4600
Ind. Eng. Chem. Res. 1997, 36, 4600-4608
Novel Catalysts for the Environmentally Friendly Synthesis of Methyl Methacrylate James. J. Spivey,*,† Makarand R. Gogate,† Joseph R. Zoeller,‡ and Richard D. Colberg‡ Research Triangle Institute, Research Triangle Park, North Carolina 27709-2194, and Eastman Chemical Company, Kingsport, Tennessee 37662-5150
The development of a process for the synthesis of methyl methacrylate (MMA) from coal-derived syngas can alleviate the environmental hazards associated with the current commercial MMA technology, the acetone cyanohydrin (ACH) process. A three-step syngas-based process consisted of synthesis of a propionic acid, its condensation with formaldehyde, and esterification of resulting methacrylic acid (MAA) to form MMA. The first two steps, propionic acid synthesis and condensation, are discussed here. The low-temperature, low-pressure process for single-step hydrocarbonylation of ethylene to propionic acid is carried out using a homogeneous iodinepromoted Mo (CO)6 catalyst at pressures (30-70 atm) and temperatures (150-200 °C) lower than those reported for other catalysts. Mechanistic investigations suggest that catalysis is initiated by a rate-limiting CO dissociation from Mo(CO)6. This dissociation appears to be followed by an inner electron-transfer process of an I atom from EtI to the coordinately unsaturated Mo(CO)5. This homogeneous catalyst for propionate synthesis represents the first case of an efficient carbonylation process based on Cr group metals. The condensation of formaldehyde with propionic acid is carried out by acid-base bifunctional catalysts. As a result of screening over 80 catalytic materials, group V metals (V, Nb, and Ta) supported on an amorphous silica are found to be most effective. A 20% Nb/SiO2 catalyst appears to be the most active and stable catalyst thus far. Preliminary relations among the reaction yield and catalyst properties indicate that a high surface area and a low overall surface acidity (270 °C) (Bertleff, 1986; Samel et al., 1993) to produce propionic acid. Alternative processes, which operate at substantially lower pressures and temperatures, generally require expensive catalysts such as Rh, Ir, or Pd, and none have been employed commercially (Bertleff, 1986; Colquhoun et al., 1991; Forster et al., 1981, Mullen, 1980; Pino et al., 1977). Due to either the harsh conditions or the expense of these alternative catalysts, most current producers of propionate derivatives use the hydroformylation-oxidation sequence to generate propionic acid, which is subsequently converted to its derivatives in a separate step. Although Cr group metals have been reported to act as “promoters” or “stabilizers” when used in combination with known carbonylation catalysts such as Co, Ni, Rh, and Ir, they have not been shown (or believed) to have significant catalytic activity in isolation. In fact, prior to the study reported here, the only known examples in which Cr group metals have been demonstrated to induce carbonylation of any substrate were limited to a stoichiometric carbonylation of alkyl iodides to esters using a Mo(CO)6/F- catalyst (Imbeaux et al., 1992) and a marginally catalytic (∼0.6 turnover/h) carbonylation of R-difluorinated iodides using Mo(CO)6 (Lichstein, 1974). In the study reported here, a stabilized Mo(CO6) homogeneous catalyst is used at 130-170 °C and 350750 psig for reaction 7. Step 2. Condensation. A second goal of this research is to find a stable and selective heterogeneous catalyst for the condensation of propionic acid and anhydride with formaldehyde. The condensation of propionate derivatives (produced in reaction 7) with formaldehyde provides a novel route to synthesize Rand β-unsaturated acids such as MAA. This condensa-
Propionate Synthesis. The rate of the hydrocarboxylation reaction was measured in a 2-L Hastelloy C overhead stirred autoclave fitted with a high-pressure condenser and a dip tube for removing samples during the course of the reaction. Gas mixtures were prepared in a stirred gas mix tank heated to 45-50 °C, and feed lines were heat-traced and maintained at 40-50 °C. The mix tank and lines must be heated to avoid ethylene liquefaction and separation, particularly when cooling begins to occur (due to expansion) either as ethylene is added to the tank during preparation of the gas mixture or as it is removed from the tank during the reaction. Liquid samples are removed every 20 min for 5 h and analyzed for ethyl iodide, ethyl propionate, propionic anhydride, and propionic acid content by gas chromatography (GC) using a Hewlett-Packard 5890 GC containing a 25-m (0.25-mm i.d., 0.25-µm film) Quadrex 007 FFAP capillary column with p-xylene as an internal standard. (A split injection was used to introduce the sample, and sample detection was accomplished with a thermal conductivity detector [TCD].) These components represent the only significant product, and all other materials detectable by a gas chromatograph/mass spectrometer (GC/MS) were only present at trace levels. Gas samples were also removed hourly and analyzed by GC to ensure that the gas mixture is consistent. The molar quantities of propionic anhydride (npan) formed were determined from the GC data using the following equation:
npan ) Xpan npa0 + nei0 130 (Xei/156) + (Xpa/74) + (Xpan/130) + (Xep/102) (8) where npan ) moles of the component, Xi ) weight fraction of the component (obtained from GC analysis), npa0 ) moles of propionyl initially present, and nei0 ) ethyl iodide initially added, with subscript designations being ei ) ethyl iodide, ep ) ethyl propionate, pa ) propionic acid, pan ) propionic anhydride. This apparatus and procedure readily allow access to studies of the effect of different gas components by changes in the gas composition. The purged system maintains consistent gas compositions above the reactor regardless of the stoichiometry or a bias in the gas absorbed. At very high reaction rates, the reactor system was tested for mass-transfer limitations by adjusting the stirring rate of the autoclave. Mass-
4602 Ind. Eng. Chem. Res., Vol. 36, No. 11, 1997 Table 1. Comparison of Different Halides and Group 6 Metal Hexacarbonyl Catalysts for the Carbonylation of Ethylene to Propionic Anhydridea metal carbonyl halide Bu4P+ salt, mmol rate turnover freq Mo(CO)6 Mo(CO)6 Mo(CO)6 Cr(CO)6 W(CO)6 Figure 2. Fixed-bed microreactor reactor for condensation.
