Ind. Eng. Chem. Res. 1987,26, 2165-2180
2165
REVIEWS
Complete Catalytic Oxidation of Volatile Organicst James J. Spivey Research Triangle Institute, Research Triangle Park, North Carolina 27709
Heterogeneous catalytic oxidation of organic compounds is a n important and intensely studied area. However, most reported research deals with the partial oxidation of petrochemical feedstocks to make products of economic value (e.g., ethylene oxide from ethylene), automotive exhaust catalysts, or CO oxidation. Complete (or “deep”) catalytic oxidation of low molecular weight volatile organic compounds (VOCs) in air has received relatively little attention. This review of heterogeneous catalytic oxidation focuses on its application t o control of VOCs at operating conditions typical of field applications. T h e parameters for this review are low t o moderate temperatures (25-400 “C), atmospheric pressure, high space velocity (103-105 h-l), and low organic reactant concentration (roughly 102-103 ppm) in air.
Preface The purpose of this review is to examine literature dealing with the heterogeneous catalytic oxidation of volatile organic compounds (VOCs). Emphasis is placed on reviewing the fundamental scientific principles general to all catalytic oxidation reactions and then showing how reported work has been, or may be, applied to the control of VOCs a t conditions of interest. Special attention has been paid to the relatively few scientific studies involving mixtures of VOCs. This is because understanding the behavior of “real” VOC-containing gas streams requires a knowledge of VOC behavior in mixtures compared to the more easily studied single-component behavior. This is extremely important if one component of a gas stream is significantly more toxic than other components, and research on its behavior in mixtures is not available. A review of reported applications of catalytic oxidation for control of VOC-type compounds is also included. The purpose is to show the types of real streams and the gross performance of catalytic oxidation systems in actual practice. Several general conclusions about the use of this technology are as follows. (1)Applications of catalytic oxidation for VOC control have primarily involved direct transfer of technology from related applications, such as the use of an automotive catalyst downstream of a burner. Little has been done to develop catalysts specifically for this need. (2) The use of catalytic oxidation as a VOC control technique is more widespread in Europe than in the United States, perhaps because of higher energy costs and more stringent environmental regulations. ’In this paper the periodic group notation in parentheses is in accord with recent actions by IUPAC and ACS nomenclature committees. A and B notation is eliminated because of wide confusion. Groups IA and IIA become groups 1 and 2. T h e d-transition elements comprise groups 3 through 12, and t h e p-block elements comprise groups 13 through 18. (Note that the former Roman number designation is preserved in the last digit of the numbering: e.g., I11 3 and 13.)
-
0888-5885/87/2626-2165$01.50/0
Table I. Approximate Range of Independent Variables for This Review temp, O C 25-400 pressure, atm 1 space velocity, h-’ 103-105 reactant concn, ppm 102-103 reactants halogenated solvents, hydrocarbons, odorous compds (amines, mercaptans, aldehydes), oxygenated hydrocarbons (ketones, esters) oxidant O2in air oxidant concn, vol 70 5-21
(3) Very little has been reported on the systematic scientific study of VOC mixtures. Reported work tends to be either well-characterized research on pure components in air or anecdotal studies of “real” streams where only gross performance is reported. (4) Both metal oxides and supported noble metals are active for many deep oxidations. The mechanism of deep catalytic oxidation involves both lattice and surface oxygen for metal oxides and probably reduced metal sites for supported noble metals. (5) Modeling of the catalytic oxidation of VOCs a t conditions of interest may be complex, especially for mixtures, because of both surface kinetic and mass-transfer effects that can vary with experimental conditions. These effects may result in inhibition or enhancement of the oxidation of a given compound.
Introduction Heterogeneous catalytic oxidation is a well-studied and industrially useful process. Industrial catalytic oxidation of vapors and gases is a very broad field and is dealt with in several texts and review articles (Thomas and Thomas, 1967; Bond, 1974; Grayson and Eckroth, 1980; Satterfield, 1980;Prasad et al., 1984). Catalytic oxidation, both partial and complete, is an important process for such reactions as the partial oxidation of ethylene and propylene, ammoxidation of propylene to acrylonitrile, maleic anhydride production, production of sulfuric acid, and oxidation of hydrocarbons in automotive exhaust catalysts. By far, the majority of oxidation catalysts and catalytic oxidation 0 1987 American Chemical Society
2166 Ind. Eng. Chem. Res., Val. 26, No. 11, 1987
Table 11.
Some
Industrial Catalytic Oxidation Reactions
.,~.--.
tvniral
reaction
-- -+- - +
typical catalyst
C2Hl + '/$02C2H,0
co + '/202 cog can,+ o2 C~H,O+ H,O NH, + C ~ + H 3 /~2 0 , CIH,CN + 3H20 2NH, + 6/20,2NO
so, + '/*O* so3
3H20
+
hydrocarbons O1 CO,. H,O . -(automotive exhaust)
Ag on ol-alumina
hopcalite (admixture of MnOl and CODDW oxides) BiMbiSiO,, Cu20 Te, Ce, Ma oxides on silica; antimony oxide-iron oxide Pt-Rh wire gauze V,0,-K,S0,/Si02 Pt-Pd-Rh on cordierite or mullite
processes have been developed for these industrially important partially oxidized products. However, there are important differences between the commercial processes and the complete catalytic oxidation of VOCs at trace concentrations in air. For instance, in partial oxidation, complete oxidation to CO, and HzO is an undesirable reaction occurring in parallel or in series to the one of interest. Other differences include the reactant concentration and temperature, the type of catalyst used, and the chemical nature of the oxidizable compound. Approximate ranges of the major independent variables of interest in this review are shown in Table I. Specifically excluded from this review are (1) industrial partial oxidation processes (e.g., maleic anhydride or ethylene oxide production) and (2) automotive catalytic oxidation. What principally distinguishes the catalytic oxidation processes considered herein from these two processes are (1)reactant (oxygen and hydrocarbon) concentration and pressure and (2) temperature, respectively. Industrial catalytic oxidation reactions are carried out at high reactant concentrations over a variety of supported metal catalysts. Because most industrial processes operate with well-characterized inlet streams (usually one reactant plus an oxidant), there has been little need to understand the complex processes that may occur in mixtures. As shown in Table 11, these reactions are typically carried out a t temperatures greater than 400 "C, with the exception of ethylene and CO oxidation. Such temperatures are generally required to achieve economical reaction rates (high activity) with minimal byproduct formation (high selectivity). In contrast to these industrial reactions, catalytic oxidation of VOCs in air is carried out a t lower reactant concentrations (often less than 1000 ppm) and with a very large stoichiometric excess of oxygen. Because oxidation is a highly exothermic reaction, and because industrial reactions are generally carried out at high reactant concentrations, these processes are usually net producers of heat. When the reactant concentration is low, the processes are net consumers of heat if the entire gas stream must he heated to an elevated temperature. It may be expensive to heat the entire gas stream to a high temperature (e.& >400 "C) to achieve a high reaction rate. As a result, catalytic oxidation of trace concentrations of VOCs is more economical if the reaction is carried out at lower temperatures. T o do this, a highly active, nonselective catalyst is required. This is in direct contrast to almost all industrial oxidation reactions where selectivity for partially oxidized products is essential. Another consideration for removal of trace contaminants is that, if the oxidation is incomplete, compounds more toxic than the trace contaminant may be formed (e.&, formation of phosgene from incomplete oxidation of vinyl chloride vapors). An additional distinction is the type of chemical compound usually oxidized in commercial chemical production. Industrial catalytic oxidation reactions (see Table 11)
temp, "C ref 260-280 Satterfield, 1980 2W50 Thomas and Thomas, 1967
-650 400-500
Satterfield, 1980 Satterfield, 1980
810-850 42MOO
Satterfield, 1980 Satterfield, 1980 Grayson and Eckroth, 1980; Satterfield, 1980
400-600
I 1 uu L VOCs
Carbon DioxideJ and Air and Water and Others
LVOCs and Air
Carbon Dioxide) and Water andothers
b) with direct flame preheater
a) without preheater
Figure I. Catalytic oxidation system configurations.
