007) Industrial catalysis: a practical guide, in Kent and Riegels Handbook of Industrial Chemistry and Biotechnology, 11th edn (ed. J. Kent), Springer, Berlin, Vol. 1, pp. 271. 4 Shoup, R., Hoekstra, K. and Farrauto, R. (1975) Thermal stability of a copper chromite autoexhaust catalysts. American Ceramic Society Bulletin, 54, 576. 5 Farrauto, R. and Wedding, B. (1973) Poisoning by SOx of some base metal oxide
6
7
8 9
10
automobile catalysts. Journal of Catalysis, 33, 249. Thompson, C., Mooney, J., Keith, C. and Mannion, W. (1979) Polyfunctional catalysts US Patent 4 157 316. Akihama, K., Takatori, Y., Inagaki, K., Sasaki, S. and Dean, A.M. (2001) Mechanism of the smokeless rich diesel combustion by reducing temperature, 200101-0655, Society of Automative Engineers, Warrendale, PA. dieselnet.com. Farrauto, R. and Voss, K. (1996) Monolithic diesel oxidation catalysts. Applied Catalysis B-Environmental, 10, 29. Farrauto, R., Voss, K. and Heck, R. (1997) Ceria–alumina oxidation catalysts US Patent 5 627 124.
References 11 Yavuz, B., Voss, K., Deeba, M., Adomaitis, J. and Farrauto, R. (2001) Zeolite addition to a diesel oxidation catalyst US Patent 6 248 684. 12 Howitt, J. and Montierth, M. (1981) Cellular ceramic diesel particulate filter 810114, Society of Automative Engineers, Warrendale, PA. 13 Byrne, J., Chen, J. and Speronello, B. (1992) Catalysis Today, 13, 33–42. 14 Bosch, H. and Janssen, F. (1988) Catalysis Today, 2, 369–521. 15 Miyoshi, M., Matsumoto, S., Katoh, K., Tanaka, T., Harada, J., Takahashi, N., Yokato, K., Sigiura, M. and Kasahara, K. (1995) A new approach to NOx reduction in lean burn engines, 950809, Society of Automative Engineers, Warrendale, PA. 16 Johnson, T. (2008) Diesel engine emissions and their control. Precious Metal Reviews, 52, 23–37. 17 Thring, R.H. (1989) Homogeneous charge compression ignition (HCCI) engines, 892068, Society of Automative Engineers, Warrendale, PA. 18 Epping, K., Aceves, S. and Bechtold, R. (2002) The potential of HCCI combustion for high efficiency and low emissions, 2002-01-
19
20
21
22
23
1923, Society of Automative Engineers, Warrendale, PA. Stanglmaier, R.H. and Roberts, C.E. (1999) Homogeneous charge compression ignition (HCCI): Benefits, compromises, and future engine applications, 1999-013682, Society of Automative Engineers, Warrendale, PA. Leet, J.A., Simescu, S.S., Froelund, K., Dodge, L.G. and Roberts, C.E. (2004) Emission solutions for 2007 and 2010 heavyduty diesel engines, 2004-01-0124, Society of Automative Engineers, Warrendale, PA. Zhao, F., Asmus, T., Assanis, D., Dec, J., Eng, J. and Najt, P. (2003) Homogeneous Charge Compression Ignition (HCCI) Engines – Key Research and Development Issues, Society of Automotive Engineers, Warrendale, PA. Farrauto, R. (2005) Introduction to solid polymer membrane fuel cells and reforming natural gas for the production of hydrogen. Applied Catalysis BEnvironmental, 56, 3. Farrauto, R., Liu, Y., Ruettinger, W., Ilinich, O. and Giroux, T. (2007) Precious metal monoliths for natural gas reforming. Catalysis Reviews-Science and Engineering, 247, 112.
j221
j223
10 Heterogeneous Catalysis for Hydrogen Production Morgan S. Scott and Hicham Idriss
10.1 Introduction
Since the industrial revolution, civilization has relied on the combustion of fossil fuels as the energy provider. The large-scale combustion of the fossil fuels coal, gas and petroleum oil is emitting vast amounts of harmful pollutants into the atmosphere and is upsetting the natural equilibrium, that of the carbon cycle. The reliance of our economy on fossil fuels is effectively promoting an inhospitable present and an apprehensive future. The harmful emissions from fossil fuel combustion include carbon dioxide, carbon monoxide, oxides of nitrogen, sulfur dioxide, particulates, heavy metals (Hg, Pb) and VOCs (volatile organic compounds) (Table 10.1). Natural gas is the cleanest of the fossil fuels; it is composed primarily of methane. Coal and oil are composed of more complex molecules some contain nitrogen and sulfur atoms, and when combusted they release NOx and SO2, in addition to large amounts of ash particulates and VOCs. VOCs are products of incomplete combustion that often deteriorate photochemically to methane in the atmosphere. Massive carbon reservoirs that have for millions to hundreds of millions of years been buried deep beneath the earth now exist as carbon dioxide in our atmosphere due to fossil fuel combustion. The International Panel on Climate Change (IPCC) reports indicate atmospheric CO2 levels have risen 100 ppm over the industrial era and 2.6 ppm in 2005. For example, from 2000 to 2005, the growth rate of carbon dioxide emissions was more than 2.5% per year, whereas in the 1990s it was less than 1% per year. The Energy Information Administration (EIA) reports that 81.2% of greenhouse gas emissions in the USA in 2000 came from CO2 directly attributed to the combustion of fossil fuels. Air quality is an important environmental issue for large metropolitan cities. Under certain geographical circumstances the photochemical smog phenomenon can arise, resulting in significant human health hazards. Smog is a product of exhaust and flue gas emissions and when photochemical interaction with sunlight occurs, highly toxic compounds are formed. Tropospheric ozone is a highly active component of the photochemical smog. Acid rain is another environmental problem
Handbook of Green Chemistry, Volume 2: Heterogeneous Catalysis. Edited by Robert H. Crabtree Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32497-2
j 10 Heterogeneous Catalysis for Hydrogen Production
224
Table 10.1 Fossil fuel emission levels [1].a
Pollutant
Natural gas
Oil
Coal
CO2 CO NOx SO2 Particulates Mercury
117 000 40 92 1 7 0.000
164 000 33 448 1 122 84 0.007
208 000 208 457 2,591 2 744 0.016
Units are in pounds (lb) per billion Btu of energy input. Btu ¼ British thermal unit: the amount of heat required to increase the temperature of 1 lb of water 1 F.
a
sourced from the combination of the photochemical smog with rain; acid rain is very damaging to crops, forests, wildlife and human health. 10.1.1 Renewable Energy
Renewable energy is energy derived from resources that are not depleted with time, which include sun, wind and water movement, or replenished with use, biomass and waste. These primary sources of energy can be converted into electricity, heat and mechanical energy in several ways. Technologies already exist for the conversion of renewable energy such as hydropower, biomass and waste combustion. Developing technologies include wind turbines, photovoltaic cells and geothermal energy. Ethanol produced from biomass is already established as a clean alternative to petroleum fuels for transportation. Ethanol is a renewable, portable, relatively nontoxic, biodegradable liquid that offers huge potential as the savior substance in the energy crisis. Ethanol–gasoline blends containing up to 85% v/v ethanol (E85) are currently being used in combustion engines with minor adjustments. Ethanol production from biomass is a well-established technique that has been developed to include production from cellulosic material, although this new method is still in the development stage. Apart from being renewable and relatively nonpolluting, another advantage of ethanol is that its use as a fuel adds no net carbon dioxide to the atmosphere. Ethanol is made by converting the carbohydrate portion of biomass into sugar, which is then converted into ethanol in a fermentation process. Carbohydrates are formed through photosynthesis via the assimilation of carbon dioxide and water from the atmosphere. Any carbon dioxide released in the use of ethanol as fuel has already been removed from the atmosphere, hence there is no net emission. The Kyoto protocol is an international agreement enforcing the reduction of greenhouse gas emissions (the six greenhouse gases stipulated by the Kyoto protocol are CO2, CH4, N2O, hydrofluorocarbons, perfluorocarbons and sulfur hexafluoride). Under the Kyoto protocol, it is economically favorable to reduce greenhouse-contributing emissions. The renewable energy cycle for photosynthetic glucose production, fermentation to ethanol by yeast and steam reforming of the crude ethanol to fuel-grade hydrogen, offers a viable choice for future energy
10.1 Introduction
production. The hydrogen can then be used by a fuel cell to produce electricity and water. One of the most attractive processes for making hydrogen from ethanol is steam reforming of ethanol to produce H2 and CO2. The overall process is an ideal renewable energy cycle that may be simplified to reaction 10.1 (the sum of reactions 10.1a and 10.1b). Reaction 10.1 shows the transformation of glucose to hydrogen gas and carbon dioxide with water as the oxidant via ethanol. C6 H12 O6 ! 2C2 H5 OH þ 2CO2
ð10:1aÞ
2C2 H5 OH þ 6H2 O ! 12H2 þ 4CO2
ð10:1bÞ
C6 H12 O6 þ 6H2 O ! 12H2 þ 6CO2
ð10:1Þ
The use of steam reforming for hydrogen production as here alleviates the necessity for the energy-intensive process of separating water from ethanol during distillation as the feed stream for the process is an azeotropic mixture of ethanol and water. 10.1.2 Hydrogen
Hydrogen is the most abundant element in the Universe; it is found as interstellar gas and as the chief constituent of main sequence stars, but very little is found on Earth in its elemental form. On Earth, hydrogen is found mainly as a part of the molecules of water and organic material. Hydrogen has the lowest density of all elements and therefore rapidly disperses and quickly ascends to the upper atmosphere. Hydrogen has the highest combustion energy release of any commonly occurring material, making it an ideal fuel. This property and its low weight make it the fuel of choice for the upper stages of multi-stage rockets. Hydrogen combustion is clean (producing only water) and very efficient (as indicated by its very high BTU [2]). Hydrogen possesses the highest energy content per unit of weight (120.7 kJ g1) compared with any of the known fuels [3]. However, there are issues with the storage and transport of hydrogen that must be addressed. Compressed gas storage of hydrogen, typically at pressures of 10–35 MPa, is the most common storage forms used today; tests are ongoing to increase pressure and hence storage capacity for passenger vehicles, while maintaining a reasonable system weight. Compression may take place at filling stations receiving hydrogen from a pipeline; however, transportation, compression and transfer are expensive operations and require a completely new, state-of-the-art infrastructure. Other alternatives include storage. Storage of liquid hydrogen requires refrigeration to a temperature of 20 K, another energy expensive method first developed for space travel. Hydrogen storage as hydrides is another area of investigation; however, hydride storage has many downsides. Typical storage densities are still only 10% or less compared with conventional fuels and a low hydrogen fraction by mass (less than 10%), casting doubt on this technique for mobile applications (Table 10.2). Heat is released when
j225
j 10 Heterogeneous Catalysis for Hydrogen Production
226
Table 10.2 Energy density by weight and volume and mass density
for various hydrogen storage forms [4]. Energy density Storage form
kJ kg1
MJ m3
Density (kg m3)
Hydrogen gas (0.1 MPa) Hydrogen gas (30 MPa) Hydrogen liquid Hydrogen in metal hydrides Hydrogen in metal hydrides, typical Methane (natural gas) at 0.1 MPa Methanol Ethanol
120 000 120 000 120 000 2 000–9 000 2 100 56 000 21 000 28 000
10 2 700 8 700 5000–15 000 11 450 37.4 17 000 22 000
0.090 22.5 71.9 5480 0.668 0.79 0.79
hydrogen enters the lattice and heat must be supplied to drive hydrogen out of the lattice, causing a potential for inefficient desorption of hydrogen. Other hydrogen storage methods include cryoadsorbed gas storage in carbon materials and storage as a hydrocarbon. Transportation, storage and safety issues may be overcome if we consider the use of ethanol as the energy carrier with conversion of ethanol to hydrogen immediately prior to use as a fuel. Onboard conversion of ethanol to hydrogen is apparently the ideal step in the efficient turnover of the petroleum economy to a hydrogen economy as there is no need for an extensive new infrastructure for fueling. Infrastructure for ethanol production and distribution is already established in many developed countries as ethanol is already used as an octane enhancer or oxygenate blended with gasoline. Both steam reforming and partial oxidation of ethanol appear as processes capable of converting ethanol to hydrogen; however, this process is still in the research and development stage. Fuel cells being developed are electrochemical cells capable of efficiently turning hydrogen and oxygen into water, creating an electric current [4]: 2H2 þ O2 ! 2H2 O þ 7:9 1019 J
ð10:2Þ
10.1.3 Hydrogen from Ethanol Decomposition
A recent topic of great interest is the catalytic decomposition of ethanol in partial oxidation and steam reforming processes. Ethanol reforming producing hydrogen for fuel cells to produce electricity appears as the ideal scenario for future energy requirements. The major issue for the production of hydrogen from ethanol is the development of a suitable catalyst. The ideal reactions for the catalytic decomposition of ethanol are the partial oxidation of ethanol (reaction 10.3) and steam reforming of ethanol (reaction 10.4). These equations are ideally described; they do not,
10.1 Introduction
however, proceed to completion and considerable amounts of carbon monoxide may be formed. CH3 CH2 OH þ 3=2O2 ! 3H2 þ 2CO2
DH ¼ 553 kJ mol1 ð10:3Þ
CH3 CH2 OH þ 3H2 O ! 6H2 þ 2CO2
DH ¼ þ 174 kJ mol1 ð10:4Þ
Ethanol can be manufactured from biomass balancing net atmospheric CO2 emissions, and has a better hydrogen yield than methane, the traditional source. An obvious obstacle in the decomposition of ethanol is that its CC bond must be broken. Depending on the interaction with the catalyst, the nature of the catalyst and the reaction conditions, one may be able to manipulate the reaction to preferred products. The process of adsorption is governed by orbital interaction of the molecule with the surface. Figure 10.1 gives a representation of the ethanol molecule and a typical ordered surface of CeO2. The electrostatic potential has a contribution from the two lone-pair electrons of the oxygen atoms in addition to the positive charge around the hydrogen atom of the hydroxyl. The interaction occurs between the hydroxyl (Od Hd þ ) group and the corresponding Ced þ Od of the surface; Ce has a
Figure 10.1 A representation of two ethanol molecules and an ordered surface of CeO2. The electrostatic potential is represented for the ethanol molecule on the left and the HOMO level is represented for the ethanol molecule on the right; also included are approximate bond strengths within the ethanol molecule.
j227
j 10 Heterogeneous Catalysis for Hydrogen Production
228
formal charge of þ 4 and O has a formal charge of 2. As a result, OH bond dissociation may occur: CH3 CH2 OH þ Ce4 þ O2 ! CH3 CH2 OCe4 þ þ OH
ð10:5Þ
The choice of the catalyst is crucial in the decomposition of ethanol for optimal hydrogen production. The catalyst must be active for the selective oxidation of both carbon atoms to CO2 without the oxidation of H2 to H2O; the ideal reaction can be shifted from CO to CO2 and from H2 to H2O. H2 oxidation has a low activation energy whereas CO oxidation has a high activation energy. Under these conditions, increasing the reaction temperature would favor the oxidation of CO. 10.1.4 Catalytic Oxidation
Total oxidation of hydrocarbons will yield CO2 and H2O and hence is not favorable in the interests of the production of hydrogen. Partial oxidation processes yielding H2 are therefore of more relevance. Steam reforming processes are also synonymous with catalytic oxidation and will be considered further in the next section. Partial oxidation systems have several distinct advantages over steam reforming systems such as fast startup and response times and the reactor is more compact than a steam reformer since it does not require the addition of a heat exchanger as partial oxidation is exothermic. Pt/CeO2 has been reported to be an efficient catalyst for the partial oxidation of natural gas for hydrogen production and recent studies report its potential for ethanol partial oxidation in the production of hydrogen [5]. The main drawback of partial oxidation is the need to separate oxygen from hydrogen molecules and, in the case when air is used, the additional separation of nitrogen and the possible formation of nitrogen-containing compounds. These additional difficulties increase the cost and complexity of the method and offset the energy gained for the exothermic reaction. 10.1.5 Steam Reforming
Steam reforming of methane is a well-established industrial technique for hydrogen production. Carbon of one form or another has long been used to extract hydrogen from water (see the water gas shift reaction in Section 10.1.7). Catalytic steam reforming of natural gas is the most energy-efficient process available for current commercial hydrogen production. Steam reforming of ethanol aims at cracking ethanol in the presence of steam over a catalyst producing as much hydrogen and CO2 as possible; see reaction 10.4. Depending on the catalyst used, there are several reaction pathways that may occur in the steam reforming of ethanol. Rh/CeO2 catalyst shows excellent activity over the temperature range 573–1073 K [3]. However, this catalyst is observed to produce significant amounts of CH4 and CO even though it has good water gas shift reactivity [3]. In general, the selectivity to hydrogen is twice as high for steam reforming than partial oxidation [6].
10.1 Introduction
10.1.6 Dry Reforming
Carbon dioxide may also be used in the presence of a catalyst at elevated temperatures to oxidize ethanol [7]. The use of carbon dioxide as an oxidant is known in methane reforming; however, it is a lesser used reforming process. Dry reforming does not excel in hydrogen production as no excess hydrogen is produced from the oxidant. CH3 CH2 OH þ CO2 ! 3H2 þ 3CO
DH ¼ 296 kJ mol1
ð10:6Þ
The mechanism proceeds by donation of one of the oxygen atoms of CO2 to a carbon atom and the subsequent evolution of CO and H2. The process is markedly endothermic. 10.1.7 Water Gas Shift Reaction (WGSR)
The WGSR is an intrinsic inclusion into the reforming process; the WGSR is seen to increase the H2/CO ratio. Most catalysts in the steam reforming of ethanol produce CO. The WGSR involves an equimolar mixture of steam and carbon monoxide and the process is moderately exothermic. It is an important step in the reforming process. One of the accepted mechanisms of WGSR occurs as follows (where represents an empty site): CO þ OðaÞ ! CO2 þ
DH ¼ 283 kJ mol1
ð10:7Þ
H2 O þ ! H2 þ OðaÞ
DH ¼ 242 kJ mol1
ð10:8Þ
Combining reactions 10.7 and 10.8 gives the WGSR: CO þ H2 O ! CO2 þ H2
DH ¼ 41 kJ mol1
ð10:9Þ
Since the WGSR is moderately exothermic, the equilibrium constant decreases with increasing temperature and high conversions are favored by low temperatures. The presence of steam in quantities greater than the stoichiometric quantity improves conversion. In addition to the WGSR, other processes may be at work to reduce the presence of CO. For example, the methanation reaction converts any residual carbon oxides to methane and water, consuming hydrogen, and is thus not the preferred method of CO elimination. Pt, Pd and Rh have been found to be equally effective catalysts in the WGSR [8]. The WGSR kinetics on CeO2-supported Pt, Pd and Rh suggest that the reaction mechanism proceeds via the metal-adsorbed CO oxidized by ceria and ceria re-oxidized by water. On Pd/CeO2, investigation of the WGSR mechanism revealed the formation of Pd b-hydride [9], which greatly suppresses CO oxidation because any oxygen reacts with H to form water, which desorbs. The low-temperature WGSR on Pd/CeO2 is hindered by less water on the ceria surface, less linearly adsorbed CO on Pd and significantly more formate species blocking the
j229
j 10 Heterogeneous Catalysis for Hydrogen Production
230
reaction in comparison with Pt/CeO2. At high temperatures, the WGSR activity of Pd/CeO2 remains lower than that of Pt/CeO2. 10.1.8 Catalytic Reforming of Methane
Industry currently produces hydrogen from steam reforming of methane/natural gas: CH4 þ H2 O ! CO þ 3H2
DH ¼ 206 kJ mol1
ð10:10Þ
This reaction requires a catalyst and typically proceeds at temperatures of about 850 C and a pressure of around 2.5 106 Pa (25 atm.). Methane reforming is endothermic and, in order to obtain a high conversion efficiency, some heat inputs are taken from cooling the reactants and from the heat outputs of the subsequent WGSR (reaction 10.9). The WGSR usually takes place in a second reactor where heat is recovered and recycled back to the first reaction. A steam/methane ratio above unity is necessary to avoid carbon and excess CO formation. Industrial steam reformers typically use total combustion of a fraction of the methane to provide the heat required for the process. Reactions 10.10 and 10.9 combine to give CH4 þ 2H2 O ! CO2 þ 4H2
DH ¼ 164 kJ mol1
ð10:11Þ
Carbon dioxide reforming of methane is another investigated pathway for hydrogen production; however, the process results in rapid carbon deposition on the catalysts causing deactivation due to sintering: CH4 þ CO2 ! 2CO þ 2H2
DH ¼ 246 kJ mol1
ð10:12Þ
Partial oxidation of methane is also a pathway viable for hydrogen production from methane. Partial oxidation produces hydrogen from the hydrocarbon only and does not include any water in the process. The process of partial oxidation is carried out on the same type of catalyst as the steam reforming process but in the presence of oxygen: CH4 þ O2 ! 2H2 þ CO2
DH ¼ 320 kJ mol1
ð10:13Þ
Significant amounts of methane are observed in the catalytic decomposition of ethanol. In addition to ethanol catalytic decomposition, it is desirable to catalyze methane oxidation to maximize hydrogen production. Recent work has established Rh/CeO2, Pt/CeO2 and Pd/CeO2 as catalysts active for the oxidation of methane in steam reforming [10]. 10.1.9 Thermodynamics
Ethanol conversion to hydrogen has been investigated by thermodynamic analyses and computer simulations [11]. The method of production considered was indirect
10.2 Catalysis
Figure 10.2 Hydrogen molar fraction to methane molar fraction ratio versus temperature at atmospheric pressure. Data are replotted from [12].
partial oxidation. The results indicate that there is an optimal water/fuel ratio for maximum hydrogen production at each operating condition. It should be noted that the H/O and H/C ratios are 4 : 1 when the water to ethanol ratio is 1 : 1. Figure 10.2 illustrates the advantages of working at low pressures, high temperatures and high water to ethanol feed ratio. However, high temperatures and low pressures also favor CO formation. Water to ethanol ratios in the feed >2 are necessary to obtain hydrogen and prevent carbon formation. An ideal catalyst must be found in order to achieve satisfactory production and selectivity of hydrogen at moderate temperatures and water to ethanol feed ratios.
10.2 Catalysis
The catalyst acts by providing an alternative reaction pathway with a lower activation energy compared with the same reaction in the absence of a catalyst and consequently raises the rate of the reaction. Heterogeneous catalysis implies that the catalyst and the reagents are in different phases (i.e. solid/gas). Heterogeneous catalysis usually depends on the reactants being adsorbed (usually chemisorbed) and modified to a form that will more readily undergo reaction. Catalytic oxidation is widely utilized in industry and in pollution control. Sometimes it is desirable to achieve total oxidation and sometimes partial oxidation is the goal. Oxidation is often catalyzed by transition metal oxides; the physical chemistry of these oxide surfaces is complex. Catalytic reforming uses a dual-function catalyst, often a noble metal supported on a semiconducting oxide. The exact role of the noble metal may not be easily understood or studied since metal–support interaction often results in alterations to the properties of both the metal and the support.
j231
j 10 Heterogeneous Catalysis for Hydrogen Production
232
10.2.1 The Noble Metals Pd and Rh
The noble metals of concern for the reforming of ethanol are Pd, Pt and Rh. They belong to the platinum group metals (rows 5 and 6 of groups 9 and 10), so called because they are outstanding in catalytic activity and platinum is the strongest oxidation catalyst. Their catalytic activity in oxidation is led by platinum; the other metals act in the same manner but to a lesser extent. However, they do vary in catalytic action; rhodium has catalytic properties that stand alone, as does palladium. Rhodium excels in scission of the carbon–carbon bond and Rh/CeO2 shows activity in ethanol decomposition at room temperature [13]. Palladium is known for its hydrogenation and a dehydrogenation catalytic activity [14] and is markedly active in the steam reforming of methane [10]. Due to rarity and volatility at high temperatures, the use of the noble metals in industrial catalysis is generally limited to Pt, Pd and Rh; the activity of these metals in oxidation reactions generally follows the order Rh < Pd < Pt. Pt shows high activity for oxidizing CO and saturated hydrocarbons whereas Pd has been found to be most active for the total oxidation of methane [15]. It has been found that combinations of metals give higher activity than the metals alone in oxidation processes [16]. These three noble metals are marked by some peculiar properties. Pt as a catalyst will initiate some chemical reactions at room temperature (e.g. oxidation of CH3OH to HCHO and the explosive reaction of H2 with O2) [17]. Pd has the unusual ability to absorb H2 to up to 900 times its own volume (of the gas at room temperature), H2 actually dissociates and permeates through the metal lattice [17]. Rh, when heated to its melting point, absorbs oxygen but does not form an oxide; when it solidifies it releases the oxygen [17]. The three noble metals display unique catalytic redox activities. Pt and Pd are very active in the catalytic function of hydrogenation, dehydrogenation and oxidation, and they are also recognized as catalysts for steam reforming. Rh is recognized to be active for CC bond cleavage and hydrogenolysis. Bimetallic PtRh and PdRh are also known for their unique catalytic activity in bond dissociation, hydrogenation, dehydrogenation and oxidation. Computation modeling using density functional theory (DFT) for ethanol decomposition on a Pt(111) surface was conducted and a reaction mechanism has been proposed [18], as represented in Scheme 10.1.
Scheme 10.1 Ethanol decomposition pathways over a Pt(111) surface.
The decomposition has two parallel paths generated by competition between CH and CC bond cleavage in the acetyl (CH3CO) intermediate. About 15% of the ethanol that reaches the acetyl stage decomposes via the ketene intermediate (CH2CO) [18]. The mechanism of ethanol decomposition on palladium is similar.