transfer (gas dissolution) limitations typically began to be observable when the rates exceeded 4 (mol/kg)/h, representing 8 (mol of gas absorbed/kg of solution)/h. Therefore, all chemical kinetic measurements in this report have been limited to reactions below 3.5 mol/ (L‚h). Note again that the experimental measurements were intentionally carried out under conditions where reaction rates are outside the mass-transfer limiting regime (ca. 3-4 mol/(L‚h)), to allow us to obtain a wide range of responses to reaction variables. These rates do not reflect limitations in these catalysts. Condensation Catalysis. Catalyst Synthesis. The V-Si-P ternary catalysts were prepared with different atomic ratios following the procedure of Ai (1987, 1990a-c). For example, V-Si-P catalyst with an atomic ratio of 1:12:2.8 was prepared as follows: 23.4 g of NH4VO3 was dissolved in 100 mL of hot water containing 20 mL of lactic acid, and 64.4 g of 85% H3PO4 was dissolved in 100 mL of hot water. The two solutions were added to 480 g of DuPont Ludox colloidal silica (Ludox SM-30), containing 30 wt % silica in water. Excess water was evaporated by stirring. The cake obtained was dried in an oven gradually heating from 50 to 200 °C, at the rate of 1 °C/min. The resulting solid was crushed to an 8-20-mesh size portion and further calcined in air at 350 °C for 4 h and then again at 450 °C for 6 h. The Nb and Ta catalysts were prepared using two different silica supports, viz., a Davison G-59 silica support and a NALCO-1034A colloidal silica support, using conventional metal salt impregnation. NbF5 and Ta3F8 were used as the starting compounds. Catalyst Characterization. The catalysts were analyzed for their surface area, pore volume, and acidbase properties. The surface area and pore volume determinations were made on a Quantchrome NOVA 1000 BET N2 surface area analyzer. The acid-base properties were determined using an Altamira AMI-100 catalyst-characterization instrument. For measurement of the acid sites and strength, NH3 TPD was used. A 10% NH3-N2 gas mixture was used as the treatment or adsorbing gas. A nominally identical charge of 150 mg of catalyst sample was used for all the measurements. NH3 adsorption was carried out with 10% NH3N2 gas mixture (25 mL/min) for 30 min at 50 °C. The desorption was carried out from 50 to 550 °C at 10 °C/ min. The TCD response was continuously monitored. For basicity measurements, an identical temperature, time, and flow profile was followed, except a 10% CO2N2 gas mixture was used as the treatment gas. Catalyst Testing. The V-Si-P, Ta, and Nb catalysts were tested in a continuous microreactor system (Figure 2). A nominally identical charge of 15, 10, and 5 g of catalyst was used for comparative experiments, corresponding to space velocities of 290, 600, and 900 cm3/(g catalyst‚h). The nominal flow rates of propionic anhydride, formaldehyde, and nitrogen were kept at 40,
I Br I I I
40 40 72 72 72
1.39 1.04 2.22 0.18 0.14
41 31 71 6 5
a Conditions: metal hexacarbonyl, 22 mmol; EtX (X ) halide), 0.7 mol; EtCOOH, 7.5 mol; Bu4PX (quantity as indicated), 160 °C, 55 atm. Gas composition: 5% H2, 45% C2H4, 50% CO. Rate ) (mol of (EtCO)2O/kg of initial solution)/h; turnover frequency ) (mol of (EtCO)2O/mol of metal carbonyl)/h.
20, and 220 mmol/h. For propionic acid and methyl propionate feeds, the mole ratio of propionyl to formaldehyde was approximately 4. The feed was prepared by dissolving 1,1,1-trioxane (a trimer of formaldehyde, solid at room temperature) into the substrate propionyl and fed to a preheater (controlled by TC1 in Figure 2) using a syringe pump (ATI Orion SAGE M361). Trioxane was used as a simple and relatively safe source of dry formaldehyde in the laboratory. The vaporized feed was passed over the catalyst charge, located centrally in a 6-mm i.d. × 356-mm-length stainless steel (SS) 316 reactor tube. The catalyst charge was held in place by glass wool and beads. Both the preheater and the reactor were mounted in a horizontal configuration in a Lindberg furnace (Blue M, Model TF55035A). A thermocouple (TC2 in Figure 2) mounted directly at the catalyst charge provided temperature control and readout. The product vapors were cooled in a water condenser. Permanent gases were analyzed on-line. Liquid products were analyzed off-line. Product Analysis. For gas analysis, a fixed-volume loop injection onto a Poropak T/molecular sieve 5A column isolation sequence in conjunction with a TCD was used. For liquid analysis, a fused silica capillary column with a 1-mm film thickness of DB-wax was used, in conjunction with a flame ionization detector (FID). Results and Discussion Propionate Synthesis. General Description of the Catalyst. The catalyst system consists of a group 6 metal, a halide salt, and an organic halide, although materials that generate the organic halide or the halide salt in situ may be used as well. Table 1 shows that the observed order of reactivity for the group 6 metal component is Mo >> W > Cr and that the halide employed may be either Br or I. Substitution of bromine for iodine only leads to a small change (∼25% decrease) in the reaction rate, suggesting an electron-transfer process (Huber et al., 1995). (Presumably the chloride would also be useful, but the volatility and toxicity of EtCl precluded it from our investigation.) The behavior of the catalyst is quite different as the nucleophilic component of the process is changed. Figure 3 shows a typical Mo-catalyzed process in which propionic acid is added to generate propionic anhydride (i.e., R ) EtCO in reaction 7), yielding nearly linear production of propionic anhydride over a large span of time. However, when water is substituted as the primary nucleophile (R ) H in reaction 7), the reaction is initially very slow but accelerates as the reaction proceeds. (Similar observations are made if methanol is added to attempt to generate methyl propionate.)
