1428
J. Phys. Chem. 1986, 90, 1428-143s
Kinetics and Reaction Pathways of Methanol Oxidation on Platinum R. W. McCabe* and D. F. McCready Physical Chemistry Department, General Motors Research Laboratories, Warren, Michigan 48090-9055 (Received: October 3, 1985)
Methanol oxidation kinetics were measured on Pt wires in a flow reactor at pressures between 30 and 130 Pa. The kinetics were measured as a function of oxygen-to-methanol equivalence ratio 4 and wire temperature. In methanol-lean feeds (4 < l), H2C0, CO,, and H 2 0 were the only products; in methanol-rich feeds (4 > l), CO, H2, H2C0, C02, and H 2 0 were observed. Experiments with ‘*02 showed that the principal methanol oxidation pathway does not involve C-O bond dissociation. However, the I8O2experiments, together with other features of the methanol oxidation data, also provided evidence for a minor oxidation pathway (accounting for less than 1% of the product C02) which proceeds through a carbon intermediate. Interpretation of the CH30H oxidation data was aided by comparison with data obtained previously for H2C0 and CO oxidation (McCabe, R. W.; McCready, D. F. Chem. Phys. Lett. 1984, 1 1 1 , 89). In particular, similarities between CH,OH oxidation and H 2 C 0 and CO oxidation indicate that the principal oxidation pathways for CH30H, H2C0, and CO are “kinetically homologous”,i.e. they share the common feature of an adsorbed CO intermediate. A mathematical model is presented which describes the principal CH,OH oxidation pathway as a series reaction involving adsorbed H 2 C 0 and CO intermediates. Formaldehyde may not be the only adsorbed H 2 C 0 intermediate. Consistent with experimental results, the model predicts that inhibition by adsorbed CO should be weaker for CH30H and H2C0 oxidation than for CO oxidation. The model leads to a simplified,generalized expression for the rates of CH30H, H2C0,and CO oxidation to C02,thus reflecting the homologous nature of these reactions. The main effect of the minor pathway is to decrease the rate of the major pathway at 4 > 0.1. Additionally, the minor pathway accounts for certain features of methanol oxidation, including rate instabilities at certain values of 4 and deactivation of Pt wires in pure methanol, observed both in the present study and in previous studies in other laboratories.
Introduction Exhaust pollutants from alcohol-fueled vehicles include unburned fuel and aldehydes in addition to C O and NO,.’-3 Catalytic converters can be used to oxidize HzCO and C H 3 0 H to C 0 2 and H20. However, laboratory catalyst studies of simulated ethanol-vehicle exhaust demonstrated that partial oxidation of ethanol to acetaldehyde is a problem with certain catalyst^.^ Analogously, the catalytic oxidation of unburned methanol to formaldehyde is a potential problem with methanol-fueled vehicles equipped with catalytic converters. This study was therefore undertaken, in part, to characterize the selectivity of Pt for partial and complete oxidation products. Previous s t u d i e ~ ~have - ~ disagreed on the selectivity of Pt catalysts for oxidizing methanol to formaldehyde; thus formaldehyde selectivity was characterized in detail as a function of temperature and 02-to-methanol equivalence ratio. The mechanism of methanol oxidation on Pt is not well-understood. Gentry et aL5 measured methanol oxidation kinetics over Pt wires between 310 and 660 K in oxidizing feeds and proposed a mechanism involving addition of 0 atoms to an HCO intermediate to form a COOH species which decomposes to yield C 0 2 . Hodges and Roselaar’ differed from Gentry et al. in proposing that H 2 C 0 , produced as an intermediate, dissociates to adsorbed C O and H atoms which are subsequently oxidized to C 0 2 and H 2 0 . In this study, methanol oxidation rates were examined on Pt wires in a flow reactor at maximum pressure near 133 Pa (1 33 Pa = 1 Torr). This ensured well-mixed reactor behavior and (1) Chui, G. K.; Anderson, R. D.; Baker, R. E.; Pinto, F. B. Proceedings of the Third International Symposium on Alcohol Fuel Technology, Paper 11-18. Asilomar, CA, May, 1979. (2) Bechtold, R.; Pullman, J. B. Society of Automotive Engineers Paper No. 800260, Feb, 1980. (3) Schuetzle, D.; Prater, T. J.; Anderson, R. D. Society of Automotive Engineers Paper No. 810430, Feb, 1981. (4) McCabe, R. W.; Mitchell, P. J. 2nd. Eng. Chem., Prod. Res. Deuel. 1983, 22, 212; 1984, 23, 196. (5) Gentry, S. J.; Jones, A. L.; Walsh, P. T. J. Chem. SOC.,Faraday Trans. 1 1983, 76, 2084. (6) Firth, J. G. Trans. Faraday SOC.1971, 67, 212. (7) Hodges, C. N.; Roselaar, L. C. J . Appl. Chem. Biotechnol. 1975, 25, 609. ( 8 ) Schwartz, A.; Holbrook, L . L.; Wise, H. J . Catal. 1971, 21, 199.