primarily involve hydrocarbons and to some degree NH, and SOz. However, the trace contaminants of concern in air streams may include organohalogens (e.g., solvents such as CHzClzor CH,CCl,) and, frequently, phosphorus-, nitrogen-, and sulfur-containing compounds (e.g., H,PO,, HCN, and HzS). These trace contaminants are, as a general rule, poisons for conventional industrial oxidation catalysts such as supported Pt or Ni. Although some catalyst suppliers and other researchers have begun to develop poison-tolerant catalysts for catalytic oxidation (e.g., Dowd and Hardison (1977)), kinetic characterization of these catalysts and understanding of the overall process have not been widely reported. Two ways in which catalytic oxidation for control of trace VOCs can be carried out are shown in Figure 1. The first uses a direct contact open flame to "preheat" the gas stream upstream of the catalyst. In this configuration, the open flame both preheats the gas stream to an elevated temperature so catalytic oxidation can take place and actually accomplishes some measure of VOC oxidation. This preheat flame may accomplish a significant portion of the overall observed oxidation (Palazzolo et al., 1985, p 78). The chemical mechanism of open-flame oxidation involves free-radical-induced homogeneous reactions and is fundamentally different from heterogeneous catalytic oxidation, which involves activated complexes formed on the catalyst surface. The second method involves only a catalyst bed, over which the gas stream passes, usually after some indirect preheating. The difference between these two configurations is the presence of an open flame, but this difference can he important because the mechanism of oxidation on a catalyst in close proximity to a flame may be different from that on a catalyst by itself. To demonstrate one such difference, Figure 2 shows the observed rate for a heterogeneous catalytic oxidation reaction as a function of temperature. At a low temperature, the reaction takes place on the catalyst surface. Although the overall reaction rate can be controlled by surface kinetics or mass transfer, the reaction still occurs on the catalyst surface. As the temperature is raised, as may occur
Ind. Eng. Chem. Res., Vol. 26, No. 11, 1987 2167
A
idation of trace concentrations of VOCs, and a discussion of mixture effects.
Catalytically Supported Homogenous Reaction
\ I
Y
Temperature Source: Prasad et al., 1984
Figure 2. Reaction r a t e as a function o f temperature. R e p r i n t e d w i t h permission f r o m Prasad e t al. (1984). Copyright 1984 M a r c e l Dekker.
in a flame, the bulk gas temperature is high enough that the homogeneous gas-phase reaction and catalytic reaction occur simultaneously (Prasad et al., 1984). Finally, the homogeneous reaction dominates. In an open-flame configuration, the homogeneous reaction must occur to some extent because the gas stream passes through an open flame at a high temperature. Thus, the downstream catalytic reaction may involve different reactant species than if the gas stream had not passed through an open flame, making study of the catalytic oxidation itself difficult. An additional complication is the possibility of chemical reaction between the fuel used for the flame and the VOC in the inlet gas stream, which could result in a variety of products. For these reasons, the focus of this review is on the catalytic oxidation process itself, as opposed to an open flame/catalytic system. Any fundamental study of the systems in Figure 1must consider the flame and catalysts separately. It is important to note that most commercial systems are of the open-flame/catalyst configuration. Almost invariably, however, it is the “overall,” Le., flame plus catalyst, destruction that is reported for such systems. One important consideration in any catalytic oxidation process at conditions of interest herein is the possible formation of hazardous incomplete oxidation products. One simple example is the formation of Clz observed in the catalytic oxidation of CH3C1 over a-Cr203/A1203 (Weldon and Senkan, 1986). Products that are even more toxic are of course possible. For all alcohols, aromatics, alkanes, aldehydes, naphthenes, ketones, and esters, the products of complete oxidation would of course be C02 and H20. Although many VOCs are from these chemical classes, other associated compounds may lead to other products: VOC contaminant mercaptans amines chlorinated solvents condensed heterocycles
chemical formula
R-SH R-NHp R - C ,I
cNg
fully oxidized products
+
SO, H2O NO, -k H 2 0 H C I -t H 2 0 NO,
-+
H2O
+ C02 + C02
+ COP + C02
Although some of the fully oxidized products may be more innocuous than is the VOC, the gaseous effluent may require further treatment. Fortunately, most oxidized products of these contaminants are acidic. Their concentration in the effluent may be reduced to acceptable levels by mild basic aqueous scrubbing. The remainder of this review summarizes the proposed mechanisms, kinetic models, applications of catalytic ox-
General Mechanisms of Catalytic Oxidation A general theory of the mechanism of the heterogeneous catalytic oxidation of low molecular weight vapors at trace concentrations in air does not exist. However, as with many catalytic reactions, certain observations have been consistently made that have led to general hypotheses about how the reaction takes place. The mechanism of complete catalytic oxidation depends on the type of catalyst used. Basically two types of conventional catalysts are used for oxidation reactions: metal oxides and noble metals (supported or unsupported). For both types of catalysts, the reaction conditions considered in this review should be kept in mind (Table I). Specifically, because oxygen is always present in large excess (from Table I, the molar ratio of O2/VOC is about 102-103),the catalyst surface concentration of oxygen is always relatively high. This also means that the oxygen concentration in the gas phase is essentially constant, and the overall rate will usually be a function of the VOC concentration only. Metal Oxides. Metal oxide catalysts are defined herein as oxides of metals occurring in groups 111-B-11-B (3-12) of the periodic table. These oxides are characterized by high electron mobility and positive oxidation states. These catalysts are generally less active than are supported noble metals, but they are somewhat more resistant to poisoning. This poison resistance may be due to the high active surface area of metal oxides compared to supported noble metals. The literature on catalytic oxidation over metal oxides is more extensive than is that on noble metals. A variety of single and mixed metal oxides have been evaluated for complete oxidation of trace levels of VOCs in oxidizing gas streams. As oxidation catalysts, these oxides are further classified in the literature. Golodets (1983) has divided them by the stability of the oxide. Those forming the most stable oxides (mO, > 65 kcal/mol of 0 )are the alkali and alkali earth metals such as Sc, Ti, V, Cr, and Mn; the rare earth metals; and the actinides Ge, In, Sn, Zn, and Al. Those oxides with intermediate stability ( A H 0 2 9 8 = 40-65 kcal/mol of 0) include Fe, Co, Ni, Cd, Sb, and Pb. Oxides that are unstable (m0,8< 40 kcal/mol of 0 )are the noble metals Ru, Rh, Pd, Pt, Ir, and