Ind. Eng. Chem. Res., Vol. 36, No. 11, 1997 4603 Table 2. Comparison of Various Salts or Potential Salt Precursors for the Carbonylation of Ethylene to Propionic Anhydridea
Figure 3. Comparison of Mo-catalyzed processes for propionic acid and propionic anhydride. Catalyst composition: EtI, 0.7 mol; Mo(CO)6, 22 mmol; Bu4Pl, 40 mmol. Gas composition and pressure: 5% H2, 50% C2H4, 45% CO; 55 atm. Temperature/ reactants/solvent: (a) propionic acid (EtCOOH) process, 190 °C, 4.5 mol of H2O, AcOH, 7.5 mol (solvent); (b) propionic anhydride ([EtCO)2O) process, 160 °C, 7.5 mol of EtCOOH (no solvent).
entry
salt or group 15 promotor
rate, (mol (EtCO)2O/ kg init soln)/h
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Bu4PI Et4PI Ph4PI Bu4NI NaI KI CsI Ph3P Ph3As Ph3N pyridine 4-methylimidazole 2,2′-dipyridyl (n-C8H15)3P (n-C8H15)3PdO Ph3PdO Bu3P (cyclo-C6H11)3P Ph2P(CH2)2PPh2 Ph2P(CH2)3PPh2 Ph2P(CH2)4PPh2 Ph2P(CH2)5PPh2
1.39 1.51 1.95 1.05 0.79 0.66 0.15 2.20 0.03 0.18 2.20 1.52 0.13 2.20 0.20 0.09 2.20 2.49 1.18 1.23 1.48 1.46
a Conditions: Mo(CO) , 22 mmol; EtI, 0.7 mol; EtCOOH, 7.5 6 mol; 160 °C, 55 atm.; salt or group 15 promotor ) 40 mmol for entries 1-18, 20 mmol for entries 19-22 (an additional 40 mmol of EtI added for none alkylated materials in entries 8-22. Gas composition: 4% H2, C2H2, 45% C2H4, 50% CO.
Figure 4. Reaction profile for the carbonylation of ethylene to propionic anhydride. Initial composition: EtI, 0.7 mol; EtCOOH, 7.5 mol; Mo(CO)6, 22 mmol; Bu4PI, 40 mmol. Conditions: 160 °C, 55 atm. Gas compositions: 5% H2, 50% C2H4, 45% CO.
This catalyst is not readily applicable to the carbonylation of non-olefinic substrates. For example, attempted carbonylations of methanol, methyl acetate, benzyl alcohol, and benzyl acetate all proceed very sluggishly, if at all, even under the best of conditions for these catalysts. Olefins other than ethylene undergo carbonylation but proceed at slower rates, yielding a mixture of carboxylic acid isomers. This study focused on the propionic anhydride process due to its apparently simpler kinetic behavior. When the process is operated with a catalyst composed of Bu4PI, Mo(CO)6, and EtI, the Mo-catalyzed carbonylation of ethylene to propionic anhydride occurs at a nearly linear rate with time, until a 75-85% conversion of the propionic acid is achieved, at which point the reaction begins to slow markedly. A typical reaction profile for this process appears in Figure 4. Unlike methanol carbonylation to acetic acid, there is little published information on the Rh- and I-catalyzed hydrocarboxylation to propionic acid, and there is virtually no information on the generation of an anhydride or an ester. Direct comparisons of the novel catalysts (Mo) reported here to noble metal catalysts are therefore not possible for propionate synthesis. For generation of acetates, the comparisons will be invalid since reaction proceeds via an entirely different mechanism. Effect of the Cation and Halide. The cation incorporated in the salt component is important and incompletely understood. A variety of materials were
tested, and these differences are evident from Table 2. Omitted from Table 2 is an entry for LiI, which initially produces a small quantity of product but was incapable of sustaining the reaction for even an hour. Therefore, measurable rates were not readily attainable with this salt. It was initially presumed that the phosphines and amines in this study underwent alkylation by reaction with ethyl iodide to form the corresponding quaternary salt. Subsequent 1H, 13C, and 31P nuclear magnetic resonance (NMR) studies of the product solutions indicated that this was true for all the monophosphines, while the amines were primarily protonated species. However, for the diphosphines (entries 19-22), there is no evidence of a diagnostic ethyl group bound to phosphorous, except in the case of 1,5-bis(diphenylphosphino)pentane (entry 22). Salts based on N and P (either added as the salt or generated in situ) generally have the highest rates, although some free phosphine or amine ligand may have been present, which would have altered the Mo activity. However, in the above-mentioned NMR studies, we saw no exchange of alkyl groups (ethyl for butyl) or free phosphine when Bu4PI was used (entry 1). Therefore, there is no evidence to support a Mo species with either a N- or P-based ligand, at least when using the monodentate precursors. Effect of Temperature and Determination of the Activation Parameters. The effect of temperature was measured between 130 and 170 °C using identical levels of gas and catalyst components throughout the full range of temperatures. The apparent activation energy (Eact) was determined from an Arrhenius plot (Figure 5, R2 ) 0.99) to be 39.3 kcal/mol. The rate expression is
k)
rate × [PCO]1.17 [Mo(CO)6]0.62[EtI]0.52[I-]
(9)
Equation 9 represent the experimentally determined
4604 Ind. Eng. Chem. Res., Vol. 36, No. 11, 1997
activation of HCHO by acidic sites of the catalyst HCHO + H+ f H2C+OH
(12)
reaction of two intermediate molecules to form an aldol, followed by dehydration R-CHX + H2C+OH f H2C(OH)C(H)RX f H2CdCRX + H2O (13)
Figure 5. Arrhenius plot. Initial composition: EtI, 0.7 mol; EtCOOH, 7.5 mol; Mo(CO)6, 22 mmol; Bu4PI, 40 mmol. Pressures: ethylene, 27.2 atm; CO, 23.8 atm, H2, 3.4 atm. Table 3. Rate as a Function of Temperaturea T, °C
measd rate, mol/(kg‚h)
calcd k′, J/(mol‚s)
130 140 150 160 170
0.04 0.17 0.58 1.39 3.49
6.56 27.9 95.1 228. 572.