10.2 Catalysis
Experimentally, ethanol is dissociated on Pd or Pt to give ethoxides that react with the surface to give acetaldehyde [2]. Acetaldehyde may, at higher temperatures, decompose to an acetyl (CH3CO) intermediate, which undergoes dehydrogenation and decarbonylation to form CO and CHx species following CC bond cleavage. These CHx groups then may hydrogenate to methane and a fraction decomposes to adsorbed C and H. Rh is unique in its surface reaction with ethanol, Rh acts to cause the adsorbed ethoxide species to form an oxametallacycle intermediate that is more stable than adsorbed acetaldehyde. Rh metal is known to activate the carbon– hydrogen sp3 bonds [19, 20] and it is observed by HREELS and calculated by DFT that over a flat Rh(111) single crystal surface the adsorbed species is tilted toward the surface [21–23]. This is postulated to be due to favorable orbital–orbital interaction between the carbon atoms of the adsorbate and the surface permitting further reactions such as dissociation. 10.2.2 Structure and Properties of Cerium Dioxide
Cerium dioxide (CeO2, ceria) has become the most significant of the rare earth oxides used in catalysis. Cerium metal has an [Xe] 4f25d06s2 electron configuration and exhibits both þ3 and þ4 oxidation states. Cerium is thermodynamically unstable in the presence of oxygen; the oxides Ce2O3 and CeO2 and intermediate oxides are formed; the stoichiometry of the final oxide is dependent on temperature and oxygen pressure. Ceria promotes noble metal catalytic oxidations and reductions and stores and releases oxygen at the surface [24]. Ceria is outstanding in its catalytic properties. According to its oxygen transport properties, ceria is considered an n-type semiconductor and has important applications in catalytic oxidation. It is also among the most efficient catalysts for CO oxidation. Ceria crystallizes in the fluorite structure (CaF2), a cubic close-packed array with face-centered cubic unit cell (space group Fm3m) of metal atoms in which all the tetrahedral holes are filled by oxygen (Figure 10.3). Each cerium cation is coordinated by eight oxygen anions at each corner of a cube; each oxygen anion is tetrahedrally coordinated to four cations. In addition, the CeO2 structure possesses large vacant octahedral holes which are significant in the movement of ions through the defect structure. Equally important, the removal of O2 ions from the lattice is easily achieved, resulting in the reduction of ceria, creating anion vacant sites (denoted &); the reaction scheme thus occurs as follows [24]: 2Ce4 þ þ O2 ! 2Ce4 þ þ 2e =& þ 0:5O2 ! 2Ce3 þ þ & þ 0:5O2
ð10:14Þ
Oxygen is removed from a tetrahedral site and electrostatic balance is maintained by reduction of two cerium cations from þ4 to þ3. Atomistic static lattice methods first modeled the (110), (310) and (111) surfaces of CeO2. The (110) and (310) surfaces are known as type one surfaces; the (111) surface of ceria is a type two surface. Type one surfaces are charge neutral with stoichiometric proportion of anions and cations in each plane, parallel to the surface. The potential for each plane is exactly zero due to the cancellation of the effects of the positive and
j233
j 10 Heterogeneous Catalysis for Hydrogen Production
234
Figure 10.3 The crystal structure of CeO2. (a) The fluorite unit cell of CeO2 where Ce atoms are forming an FCC structure. (b) Top view of the (111) surface of CeO2. (c) Side view of the (111) surface. The lattice constant is 3.87 Å.
negative charges and therefore there is no dipole moment perpendicular to the surface [26]. In a type two surface, the surface terminates with a single anion (O2) plane and consists of a neutral three-plane repeat unit. Based on the energetic criteria displayed in Table 10.3, the relative stability of the surfaces of ceria, despite considerable differences in absolute numbers depending on the computation method (for more details, see ref. [25]) decreases in the order (111) > (110) > (310) and remains the same before and after relaxation. The redox properties of ceria and the high lability of the lattice oxygen are reliant on structural features and contribute to the catalytic activity of ceria in total oxidation processes. Ceria displays both electronic and ionic conduction. The electrical properties of ceria are strongly dependent on temperature, oxygen partial pressures and the presence of impurities or dopants. These variables affect the charge carrier Table 10.3 Surface energy of the fluorite structure, computed
using a fully ionic model [24] and (in parentheses) using PWDFT–PBE [25]. Energy (J m2) Surface
Unrelaxed
Relaxed
(111) (110) (310)
1.707 (0.55) 3.597 (1.09) 11.577 (6.06)
1.195 (0.45) 1.575 (0.86) 2.475 (1.68)
10.2 Catalysis
concentration, which ultimately when coupled with charge carrier mobility determines the electrical conductivity. At high temperature and low oxygen partial pressures, ceria behaves as an n-type semiconductor and the electrons liberated following reduction are the primary charge carriers. The diffusion of O atoms through the ceria lattice leads to non-stoichiometry. 10.2.3 Noble Metal/Ceria Catalysts
Supporting noble metals on ceria considerably enhances the redox activity. The addition of a few percent of noble metal to a ceria support is reported to promote effectively the reduction of surface oxygen only. XPS studies have indicated that the addition of Pt partially reduces the CeO2 surface, as evidenced by a decrease in the O/Ce ratio from the stoichiometric value 2 to 1.59 [27]. Also noted is that following treatment at high temperature in an oxygen-containing atmosphere, the oxidation of the metal such as Pd or Pt in M/CeO2 is promoted by ceria to an MO complex. The activity of the Pt/CeO2 system is attributed to enhanced oxygen species mobility on to ceria-containing solids. Following investigations of noble metal–ceria interactions by XRD and XAFS techniques using a range of calcination temperatures, an affinity order Rh > Pd > Pt of the noble metal–ceria interaction was established [28]. The ceria was found to provide oxygen to the metal-ceria (MOCe) interface and to stabilize the noble metal. In the case of platinum, calcination at 500 C showed mainly PtO2, but at 800 C a mixture of PtO2 and Pt metal and above 800 C mainly Pt metal. PdO was found to be stable to 800 C, above which Pd metal was observed. Decomposition of Rh oxide to the metal was not established. It was postulated that the formation of MOCe bonds may be the major factor attributed to higher catalytic activities. A metal–support interaction phenomenon occurs in noble metal/ceria catalysts. There is a great deal of variation in the literature regarding the nature of the metal–support interaction in noble metal/ceria catalysts. Discovery of the SMSI (strong metal support interaction) effect in 1978 [29–31] revealed a decrease in the chemisorption capacity of CO and especially H2 when using a reducible oxide support; historically it is a titanium dioxide support, without a decrease in dispersion. Other studies have later shown that this is the case for other supports, including CeO2 [32]. The SMSI effect is characterized by four features: first, it is associated with reducible supports; second; it is induced by high temperature (Tredn 723 K) reduction; third, the onset of the SMSI state causes great disturbances in the chemical properties of the dispersed metal phase, and strong inhibition of the chemisorptive properties and significant changes in the catalytic behavior occur; and fourth, reversibility, hightemperature reoxidation (Treoxn 723 K) followed by a mild reduction treatment may recover the conventional behavior of the supported metal phase. The discovery of the observed decrease in chemisorption capacity was found to be the result of reduction of the support followed by migration of the reduced region on to the metal surface (decoration), causing physical blockage of the surface sites. The SMSI state is typically induced by high-temperature reduction in H2.
j235
j 10 Heterogeneous Catalysis for Hydrogen Production
236
Figure 10.4 Adsorption of ethoxide species on ceria in the presence of a metal such as platinum [33].
10.2.4 Adsorption of Ethanol
Scheme 10.2 is a representation of the mode of adsorption of ethanol on the ceria surface that leads to dissociation of the OH bond. Surface adsorption of ethanol leads to ethoxide formation. Figure 10.4 shows a representation of the modes of ethoxide species on CeO2 and in the presence of a metal cluster
Scheme 10.2 Adsorption of alcohol on a ceria surface.
Adsorption of ethanol on the ceria surface occurs readily and completely, resulting in surface coverage by the ethoxide species. Further reaction with surface oxygen may result in stable acetate species as shown in Figure 10.4. 10.2.5 Adsorption of Water
Surface coverage by water is stabilized by the reduced Ce3 þ ; studies of the redox properties of water indicate no oxidation by water occurred on reduced ceria surface at temperatures below 650 K [34]. In fact, water is found to stabilize the Ce3 þ cations [34]. Water accumulates on the ceria support at 363–423 K and makes hydrogen bonds (Figure 10.5). It has been proposed that the reduced Ce3 þ ceria centers are observed to move from the surface to the bulk of the ceria lattice, as the surface appears to reoxidize [35]. With relevance to hydrogen production, it appears that adsorbed water on a Pt/CeO2 system, suppresses hydrogen oxidation to water while CO oxidation still occurs; this is most likely because the Pt is covered by CO and oxygen spillover to the metallic site is not impeded by the hydrogen-bonded hydroxyl structure. It is proposed that at the
10.2 Catalysis
Figure 10.5 Adsorbed water molecules on CeO2 (111). The various degrees of hydrogen bonding are shown: from the left, two H nods, one H bond, no H bond [34].
perimeter of the Pt crystallites the low-temperature WGSR takes place where surface water reacts with linearly bonded CO molecules producing CO2 and H2. 10.2.6 Adsorption of Carbon Oxides
XPS studies of the interaction of CO and CO2 with model Pt/CeO2 catalysts show that CO adsorption on the noble metal is rapid and extensive; it was also shown that heating to 300 C results in desorption of the Pt-adsorbed CO [36]. No interaction of CO with ceria was observed under the reaction conditions employed (<573 K). Pt/CeO2 exhibits high activity for CO to CO2 conversion under these reaction conditions with 100% conversion by 523 K [36]. These observations serve as proof of CO poisoning of Pt at temperatures below 573 K where the surface can become blocked by complete CO adsorption. Extensive CO adsorption on the Pt metal of Pt/CeO2 catalysts was also observed in other studies [35] where it reacted with spilled over oxygen to form CO2. It has been reported that any CO present on support sites is apparently rapidly oxidized to carbonate species and CO interaction with support OH groups results in the formation of formate and bicarbonate species (Figure 10.6) [35]. Disproportionation of CO is another mechanism of CO2 evolution on Pt/CeO2 at high temperatures; however, the reaction mechanism is not well understood [35]. 2CO ! C þ CO2
DH ¼ 172 kJ mol1
ð10:15Þ
10.2.7 Hydrides
Palladium has the highest affinity to hydrogen of the noble metals. At low temperatures, H dissolves in the Pd particles to form a hydride-like structure that may be oxidized and desorbs or moves to the ceria support. An increase in temperature results in decomposition of the Pd b-hydride to metallic palladium.
j237
j 10 Heterogeneous Catalysis for Hydrogen Production
238
Figure 10.6 Schematic representations of adsorbed formates, carbonates and bicarbonates.
Rh has a high affinity for hydrogen adsorption, although the adsorption does not proceed to the bulk of the metal. Rhodium hydride has low stability. Hydride adsorption sites include the fourfold hollow sites at one monolayer and subsurface sites [38]. Desorption from the fourfold sites occurs at around 300 K, whereas desorption at the subsurface sites occurs at around 120 K. The subsurface adsorption occurs via a hydride of rhodium.
10.3 Catalytic Decomposition of Ethanol 10.3.1 Ethanol on Metal Oxides
Reactions of ethanol on catalytic surfaces have been investigated for more than 30 years. On metal oxides, the reactions of ethanol have been reported over a wide range of oxides including [39] Fe2O3, Fe3O4, CaO, TiO2 and to a lesser extent SiO2. The main product is acetaldehyde, the secondary products being acetone and ethyl acetate. Acetaldehyde formation proceeds via the oxidative dehydrogenation of ethanol following transformation to adsorbed ethoxide and adsorbed proton on the metal oxide surface. Acetaldehyde formation liberates two hydrogen atoms from the ethanol molecule that desorb as either H2O or H2. In a rich O2 environment, H2 is easily oxidized to H2O. Acetone formation represents a further oxidation route of acetaldehyde; on TiO2 and Fe oxides, acetone forms once surface acetate appears. The surface must be capable of donating its oxygen to an adsorbed species (SiO2 shows no acetone formation) and one surface cation must be capable of accommodating two acetate molecules as is exhibited on TiO2 and Fe oxides.
10.3 Catalytic Decomposition of Ethanol
Alumina (Al2O3) and vanadia (V2O5) show high activity for ethanol at 623 K [3]. Small amounts of H2 are produced with large amounts of ethene (C2H4) and/or acetaldehyde (C2H4O). Al2O3 is very active for ethanol but the reaction pathway is dominated by ethanol dehydration to ethene. ZnO shows good potential for steam reforming of ethanol to produce hydrogen [3]. Other studies have reported a high hydrogen yield of the ZnO catalyst and also MgO, SiO2, TiO2, V2O5, La2O3, Sm2O3 and CeO2 [40]. Hydrogen production via steam reforming of ethanol over Ni, Co, Ni/Cu and noble metal supported catalysts shows promise [41]. Ethanol steam reforming has been observed to occur to a large extent over ZnO-, La2O3-, Sm2O3- and CeO2-supported catalysts, where CO-free H2 is produced with selectivity of up to 73.8% for H2 and 24.2% for CO2 [42]. Similar results have been observed in other studies for Cu/ZnO, Rh/Al2O3, Rh/CeO2 and Ni/La2O3–Al2O3 [3]. Surface activity for ethanol has also been reported for MgO and TiO2 [43]. Alumina-supported catalysts show activity at low temperatures for the dehydration of ethanol to ethene, which at higher temperatures is converted to H2, CO and CO2 with CH4 as a minor product; the order of activity of the metals is Rh > Pd > Ni ¼ Pt [44]. With a ceria/zirconia-supported catalyst, the formation of ethene is not observed and the order of activity at higher temperature is Pt Rh > Pd [44]. Combinations of ceria/zirconia-supported metal catalysts with the inclusion of alumina yield ethene [44]. Ceria is an active component of the three-way catalytic converters that make a significant contribution to reducing emission levels by simultaneously converting hydrocarbons, carbon monoxide and nitrogen oxides into non-toxic products. Ceria acts as an oxygen storage component, promotes WGSR activity, acts to promote CO oxidation and is associated with dispersion of the active noble metal phase. Previous studies with CeO2 include methanol oxidation over ceria-containing catalysts. The activity of the reactions of ethanol over reduced and unreduced ceria have been investigated [45]. It has been shown that ethanol reaction on the surface of ceria produces acetaldehyde, acetone and methane in addition to CO2 and CO [45]. The reducing properties of the noble metals when used in conjunction with a suitable support have favorable effects on the hydrogen yield. Noble metals appear as the catalyst of choice for the production of hydrogen from ethanol, when used in conjunction with a ceria support where the combined catalytic activity may be ideal. Studies of the reactions of ethanol on noble metal/ceria catalysts have included Pd [2, 45], Pt [2, 27], Rh [2, 46], Au [2, 47], Pt–Pd [2], Au–Rh [2, 48] and Pt–Rh [2]. It is observed the adsorption of ethanol on Rh/CeO2 [2], Au–Rh/CeO2 [2, 48] and Pt–Rh/ CeO2 [2, 13] leads to scission of the carbon–carbon bond of a fraction of the adsorbed ethanol whereas catalysts without Rh do not. 10.3.2 Ethanol on a Noble Metal/Ceria Surface
From the high lability of the lattice oxygen, we may infer that the surface of ceria is an excellent oxidant. Many studies have been carried out investigating ceria associated with a noble metal for boosting activity in catalytic combustion. As the process is an oxidation reaction (e.g. ethanol to CO2), the oxidation states of both the metal and the support may change. Hence it is worth studying the effect of prior reduction of the
j239
j 10 Heterogeneous Catalysis for Hydrogen Production
240
catalyst on the outcome of ethanol decomposition. Regarding the effects of prior reduction of the ceria surface for a series of various metals, it was observed that Rh/ CeO2 catalyst had the least DO value (where DO is defined as the deviation from the stoichiometric value of 2 as analyzed by their XPS O1s lines). Previous studies [13] have found that prior reduction of Rh/CeO2 catalysts shifts the ethanol main reaction products from CH3CHO formation to CO and CH4. FT-IR studies have shown that Rh/CeO2 can dissociate the CC bond of ethanol at room temperature to form adsorbed CO [46]. Moreover, over Rh(111) adsorbed ethanol does not dehydrogenate to a surface acetaldehyde as is the case with Pt and Pd. Ethoxides are initially formed, then they are converted to an oxametallacycle](a)-OCH2CH2-(a)] intermediate [where (a) means adsorbed]. The adsorbed oxametallacycle then decomposes to CO and CH4 upon further annealing, followed by oxidation of products.
ð10:16Þ
½O
CO þ CH4 ! CO2 þ CO þ H2 þ H2 O
ð10:17Þ
Considerable differences are found in the reaction pathways and products of ethanol over CeO2 and Pd/CeO2 [45]. Ethanol decomposition over Pd/CeO2 (or Pt/CeO2) occurs via the formation of surface-adsorbed ethoxide species followed by dehydrogenation to acetaldehyde and further reaction to other desorbed products. Coupling reactions can occur to produce higher hydrocarbons. Studies of ethanol adsorption on unreduced Pt/CeO2 by temperature-programmed desorption gave the following products [27]: unreacted ethanol, acetaldehyde, CO, CO2, methane, benzene, water and some hydrogen. Formation of water occurs via combination of hydrogen atom with surface hydroxyls: HðadsÞ þ OHðsÞ ! H2 O
ð10:18Þ
Ethanol conversion to acetaldehyde occurs via adsorbed ethoxide species: CH3 CH2 OH ! CH3 CH2 OðadsÞ þ HðadsÞ
ð10:19Þ
and subsequent oxidative dehydrogenation of the adsorbed species to acetaldehyde (Scheme 10.3).
Scheme 10.3 Dehydrogenation of ethoxide to adsorbed acetaldehyde.
10.3 Catalytic Decomposition of Ethanol
Methane formation occurs via the breaking of the carbon–carbon bond followed by recombination of the methyl species with hydrogen: CH3 CH ¼ ¼ OðadsÞ ! CH4 þ CO
ð10:20Þ
Formation of acetone also occurs. Two routes may give acetone one on the oxide and the other on the metal shown in (i) reactions 10.21 and (ii) reactions 10.22. (i) Adsorbed acetaldehyde may be oxidized to acetate on CeO2 and two molecules of acetates may combine in the so-called ketonization reaction to give one molecule of acetone and one molecule of CO2. The other mechanism (ii) is via the further reaction of an acetyl species with another methyl group of another acetyl. 2CH3 CHOðadsÞ þ 3ðOÞs ! 2CH3 COOðadsÞ þ H2 O þ VO 2CH3 COOðadsÞ ! CH3 COCH3ðgÞ þ CO2 þ ðOÞs
ð10:21Þ
where VO is a surface oxygen vacancy. ¼ OðadsÞ ! 2CH3 C ¼ ¼ OðadsÞ þ H2 2CH3 CH ¼ CH3 C ¼ ¼ OðadsÞ ! CH3 COCH3ðgÞ þ COðgÞ
ð10:22Þ
b-Aldolization of acetaldehyde on the ceria surface yields crotonaldehyde: 2CH3 CH ¼ ¼ OðadsÞ ! CH3 CH ¼ ¼ CHCHOðadsÞ þ H2 O
ð10:23Þ
b-Aldolization of crotonaldehyde with acetaldehyde on ceria again gives 2,4hexadienal: ¼ CHCHOðadsÞ þ CH3 CH ¼ ¼ OðadsÞ ! CH3 CH ¼ ¼ CHCH CH3 CH ¼ ¼ ¼ CHCHOðadsÞ þ H2 O
ð10:24Þ
CH bond dissociation coupled with intramolecular cyclization and H2O elimination gives benzene, as shown in Scheme 10.4.
Scheme 10.4 Possible formation of benzene from 2,4-hexadienal on a Pt/CeO2 surface.
Previous studies [2, 13] showed that bimetallic catalysts containing Rh are better suited for ethanol decomposition and subsequent hydrogen production. Although rhodium appears as an essential constituent of the ideal catalyst, alone it is not sufficiently active for total ethanol decomposition due to coking [13] and a second metal (Pt, Pd) is needed to enhance either the hydrogenation or oxidation pathways. Hydrogen production from ethanol over Rh–Pt/CeO2 catalysts at a reaction temperature above 600 K and at an oxygen to ethanol ratio of 2 : 1 shows total conversion of ethanol with an H2 yield approaching 25 mol% [13]. Mole percentage yields of
j241
j 10 Heterogeneous Catalysis for Hydrogen Production
242
Table 10.4 Products (mol%) from partial oxidation of ethanol over Rh–Pt/CeO2 at different temperatures [13].
Product
473 K
573 K
673 K
773 K
873 K
973 K
1073 K
CH3CHO CH4 CO CO2 H2 Conversion (%)
45.9 5.1 — 23.8 25.2 10.5
0.06 10.7 6.8 64.6 17.9 92.7
0.45 11.1 10.5 63.1 14.3 96.8
0.02 9.0 7.4 64 19.2 98.2
0.01 6.6 9.0 63.2 21.2 99.2
0.01 8.3 10.6 60.7 20.3 99.2
0.01 10.4 14 61.8 13.8 99.6
products from the decomposition of ethanol by partial oxidation over the bimetallic Rh-Pt/CeO2 catalysts are given in Table 10.4. 10.3.3 Catalytic Oxidation of Ethanol
Steam reforming is an endothermic process that requires a high energy input whereas partial oxidation is exothermic. Both processes proceed via a complex set of reactions, not well understood. In the interests of energy-efficient hydrogen production, a combination of steam reforming and partial oxidation may be appropriate. Energy input for distillation in the purification of ethanol is not required in steam reforming and the energy input for the endothermic steam reforming may be alleviated if we initiate and maintain the decomposition with some exothermic partial oxidation. Thus, beginning with a water–ethanol stream of molar ratio less than 3 : 1 we may find a mechanism that is a combination of steam reforming and partial oxidation. If we consider the process over an Rh-containing CeO2-supported catalyst: CH3 CH2 OH ! CO þ CH4 þ H2
DH ¼ 50 kJ mol1
ð10:25Þ
We need only consider the WGSR: CO þ H2 O ! CO2 þ H2
DH ¼ 41 kJ mol1
ð10:9Þ
and the catalytic oxidation of methane occurring via steam reforming or partial oxidation: CH4 þ 2H2 O ! CO2 þ 4H2
DH ¼ 165 kJ mol1
ð10:11Þ
CH4 þ O2 ! 2H2 þ CO2
DH ¼ 320 kJ mol1
ð10:13Þ
Thus two stoichiometric equations may be deduced that incorporate Rh/CeO2catalyzed decomposition of ethanol and either methane steam reforming or methane partial oxidation: CH3 CH2 OH þ 3H2 O ! 6H2 þ 2CO2
DH ¼ 174 kJ mol1
ð10:13Þ
10.3 Catalytic Decomposition of Ethanol
CH3 CH2 OH þ H2 O þ O2 ! 2CO2 þ 4H2
DH ¼ 311 kJ mol1 ð10:26Þ
However, the oxidation of CO may proceed via a direct oxidation route as opposed to the WGSR. Previous work [49] on Rh/Al2O3 suggests the decomposition of ethanol proceeds by dehydrogenation of ethanol, forming adsorbed acetaldehyde, which subsequently decomposes to CH4 and CO: CH3 CHO ! CH4 þ CO
ð10:27Þ
or undergoes steam reforming: CH3 CHO þ H2 O ! 2CO þ 3H2
ð10:28Þ
Methane then undergoes reformation and CO undergoes the WGSR. Decomposition of acetaldehyde to methane and CO was proposed as the true reaction mechanism and the formation of products was mainly controlled by methane steam reforming and the WGSR. 10.3.4 Catalytic Reforming of Ethanol
Figure 10.7a gives representative data for steady-state ethanol reforming by Rh–Pd/ CeO2 catalysts over a wide range of temperatures. Many studies were conducted with changes to both the metal loading and their ratio. Here we present the results for one of the active catalysts that contain a minimum amount of transition metals (0.5 wt% for each). At each temperature the catalyst was operated until the steady state was reached. There are three regions in the figure labeled I, II and III. In region I, similar amounts of CH4 and CO were produced with a slightly higher contribution from hydrogen. This region is described by reaction 10.25, with possibly some reforming of methane occurring, resulting in higher hydrogen and lower methane yield. Region II is characterized by the onset of the WGSR leading to increased hydrogen and CO2 production at the expense of CO (reaction 10.9). Region III is dominated by reforming of methane in addition to the WGSR; hydrogen increases at the expense of methane (reactions 10.11 and 10.9). The effect on hydrogen selectivity is best presented while keeping the amount of Rh fixed and varying the amount of Pd. Figure 10.7b shows representative data over a catalyst composed of 0.5% Rh and x wt% Pd/CeO2 with x ¼ 0, 0.5, 1 and 2. The data are reported for a high throughput of liquid ethanol–water where nearly maximum conversion is seen. The selectivity to hydrogen steadily increased with increasing Pd content but the activity decreased considerably when Pd was in excess. The most obvious reason is coating of Rh particles by Pd particles when the latter is in excess compared with the former. Figure 10.7b gives direct evidence that the hydrogen– hydrogen recombination reaction becomes more efficient in the presence of
j243
j 10 Heterogeneous Catalysis for Hydrogen Production
244
Figure 10.7 (a) Product distribution in vol.% (or mol%) as a function of reaction temperature over 0.5 wt% Rh–0.5 wt% Pd0 CeO2 during the steam reforming of ethanol with a 1 : 6 molar ratio. The total upper curve is the sum of H2, CH4, CO and CO2. The deviation from 100% of the total observed products in region I is mainly due to the production of acetaldehyde at low temperature. The three regions are explained in the text. (b) Conversion and selectivity changes with increasing Pd content of Rh/CeO2 at 773 K
during ethanol–water reaction with a molar ratio of 1 : 6. It is noted that the selectivity increases with increasing Pd loading. The selectivity is based on the outlet volume that is composed of H2 and CO2. Because of the reaction stoichiometry, the maximum volume of hydrogen would be 75%. The outlet flow rate is typically 200 mL min1 (g cat.)1. The flow rate indicated is that of the premixed ethanol–water solution.
increasing amounts of Pd. The activity of the 0.5 wt% Rh–0.5 wt% Pd/CeO2 appears higher than that with more Pd and may be linked to even smaller particles dispersed on the CeO2 surface.
10.4 Conclusions
Heterogeneous catalytic production of hydrogen from ethanol by steam reforming involves a complex set of reactions including dehydrogenation, water gas shift and oxidation. Among the most stable and active catalysts are those based on transition metal systems deposited on nanoparticles of cerium oxide. For efficient conversion and selectivity, two types of metals are needed: Rh for breaking the carbon–carbon bond and Pd or Pt for fast hydrogen–hydrogen recombination and for further reforming of methane.
References
References 1 Energy Information Administration, (1998). 2 Idriss, H. (2004) Platinum Metals Review, 48, 105–115. 3 Haryanto, A., Fernando, S., Murali, N. and Adhikari, S. (2005) Energy and Fuels, 19, 2098–2106. 4 Sorensen, B. (2005) Hydrogen and Fuel cells: Emerging Technologies and Applications, Elsevier Academic Press, Amsterdam. 5 Mattos, L.V. and Noronha, F.B. (2005) Journal of Catalysis, 233, 453–463. 6 Brown, L.F. (2001) International Journal of Hydrogen Energy, 26, 381–397. 7 Tsiakaras, P. and Demin, A. (2001) Journal of Power Sources, 102, 210–217. 8 Bunluesin, T., Gorte, R.J. and Graham, G.W. (1998) Applied Catalysis B, 15, 107–114. 9 Pozdnyakova, O., Teschner, D., Wootsch, A., Krohnert, J., Steinhauer, B., Sauer, H., Toth, L., Jentoft, F.C., Knop-Geriche, A., Paal, Z. and Schlogl, R. (2006) Journal of Catalysis, 237, 17–28. 10 Cracium, R., Shereck, B. and Gorte, R.J. (1998) Catalysis Letters, 51, 149–153. 11 Orucu, E., Karakaya, M., Avci, A.K. and Onsan, Z.I. (2005) Journal of Chemical Technology and Biotechnology, 80, 1103–1110. 12 Garcia, E.Y. and Laborde, M.A. (1991) International Journal of Hydrogen Energy, 16, 307–312. 13 Sheng, P.-Y., Yee, A., Bowmaker, G.A. and Idriss, H. (2002) Journal of Catalysis, 208, 393–403. 14 Emsley, J. (2001) Natures Building Blocks: an A–Z Guide to the Elements, Oxford University Press, Oxford. 15 Birot, P. and Primet, M. (1991) Applied Catalysis, 68, 301–314. 16 Trimm, D.L. (1983) Applied Catalysis, 7, 249–282. 17 Freund, H. J. (2008) Topics in catalysis, 48, 137–144. 18 Vesselli, E., Coslovich, G., Comelli, G. and Rosei, R. (2005) Journal of Physics: Condensed Matter, 17, 6139–6148.