Au, as well as Ag. The usefulness of this criterion for classifying oxidation catalysts is that presumably the metals that do not form stable bulk oxides remain as reduced metals during oxidation reactions a t moderate temperatures. This suggests that the mechanism of oxidation, even when these metals are supported on refractory oxides such as Si02or A1203,may involve only molecular O2 in the incoming gas stream. By contrast, the lattice oxygen of some metals forming stable or intermediately stable oxides is known to be involved in the oxidation of hydrocarbons and other reactants in 02-containing gas streams. This has been demonstrated over numerous metal oxides by using 1802 in the gas stream and measuring the l6O and l80content of the oxidized products. Another consequence of the classification of oxide catalysts by oxide stability is that there is some optimum level of metal-oxygen interaction in an oxide catalyst. This is because “the catalytic activity [of a metal oxide catalyst] is inversely related to the strength of chemisorption of [the VOC and oxygen], provided that adsorption is sufficiently strong for [the VOC and oxygen] to achieve a high surface coverage” (Bond, 1974). This statement suggests the qualitative behavior shown in Figure 3, often known as a
2168 Ind. Eng. Chem. Res., Vol. 26, No. 11, 1987 I Adsorption Too Weak -Too
I
Adsorption Strong . I
as described by Bond (1974). On p-type oxides such as NiO, adsorbed 0-,which is formed upon O2 absorption. reacts with adsorbed CO via
-
O2 + 4Ni2+ co(g)
4Ni3+ + 202-
[0---Ni3+] + CO(&)
Parameter Measuring Strength of Adsorption Source: Bond, 1974
Figure 3. Catalytic activity as a function of adsorption strength. Reprinted with permission from Bond (1974). Copyright 1974 Oxford University Press.
“volcano” plot. If the chemisorption is too strong, the catalyst will be quickly deactivated as active sites are irreversibly covered (this is one definition of poisoning). If chemisorption is too weak, only a small fraction of the surface is covered and the catalytic activity is very low. It is interesting to examine the quantities used for the abscissa of Figure 3. Bond (1974) suggests using the initial heat of adsorption. Satterfield (19801, discussing metal oxide catalysts more specifically, suggests using the heat of reaction, Qo, for reoxidation of the catalyst. Balandin (1958, 1969) specified that the maximum VOC oxidation rate would occur when Qo QJ2, where Q, is the overall heat of reaction for conversion of VOC to products (e.g., VOC + 0 2 CO, + H2O). Metal oxide catalysts can be generally divided into three groups insofar as catalytic oxidation reactions are concerned. These are n-type semiconductors, p-type semiconductors, and insulators. The basis for this classification is electrical conductivity (which is related to their catalytic properties). In n-type metal oxides, electrical conductivity arises by means of quasi-free electrons that exist because of an excess of electrons present in the lattice. N-type metal oxides are generally not active oxidation catalysts, although vanadium pentoxide (V20,) is a notable exception. P-type metal oxides are electron-deficient in the lattice and conduct electrons by means of positive “holes”. These oxides are generally active oxidation catalysts. Insulators have very low electrical conductivities because of the strictly stoichiometric metal-oxygen ratio in the lattice and very low electron (or “positive hole”) mobility and are generally not active catalysts. However, insulators are often used as catalyst supports. One direct result of the different electrical and chemical properties of n- and p-type oxides is that n-type oxides lose oxygen upon heating in air, whereas p-type oxides gain oxygen. Fierro and de la Banda (1986) discuss the desorption of oxygen from various classes of metal oxides more quantitatively, showing that the less stable the metal oxide is (as measured by the heat of formation per oxygen atom), the more easily the surface is reduced to form oxygen adsorption sites. This is important because the difference in their catalytic properties is a direct result of their respective interaction with oxygen at reaction conditions. Oxygen adsorption occurs far more readily on p-type oxides because electrons can be easily removed from the metal cations to form active species such as 0-,whereas no such mechanism is available on an n-type oxide. On n-type oxides, oxygen adsorption occurs only on prereduced surfaces, i.e., replacing oxide ions removed in a reducing pretreatment. As an example, consider the mechanism of CO oxidation
-
-
-
Co(ads)
C02(ads)+ Ni2-
On n-type oxides such as ZnO, CO oxidation occurs via lattice 02CO + 2 0 2 - C032-+ 2e-
- coz + + + +
02-
co32-
followed by oxide regeneration 2e- Zn2+
Zn”
O2 2Zn2+ Z 0 2 The difference between these mechanisms for the two types of oxides, p-type oxides involving adsorbed 0- and n-type oxides involving lattice 02-,leads to profoundly different activity for deep oxidation reactions. Because adsorbed oxygen species are more reactive than are lattice oxide ions, p-type oxides are generally more active, especially for deep oxidation. The oxidation of CO over metal oxides demonstrates the differences in these oxide catalysts as opposed to noble metals. First, over noble metals, CO(ads)is not thought to be active in the catalytic reaction (the mechanism is Eley-Rideal, i.e., involving a reaction between adsorbed oxygen and gas-phase CO). Second, and partly as a consequence of this, the formation of an adsorbed carbonate intermediate does not occur over noble metals (Golodets, 1983, p 283) but has been shown to be present at least on NiO at some conditions (Thomas and Thomas, 1967, p 372). This adsorbed carbonate intermediate, in a related study on HCHO oxidation on NiO, is shown to be autocatalytic; Le., the presence of C02in the feed to the catalyst increases the reaction rate by forming carbonate sites that oxidize HCHO directly (Foster and Masel, 1986). As shown by Golodets (1983, p 2891, the formation of the carbonate intermediate also may be interpreted in terms of the more conventional redox or Mars-van Krevelen cycle, which explicitly involves the oxidation and reduction of the metal oxide (Satterfield, 1980). The Mars-van Krevelen mechanism of catalytic oxidations over metal oxides is a redox mechanism involving both gasphase and lattice oxygen Me0 + R - RO + Me (1) 2Me + O2 2Me0 (2) where Me is a metal cation and R is a hydrocarbon reactant. This mechanism and the resulting kinetic model account for the observed zero order in hydrocarbon partial pressure (assuming step 2 is rate controlling) and the zero order in oxygen partial pressure at high temperatures (because of the lower activation energy of step 1) for some reactions (Satterfield, 1980, p 183). Further, Pearce and Patterson (1981, p 295) point out that, for many reactions, if the oxygen supply is discontinued over a metal oxide catalyst that has been brought to steady state, the catalyst will continue to oxidize the reactant at the same selectivity (although the activity will decline). If the O2 supply is started again, the original activity is restored. The activity of the catalyst for deep oxidation may depend on the type of oxygen involved in the reaction. The Zn”
-