0022-3654/86/2090-1428$01 .SO/O
eliminated potential complications in the kinetic analysis due to heat and mass transfer effects. Methanol oxidation kinetics are compared to literature data and also to H,CO and C O oxidation kinetics obtained previously in our laboratory under similar conditions using the same Pt wires.g A surface mechanism for Pt-catalyzed methanol oxidation is proposed which involves both a major and a minor pathway. Less than 1% of the methanol reacts by the minor pathway, but it accounts for some unusual aspects of the kinetics including instability in the methanol oxidation rate at certain conditions.
Experimental Section Methanol-oxygen mixtures were flowed over resistively heated Pt wires (99.99% pure) in a stainless steel reactor. The reactor was connected by a leak valve to a turbomolecular-pumped high vacuum chamber containing a quadrupole mass spectrometer for measuring methanol conversions and product yields. Details of the apparatus and experimental procedure have been reported previo~sly.~*’~ In particular, the methanol oxidation experiments were carried out in similar fashion to H 2 C 0 and C O oxidation experiments reported previously using the same Pt wire sample^.^ Methanol was supplied as vapor over liquid methanol (analytical reagent grade) held at 273 K. O2 (99.99%) was supplied from gas cylinders. Isotopically labeled oxygen (99% oxygen 18-Stohler Isotope Chemicals) was used in a few experiments. The Pt wires were cleaned by heating between 1325 and 1375 K in flowing O2 at 133 Pa for 20-30 min. This led to reproducible methanol oxidation kinetics and produced wires which, when transferred to a UHV analysis chamber, showed only traces of Ca and Si oxides by Auger ~pectroscopy.~ Results Methanol oxidation kinetics were studied over a wide range of 0,-to-CH30H ratios and temperatures. Results obtained under conditions of integral conversion are presented first. These include the broader aspects of the kinetics, namely product analysis, identification of kinetic regimes, observations of rate instabilities, and experiments examining deactivation and reactivation of the Pt wires. Results obtained under conditions of differential con(9) McCabe, R. W.; McCready, D. F. Chem. Phys. Lett. 1984, 111, 89. (10) McCabe, R. W. J . Carol. 1983, 79, 445.
0 1986 American Chemical Society
The Journal of Physical Chemistry, Vol. 90, No. 7, 1986 1429
Methanol Oxidation on Platinum
0.lC'
'
1
I I ' ' ' ' 1
1
I 1 " " f
T
4
- C H 3 0 H Conversion A
H2CO Yield
o COP Yield
10.4 Z H Q O H 117 Pa 0 2
I
0 400
500
600
-
700
Temperature (K)
s
Figure 1. Steady-state methanol conversion (solid curve) and yields of carbon-containing products as a function of Pt wire temperature for a feed containing 10.4 Pa of CH,OH and 117 Pa of O2(4 = 0.13). H2C0 (A)and C 0 2 (0)were the only carbon-containingproducts. Yields are expressed as percent of the feed methanol converted to each species.