a Initial composition: EtI,0.7 mol; EtCOOH, 7.5 mol; Mo(CO) , 6 22 mmol; Bu4PI, 40 mmol. Pressures: ethylene, 27.2 atm; CO, 23.8 atm, H2, 3.4 atm.
rate orders for Mo(CO)6, Bu4PI, EtI, and CO as determined from the best fit of the data obtained in the studies of the gas and catalyst component dependencies. The enthalpy of activation (∆Hq) and entropy of activation (∆Sq) were determined from the Eyering plot [ln (k/T vs ln (l/T)]: ∆Hq was found to be +38.4 kcal/mol, and ∆Sq was estimated to be +40 (cal/mol)/K. (It is important to indicate that, by keeping the CO pressure and the catalyst components constant between experiments, inaccuracies in the reaction orders used within eq 9 have no effect upon the measured ∆Hq and only a negligible effect upon the magnitude of ∆Sq.) The raw data for this determination appear in Table 3. These activation parameters suggest a mechanism. The large, positive ∆Sq is indicative of a rate-determining dissociative step. Further, the most recent (and reliable) estimate of the dissociation energy of the MoCO bond in Mo(CO)6 (eq 10) is about 40 kcal/mol (Ehlers
Mo(CO)6 h Mo(CO)5 + CO
(10)
and Frenking, 1993, 1994). This is almost identical to the Eact determined for our tests and, along with the positive ∆Sq, indicates that the rate-limiting step is the dissociation of CO from Mo(CO)6. The experimentally determined differences in Eact and ∆Hq are also consistent with this dissociation process. In a process that generates a mole of gas, as in eq 10, the expected difference between these two values would be 0.84 kcal/ mol at 150 °C, which coincides with our observed difference. Condensation Catalysis. The mechanism for the condensation reaction is as follows (Ai, 1996):
activation of the reactant by basic sites of the catalyst RCH2X + B f R-CHX + BH+ (B ) O- or OH-, R ) CH3, X ) COOCH3) (11)
Conceptually, there are advantages to using the carboxylic acid anhydrides in the condensation reaction with formaldehyde to yield the corresponding R- and β-unsaturated acid anhydrides. For example, the condensation of formaldehyde with a carboxylic acid (or an ester) generates 1 mol of water as shown by
RCH2COOR′ + HCHO f RC(dCH2)COOR′ + H2O (14) The liberated water inhibits the further reaction of formaldehyde and the carboxylic acid and, in the case of ester feedstocks, leads to the hydrolysis of ester. When the reaction is run using an acid anhydride, water is consumed in a subsequent reaction to form free acid, with the overall condensation reaction represented by
(RCH2CO)2O + HCHO f RCH(dCH2)COOH + RCH2COOH (15) The product R- and β-unsaturated acids may then be separated from the acid product. One further advantage of using the anhydride is that the saturated acid coproduct in eq 15 may be used to regenerate the corresponding carboxylic acid anhydride by any of the several known processes (Cook, 1993; Forster and Hershman, 1976; Gresham and Brooks, 1950; Reppe and Schweckendiek, 1953; Rizkalla, 1984). Two particularly useful approaches for acetic anhydride and propionic anhydride are shown respectively in
CH3COOH + CH3OH f CH3COOOCH3 (+CO) f (CH3CO)2O (16) CH3CH2COOH + CO + CH2dCH2 f (CH3CH2CO)2O (17) Unfortunately, as noted by Holmes (1978), the reaction of anhydrides with aliphatic aldehydes (including formaldehyde) in the condensed phase generally leads not to condensation but to the formation of 1,1-dicarboxylates according to
(RCH2CO)2O + R′CHO f (RCH2CO)2CHR′
(18)
As a consequence, the condensation of anhydride with aldehydes has seen limited application outside of the well-documented condensation of aromatic aldehydes (which do not form diesters readily) with formaldehyde, which is known as the Perkin reaction. The catalysts that prevent the production of the diester or operate on the diester to produce the corresponding R- and β-unsaturated acids are thus the exact bill for the condensation of anhydrides with aliphatic aldehydes. Despite these potential advantages, an exhaustive search indicates that only one reference has alluded to the catalysts capable of this transformation
Ind. Eng. Chem. Res., Vol. 36, No. 11, 1997 4605 Table 4. Optimization of the V-Si-P Ternary Oxide Catalysta catalyst
atomic ratio
surf area (BET N2), m2/g
MAA yieldb
MAA yieldc
V-Si-P V-Si-P V-Si-P V-Si-P V-Si-P
1:12:2.8 1:10:2.8 1:2.8:2.8 1:3.57:1 1:10:10
96.5 94.2 24.5 114.2 3.24
52.7 55.8 38.4 38.9 9.46
22.6 24.2 16.5 16.6 4.06
a Reaction conditions: T ) 300 °C, P ) 2 atm (30 psi in-house nitrogen), mole flow rates of propionic anhydride:formaldehyde: nitrogen ) 41:17:220 mmol/h. b Yield based on charged HCHO. c Yield based on charged propionic anhydride.
Table 5. Condensation of Propionates with Formaldehydea MMA yield, % type of space velocity, propionyl cm3/(g cat‚h) based on HCHO based on propionyl PPAb PPA PAA PAc PA PA MPd MP MP
290 480 900 320 480 900 320 320 900
76.8 36.4 30.1 60.5 55.3 45.8 22.2d 21.7 9.61
36.4 16.7 14.7 12.9 11.3 9.82 4.83 4.72 2.08
a Reaction conditions: T ) 300 °C, P ) 2 atm (30 psi in-house nitrogen). b PAA ) propionic anhydride; nominal mole flow rates of PAA:formaldehyde:nitrogen ) 41:17:220 mmol/h. c PA ) propionic acid; PA:formaldehyde:nitrogen ) 72:15.5:220 mmol/h. d MP ) methyl propionate; MP:formaldehyde:nitrogen ) 61:13.3:220 mmol/h; product contains both MMA and MAA; MMA/(MMA + MAA) ) 0.69.