Ind. Eng. Chem. Res., Vol. 26, No. 11, 1987 2169 various types of oxygen present on the surface of metal oxide catalysts have been the subject of much study. The existence of 02-, atomic 0-,and of the regular 02ions in the oxide lattice is beyond doubt (Margolis, 1973; see also Sachtler (1970) who discusses the participation of various forms of oxygen in some hydrocarbon oxidation reactions). Low-temperature ESR studies have shown that the interaction of gaseous diatomic oxygen with an oxide surface proceeds through the following steps (Kon’ et al., 1972a,b):
02-+ e0-
+ e-
-
-
20-
02-
where 02-is directly incorporated into the oxide lattice (see step 2 in the Mars-van Krevelen mechanism above). For complete oxidation of trace contaminants at the conditions of Table I, the extent to which lattice oxygen (02-) and chemisorbed oxygen (OL, 0-)participate in the mechanism is not clear. This is in contrast to industrial partial oxidation reactions where selective oxidation is dependent only on lattice oxygen (Pearce and Patterson, 1981, p 295). Although complete oxidation to C 0 2 is possible when no oxygen is present in the gas stream, the consensus seems to be that both chemisorbed and lattice oxygen can participate in complete oxidation, although lattice oxygen certainly accounts for selective oxidation (Fierro and de la Banda, 1986). Additional views on complete oxidation as compared to selective oxidation center primarily on oxygen mobility which results from so-called “weak” metal-oxygen bonds. Weak metal-oxygen bonds, i.e., low energies of oxygen binding to the oxide surface, are necessary for catalysts that completely oxidize reactants (Simons et al., 1968; Roiter et al., 1971; Boreskov et al., 1971). Haber (1975) postulates that “surface adsorbed” oxygen may generally lead to complete oxidation, whereas lattice oxygen is needed for partially oxidized products. Presumably, this surface-adsorbed oxygen is more mobile than is lattice oxygen. Though in principle either 02-or 0- may promote complete oxidation, 0- may be the more reactive (i.e., “mobile”) because it is known to oxidize CO and H2 even a t -196 “C (Margolis, 1973). Oxidation on n-type oxides is thought to involve lattice oxygen, while on p-type oxides the reaction involves an adsorbed oxygen (Bond, 1974). Haber’s postulate suggests that p-type oxides are active for complete oxidation, but at least some n-type oxides can be used for partial oxidation. This is generally true. Finally, Satterfield (1980, p 185) states that “highly mobile” oxygen should result in a highly active, nonselective catalyst. Germain (1972) in fact says that for simple metal oxides, there is a direct, but limited, correlation between activity and oxygen mobility. Satterfield also points out, however, that oxygen mobility as a sole criterion for catalyst activity and selectivity is somewhat limited. For instance, this concept does not account for the effect of partially oxidized intermediate adsorption on selectivity in series-type reactions nor for the effects of mixed oxide composition and catalyst surface defects on catalytic reactivity, especially in partial oxidation reactions. Mixed metal oxides are used quite often in industrial partial oxidation reactions, examples being bismuthmolybdate for the oxidation of propylene to acrolein and vanadia-molybdate for oxidation of benzene to maleic anhydride. Some mixed oxides also are quite active deep oxidation catalysts, a good example being Mn02-CuO. The difficulties in understanding mixed oxides are of course more formidable than they are for single metal
oxides. I t is a well-established empirical fact that mixed oxides behave quite differently than as individual oxides in most catalytic reactions. This situation is further complicated by the often dramatic effect of “promoters,” such as alkali metal oxides that are added to the catalyst intentionally. The mechanism of catalytic oxidation on mixed metal oxides is thought to be similar, at least in principle, to that on single metal oxides. The interesting aspect of these mixed oxides is their generally higher activity compared to the single oxide components. This is thought to be due to the readily available multiple energy levels of the metals and their associated oxygen anions, which makes available to the organic reactant more accessible active oxygen anions. This also may result in higher surface mobility of oxygen and/or the activated complex as well as electron transport through the lattice. Although many attempts to examine the mechanism of catalytic oxidation have been reported, the exact mechanism even for intensely studied reactions such as CO oxidation is not completely understood. However, certain general statements about p-type oxides and Vz05deep oxidation catalysts can be made. (1)High activity catalysts are generally metal oxides in which the metal can assume more than one valence state, are p-type semiconductors, and produce highly mobile chemisorbed surface oxygen (a consequence of this last characteristic is an intermediate heat of adsorption of O2 that produces the familiar volcano plot). (2) Mixed metal oxides often have unique properties relative to the individual oxide constituents, primarily in terms of activity and stability (Prasad et al., 1984, p 20). (3) Lattice oxygen, as well as gas-phase oxygen, participates in the catalytic oxidation process. Various attempts to explain exactly how the process occurs usually include a redox cycle at the oxide surface with anionic oxygen from the surface (either chemisorbed or lattice oxygen) reacting with a chemisorbed or gas-phase organic reactant. Whether the reactant is oxidized directly in the gas phase by an adsorbed oxygen species (a so-called Eley-Rideal mechanism) or adsorbs and then reacts (a LangmuirHinshelwood mechanism) is especially important in considering mixture effects because the selectivity for a given reactant in a mixture would presumably be different if it must be adsorbed prior to being oxidized. Such mixture effects are difficult to study at the oxide surface, and little is reported. (4) Various attempts have been made to relate the activity of metal oxides for catalytic oxidation with various thermodynamic properties that would seem to be important, knowing that in a continuous system oxygen must adsorb from the gas phase and interact with the metal oxide lattice. Such correlations include comparison of catalytic activity with heat of formation of the oxide from elemental metal (per oxygen atom usually), initial heat of chemisorption of oxygen on evaporated metal films, heat of dissociation of the first oxygen atom from the oxide, grams of oxygen per gram of metal oxide, and others. One of the most widely used comparisons is that of activity versus the heat of reaction for reoxidation of the catalyst. Such a plot has a maximum (see Figure 3), suggesting that if oxygen is chemisorbed either too weakly or too strongly, the activity is not as great as for some intermediate strength of chemisorption that might correspond to activated yet mobile surface oxygen species (probably anionic). (5) The most active single metal oxide catalysts for complete oxidation for a variety of oxidation reactions, as calculated from the above techniques and confirmed by
2170 Ind. Eng. Chem. Res., Vol. 26, No. 11, 1987