19 Atherton, M.J., Fawcett, J., Holloway, J.H., Hope, E.G., Martin, S.M., Russell, D.R. and Saunders, G.C. (1998) Journal of Organometallic Chemistry, 555, 67–79. 20 Atherton, M.J., Fawcett, J., Holloway, J.H., Hope, E.G., Karakar, A., Russell, D.R. and Saunders, G.C. (1996) Dalton Transactions, 3215. 21 Brown, N.F. and Barteau, M.A. (1995) Langmuir, 11, 1184–1189. 22 Jones, G.S., Mavrikakis, M., Barteau, M.A. and Vohs, J.M. (1998) Journal of the American Chemical Society, 120, 3196–3204. 23 Mavrikakis, M., Doren, D.J. and Barteau, M.A. (1998) Journal of Physical Chemistry B, 102, 394–399. 24 Trovarelli, A. (2002) Catalysis by Ceria and Related Materials, Imperial College Press, London. 25 Jianga, Y., Adams, J.B. and van Schilfgaarde, M. (2005) Journal of Chemical Physics, 123, 064701. 26 Tasker, P.W. (1979) Surface Science, 78, 315–324. 27 Yee, A., Morrison, S.J. and Idriss, H. (2000) Journal of Catalysis, 191, 30–45. 28 Hosokawa, S., Taniguchi, M., Utani, K., Kanai, H. and Imamura, S. (2005) Applied Catalysis A, 289, 115–120. 29 Tauster, S.J., Fung, S.C. and Garten, R.L. (1978) Journal of the American Chemical Society, 100, 170. 30 Baker, R.T.K., Prestridge, E.B. and Garten, R.L. (1979) Journal of Catalysis, 56, 390. 31 Baker, R.T.K., Prestridge, E.B. and Garten, R.L. (1979) Journal of Catalysis, 59, 293. 32 Bernal, S., Calvino, J.J., Cauqui, M.A., Gatica, J.M., Larese, C., Omil, J.A.P. and Pintado, J.M. (1999) Catalysis Today, 50, 175–206. 33 Idriss, H. and Barteau, M.A. (2000) Advances in Catalysis, 45, 261–331. 34 Henderson, M.A., Perkins, C.L., Engelhard, M.H., Thevuthasan, S. and Peden, C.H.F. (2003) Surface Science, 526, 1–18.
j245
j 10 Heterogeneous Catalysis for Hydrogen Production
246
35 Pozdnyakova, O., Teschner, D., Wootsch, A., Krohnert, J., Steinhauer, B., Sauer, H., Toth, L., Jentoft, F.C., Knop-Geriche, A., Paal, Z. and Schlogl, R. (2006) Journal of Catalysis, 237, 1–16. 36 Zarraga-Colina, J. and Nix, R.M. (2006) Surface Science, 600, 3058–3071. 37 Bunnik, B.S. and Kramer, G.J. (2006) Journal of Catalysis, 242, 309–318. 38 Pauer, G., Eichler, A., Sock, M., Ramsey, M.G., Netzer, F. and Winkler, A. (2003) Journal of Chemical Physics, 119, 5253. 39 Idriss, H. and Seebauer, E.G. (2000) Journal of Molecular Catalysis A, 152, 201–212. 40 Llorca, J., Piscina, P.R.d.l., Sales, J. and Homs, N. (2001) Chemical Communications, 641–642. 41 Vaidya, P.D. and Rodrigues, A.E. (2006) Chemical Engineering Journal, 117, 39–49.
42 Llorca, J., Homs, N., Sales, J. and Piscina, P.R.d.l., (2002) Journal of Catalysis, 209, 306–317. 43 Liguras, D.K., Kondarides, D.I. and Verykios, X.E. (2003) Applied Catalysis B, 43, 345–354. 44 Breen, J.P., Burch, R. and Coleman, H.M. (2002) Applied Catalysis B, 39, 65–74. 45 Yee, A., Morrison, S.J. and Idriss, H. (1999) Journal of Catalysis, 186, 279–295. 46 Yee, A., Morrison, S.J. and Idriss, H. (2000) Catalysis Today, 63, 327–335. 47 Sheng, P.-Y., Bowmaker, G.A. and Idriss, H. (2004) Applied Catalysis A, 261, 171–181. 48 Sheng, P.Y. and Idriss, H. (2004) Journal of Vacuum Science Technology A, 22, 1652–1658. 49 Cavallaro, S., Chiodo, V., Freni, S., Mondello, N. and Frusteri, F. (2003) Applied Catalysis A, 249, 119.
j247
11 High-Throughput Screening of Catalyst Libraries for Emissions Control Stephen Cypes, Joel Cizeron, Alfred Hagemeyer, and Anthony Volpe
11.1 Introduction 11.1.1 Introduction to High-Throughput Heterogeneous Catalysis
The use of high-throughput synthesis and screening methodologies to accelerate research and development in heterogeneous catalysis has become well established in many industrial and academic laboratories around the world. These methodologies are providing significant advantages to their practitioners, including a reduction in time-to-market for new and optimized catalysts, increased probabilities of success due to the ability to perform very large numbers of experiments, better intellectual property protection, execution of more and shorter projects per unit time and increased organizational efficiency due to improved data storage, access, analysis and sharing. Traditional methods for catalyst discovery and optimization are not efficient because they are primarily empirical processes. The ability to predict required catalyst compositions, structures and formulations for specific chemical processes is low and, for complex multicomponent catalysts, almost non-existent. In addition, there are many variables that affect a catalyst beyond the elemental composition, such as element precursor types, wet synthesis variables, post-treatment methods such as calcination conditions, type and shape of the binder and/or support and process operating conditions such as temperature, pressure, space velocity and reactant composition. These variables are too numerous to explore comprehensively using conventional preparation and testing methods. The high-throughput experimental process in heterogeneous catalysis involves the software-assisted design of diverse, high-density arrays of potential catalytic materials (referred to as libraries) and ultra-fast synthesis, characterization and screening techniques characterized by the use of computer-controlled robotics. The integrated synthesis and screening of a plurality of catalysts in library format has been recognized as an essential factor. Equipment miniaturization and integrated data
Handbook of Green Chemistry, Volume 2: Heterogeneous Catalysis. Edited by Robert H. Crabtree Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32497-2
j 11 High-Throughput Screening of Catalyst Libraries for Emissions Control
248
management systems are also key aspects of successful workflows. The development and implementation of these methods require the application of unconventional engineering and software expertise not commonly available at chemical and refining companies or catalyst manufacturers. High-throughput research is optimally performed by interdisciplinary teams made up of chemists, engineers and programmers. Work beginning in the mid-1990s, led by Symyx Technologies, has significantly advanced the field of high-throughput experimentation and initiated many of the efforts that are currently in place or under way [1–22]. The number of experiments that can now be performed using state-of-the-art high-throughput workflows can be an order of magnitude or more higher than was possible using conventional research. For example, a high-throughput program can yield 50 000 experiments per year compared with 500–1000 experiments using traditional methods. This chapter describes state-of-the-art synthesis and screening techniques for highthroughput experimentation in heterogeneous catalysis applied to emissions control. 11.1.2 The Hierarchical Workflow in Heterogeneous Catalysis
The research and scale-up phases leading to commercialization are shown in Figure 11.1. The high-throughput workflow can be divided into primary, secondary and, in some cases, tertiary screening (and corresponding synthesis). Primary screening, which is most often focused on discovery, is typically a very highthroughput qualitative or semiquantitative screen performed on small samples often using unconventional reactor designs. The objective during this phase is broadly to prepare and screen a large and diverse set of material types that may perform the desired catalytic reaction. Discoveries, termed hits, are then taken to the secondary screening stage and compositional space that is not useful is discarded. Although it is critical that primary screening results correlate with the real catalytic process, this can
Figure 11.1 Stages in catalyst discovery, scale-up and commercialization.
11.1 Introduction
usually be accomplished by screening for qualitative or semiquantitative trends and using relative performance rankings generated from a simplified analog of the real process parameter. The primary screen must be designed to minimize false positives and especially false negatives. Secondary screening is used for the confirmation and optimization of primary hits or for existing catalyst optimization. In contrast to the primary screen, the catalyst form, reactor and process analytics are designed to be similar to those of the real bench-scale material and reaction. The data quality and precision should be equivalent to those of a standard laboratory reactor or even a pilot unit, since the goal is to observe small improvements in performance as a function of catalyst modification. Leads from the secondary stage are taken to the tertiary screens to generate commercial development candidates. If necessary, tertiary screening can be performed using conventional fixed-bed microreactors or pilot units with full reactant and product detection and full mass balance, and in some cases is also parallelized. Note that secondary synthesis and screening technology has, in many cases, now evolved to the point where the quality of the data obtained is equivalent to that obtained using conventional laboratory or pilot technologies. A generalized set of high-throughput workflows used in heterogeneous catalysis is depicted in Figure 11.2. The cycle of library design, synthesis, screening and data analysis is illustrated. Software tools and databasing of synthesis and performance data are key factors [2]. 11.1.3 Applications to Green Chemistry
Catalysis plays a very important role in sustainable and green chemistry applications such as renewable resources for energy and fuels, waste recovery and recycling and low- or even zero-emission power plants, chemical production sites and vehicles.
Figure 11.2 High-throughput catalyst discovery workflow.
j249
j 11 High-Throughput Screening of Catalyst Libraries for Emissions Control
250
Emissions control and environment protection are new challenges that call for novel and far more efficient catalysts to be developed in short time periods. Here, we discuss examples of the application of high-throughput methods to heterogeneously catalyzed green chemistries. Combinatorial catalysis is ideally suited for the discovery of novel noble metal and mixed metal oxide catalyst formulations for total combustion/volatile organic compound (VOC) removal, emissions control from stationary and mobile sources (NOx abatement, CO oxidation, automotive three-way catalysis) because simple (gas-phase) feeds and product mixes allows truly parallel detection and very high sample throughputs via primary screening. Furthermore, combinatorial methods are advantageously applicable to multicomponent catalysts (mixed metal oxides, alloys) for selective oxidation, hydrogenation, dehydrogenation and refining, including desulfurization and denitrification. Additionally, improved fast analytical techniques are making it increasingly possible to perform high-throughput experimentation on complex real plant or vehicle feeds, especially in secondary screening. In this chapter, we focus on the primary screening of metal oxides for lowtemperature CO oxidation and VOC combustion using infrared thermography reactors and SCR DeNOx of wafer-formatted catalyst libraries using scanning mass spectrometry. Other published examples of gas-phase combinatorial screening applied to emissions control include combustion catalysis with better low-temperature activity (Maier) in exhaust gas and air cleaning, automotive emissions, stoves and explosion prevention sensors, the water gas shift reaction aiming at high activity nonpyrophoric catalysts for fuel processors in future fuel cell-driven vehicles (Symyx, Mirodatos), PROX (preferential CO oxidation in excess of hydrogen) as the final clean-up stage of hydrogen streams in fuel processors (Margitfalvi, Mirodatos), CO cleanup by methanation (Maier), diesel tailpipe emissions control (HTE), NSR (NOx storage/reduction) materials (Lauterbach) and DeNOx in wall-coated honeycomb reactors (Claus). [2] and references therein, [18–22]
11.2 Experimental Techniques and Equipment 11.2.1 Overview of Hardware and Methodologies for Combinatorial Heterogeneous Catalysis
Since the mid-1990s, hardware and software engineering efforts both within academia and in industry have resulted in various synthesis and reactor designs for the purpose of primary, secondary and tertiary screening across broad applications within heterogeneous catalysis. Platforms that are designed to allow for tens of thousands of experiments per year now exist that are amenable to catalyst and process optimization in application areas spanning refining, basic chemicals, fine chemicals and specialty applications. For primary screening, where the throughput of catalyst and process conditions may approach hundreds of thousands of experiments per year, catalyst candidates are
11.2 Experimental Techniques and Equipment
typically synthesized by deposition on a substrate to define an array. This deposition may involve evaporation of a solution of metal precursors to yield candidate catalyst films, vapor deposition of metals to define a candidate catalyst alloy or slurry deposition of a solid catalyst with subsequent drying also to leave a film. Arrays can vary in format depending on the reactor choice, but typically are either 256element (16 16) or 64-element (8 8 or 4 16) arrays on a single quartz wafer substrate, with only about 1 mg or less of each catalyst needed (while being amenable to rapid automated characterization such as X-ray diffraction). Due to the large number of catalysts per array, liquid handling robots are typically used to allow for rapid, precise deposition of required catalyst compositions to each well (Figure 11.3). After deposition and drying, all wells are pretreated under identical conditions, typically for the purpose of calcination, reduction or sulfidation. Depending on the pretreatment, this can be done on multiple substrates simultaneously in standard ovens or in tube furnaces. Screening of wafer-based catalyst arrays at the highest throughput as a primary screen has resulted in the need for non-conventional reactor and analytical designs. Symyx has published several platforms that make these semiquantitative screens possible in either rapid serial or truly parallel fashion. For example, a scanning mass spectrometer reactor system (Figure 11.4) allows for the rapid serial evaluation of
Figure 11.3 Liquid handling robots used for rapid, precise deposition of required catalyst compositions to each well on the library wafer.
j251
j 11 High-Throughput Screening of Catalyst Libraries for Emissions Control
252
Figure 11.4 Schematic of the fluidic routing for the scanning mass spectrometer (SMS) system and a photograph of the SMS system.
catalysts on a substrate by moving a sampling/sniffer probe over a single candidate catalyst, heating the catalyst locally using a CO2 laser to a user-defined temperature, flowing reactant gas over the catalyst and sniffing the product gas from the catalyst surface to a quadrupole mass spectrometer to evaluate activity and selectivity of the catalyst. In two other examples, microfluidic-based reactor platforms sandwich the substrate containing all of the candidate catalysts to a flow distribution manifold, such that all catalysts see the same feed and process conditions simultaneously. In one of these systems, an absorbent (e.g. TLC) plate is placed just downstream of the catalyst wafer such that reaction products adsorb on the plate locally, corresponding to the location of each catalyst on the quartz wafer. Selective dyes can then be used to bind to the desired reaction product off-line and an image of the dyed absorbent plate can be taken to determine which catalysts are the most active/selective (Figure 11.5). In another microfluidic system, products from each of the candidate catalysts simultaneously flow to an assembly of IR waveguides (e.g. gas sampling cells) where
11.2 Experimental Techniques and Equipment
Figure 11.5 Schematic of the microfluidic reactor feed system, reactor station and analytics workflow (a). Photograph of the MEMS reactor hardware (b) and an example library array of catalysts and a resulting image of a dyed TLC plate for catalyst activity (c).
an IR camera with a focal plane array sensitive to mid-IR light is directed towards the multiple gas samples and modulated broadband mid-IR light is directed from an FTIR spectrometer to the gas samples and on to the camera such that spectral information for each product gas is simultaneously determined, giving activity and selectivity information for all catalysts simultaneously (Figure 11.6). In a final example of a primary screening reactor, an entire catalyst wafer may be heated to reaction conditions in a pressurized chamber of process gas, the top of the chamber being fitted with an IR-transparent window and an IR camera to measure the apparent temperature of each catalyst simultaneously under the reaction conditions. This technique gives activity information for exothermic or endothermic reactions and has been shown to be ideally suited to screening reactions such as oxidations. Typical operating conditions for the aforementioned primary screens are temperatures up to 500 C and pressures up to 200 psig and the system allows for the use of feeds and products that contain condensable species that are to be kept in the vapor phase. Secondary screening systems require quantitative data corresponding to the quality and differentiating power of traditional laboratory-scale reaction systems so
j253
j 11 High-Throughput Screening of Catalyst Libraries for Emissions Control
254
Figure 11.6 Photograph of a microfluidic screening reactor, with microfluidic channels etched into a quartz plate (a) together with a photograph of the reactor system itself for simultaneous reaction on 64 different catalyst candidates (b). A schematic cut-away of the waveguide assembly for simultaneous analysis of 64 product gases using FTIR imaging is also shown (c).
that the researcher will be confident that the results can readily scale to the pilot plant level. At this point, catalyst synthesis techniques should be comparable to the traditional techniques used in heterogeneous catalysis, such as incipient or excess wetness impregnation, coprecipitation, hydrothermal synthesis and precipitation via evaporation. Since the throughput of catalysts at this screening stage may still be within the range 5000–10 000 per year and typically require 100–1000 mL of catalyst, liquid and solid handling robotics are still necessary and should be designed with features that allow as near a traditional synthesis technique as possible (Figure 11.7). The synthesis of the candidate catalyst formulations, including parallel drying and pretreatment, is just one step to prepare for screening in a miniaturized reactor. At the secondary screening stage, hydrodynamics in the screening reactor that model a large-scale, plug-flow reactor are critical in order to ensure that the inherent catalytic performance of a candidate is not masked by axial dispersion or heat transfer limitations. This is achieved by ensuring that the reactor diameter is at least 10 times the average catalyst particle size while ensuring that the reactor length is between five and 100 times its diameter (depending on the chemistry). Sizing of 5000–10 000 catalysts per year to a range consistent with the secondary screening
11.2 Experimental Techniques and Equipment
Figure 11.6 (Continued ).
Figure 11.7 Images of various high-throughput catalysts synthesis platforms.
j255
j 11 High-Throughput Screening of Catalyst Libraries for Emissions Control
256
Figure 11.8 Workflow for parallel pressing and sizing of catalysts.
reactors requires additional parallel hardware and systems have been designed to pelletize powders isostatically from bulk syntheses (e.g. precipitation, hydrothermal) and size up to 24 samples simultaneously using a sieve assembly (Figure 11.8). In cases where the support is shaped (e.g. pellets, extrudates) and impregnation synthesis is used, these materials may be crushed and sized before or after the impregnation. Secondary screening reactors are designed to allow for the study of catalysts under real processing conditions, with real feeds and at temperatures, pressures and flow rates consistent with the manufacturing scale. In addition, the geometry of the catalyst bed, if a fixed-bed process, is such that axial dispersion is negligible, as described above, so that plug-flow behavior is achieved. Furthermore, to minimize the amount of material needed to be synthesized, reactor volumes are kept small, typically requiring 100–1000 mL of catalyst per experiment, thus equating to reactor diameters of about 2–5 mm and lengths of about 1–20 cm. These dimensions also allow for very efficient heat transfer for exothermic and endothermic reactions, thus allowing for intrinsic kinetic parameters of the candidate catalysts to be obtained under real conditions, something not possible at larger scales where hot- or cold-spots may exist within the catalyst bed. The number of channels or catalysts screened simultaneously within one secondary screening reactor system can vary substantially depending on the throughput needs for the application, but both fixed-bed and fluidized-bed reactors have been typically designed as 6, 8, 16 or 48 channels (with intermediate channel numbers also possible). Common to all of these reactor configurations is the need to split the reactant gas flow evenly amongst all the individual channels such that one feed system (which is typically comprised of expensive mass flow controllers and liquid pumps) can be used for the entire system. This flow splitting is typically achieved by introducing a flow restriction into the system along each channel that is very uniform among all the channels and is far greater than any variable flow restriction that may exist between channels due to, say, variations in catalyst packing density or
11.2 Experimental Techniques and Equipment
Figure 11.9 A microfluidic flow splitter consisting of one inlet (center) and 16 identical, concentric capillary spirals out from the center to 16 individual outlets to achieve uniform flow distribution among all reactor channels from a common feed system.
imperfections in the small fluidic tubing within the reactor. Symyx has published the use of microfluidic flow splitters that achieve this need (Figure 11.9), which are placed just upstream from the catalyst beds. Analysis of the reaction products is typically similar to the techniques used in traditional laboratory-scale reactors (although analysis times are kept as short as possible using rapid analysis techniques) and have most often included rapid serial FTIR, gas chromatography and mass spectrometry. The use of stream selection valves are common among all reactor platforms to direct one of the product gases to the analytical equipment, although for larger channel-count systems multiple identical analytical devices may be integrated into the single system (e.g. six gas chromatographs for a 48-channel system). Also common to all platforms is the use of automated, computer-integrated temperature, pressure, flow and analytical device control such that a user does not need to be present during data collection. As mentioned previously in this chapter, data quantity is substantial from these parallel systems using traditional analytical techniques and therefore automation alongside data capture to a searchable database is critical. For the latest generation of secondary screening reactors, typical operating conditions are temperatures up to 650 C, pressures up to 1500 psig and space velocity from 100 to 250 000 h1. As a final point, the secondary screening reactors have also been designed to keep downtime to a minimum, such that features to allow for rapid changing of catalysts between experiments (such as quick-replace filters and rapid catalyst loading tools) and also robust pressure sealing in a system in which up to 48 reactors must be simultaneously sealed under reaction conditions (by housing, for example, the majority of the seals within a nitrogen-pressurized container, thus keeping the differential
j257
j 11 High-Throughput Screening of Catalyst Libraries for Emissions Control
258
Figure 11.10 Images of a parallel fixed-bed reactor for secondary screening.
pressure across each seal low) are included as features that allow the maximum throughput possible for the hardware (Figure 11.10). Tertiary screening, mentioned previously in this chapter, may be used additionally, although this is applied less often due to the high data quality from the latest generation of secondary screening reactors. This screening level, when applied, typically is used to screen the final pilot-scale catalyst formulation in its final support geometry, such as in an extrudate, bead, ring, etc. Benchtop units to rapidly screen bead or ring geometries (Figure 11.11) up to parallel pilot-plant systems requiring
11.2 Experimental Techniques and Equipment
Figure 11.11 Images of a tertiary screening bench-top reactor to analyze the performance of industrially relevant form factors of catalyst (in the case shown, 7 mm bead catalysts).
only 4 mL of extrudate catalyst (Figure 11.12) have been designed and employed for the rapid characterization of a candidate pilot-scale catalyst formulation. 11.2.2 Experimental High-Throughput Workflow for Low-Temperature CO Oxidation and VOC Combustion
The specific workflow employed in this work is shown in Figure 11.6 and schematically in Figure 11.13. The equipment and procedures used allow for quick
Figure 11.12 Image of a tertiary screening pilot-plant reactor (32 channels total).
j259
j 11 High-Throughput Screening of Catalyst Libraries for Emissions Control
260
Figure 11.13 IR thermography reactor.
identification of active and selective catalyst lead compounds within a large compositional space. 11.2.2.1 Primary Synthesis Methods Wafer-based catalyst libraries were designed using Symyx software. Commercial 4 in circular quartz wafers were cleaned, silanized and bead blasted through steel masks to produce 16 16 arrays of wells for catalyst synthesis. These wells were then optionally precoated with a metal oxide carrier or commercial support layer using a slurry dispensing technique. Catalyst library wafers were defined within Library Studio software as arrays of wells to which a portion of premixed metal precursor solutions was transferred. Once the library design was complete, the software generated instructions used to control liquid dispensing robots that physically prepared the wafer libraries (Figure 11.3). Pre-made external standards (0.5% Pt/ Al2O3) were slurry dispensed into several first/last row and last column wells. The remaining spots in first and last row and last column were left blank to correct for the background signal during data processing. Each well contained 250 mg of catalyst, resulting in a catalyst film approximately 10 mm thick and 3 mm in diameter. The specific carriers and metal precursors used and the pretreatment conditions applied to the particular wafers discussed here are given in the figure captions. Catalytic testing was performed directly on the wafers (see Section 11.2.2.3). 11.2.2.2 Secondary Synthesis Methods Preformed carrier pellets were impregnated with aqueous metal solutions to incipient wetness. Incompatible solutions were sequentially impregnated.
11.2 Experimental Techniques and Equipment
11.2.2.3 IR Thermography Reactor The use of IR thermography has previously been applied to the discovery and optimization of heterogeneous catalysts, specifically for CO oxidation and for the total combustion of hydrocarbons. The system in use at Symyx consists of a reactor chamber, a downward-looking IR imaging camera (with associated optics and readout electronics), a gas feed manifold and a means to heat the catalysts to be screened. It is shown in Figure 11.13. The current reactor and optical system can be used to screen reactions up to 500 C at atmospheric pressure and the fitment of the reactor chamber with an IR-transparent window allows the system to operate at pressures greater than 200 psig. IR detection from the catalyst array is achieved using a liquid nitrogen-cooled 256 256 pixel InSb detector with electronic shutter speeds and electronic readout controlled by CamIRa software (SE-IR Corporation). To screen a catalyst library for CO oxidation activity, a candidate wafer was placed on the wafer heater in the reactor chamber and heated to 105 C with 1.0 standard liter per minute (SLM) clean, dry air (CDA) flow into the reactor chamber. A thermographic image of the candidate wafer was acquired by averaging 20 individual images. The wafer was then heated to 115 C with 1 SLM CDA flow and another 20-frame average image was taken. A linear calibration of temperature versus pixel response was assumed for each pixel in the detector and this two-point calibration acted as the non-uniformity correction to account for varying emissivities of materials on the wafer. The wafer was then cooled to 110 C, the desired reaction temperature, and 2% CO in CDA was fed to the reactor chamber at 1 SLM. The candidate catalysts were allowed to equilibrate in the chamber for approximately 30 min and a 20-frame average image was taken of the catalysts in the presence of CO. The reactor was then purged by feeding 1 SLM of CDA for approximately 30 min while keeping the wafer at 110 C, and a final 20-frame average image was taken. This final image was used to ensure that all hot catalysts returned to the wafer set point temperature upon evacuating CO from the chamber, a sign that the presence of CO caused a true temperature rise on the catalyst rather than an irreversible emissivity change of the material. To illustrate this important issue of emissivity change further, see Figures 11.14 and 11.15. Figure 11.14 shows a discovery wafer (wafer design: carrier Co3O4 impregnated with 3% Ru; linear vertical gradients from 1 to 10% of Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Ge in the upper half and Zr, Nb, Mo, Ag, Sn, Sb, W, Ce, K, Re in the lower half; Ru/Co3O4 was a hit detected in combination with Ni, Ag, Sn or Ce). Figure 11.15 shows a focus wafer [wafer design: different carriers: Co3O4 in columns 1–3, activated carbon Aldrich 22.164-3 in columns 5–7, SiO2 Siliperl AF 125 in columns 9–11, TiO2 NorPro XT 25376 in columns 13–15; 10% Co as Co(NO3)2 uniform, countercurrent gradients of 0–6% Ru as Ru(NO)(NO3)3 and 12–0% Ag ex Ag lactate; activated carbon is the most active carrier]. The temperature change upon exposure to CO (30 min) is shown in Figures 11.14a and 11.15a and the temperature change upon repurging with air (30 min) is shown in Figures 11.14b and 11.15b. As can be seen, the hot-spots return to the original temperature upon repurging, indicating the absence of irreversible emissivity changes, i.e. the temperature changes are entirely caused by the catalytic reaction.
j261
j 11 High-Throughput Screening of Catalyst Libraries for Emissions Control
262
Figure 11.14 16 16 array discovery wafer Ru–M/Co3O4 comprising 30 seven-point vertical gradients together with spotted Pt/Al2O3 standards in the first and last row and also the last column. The Co3O4 carrier was slurried and then impregnated with 3% Ru. Metal gradients from 1 to 10% of Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Ge
in the upper half and Zr, Nb, Mo, Ag, Sn, Sb, W, Ce, K, Re in the lower half of the wafer. Ru/Co3O4 as hit detected in combination with Ni, Ag, Sn or Ce. (a) Temperature change upon exposure to CO (30 min); (b) temperature change upon repurging with air (30 min).
Figure 11.15 16 16 array focus wafer of RuCoAg lead on various carriers comprising 15 14-point vertical gradients: Co3O4 (Aldrich, row 1–3), activated carbon (Aldrich 22.164-3, rows 5–7), SiO2 (Siliperl AF 125, rows 9–11), TiO2 (NorPro XT 25376, rows 13–15). Uniform layer of
10% Co ex Co(NO3)2 uniform and countercurrent gradients of 0–6% Ru ex Ru(NO)NO3 and 12–0% Ag ex Ag lactate. Active carbon is most active carrier. (a) Temperature change upon exposure to CO (30 min); (b) temperature change upon repurging with air (30 min).