(Holmes, 1978). The literature on the synthesis of MAA and MMA (Ai, 1987, 1989, 1990a-c) deals primarily with the condensation of propionic acid and methyl propionate with formaldehyde, and not propionic anhydride. As illustrated in reactions 2-4, condensation of propionic anhydride does not produce water (which can degrade the product and inhibit the reaction) and thus produces an easily separable product mix; our research effort has been initially focused on the anhydride. The V-Si-P ternary oxides and the Ta and Nb/Si catalysts discovered in this work are thus a class of catalysts capable of condensing a number of anhydrides and acids into the corresponding R- and β-unsaturated acids. Over 80 acid-base catalysts were tested in the fixedbed microreactor system (Figure 2). Out of this family of catalysts, V-Si-P ternary metal oxide catalysts and niobium and tantalum series of catalysts were found to give the highest MAA yields and were chosen for detailed catalyst optimization, long-term deactivation, and regeneration studies (Spivey et al., 1995; Gogate et al., 1996b, 1997). To optimize the catalyst atomic ratio in the V-Si-P ternary metal oxide catalysts, five different catalysts were prepared following the procedure in the Experimental Section. The condensation of propionic anhydride with formaldehyde was carried out over these catalysts, and the results are shown in Table 4. The most active catalyst [V-Si-P ) 1:10:2.8] was selected for runs with propionic acid and methyl propionate. The catalyst activity was defined as the yield of MAA based on propionyl moiety and formaldehyde fed to the reactor. The activity of five different V-Si-P catalysts for the condensation of formaldehyde with propionic anhydride is summarized in Table 4. The results summarized in Table 5 compare the yields of MAA at nominally identical operating conditions. The yields (moles of MAA product/mole of propionate and
HCHO charged) are higher for propionic anhydride and propionic acid than the ester, methyl propionate, indicating that while the acid and anhydride are relatively reactive, the ester is difficult to condense (Gogate et al., 1996a). A thermodynamic analysis (Shreiber et al., 1996) has shown that there are no thermodynamic limitations for the condensation of the ester, however. Since the MAA yields based on methyl propionate were lower than either the anhydride or the acid, further reaction studies did not focus on methyl propionate. Propionic acid and propionic anhydride were chosen as substrates for the condensation reactions with formaldehyde. Note that the MAA yield is defined here as based on charged reactants and does not account for unreacted material. Measurement of Acid-Base Properties. The mechanism of condensation (reactions 11-13) shows that both acid and base properties are needed for an active and selective catalyst, suggesting a correlation between the acid-base site distribution and the condensation yields (Spivey et al., 1996; Gogate et al., 1997). This correlation was explored by measuring the acid-base properties of the catalysts tested for the condensation reaction. The results were quantified in terms of micromoles (NH3 or CO2) adsorbed/g of catalyst and are summarized in Table 6. For all experimental runs, chemical selectivity to MAA was 99+%, making reactant conversion and product yield numerically identical. The characteristic NH3 and CO2 TPD patterns for each of the five catalysts in Table 4 have been compared in Figures 6 and 7. The high-temperature desorption peaks correspond to high-strength acid and base sites. Figure 6 gives the NH3 TPD patterns (measuring acidity) and Figure 7 gives the CO2 TPD patterns (measuring basicity). The optimum 1:10:2.8 V-Si-P and 10% TaNALCO Si catalysts do not exhibit a prominent hightemperature NH3 and CO2 desorption peak in the range of 573-823 K, while 1:2.8:2.8 and 1:10:10 V-Si-P show characteristic high-temperature desorption peaks for both NH3 and CO2. The presence of a high-temperature peak, i.e., strong acidic and basic sites, does not correspond to high condensation yields. This suggests that the high-strength acid and base sites react irreversibly with the reactant molecules (propionic anhydride, propionic acid, and formaldehyde). Although the hightemperature peak is also present in 2.8:10:2.8 V-SiP, the peak is not as strong as that for 1:2.8:2.8 and 1:10:10 V-Si-P. Further, the 2.8:10:2.8 V-Si-P catalyst also has a prominent acid-base peak in the medium-temperature region, corresponding to 423-573 K. The data in Table 6 suggest that the presence of strong acid sites (high-temperature peak [(573 < T < 823 K]) limits the condensation yields. The site distribution of both acid and base sites as a function of temperature (i.e., as a function of strength) can be quantified by deconvoluting the TPD spectrum (NH3 TPD pattern for distribution of acid sites as a function of temperature and CO2 TPD pattern for the distribution of base sites as a function of temperature). The distribution of the catalyst sites of a particular type (acid or base), as a function of temperature range, can be quantified by a ratio of the form (Kanno and Kobayashi, 1994)
q ) A(desorbed from 323 < T < 573 K)/ A(desorbed from 323 to 823 K) (19) where A(desorbed at 323-823 K), the denominator of
4606 Ind. Eng. Chem. Res., Vol. 36, No. 11, 1997 Table 6. Effect of Catalyst Acid-Base Properties on Methacrylic Acid Yielda
catalyst (atomic ratio)
surf area (BET N2), m2/g
total acidity, µmol NH3/g cat
total basicity, µmol CO2/g cat
MAA yieldb
MAA yieldc
q ratio
V-Si-P (1:12:2.8) V-Si-P (1:10:2.8) V-Si-P (1:2.8:2.8) V-Si-P (1:3.57:1) V-Si-P (1:10:10) Ta (10%)
96.5 94.2 24.5 114.2 3.24 132.1
155.7 150.8 124.5 174.9 81.8 24.52
100.2 96.5 99.8 59.1 74.0 26.0
52.7 55.8 38.4 38.9 9.46 46.4
22.6 24.2 16.5 16.6 4.06 19.9
0.39 0.43 0.08 0.27 0.06 0.28
a Reaction conditions: T ) 300 °C, P ) 2 atm (30 psi in-house nitrogen), mole flow rates of propionic anhydride:formaldehyde:nitrogen ) 41:17:220 mmol/h. b Yield based on charged HCHO, i.e., mole of MAA/mole of charged HCHO × 100. c Yield based on charged propionic anhydride.