experiment, are usually found to be oxides of V, Cr, Mn, Fe, Co, Ni, and Cu. The mechanism of oxidation on these p-type oxide catalyste is generally thought to involve strong adsorption of the organic compound at an anionic oxygen site in the oxide lattice leading to the formation of an activated complex. This complex can then react further to yield products of complete combustion. Some authors have suggested that a degree of oxygen mobility on the surface is necessary for such a mechanism, with the successive desorption of complete oxidation products C02 and H20 and movement of the remaining activated complex fragment to a nearby available oxygen anion for further reaction (Germain, 1972). (6) Metal oxides generally have somewhat lower activity than do noble metal catalysts, but they have greater resistance to certain poisons (especially halogens, As, Pb, and PI. Noble Metals. Noble metals are defined herein as Pt, Pd, Ag,and Au. These noble metals are frequently alloyed with the closely related metals Ru, Rh, Os, and Ir, and they are usually supported on an oxide support such as 7-A1203 or Si02. Although in principle any of these metals may be used as oxidation catalysts, in most practical systems, only Pt, Pd, and a few alloys are used because the generally high temperatures employed for most oxidation catalyst applications (e.g., catalytic incineration, automotive exhaust catalysts) can cause sintering, volatility loss, and irreversible oxidation of the other metals. Limited supply and the resulting high cost of these other metals also minimize their use (Prasad et al., 1984). As a result, most reported research deals with supported Pt and P d (and alloys of these other metals) catalysts. In general, much less research has been reported on catalytic oxidation at conditions of interest herein on noble metals than has been reported on metal oxides, especially research dealing with the mechanism of complete oxidation. Although supported noble metal oxidation catalysts (apparently based on formulations developed for automotive applications) have been widely applied for control of exhaust from various industrial processes, most reported information deals primarily with overall performance, not the mechanism of oxidation. One reaction over noble metals that has been extensively studied is the oxidation of CO. Although it is not within the scope of this review to examine the vast literature dealing with this reaction, others who have done so (Wei, 1975) have reached several general conclusions of interest. (1)A change in reaction order in CO from positive at low CO concentrations to negative at high CO concentrations and multiple steady states has been both predicted and observed (Hegedus and Baron, 1977). (2) Several surface species of adsorbed CO have been observed during oxidation on Pt/Al2O3. Barshad et al. (1985) conclude that CO adsorbed on oxidized Pt was not taking part in the reaction but that CO adsorbed on a Pt atom sharing an adsorbed oxygen with a neighbor accounts for their results. (3) Apparently, 0, adsorbs onto adjacent Pt atoms, and CO diffuses along the surface to the active 0-containing sites. O2 adsorption has been postulated as being ratedetermining in at least some cases (Barshad et al., 1985). Despite the extensive research on this specific reaction, it is unclear whether these results can be directly extended to oxidation of other organic compounds. However, it is interesting that, at moderate CO concentrations (10 vol % CO in 15% 02/75% N,) and low temperatures (150 “C), the rate-determining step is O2adsorption on adjacent Pt sites (Barshad et al., 1985). Above 230 “C. Langmuir
(1921) postulated that the mechanism of CO oxidation over Pt changed to a reaction between adsorbed O2and gaseous CO (an Eley-Rideal mechanism). Further, in a large excess of 02,above 350-400 “C, Sklyarov et al. (1969) and Tretyakov et al. (1970,1971) show that on pure Pt metal, the rate law is first order in CO and zero order in 02.The above observations suggest that, as the temperature is increased, the surface concentration of adsorbed oxygen increases while that of CO decreases. Because the carbon atom of CO is known to adsorb to the Pt surface, this postulate may hold for chemically similar VOCs and is consistent with the rate laws that are zero order in O2 at the conditions of Table I. The noble metals Pt and Au are said to function in the reduced state at all conditions (Germain, 1969). This may be true for P d as well, at least a t relatively low temperatures. At temperatures above about 450 “C, prolonged exposure of supported P d to oxygen has been observed to cause structural changes in the P d metal that result in a loss of catalytic activity in the oxidation of methane (Cullis and Willatt, 1983). In this same study, supported Pt did not undergo detrimental structural changes at these temperatures. Noble metals generally form quite unstable oxides (e.g., for Au,03 = 0). However, oscillatory behavior in catalytic oxidation of n-heptane on Pt/yA120, at 178 “C has been linked to the presence of relatively active reduced metallic Pt sites and relatively inactive oxidized [Pt’”] (Volter et al., 1987), suggesting that even at moderate temperatures, there may be a complex interaction between the Pt and oxygen that affects the hydrocarbon conversion. Thus, the mechanism for oxidation on noble metals may be different from that on metal oxides. Literature on the complete oxidation of organic compounds over noble metals or supported noble metals shows the following. (1) Noble metals may follow either a Langmuir-Hinshelwood type of mechanism (reaction between adsorbed oxygen and an adsorbed reactant) or an Eley-Rideal mechanism (reaction between adsorbed oxygen and a gas-phase reactant molecule). In the case of nucleophilic organic reactants (e.g., CO or olefins), both the oxygen and reactant are adsorbed and react on the surface. ( 2 ) On Ag, oxygen chemisorption is relatively strong, with a transfer of electrons to the oxygen taking place (Dixon and Longfield, 1960). On Pd, however, oxygen adsorption is relatively weak. This implies that the reaction mechanism on Ag involves a redox mechanism similar to that for metal oxides. On Pd, however, anionic surface oxygen species do not exist and the mechanism may proceed through direct electron transfer between the reactant and oxygen, with the metal surface providing an energy-modifying function. On Pt, Volter et al. (1987) show that oxidation occurs faster in air than in pure oxygen due to the formation of a relatively inactive “oxidic” [Pt”] species in highly oxidizing environments. (3) Pt and P d have a high activity for total oxidation of C2-C8 paraffins (Golodets, 1983, p 455). As a general rule, these noble metals have a higher activity than do metal oxides for complete oxidation. The mechanism of complete oxidation of paraffins on these metals has not been well studied. (4) The general mechanism of oxidation over noble metals is thought to involve the dissociative adsorption of oxygen 0 2
+ [I
-*
[O,l-
fast
2[01
where [ ] represents a “surface site”. This step is then followed by direct reaction of the gaseous organic reactant
Ind. Eng. Chem. Res., Vol. 26, No. 11, 1987 2171 with [O], or the reactant may be weakly adsorbed first. Both Golodets (1983) and Germain (1969, 1972) have proposed that oxidation often occurs by a parallel-series mechanism: CO?