11.2 Experimental Techniques and Equipment
11.2.2.4 Multi-Channel Fixed-bed Reactor The secondary screening reactor consisted of eight parallel fixed beds with a flowdistributed feed, in which a stream selection valve selected one of the reactor effluents for rapid serial GC analysis. The feed consisted of 1% CO and optionally 0.15% propylene in CDA at a space velocity of 25 000 h1. The reactor temperature was increased from 40 to 200 C in 10–20 C increments in order to produce light-off curves for CO oxidation as pure component in CDA as well as a mixed CO-propyleneCDA feed. GC analysis used a TCD for quantitative detection of CO, CO2, H2O, propylene and O2. Light-off curves were produced for the CO oxidation and VOC combustion analysis of all eight candidate catalysts in approximately 12 h, using automated Symyx software control. 11.2.3 Experimental High-Throughput Workflow for NOx Abatement
The scanning mass spectrometer (SMS) described earlier was used for NOx abatement catalyst discovery and optimization. The temperature range 200–600 C is accessible. Typically, 256 catalysts can be screened per day and about 100 000 experiments run annually. Details of Symyx SMS have been published previously. The reactor configuration used in this work permitted the introduction of up to four different gases and one liquid feed. 11.2.3.1 Primary Synthesis Methods Wafer-based catalyst libraries were designed using Symyx software and synthesized consistent with the general procedure described in the previous section for waferbased primary screening of CO oxidation catalyst candidates. 11.2.3.2 Primary Screening Methods For screening, the wafers were mounted on an xy stage that registers the catalyst to be tested below the reactor head. The catalyst being tested was heated with a CO2 laser to the desired reaction temperature. Temperatures in the range 180–500 C were obtained using standard quartz wafers, with a thickness of approximately 1300 mm. Temperatures up to 700 C were accessed using thinner wafers of <500 mm. The chamber was maintained at a constant pressure of 30 psi using argon dilution gas and a pressure controller. The reactants were mixed in the SMS head and delivered through two small ports on either side of the catalyst element being analyzed. At the center of the SMS head, a small fraction of the reaction gas was removed through a heated capillary. This capillary delivers a constant sample stream to a Bruker quadrupole mass spectrometer. The complete setup is shown in Figure 11.4. For SCR DeNOx, an NO/NH3 model feed with and without O2 cofeed in the absence of moisture and sulfur has been used. Custom Symyx control software permits independent reactor operation (with one wafer) for periods of over 1 week. Up to 20 quadrupole masses can be monitored for a given experiment. The measurement required 1–3 min per sample, depending on the catalyst stabilization time and the number of masses being analyzed.
j263
j 11 High-Throughput Screening of Catalyst Libraries for Emissions Control
264
11.2.3.3 Data Analysis for NOx Abatement from SMS Premixed gas tanks of reactants with Kr as the internal gas standard were used. The raw data were first normalized to the Kr signal, e.g. Norm CO2 ¼ raw CO2/raw Kr (repeated for all masses). The normalization of each mass to the internal gas standard Kr compensated for systematic errors (e.g. pressure fluctuations, gas flow fluctuations, varying wafer–nozzle gap) and improved the wafer-to-wafer reproducibility. Next, the signals of the (external, commercial) catalyst standards or (internal, synthesized) catalyst standards (in the first row and last column of the wafer) and all blank elements were averaged. The external standard is added to the catalyst library wafers after thermal processing and prior to screening and allows checking of the wafer to wafer reproducibility. The internal standard is a synthesized on the wafer and undergoes all the thermal pretreatments of the wafer to allow uniformity checks across the wafers (it also provides a benchmark for relative comparisons with catalyst library elements). Catalysts were not screened at full conversion/equilibrium conversion in order to discriminate and rank the catalysts better. Catalyst amounts on the wafer were adjusted to achieve the desired conversion. The data analysis is based on mass balance plots: Figure 11.16 shows a generalized plot where reactant conversion is plotted versus product production (e.g. uncalibrated but Kr-normalized MS signals). Since the DeNOx reaction converts one NO molecule into one N2 molecule, selective catalysts would lie on a straight line with a slope of unity, whereas any deviation from the diagonal is indicative of unselective side reactions. The relevance of primary SMS screening and scalability of hits has already been demonstrated for the low-temperature CO oxidation for cold start automotive emission control, SCR (selective catalytic reduction) and SCD (selective catalytic decomposition, direct decomposition) DeNOx and VOC removal [2, 9, 11, 12].
Figure 11.16 Data analysis by mass balance plots as described in the text.
11.3 Low-Temperature CO Oxidation and VOC Combustion
11.3 Low-Temperature CO Oxidation and VOC Combustion
The catalytic oxidation of carbon monoxide to carbon dioxide is a key process for respiratory protection, industrial air purification, automotive emissions control and CO clean-up in flue gases and fuel cells. For automotive emissions control, CO oxidation along with VOC combustion and DeNOx catalytic processes are important for cold-start emissions control using three-way catalytic converters. Extensive research has focused on the improvement of catalytic activity at low temperatures for these three co-current catalytic processes [23–28]. Numerous catalytic systems have been proposed, including those based on Pt, Pd, Rh, Ru, Au, Ag and Cu, supported on refractory or reducible carriers or dispersed in perovskites. Due to the cost of the noble metals, much attention has been given to base metals as catalysts. A large number of papers have been published on the low-temperature oxidation of CO over various transition metal oxides. Our interest is in the metal oxides of ruthenium, cobalt, copper, tin and cerium [29–31]. Other well-known commercial catalyst formulations for room temperature CO oxidation are based on CuMn2O4 (hopcalite) and CuCoAgMnOx mixed oxides. However, these systems are problematic in that they rapidly deactivate in the presence of moisture. For CO oxidation, primary screening experiments were carried out using parallel IR thermography with equipment capable of screening >10 000 catalysts per month in microgram quantities. More than 20 000 catalysts were tested in 4 months for CO oxidation activity via this primary screen. Families of Ru catalysts, in particular RuCo, RuSn and RuCu, were discovered to be very active and were further optimized. When applying IR thermography, one has to make sure that the changes observed on the catalyst sample spot originate from reaction-induced temperature changes but not from emissivity changes of the material. This important point has been illustrated in the experimental section above. The IR thermography reactor was initially used to screen pure and binary metals on supports and bulk oxides, as a function of variables such as precursor salt and calcination/reduction conditions. This included redox metals from group VB–IB and rare earths and dopants/promoters from transition, rare earth and main group metals and supports including numerous commercial silicas, aluminas, titanias, zirconias, cerias, carbons and niobias, in addition to proprietary high-surface area CeO2, Y2O3, SnO2 and Co3O4 [32]. Figures 11.17–11.22 show typical thermographic images of 256-element libraries with CO fed into the reactor chamber. Figures 11.17 and 11.18 show wafer design and thermographic images for RuCo supported on activated carbons and a selection of oxidic carriers. Details of the carriers are given in the captions and in the legends. In Figure 11.17 countercurrent gradients of Co and Ru were used whereas in Figure 11.18 the Ru loading was fixed (a uniform Ru layer across the wafer) with a linear vertical Co gradient. The library design in Figure 11.17 consists of 14-point CoRu binaries supported on five different carriers (carbon and silica/ bentonite) grouped in three columns (with ZrO2 added as binder). Three different
j265
j 11 High-Throughput Screening of Catalyst Libraries for Emissions Control
266
Figure 11.17 (a) Library design for 14-point CoRu binaries supported on five different activated carbon carriers grouped in three columns (with ZrO2 powder added as binder). Three different Co gradients from bottom to top from 0 to 20, 40, 60 mmol corresponding to 0 to 3.4, 6.6, 9.6% and three different countercurrent Ru gradients from top to bottom from 0 to 20, 40, 60 mmol corresponding to 0 to 5.7, 10.8, 15.4%. Dispense volume, 8 ml per well. The wafer was calcined at 250 C for 3 h in air. Columns 1–3: 1 g Darco, KB-B (Aldrich, #27,810-6) þ 1 g ZrO2
FZO 935/8 mL H2O. Columns 4–6: 1 g Darco, G-60 (Aldrich, #24,227-6) þ 1 g ZrO2 FZO 935/8 mL H2O. Columns 7–9: 1 g bentonite, Sample G, K-carrier, (S€ ud-Chemie) þ 1 g ZrO2 FZO 935/8 mL H2O. Columns 10–12: 1 g Norit, RBI (Aldrich #33,128-7) þ 1 g ZrO2 FZO 935/8 mL H2O. Columns 13–15: 1 g SC 40 (Silcarbon Aktivkohle) þ 1 g ZrO2 FZO 935/8 mL H2O. 5% Pt/Al2O3 standards (0.5 g/4 mL H2O, Aldrich #31-132,4) slurried in several first/last row and last column positions. (b) IR thermogram upon exposure to CO.
linear vertical Co gradients from bottom to top from 0 to 3.4, 6.6 and 9.6% and three different countercurrent Ru gradients from 0 to 5.7, 10.8 and 15.4%, respectively, were used. Pt/Al2O3 standards were slurried in several first/last row and last column positions. The library design in Figure 11.18 consists of 14-point RuCo binaries on 15 different carriers (metal oxides, clays, Y-zeolite). A linear vertical Co gradient from 0 to 14% and a uniform Ru loading of 6% were used. Co nitrate and Ru nitrosyl nitrate were the precursors. Active carbon is by far the most active carrier (the Pt/Al2O3 standards in the last row, cf. Figure 11.17, show low activity in comparison). High surface area zeolites, silica and SnO2 and also CeO2 are also very active. Figure 11.19 shows the ternaries RuCo–(no element or Sn or Ce or Zr) supported on Aerolyst 350 silica. The library design consists of countercurrent horizontal Co and dopant gradients and a vertical Ru gradient. The wafer is made up of two 7 7 and two 8 7 ternaries. The linear gradients range from 1 to 6% for Ru, 1.4 to 14.1% for Co, 2.85 to 28.5% for Sn, 3.36 to 33.6% for Ce and 2.19 to 21.9% for Zr. The undoped RuCo reference ternary (upper left corner of the wafer) features orthogonal
11.3 Low-Temperature CO Oxidation and VOC Combustion
Figure 11.18 (a) Library design for 14-point RuCo binaries on 15 different carriers. Uniform Ru layer (6% Ru constant) and vertical Co gradient (0–4%). The wafer was calcined at 300 C for 2 h in air. The carriers are as follows. Column 1: Sepiogel-A (IMV Nevada): 8 mL dispense volume per well. Column 2: SiO2 bentonite, Sample G, K-carrier, (S€ ud-Chemie), BET: 266 m2 g1: 8 mL dispense volume. Column 3: Sepiogel-A (IMV Nevada): 8 mL dispense volume. Column 4: CeO2 (proprietary recipe 1): 2.67 mL dispense volume. Column 5: Ta2O5, (H. C. Starck): 4 mL dispense volume. Column 6: CeO2 (proprietary recipe 2), BET: 171 m2 g1: 4 mL dispense volume. Column 7: Nb2O5 (H.C. Starck): 4 mL dispense volume. Column 8: CeO2 (proprietary recipe 3), BET: 146 m2 g1: 4 mL dispense volume. Column 9: SiO2 Aerolyst 350 (Degussa), BET: 175 m2 g1: 4 mL dispense volume. Column 10: ZrO2 XZ16154 (NorPro), BET: 269 m2 g1: 4 mL dispense volume. Column 11: SnO2 (by precipitation using hydrazine
method ex SnCl4, calcination at 300 C), BET: 250 m2 g1: 8 mL dispense volume. Column 12: ZrO2 XZ16122 (NorPro), BET: 135 m2 g1: 4 mL dispense volume. Column 13: silica gel, Davisil, grade 635 (Aldrich): 4 mL dispense volume. Column 14: SnO2 (by precipitation using hydrazine method ex SnCl4, calcination at 400 C), BET: 228 m2 g1: 8 mL dispense volume. Column 15: zeolite, type Y, molecular sieve, NH4 þ ion (Alfa 41901): 4 mL dispense volume. The following slurries were used for carrier dispensing (with a-alumina added as binder). Columns 1–3, 11, 14: 0.125 g mL1 H2O and 0.25 g a-Al2O3 g1 carrier. Columns 5–10, 12, 13, 15: 0.25 g mL1 H2O and 0.25 g a-Al2O3 g1 carrier. Column 4: 0.375 g mL1 mixed solvent (30% ethylene glycol, 30% H2O, 40% MeOH). 5% Pt/Al2O3 standards (0.5 g/4 mL H2O, Aldrich #31-132,4) slurried in several first/last row and last column positions. (b) IR thermogram upon exposure to CO.
Ru and Co gradients, i.e. the Ru/Co ratio and the total metal loading are varied at the same time. Zr and Sn are identified as the most active. The highest activity is found in the Corich areas of the wafer with a total temperature rise higher than the Pt on alumina standards. Figure 11.20 shows Ru doped with 11 metals (29 different metal precursors including undoped Ru for comparison) on SiO2 bentonite carrier (supplied by
j267
j 11 High-Throughput Screening of Catalyst Libraries for Emissions Control
268
Figure 11.19 (a) Library design for four RuCo–(no element, Sn, Ce, Zr) 7 7 and 8 7 ternaries on SiO2 Aerolyst 350 carrier (Degussa), BET ¼ 175 m2 g1, slurry ¼ 1 g/4 mL H2O þ 0.25 g a-Al2O3, 4 mL dispense volume per well. Linear gradients: Ru 1 to 6%, Co 1.4 to 14.1%, Sn, Ce and Zr 6 to 6 mmol (Sn 2.85 to
28.5%; Ce 3.36 to 33.6%; Zr 2.19 to 21.9%). The wafer was calcined at 300 C for 2 h in air. 5% Pt/ Al2O3 standards (0.5 g/4 mL H2O, Aldrich #31132,4) slurried in several first/last row and last column positions. (b) IR thermogram upon exposure to CO.
S€ ud-Chemie). The design consists of a constant Ru layer (6% Ru) and metal gradients from 6 to 60 mmol, calcined at 250 C for 3 h). The wafer contains 30 seven-point and eight-point vertical metal gradients, i.e. 30 Ru–M binaries. The metal precursors used are given in the caption. RuSn using SnO2 colloidal dispersion and also Sn oxalate as Sn precursors are seen to be very active and compare favorably with the Pt/Al2O3 standard. Figure 11.21 shows Ru systems doped with 11 metals (29 different metal precursors in addition to the undoped Ru system) on g-alumina SBa 150 carrier. The Ru loading is constant across the wafer. Seven-point vertical metal gradients are used for the dopants. The wafer has been reduced at 400 C for 2 h in 5%H2/N2 and then passivated with 0.5%O2/N2 at room temperature for 15 min. RuCu in first lower column using Cu malicate as Cu precursor is seen to be very active in comparison to nitrate and Cu formate precursors and compares favorably with the Pt/Al2O3 standard.
11.3 Low-Temperature CO Oxidation and VOC Combustion
Figure 11.20 (a) Library design for 30 sevenpoint and eight-point Ru–11 metal binaries on SiO2 bentonite, sample G, K-carrier (S€ udChemie). Carrier slurry ¼ 1 g SiO2 bentonite þ 0.25 g a-Al2O3 in 8 mL H2O, 8 mL dispense volume per well. Uniform Ru layer (6% constant, 20 mmol) and metal gradients from top to bottom from 6 to 60 mmol for 11 metals (various metal precursors for individual
metals). Maximum metal content: Ti 8.6%, Zr 16.4%, Co 10.6%, Cu 11.4%, Ni 10.6%, Fe 10%, Ce 25.2%, Sn 21.4%, Nb:16.7%, Sb 21.9%, Mn 9.9%. The wafer was calcined at 250 C for 3 h in air. 5% Pt/Al2O3 standards (0.5 g/4 mL H2O, Aldrich #31-132,4) slurried in several first/last row and last column positions. (b) IR thermogram upon exposure to CO.
j269
j 11 High-Throughput Screening of Catalyst Libraries for Emissions Control
270
Figure 11.21 (a) Library design for 30 sevenpoint and eight-point Ru–11metal binaries on g-Al2O3, SBa-150 Puralox (Condea), BET ¼ 150 m2 g1, carrier slurry ¼ 1 g/4 mL mixed solvent (50% ethylene glycol, 50% H2O), 8 mL dispense volume per well. Uniform Ru layer (6% constant, 20 mmol) and metal gradients from top to bottom from 6 to 60 mmol for 11 metals (various metal precursors for individual
metals). Maximum metal content: Ti 8.6%, Zr 16.4%, Co 10.6%, Cu 11.4%, Ni 10.6%, Fe 10%, Ce 25.2%, Sn 21.4%, Nb 16.7%, Sb 21.9%, Mn 9.9%. The wafer was reduced at 400 C for 2 h in a continuous flow of 5% H2–N2. 5% Pt/Al2O3 standards (0.5 g/4 mL H2O, Aldrich #31-132,4) slurried in several first/last row and last column positions. (b) IR thermogram upon exposure to CO.
11.3 Low-Temperature CO Oxidation and VOC Combustion
Figure 11.22 (a) Library design for 30 sevenpoint and eight-point Ru–29 metal binaries on Co3O4 carrier. Top to bottom Ru gradient (2–10 mmol; 0.6–3%) and countercurrent bottom to top metal gradients (4–20 mmol). 29 metals as given in the legend and undoped
reference in cells I15–O15. The wafer was calcined at 500 C for 2 h in air. 5% Pt/Al2O3 standards (0.5 g/4 mL H2O, Aldrich #31-132,4) slurried in several first/last row and last column positions. (b) IR thermogram upon exposure to CO.
Figure 11.22 shows countercurrent gradients of Ru and 29 different metal precursors on Co3O4 carrier calcined at 500 C for 2 h in air. For instance, columns 10 and 11 in the upper half of the wafer are composed of Co citrate on Co3O4 carrier and Co nitrate on Co3O4 carrier, respectively. The highest activity is found in upper columns 10–14, i.e. Co, Ni and Cu. Also active are Fe, Sn, Sb and W.
j271
j 11 High-Throughput Screening of Catalyst Libraries for Emissions Control
272
Figure 11.23 Light-off curves (CO conversion versus reaction temperature) for RuCo leads in eight-channel fixed-bed reactor. RuCo/ceria and active carbon are more active than the Pt/Al2O3 benchmark whereas RuCo/SnO2 lights off at comparable temperature.
Additional secondary screening of catalyst hits was performed in an eightchannel parallel fixed-bed reactor with GC analytics. Figure 11.23 shows the resulting light-off curves for CO oxidation for several potential catalysts using a CO-CDA feed. It can be seen from the plots that RuCo on ceria and carbon outperform the Pt/Al2O3 standard. RuCo on SnO2 is at least comparable to the benchmark. In conclusion, state-of-the-art catalyst performance could quickly be improved by screening higher dimensional compositional spaces. Single-element Ru, Ce and Co (bulk and supported) compositions, some binaries such as CoMn and CuCe and a few ternaries such as CoCuCe have been reported previously for this chemistry. High-throughput screening, however, has allowed us quickly to find novel, synergistic and better performing RuCo/C and RuCoCe ternaries. Compared with hopcalite CuMn catalysts, RuCo systems are more water resistant and are also suitable for hydrocarbon combustion, i.e. for combined CO oxidation and VOC removal. We have also developed high surface area ceria carriers that are expected to boost further the activity of these systems [32]. In addition, when further doping with Y as a stabilizer, a quaternary RuCoCeY formulation was developed and scaled up to several grams.
11.4 NOx Abatement
Another family of active hits that compare favorably with the Pt/Al2O3 benchmark is based on RuSn, where Sn can be applied as a dopant (e.g. RuSn/SiO2) and/or as a high surface area carrier [32] (e.g. SnO2 or Sn-containing mixed metal oxides). Also, RuCu binary compositions were found to be active after a reduction pretreatment with hydrogen. whereas oxidic copper (after a high-temperature calcination pretreatment) was found to be inactive.
11.4 NOx Abatement
Nitric oxide and nitrous oxide are combustion byproducts which need to be removed from exhaust streams. For stationary power sources, ammonia-selective catalytic reduction(NH3-SCR) is used for NOx abatement. This chemistry has been extensively studied and reviewed [33–35]. 4NO þ 4NH3 þ O2 ! 4N2 þ 6H2 O or NO þ NO2 þ 2NH3 ! 2N2 þ 3H2 O
Figure 11.24 NH3-SCR catalysts NO removal versus N2 formation at (a) 190 and (b) 300 C. Plot (a) presents the activity of 30 metal binary combinations with eight loading levels (color by metal binary). Plot (b) presents the activity of 15 metal–support combinations at 15 metal
loadings each (color by metal–support combination). The inset in (b) shows the library design used with 15 points vertical gradient of metal loading for three metals impregnated on five carriers.
j273
j 11 High-Throughput Screening of Catalyst Libraries for Emissions Control
274
The commercial honeycomb catalyst for NH3-SCR stack DeNOx used in the clean-up of exhaust gas streams in stationary power plants at low to medium temperatures basically consists of V2O5/WO3/TiO2 and allows up to 90% NOx removal to be achieved with less than about 5 ppm ammonia slip. Although this technology is well suited for fixed sources, it requires large catalyst amounts and significant gains in activity would be advantageous. A broad screening program encompassing supported redox, noble and base metal catalysts for NH3-SCR DeNOx using high-throughput scanning mass spectrometers was conducted. The goals were to discover novel compositions with drastically enhanced activity or a broader dynamic temperature window, to provide an activity ranking of catalyst compositions and to identify general trends. Desired gains in activity are approximately an order of magnitude for SCR stack DeNOx catalysts operating in the temperature range 200–300 C. A large number of catalyst libraries consisting of binary, ternary and quaternary catalyst compositions were synthesized on quartz wafers and screened. Figure 11.24 shows NO conversion plotted versus N2 production for various metal/ carrier and supported binary combinations at (a) 190 and (b) 300 C, colored by metal and by support. This representation allows the rapid evaluation of selectivity towards
Figure 11.25 Effect of temperature on NH3-SCR performance: (a) N2 formation and (b) N2O formation. Selected metals were impregnated on 30 different supports. The N2 and N2O production for the reference sample V/TiO2 is highlighted for comparison.
11.4 NOx Abatement
N2 formation. All catalysts are selective for SCR with negligible N2O byproduct breakthrough. The NO conversions are high and range from approximately 10 to 60%. Catalysts are not screened at full conversion in order to discriminate and rank the catalysts better. Space velocity is adjusted by selection of the catalyst loading to achieve the desired conversion range. The reference catalyst, V/TiO2 in column 16, would achieve nearly quantitative conversion under plant stack-DeNOx conditions (3000 GHSV, 1 bar). Our data shows that many catalyst formulations outperform the standard at both temperatures. Catalyst activities for the different compositions cover a broad range of conversions and are therefore easily ranked. Figure 11.25 shows (a) the N2 production and (b) the N2O production as a function of reaction temperature, colored by support, for a wafer consisting of 30 different supports (carrier screening). From the N2O production plot, it can be seen that for this set of samples the reference catalyst V/TiO2 is less selective than the other samples at and above 320 C. Figure 11.26 demonstrates the much higher activity of a high-temperature SCR catalyst for temperatures >260 C and the broader operational temperature window for a full-range SCR catalyst, with respect to the V/TiO2 standard. N2O production for the new catalysts is also lower.
Figure 11.26 Selected catalyst performance on TiO2 (sets of duplicate samples) as a function of temperature compared with the V/TiO2 standard.
j275
j 11 High-Throughput Screening of Catalyst Libraries for Emissions Control
276
Figure 11.27 NO conversion for 24 single metals (30 metal precursors) impregnated on g-alumina for 7–8 loadings each. Samples presented by metal precursor (x-axis). Conversion at 250 C.
The NO removal activity of another representative wafer containing 24 metals supported on a g-alumina carrier is shown in Figures 11.27 and 11.28. g-Alumina was identified as active carrier. Each column represents a different metal precursor used and the multiple data points for each column represent different loadings. For clarity the data were split into two graphs. Note that different precursors have been used for some metals (Co, Cu, Fe, Ce, V). Reduced selectivity with increased N2O production was found for Mn > Co > Mo (not presented here). Different metals have clearly different activities; however, only modest differences were found for samples prepared with the same metal from different precursors. For example, Co and Cu carboxylates appear to form more active catalysts than Co and Cu nitrates. For 250 C, the activity ranking is Cu (both precursors) V, Co, Fe, Mn, whereas for 300 C, the ranking is Ru > Co Cu > V Fe > Mn. Ru is inactive at low temperatures but highly active at medium and high temperatures, presumably due to in situ reduction of Ru oxide by NH3 at higher temperatures. NiO is inactive (Ni metal is required), but Co oxide is active. In conclusion, high-throughput screening of catalyst formulations with the scanning mass spectrometer allowed the identification of a number of hits for this chemistry. The most promising samples contain Cu and Ru as main active elements. Addition of second elements and the selection of the support are found to be important in promoting selective NOx reduction by NH3. The most active samples
11.5 Conclusion
Figure 11.28 NO conversion for 24 single metals (30 metal precursors) impregnated on g-alumina for 7–8 loadings each. Samples presented by metal precursor (x-axis). Conversion at 300 C.
identified had activity down to 190 C (low-temperature SCR) or large improvement in activity above 250 C compared with the reference V/TiO2 samples.
11.5 Conclusion 11.5.1 Application of High-Throughput Screening to Emissions Control
In this chapter, we have discussed methodologies and tools for the high-throughput screening of libraries for gas–solid reactions and presented case studies in the area of emissions control. The screening workflows were applied to the discovery of more efficient catalysts for the low-temperature CO oxidation. The workflow used was capable of identifying novel groups of Pt-free RuCo-, RuSn- and RuCu-based CO oxidation catalysts. RuCo supported on active carbon, high surface area CeO2 and bulk RuCoCe mixed oxides have been scaled up and confirmed by secondary screening. In addition, RuSn and RuCu systems were active hits and are proposed for further investigation. These discoveries were accomplished by screening >20 000
j277
j 11 High-Throughput Screening of Catalyst Libraries for Emissions Control
278
potential catalysts via a primary screen utilizing IR thermography. The discoveries made in this study are also suitable for further development towards combined lowtemperature CO oxidation and total VOC combustion [11, 29–31]. As discussed above, evaluation of NH3-SCR catalysts resulted in the identification of a number of catalysts leads superior to the benchmark sample. Secondary screening to assess these materials further is in progress. 11.5.2 Future Trends in Combinatorial Catalysis
Looking forward, it is clear that the field of high-throughput and combinatorial heterogeneous catalysis will continue to advance and be applied at an increasing rate. Areas in which advances are needed and are being pursued include: acceleration of scale-up activities using commercial-sized catalysts (e.g. large commercial-sized pellets with control over active catalyst distribution in the support); performing process optimization in high-throughput reactors; development of new equipment that can handle harsher reaction feeds and conditions such as high temperatures, pressures, corrosive and heavy feeds, etc; improved and faster analytical methods for analyzing reaction products from screening (including analyzing complicated real plant feeds and exhaust gas streams); and for faster/better characterization of catalyst materials.
Acknowledgments
We are pleased to thank all Symyx co-workers who made this work possible. In particular, we wish to thank Karin Yaccato for reactor operation and analytical work, Andreas Lesik and Guido Streukens for synthesizing and screening catalyst libraries, Marco Schlichter and Carola Fink for synthesizing catalyst carriers, Zach Hogan for the design and construction of synthesis hardware, Valery Sokolovskii for helpful discussions and Victor Wong for technical support. We also wish to thank Honda for permission to publish some primary screening catalyst data.