Table 7. Nb-Si Catalysts for Condensation Reactionsa catalyst (atomic ratio) 1:99 (1%) 2:98 (2%) 5:95 (5%) 10:90 (10%) 20:80 (20%) V-Si-P (1:10:2.8) (comparison)
MAA yield, % surf area, surf acidity, q m2/g µmol/g ratio HCHO PA 110.9 100.3 119.1 86.5 NA 96.5
14.9 23.2 28.6 27.3 34.8 150.8
0.763 0.655 0.801 0.806 0.746 0.444
22.9 NAb 58.2 58.6 71.0 47.0
4.91 NA 12.7 12.5 15.2 10.1
a Reaction conditions: T ) 300 °C, P ) 2 atm (30 psi in-house nitrogen), mole flow rates of propionic acid formaldehyde:nitrogen ∼ 72:16:220 mmol/h, 5 g of catalyst charge, ∼1080 cm3/(g catalyst‚h). b NA ) not available.
Figure 6. NH3 TPD spectra of V-Si-P and 10% Ta-Si catalysts. (‚‚‚) 1:12:2.8 V-Si-P, (s) 1:2.8:2.8 V-Si-P, (- - -) 10% Ta on NALCO, (- -) 2.8:10:2.8 V-Si-P, (- ‚ -) 1:10:10 V-Si-P.
Figure 7. CO2 TPD spectra of V-Si-P and 10% Ta-Si condensation catalysts. (‚‚‚) 1:10:2.8 V-Si-P, (s) 1:2.8:2.8 V-Si-P, (- - -) 10% Ta on NALCO, (- -) 2.8:20:2.8 V-Si-P, (- ‚ -) 1:10: 10 V-Si-P.
eq 19 above, is the total area under the TPD spectrum. The results of the deconvoluted CO2 TPD profiles in terms of the q ratio have also been summarized in Table 5. The q ratios for the V-Si-P series of catalysts (1:12: 2.8, 1:10:2.8, and 1:3.57:1) are 0.388, 0.426, and 0.271, respectively. The q ratio for the 10% Ta-NALCO SiO2 catalyst is also high, at 0.278. The lower q ratios contribute directly to lower condensation yields as evidenced by catalyst 1:10:10 V-Si-P. The yield of the catalyst 1:2.8:2.8 V-Si-P is also high at 38.4%, although this catalyst has a characteristically low q value
at 0.086. The catalyst 1:10:10 V-Si-P which gave the lowest MAA yield, has both a low q ratio (of 0.063) and a very low surface area (3.24 m2/g). Thus, a direct correlation between the surface area, the q ratio, and the MAA yield is observed from these observations. Studies on Niobium Catalysts. A Nb series of catalysts were also tested for their condensation activity in a fixed-bed continuous-flow microreactor system (Figure 2). Five different metal catalyst loadings from 1% to 20% were synthesized. The surface area, surface acidity (micromoles of NH3/gram of catalyst), and q ratio (fractional surface acidity of the low-strength type) were measured using BET N2 and NH3 TPD (Table 7). Twenty percent Nb/Si catalyst gives the best result in terms of condensation yield, with the MAA yield approaching 71% (based on charged formaldehyde). The MAA yield based on PA is at 15.2%. For comparison, the V-Si-P catalyst gives MAA yields of 47% and 10.1%, respectively. [While the Ta catalyst gives the yields of 46% and 19.9%, these numbers are with propionic anhydride as the feed and therefore are not a direct comparison of the catalyst performance because the feed (propionic acid) is different for the Nb/Si catalyst.] Nevertheless, the MAA yield for the Nb series of catalysts is higher for the V-Si-P and Ta catalysts. Because 20% Nb/Si gave the maximum MAA yield, corresponding analogues of two other group 5 metals, V and Ta, were synthesized: a 20% V/Si and a 20% Ta/ Si catalyst. The results have been summarized in Table 8, which also shows the 1:10:2.8 V-Si-P catalyst for comparison. The 20% Nb/Si catalyst has a significantly higher MAA yield than the other three. The surface area, the q ratio, and the surface acidity of the Nb series of catalysts suggest that, while high values of surface area and q ratio are clearly desirable, they still do not completely determine the MAA yield. Further studies are underway to investigate these correlations. A high overall surface acidity is suspected to cause the rapid deactivation for the V-Si-P catalyst (Figure 8), and the Nb catalysts preserve their activity at least 3 times as long as the V-Si-P catalysts (Figure 9).
Ind. Eng. Chem. Res., Vol. 36, No. 11, 1997 4607 Table 8. Group 5 Catalyst for Condensation Reactionsa catalyst (atomic ratio) Nb-Si (1:4, 20%) Ta-Si (1:4, 20%) V-Si (1:4, 20%) V-Si-P (1:10:2.8)
MAA yield, % surf acidity, surf basicity, q µmol/g µmol/g ratio HCHO PA 34.8 34.2 54.9 150.8
32.6 NA NA 96.5
0.75 0.66 0.74 0.44
71.0 26.1 14.1 47.0
15.2 6.0 3.0 10.1
a Reaction conditions: T ) 300 °C, P ) 2 atm (30 psi in-house nitrogen), mole flow rates of propionic acid formaldehyde:nitrogen ∼ 72:16:220 mmol/h, 5 g of catalyst charge, ∼1080 cm3/(g of catalyst‚h). NA ) not available.
Figure 9. Long-term activity check on 1:10:2.8 V-Si-P, 10% TaSi, and Nb-Si catalysts. For V-Si-P, Ta-Si: 300 °C, 2 atm, 15 g of catalyst charge, 41:17:220 mmol/h. PAA:HCHO:nitrogen, 290 cm3/(g of catalyst‚h). For Nb/Si: 300 °C, 2 atm, 5 g of catalyst charge, 72:16:220 mmol/h. PA:HCHO:nitrogen, 1080 cm3/(g of catalyst‚h).