R
+
HzO
+ C01 2 RO
Models The overall process of any catalytic reaction is a combination of mass transfer (describing transport of reactants and products to and from the interior of a solid catalyst) and chemical reaction kinetics (describing chemical reaction sequences on the catalyst surface). The overall process is (1)transport of reactants from the bulk fluid through the gas film boundary layer to the surface of the particle, (2) transport of reactants into the catalyst particle by diffusion through the catalyst pores, (3) chemisorption of a t least one reactant on the catalyst surface, (4) chemical reaction between chemisorbed species or between a chemisorbed species and a physisorbed or fluid-phase reactant, (5) desorption of reaction products from the catalyst surface, (6) diffusive transport of products through the catalyst pores to the surface of the catalyst particle, and (7) diffusion of products through the exterior gas film to the bulk fluid. In principle, any of these steps, or some combination, can be rate controlling. The rate-controlling step for a given reaction and catalyst can vary depending on temperature, flow rate, gas composition, and catalyst geometry. The development of a cost-effective catalytic oxidation system requires use of a solid catalyst material with a high specific surface area, i.e., high surface area per net weight of catalyst. To achieve this high specific surface area, a highly porous material is required. The presence of many small pores necessarily introduces pore transport diffusion resistance as a factor in the overall, or global, kinetics. Fott and Schneider (1984) point out that the overall rate of most industrial gas-phase reactions is controlled by mass transport within the catalyst particle or by the heat transport between the particle and the flowing gas. Any comprehensive analysis of actual catalytic oxidation systems of practical interest must include a quantitative understanding of the relative effects of mass transfer (steps 1,2,6, and 7) and surface reaction (steps 3,4, and 5). The discussion herein focuses solely on surface reactions, recognizing that mass transfer may be rate determining in many practical situations. The reader is referred to Satterfield (1970) for a discussion of the criteria for determining the relative importance of mass transfer and surface reaction. Kinetic models of catalytic oxidation can take one of two general forms. The first and simplest is a power law rate equation that expresses the rate as a product of a "rate constant" and the reactant concentrations raised to a power; e.g., for the catalytic oxidation of methane, the rate would be expressed as r = kPCHtPO:
much larger than the organic reactant partial pressure (if methane a t 1000 ppm is reacted in air, Po,/PcH4 4 210). Thus, because the oxygen partial pressure will be essentially constant during the reaction, the observed rate will be proportional only to the reactant partial pressure. The rate dependence on the VOC concentration at conditions of interest herein is typically near first order. This very simple type of rate law, although useful for quick comparisons, does not provide any insight into what is occurring a t the catalyst surface. However, the degree to which any rate equation yields information about a reaction mechanism is frequently open to question. The fact that some proposed mechanism yields a rate expression that is consistent with experiment does not necessarily make that mechanism (or rate expression) uniquely correct. Nonetheless, rate expressions more complex than a simple power law are sometimes useful. For example, a power law expression does not provide any insight into the reasons for changing reactant order (i.e., a changing value of a) with temperature or organic reactant concentration. However, such effects are frequently observed in oxidation reactions and are often consistent with more fundamentally based rate expressions. Consider, for example, what one would suppose to be the simple oxidation of methane. Golodets (1983, p 445) states that methane oxidation over metal oxide catalysts may be interpreted by the following mechanism: 1.
0 2
2.
3.
+ [1
[Ozl + CH4
[I
--
[CHZI + [OI
5.
[CHZO]
7.
2[01
+ [O] -.+ [CH,] + HzO
4. 6.
[Ozl
+
[CHZOI
HCHO
[CHZO] + [O] -* [HCOOH] [HCOOH] + [O]
-+
COZ + HzO
(11)
This is a so-called Eley-Rideal mechanism, meaning that CH, from the gas phase reacts directly with an adsorbed oxygen species (step 3). The rate expression from the above mechanism is
where v is the stoichiometric coefficient of oxygen in the overall reaction (v = 2 for complete oxidation via the reaction CH4 + 202 COz + 2Hz0) and ki's are the individual rate constants for the reactions in the mechanism above. With this more fundmentally based (and more complex) rate expression, the following can be explained. (1) Fractional Reaction Orders in P C H 4 . By use of a power law expression, fractional values of the reaction order in PCH4,which often depend on the concentration range in which the rate is measured, are frequently observed (Le., 0 < a < 1 in eq I). These can be explained by equating eq I and I11
-
(1)
where r = reaction rate (g/s), k = rate constant (g/(sP C H 4 = partial pressure of methane (atm), Po, atm(a+b))), = partial pressure of oxygen (atm), and a, b = empirically determined reaction orders (dimensionless). For the conditions encountered in most VOC control applications, the oxygen partial pressure will always be
and noting that, by forcing a power law model on any given set of data, a and b could be fractional numbers. (2) Decrease in the Rate of Methane Oxidation in the Presence of Formaldehyde. Note that formaldehyde is an intermediate in the mechanism shown in eq 11. It interacts more rapidly with adsorbed oxygen than does
2172 Ind. Eng. Chem. Res., Vol. 26, No. 11, 1987
methane, thus inhibiting step 3 of eq I1 by lowering [O], k S ,and thus r. (3) Zero Reaction Order in Oxygen Pressure at High Values. As Po, becomes large in eq 111,
this indeed is observed over most oxidation catalysts. (4) Decreasing Oxidation Rate with Increasing Strength of the Metal-Oxygen Bond in Metal Oxide Catalysts. Because a t high values of Po,, r = k$CH4, and because step 3 of eq I1 involves breaking oxygen-metal bonds, this effect would be expected. It is, in fact, observed for single and mixed metal oxides (Golodeta, 1983, p 447) as well as noble metals (Golodets, 1983, p 439). Other complications include inhibition or enhancement by oxidation products C02 and H 2 0 (e.g., Foster and Masel, 1986). Rate expressions accounting for this have been developed (e.g., Golodets, 1983, p 132). It is interesting that the mechanism of eq I1 can explain many observations. However, this may be simply fortuitous because other mechanisms have been proposed that can explain more readily other phenomena (e.g., the selectivity to formaldehyde formation a t low conversions) but that present other phenomenological problems. The conclusion to be drawn is that the formulation of a complex reaction model, however consistent it may be with experiment, may be unnecessary at best and misleading a t worst. In a practical situation, the best model is the simplest one that is consistent with the available data and that allows reasonable extrapolations to conditions of interest to be made.