References 1 Xiang, X.-D., Sun, X.-D., Briceno, G., Lou, Y., Wang, K.A., Chang, H., WallaceFreedman, W.G., Chen, S.-W., and Schultz, P.G. (1995) A combinatorial approach to material discovery. Science, 268, 1738–1740. 2 Hagemeyer, A. Strasser, P. and Volpe, A. (eds), (2004) High Throughput Screening in Chemical Catalysis: Technologies, Strategies
and Applications, Wiley-VCH Verlag GmbH, Weinheim. 3 Hagemeyer, A., Jandeleit, B., Liu, Y., Poojary, D.M., Turner, H.W., Volpe, A.F., Jr. and Weinberg, W.H. (2001) Applications of combinatorial methods in catalysis. Applied Catalysis A-General, 221, 23–43. 4 Murphy, V., Volpe, A.F., Jr. and Weinberg, W.H. (2003) High throughput approaches
References
5
6
7
8
9
10
11
12
to catalyst discovery. Current Opinion in Chemical Biology, 7, 427–433. Senkan, S. (2001) Combinatorial heterogeneous catalysis – a new path in an old field. Angewandte Chemie-International Edition, 40, 312–329. Sch€ uth, F., Busch, O., Hoffmann, C., Johann, T., Kiener, C., Demuth, D., Klein, J., Schunk, S., Strehlau, W. and Zech, T. (2002) High-throughput experimentation in oxidation catalysis. Topics in Catalysis, 21, 55–66. Cawse, J.N. (2001) Experimental strategies for combinatorial and high-throughput materials development. Accounts of Chemical Research, 34, 213–221. Bergh, S., Cong, P., Ehnebuske, B., Guan, S., Hagemeyer, A., Liu, Y., Lugmair, C.G., Turner, H.W., Volpe, A.F., Jr. Weinberg, W.H., Woo, L. and Zysk, J. (2003) Combinatorial heterogeneous catalysis: oxidative dehydrogenation of ethane to ethylene, selective oxidation of ethane to acetic acid and selective ammoxidation of propane to acrylonitrile. Topics in Catalysis, 23, 65–79. Guram, A., Hagemeyer, A., Lugmair, C.G., Turner, H.W., Volpe, A.F., Jr. Weinberg, W.H. and Yaccato, K. (2004) Application of high throughput screening to heterogeneous liquid and gas phase oxidation catalysis. Advanced Synthesis and Catalysis, 346, 215–230. Jahnisch, K., Hessel, V., Lowe, H. and Baerns, M. (2004) Chemistry in microstructured reactors. Angewandte Chemie-International Edition, 43, 406–446. Cypes, S., Hagemeyer, A., Hogan, Z., Lesik, A., Streukens, G., Volpe, A.F., Jr. Weinberg, W.H. and Yaccato, K. (2007) High throughput screening of low temperature CO oxidation catalysts using IR thermography. Combinatorial Chemistry & High Throughput Screening, 10, 25–35. Hagemeyer, A., Lesik, A., Streukens, G., Volpe, A.F., Jr. Turner, H.W., Weinberg, W.H., Yaccato, K. and Brooks, C. (2007) Discovery of novel catalytic materials for
13
14
15
16
17
18
19
20 21
22
emissions control using high throughput scanning mass spectrometry. Combinatorial Chemistry & High Throughput Screening, 10, 135–147. Maier, W.F., St€owe, K. and Sieg, S. (2007) Combinatorial and high throughput techniques in materials research. Angewandte Chemie, 119, 6122–6179. Tompos, A., Margitfalvi, J.L., Tfirst, E. and Vegvari, L. (2006) Evaluation of catalyst library optimization algorithms Comparison of the holographic research strategy and the genetic algorithm in virtual experiments. Applied Catalysis AGeneral, 303, 72. Tompos, A., Margitfalvi, J.L., Hegedu´´ s, M., Szegedi, Á., Fierro, J.L.G. and Rojas, S. (2007) Characterization of trimetallic Pt–Pd–Au/CeO2 catalysts combinatorialy designed for methane total oxidation. Combinatorial Chemistry & High Throughput Screening, 10, 71. Tompos, A., Margitfalvi, J.L., Vegvari, L. and Tfirst, E. (2007) Assessment of predictive ability of artificial neural network using the holographic mapping. Combinatorial Chemistry & High Throughput Screening, 10, 121. Tompos, A., Hegedu´´ s, M., Margitfalvi, J.L., Szabó, E. Gy. and Vegvari, L., (2008) Multicomponent Au/MgO catalysts designed for selective oxidation of carbon monoxide Application of a combinatorial approach. Applied Catalysis A-General, 334, 348. Guliants, V.V. (ed.) (2001) Current developments in combinatorial heterogeneous catalysis, Catalysis Today, 67. Perego, C. (ed.) (2003) Proceedings of the EuroCombiCat, Ischia, June 2–5 2002. Catalysis Today, 81. Maier, W.F. (ed.) (2003) Combinatorial catalysis. Applied Catalysis, 254. Margitfalvi, J. (ed.) (2007) Combinatorial catalysis. Combinatorial Chemistry and High Throughout Screening, 10. Mirodatos C. and Maier, W.F., (eds) (2008) Proceedings of the EuroCombiCat, Bari, April 22–25 2007. Catalysis Today, in press.
j279
j 11 High-Throughput Screening of Catalyst Libraries for Emissions Control
280
23 Orschel, M., Klein, J., Schmidt, H.-W. and Maier, W.F. (1999) Angewandte ChemieInternational Edition, 38, 2791. 24 Kirchner, T., Donnerstag, A., K€onig, A. and Eigenberger, G. (1998) Studies in Surface Science and Catalysis, 116, 125. 25 Lenaers, G. (1996) The Science of the Total Environment, 190, 139. 26 Holzwarth, A., Schmidt, H. and Maier, W. (1998) Detection of catalytic activity in combinatorial libraries of heterogeneous catalysts by IR thermography. Angewandte Chemie-International Edition, 37, 2644. 27 Li, W., Gracia, F.J. and Wolf, E.E. (2003) Catalysis Today, 81, 437–447. 28 Kirsten, G. and Maier, W.F. (2004) in High Throughput Screening in Chemical Catalysis: Technologies, Strategies and Applications, (eds A. Hagemeyer, P. Strasser and A. Volpe), Wiley-VCH Verlag GmbH, Weinheim, pp. 175–187. 29 Cypes, S., Hagemeyer, A., Hogan, Z., Lesik, A., Sokolovskii, V., Streukens, G., Volpe, A.F., Jr. and Weinberg, W.H. (13–17 March 2005) High-throughput screening of low-temperature CO oxidation catalysts
30
31
32
33
34
35
in IR thermography reactor. ACS Spring Meeting, San Diego, CA. Cypes, S., Bergh, H.S., Hagemeyer, A., Hogan, Z., Lesik, A., Sokolovskii, V., Streukens, G. and Weinberg, W.H. (2005) High-throughput screening of lowtemperature CO oxidation catalysts. AIChE Spring Meeting, Atlanta, GA, 10–14 April. Streukens, G., Cypes, S., Lesik, A. and Weinberg, W.H. (2005) High-throughput screening of low-temperature CO oxidation catalysts in IR thermography reactor. 4th International Conference on Environmental Catalysis, Heidelberg, 5–8 June. Fink, C., Hagemeyer, A., Schlichter, M. and Volpe, A. (2007) Aqueous synthesis of high surface area metal oxides. EuroCombiCat, Bari, 22–25 April. Farrauto, R.J. and Bartholomew, C.H. (1997) Fundamentals of Industrial Catalytic Processes, Chapman and Hall, London. Busca, G., Lietti, L., Ramis, G. and Berti, F. (1998) Applied Catalysis B-Environmental, 18, 1–36. Forzatti, P. (2001) Applied Catalysis AGeneral, 222, 221–236.
j281
12 Catalytic Conversion of High-Moisture Biomass to Synthetic Natural Gas in Supercritical Water Frederic Vogel
12.1 Introduction 12.1.1 Heterogeneous Catalysis in Hydrothermal Medium at the Origin of Life?
The hydrothermal conditions prevailing in hot cracks in the deep sea (hydrothermal vents) have been proposed as a major factor in the origin of life on Earth [1]. Why? One would expect mild conditions, more compatible with biological systems and the presence of light rather than high pressures and elevated temperatures, to be at the origin of life. Apparently, some uncommon features make hot pressurized water a very interesting environment for chemical reactions. A key step is the formation of simple organic molecules from inorganic reactants available from the early planetary atmosphere, that is, CO2, CH4, H2, H2O, NH3, CO, H2S and N2, and from minerals (silicates, carbonates, phosphates, etc.). Since the formation of planets occurred at very high temperatures, that is, from a plasma, the resulting compounds were formed from the atoms and radicals by recombination at lower temperatures. These conditions did not allow for the formation of organic compounds other than methane. Organic building blocks of biological systems, for example amino acids, carbohydrates, phenols, fatty acids and alcohols, must have been synthesized from the inorganic carbon oxides, methane, hydrogen, water, hydrogen sulfide, ammonia and/or nitrogen, as well as from minerals. Different hypotheses have been proposed for the conditions that may have led to the synthesis of these compounds. One of these hypotheses was proposed by Foustoukos and Seyfried [2] to be a Fischer–Tropsch-like pathway occurring in hydrothermal vents, catalyzed by Fe- and Cr-containing minerals. They were able to synthesize methane, ethane and propane at 390 C and 40 MPa using FeO–Cr2O3 as catalyst. More recently, Tian et al. [3] succeeded in synthesizing phenol from NaHCO3 at 200 C and 1.8 MPa using iron powder as catalyst. Hydrothermal vents, first discovered in 1977, are found in a depth of 2000–3000 m below sea level along the mid-ocean ridge. Water exits from cracks
Handbook of Green Chemistry, Volume 2: Heterogeneous Catalysis. Edited by Robert H. Crabtree Copyright 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 978-3-527-32497-2
j 12 Catalytic Conversion of High-Moisture Biomass to Synthetic Natural Gas in Supercritical Water
282
in the seafloor at near-critical conditions up to 400 C and at the hydrostatic pressure reigning at those depths, that is, 20–30 MPa. This water, rich in minerals that were leached out of the rocks, mixes with the cold (ca 2–4 C) surrounding water. A shock crystallization occurs, as the solubility of the minerals is much lower in the cold water than near the critical point of the fluid. Iron sulfide forms a smoke of black particles, which is why some hydrothermal vents are termed black smokers. Precipitation of carbonates yields a white smoke (white smokers). The warm surroundings of these vents are inhabited by bacteria, tubeworms, clams and eyeless shrimps. From a chemical point of view, it is interesting that higher hydrocarbons can be formed via Fischer–Tropsch synthesis from CO2 and H2 in the presence of a high partial pressure of water that drives the equilibrium to the side of the reactants. Geochemical time scales are not practical for carrying out technical reactions. Catalysis under hydrothermal conditions can be used to accelerate the formation of fuels from biomass similar to their fossil analogs such as oil, coal and natural gas. The time scales of these catalytic transformations range from minutes to seconds, matching the requirements of our fast-paced society. In this chapter, we focus on the uncommon chemical and physical properties of water near its critical point (i.e. 647 K, 22.1 MPa) and discuss the implications and challenges this unusual chemical environment has for the heterogeneously catalyzed synthesis of methane from biomass. 12.1.2 Biomethane – a Green and Sustainable Fuel
Chemical fuels (derived from coal, petroleum, natural gas, peat and biomass, including organic wastes) are used to cover 90% of our energy needs such as heating, cooling, power, lighting, communication and traction. Whereas traditional fossil fuels such as coal, natural gas and petroleum are not renewable within reasonable time scales (i.e. decades), biofuels are enjoying fast-growing popularity because they are thought to help counteract global warming. However, if biofuels are not produced according to sustainability criteria, they may promote global warming rather than alleviate it. Also, the United Nations recently issued reservations against producing biofuels from edible biomass, asking for a moratorium for biofuels production. Hydrogen is considered the cleanest fuel because its use produces neither greenhouse gas emissions nor other local pollutants. Unfortunately, hydrogen is not a primary fuel present as such on our planet, and has to be produced from other sources. Depending on this production path, there will be emissions of greenhouse gases and pollutants. Methane is available from natural sources as natural gas, but its reserves will last for only ca 64 years if we continue to use it at the current rate. Marine and terrestrial methane hydrates might extend this period if viable technologies for tapping these reservoirs can be developed. From the carbonaceous fuels, biomethane and natural gas have the lowest CO2 emissions per MJ of lower heating value (CH4, 55; gasoline, 73; wood, 99; lignite, 106 g CO2 MJ1).
12.1 Introduction
Methane is attractive because of its beneficial combustion properties, the availability of a transportation and distribution infrastructure, its accepted and established use as a heating fuel for residential and industrial applications, its use in CNG cars and its use as a fuel for gas turbines for power generation. Biomethane is renewable. 12.1.3 Energetic Potentials
Since our hunger for chemical fuels is huge (ca 420 EJ of primary energy in 2004), only those alternative fuels with a potential to substitute a significant amount of fossil fuels are worth considering, even if we manage to reduce the energy consumption per capita. This energetic potential is defined by the total amount of feedstock available and the conversion efficiency to the secondary fuel, for example, CH4. Table 12.1 lists some estimates for the annual energetic potentials of residual and waste biomass worldwide, for Europe, Asia and North America in 2050. Focusing solely on residual and waste biomass is a prerequisite for reaching a sustainable energy system and avoiding the competition between food and energy. A category that is not well characterized in terms of available potential is highmoisture waste biomass, for example manure. In Switzerland, liquid and solid manure amounts to ca 16% of the annual energetic potential of biomass. Data for the large regions in the world are difficult to find. A very rough estimate would be to assume the same relative contribution to the total biomass potential as for Switzerland. This estimate has been added in Table 12.1. These numbers reveal that at most ca 30% of the primary energy consumption for fuels could be substituted by waste biomass in 2050, assuming no increase in total fuel consumption from 2004 to 2050. Taking into account the lower average conversion efficiency from biomass to the actual fuels (methane, Fischer–Tropsch diesel, methanol, ethanol), an upper limit of ca 15% seems more realistic. Feedstocks with a water content of 90% may be blended with residues of lower moisture content, for example wood residues, to yield a pumpable feed with a dry matter content suitable for hydrothermal processing (15–30%). In principle, municipal solid waste, which has an average water content of 30–40% in European countries, may also be used as feedstock for blending after shredding and macerating [6]. Table 12.1 Energetic potentials for agricultural and forestry residues and wastes in 2050, in EJ yr1 [4].
Biomass
World
Europe
Asia
North America
Crop harvest residue Crop process residue Wood residues and waste Manure (own estimate) Total
49–69 (12.5) 16 30 (13.7) 18–22 (5.1) 113–137 (31.2)
6–8 (1.3) 1 8 (2) 3 (0.5) 18–20 (3.8)
9–12 9 7 5 30–33
5–10 (1.7) 1 10 (3.8) 3–4 (0.4) 19–25 (5.9)
Data in parentheses are projections from 1993 cited in [5].
j283
j 12 Catalytic Conversion of High-Moisture Biomass to Synthetic Natural Gas in Supercritical Water
284
Table 12.2 Characteristic data for selected biomass types and their theoretical methane and hydrogen yields from gasification in hot compressed water.
Biomass
Water content (typical) (wt%)
Formula (daf )
Inorganic content (wt% of DM)
LHV (dry) (MJ kg1)
YCH4,max (kg CH4 kg oDM1)
YH2,max (kg H2 kg oDM1)
Wood (spruce) Manure (pig)a Sewage sludgeb Straw (wheat) Switchgrass Coal (lignite)c
50 84 70 10 13 36
CH1.50O0.67 CH1.75O0.56 CH1.96O0.53 CH1.61O0.66 CH1.46O0.67 CH0.80O0.42
0.3 29 39.1 7.8 3.6 17.7
18 14.6 13.3 17.3 17.4 17.6
0.34 0.41 0.44 0.36 0.34 0.41
0.17 0.20 0.22 0.18 0.17 0.20
DM, dry matter; oDM, organic dry matter; daf, dry ash-free. a Mechanically dewatered (curved screen). b Residential/urban, secondary sludge, undigested, mechanically dewatered. c For comparison (Great Plains).
Another important question is the maximum yield of the desired secondary fuel that can be produced from a certain feedstock. Table 12.2 lists the maximum theoretical yields for CH4 and H2 from a variety of biomass and waste types. The methane and hydrogen yields indicated in Table 12.2 are theoretical upper limits. In hydrothermal gasification, these yields will be lower because: . .
.
The methane and hydrogen yields are limited by chemical equilibrium, which leads to slightly lower yields than those in Table 12.2. The organically bound nitrogen contained in the biomass (neglected in the formulae shown in Table 12.2) will react to give ammonium (NH4 þ ), which will bind some of the hydrogen, making it unavailable for the synthesis of methane. Nitrates may react with the organic fraction under hydrothermal conditions and therefore the oxygen in the nitrate will lower the maximum methane and hydrogen yields, forming more CO2.
The yield of hydrogen includes also the hydrogen from the water taking part in the gasification. How much water reacts depends on the elemental composition of the organic fraction. From a green chemistry point of view, the production of CH4 has a better carbon atom economy, as roughly 50–60% of the carbon atoms in the biomass remain in the product CH4. In the case of hydrogen, all carbon atoms in the biomass are emitted as CO2. 12.1.4 Nutrient Cycles
As shown in Table 12.2, most biomass types contain a large fraction of inorganics. These comprise salts (nitrates, sulfates, chlorides, phosphates of potassium, sodium, calcium, ammonium, magnesium, etc.) and oxides (SiO2, CaO, Al2O3, MgO, Fe2O3,
12.2 Survey of Different Technologies for the Production of Methane
etc.). Many of the elements making up these substances are essential for plant growth. If considering algae as biomass feedstock, nutrients must also be supplied for optimal growth. Among these, N, P and K are the ones to be supplied as fertilizers to most plants and crops. The production of these fertilizers consumes a lot of fossil energy: 45 GJ per ton of N and 29 GJ per ton of P [7]. The energetic use of biomass with a high content of inorganics must also include the separation and recovery of these substances for reuse in agriculture. If not, the life cycle performance of such a biomass to energy process might become negative. Furthermore, as phosphorus reserves are finite, recovery of P from, for example, sewage sludge has become a priority in some countries [8]. When growing, biomass takes up nutrients from the soil, dissolved in water. These nutrients are incorporated into the biomass during the biochemical metabolism. Nitrogen is the key element for building amino acids and proteins. Plants take up nitrogen either as ammonium or as nitrate. Nitrogen bound in organic molecules such as pyridine, aniline or urea cannot be assimilated. It must first be converted into ammonium or nitrate by microorganisms. In the absence of oxygen, nitrate will be reduced to N2 (denitrification), which will be lost into the atmosphere. Phosphorus is the key element for the intermediate storage and release of energy in the cell via adenosine triphosphate (ATP). When the cell dies, the phosphate esters are readily hydrolyzed, liberating phosphate ions. Their mobility in soils is limited due to the formation of insoluble salts. If they are washed out into rivers and lakes, they may lead to eutrophication. During storage and spreading of manure, loss of N as NH3 and N2O into the atmosphere is significant. Only about one-quarter to one-third of the nitrogen contained in manure is taken up by the plants. In high-temperature thermal processing of biomass (combustion, gasification, pyrolysis), nitrogen is usually completely lost to the air as N2, as nitrous oxides (NOx) or as N2O, because both ammonium and nitrate are not stable under these conditions. In addition to the loss of nitrogen as fertilizer, the formation of NOx and N2O represents a negative environmental impact. Note that N2O has a global warming potential 310 times higher than CO2. P and K are usually recovered as oxides (P2O5, K2O) or salts from the ash of thermal processes. A sustainable biomass conversion process will allow for the complete recovery of N, P, K and other important elements and avoid noxious emissions such as NOx, N2O or NH3 to the air and phosphate and nitrate to the effluent.
12.2 Survey of Different Technologies for the Production of Methane from Carbonaceous Feedstocks 12.2.1 Anaerobic Digestion
Methane in the form of biogas is produced today in small- to medium-sized plants by anaerobic digestion of sewage sludge, manure, food wastes, oils and fats, kitchen
j285
j 12 Catalytic Conversion of High-Moisture Biomass to Synthetic Natural Gas in Supercritical Water
286
waste and other digestible organic waste fractions. There are two biochemical pathways of methane formation: . .
via acetate formation and subsequent enzymatic decarboxylation to CO2 and CH4 via CO2 and H2 in an enzyme-catalyzed methanation reaction (see Equation 12.4).
About 70% of the biogenic methane is produced by methanogenic bacteria according to the first pathway [9]. Since only dissolved gases can react in a biological system, factors that influence their concentration in the aqueous phase will also determine the formation of methane. The most important ones are: partial pressure of the reactants, pH and temperature. Microorganisms can only digest some fractions of the wastes, namely oils, fats, carbohydrates and proteins, but not lignin. The non-digested part ends up as residual sludge that contains also the nutrients and is usually used as fertilizer. Typical process conditions are a temperature of 30–40 C (mesophilic), sometimes up to 60 C (thermophilic range), and a residence time of 20–40 days. For cow manure, the methane yield for typical process conditions [7% dry matter (DM), 30–37 C, residence time 20 days] is about 0.15 g CH4 g1organic dry matter (oDM), which is less than half of the theoretical maximum of 0.41 g CH4 g1 oDM (compare Table 12.2). The biogas produced has usually a high concentration of methane (60 vol.%). The hydrogen concentration in the biogas is usually below 1 vol.%. The remainder is CO2, N2 and some trace compounds such as H2S, NH3 and organosilicon compounds. The latter represent a problem for downstream equipment such as a gas engines or a gas turbines and must be removed before the gas can be used for power generation or be fed into the natural gas grid. 12.2.2 Thermal Processes
Substantial efforts have been made since the 1960s to produce synthetic natural gas (SNG), mainly from coal or from naphtha, industrially at a large scale. The largest SNG plant in operation since 1984, the Great Plains Synfuels plant, is located in North Dakota (USA) and produces 3.7 106 m3 d1 (25 C, 0.1 MPa) or 2380 t d1 of SNG that is fed into the natural gas pipeline. Several by-products from gas cleaning (ammonium sulfate, cresylic acid, phenol) and from air separation (liquid nitrogen, krypton, xenon) are also sold, in addition to the CO2. The technology employed in this plant is a good example of a sequential approach using known and mature unit operations. Fourteen Lurgi Mark IV pressurized moving bed gasifiers are fed with 18 500 t d1 of lignite coal. The coal is gasified with steam and oxygen from a cryogenic air distillation unit. After cooling the gas and condensing out most of the water and also tars, the ratio of CO and H2 in the raw gas is adjusted in a water gas shift unit. After a cold methanol scrubbing unit (Rectisol) to remove CO2 and sulfur compounds, the cleaned gas is sent to the methanation reactors. These are cooled fixed-bed catalytic reactors. The average production of the plant corresponds to a methane
12.2 Survey of Different Technologies for the Production of Methane
yield of ca 0.24 kg of CH4 per kg of oDM, which is ca 60% of the upper limit given in Table 12.2 for lignite. Biomass gasification emerged initially from coal gasification technologies but, due to the peculiarities of biomass, new gasification technologies were developed. In G€ ussing (Austria), a wood gasifier with a thermal input of 8 MW, based on developments at the Technical University of Vienna, has been successfully operated since 2002. The indirect gasification technology is a combination of a bubbling bed steam gasifier and a circulating fluidized bed (CFB) combustor with a bed material (olivine) acting as heat carrier between the two units. Char from the gasifier is combusted in the CFB combustor. This technology is called fast internally circulating fluidized bed (FICFB), although it has undergone several changes since the initial development. The key advantage for synthesis applications is a nearly nitrogen-free producer gas. In the research project Bio-SNG of the European Union, the Paul Scherrer Institut (Switzerland), together with industrial and academic partners, is building a 2 MWth methanation process demonstration unit, based on a low-pressure fluidized bed technology [10]. When dealing with high-moisture biomass, conventional gas-phase gasification processes will yield very low thermal efficiencies. Schuster et al. [11] performed a parametric modeling study on the steam gasification of biomass, based on the FICFB technology implemented in G€ ussing. In this model, the gasification temperature of 800 C is maintained by recirculating some of the product gas to the combustion zone of the indirect gasifier. The cold gas efficiency, hchem, drops markedly with increasing water content of the biomass (Figure 12.1). This is mainly due to the increasing amount of product gas that must be recirculated in order to evaporate the water. At a water content of 66%, the efficiency drops to zero, that is, all of the product gas must be combusted in order to keep the gasifier at 800 C.
Figure 12.1 Operational parameters of a biomass steam gasifier as a function of the biomass water content. Tgasifier ¼ 800 C, p ¼ 0.1 MPa, steam-to-carbon ratio, not known. Reprinted from [11] with permission.
j287
j 12 Catalytic Conversion of High-Moisture Biomass to Synthetic Natural Gas in Supercritical Water
288
12.3 Water as Solvent and Reactant
Water is the greenest of all solvents because it is carbon-free, non-toxic and it does not have any negative impact on the environment. Actually, water is the only solvent that is compatible with biological systems. Most plants and living creatures are made up of more than 50% of water. Upon heating pressurized water, its hydrogen-bridged network of water molecules loosens and oligomers (H2O)n start to form a second, less dense phase, the vapor phase. Along the saturation line, both temperature and pressure increase. The population of the vapor phase increases and consequently also its density, whereas the population and the density of the liquid phase decrease. At the critical point, the two phases reach the same density, forming a single-phase system, which persists in the supercritical region (Figure 12.2). An interesting region is the one located between the line T ¼ 374 C and p > 22.1 MPa and the line of the pseudocritical points. This is a dense supercritical fluid (r ¼ 300–500 kg m3) that has not attracted much attention so far. The properties of near- and supercritical water have been the subject of excellent reviews [12, 13]. For the conversion of biomass, the following two roles of water are most important: .
as a solvent – for weakly polar and non-polar organic compounds (BTX, phenolics, fatty acids, naphthalene)
Figure 12.2 Vapor–liquid saturation line for pure water. Pictures represent molecular dynamic (MD) simulations of single point charge water molecules (red and white) and solvated Na þ (blue) and Cl (yellow). The dashed line extending the saturation line into the supercritical region represents the pseudocritical points. CP, critical point. MD pictures reproduced from [109] with permission.
12.3 Water as Solvent and Reactant
– for gases – for inorganics such as salts and soluble oxides .
as a reactant (source of H2O, OH, H, H þ , OH) – hydrolyzing agent – gasifying agent – Oxidant (e.g. for CO, metals).
12.3.1 Solubility of Organic compounds and Gases
As the hydrogen bonding of liquid water is weakened, its cohesive forces are also weakened, which results in a lower enthalpy of vaporization. This in turn changes the solvation properties of liquid water, which can be described with the Hildebrand solubility parameter d. For the supercritical fluid, other correlations are more appropriate [14]. In Figure 12.3, d and the dielectric constant e for saturated liquid water up to 370 C are plotted. Both follow the same trend, although with opposite curvature. Values for some common organic solvents at room temperature are also indicated in Figure 12.3. This illustrates that near-critical water has solvent properties similar to those of non-polar organic solvents under ambient conditions. This is a very important feature, as near- and supercritical water dissolve organic coke precursors, that is, phenolics and furfurals, preventing their deposition on catalysts and reactor walls.
Figure 12.3 Hildebrand solubility parameter d (solid line) and dielectric constant e (dashed line) for saturated liquid water. The e data are taken from [15]; d calculated using data from [15].
j289
j 12 Catalytic Conversion of High-Moisture Biomass to Synthetic Natural Gas in Supercritical Water
290
On the other hand, the solubility of gases such as H2, O2, CH4 and CO2 increases strongly towards the critical point. Supercritical water is completely miscible with these gases, forming a single-phase fluid. Furthermore, supercritical water provides higher diffusivities than liquids and better heat transfer than gases due to its relatively high density and low viscosity. The transition from liquid water to the supercritical fluid is a l phase transition, as opposed to a first-order phase transition. This means that no latent heat of vaporization has to be supplied. The change in enthalpy from liquid to supercritical water exhibits an S-shaped curve with the point of inflection located at the pseudocritical temperature. At temperatures >600 C, the difference between the enthalpy of supercritical water at 30 MPa and that of superheated steam at 0.1 MPa becomes small. 12.3.2 Solubility of Salts
Most biomass contains significant amounts of dissolved salts, either in the water phase between the cells or in the cell liquor (cf. Table 12.2). The latter are set free upon hydrolysis of the cells as the biomass is heated. Other inorganic compounds are also found in biomass, mainly metal oxides (SiO2, CaO, Al2O3, Fe2O3, MgO). None of these compounds contributes to the production of gas, but they have important implications for the chemistry during hydrolysis and gasification. Many salts exhibit an increase in solubility up to the pseudocritical point, where a sharp drop in solubility occurs (Figure 12.4). In general, alkali metal halides are more soluble in supercritical water than alkaline earth metal halides, and also sulfates and phosphates. Some metal oxides exhibit substantial solubilities in supercritical water such as SiO2 (2600 ppm at 500 C and 103 MPa [16]). The phase behavior of salt–water systems is very complex and has been reviewed by Valyashko [17]. For practical applications, it has proven useful to classify the salts into two categories named type I and type II [18]. The solubility of type I salts increases continuously with increase in temperature up to the melting point, whereas type II salts have solubilities that decrease over the entire temperature range or after passing through a maximum, becoming very small at the critical point. Another way to distinguish type I and II salts is the formation of a solid and often sticky, phase under supercritical conditions for type II salts, whereas type I salts form a liquid phase. Examples of type I salts include NaCl, KCl, CaCl2, KNO3 and NaNO3. Some type II salts are: Na2SO4, CaSO4, K2SO4 and CaCO3. 12.4 The Role of Heterogeneous Catalysis 12.4.1 Experimental Methods
Most of our experiments reported in this chapter were performed in three reactor systems: (i) fused quartz capillaries, (ii) an unstirred batch reactor with a volume of
12.4 The Role of Heterogeneous Catalysis
Figure 12.4 Salt solubilities as a function of temperature at a constant pressure of 25 MPa. Note the logarithmic scale of the solubility axis. The vertical line at T ¼ 385 C corresponds to the pseudocritical temperature for pure water at 25 MPa. Adapted from [19] with permission.
