2438
J. Phys. Chem. 1982, 86,2438-2441
Ethanol Formation Mechanism from CO 4- H, Atsushl Takeuchit and James R. Katzer" Center for Catalytic Science and Technology, Department of Chemical Engineering, University of Delaware, Newark, Delaware 1971 1 (Received August 17, 1981; In Final Form: February 11, 1982)
The mechanism of ethanol formation from CO + Hzwas studied by the isotopic tracer method. Enol condensation and CO insertion into surface-bound CH3 does not explain the isotopic composition of the ethanol product. A substitute mechanism involving CO insertion into an adsorbed carbene followed by isotopic scrambling in the adsorbed intermediate is proposed.
Introduction The Fischer-Tropsch synthesis has been carefully and extensively examined recently. Hydrocarbons are produced from CO + H, catalyzed by Co, Ni, and Ru;' Fe and Rh catalyze the formation of oxygen-containing compounds, such as alcohols and aldehydes, in addition to hydrocarbon^.^,^ Pd, Pt, and Ir can exhibit very high selectivity to methanol synthesis from CO H2.4 However, transition metal oxides are highly selective to alcohols synthesis, e.g., oxides of Zn and Cu, and iron oxide catalyze the formation of alcohol^.^ Thus, CO hydrogenation catalysts are roughly classified into these four groups. Alcohols have been considered to be hydrocarbon precursor~.~ The mechanism of methanol formation catalyzed by Rh has been shown to involve a nondissociative CO hydrogen mechanism.' Rh also catalyzes ethanol formation from CO + H2. It is important to elucidate the mechanism of ethanol formation in order to better understand the chain propagation reaction in FischerTropsch syntheses of higher alcohols and hydrocarbons.
TABLE I: Iostopic Composition of CO before and after Reaction with H, Catalyzed b y Rh/TiO," composition, mol % isotopic species lZC160 13C160 lZC180 13C180
+
Experimental Methods An all-glass internal-recycle reactor operated in the batch mode was used. Rh/TiOz (3 w t 70,O.lg) was placed in the reactor and reduced in situ at 473 and 573 K. The catalysts had been used repeatedly for CO hydrogenation prior to this isotopic study in which an approximately 5050 mixture of l2Cl80and 13C160was used. Experimental details are described el~ewhere.~ Isotopic ethanol was analyzed by chemical ionization mass spectrometry to take advantage of the low fragmentation of ethanol that is achievable. Methane was used satisfactorily as the chemical ionization reactant for this analysis, although isobutane was recommended in the literature.8 The ethanol analysis was carried out by gas chromatography/chemical ionization mass spectrometry with a gas chromatograph (Varian Model 2740) and a chemical ionization mass spectrometer (DuPont 21-492 B). The ethanol and methanol were separated on a 6 f t X 'I8 in. glass column of 0.2% Carbowax 1500 on 60/80 mesh Carbopack at 373 K. Much of the He carrier gas was removed by the separator; the remaining He and sample were mixed with CHI reagent gas in the source of the mass spectrometer. Scans of the background ion current preceding and following each experiment showed no interferences at the masses (M + H)+of interest, 47-51. Traces of the individual ions were obtained as functions of time 'Materials Research Laboratory, SRI International, Menlo Park, CA 14025. * Mobil Research and Development Corporation, Central Research Division, Princeton, NJ 08540.
a
wt
for 0% CO convrn
for 48.2% CO conwn
28 29 30 31
2.3 i 0.2 45.1 i 0.2 51.5 0.4 1.1 i. 0.2
11.0 * 41.4 i 38.8 i 8.8 *
mol
*
1.4 4.0
4.0 1.0
Reaction conditions: batch reactor; initial pressures
PCO 7 30.5 torr, P, = 681 torr; temperature = 4 2 3 K, Rh/TiO, Prereduced in H, at 573 K.