Applications Reported applications for low-temperature catalytic oxidation include gasoline vapor removal; polymer processing vapor control (Budd, 1978; Anonymous, 1975); coating operations (Bonacci and Heck, 1983);odor control from sewage treatment and food processing (Balough et al., 1975; Roberts and Roberts, 1976); and VOC removal from processes such as spray painting, offset printing, and coating operations (Jennings et al., 1985). Almost all reported applications of this technology are for removal of a trace “contaminant” from a process gas stream discharged to the atmosphere. A summary of reported experimental catalytic oxidation research is presented in Table 111. Because the composition of the gas stream containing the oxidizable compounds is very complex in most of these systems, the complete chemical composition is not reported. Often, only generic analyses are given (e.g., 40% “aromatic”, 60% “aldehyde”). Also, the overall chemical reaction rate and conversion are frequently given only in terms of “odor reduction” or “percent carbon removal”. Thus, although helpful, such reports generally do not provide sufficient specific information to extract rate constants or to assess quantitatively any mixture effects. It is thus interesting that, in spite of a lack of fundamental understanding of the complex processes occurring in actual installations, catalytic oxidation is widely used for trace contaminant removal. This is especially true in Europe, where (presumably) the high cost of energy and relatively strict environmental regulations (Acres, 1970) make catalytic oxidation a practical means of achieving acceptable emission reduction a t a reasonable price (the primary alternatives being thermal incineration with high energy costs and adsorption and/or absorption systems in which the contaminant is only concentrated-and must then be destroyed or disposed, at additional cost). It seems reasonable that, as more fundamental knowledge of the
complex reactions of trace VOC mixtures on catalysts is gained, substantial improvements in low-temperature catalytic oxidation can be achieved. This will most likely be seen in a reduction in the temperature needed to oxidize completely all contaminants in a given gas stream. An ultimate goal of such research would be a catalyst and a contacting process that would oxidize (or otherwise destroy) all contaminants in the gas stream at its inlet temperature (often near ambient 25 “C). Catalytic oxidation of trace contaminants in air also has been used in analytical chemistry. One process to date has been developed to analyze for total non-methane hydrocarbons emissions from stationary sources (EPA method 25). This technique involves drawing a gas sample containing hydrocarbons over a catalyst and oxidizing the hydrocarbons to COz, which is then measured. The best catalysts for this process were found to be Pt or Pd on alumina (EPA, 1984).
Mixture Effects The extant literature on the complete catalytic oxidation of hydrocarbons and substituted hydrocarbons has been written from two basic perspectives-in-depth laboratory investigations of pure components and more qualitative investigations of very complex and ill-characterized mixtures. The laboratory studies involve a wide range of reactant concentrations, temperatures, pressures, and reaction conditions. The investigations of “real” mixtures also involve a similarly wide range of conditions because the usual application is for end-of-pipe air pollution or odor control from various industrial sources. However, unlike most laboratory research, many studies of real mixtures report only overall gross performance and are not helpful in understanding the fundamental processes involved. One notable exception has been the development of the catalytic exhaust system for automobiles, one of the most intense catalyst development efforts ever undertaken. An automotive catalyst normally consists of Pt/Pd and some Rh on a ceramic support. Catalytic exhaust control systems function under severe and rapidly changing conditions and must be active for several reactions that reduce automotive emissions-C0 oxidation, hydrocarbon oxidation, and NO, reduction (this is the so-called “three-way” catalyst). Typical operating conditions are temperatures of 400-600 O C (or much greater under certain conditions) and 150 000 h-l space velocity. Numerous reviews of the development and performance of these catalysts are available, and these catalysts are of interest because they are frequently used for control of VOC emissions, particularly in conjunction with open-flame “preheaters”. Unfortunately, these catalysts are not designed to resist poisoning by many VOC-type compounds, particularly those containing chlorine and sulfur. Very little has been reported in the literature on research into the catalytic oxidation of mixtures as compared to the catalytic oxidation of individual components. Often, but not always, the catalytic reaction of a component in a mixture cannot be predicted solely from the behavior of the individual components. This has been shown for nonoxidation catalytic reactions by several authors (e.g., Spivey and Bryant (1982)) and for catalytic oxidation reactions by Palazzolo et al. (1985). Dixon and Longfield (1960) show that the yield of partially oxidized products in the catalytic oxidation of mixtures of methylnaphthalene and naphthalene over a vanadium oxide catalyst “at optimum conditions” can be predicted directly and linearly from the feed composition, suggesting no anomalies at least for these closely related compounds. However, Satterfield (1980) uses examples from the lit-
Ind. Eng. Chem. Res., Vol. 26, No. 11, 1987 2173 erature to show that nonlinear anomalies occur in other types of reaction mixtures. These anomalies are interpreted in terms of competitive adsorption for hydrogenation and hydrodenitrogenation. He also suggests that accompanying homogeneous or free-radical effects, which may occur in catalytic oxidation reactions, can cause anomalous mixture behavior. These examples of Dixon and Longfield and of Satterfield are not necessarily contradictory, with the example of Dixon and Longfield being what might be expected for closely related compounds that may not exhibit any selective adsorptivity. The relatively few scientific studies of the oxidation of a compound by itself and in a mixture generally show retardation of the compound of interest, but there are exceptions. Yao (1973) studied the oxidation of C2H4, C3H6,CzH& C3H8,and CO over various a-Cr203crystal faces. He found that the oxidation rate decreased in the order C3H6 > co C2H4 > CzH6both as single components and in various binary mixtures. This is fairly consistent with results of other deep oxidation catalysts over this catalyst; C2H4 was retarded much less than was C& (ina binary mixture) and C3H6 much less than C2H4 (also in a binary mixture). He suggests that the results can be explained by competition for adsorbed O2 or by competition for undefined “adsorption sites” among the hydrocarbons. Interestingly, Yao and Kummer (1973) also added trace quantities of possible propylene oxidation intermediates CH3COOH and CH,CHO to see if they affected the overall complete oxidation of propylene on NiO catalysts. Although there was a temporary and reversible suppression of the propylene oxidation rate in the presence of these intermediates, the