24 mL and (iii) a continuous catalyst test rig using a fixed-bed reactor. Details of these reactors have been published elsewhere [20–23]. For the tests with g-alumina reported in Section 12.4.7, three different quartz capillaries with an inner diameter of 3.3 mm and a length of 144–147 mm were used; 417–434 mg of water and 51 mg of g-alumina powder were loaded and the capillaries were sealed in a propane–oxygen flame. The capillaries were then placed in a grooved heating plate and covered with a quartz window to reduce heat losses. After predetermined times (90, 180 and 270 min), the capillaries were removed from the heating plate, quenched in cold water and opened for collecting the alumina. Recovery of the liquid phase was not attempted. 12.4.2 Thermodynamic Stability of Methane under Hydrothermal Conditions
Methane as the most stable hydrocarbon requires high temperatures to form more stable products such as CO and H2. Thermodynamic calculations of equilibrium
j291
j 12 Catalytic Conversion of High-Moisture Biomass to Synthetic Natural Gas in Supercritical Water
292
Figure 12.5 Molar yields of gaseous products at thermodynamic equilibrium for a mixture of 20 wt% CH1.50O0.67 and 80 wt% H2O. Pressure: 0.1 MPa (a) and 30 MPa (b). Solid carbon was excluded from the products.
compositions for a mixture of 20 wt% wood and 80 wt% water were performed with the software Aspen plus (Version 2006) (Aspen Technology, Cambridge, MA, USA). As can be seen from Figure 12.5, thermodynamic equilibrium for a mixture of 20 wt% wood and 80 wt% water (corresponding to a steam-to-carbon ratio of 5.4) would yield almost exclusively CH4 and CO2 up to ca 400 C at 30 MPa, whereas at 0.1 MPa CH4 is stable only below ca 300 C. From this analysis, one can conclude that in a high-pressure aqueous environment methane can be formed even at high temperatures.
12.4 The Role of Heterogeneous Catalysis
In practice, not thermodynamics but the relative kinetic rates of the different reactions involved will determine the gas composition. Because of the relatively low molar steam-to-carbon ratio of ca 1–3 at which biomass steam gasifiers are operated [24, 25], tars and char will also be formed. In fact, for the atmospheric gasification of woody biomass, these substances occur frequently and require extensive cleanup of the raw gas [26, 27]. 12.4.3 Main Reactions of Biomass Gasification
A simplified model of wood gasification includes drying of the wood particle (100–150 C); pyrolysis of the dry particle to gases, pyrolysis vapors and charcoal (200–500 C); and gasification of the charcoal with steam and/or oxygen to gases, leaving just ash behind (600–1000 C). Kinetic models for the pyrolysis of biomass have been reviewed recently by Di Blasi [28]. Since pyrolysis of the solid particle is influenced by heat and mass transport, the pyrolysis products are far from chemical equilibrium, that is, only little methane is formed. The main reactions involved in the gasification of the charcoal can be summarized in the following seven reactions: C þ H2 O ! CO þ H2
ð12:1Þ
CO þ H2 O $ CO2 þ H2
ð12:2Þ
CO þ 3H2 $ CH4 þ H2 O
ð12:3Þ
CO2 þ 4H2 ! CH4 þ 2H2 O
ð12:4Þ
C þ CO2 $ 2CO
ð12:5Þ
C þ 2H2 ! CH4
ð12:6Þ
CH4 þ CO2 $ 2CO þ 2H2
ð12:7Þ
where charcoal is denoted as C, although charcoal does not consist of pure carbon. The two main reactions for the formation of methane are the hydrogenation of CO (12.3) and CO2(12.4), in addition to the hydrogasification of the char (12.6). These reactions exhibit high activation energies and require 3 and 4 mol of hydrogen, respectively, for 1 mol of methane. Without a catalyst, reactions 12.3 and 12.4 will therefore not attain high rates and the methane concentration is usually far from the equilibrium concentration. The actual gasification reactions, steam gasification (12.1), Boudouard reaction (12.5) and hydrogasification (12.6), are reactions between a solid and a gas. An effective way to catalyze such reactions is by impregnating the biomass solid with alkali metal carbonates. K2CO3 and Na2CO3 were very effective in suppressing carbon deposition and tar formation in the steam
j293
j 12 Catalytic Conversion of High-Moisture Biomass to Synthetic Natural Gas in Supercritical Water
294
gasification of wood at 750 C and a molar steam-to-carbon ratio of 1.3 [29]. For the same reason, Ni catalysts for the steam reforming of hydrocarbons are frequently doped with alkali metals to reduce coking [30]. 12.4.4 Homogeneous, Non-catalyzed Pathways in Hot Compressed Water
In an aqueous environment, pyrolysis is no longer the sole reaction path at low temperatures. Hydrolysis becomes the dominant route because of the high density of the aqueous medium, as opposed to low-density steam. Furthermore, in liquid water H þ and OH exist, catalyzing the hydrolysis and the ion product Kw exhibits a maximum around 250 C. Ester, ether and amide (or peptide) linkages are particularly prone to hydrolysis in hot compressed water [31]. Methoxy groups, present for example in lignin, will yield methanol upon hydrolysis. In a hydrothermal gasification process, the main path does not produce char as an intermediate that is subsequently gasified. However, under certain conditions, char and tar formation may also occur in a hydrothermal environment. To illustrate this phenomenon, let us consider a small solid biomass particle, for example wood, surrounded by pressurized liquid water (Figure 12.6). As the water and the particle are heated, the water attacks the surface of the wood particle. Cellulose, hemicellulose and lignin are hydrolyzed to their building blocks, that is, sugars and substituted phenols, forming a liquid film around the particle. The water molecules must diffuse through this film to access the wood particle. In near-critical water, the oily film dissolves and is continuously removed, exposing fresh wood surface to the water. Note that hot compressed water has solvation properties similar to those of methanol above ca 220 C and similar to those of acetone above ca 300 C. Finally, the wood particle is liquefied completely. If an active catalyst is present, the dissolved oily compounds can easily access the catalytic sites, where they are gasified. Without an active catalyst, secondary coke and tar formation reactions may occur. In a steam atmosphere, the sugars and phenolics in the liquid film around the particle polymerize to tars. As the oily film covers the whole particle, no more water can penetrate and hydrolysis stops. The wood particle is surrounded by a diffusion
Figure 12.6 Schematic representation of a solid biomass particle dissolving in near-critical water and forming coke in steam.
12.4 The Role of Heterogeneous Catalysis
barrier for water and finally pyrolyzes to tars, coke and gases. The tars are not volatile enough at these temperatures to access the catalytic sites of a catalyst. They will just stick to the walls of the reactor and on the outer surface of the catalyst particles. Much higher temperatures (>600 C) are needed to crack the tars by steam reforming. Coke is not reactive at these low temperatures either. The two main reaction paths for the direct formation of gaseous products from the hydrolysis products are the decarboxylation to CO2 and the decarbonylation to CO. This is illustrated for the decarboxylation of malonic acid to acetic acid and CO2 [32] and for the decarbonylation of acetaldehyde [108]: HOOCCH2 COOH ! CH3 COOH þ CO2
ð12:8Þ
CH3 CHO ! CO þ CH4
ð12:9Þ
In an inert environment, for instance in sealed quartz capillaries, Kersten et al. [33] found CO to be produced in significant quantities from glucose, whereas similar experiments conducted in stainless steel or Inconel reactors with a high surface-tovolume ratio yielded much less CO. This is likely to be due to the metal walls catalyzing the water gas shift (WGS) reaction of CO to CO2. This work unequivocally demonstrated that CO must be considered as a primary product even at the high partial pressure of water in hydrothermal gasification. The same conclusion was drawn by Penninger and Rep [34] from continuous gasification experiments on an aqueous wood pyrolysis condensate in a tubular reactor made from Incoloy 825 (a nickel base alloy). CO was found to be a primary product, formed not from steam reforming but from decarbonylation reactions. The relatively low surface-to-volume ratio of the tubular reactor was assumed not to catalyze the reactions significantly. In an aqueous environment, the WGS reaction can also be catalyzed homogeneously by alkali metal salts [35], but the mechanism is not well understood. The steam reforming of larger aromatic molecules (e.g. phenols) occurs as a sequence of several steps, including ring opening, to compounds of lower molecular weight, some of these steps liberating CO or CO2. Detailed chemical kinetic models are usually derived from high-temperature combustion [36] or pyrolysis [37] mechanisms. Unfortunately, these mechanisms do not represent accurately the conditions encountered in hydrothermal processing at temperatures below ca 500 C. For instance, stable polar intermediates, such as acetic acid and formic acid, are usually not represented in mechanisms derived from combustion chemistry. A more successful approach is to combine both homolytic (radical) and heterolytic (ionic) pathways [38]. In wet oxidation under mild conditions, phenol was shown to decompose to acetic acid and formic acid as final stable organic compounds [39]. Under the conditions of hydrothermal gasification, these compounds may decompose to form CO, CO2, CH4, H2 and H2O: CH3 COOH ! CO2 þ CH4
ð12:10Þ
CH3 COOH ! 2CO þ 2H2
ð12:11Þ
HCOOH ! CO þ H2 O
ð12:12Þ
j295
j 12 Catalytic Conversion of High-Moisture Biomass to Synthetic Natural Gas in Supercritical Water
296
Acetic acid, acetaldehyde, formaldehyde and small amounts of formic acid were found in the liquid phase after the uncatalyzed gasification of biomass at 330–410 C [40]. Acetaldehyde and acetic acid were both detected in significant concentrations in the aqueous phase after gasifying wood sawdust with a skeletal nickel catalyst [20], confirming that these compounds are important intermediates on the way to gaseous products. Reaction (12.10) is of particular importance for understanding the geological synthesis of natural gas from organic matter [41]. According to Palmer and Drummond [42], the non-catalyzed decomposition of acetic acid is a very slow reaction. Only ca 5% was decomposed after 141 h at 400 C in a titanium reactor. Interestingly, gold and titanium were found to be the most inert materials for the decomposition of acetic acid, whereas stainless steel, silica and magnetite showed marked catalytic effects. The main products were in all cases CH4 and CO2 (Equation 12.10), but traces of CO, H2 and volatile hydrocarbons were also detected (Equation 12.11). In fact, acetic acid is one of the most stable compounds in hot pressurized water in the absence of catalysts, which may just be metal walls of the reactor [43]. Acetaldehyde decomposition was studied in quartz ampoules by Nagai et al. [44]. The main products at 400 C in the presence of pressurized water at a density of 500 kg m3 were CH4 and CO2, followed by H2. The authors suggested that acetaldehyde decarbonylated according to Equation 12.9 to CH4 and CO and that CO2 and H2 were formed subsequently from CO by the WGS reaction. This is consistent with the results from Arita et al. obtained at 500 C [108]. The decomposition of formic acid in hot pressurized water is still debated in the literature. It seems that in an inert reactor, for example quartz, CO is the main product (see Equation 12.12) [45]. The pH plays an important role, as acidic conditions favor CO formation and alkaline conditions CO2 formation [46]. In a metal reactor (Hastelloy C-276), Yu and Savage oll et al. [48] found CO to be the main found CO2 to be the main product [47]. Br€ product below ca 320 C at 30 MPa in a view cell made of Inconel 625. Again, catalysis by the metal walls seems to play an important role in the decomposition of formic acid. Based on these observations, the hydrothermal gasification of organic compounds in the absence of a catalyst may be described as a sequence of three steps: . . .
conversion of the organic compound with water to form mainly acetic acid, acetaldehyde and formic acid decomposition of these C2 and C1 compounds to CO, CO2, CH4 and H2O formation of H2 (and CO2) from CO by the WGS reaction.
The reforming of organic compounds in a hydrothermal environment will therefore in general lead to a mixture of CO, CO2, CH4 and H2. In the absence of a catalyst, methane will thus be formed directly from acetic acid and/or acetaldehyde (primary methane). Without a catalyst, steam reforming is slow below ca 600 C. As an example, DJesus et al. [49] achieved 41% conversion of corn starch to gaseous products at 550 C with a residence time of ca 2 min., whereas at 700 C and a similar residence time the conversion increased to 92%. Note, however, that these results are affected
12.4 The Role of Heterogeneous Catalysis
by catalysis from the walls of their tubular Inconel 625 reactor and also from adding KHCO3 as a homogeneous catalyst. The slow rates of the (truly) uncatalyzed reforming reaction become evident on looking at the results of DiLeo et al. [50]. The gasification of phenol at 600 C and 27.5 MPa yielded a conversion of only 68% after 1 h in a sealed quartz capillary, with the gas composition being far from equilibrium. The reactivity of oxygenated molecules such as cresol and phenol is higher than that of char. In general, the more carbon-like the molecule, the more difficult it is for it to undergo steam reforming (i.e. the slower the rate). Pure carbon such as graphite or amorphous carbon is chemically very stable in a non-oxidizing hydrothermal environment. In fact, they can be used as a catalyst support (see Section 12.4.7). Under the conditions of supercritical water oxidation (SCWO), that is, at high temperatures in an oxidizing environment, carbon is fully converted to CO2. This is used in a commercial process to recover precious metals from spent catalysts with carbon supports (AquaCat [51]). Tests at General Atomics with supercritical water partial oxidation (SWPO) led to the conclusion that coal cannot be sufficiently gasified in supercritical water at 23.5 MPa up to 800 C [52]. Others have studied the gasification of coal in supercritical water with greater success [53]. Cheng et al. [54] were able to gasify a reactive fraction of the coal, corresponding to its volatile matter, but a substantial amount of less reactive solid residue remained. 12.4.5 Heterogeneously Catalyzed Pathways in Hot Compressed Water
If H2 is to be the main product, the WGS reaction is the main provider of H2 and the methanation reactions must be suppressed. Gadhe and Gupta [55] discussed the influence of several parameters for suppressing methane formation during the gasification of methanol in supercritical water at 700 C. The role of heterogeneous catalysts for the production of CH4 is primarily to increase the rates of the methanation reactions 12.3 and 12.4, and also of the WGS reaction 12.2, to provide the necessary hydrogen. Accelerating the preceding steps of hydrolysis and steam reforming will, of course, also accelerate the rate of the overall conversion to methane. Hydrolysis of the biomass is the reaction of a solid with liquid water and therefore homogeneous catalysis is more effective than heterogeneous systems. Only once lower molecular weight compounds have been formed that can access catalytic sites do heterogeneous catalysts become effective. Steam reforming to small aliphatic molecules such as acetic acid, formic acid and acetaldehyde becomes then the dominant reaction. The relative ranking of the rates of the three main reactions in pressurized water in the presence of a catalyst is WGS methanation > steam reforming. Without a catalyst, chemical equilibrium is usually not attained even at 600 C. This relative ranking explains why in batch experiments with incomplete conversion of the organic feed a high concentration of methane is still observed. On the other hand, in the absence of a catalyst, full conversion can be reached at high temperatures (>600 C) and/or long residence times but with a gas composition still far from
j297
j 12 Catalytic Conversion of High-Moisture Biomass to Synthetic Natural Gas in Supercritical Water
298
equilibrium. However, catalysis by the metal walls of the reactor will often accelerate the WGS reaction and the methanation to some extent [55]. An important difference between the non-catalyzed and the catalyzed pathways lies in the formation of methane. For the non-catalyzed situation, reactions 12.10 and 12.9 are the predominant routes to methane (primary methane). Temporal sampling and analysis of the gas phase during the hydrothermal gasification using a Ru/C catalyst did not reveal a significant amount of primary methane, either from spruce or from a mixture of ethanol, acetic acid, formic acid, anisole and phenol [22]. We therefore conclude that acetic acid decomposes primarily according to reaction 12.11 in the presence of a Ru catalyst, forming CO and H2. The mechanism of this pathway is believed to follow a redox cycle with the metal shuttling between two oxidation states (Equations 12.13a and b). Methane is then formed almost exclusively by methanation of CO and/or CO2 (secondary methane). CH3 COOH þ 2RuO2 ! 2CO þ 2RuO þ 2H2 O
ð12:13aÞ
RuO þ H2 O ! RuO2 þ H2
ð12:13bÞ
The question of whether CO or CO2 is hydrogenated in a hydrothermal environment is still debated in the literature. CO2 methanation was shown to occur by Minowa and Fang [56]. In an experiment with CO2 and H2 as reactants, they loaded 45 mmol of CO2 and 86 mmol of H2, together with water and a Ni catalyst, into a batch reactor, heated the mixture to 350 C and held it at that temperature for 30 min. The gas phase after the experiment consisted of 53 mmol of H2, 41 mmol of CO2 and 6 mmol of CH4. No CO was detected. Taking into account that on transition and noble metal catalysts CO is adsorbed much more strongly than CO2, the surface coverages of both CO and H2O will be high in a hydrothermal environment. This will favor fast WGS of CO to CO2 and H2, explaining why CO is usually not detected as an intermediate in the presence of a catalyst. On the other hand, H2, formed as a secondary product, will never reach the same high coverage as H2O and therefore the rate of CO hydrogenation is very likely much smaller than the rate of WGS. This line of argumentation would support the hydrogenation of CO2 as the main path for the formation of CH4 in a hydrothermal environment in the presence of a catalyst. A comparison of CO2 hydrogenation on reduced and oxidized Ru catalysts yielded much higher rates on the oxidized Ru. The proposed mechanism leads directly to methane and does not produce CO as an intermediate [57]. A similar observation has been reported for Rh [58]. 12.4.6 Active Metals Suited to Hydrothermal Conditions
Using different techniques, the onset of gas formation from wood sawdust in hot pressurized water in the presence of a catalyst was located between 250 and 290 C [59]. This range sets the stage for the catalysts. Unlike in atmospheric gasification, where the gasification of the char does not start below ca 600 C, hydrothermal gasification and methanation require catalysts with a high activity
12.4 The Role of Heterogeneous Catalysis
below 300 C. Catalysts that are not active below this temperature will lead to tar and secondary coke formation (see Section 12.4.4). Drawing from the experience of Elliott and co-workers [60, 61] and on our own screening program [59], skeletal nickel and Ru supported on TiO2, C and ZrO2 were found to be active and selective catalysts for the gasification and methanation in supercritical water at temperatures of 350–450 C. 12.4.6.1 Methanation and Steam Reforming Catalysts A more general approach is to test systematically combinations of useful metals and useful supports. If limited to one metal, Vannice [62] lists the following metals in order of decreasing activity for the methanation of CO: Ru > Fe > Ni > Co > Rh > Pd > Pt > Ir. Vannice conducted this study on h-alumina as support (SBET ¼ 200 m2 g1), which is not stable under hydrothermal conditions (see Section 12.4.7). Nevertheless, this study gives useful hints about the relative rates for different common metals. We have replotted the data of Vannice in a selectivity versus activity diagram in Figure 12.7. Ru is by far the most active metal, followed by Fe with about half the activity of Ru, based on the turnover number for CO conversion. Ni as very popular methanation catalyst exhibits only ca 10% of the activity of Ru, but the selectivity towards methane is higher for Ni. Pd has the highest selectivity but a low activity. It is no surprise that both Fe and Ru are used as Fischer–Tropsch catalysts for the synthesis of higher hydrocarbons, as they exhibited a significant selectivity to these products in the studies of Vannice. Iron has about half the activity of Ru and the lowest selectivity of all catalysts tested, but its resistance to poisons is superior to nickel [63]. Rh was found to be a good catalyst for the hydrogenation of CO2, exhibiting activation energies much lower than for CO hydrogenation on the same catalyst. This was explained by partial oxidation of the Rh surface by CO2 [58].
Figure 12.7 Activity–selectivity plot for metals supported on h-alumina using data from Vannice [62] for the hydrogenation of CO at 275 C and atmospheric pressure.
j299
j 12 Catalytic Conversion of High-Moisture Biomass to Synthetic Natural Gas in Supercritical Water
300
All the metals studied by Vannice have also been tested for their performance under hydrothermal conditions. Of the metals tested by Vannice, Fe, Co, Pt, Pd and Ir were found by Elliott and co-workers to be inactive in batch screening tests with pcresol and phenol at 350 C and 20 MPa [60, 61]. However, there was an important difference in these two studies. Whereas Vannice studied only the methanation of CO from a CO–H2 mixture, the study of Elliott and co-workers included also the gasification of the feed compound p-cresol or phenol to CO and H2 as the first step before the methanation. To compare the results of Vannice and Elliott and coworkers, the performance of these catalysts in steam reforming of organic molecules such as cresol must be taken into account. Rostrup-Nielsen [30] tested many different catalysts for their activity in the steam reforming of ethane, the simplest hydrocarbon with a C–C bond. From this study, the following order of activity was found for the metals supported on alumina or magnesia: Rh, Ru > Ni, Pd, Pt > Re > (Ni0.7Cu1.3) > Co. Fe was not tested, but Satterfield [63] attributes it an activity similar to Co. Cu is not able to break C–C bonds and it therefore cannot be used for steam reforming of higher hydrocarbons. For reforming methanol, however, Cu possesses a remarkable activity [64]. In Rostrup-Nielsens study, coking did not occur on supported Ni–Cu alloy, Co, Pt, Ru and Rh, all on alumina, even at steam-to-carbon ratios as low as 0.3–0.6. Cobalt has not attracted much attention in hydrothermal gasification. Elliott and co-workers [60, 61] tested a catalyst composed of 39% Co oxide on kieselguhr from Harshaw. This catalyst yielded a carbon conversion to gases of only 0.08% after 90 min at 350 C and 20 MPa and was therefore not considered active. Furusawa et al. [65] prepared a 10% Co/MgO catalyst (metal surface area 2 m2 g1, crystallite size 99 nm) for the gasification of organosolv lignin at 400 C in a batch reactor. The carbon conversion was only 5% after 2 h and the yield of gaseous products was low. In our own screening program, a skeletal Co catalyst (Raney Co 2700 from Grace) showed a better performance. Wood sawdust (10 wt% in water) was gasified at 388 C and 28 MPa for ca 45 min. with about half the carbon retrieved in the gas phase, consisting of 54 vol.% H2, 40 vol.% CO2, 6 vol.% CH4 and 0.3 vol.% CO. The remaining liquid phase was not analyzed for dissolved Co. Although skeletal Ni catalysts performed much better, the Co catalyst gave better results than 1% Ru/TiO2 tested under similar conditions. In particular, the 1% Ru/TiO2 yielded 4.4 vol.% CO, indicating a slow WGS reaction. Note that this catalyst yielded a similar CO concentration in a continuous gasification run with 30 wt% ethanol (see Table 12.6). The results of Elliott and co-workers are difficult to interpret because they used many different combinations of metals, metal loadings and supports. Pt, as an example, was tested with a loading of 5 wt% on g-Al2O3 only, whereas Ru was tested with seven different supports and loadings. A study performed by Osada et al. [66] took into account also the metal dispersion and calculated turnover numbers for the gasification of lignin at 400 C and 37 MPa. The results from the nine catalysts tested are plotted in Figure 12.8 in a similar manner to the data of Vannice (cf. Figure 12.7). For Osada et al.s data, however, selectivity to methane is calculated as carbon selectivity from the measured gas composition, that is, the fraction of carbon in all gaseous products
12.4 The Role of Heterogeneous Catalysis
Figure 12.8 Activity–selectivity plot for supported metals using data from Osada et al. [66] for the gasification of organosolv lignin at 400 C and 37 MPa. 1, 2% Ru/TiO2; 2, 5% Ru/Al2O3; 3, 5% Ru/ C; 4, 5% Rh/C; 5, 5% Pt/C; 6, 2% Pt/Al2O3; 7, 5% Pd/C; 8, 5% Pd/ Al2O3; 9, 17% Ni/Al2O3.
present as methane. The maximum selectivity is limited by the feed stoichiometry at ca 0.5. The three Ru catalysts (1, 2, 3) gave the best performance. Interestingly, in Osada et al.s study the Ni catalyst (9) yielded the poorest performance (lowest activity and lowest selectivity to methane). This might be explained by the relatively low Ni loading of 17 wt% and the very low dispersion of 5%, leading to a low metal surface area for this catalyst. We have tested a methanation catalyst from Degussa (type H 14174 RS-15%), 15% Ni on a-Al2O3 (Ni surface area 0.92 m2 g1), that also exhibited a very low activity around 400 C and 30 MPa [67]. A similar result was obtained by Elliott and co-workers for 20% NiO on a-Al2O3. Highly loaded Ni catalysts (> ca 45% Ni) yielded good results [60, 68, 69]. The low activity of iron for steam reforming might be explained by its phase behavior in a hydrothermal environment. Under typical conditions of 350–450 C, p (H2O) ¼ 20–30 MPa, iron is present as iron(III) oxide, Fe2O3 (hematite), which has no driving force towards the reduction to Fe(II) oxide (cf. Table 12.4). The surprising inactivity of Pt (5% Pt on g-alumina from Engelhard, SBET ¼ 125 m2 g1) found by Elliott and co-workers is in contrast to the findings of Dumesics group. They produced H2 from oxygenates such as ethylene glycol in aqueous solution at 210–250 C and ca 3 MPa using supported Pt catalysts, and also skeletal Ni/Sn catalysts [70]. The order of decreasing activity at 210 C found by Dumesics group was Pt > Ni > Ru > Rh Pd for these metals supported on silica [71]. The Pt catalysts tested by Osada et al. exhibited an intermediate performance, with Pt/C showing similar activity to Ru/C but with a lower selectivity to methane.
j301
j 12 Catalytic Conversion of High-Moisture Biomass to Synthetic Natural Gas in Supercritical Water
302
Table 12.3 Summary of useful metals for the gasification of biomass
under hydrothermal conditions. Alloys (PtRu, NiSn, etc.) have not been considered. Metal
Oxidizing
Reducing
Ni (stabilized) Ru Rh Pt Pd (promoted by Fe [71])
þ ? (þ) (þ)
þ þ þ þ þ
? ¼ not known; ( þ ) ¼ no conclusive information available.
Palladium has been reported by several groups to exhibit only low activity in hydrothermal gasification and methanation [60, 72, 73, 66, 74]. In contrast, Pd–Fe systems, prepared as 6% Pd on high surface area Fe2O3 (SBET ¼ 250 m2 g1), exhibited a high activity for the aqueous phase reforming of ethylene glycol at 210 C [71]. The explanation offered by the authors suggests iron to act as WGS promoter, keeping the CO coverage of the Pd surface low and thus active for reforming (i.e. C–C bond splitting) reactions. Therefore, the low activity of Pd could be due to its low WGS activity, leading to its self-poisoning by CO. Other metals tested by Elliott and co-workers that showed no activity for gasifying p-cresol or phenol were Cu, Zn, Cr, W, Mo, Re, Ag, Sn and Pb. This is not surprising, as these metals do not exhibit an activity for the steam reforming of organic molecules (Cu, Re) or they are readily oxidized under hydrothermal conditions (Zn, Cr, W, Mo, Sn, Pb). Silver has not been tested as a gasification catalyst under hydrothermal conditions so far. Table 12.3 summarizes the metals that were found to exhibit a useful performance, that is, high activity and selectivity and good stability, for the gasification and methanation of modelcompounds andreal biomass.Oxidizing means conditions encountered in partial oxidation (substoichiometric amount of oxygen). Subcritical conditions typical for wetoxidation are not consideredin this chapter. Catalysts forthese conditions have been reviewed by Ding et al. [16], Bhargava et al. [75] and Levec and Pintar [76]. Two metals, nickel and ruthenium, are discussed in more detail below. 12.4.6.2 Nickel Nickel is the catalyst of choice for the steam reforming of methane and higher hydrocarbons (e.g. naphtha), and also for the methanation of CO and CO2 [62, 63]. It is relatively cheap, active and selective and very well studied. The active phase for steam reforming and also for methanation was reported to be reduced nickel [58]. At high partial pressures of water, that is, in a hydrothermal environment, the predominant nickel phase is NiO, as shown in Figure 12.9. Reduced nickel, denoted Ni(FCC) in Figure 12.9, will be thermodynamically stable under reducing conditions, that is, at high partial pressures of H2 and CO above 800 C. This prediction was verified for a skeletal nickel catalyst used for the hydrothermal gasification of wood sawdust around 400 C.