0.02 0.05 0.1 0.2
0 . 5 1.0
4
Figure 3. (a) Plot of the transition temperature between kinetic regimes, AT,,, as a function of 02-methanol equivalence ratio @ for both CO oxidation (0)and CH30H oxidation (0). For methanol oxidation at @ = 0.48, AT,, was multi-valued (see Figure 4). (b) A plot of the maximum selectivity for methanol oxidation to H2C0observed at various @. Selectivity is defined as the mole fraction of the feed methanol converted
to H2CO.
I
/Do
1 3 . 3 Pa CH3OH 6.1 Pa 0 2
(6= 3.3)
400
600
E00
700
800
Temperature (K) Figure 2. (a) Steady-statemethanol conversion (solid curve) and product yields [A,H2CO;0, CO,; 0,CO)as a function of Pt wire temperature for a feed containing 13.3 Pa of CH,OH and 6.1 Pa of O2 (4 = 3.3). Yields are expressed as per Figure 1. (b) O2consumption for the same experiment.
version in the low-temperature regime are presented second. These include the determination of kinetic orders and apparent activation energies. Products of Methanol Oxidation over Pt. C 0 2 , H 2 C 0 , and H 2 0 were the only gaseous products in feeds containing excess oxygen.33 Methyl formate-an additional product of methanol oxidation over highly dispersed alumina-supported Pt catalysts" -was not observed. Figure 1 shows steady-state methanol conversions and yields of H 2 C 0 and C02as a function of temperature for a feed containing 10.4 Pa of C H 3 0 H and 117 Pa of O2 (4 = 0.13). C 0 2 and H 2 C 0 formed simultaneously at low methanol conversions. The H 2 C 0 yield reached a maximum level of 8% of the feed methanol concentration near 465 K and decreased sharply at higher temperatures as the methanol conversion increased sharply and C 0 2 became the only carbon-containing product. In feeds containing excess methanol, C O and H2 were formed in addition to C 0 2 ,H 2 C 0 , and H20. For example, Figure 2a shows steady-state methanol conversions and yields of carboncontaining products for a feed containing 13.3 Pa of CH,OH and 6.1 Pa of O2 (4 = 3.3). At temperatures below 600 K, C 0 2 and H 2 0were the principal products, consistent with results for lean feeds. H 2 C 0 also formed, but at lower yield than in the 4 = 0.13 ~~
(11) McCabe, R. W.; Mitchell, P. J., to be submitted for publication.
feed. Above 600 K the C 0 2 yield decreased and C O became the major carbon-containing product; H2 also formed. Figure 2b shows that nearly all of the oxygen was consumed at temperatures above 700 K where CO was the principal carbon-containing product. Under those conditions, nearly all of the O2was reacted to H20. The high-temperature product mixture is kinetically determined and reflects competition between intermediate species (specifically hydrogen and carbon monoxide) for atomic oxygen; it is different from that predicted by water-gas shift equilibrium. Methanol was also reacted over the Pt wires in the absence of O2 at temperatures above 550 K. C O and H 2 were observed initially, but the rate decreased to undetectable levels after a few minutes. IdentiJication of Kinetic Regimes for Methanol Oxidation ooer Pt. Integral conversion vs. temperature data in all lean feeds were qualitatively similar to the data of Figure 1 for 4 = 0.13. Two kinetic regimes were observed-a low-temperature regime characterized by low conversions of methanol to both H 2 C 0 and C 0 2 , and a high-temperature regime characterized by high conversions of methanol to C 0 2 as the only carbon-containing product. Conversions increased gradually with temperature in the lowtemperature regime until a transition temperature was reached where the conversion increased sharply to high levels and remained nearly constant at higher temperatures. Thus the conversion vs. temperature plots obtained for all lean feeds showed the characteristic sigmoidal shape of Figure 1; they differed, however, in the temperatures of transition between the kinetic regimes and the yields of H2C0. The effects of 4 on transition temperatures and HzCO yields are summarized in Figure 3. Figure 3a shows the dependence of the transition temperature ATt, on 4. The delta symbol indicates that the transition is not a step function but instead occurs over a range of temperatures. However, for ease of discussion, we identify AT,, simply by the temperature of maximum slope in the conversion vs. temperature curve. AT,, was multivalued at 4 = 0.48, as illustrated by the data of Figure 4. Thus, two points are shown in Figure 3a at 4 = 0.48, one corresponding to the path ACD and the other corresponding to the path ABD in Figure 4. The curve describing the AT,, vs. 4 relationship in Figure 3a was arbitrarily drawn between the values of AT, observed at 4 = 0.48. The multivalued nature of the transition temperature at 4 = 0.48 will be described in more detail in the following section.