TABLE 11: Isotopic Composition of Ethanol Produced by t h e Rh/TiO, Catalyzed Hydrogenation of Labeled CO, 45% l3CI60-52% 1zC'80 composition," mol % ethanol
mol wt
12CH,1ZCH,'60H 46 iZCH;13CH;160H 47 13CH,'ZCH,'60H 13CH;'3CH;'60H 48 12CH,'ZCH.'80H 12CH;13CH;180H 49 '3CH,1ZCH,180H '3CH313CH,'80H 50
for 48.2%
for 98.7%
24.2
f
0.5
23.2
i
0.6
26.1
f
0.4
27.2
i
0.6
20.7
i
0.5
co conwnb co convrnc 18.4 * 1.0 18.6 i 0.8
21.8 c 1.4 9.5
i
0.3
10.3 i 0.2
a Analyzed by DuPont 21-492B chemical ionization mass spectrometer; reactant, methane; GC column, carbowax 1 5 0 0 ; electron energy, 75 eV. Initial reactant pressure of PCO = 30.5 torr, PH = 681 torr; reaction temperature 423 K ; Rh/TiO, prereduced in H, at 573 K. Initial reactant pressure of P c p = 29.8 torr, PH, = 679 torr; temperature 422 K ; Rh/TiO, prereduced in H, a t 473 K.
during each chromatographic analysis and integration of the ion current peaks was carried out. No significant interferences existed for the higher mass species; however, corrections of (M - H)+ions were made to the ion currents of the lighter species. (1) Denny, P. J.; Whan, D. A. 'Catalysis"; The Chemical Society: London, 1978; Vol. 2, Chapter 3. (2) Ichikawa, M. Bull. Chem. SOC.Jpn. 1978,51, 2268, 2273. Katzer, J. R.; Sleight,A. W.; Gajardo, P.; Michel, J.; Gleason, E. F.; McMillan, S. Paper submitted to Trans. Faraday SOC. (3) Henrici-OlivB,G.; OlivB, S. Angew. Chem., Int. E d . Engl. 1976,15, 136. (4)
Poutsma, M. L.; Elek, L. F.; Ibarbia, P. A,; Risch, A. P.; Rabo, J. A. J . Catal. 1978, 52, 157. (5) Stiles, A. B. AIChE J. 1977, 23, 362. (6) Gall, D.; Gibson, E. J.; Hall, C. C. J . Appl. Chem. 1952, 2, 371. (7) Takeuchi, A.; Katzer, J. R. J. Phys. Chem. 1981, 85, 937. (8) Field, F. H. J. Am. Chem. SOC.1970, 92, 2672.
0022-3654/82/2086-2438$0 1.2510 0 1982 American Chemical Society
Ethanol Formatlon Mechanism
The Journal of Physical Chemistry, Vol. 86, No. 13, 1982 2439
Results In one experiment the isotopic mixture of CO (30.5 torr) was reacted with H2 (681 torr) at 425 K catalyzed by Rh/Ti02, prereduced in H2 at 573 K, and the products were analyzed after 48.2% of the CO had been converted. In another experiment the isotopic mixture of CO (29.8 torr) and H2 (679 torr) was reacted at 422 K, catalyzed by Rh/Ti02 which had been reduced at 473 K. Products were sampled and analyzed after 98.7% of the CO had been converted. The product distribution was similar to that observed in previous experiments,' although the olefins had been completely hydrogenated to their corresponding paraffins. Isotopic composition of the CO is shown in Table I at 0 and 48.2% conversion; the isotopic composition for the run having 98.7% conversion of CO could not be obtained because the CO concentration was too low for analysis. There is exchange of l2C180with l60from the catalyst, but, as shown previously, this is slow and has little effect on data analysis.' The isotopic composition of the ethanol produced is summarized in Table 11. The isotopic compositions of the product ethanol for the two experiments are almost identical although the CO conversion and the Rh/Ti02 prereduction temperature were different. Triplicate analyses were done on each sample with satisfactory precision; uncertainties are given in the tables. Because of the necessity for rapid scanning of the chromatographic peaks, it was necessary to operate the mass spectrometer at low resolution; consequently,no distinction can be made between 13C2H5160H and 12C2H6160H. Discussion The most frequently discussed mechanisms for ethanol and higher alcohol synthesis are the condensation reaction of enol intermediates (1)9 (eq 1) and CO insertion into adsorbed alkyls (2)3J0(eq 2). H
H
OH
OH H
OH
TABLE 111: Calculated Isotopic Composition of Ethanol Synthesized from Isotopic CO Mixture According to Fully Dissociative and Partially Dissociative Reaction Mechanisms possible isotopic composition mol of ethanol wt
model Aa
model Bb
A,
A,
B,
B,
IzC + 12C + l6O
46
13.7
13.0
1.2
5.5
'"
47
23.6
26.2
25.4
26.1
+
48
25.3
25.0
48.5
40.1
+
49
26.2
23.8
24.4
23.9
50
11.2
12.0
0.5
4.4
I3C + ',C
+ l6O 'ZC + lzc+ I 8 0 13c "O + + '*O 13C + I3C + +
13c +
lZc
+
13'
+
"'
a Model A, calculated product composition, mol %. Fully dissociative model, adsorbed CO completely dissociates t o adsorbed C and 0 atoms which recombine (with needed H) t o produce ethanol; recombination is assumed to occur statistically; A , is calculated from the isotopic composition of CO at 0% conversion; A, is calculated from the isotopic composition of CO a t 48.2% conversion Model B, calculated product composition, (Table I). mol %. Partially dissociative model, one CO dissociates and the other CO maintains its molecular identity in producing ethanol; B, is calculated from the isotopic composition of CO a t 0% conversion; B, is calculated from the isotopic composition of CO a t 48.2% conversion (Table I).