12.4 The Role of Heterogeneous Catalysis
Figure 12.9 Equilibrium diagram for Ni, biomass (CH2O assumed) and water. Ni/H2O ¼ 0.0134 (mol/mol), S/C ¼ 12.4 (10 wt% CH2O), total pressure ¼ 30 MPa.
After the experiment, the catalyst was first washed with water and methanol to remove salts and organic deposits. XPS analysis revealed oxidized Ni to be the predominant Ni species on the surface, either as Ni(III) oxide or Ni(II) hydroxide. Most of the surface was covered by aluminum hydroxide, typical for skeletal Ni catalysts. Interestingly, the bulk of the catalyst was still composed of reduced Ni, as determined by XRD. The aluminum hydroxide layer apparently prevents bulk oxidation of the nickel by water. Reduced nickel as the main phase was also observed by Elliott et al. [60] for supported nickel catalysts, reduced before use, after hydrothermal gasification. Furthermore, both XPS and TPO of the spent catalyst revealed some carbonaceous deposits on the surface. However, already the fresh catalyst exhibited surface carbon (ca. 10 at.% versus 15 at.% on the spent catalyst). This might be explained by the fact that the catalyst was exposed to the laboratory air during sampling and analysis, taking up volatile organics on its surface. This is illustrated in Figure 12.10. For metals whose predominant form is an oxide or hydroxide, steam reforming in near- and supercritical water most likely occurs according to a Mars–van Krevelen redox mechanism: CH2 O þ Mðþ nÞO2 ! CO þ H2 O þ Mðþ n2ÞO
ð12:14aÞ
Mðþ n2ÞO þ H2 O ! Mðþ nÞO2 þ H2
ð12:14bÞ
Although skeletal nickel catalysts exhibited a very high initial activity, they lost most of their activity within a few hours or possibly even less time on-stream. The main cause was found to be sintering of the nickel crystallites [23]. This would not be
j303
j 12 Catalytic Conversion of High-Moisture Biomass to Synthetic Natural Gas in Supercritical Water
304
Figure 12.10 Schematic representation of the skeletal Ni surface after a hydrothermal gasification experiment.
expected from a thermal treatment of supported nickel catalysts, as the Tammann temperature of nickel is 590 C [77]. However, skeletal nickel catalysts behave differently and thermal sintering is expected at temperatures lower than the Tammann limit [78]. Furthermore, in a hydrothermal environment dissolution and recrystallization constitute another likely sintering mechanism, in particular for the alumina present on the surface of the skeletal Ni catalyst. Elliott and Hart succeeded in stabilizing the nickel crystallites of a commercial catalyst from BASF (G1-80 used for adiabatic prereforming, 50% nickel on a proprietary support) by doping the nickel with several other metals including Ru, Cu, Re, Sn, Pb and Ag [79]. All these metals, added at 1 wt% or (Ru) 5 wt%, stabilized the initial nickel particle size of 6 nm to a value between 10 and 19 nm after 65 h in a batch autoclave at 350 C and 20 MPa. The tin and the lead, however, reduced the catalytic activity of the catalyst compared with the undoped catalyst. In long-term tests over 24 weeks, the Ru-doped catalyst proved to be the most stable variant. Building on these findings, we tried to stabilize skeletal nickel catalysts by having them doped by the manufacturer individually with Ru, Cu and Mo. As opposed to Elliott and Hart, these additions did not increase the lifetime of the skeletal Ni catalyst significantly [59]. This difference might be due to the higher amounts of doping metals added and the lower temperatures (350 C) used in their work. A promising nickel catalyst was reported by Nakagawa et al. [69]. They impregnated methacrylic ion-exchange resin particles of diameter ca 0.5 mm with a solution containing Ni(II). After drying and pyrolysis in N2 at 500 C, spherical catalyst particles of diameter 0.2–0.3 mm were obtained. The nickel content was 47 wt% and SBET was 170 m2 g1 with pores in the size range 1–10 nm. This catalyst was successfully tested for the gasification of phenol, lignin and other compounds and did not show any signs of deactivation after 200 h on-stream [80]. Attempts to use powdered nickel as a catalyst met with limited success. This is not surprising, as the available Ni surface area of a non-porous powder is given by its external surface area. For 10 mm Ni spheres this area is only 0.07 m2 g1, whereas for 0.5 mm particles it reaches 1.3 m2 g1, which is in the range of Ni on a-alumina. Elliott et al. [60] tested a Ni powder with particle size <150 mm that showed no activity for the
12.4 The Role of Heterogeneous Catalysis
gasification of p-cresol at 350 C. Minowa and Inoue [81] were able to reach gasification yields of ca 75% using 0.5 mm Ni powder for the gasification of cellulose at 350 C. In other words, in order to reach a similar ratio of surface Ni to biomass, a very large amount of Ni powder would have to be added. As an example, 148 g of the 10 mm particles would have to be added to provide an Ni surface area of 10 m2. The same area would be provided by 1 g of a supported Ni catalyst with 10 m2 of Ni per gram of catalyst. For the same reason, metal walls of reactors with a low surface-to-volume ratio will not have a significant catalytic effect unless they have been pretreated (on purpose or inadvertently) to form small crystallites, and thus a large specific surface area, on their surface. 12.4.6.3 Ruthenium Much of what has been found for the skeletal nickel catalysts relating to the pathways is also valid for ruthenium catalysts. Park and Tomiyasu [82] demonstrated that unsupported RuO2 is active in the hydrothermal gasification of naphthalene, polyethylene, cellulose and other compounds at 450 C. Labeling experiments with D2O supported a redox mechanism as described by the following equations:
CH2 O þ RuO2 ! CO þ RuO þ H2 O
ð12:15aÞ
RuO þ H2 O ! RuO2 þ H2
ð12:15bÞ
We confirmed the findings of Park and Tomiyasu by performing a similar experiment in our batch reactor using powdered RuO2 (Riedel de Ha€en) for the hydrothermal gasification of a model mixture of five organic compounds. The gasification efficiency after 60 min at 400 C and ca 30 MPa was 97.8% and the gas composition was 40 vol.% CH4, 47 vol.% CO2, 13 vol.% H2, 0.3 vol.% CO and 0.3 vol.% C2H6. Although powdered RuO2 does not have a high specific surface area (ca 1 m2 g1 for a 1 mm sphere), it is active for the hydrothermal gasification and methanation. This suggests that, in addition to the number of available surface sites, another factor must be important for determining the catalysts activity. Considering the redox mechanism for the gasification of the organics, the reducibility of the metal in the higher oxidation state and the oxidizability of the metal in the lower oxidation state are believed to play an important role. In other words, the rates of reactions 15a and 15b are determined by the redox kinetics of the metal oxides. If the reduction of the higher oxidation state of the metal by the organics is too slow, the organics will not be gasified and they may form secondary coke. The measure for the driving force for bulk metal reduction in aqueous phase is the reduction potential at standard conditions. These values are listed for some metals in Table 12.4. Metals that can be reduced by organic compounds in aqueous solutions should also be suited to reforming according to the Mars–van Krevelen mechanism (Equations 12.14a and b). Of course, E values are a driving force only and do not carry any kinetic information, but they can be used to identify potentially well reducible metals. If the E value is negative, that is, if there is no driving force for the
j305
j 12 Catalytic Conversion of High-Moisture Biomass to Synthetic Natural Gas in Supercritical Water
306
Table 12.4 Standard reduction potentials and reduction potentials at pH 3 for selected bulk metal oxides [83].
No.
Redox pair
E (V)
E (pH 3)
1 2 3 4 5 6 7
2RuO2 þ 2H þ þ 2e $ Ru2 O3 þ H2 O Rh2 O3 þ 4H þ þ 4e $ Rh2 O þ 2H2 O Ni3 O4 þ 2H þ þ 2e $ 3NiO þ H2 O Fe2 O3 þ 4H þ þ 2e $ 2FeOH þ þ H2 O Co3 O4 þ 2H þ þ 2e $ 3CoO þ H2 O PdO2 þ 2H þ þ 2e $ PdO þ H2 O PtO2 þ 2H þ þ 2e $ PtO þ H2 O
0.937 0.877 0.876 0.16 0.777 1.263 1.01
0.7597 0.6997 0.6987 0.194 0.600 1.086 0.833
reduction, the metal will not be able to participate in a mechanism such as Equations 12.14a and b. During the liquefaction of the biomass, the pH of the solution will decrease, as carboxylic acids are formed. When starting from a neutral solution containing no buffer salts, this means that the environment will become acidic. After liquefying spruce sawdust, the pH of the remaining solution was around 3. Table 12.4 reveals that some oxides of Ni, Ru, Pd, Rh and Pt have values of E in the range 0.7–1 V at pH 3. Co has a lower value of E than the other metals considered here, except Fe. Iron(III) oxide has a slightly positive value of E , which becomes negative at pH 3. The order with respect to the driving force for the reduction of an oxide is Pd > Pt > Ru > Rh Ni > Co Fe. Interestingly, this order is in line with the observation that platinum group metals are good catalysts for aqueous phase reforming, that nickel is better than cobalt and that iron is not able to participate in such a mechanism. Pd is an exception, which could be due to the self-poisoning effect by CO discussed above. This comparison suggests that reduction potentials of bulk metal oxides can be used to evaluate the potential of a certain metal oxide to undergo a reduction by dissolved organic compounds according to a Mars–van Krevelen redox mechanism. Although in reality only the surface rather than the bulk oxide participates in the redox cycle, we believe that the reduction potentials of the bulk oxides correlate with the behavior of the surface oxide. To corroborate this hypothesis, well-defined and systematic experiments for determining turnover frequencies of these metals on different supports and also unsupported (skeletal or sponge) types, not biased to a particular metal or support, are recommended. 12.4.7 Catalyst Supports Suited to Hydrothermal Conditions
The function of the support is to provide a surface for dispersing small crystallites of the active metal and to provide a physical structure for the reactants to access the crystallites (open pores). In some cases, the support material was found to enhance the effect of the active metal [strong metal–support interaction (SMSI)]. In particular, titania exhibited a strong accelerating effect on Ni and Rh catalysts at a coverage of about half a monolayer. The SMSI effect was observed for Ni, Co, Fe and Rh on several
12.4 The Role of Heterogeneous Catalysis
Figure 12.11 X-ray diffractograms of g-alumina exposed to supercritical water for different times. Traces of the fresh sample and of a sample treated in air at 400 C are also included for comparison.
oxides including TiO2, La2O3, Nb2O and Ta2O5 [58]. All these factors have been studied and optimized mostly for gas-phase reactions and to a much smaller extent for liquid-phase reactions. Since sub- and supercritical water exhibit properties very different from those of gases and liquids, the behavior of common catalyst support materials such as SiO2, Al2O3, CeO, TiO2, MgO, C and so forth must be known in order to evaluate their suitability. As an example, we exposed amorphous g-alumina to hot compressed water in sealed quartz tubes and measured the XRD (ex situ) and the BET surface area of the samples exposed for different times. Figure 12.11 shows the diffraction patterns for different exposure times, starting with the fresh amorphous alumina. A control sample treated for 270 min. at 400 C without water is also shown. Already after 90 min of exposure to hot compressed water distinct reflexes appear corresponding to hydrated alumina AlO(OH) or boehmite. These reflexes increase up to 180 min and then decrease again. This is depicted in Figure 12.12 for a typical reflex of boehmite. Also shown in Figure 12.12 is the BET surface area of the samples. Within the first 90 min, the BET surface area SBET decreases from ca 220 to between 50 and 60 m2 g1. The control sample in air does not shown any new reflexes but its SBET value increases to ca 285 m2 g1. The thermodynamically stable form of alumina at 400 C and 30 MPa is not fully elucidated. Figure 12.13 is taken from an often-cited reference [84] that states that under these conditions corundum is the stable phase, but frequently diaspore is considered to be the stable phase. By heating the alumina in the sealed quartz tube, the trajectory followed is close to the saturation line of pure water. This line crosses the metastable region for boehmite, which was determined by XRD. It seems that the transition of the amorphous alumina to the hydrated boehmite form is fast, but the further transformation into the stable form, diaspore or corundum, is slow.
j307
j 12 Catalytic Conversion of High-Moisture Biomass to Synthetic Natural Gas in Supercritical Water
308
Figure 12.12 Intensity of one of the AlO(OH) reflexes (diamonds and dotted line) and the BET surface area (squares and solid line) of g-alumina exposed to supercritical water for different times. The triangle at 270 min. corresponds to a sample treated in air at 400 C.
As the treatment of the g-alumina in air at 400 C did not reveal a different XRD pattern, thermal sintering can be ruled out. In fact, g-alumina is used as a catalyst support up to ca 600 C [63]. A phase change to a-alumina occurs only at temperatures >1100 C [84]. The mechanism in a hydrothermal environment most likely involves dissolution–recrystallization, initiated by hydration of the alumina to AlO (OH), which exhibits a certain solubility in water. Upon cooling the quartz ampoules,
Figure 12.13 Phase diagram of the alumina–water system. Reprinted from [84] with permission.
12.4 The Role of Heterogeneous Catalysis
the dissolved AlO(OH) will recrystallize to the boehmite form. Note that the XRD patterns of the samples were measured ex situ after drying the sample. Elliott et al. tested a range of supports at 350 C and 20 MPa, including a-, g-, d- and h-alumina. All transition alumina types, that is, the g, d and h forms, were transformed to the boehmite phase. This transformation was accompanied by a loss of surface area for the d type and a loss of crush strength for the g type [60]. The collapse of the BET surface area in less than 90 min. (see Figure 12.12) demonstrates that g-alumina is not a suitable support under hydrothermal conditions. In addition to the phase transformations, another key parameter is the solubility of the support material in supercritical water. If the support dissolves even in small amounts, this will lead to a breakdown of the catalyst particle after a certain time on-stream. Such materials include SiO2, V2O5 and Fe2O3 [16]. We have compiled from the literature and from our own experience a list of stable supports in near- and supercritical water, for both oxidizing and reducing conditions. The range of temperatures covered by Table 12.5 is ca 300–500 C. Carbon supports tend to gasify at higher temperatures. The pH and the presence of salts will also affect the stability (solubility) of the oxide materials. For mild hydrothermal conditions, for example in aqueous phase reforming, other supports such as silica, g-alumina and Fe2O3 may be suitable because their decomposition (dissolution) or phase change is very slow. However, these conditions have not been included in Table 12.5. Some carbon supports were found to be useful in a hydrothermal environment. One of their main advantages is the absence of sintering. Of the different forms of carbonaceous supports, pyrolytic carbon obtained from an ion-exchange resin and activated carbon obtained from coconut shells were both used successfully in Table 12.5 Useful supports for hydrothermal gasification and partial oxidation in the range 300–500 C. Alloys (PtRu, NiSn, etc.) have not been considered.
Support
Oxidizing
Reducing
References
a-Al2O3
(þ)
[60], [67], [85]
HfO2 MnO MnO2 Nb2O5 Ta2O5 (orthorhombic) TiO2 (rutile) UO2 ZrO2 (monoclinic) ZrO2 (stabilized by yttria or magnesia) C (graphitic) C, activated (coconut shell)
þ ? (þ) þ þ þ þ þ þ (neutral pH)
þ (at neutral to low pH) ? þ – ? ? þ ? þ þ (neutral pH)
[60], [61], [85], [87] [88], [85], [89]
C, pyrolytic
?
(þ) þ (low T) (high T) þ
[23], [60] [23] [90] [69]
(þ)
? ¼ not known; ( þ ) ¼ no conclusive information available.
[85] [86] [16] [23], [61], [85]
j309
j 12 Catalytic Conversion of High-Moisture Biomass to Synthetic Natural Gas in Supercritical Water
310
hydrothermal gasification at low temperatures. Powdered graphite with a high specific surface area was tested as a support for ruthenium in our screening program but this catalyst did not perform well [23]. At high temperatures, carbon supports are prone to gasification. Xu et al. [90] determined a carbon gasification rate at 600 C and 34.5 MPa of 5 105 mol min1 from a bed of 3.6 g of coconut shell activated carbon. This rate corresponds to complete gasification of the 3.6 g after ca 100 h time on-stream, assuming zeroth-order kinetics, which is not acceptable for a technical process with annual operating times of 7000–8000 h. Metals supported on the carbon support may also catalyze the gasification of their own support, in particular at the metal–support interface. Our assessment of this self-gasification led to the hypothesis that this might become important at high levels of conversion. As long as the metal sites are busy gasifying the feed molecules, they do not seem to attack their own support. The support materials listed in Table 12.5 are insoluble in near- and supercritical water and do not change their phase, that is, the most stable phase under hydrothermal conditions was selected. For alumina, this is the a phase (corundum), also obtained at high temperatures in a gaseous atmosphere. a-Alumina does not have a high specific surface area and is therefore of limited use for achieving finely dispersed metal clusters on the support. Furthermore, it should be kept in mind that a-alumina will dissolve slightly in a neutral [85] and heavily in an alkaline hydrothermal medium. At temperatures below ca 360 C, diaspore becomes the stable alumina phase (see Figure 12.13). HfO2 and UO2 were both included in Table 12.5 because they do not dissolve in hot compressed water and both were reported to form stable phases under oxidative conditions [16]. HfO2 was used as a support for Pt under oxidizing conditions [85]. UO2 has not yet been investigated. Also, their stability under reducing conditions is not known. MnO2 was found to be reduced to MnO in a hydrothermal environment, but it was stable under oxidizing conditions. Likewise, MnO is stable under reducing conditions but will probably be oxidized to MnO2 in an oxidizing environment (e.g. in SCWO). Ta2O5 is a very stable compound forming a hydrothermally stable orthorhombic phase. However, its behavior under reducing conditions has not been investigated so far. Ding et al. [16] reported a solubility of 30 ppm for this oxide at 500 C and 103.4 MPa. It might therefore be used at the lower temperatures and pressures of hydrothermal gasification for the production of methane. Titania is available in two major modifications, anatase and rutile. Some commercial supports, such as Degussas P-25, are mixtures of 20% rutile and 80% anatase. Anatase was reported to be converted to rutile at 350 C and 20 MPa [91]. Elliott et al. used 3%Ru on rutile (Degussa) successfully over 19 weeks on-stream for gasifying a solution with 10 wt% phenol [61]. Others have claimed that anatase is the stable hydrothermal form below 600 C [16]. We have used a 1%Ru on rutile catalyst from Degussa to gasify 30 wt% ethanol at 345–442 C and 30 MPa and high space velocities (WHSV) up to 36 gorg gcat1 h1. Turnover frequencies calculated on a carbon basis suggest that a progressive deactivation had taken place. The values in Table 12.6 are listed in the same order as the experimental conditions were changed. After a high initial turnover frequency (TOF) of 16 mol C mol1 s1 at
12.4 The Role of Heterogeneous Catalysis Table 12.6 Turnover frequencies (TOF), first-order rate constant k
and gas composition for the gasification of ethanol on 1% Ru/TiO2.
Tavg ( C)
WHSV (gorg gcat1 h1)
TOF (mol C mol Ru1 s1)
442 345 400
36 36 18
16.2 5.9 4.5
k (s1)
Ygas (L gas g EtOH1)
CH4 (vol.%)
H2 (vol.%)
CO2 (vol.%)
CO (vol.%)
1.91 0.13 0.20
1.09 0.11 0.58
65 66 60
13 19 20
22 9 17
1 4 4
442 C average catalyst bed temperature, the TOF decreased markedly at 345 C. On increasing the temperature to 400 C by keeping the residence time nearly constant (the decrease in fluid density was compensated by decreasing the WHSV), the TOF decreased even further. Also, the gas composition at 400 C was further away from equilibrium than at 345 C and the same residence time. Interestingly, the first-order rate constant k, calculated from the carbon conversion and the residence time under the operating conditions, follows the same trend as the TOF values, although no information about the Ru surface sites was used for calculating k. As the k values deviate significantly from the expected Arrhenius behavior, we interpret this as a further hint that some deactivation took place. The axial temperature profiles measured during the different runs reveal the probable cause of this deactivation (Figure 12.14).
Figure 12.14 Axial temperature profiles measured during the gasification of ethanol on 1% Ru/TiO2.
j311
j 12 Catalytic Conversion of High-Moisture Biomass to Synthetic Natural Gas in Supercritical Water
312
The first run with a reactor inlet temperature of ca 430 C led to a hot-spot of ca 490 C, measured after 1 h of time on-stream. After a further 18 h, the temperature profile had flattened with the hot-spot located further downstream at ca 475 C. These hot-spots were created by the high feed concentration of 30 wt% and the high space velocities. Note that the catalyst bed was diluted 1 : 1 with a-alumina. The moving hotspot clearly indicates a change in the catalysts activity. We believe that these hot-spots damaged the catalyst irreversibly. It is known that TiO2 rearranges in the presence of hydrogen and forms layers that partially cover the ruthenium crystallites above ca 300 C [59]. Taking into account that the activity at 400 C was still significant and that the slope of total gas production versus time on-stream was linear at all temperatures [21], the observed decrease in activity may be attributed to the catalyst accommodating to the hydrothermal environment rather than to a progressive deactivation. Unfortunately, no post-reaction analyses of the catalyst are available to verify this hypothesis and to rule out sintering. Remarkable also is the relatively high concentration of CO. Similar values were measured with other catalysts only when they were almost completely deactivated. ZrO2 in its monoclinic form is hydrothermally stable at neutral pH and a good catalyst support. Alternatively, other forms of ZrO2 may be stabilized by dopants such as yttria and magnesia [88, 89]. We have tested a 2%Ru/ZrO2 (monoclinic) catalyst prepared in-house for gasifying algae such as Spirulina platensis with good results. Interestingly, this catalyst exhibited an extraordinary tolerance towards sulfate. The yield of methane from glycerol in batch experiments was not affected by adding K2SO4 (0.06 M) to the feed solution, whereas 2%Ru/C almost completely deactivated under the same conditions [87]. 12.4.8 Deactivation Mechanisms in a Hydrothermal Environment
Catalysis in a hydrothermal environment is subject to the same modes of deactivation as in gas-phase and liquid-phase heterogeneous catalysis. However, some effects have a different magnitude due to the properties of hot compressed water and the high partial pressures. In Table 12.7 are compiled reports on catalyst deactivation in near- and supercritical water. 12.4.8.1 Coke Formation A common mode of deactivation in steam reforming of hydrocarbons is coke formation [30]. Coking can be reduced by using higher steam-to-carbon (S/C) ratios. Hydrothermal gasification takes place in a large excess of water. A concentrated biomass slurry with 20 wt% dry matter corresponds to an S/C ratio of ca 5–7, depending on the carbon content of the dry matter. At these high S/C ratios, coking rates are very low. The high partial pressure of water in hydrothermal gasification leads to high rates of carbon gasification at higher temperatures only, as carbon is fairly unreactive at temperatures below ca 600 C without suitable catalysts, for example alkali metal salts. An important reason why coke is usually not observed in a hydrothermal environment even in the absence of a catalyst is the high solubility of
12.4 The Role of Heterogeneous Catalysis Table 12.7 Modes of deactivation in a hydrothermal environment.
Mode
References
Comments
Dissolution of active metal
[92]
Reduction of SBET
[60] [92] [61] [23] [60] [20] [91]
Pronounced in the presence of salts or at high T Pronounced due to phase changes
Coke deposition Fouling Poisoning
Sintering of active metal (reduction of SMetal)
Oxidation or reduction of active metal
Dissolution, attrition or gasification of the support
[91] [61] [93] [23] [94] [60] [95] [61] [23]
[60] [16] [86] [60]
[90] [96] [89] [88] [97] [61] [98] YSZ: yttria-stabilized zirconia.
Skeletal Ni (dissolution and recrystallization) Severe for Pt and Pd on g-alumina Skeletal Ni: could not be related to deactivation No clear evidence as to cause of deactivation Salt precipitation is most serious problem Poisoning in the presence of sulfate H3PO2 (inconclusive) Poisoning in the presence of sulfuric acid Poisoning in the presence of sulfate Poison: various S compounds Pronounced for Ni (supported and skeletal)
More pronounced due to dissolution and re-deposition of active metal/metal oxide Reported for Co, Fe, Cr, Mo, W, Zn MnO2 Pronounced due to high p(H2O) Pronounced for SiO2; slow gasification of C support (catalyzed by supported Ru) Gasification of coconut carbon at 600 C
Oxidizing conditions, pH 1.3, 465 C, 25 MPa: SiC, B4C, TiB2, Y2O3, BN, YSZ, Si3N4, AlN YSZ in acidic and alkaline solutions at 600 C and 100 MPa Ni/MgO was transformed to Mg(OH)2 Loss of catalyst dust due to attrition Hydrolysis of asbestos fibers
j313
j 12 Catalytic Conversion of High-Moisture Biomass to Synthetic Natural Gas in Supercritical Water
314
coke precursors in near- and supercritical water, permitting effective gasification of these compounds before they can form coke. Coke formation was also found to be affected by the acidity of the support and by the type of active metals. In RostrupNielsens study, coking did not occur on supported Ni–Cu alloy, Co, Pt, Ru and Rh, all on alumina, even at low S/C ratios of 0.3–0.6 [30]. 12.4.8.2 Sintering Sintering in gas-phase processes is usually observed at high temperatures where the mobility of the surface atoms becomes significant. An empirical rule was found by Tammann, according to which sintering should be expected at temperatures above 0.5 times the melting point (in Kelvin) of the active metal [63]. For bulk nickel, this threshold is at 590 C and for bulk ruthenium it is at 880 C. In a hydrothermal environment, dissolution and recrystallization of metals as oxides, hydroxides or other complexes is more likely than pure thermal sintering due to the solvation properties of near- and supercritical water and the presence of ligands. Sintering of the support may proceed according to the same mechanism. For carbon supports, sintering is not an issue but rather its gasification (see Section 12.4.7). 12.4.8.3 Poisoning Sulfur poisoning of nickel catalysts has been the subject of many studies [30]. For other metals, sulfur is in most cases also considered a poison. Under the conditions of steam reforming, the actual poison is H2S, as organic sulfur compounds such as thiophene are known to decompose to H2S. In a hydrothermal environment, aqueous chemistry of the sulfur compounds must also be taken into account. In the absence of oxygen, some of these reactions include the following: 2 þ 4H2 O SO2 4 þ 4H2 ! S
ð12:16Þ
2 SO2 4 þ H2 ! SO3 þ H2 O
ð12:17Þ
2 þ 3H2 O SO2 3 þ 3H2 ! S
ð12:18Þ
þ 2 2SO2 3 þ 2H2 þ 2H ! S2 O3 þ 3H2 O
ð12:19Þ
2 S2 O2 3 ! 1=8S8 ðsÞ þ SO3
ð12:20Þ
In addition to sulfide, also sulfate, sulfite, thiosulfate and elemental sulfur may coexist under hydrothermal conditions, in addition to their partly or fully protonated species. The reduction of sulfate and sulfite may occur with hydrogen or with organic compounds, for example aldehydes or alcohols, as reducing agents. Another possibility is that sulfate or sulfite take part in the redox cycle as oxidants for the metal oxide instead of water. Equation 12.14b would then be replaced by 2 4Mðþ n2ÞO þ SO2 4 ! 4Mðþ nÞO2 þ S
ð12:14cÞ
Poisoning by H2S involves the two lone pairs of the sulfur atom overlapping with the empty d-orbitals of the reduced metal atom. In the redox cycle proposed for hydro-
12.5 Continuous Catalytic Hydrothermal Process for the Production of Methane
thermal reactions, ionic metal species play an important role. Therefore, sulfate, sulfite or sulfide may form stable complexes with oxidized metal sites, leading to deactivation. If elemental sulfur is formed, it would deactivate the catalyst by precipitating and fouling the surface, similar to precipitating salts. No conclusive results have been reported so far regarding poisoning by sulfur compounds in a hydrothermal environment. Some workers have reported poisoning by sulfate [23, 91, 94], but this diagnosis was obtained by post mortem and ex situ analyses using XPS and SEM–EDX. Oxidation of the adsorbed sulfur species by air during transfer of the catalyst from the reactor to the analytical equipment must be expected and hence no conclusive information on the species present under the reaction conditions is available from these studies. This requires an in situ investigation under the operating conditions. We are exploring X-ray absorption spectroscopy (XAS) with this aim.