1430 The Journal of Physical Chemistry, Vol. 90, No. 7, 1986
McCabe and McCready 100
p
I
300
400
I
I
500
600
I
1
700
800
c
> C
39.6Pa 02 (d= 0.48)
Temperature (K) Figure 4. Plot of steady-statemethanol conversion to C02as a function of Pt wire temperature for a feed containing 12.8 Pa of CH30H and 39.6 Pa of O2 (6 = 0.48). The solid curve shows the course traced in a particular experiment carried out in the direction of decreasing temperature. The dashed curves show other paths traced in different ex-
periments carried out at the same conditions. Figure 3a also shows AT,, vs. 4 data obtained previously for C O oxidation.' In contrast to methanol oxidation, AT,, for CO oxidation increased gradually at low 4 and sharply for 4 > 0.2. Consequently, ATt, for methanol oxidation is nearly equal to, or greater than, AT,, for CO oxidation at 4 between 0.13 and 0.5. Figure 3b shows the 4 dependence of the maximum H,CO yield observed in each conversion vs. temperature experiment (i.e. each point in Figure 3b corresponds to the maximum H 2 C 0 yield observed in an integral conversion experiment at a particular 4, such as the maximum yield near 0.08 shown in Figure 1 for 4 = 0.13). The maximum HzCO yield increased to -8% of the feed methanol concentration at 4 = 0.13 and decreased at larger
(K)
Temperature
Figure 5. Conversion of methanol to C 0 2as a function of Pt wire temperature in a feed containing 12.6 Pa of CH30H and 40.6 Pa of O2 (6 = 0.47). The wire was initially deactivated by heating in 16 Pa of CH30H at 773 K. Path A (open symbols) shows the conversion vs. temperature profile after heating the deactivated wire in the 6 = 0.47 feed starting from rmm temperature. Path B (closed symbols) shows the conversion vs. temperature profile obtained during stepwise cooling back to room temperature.
4.
Instability in Methanol Oxidation near 4 = 0.5. Figure 4 shows methanol conversions and product yields as a function of temperature for a feed containing 12.8 Pa of CH30H/39.6 Pa of O2 (4 = 0.48). The region between A and D was characterized by unstable conversion of methanol to C 0 2 . The data were obtained by starting at 785 K and decreasing the temperature slowly, in stepwise fashion, while obtaining steady-state data at the temperatures indicated. When the temperature was decreased from 570 to 557 K, the conversion jumped to point C rather than continuing the decreasing trend established between A and B. The conversion followed the path from C to D as the temperature was reduced further. In other experiments at 4 = 0.48, similar sharp increases or decreases in conversion were observed in traversing the region between A and D starting either from low or high temperatures. Taken together, the experiments at 4 = 0.48 define the curves ABD and ACD for conversion vs. temperature between 535 and 610 K. Instability in the conversion vs. temperature data was most pronounced near 4 = 0.5, but also observed a t slightly lower and higher 4. Similar experiments involving C O and H 2 C 0 oxidation resulted in well-behaved hysteresis loops obtained by approaching the transition regime from both the low-temperature and high-temperature directions. However, the instabilities of the type shown in Figure 4, involving sharp increases or decreases in conversion with slight changes in temperature, were unique to methanol oxidation. The absence of similar instabilities in H 2 C 0 ,~ the same wires and identical experiand CO o ~ i d a t i o nwhere mental techniques were employed, ensures that the instabilities observed in methanol oxidation are a real feature of the surface reactions and not an experimental artifact. Experiments with Deactivated Pt Wires. Methanol/oxygen mixtures were flowed over Pt wires which had been deactivated by heating for a few minutes in 16 Pa of C H 3 0 H at 773 K. Figure 5 shows C 0 2 yield as a function of temperature after exposing a deactivated wire to a 12.6 Pa of CH30H/40.6 Pa of 0, feed. The arrows indicate the temperature course of the experiment. Starting from 296 K, the C 0 2 yield followed path A (open symbols) with increasing temperature. Essentially no C 0 2was produced until 540 K. Between 540 and 570 K, the C 0 2 production increased sharply to high levels and remained relatively unchanged
0 ' 0 P ' v) v) v)
I I I
I
I
I
I
I
Figure 6. Mass spectrometer responses following addition of O2 to methanol flowing over a deactivated Pt wire at 585 K. The methanol pressure was 16 Pa. 0, was introduced in two steps to a final pressure of 60 Pa. Masses 31 (CH,OH fragment),44 (CO,), 32 (OJ, and 40 (Ar tracer in the 0,) were monitored. The mass spectrometer signals are arbitrarily scaled. Prior to time-zero the Ar, 02,and C02 signals represent background levels of those gases arbitrarily positioned on the vertical axis. The CH30H signal prior to time-zero represents the mass 31 signal associated with 16 Pa of CH,OH, also arbitrarily positioned on the vertical axis.