Figure 8. Long-term deactivation pattern of V-Si-P and Ta catalysts.
Deactivation of the V-Si-P (1:10:2.8), Ta/Si, and Nb/Si Catalysts. To investigate the long-term activity of the 10% Nb/Si (using propionic acid) and the V-Si-P (1:10:2.8) and 10% Ta/Si catalysts (using propionic anhydride), an extended experiment was carried out with all three catalysts. Figure 8 compares the V-Si-P and Ta/Si catalysts. The initial MAA yield (based on charged HCHO) is higher for the V-Si-P catalyst, at 56%, while for the 10% Ta/Si catalyst it is 46.4%. After a 180-h period, the catalyst activity drops significantly for both catalysts. For the V-Si-P catalyst, the MAA yield dropped to 2.74%. For 10% Ta/Si, the activity drops to 28%. The catalyst deactivation could be caused by the loss of acidity by hydroxyl groups, loss of surface area and/or porosity, and carbon deposition due to coking. These causes were investigated as follows. 1. Catalyst Hydration. If the loss in the catalyst activity were due to the loss in the surface acidity, it could be due to the dehydration of the surface, which could be restored by rehydration. The deactivated V-Si-P catalyst was steam-treated by exposure to a 33% steam/66% N2 atmosphere at a space velocity of 290 cm3/(g catalyst‚h). The catalyst activity was then checked. The results (Figure 7) show that this was ineffective in restoring the catalyst activity for the V-Si-P catalyst. This also suggests that the loss in catalyst activity may not be due to the loss of surface hydroxyl groups. 2. Surface Area. The bulk surface areas of the fresh and deactivated catalysts did not reveal any loss of bulk surface due to deactivation. The surface area of 1:10: 2.8 V-Si-P and Ta/Si increased marginally from 94.2 and 132.1 m2/g to 113.4 and 150.7 m2/g, respectively, as a result of testing under the reaction conditions summarized in Figure 7. (The bulk surface area of Nb/ Si was not checked following long-term testing under propionic acid feed conditions.) This suggests that loss of surface area did not account for the deactivation. 3. Oxidative Regeneration at 300 °C. The deactivated catalyst, which had been hydrated, was treated at 300 °C, in flowing air (ca. 290 cm3/(g catalyst‚h). The MAA yield (based on charged HCHO) was restored to 40.8%, 78% of its original activity. For the 10% Ta/Si
Figure 10. TGA test on deactivated 10% Nb-Si catalyst. Coke burnoff occurs at ∼370 °C.
catalyst, the regeneration restored the catalyst activity from 28% to about 41.1% yield of MAA, 88% of the original. However, the deactivation pattern continues after the oxidation regeneration for both V-Si-P and Ta catalysts (Figure 8). The tantalum catalyst is more stable and has a higher on-stream life, based on these catalyst long-term activity studies. We also conclude that coke was the primary cause of deactivation on these two catalysts and that oxidation does not completely regenerate the catalyst surface. The steady loss in activity also seems to continue, after regeneration. The long-term activity of the Nb catalyst was checked with propionic acid as the feed (Figure 9). (The results for the V-Si-P and Ta catalysts are also given for comparison, but recall that these catalysts were tested with propionic anhydride.) The yield of MAA based on charged formaldehyde declines from ca. 58% to ca. 20%, over a 180-h test period. Coke burnoff at 300 °C in flowing air (ca. 290 cm3/(g catalyst‚h) does not restore the catalyst activity as it did for the V-Si-P and Ta catalysts, suggesting that either the nature of the coke and/or its content are different for the Nb catalyst than for the V-Si-P or Ta catalyst. A TGA test on the deactivated Nb catalyst (Figure 10) shows that coke burnoff occurs at 370 °C; therefore, the 300 °C regeneration temperature was not sufficient for coke oxidation and burnoff. Coke loading is also higher for the Nb case (ca. 12-13%) compared to the V-Si-P case (ca. 6 %), which explains the difficulty in regeneration. Further work correlating the acid-base properties, MAA yield,
4608 Ind. Eng. Chem. Res., Vol. 36, No. 11, 1997
and long-term activity on the V-Si-P, Nb, and Ta family of catalysts is currently under way. Although these condensation catalysts do suffer from deactivation, several measures to alleviate aging are currently being studied, including reaction-regeneration cycles and incorporating stabilizing ions in the catalyst matrix. Conclusions Propionate Synthesis. The homogeneously catalyzed carbonylation of olefins, particularly ethylene, can be carried out with a halide-promoted Mo(CO)6 catalyst. This represents the first efficient carbonylation process using a Cr group metal as the active catalytic species. Detailed mechanistic examinations suggest that the reaction proceeds via a free-radical pathway which is initiated by a rate-limiting dissociation of CO from Mo(CO)6. This initial dissociation is likely followed by a subsequent radical generation step involving the interaction of EtI with the coordinatively unsaturated species, Mo(CO)5, to ultimately form ethyl radicals. The resultant ethyl radicals would be captured by Mo(CO)6 to form the pivotal complex [•Mo(EtCO)(CO)5], which allows entry into a chain propagation sequence which is responsible for catalysis. The addition of iodide to the complex [•Mo(EtCO)(CO)5] generates EtCOI, a precursor to the propionate derivatives and allows eventual regeneration of the pivotal complex, [•Mo(EtCO)(CO)5]. The specific propionate derivative is determined by the nature of the nucleophilic component. Condensation Catalysis. The V-Si-P and Tasilica catalysts are active and selective for the condensation of propionic anhydride with formaldehyde. The Nb-Si catalyst is also active for the condensation of propionic acid with formaldehyde. Measurement at the acid-base properties of these catalysts shows that catalysts that exhibit higher condensation yields have characteristic low-temperature acid and base TPD desorption peaks in the temperature range of 323-573 K and an absence of a high-temperature peak in the range of 573-823 K. The q ratio, which is a measure of the strength of weak acid/base sites, is directly related to the MAA yield; the higher the q ratio (i.e., the higher the proportion of weak acid/base sites), the higher the MAA yield. The 1:10:2.8 V-Si-P and Ta/Si catalysts have high surface areas (94.2 and 132.1 m2/g, respectively) and a high q ratio. Both exhibit high MAA yields. The activation energy of the high-strength acid sites is estimated to be 25 kcal/mol. The estimations of the acid-base correlations and activation energies will be used to “design” a second generation of more active condensation catalysts. Acknowledgment We thank the U.S. Department of Energy (DOE), Pittsburgh Energy Technology Center, and Eastman Chemical Company for their shared support of this work under DOE Contract No. DE-AC22-94PC94065. Literature Cited Ai, Ai, Ai, Ai, Ai,