12.5 Continuous Catalytic Hydrothermal Process for the Production of Methane
The idea of using near- and supercritical water to gasify biomass was proposed by Modell at MIT in the 1970s. He observed that glucose and wood flour could be gasified in supercritical water without the formation of tars and char, but the yield of gaseous products was low. Modell also studied the effect of adding different catalysts, with limited success [99]. 12.5.1 Overview of Processes
Pioneering work in developing a catalytic hydrothermal gasification process for the production of a methane-rich gas from wet organic waste streams was carried out at the Pacific Northwest National Laboratory in the USA and resulted in the TEES (Thermochemical Environmental Energy System) process [60, 100]. Typical TEES conditions are in the subcritical region (350 C, 20 MPa). The preferred catalysts are a stabilized nickel catalyst from BASF, Ru/C and Ru/TiO2 [61]. Other related hydrothermal gasification processes, targeted at the production of hydrogen, have been proposed by the Forschungszentrum Karlsruhe in Germany [101], by the Biomass Technology Group and the University of Twente in The Netherlands [102] and by General Atomics in the USA [52]. We have not included a more detailed discussion of these processes because they do not involve a heterogeneous catalyst, which is the main scope of this chapter. 12.5.2 PSIs Catalytic Hydrothermal Gasification Process
Our research work at the Paul Scherrer Institut started in 2002 and led to the development of a catalytic process for the production of a methane-rich gas from biomass and organic waste streams [103]. A simplified flowsheet is shown in Figure 12.15.
j315
j 12 Catalytic Conversion of High-Moisture Biomass to Synthetic Natural Gas in Supercritical Water
316
Figure 12.15 Schematic of PSIs catalytic hydrothermal biomass gasification process.
A pumpable slurry with 15–30 wt% dry matter is continuously pumped and pressurized to the process pressure of about 30 MPa. The slurry is heated to ca 350 C by recovering heat from the hot reactor effluent stream, thereby the biomass constituents undergo hydrolysis to lower molecular weight compounds (e.g. glucose, phenols and carboxylic acids). In an externally heated separator vessel, the temperature of the stream is raised above the mixtures critical point to ca 450 C. Salts precipitate instantaneously above the critical point and are continuously recovered, together with other solids, in a separator vessel. The stream free of most salts and solids then passes over a catalyst bed where the organics are gasified completely to methane, CO2 and hydrogen. CO is formed only in trace amounts. After cooling the reactor effluent, the liquid water phase is separated from the gases. Conventional gas separation such as pressure swing adsorption (PSA) or pressurized water scrubbing to remove CO2 yields a fuel gas containing mainly CH4 and small amounts of H2. A fraction of the gas is combusted in an atmospheric burner and the hot flue gases are used to heat the salt separator to the process temperature. The salt brine from the separator vessel is processed further to remove heavy metals selectively if required. Additional process heat may be recovered from the hot salt brine. The net thermal process efficiency was calculated to be in the range 60–70%, depending mainly on the composition of the biomass. 12.5.2.1 Continuous Salt Precipitation and Separation The main feature of the PSI process is the separation of salts and insoluble oxides from the supercritical fluid before the catalytic reactor. Because both type I and II salts will precipitate under supercritical conditions, either as liquid or as solid phase, once saturation is reached it is important to deal with them in order to avoid plugging or poisoning of the catalyst. The process separates most of the salts from the supercritical fluid before it reaches the catalyst. This is achieved in a separator by a controlled
12.5 Continuous Catalytic Hydrothermal Process for the Production of Methane
precipitation: the salts are denser than the supercritical fluid and can be separated by gravity. Our current separator design consists of a high-pressure vessel with a dip tube for the feed stream entering axially at the top and leaving by an exit port in the radial direction at the top. The accumulated salt solution is withdrawn as brine from the bottom. We have used neutron radiography to study the phenomena inside the salt separator [104]. Neutron radiography is similar to imaging by X-rays but employs a beam of cold neutrons instead of X-rays. By choosing suitable materials of construction and experimental conditions, one can image the inside of a high-pressure vessel without having to use windows, which are likely to be covered by salt deposits. We have studied the precipitation and continuous removal of various salts from supercritical water. As an example, Figure 12.16 shows the result obtained when feeding 1 kg h1 of a mixture of KH2PO4 and K2HPO4 (both type I) at a total concentration of 0.1 M. A constant brine flow of ca 0.16 kg h1 was withdrawn from the bottom of the salt separator. The total pressure of the system was maintained at 30 MPa. The green curve represents the electrical conductivity of the brine stream and the orange curve that of the effluent to the reactor. The blue horizontal line represents the conductivity of the feed solution. At temperatures of the fluid inside the salt separator below the critical point, no separation occurred. When the temperature of the fluid inside the salt separator was raised to ca 376 C, precipitation and accumulation of the salts at the bottom of the separator started, leading to an increase in the brine conductivity. A further increase in the fluid temperature resulted in an increased recovery of the salts in the brine and in an effluent stream depleted in salts.
Figure 12.16 Continuous precipitation and removal of a mixture of KH2PO4 and K2HPO4 from supercritical water. Electrical conductivity of salt brine (0.16 kg h1; green line) and separator effluent (0.84 kg h1; orange line) for different fluid temperatures. A, 366 C; B, 376 C; C,
388 C; D, 398 C; E, 403 C; F, 406 C. Total pressure was maintained at 30 MPa. The horizontal blue line is the conductivity of the feed solution. The temperatures at the top represent heater setpoints.
j317
j 12 Catalytic Conversion of High-Moisture Biomass to Synthetic Natural Gas in Supercritical Water
318
Figure 12.17 PSIs laboratory-scale process development unit KONTI-2. A, feed container; B, high-pressure feed pump; C, preheater; D, salt separator/superheater; E, fixed-bed reactor; F, effluent coolers; G, pressure letdown and gas–liquid phase separation; H, on-line gas analysis (GC).
12.5.2.2 Status The PSI process has been demonstrated in a laboratory-scale continuous process development unit with a maximum feed rate of 1 kg h1, depicted in Figure 12.17. The main features of the process, that is, preheating/hydrolysis, superheating/salt separation and catalytic conversion to a methane-rich gas, have been studied with model solutions (e.g. ethanol, glycerol, salt mixtures) and with liquid biomass (wood pyrolysis condensate, micro algae slurries).
12.6 Summary and Conclusions
Methane produced from waste biomass is a renewable and clean biofuel that can be distributed using the existing natural gas infrastructure. It can be used for heat and power generation and as a transportation fuel. High-moisture biomass is a relatively untapped resource with a significant energetic potential and attractive costs. However, new technologies are needed for converting high-moisture biomass efficiently into methane and recovering the nutrients for use as a fertilizer. Gasification of the biomass in a hydrothermal environment is an emerging technology that offers many advantages over gas-phase conversion processes or anaerobic digestion. Heterogeneous catalysis is the key to a successful hydrothermal gasification process for the synthesis of methane. Only a few metals, including Ru, Ni, Rh and Pt, are useful under these conditions. Pd and Co catalysts might also be suitable but conclusive data are lacking. Alloying is another approach that holds promise to yield active and stable catalysts. The choice of hydrothermally stable supports is limited to some insoluble
12.7 Outlook for Future Developments
oxides and carbon. Some of these oxides have not yet been tested as catalyst supports (e.g. Nb2O5, Ta2O5 and UO2) and might prove useful. The mechanism for the gasification of the organic compounds to CO and H2 is likely to follow a Mars– van-Krevelen redox cycle with two oxides of the catalytic metal involved. The strongest evidence for such a mechanism was found for RuO2, but specific in situ studies are needed for corroborating this hypothesis and determining the actual oxidation states involved in the mechanism. Deactivation in hydrothermal gasification follows the same modes as in gas-phase and liquid-phase catalysis. Coke deposition is not a primary cause of deactivation due to the high partial pressure of water and the high solubility of coke precursors in near- and supercritical water. Salts play a crucial role in catalyst deactivation. Sulfate was found to be a strong poison for Ru catalysts, but the actual poison might be sulfide, formed by reduction of the sulfate with hydrogen or organic compounds. Based on this knowledge, a continuous catalytic hydrothermal gasification process is under development at PSI featuring continuous on-line salt precipitation and removal before the catalytic reactor.
12.7 Outlook for Future Developments
In addition to producing biomethane, the need for green electricity is growing fast. Power plants based on coal combustion are expected to reach 50–53% of electrical efficiency by drying the lignite and by utilizing supercritical steam cycles up to 700 C and 35 MPa in the near future. This gain in efficiency will be partly compensated by the measures for capturing the CO2. Electricity production from high-moisture biomass via anaerobic fermentation and combustion of the biogas in a gas engine is available at a much smaller scale (a few megawatts of thermal input) and has much lower electrical efficiencies in the order of 10–15%. With the development of durable and efficient fuel cells, this number will increase in the future. Integration of a high-temperature fuel cell [molten carbonate (MCFC) or solid oxide (SOFC) types] with the hydrothermal gasification process could result in high electrical efficiencies up to ca 43% even for relatively small plants in the megawatt range. The heat necessary for reaching supercritical conditions of 400 C would then be provided by the waste heat of the fuel cell, available at ca 600 and 900 C for the MCFC and SOFC, respectively. No product gas will have to be burned to provide this heat and therefore higher efficiencies for the hydrothermal conversion of the biomass into methane in the range 80–85% will be possible. Combined with an electrical efficiency of ca 50% for the fuel cell with methane as fuel, an overall efficiency of 40–43% would result. 12.7.1 A Self-sustaining Biomass Vision (SunCHem)
A visionary concept for the self-sustaining bioenergy production using micro algae has been proposed by researchers in Japan [105, 106]. Micro algae are cultivated,
j319
j 12 Catalytic Conversion of High-Moisture Biomass to Synthetic Natural Gas in Supercritical Water
320
collected and gasified under hydrothermal conditions to yield a methane-rich gas. The CO2 and the aqueous effluent with the nutrients are separated and recycled to the algae. We have taken this concept one step further in the SunCHem process [107]. In this process, the nutrients are separated as a concentrated brine and recycled to the algae. Before gasifying the algae, valuable chemicals may be extracted and purified. The makeup stream of CO2, corresponding to the net molar flow of methane that is produced, may be taken from exhaust fumes of fossil power plants or eventually from the atmosphere. One advantage of such a self-sustaining bioenergy system is the lack of transportation of the biomass to the plant. Furthermore, the plant can be located in desert areas not suitable for agricultural production and thus not competing with food production.
Acknowledgments
The following colleagues have contributed to the work presented in this chapter: Dr Stefan Rabe, Dr Maurice Waldner, Dr Ashaki Rouff, Martin Schubert, Martin Brandenberger, Johann Regler, Erich De Boni, Thanh-Binh Truong, Dr Samuel Stucki, Dr Christian Ludwig, Tiina Vuoti, Alwin Frei, Dr Fabio Raimondi, Dr Alexander Wokaun, Andrew Peterson (MIT) and Dr Jefferson Tester (MIT).
References 1 Hazen, R.M., Boctor, N., Brandes, J.A., Cody, G.D., Hemley, R.J., Sharma, A. and Yoder, H.S. Jr, (2002) Journal of Physics: Condensed Matter, 14, 11489–11494. 2 Foustoukos, D.I. and Seyfried, W.E. Jr., (2004) Science, 304, 1002–1005. 3 Tian, G., Yuan, H., Mu, Y., He, C. and Feng, S. (2007) Organic Letters, 9, 2019–2021. 4 Smeets, E.M.W., Faaij, A.P.C., Lewandowski, I.M. and Turkenburg, W.C. (2007) Progress in Energy and Combustion Science, 33, 56–106. 5 Tester, J.W., Drake, E.M., Driscoll, M.J., Golay, M.W. and Peters, W.A. (2005) Sustainable Energy – Choosing Among Options, MIT Press, Cambridge, MA. 6 Vogel, F. (2003) Hydrothermal processes, in Municipal Solid Waste Management (eds C. Ludwig, S. Hellweg and S. Stucki), Springer, Berlin, pp. 204–210. 7 Maurer, M., Schwegler, P. and Larsen, T.A. (2003) Water Science and Technology: A
8
9
10
11
12
13 14
Journal of the International Association on Water Pollution Research, 48, 37–46. Stendahl, K. and J€afverstr€om, S. (2003) Water Science and Technology, 48, 185–191. Schlegel, H.G. (1992) Allgemeine Mikrobiologie, 7th edn Georg Thieme, Stuttgart. Seemann, M.C., Schildhauer, T.J., Biollaz, S.M.A., Stucki, S. and Wokaun, A. (2006) Applied Catalysis A-General, 313, 14–21. Schuster, G., L€offler, G., Weigl, K. and Hofbauer, H. (2001) Bioresource Technology, 77, 71–79. Savage, P.E., Gopalan, S., Mizan, T.I., Martino, C.J. and Brock, E.E. (1995) AIChE Journal, 41, 1723–1778. Kruse, A. and Dinjus, E. (2007) Journal of Supercritical Fluids, 41, 361–379. King, J.W. and Friedrich, J.P. (1990) Journal of Chromatography, 517, 449–458.
References 15 Harvey, A.H., Peskin, A.P. and Klein, S.A. (1996) NIST/ASME Steam v 2.2, NIST Standard Reference Database 10. 16 Ding, Z.Y., Frisch, M.A., Li, L. and Gloyna, E.F. (1996) Industrial & Engineering Chemistry Research, 35, 3257–3279. 17 Valyashko, V.M. (1997) Pure and Applied Chemistry, 69, 2271–2280. 18 Rumyantsev, V.N. (1977) Soviet Physics, Crystallography, 22, 610–613. 19 Armellini, F.J. (1993) PhD Thesis, Department of Chemical Engineering, Massachusetts Institute of Technology. 20 Waldner, M.H. and Vogel, F. (2005) Industrial & Engineering Chemistry Research, 44, 4543–4551. 21 Vogel, F., Waldner, M.H. and De Boni, E. (2005) in Proceedings of Joint 20th AIRAPT–43rd EHPRG, Karlsruhe. 22 Vogel, F., Waldner, M.H., Rouff, A.A. and Rabe, S. (2007) Green Chemistry, 9, 616–619. 23 Waldner, M.H., Krumeich, F. and Vogel, F. (2007) Journal of Supercritical Fluids, 43, 91–105. 24 Craig, K.R. and Mann, M.K. (1996) Cost and Performance Analysis of Biomass-based Integrated Gasification Combined-cycle (BIGCC) Power Systems, NREL/TP-43021657, October. 25 Bain, R.L., Dayton, D.C., Carpenter, D.L., Czernik, S.R., Feik, C.J., French, R.J., Magrini-Bair, K.A. and Phillips, S.D. (2005) Industrial & Engineering Chemistry Research, 44, 7945–7956. 26 Bridgwater, A.V. (2003) Chemical Engineering Journal, 91, 87–102. 27 Devi, L., Ptasinksi, K.J. and Janssen, F.J.J.G. (2003) Biomass Bioenergy, 24, 125–140. 28 Di Blasi, C. (2008) Progress in Energy and Combustion Science, 34, 47–90. 29 Sutton, D., Kelleher, B. and Ross, J.R.H. (2001) Fuel Processing Technology, 73, 155–173. 30 Rostrup-Nielsen, J.R. (1975) Steam Reforming Catalysts, Teknisk Forlag, Copenhagen.
31 Comisar, C.M., Hunter, S.E., Walton, A. and Savage, P.E. (2008) Industrial & Engineering Chemistry Research, 47, 577–584. 32 Maiella, P.G. and Brill, T.B. (1996) The Journal of Physical Chemistry, 100, 14352–14355. 33 Kersten, S.R.A., Potic, B., Prins, W. and Van Swaaij, W.P.M. (2006) Industrial & Engineering Chemistry Research, 45, 4169–4177. 34 Penninger, J.M.L. and Rep, M. (2006) International Journal of Hydrogen Energy, 31, 1597–1606. 35 Kruse, A. and Dinjus, E. (2005) Zeitschrift €r Physikalische Chemie-International fu Journal of Research in Physical Chemistry & Chemical Physics, 219, 341–366. 36 DiNaro, J.L., Howard, J.B., Green, W.H., Tester, J.W. and Bozzelli, J.W. (2000) Journal of Physical Chemistry A, 104, 10576–10586. 37 Ederer, H.J., Kruse, A., Mas, C. and Ebert, K.H. (1999) Journal of Supercritical Fluids, 15, 191–204. 38 B€ uhler, W., Dinjus, E., Ederer, H.J., Kruse, A. and Mas, C. (2002) Journal of Supercritical Fluids, 22, 37–53. 39 Vogel, F., Harf, J., Hug, A. and Rudolf von Rohr, P. (1999) Environmental Progress, 18, 7–13. 40 Kruse, A. and Gawlik, A. (2003) Industrial & Engineering Chemistry Research, 42, 267–279. 41 Kharaka, Y.K., Carothers, W.W. and Rosenbauer, R.J. (1983) Geochimica et Cosmochimica Acta, 47, 397–402. 42 Palmer, D.A. and Drummond, S.E. (1986) Geochimica et Cosmochimica Acta, 50, 813–823. 43 Di Blasi, C., Branca, C., Galgano, A., Meier, D., Brodzinski, I. and Malmros, O. (2007) Biomass Bioenergy, 31, 802–811. 44 Nagai, Y., Morooka, S., Matubayasi, N. and Nakahara, M. (2004) Journal of Physical Chemistry A, 108, 11635–11643. 45 Wakai, C., Yoshida, K., Tsujino, Y., Matubayasi, N. and Nakahara, M. (2004) Chemistry Letters, 33, 572–573.
j321
j 12 Catalytic Conversion of High-Moisture Biomass to Synthetic Natural Gas in Supercritical Water
322
46 Yasaka, Y., Yoshida, K., Wakai, C., Matubayasi, N. and Nakahara, M. (2006) Journal of Physical Chemistry A, 110, 11082–11090. 47 Yu, J. and Savage, P.E. (1998) Industrial & Engineering Chemistry Research, 37, 2–10. 48 Br€oll, D., Kaul, C., Kr€amer, A., Kammer, P., Richter, T., Jung, M., Vogel, H. and Zehner, P. (1999) Angewandte Chemie-International Edition, 38, 2998–3014. 49 DJesus, P., Boukis, N., KraushaarCzarnetzki, B. and Dinjus, E. (2006) Fuel, 85, 1032–1038. 50 DiLeo, G.J., Neff, M.E. and Savage, P.E. (2007) Energy Fuels, 21, 2340–2345. 51 Grumett, P. (2003) Platinum Metal Review, 47, 163–166. 52 Hong, G.T. and Spritzer, M.H. (2002) Proceedings of the 2002 US DOE Hydrogen Program Review, NREL/CP610-32405. 53 Vostrikov, A.A., Psarov, S.A., Dubov, D.Y., Fedyaeva, O.N. and Sokol, M.Y. (2007) Energy Fuels, 21, 2840–2845. 54 Cheng, L., Zhang, R. and Bi, J. (2004) Fuel Processing Technology, 85, 921–932. 55 Gadhe, J.B. and Gupta, R.B. (2005) Industrial & Engineering Chemistry Research, 44, 4577–4585. 56 Minowa, T. and Fang, Z. (1998) Journal of Chemical Engineering of Japan, 31, 488–491. 57 Dao, H.Q. and Pruchnik, F. (1987) Polish Journal of Chemistry, 61, 537–545. 58 Somorjai, G.A. (1994) Introduction to Surface Chemistry and Catalysis, John Wiley & Sons, Inc., New York. 59 Waldner, M.H. (2007) PhD Thesis, ETH Z€ urich. 60 Elliott, D.C., Sealock, L.J., Jr. and Baker, E.G. (1993) Industrial & Engineering Chemistry Research, 32, 1542–1548. 61 Elliott, D.C., Hart, T.R. and Neuenschwander, G.G. (2006) Industrial & Engineering Chemistry Research, 45, 3776–3781. 62 Vannice, M.A. (1975) Journal of Catalysis, 37, 449–461.
63 Satterfield, C.N. (1991) Heterogeneous Catalysis in Industrial Practice, 2nd edn McGraw-Hill, New York. 64 Jiang, C.J., Trimm, D.L. and Wainwright, M.S. (1993) Applied Catalysis A-General, 93, 245–255. 65 Furusawa, T., Sato, T., Sugito, H., Miura, Y., Ishiyama, Y., Sato, M., Itoh, N. and Suzuki, N. (2007) International Journal of Hydrogen Energy, 32, 699–704. 66 Osada, M., Sato, O., Watanabe, M., Arai, K. and Shirai, M. (2006) Energy Fuels, 20, 930–935. 67 Vogel, F. and Hildebrand, F. (2002) in Proceedings of the 4th International Symposium on High Pressure Process Technology and Chemical Engineering, EFCE, Venice, 22–25 September 2002, (ed. A. Bertucco), Chem Eng Trans. 2, 771–777. 68 Minowa, T. and Ogi, T. (1998) Catalysis Today, 45, 411–416. 69 Nakagawa, H., Namba, A., B€ohlmann, M. and Miura, K. (2004) Fuel, 83, 719–725. 70 Huber, G.W., Shabaker, J.W. and Dumesic, J.A. (2003) Science, 300, 2075–2077. 71 Huber, G.W., Shabaker, J.W., Evans, S.T. and Dumesic, J.A. (2006) Applied Catalysis B-Environmental, 62, 226–235. 72 Sato, T., Osada, M., Watanabe, M., Shirai, M. and Arai, K. (2003) Industrial & Engineering Chemistry Research, 42, 4277–4282. 73 Davda, R.R. and Dumesic, J.A. (2003) Angewandte Chemie-International Edition, 42, 4068–4071. 74 Tanksale, A., Wong, Y., Beltramini, J.N. and Lu, G.Q. (2007) International Journal of Hydrogen Energy, 33, 717–724. 75 Bhargava, S.K., Tardio, J., Prasad, J., F€oger, K., Akolekar, D.B. and Grocott, S.C. (2006) Industrial & Engineering Chemistry Research, 45, 1221–1258. 76 Levec, J. and Pintar, A. (2007) Catalysis Today, 124, 172–184. 77 Golunski, S.E. (2007) Platinum Metal Review, 51, 162. 78 Kelley, R.D., Candela, G.A., Madey, T.E., Mewbury, D.E. and Schehl, R.R. (1983) Journal of Catalysis, 80, 235–248.
References 79 Elliott, D.C. and Hart, T.R. (2000) US Patent 6 152 975. 80 Sharma, A., Nakagawa, H. and Miura, K. (2006) Fuel, 85, 179–184. 81 Minowa, T. and Inoue, S. (1999) Renewable Energy, 16, 1114–1117. 82 Park, K.C. and Tomiyasu, H. (2003) Chemical Communications, 694–695. 83 Pourbaix, M. (1974) Atlas of Electrochemical Equilibria in Aqueous Solutions, 2nd edn National Association of Corrosion Engineers, Houston, TX. 84 Wefers, K. and Misra, C. (1987) Alcoa Technical Paper No. 19, revised, Alcoa Laboratories, Pittsburgh, PA. 85 Frisch, M.A., Li, L. and Gloyna, E.F. (1994) in Proceedings of the 49th Purdue Industrial Waste Conference, Lewis Publishers, Chelsea, MI. 86 Armbruster, U., Martin, A. and Krepel, A. (2001) Applied Catalysis B-Environmental, 31, 263–273. 87 Brandenberger, M. (2007) Masters Thesis, Ecole Polytechnique Federale de Lausanne (EPFL). 88 Yoshimura, M., Hiuga, T. and Somiya, S. (1986) Journal of the American Ceramic Society, 69, 583–584. 89 Boukis, N., Claussen, N., Ebert, K., Janssen, R. and Schacht, M. (1997) Journal of the European Ceramic Society, 17, 71–76. 90 Xu, X., Matsumura, Y., Stenberg, J. and Antal, M.J., Jr. (1996) Industrial & Engineering Chemistry Research, 35, 2522–2530. 91 Elliott, D.C., Neuenschwander, G.G., Hart, T.R., Butner, R.S., Zacher, A.H., Engelhard, M.H., Young, J.S. and McCready, D.E. (2004) Industrial & Engineering Chemistry Research, 43, 1999–2004. 92 Yu, J. and Savage, P.E. (2001) Applied Catalysis B-Environmental, 31, 123–132. 93 Valenzuela, M.B., Jones, C.W. and Agrawal, P.K. (2006) Energy Fuels, 20, 1744–1752. 94 Osada, M., Hiyoshi, N., Sato, O., Arai, K. and Shirai, M. (2007) Energy Fuels, 21, 1400–1405.
95 Elliott, D.C., Phelps, M.R., Sealock, L.J., Jr. and Baker, E.G. (1994) Industrial & Engineering Chemistry Research, 33, 566–574. 96 Matsumura, Y., Xu, X. and Antal, M.J., Jr. (1997) Carbon, 35, 819–824. 97 Furusawa, T., Sato, T., Saito, M., Ishiyama, Y., Sato, M., Itoh, N. and Suzuki, N. (2007) Applied Catalysis A-General, 327, 300–310. 98 Sigon, F., Servida, A. and Grassi, S. (2005) Patent WO 2005/000490 A2. 99 Modell, M. (1985) Gasification and liquefaction of forest products in supercritical water, in Fundamentals of Thermochemical Biomass Conversion (eds R.P. Overend, T.A. Milne and L.K., Mudgel Jr.), Elsevier, Amsterdam, pp. 95–119. 100 Sealock, L.J., Jr. and Elliott, D.C. (1991) Method for the catalytic conversion of lignocellulosic materials, US Patent 5 019 135. 101 Boukis, N., Galla, U., DJesus, P., M€ uller, H. and Dinjus, E. (17–21 October 2005) in Proceedings of the 14th European Biomass Conference, Paris, pp. 964–967. 102 Matsumura, Y., Minowa, T., Potic, B., Kersten, S.R.A., Prins, W., Van Swaaij, W.P.N., van de Beld, B., Elliott, D.C., Neuenschwander, G.G., Kruse, A. and Antal, M.J. Jr. (2005) Biomass Bioenergy, 29, 269–292. 103 Vogel, F., Waldner, M.H., Truong, T.-B., De Boni, E. and Stucki, S. (2005). PCT 05021601.9/EP 05021601. 104 Peterson, A.A., Vontobel, P., Vogel, F. and Tester, J.W. (2008) Journal of Supercritical Fluids, 43, 490–499. 105 Minowa, T. and Sawayama, S. (1999) Fuel, 78, 1213–1215. 106 Tsukahara, K., Kimura, T., Minowa, T., Sawayama, S., Yagishita, T., Inoue, S., Hanaoka, T., Usui, Y. and Ogi, T. (2001) Journal of Bioscience and Bioengineering, 91, 311–313.
j323
j 12 Catalytic Conversion of High-Moisture Biomass to Synthetic Natural Gas in Supercritical Water
324
107 Haiduc, A.G., Brandenberger, M., Suquet, S., Ludwig, C., Vogel, F., Bernier-Latmani, R. and Stucki, S. (2007) in Proceedings of R07 World Congress – Recovery of Materials and Energy for Resource Efficiency, Davos, Switzerland.
108 Arita, T., Nakahara, K., Nagami, K. and Kajimoto, O. (2003) Tetrahedron Letters, 44, 1083–1086. 109 Reagan, M.T. (2000) PhD Thesis, Department of Chemical Engineering, Massachusetts Institute of Technology.