at higher temperatures. After maintaining the wire at 783 K for 15 min the temperature was decreased in stepwise fashion yielding the steady-state CO,.yield curve described by path B (closed symbols). The transition between kinetic regimes was much less abrupt in the direction of decreasing temperature than in the direction of increasing temperature and CO, was detected down to -420 K. In subsequent heating and cooling cycles, the COz yield curves were similar to curve B although instabilities of the type shown in Figure 4 were observed between 540 and 600 K. Behavior of the type shown in curve A, where CO, yield was negligible below 540 K, was only observed during the first heating sequence following deactivation of the wire in pure C H 3 0 H at 773 K. In another experiment, the Pt wire was deactivated by heating in flowing methanol at 773 K and then O2 was abruptly added to the methanol feed while maintaining the wire a t 585 K. The transient mass spectrometer responses for CH,OH, CO,, and 0, containing Ar tracer are shown in Figure 6. The Ar signal shows the pumping response of the system to the introduction of 0,. The O2was introduced in two steps labeled 1st and 2nd in Figure 6.
The Journal of Physical Chemistry, Vol. 90, No. 7, 1986 1431
Methanol Oxidation on Platinum
PCH3 OH (pa)
TABLE I: Methanol Oxidation Experiments at 773 K with ' * 0 2 reaction
partial Dress.
W H oxidation H 2 C 0 oxidation CO oxidation l80exchange with C'60160
12.6 Pa of CH,OH/45 Pa of I8O2 5.2 Pa of H2C0/13.8 Pa of ' * 0 2 13.3 Pa of C0/14.2 Pa of 1 8 0 2 9.7 Pa of co2/14.6 of Pa I8O2
selectivity to Cl80l80'
0.008 k 0.0005 0.006 0.0005 0.002 0.0005 no c160180 observed
*
-
3
5
10
I
I
I
4o
S=O.56?0.10
391
o
20
I/
I
1
r
a
x
conversion
"The selectivity to C ' 8 0 ' 8 0is expressed as a ratio of mass spectrometer peak intensities uncorrected for sensitivity differences between the various isotopes of C 0 2 , i.e.