M. M. M. M. M.
J. Catal. 1987, 107, 201. Appl. Catal. 1989, 54, 29-36. Appl. Catal. 1990a, 63, 39. J. Catal. 1990b, 124, 293. Appl. Catal. 1990c, 63, 29.
Ai, M. Vapor Phase Condensation Reactions Using Formaldehyde or Methanol. Catalysis; The Royal Society of Chemistry, Athenaeum: London, 1996; Vol. 12, Chapter 5, pp 152-198. Bailey, O. H.; Montag, R. A.; Yoo, J. S. Appl. Catal. A: Gen. 1992, 88, 163. Bertleff, W. Carbonylation. In Ullman’s Encyclopedia of Industrial Chemistry, 5th ed.; VCH: New York, 1986; Vol. A5, p 223. Colquhoun, H. M.; Thompson, D. J.; Twigg, M. V. CarbonylationsDirect Synthesis of Carbonyl Compounds; Plenum: New York, 1991; pp 102-106, 119-130. Cook, S. L. Acetic Anhydride. In Acetic Acid and Its Derivatives; Agreda, V. H., Zoeller, J. R., Eds.; Marcel Dekker: New York, 1993; Chapter 9. Ehlers, A. W.; Frenking, G. J. Chem. Soc., Chem. Commun. 1993, 1709 and references cited therein. Ehlers, A. W.; Frenking, G. J. Am. Chem. Soc. 1994, 116, 1514 and references cited therein. Forster, D.; Hershman, A. U.S. Patent 3,989,751, 1976. Forster, D.; Hershman, A.; Morris, D. E. Catal. Rev.sSci. Eng. 1981, 23, 89. Gogate, M. R.; Spivey, J. J.; Zoeller, J. R. In Proceedings of the ACS National Meeting, Division of Petroleum Chemistry, Syngas Conversion to High Value Chemicals, New Orleans, LA, 1996a; pp 216-219. Gogate, M. R.; Spivey, J. J.; Zoeller, J. R.; Colberg, R. D.; Choi, G. N.; Tam, S. S. Synthesis of Methyl Methacrylate from Coalderived Syngas. Annual Technical Report for DE-AC2294PC94065, U.S. Department of Energy, Pittsburgh Energy Technology Center, Pittsburgh, PA, Dec 1996b. Gogate, M. R.; Spivey, J. J.; Zoeller, J. R. Catal. Today 1997, 36 (3), 243-254. Gresham, W. F.; Brooks, R. E. U.S. Patent 2,497,304, 1950. Holmes, J. D. U.S. Patent 4,085,143, 1978. Huber, T. A.; Macartney, D. H.; Baird, M. C. Organometallics 1995, 14, 592 and references cited therein. Imbeaux, M.; Mestdagh, H.; Moughamir, K.; Rolando, C. J. Chem. Soc., Chem. Commun. 1992, 1678-1679. Kanno, T.; Kobayashi, M. In Acid Base Catalysis II; Hattori, H., Misono, M., Ono, Y. Eds.; Elsevier: Tokyo, Japan, 1994; Chapter 2.14. Lichstein, B. M. U.S. Patent 3,790,607, 1974. McKetta, J. J., Ed. Encyclopedia of Chemical Processing and Design; 1989; Vol. 30. Mullen, A. In New Syntheses with Carbon Monoxide; Falbe, J., Ed.; Springer-Verlag: Berlin, Germany, 1980; pp 275-286. Pino, P.; Piacenti, F.; Bianchi, M. In Organic Syntheses via Metal Carbonyls; Wender, I., Pino, P., Eds.; John Wiley & Sons, Inc.: New York, 1977; Vol. 2, pp 223-296. Reppe, W.; Schweckendiek, W. U.S. Patent 2,658,075, 1953. Rizkalla, N. U.S. Patent 4,483,803, 1984. Samel, U. R.; Kohler, W.; Gamer, A. O.; Keuser, U. Propionic Acid and Derivatives. In Ullman’s Encyclopedia of Industrial Chemistry, 5th ed.; VCH: New York, 1993; Vol. A22, p 223. Shreiber, E. H.; Mullen, J. R.; Gogate, M. R.; Spivey, J. J.; Roberts, G. W. Ind. Eng. Chem. Res. 1996, 35 (7), 2444-2452. Spivey, J. J.; Gogate, M. R.; Jang, B. W. L.; Middlemas, E. D.; Zoeller, J. R.; Tam, S. S.; Choi, G. N. Synthesis of Acrylates and Methacrylates from Coal-derived Syngas. Proceedings of Contractors’ Review Conference on Coal Liquefaction and Gas Conversion, U.S. Department of Energy, Pittsburgh Energy Technology Center, Pittsburgh, PA, 1995; pp 385-395. Spivey, J. J.; Gogate, M. R.; Zoeller, J. R. Proceedings of the ACS National Meeting, Division of Petroleum Chemistry, Symposium of Syngas Conversion to High Value Chemicals, New Orleans, LA, 1996; pp 233-237. Yoo, J. S. Appl. Catal. A: Gen. 1993, 102, 215.
Received for review February 18, 1997 Revised manuscript received July 25, 1997 Accepted August 4, 1997X IE970139R
X Abstract published in Advance ACS Abstracts, October 15, 1997.