intermediate in the Fischer-Tropsch synthesis catalyzed by Rh.12 H
O
2
(4) M 2
1
1 H C
OH
M
v 2
c =o
I
M
In the latter case, MCH3 (the adsorbed alkyl (2)) can be produced either by the hydrogenative dehydration of an oxygenated intermediate (eq 3)3J0or by the hydrogenation of a surface carbene formed by the dissociation of adsorbed CO" (eq 4) although, in contrast to Ni and Co, there is a possibility that surface carbon atoms do not act as the only (9)Storch, H.H.;Columbic, N.; Anderson, R. B. "The Fischer-Tropsch and Related Synthesis"; Wiley: New York, 1951. Kummer, J. T.; Emmett, P. H.J. Am. Chem. SOC.1953,75,5177. Hall,W. K.; Kokes, R. J.; Emmett, Pp. H. Ibid. 1960,82,1027. Kummer, J. T.; Podgurski, H. H.; Spencer, W. B.; Emmett, P. H. Ibid. 1951, 73,564. Vannice, M.A. J. Catal. 1976,37,462. (10)Pichler, H.; Schulz, H. Chem. Ing. Tech. 1970,18,1162. (11)Biloen, P.;Sachtler, W. M. H.Adu. Catal. In press.
One CO molecule must dissociate, and the other CO molecule must conserve its molecular and isotopic identity in both the condensation and the CO insertion mechanisms leading to the formation of ethanol. This type of reaction mechanism can be modeled statistically and will here be referred to as the partially dissociative model (model B). This model represents any reaction mechanism in which one CO molecule effectively dissociates and the other maintains its isotopic identity in producing one molecule of product ethanol. It is assumed that there are no isotopic effects on reaction rates that significantly affect the isotopic distribution in the ethanol. The calculated isotopic composition of product ethanol for the partially dissociative model is given in Table 111, model B. The isotopic composition of product ethanol was calculated for the isotopic composition of CO at 0% conversion-(B,) and at 48.2% conversion (B2). A second statistical model referred to here as the fully dissociative model (model A) was proposed and tested. In this model all CO molecules are assumed to completely dissociate to adsorbed C and 0 atoms which then recombine statistically (with needed H) to produce ethanol. This model corresponds to isotopic equilibrium in product ethanol. The calculated isotopic composition of product ethanol is given in Table 111, model A; again composition is calculated for the isotopic CO composition at 0% con(12)Takeuchi, A.;Katzer, J. R. Submitted to J . Catal.
2440
The Journal of Physical Chemlsby, Vol. 86,No. 13, 1982
-
Takeuchi and Katzer
Furthermore, the carbon number distribution of the hydrocarbons and of the alcohols separately did not show good Schultz-Flory-Anderson behavior, but the carbon number distribution of the hydrocarbons plus alcohols did indicating that the product species may have been produced from a common intermediate.12 On the basis of the above information, a reaction scheme can be proposed which explains the formation of CHBOH (eq 5 ) , CHI (eq 6) (and higher hydrocarbons), and ethanol
Experimental
50
H2
46
40 49 MOLECULAR WEIGHT
47
--
2Ho3
50
Flgure 1. Isotopic composition of ethanol; the experimental result is for the experiment with 48.2% CO conversion; A,, A, E,, and E, are defined in Table 111.