oxidation rates of the acid and aldehyde themselves were suppressed a t steady state over their respective pure component values. Similarly, Cullis et al. (1970) studied the oxidation of methane over palladium catalysts by adding small quantities of possible methane oxidation intermediates (CH30H and CH20) to a methane/oxygen mixture, much in the manner of Yao (1973). Contrary to Yao, however, both intermediates retarded methane oxidation at steady state. As with Yao and Kummer’s results, both intermediates were themselves completely oxidized. Interestingly, propane retarded methane oxidation when added to a methane/oxygen mixture a t similar conditions but was itself oxidized very little. Cullis et al. (1970) also studied the effect of halomethanes on methane oxidation at the same conditions. All halomethanes retarded methane oxidation, with the retardation increasing with halomethane concentration. Of particular interest herein is that CH20 production from methane increased as the concentration of CH2C12increased up to 0.9 mol % . This is especially significant because it shows that, a t least for this catalyst, this halogenated hydrocarbon (CH2C12)not only decreases the overall primary hydrocarbon (CH,) conversion but also actually increases the yield of an environmentally hazardous compound, CH20, a t the same time (this was the only partial oxidation product observed). The oxidation of CH2C12itself is not discussed in detail, though CH2C12 is said to be “completely consumed” a t concentrations less than about 0.24mol %. Also of interest is that the activity for methane oxidation and selectivity to deep oxidation products C 0 2 and H 2 0 could be completely restored, gradually, by eliminating CH2C12from the feed. Another possible catalyst poison, sulfur, was studied by Pope et al. (1976), who examined the oxidation of CO and some odorous organic compounds over supported and unsupported Co304. In CO oxidation, they showed that the addition of dimethyl sulfide, (CH3I2S,even in trace
-
quantities, retarded CO oxidation significantly and irreversibly while the sulfur became incorporated in the catalyst. However, one objective of this research was to develop a sulfur-tolerant catalyst. Thus, by choosing a solid metal oxide catalyst, the authors claim that “the c0304 surface is renewed as sulfur becomes incorporated [into the catalyst] by diffusion to the interior”. Thus, the “catalyst” acts as both a sorbent (for sulfur) and a catalyst (for CO oxidation). This is an interesting principle because, if the “catalyst” can either be cheaply regenerated or disposed, any undesired oxidized sulfur products would not be produced, as would be the case for, say, most noble metal deep oxidation catalysts. Questions remain, however, as to whether the slightly lower activity of the catalyst in the presence of sulfur is economically acceptable. The authors also found that for butyric acid oxidation, the effect of (CH3)2Saddition was reversible, in contrast to the results for CO oxidation. A closely related study from the same laboratory examined CuO, Co304,Mn02, V205,and Pt-Torvex for the oxidation of n-butanal and methylmercaptan (CH3SH) mixtures at moderate temperatures at high space velocities (Heyes et al., 1982). As the temperature was raised, the conversion of CH3SH to SOz over Co304and Mn02 went through a maximum, suggesting that the incorporation of sulfur into the catalyst, as proposed by Pope et al. (19761, takes place primarily a t temperatures either higher or lower than this maximum, perhaps in two chemical forms. Mixtures of n-butanal and CH3SH were oxidized in extended continuous tests at a temperature high enough that no SO2 was formed in tests done for only a limited time. As the catalyst remained on stream, CH,SH conversion to SO2 increased from 0 to over 50%. Also, quite interestingly, over Co304,CH3SH conversion remained a t essentially 100% for the life of the test, while butanal conversion dropped from an initial value of 100% to 20% in a short time. Over Mn02, which was a more active catalyst for CH3SH oxidation as a single component than was Co304,CH3SH conversion also remained at 100% with time, whereas n-butanal conversion dropped to about 85% over a comparable time frame. Over CuO, n-butanal and CH3SH conversions remained a t essentially 100% throughout the tests. The results of Heyes et al. (1982) suggest that the oxidation of even this simple binary mixture is a complicated function of temperature, time, and catalyst with dramatic changes in activity and selectivity t~ complete oxidation products being observed. The situation is further compounded by the incorporation of sulfur into the catalyst itself. It appears, from X-ray analysis, that sulfur is incorporated into the catalyst, rather than simply building upon the active sites. The effect of this sulfiding action ranged from severe deactivation for n-butanal oxidation over Co304to essentially no effect over CUO. The catalytic oxidation of hydrocarbons in mixtures with nitrogen-, sulfur-, and chlorine-containing compounds has shown how hydrocarbon oxidation is generally inhibited in such mixtures, with the inhibition related to the rate at which the non-hydrocarbon contaminant is itself oxidized. Pope et al. (1978) studied such effects over a PtThermacomb catalyst, which is used commercially for catalytic oxidation to trace contaminants in air. The hydrocarbons were generally more easily oxidized than the nitrogen-, sulfur-, and chlorine-containing compounds, both individually and in mixtures. Also, incomplete mass balance closure for all compounds suggested that partial oxidation products were present in the outlet gas, though these were not identified. Of greater interest is a series
2174 Ind. Eng. Chem. Res., Vol. 26, No. 11, 1987 Table 111. Summary of Reported Low-TemperatureCatalytic Oxidation Research"
catalvst
gas
stream comuosition
gas stream temp, "C
Dress.*
space velocity, vol/h/vol
ref
aoolication ~~
Ag powder chabazite tuff (a natural zeolite containing Ca and K oxides) 0.5% P t on 'js-in. support
H-ZSM5 acidified FeS04 MzO/BaO MiO' Pd/Si02 Pd-Ag/ A1203 Pt-alumina non-noble metal perovskite-type metal oxides, 1% Pt-alumina
vzos
V205-Cr203 ViOr-MoOi
p;,o,
2-9 vol % C2-Cs olefins 10-20 vol % o2 71-88 V O ~% Nz 2.5-10.2% H2S 4.9-17.8% O2 (balance N2)
230-245
1 atm
NRc
mechanistic study of olefin oxidation
Akimoto et al., 1982
200-250
NR
NR
evaluation of a natural zeolite
Alabiso et al., 1979
0.12-1.0% CO 0.17-2.0% 02 0-270 coz 74% Hz 24% Nz 0 . 2 2 % CH4 CHdOz, mole ratio 99/1 to 22/78
139-156
130-180 10000 h-' Psig
CO removal from NH3 synthesis gas
Andersen and Green, 1961
183-518
1 atm
1080120000 h-l
CHI partial oxidation
Anderson and Tsai, 1985
diamine in air
230-300
1 atm
NR
anonymous, 1975
1% pollutants (presumably in air) 2 vol % CHI in air
260
NR
20000 h-l
200-800
NR
Arai et al., 1986
air/benzene = 9.911 to 23.6/1
300-525
NR
4500050000 h-' NR
odor removal from nylon production gases pollutant removal from acrylonitrile manufacture CHI combustion production of phenol from benzene
Badarinarayana et al., 1967
pseudocumene, CgH12 0.071-0.125 mol/m3; 02 0.91-2.9 mol/m3 500-5000 ppm CH3CH0 in air 600-8000 ppm CH30H in air 380-1350 ppm benzene in air 250-300 ppm dioctyl phthalate in air 70-400 ppm dibutyl and diheptyl phthalate in air 0-20% 02 0-2070 co 0-40% CO2 balance He trace amts of CH3SH; 226 mg/m3 organic C plus 1 ppm H2S,