4
S=
C ' 8 0 1 8 0(48 amu)
CL80180 (48 amu)
-
+ C'60180(46 amu) + Cl6OI6O(44 amu)
The time of the initial valve opening is designated time-zero in Figure 6 . After 15 s the valve was opened wider. Measurable reaction rates were not observed until approximately 20 s after O2was first introduced. The onset of rapid reaction was signified by sharp decreases in the CH30H and O2 signals and a sharp increase in C 0 2 signal above the background level. Oxidation Experiments with I8O2. Methanol (CH3I60H)was oxidized with I8O2to determine whether or not C02is produced by a pathway involving C-I6O bond dissociation. As shown in Table I, very little C ' 8 0 1 8 0 formed a t 773 K, proving that methanol is oxidized to C02principally by a pathway which leaves the C-I6O bond intact. Table I also shows C 1 8 0 1 8 0yields in similar lSOzoxidation experiments with H2CI60and CI6O. As in methanol oxidation, only small amounts of C180180 were observed. Thus H 2 C 0 and C O also react principally by mechanisms which leave the carbon-oxygen bond intact. The small yields of C ' 8 0 1 8 0in C H 3 0 H , H 2 C 0 , and C O oxidation suggest that all three reactions have minor pathways involving carbon-oxygen bond d i ~ s o c i a t i o n . However, ~~ given the low C 1 8 0 1 8 0yields, special care was taken to ensure that experimental artifacts did not account for the C'801s0product. To this end, the following sources tests were conducted: (1) CH,OH, H 2 C 0 , CO, and '*02 were carefully checked for impurity species by mass spectrometry, (2) I8O2and either C H 3 0 H , H 2 C 0 , or C O were flowed over the Pt wire at room temperature to check for the possibility of reactions occurring on surfaces other than the Pt wire such as the ion gauge or mass spectrometer filaments, and (3) mixtures of CO, and ISO2were flowed over the hot Pt wire to check for exchange of oxygen with C 0 2 . The latter test was conducted to determine if the C180180 product could be attributed to oxygen exchange between the principal product, C'60180, and lsO. Because we had no feed source of C 1 8 0 1 8 0we carried out the and I8O2and looked for analogous reaction between C160160 C 1 6 0 1 8 0product. All tests were negative; but we note that, in test ( l ) , possible trace contamination by ethylene and ethane cannot be ruled out since those species have parent mass spectrometer peaks coinciding with either parent or fragment peaks of CH,OH, H 2 C 0 , and CO. However, such contamination is unlikely because H 2 C 0 could be dissociated for long periods without deactivating the Pt wire. This would only be expected in the absence of significant hydrocarbon contamination in the background gases and in the H 2 C 0 feed. Effects of Methanol Pressure, Oxygen Pressure, and Temperature on Methanol Oxidation Kinetics. Methanol oxidation experiments were conducted a t differential conditions ( 1 indicates CH30H-rich feeds. 4 = 1 corresponds to a feed 0 2 / C H , 0 H mole ratio of 1.5 obtained from the stoichiometric reaction CH,OH 1.502 -- C 0 2 2H20. (34) C’B0180 could also form from reaction of CH3I80Hwith l80,without C - 0 bond dissociation. However, no CH3I80Hwas detected in our methanol feed at the mass spectrometer sensitivities employed in these experiments.
+
+
The Journal of Physical Chemistry, Vol. 90, No. 7, 1986 1435
Methanol Oxidation on Platinum termediates, and a minor pathway involving an elemental carbon intermediate. The reaction products, COz,H 2 C 0 , HzO, Hz, and CO, depend strongly on reaction conditions, namely feed stoichiometry and reaction temperature. Very little methanol reacts via the minor pathway (> 2kotdsPO2and eq 6 can be approximated closely by 2ko;dsP02 rcol = (7) 1 + Kcopco which, for KcOpc0 >> 1, accurately predicts the negative-firstorder CO pressure dependence and positive-first-order O2pressure dependence observed at low temperatures and/or high C O pressure. B. H 2 C 0 and CH30H Oxidation. Reactions A-D are combined with reactions E and F to account for H 2 C 0 oxidation via the pathway outlined in Figure 1 1 . C H 3 0 H oxidation is described by reactions A-H. The derivations of the rate expressions for H 2 C 0 and C H 3 0 H oxidation follow the approach outlined above for C O oxidation making similar simplifying assumptions. H,CO(g)
+S
kHfl"
H2CO(S)
(3)
In deriving eq 1-3, the following assumptions were made: ( 1 ) C O is present at high coverage, thus the fraction of empty surface sites, S, is closely approximated by (1 - Oca). ( 2 ) 0-atom recombination does not occur at significant rates.
(5)
to give
1 - Oco)]Oo, ( 2 )
- k,OcoOo
(4)
Registry No. Methanol, 67-56-1; platinum, 7440-06-4.
(E)