version (A,) and at 48.2% conversion (Az). These calculated isotopic compositions of product ethanol are plotted along with the experimental results for 48.2% CO conversion in Figure 1. The experimentally determined isotopic composition is completely different from those of the partially dissociative model (B,+ Bz), and it is difficult to explain this result if the enol condensation or CO insertion mechanisms are operative. One possibility is that the isotopic exchange of CO occurs rapidly on the catalyst surface, resulting in isotopic mixing in product ethanol. However, CO adsorption and desorption occur very rapidly in the presence of gas-phase C0,7J3J4and therefore the presence of isotopic exchange of CO on the surface must be reflected in the isotopic composition of gas-phase CO which was not observed (Table I). However, B2,calculated from the isotopic composition of product ethanol at 48.2% CO conversion, is somewhat different from B1 but still differs markedly from the experimental results. If condensation or CO insertion mechanisms were operative, the experimental results would have been expected to lie between B1and B2. The results, on the other hand, lie very close to the isotopic composition predicted by the fully dissociative model (A, + A,) (Figure 1, Table 11). However, these results do not prove that the physical picture used in constructing the fully dissociative model is operative; other reaction mechanisms could achieve the same net result, but the results cast very serious doubt on the operability of either the enol condensation or alkyl (2)-CO insertion mechanisms in ethanol formation from CO + H2. If complete dissociation of CO occurs followed by recombination in ethanol formation similar behavior might be expected in methanol formation but has been shown not to occur; methanol synthesis occurs by a totally nondissociative mechanism with respect to CO.' If the fully dissociative model were operative, isotopic equilibrium of CO would be expected in contrast to the experimental results (Table I) although it is necessary that the CO dissociation step not be rate determining. (13) Yates, J. T.,Jr.; Duncan, T. M.; Worley, S. D.; Vaughan, R. W. J. Chem. Phys. 1979, 70, 1219. (14) Liu, K.; Katzer, J. R. Paper submitted to J. Catal.
CZH6
10
2
5
11
HjC-CHzOH 8
CH3CHad
9
(eq 8 and 13) (and higher alcohols) catalyzed by supported Rh. We speculate that ethanol formation involves CO insertion into an adsorbed carbene (3, eq 7),15J6instead of CO insertion into an adsorbed alkyl (2, eq 2); this results in the formation of ketene (4). Somorjai has reported the presence of carbene as a stable intermediate on Rh as observed by high-resolution electron energy loss spectroscopy." Herrmann et al. and Wilkinson et al. support the carbene ketene ethanol route.16 Furthermore, oxirene (5) is a known intermediate in organic synthesis and is convertible to ketene (4).18 Ketene (4) and oxirene
-
-
(15) Dorrer, B.;Fischer, E. 0. Chem. Ber. 1974,107,2683. Master, C. Adu. Organomet. Chem. 1979,17,61.
(16) Daroda, R. J.; Blackworow, J. R.; Wilkinson, G. J . Chem. SOC., Chem. Commun. 1980,1098. Herrmann, W. A,; Plank, J. Angeur. Chem., Znt. Ed. Engl. 1978, 17, 525. (17) Somorjai, G. A. 'Chemistry in Two Dimensions: Surfaces"; Cornel1 University Press: Ithica, 1981; p 541.
J. phys. Chem. 1982, 86, 2441-2442
(5) may be hydrogenated respectively to an adsorbed enol (6) and to an adsorbed oxirane (7, (ethylene oxide),& Intermediate 6 could be hydrogenated to give ethanol (8, eq 8). Ethylene oxide can be catalytically hydrogenated to ethanol (8, eq 13).19 Oxirene (5) may be hydrogenated and then hydrated to produce ethylene glycol (11) at high pressure (eq 12).,O Intermediates 6 or 7 may be hydrogenated, then dehydrated to produce 9 (eq 9 and 14),which acta as an intermediate in the formation of C2H4and C2H6 (eq l l ) , or as an intermediate in a chain-propagating step to produce C3+compounds (eq 10). The isotopic composition of model B is predicted by the reaction mechanism involving reactions 1and 2 for ethanol formation; this is not consistent with experimental results. Statistical scrambling of 0 and C may result from the reversibility of eq 7 or the exchange reactivity of the ring species 5 and 7. Oxirene (5) is expected to be more reactive than oxirane (7) (adsorbed ethylene oxide) because 5 contains the C = C moiety and is extremely stained. The 0 atoms in ethylene oxide and propylene oxide are readily transferred to another ethylene or propylene molecule (eq 15) and the 0 readily exchanges between propylene oxide and oxygen (eq 16).,l
2441
Ethanol is produced mainly from CO + H, directly but a small portion of the ethanol is produced by the homologation of methanol.22 This minor route contributes to composition B and may be responsible for the small maximum at molecular weight 48. Thus we conclude that the mechanism of ethanol formation involves insertion of CO into a surface carbene rather than into a surface alkyl group. This leads to the formation of surface intermediates that give isotopic scrambling in the product ethanol. A catalyst that is to produce methanol in high selectivity, e.g., Pd,4 must not readily dissociate CO? A hydrocarbon synthesis catalyst must rapidly dissociate the C-0 bond to give suri'ace carbide, carbene, and alkyl groups which readily lead to hydrocarbon products.ll A catalyst that is to synthesize ethanol and higher alcohols with high selectivity must dissociate the C-0 bond a t only a moderate rate so that the catalyst surface contains undissociated adsorbed CO and surface carbene species for the bimolecular insertion reaction. Thus Rh, which lies between Pd (unable to readily dissociate the C-0 bond and produces methanol in high selectivity) and Ru (readily dissociates the C-0 bond producing exclusively hydrocarbons), shows an intermediate rate of C-0 bond dissociation and exhibits intermediate behavior producing methanol, ethanol, and higher alcohols, and methane and higher hydrocarbons.
Acknowledgment. We are grateful to Dr. G. C. A. Schuit for enlightening discussions, and Dr. B. Munson and P. Rudewicz of the Chemistry Department for chemical ionization mass analyses. Support for the work was from the Center Industrial Sponsors Program and is gratefully acknowledged. (18) Acheeon, R. M. "An Introductionto the Chemistry of Heterocyclic Compounds"; Wiley-Interscience: New York, 1976; p 25. (19) Kirk-Other "Encyclopedia of Chemical Technology";Wiley-Interscience: New York, 1980; 3rd ed, Vol. 9, p 435. (20) Pruett, R. L.; Walker, W. E. US.Patent 3833~634,1974.Kaplan, L. US.Patent 3944588,1976. Pruett, R. L.; Walker, W. E. US.Patent 4 133 776, 1979.
Reduction of the (2
(21) Manara, G.; Parravano, G. J.Catal. 1971,23,379. Zbid. 1974,32, 72. Tezuka, Y.; Takeuchi, T. 2.Phys. Chem. (Frankfurt am Main) 1975, 97, 321. (22) Takeuchi, A.; Katzer, J. R.; Crecery, R. W. to be submitted for publication.
+ 2) Dimer of [8]Annulene To Yield the Anions of [IG]Annulene
Gerald R. Stevenson, James B. Sedgwlck, and Randy Miller Department of Chemlstty, Illinds State University, Normal, Illinois 6176 1 (Received: November 10, 198 1)
The addition of the (2 + 2) dimer of cyclooctatetraeneto the dianion of cyclooctatetraene (cot) in HMPA results in a solution that yields the ESR signal for both the anion radicals of cot and [16]annulene simultaneously. However, the electron transfer from the cot dianion to form the [16]annuleneanion radical is not quantitative, and some polymeric material is formed. The formation of the polymeric material also takes place when the dimer is reduced in HMPA or in tetrahydrofuran via alkali metals. The reduction of the dimer takes place via two different pathways.
++
Vincow and Concepcion' reported the fact that the (2
+ 2) dimer of cyclooctatetraene (cot) can be reduced by
alkali metals in tetrahydrofuran (THF) or in hexamethylphosphoramide (HMPA) to yield the anion radical and dianion of [16]annulene, eq 1. Their reaction has proven to be a convenient source of the anion radical of [16]annulene ([16]--),since it does not require the laborious synthesis of neutral [16]annulene ([16]). Although the Vincow-Concepcion route to [16]-- and [l6]*- is most (1) Concepcion, J. G.; Vincow, G. J. Phys. Chem. 1975, 79, 2037. 0022-3654/82/2086-2441$01.25/0
-
-
I
\e
(1)
polymer
useful, we report that the conversion of the dimer I via reduction is not quantitative as has been claimed. Further, due to the nonquantitative nature of the reaction, the 0 1982 American Chemical Society