Ethylene formation by oxidative dehydrogenation of ethane over...
0 downloads
160 Views
857KB Size
J. Phys. Chem. 1993,97, 11815-11822
11815
Ethylene Formation by Oxidative Dehydrogenation of Ethane over Monoliths at Very Short Contact Timed M. H d and L. D. Schmidt' Department of Chemical Engineering and Materials Science, University of Minnesota, Minneapolis, Minnesota 55455 Received: July 27, 1993; In Final Form: September 14, 19930
The production of ethylene or CO and H2 from ethane in the presence of air or 0 2 at atmospheric pressure has been examined over ceramic foam monoliths coated with Pt, Rh, and Pd at contact times on the order of milliseconds. In a fuel-rich regime (C~Ha/02> 1.5) on Pt, we observe selectivities to ethylene up to 70% with conversions above 80%. On Rh, CO and H2 (syngas) productiondominates, while on Pd, heavy carbon deposition rapidly occurs. Optimum production of ethylene from ethane on Pt is obtained by reacting ethane with a mixture of air and 0 2 a t a C2H6/02 ratio of -1.7 at contact times < 10 ms. Optimum production of syngas is obtained on Rh a t a C2H6/02 ratio of 1.0, with -70% selectivity and >95% conversion of C2H6. These high selectivities to specific products are strong evidence that very simple reaction pathways dominate. The formation of C2H4 and the complete reaction Of C2H6 strongly argue that the process is initiated by oxidative dehydrogenation. On Pt, this must be followed by &hydrogen elimination from adsorbed ethyl and C2H4 desorption. On Rh and Pd, @-hydrogenelimination is not preferred, resulting in the complete pyrolysis of the adsorbed hydrocarbon and ultimately syngas production (Rh) or carbon deposition (Pd).
Introduction
- + 2co - c, + + - +
C2H4
There is an abundant supply of light alkanes and relatively few means of converting them to more valuable products. Although natural gas is predominantly CH4, it also contains from 5 to 30% CZ&, C3Hs, and C4H10,l with C2Ha the most abundant secondary component. Natural gas is currently underutilized primarily due to difficulty in transportation from the source. Much research has been devoted to conversion of methane to more easily transportable productsvia synthesisgas (CO and H z ) . ~Although methane is the major component of natural gas, it is alsonecessary to understand the reactions of ethane to determine whether generalization of the CH4 reactions to natural gas will result in the production of solid carbon. Synthesisgas can be easily reacted to form liquids such as methanol over a Cu/ZnO catalyst3 or through the Fischer-Tropsch process over Fe and Ru catalysts to form long-chain h y d r ~ a r b o n s . ~ The oxidationof ethane (either heterogeneousor homogeneous) can lead to complete combustion,
C2H6 + ' / 2 0 2
-
2 c 0 2 + 3H2O
(1)
partial oxidation to synthesis gas,
C2H, + 0,
-
2CO + 3H2
(2)
dehydrogenation, or oxidative dehydrogenation reactions to ethylene.
These product species can also react to form solid carbon, C,. f This research was partially supported by DOE under Grant DE-FGO288ER13878-AO2. t Supported by DARPA-NDSEG Graduate Fellowship. To whom correspondence should be addressed. Abstract published in Advance ACS Absrmcrs, October 15, 1993.
0077-3654/93 /2097-11815%04.00/0
2C,
2H2
CO,
CO
H2
C,
H20
(5) (6) (7)
These reactions are listed in Table I with corresponding heats of reaction and equilibrium constants. The goals of this investigation are the determination of the reaction conditions and catalysts that promote the selective formation of olefins, syngas, and surface carbon and the mechanisms by which these occur. Ethylene has been shown to be formed from ethane by several mechanisms including thermal dehydrogenation (eq 3) and oxidative dehydrogenation (eq 4). Thermal (homogeneous) dehydrogenation of ethane exhibits a high se1ectivi.y to ethylene (- 80%) with a fairly high conversion (- 60% per pass): and these reactions represent the main procesw for commercialolefin production. However, because this process is very endothermic, ethane must be heated to 750-900 OC in a tube furnace for this reaction to proceed, and coke formation can only be minimized by adding H20 to steam reform the coke and alkanes. The oxidativedehydrogenation of ethane over oxide catalystss12 such as VzOs/SiOz has also been shown to be fairly selective to ethylene at low conversions. These catalysts are 100%selective to ethylene at conversions of ethane 40% conversion of CZH6.l4 Since these processes are operated under severely fuel-rich conditions, carbon deposition and consequently catalyst deactivation can be major problems, and this contributes to the poor conversions of many processes. At higher conversions, not only does selectivity decrease but also coke formation becomes an issue.s In fact, under the reaction conditions of this study, the products predicted at thermodynamic equilibrium would be predominantly CO, Hz, and solid carbon. We will show that when Pt isused as the catalyst,coke formation is totally suppressed even at high temperatures. @I 1993 American Chemical Society
Huff and Schmidt
11816 The Journal of Physical Chemistry, Vol. 97, No. 45, 1993
TABLE I: Rerctiom of Ethfae %C&
- --
(1) C2& + '/202 2C02 + 3Hz0 (2) C2H6 + 0 2 2 c 0 + 3H2 (3) C2H6 C2)4 + H2 (4) C2H6 + '/20~C2H4 + H20 (5) CzH4 2C, + 2H2 (6) 2CO C, + C02 (7) CO + H2-+C, + H2O +
-
+
inair
complete combustion partial oxidation to syngas dehydrogenation oxidative dehydrogenation cracking CO disproportionation (Boudouard) reverse steam reforming of carbon
Experimental Section The reactor was essentially identical to that described previously for syngas production by direct oxidation of CH4.2 It consisted of a quartz tube with an inside diameter of 18 mm containing the catalytic monolith which was sealed into the tube with hightemperature alumina-silica cloth that prevented bypass of the reactant gases around the edges of the catalyst. To reduce radiation heat loss and better approximate adiabatic operation, the catalyst was immediately preceded and followed by inert alumina extruded monolith heat shields. The outside of the tube near the reaction zone was insulated. The catalyst samples were prepared by impregnation of a-Al203 foam monolith disks 17 mm X 1 cm long with saturated solution of metal salts. For the pt catalysts, a saturated solution of H2PtC16 in water was dripped onto a clean and dry A1203 foam monolith with 45 or 80 pores per inch (ppi) until the monolith was saturated with liquid. After the catalysts had been dried under N z , they were calcined in air at 600 O C and then reduced in H2. The Rh and Pd catalysts were prepared similarly using saturated solutions of rhodium acetylacetonate and palladium acetate in acetone. Singleimpregnationstypically produced metal loadings of 1-5 wt 4%. Gas flow into the reactor was controlled by mass flow controllers which had an accuracy of *O.l SLPM for all gases. The feed flow rates ranged from 2 to 12 SLPM total flow, corresponding to 13 to 79 cm/s superficial velocity (i.e., the velocity of the feed gases upstream from the catalyst) at room temperature and atmospheric pressure. In all experiments, the reactor pressure was maintained at 1.4atm. The reaction temperaturewas lo00 OC, and contact times were from 7 to 40 ms. Product gases were fed through heated stainless steel lines to an automated gas chromatograph. For quantitative determination of concentrations, standards were used for all species except H 2 0 , which was obtained most reliably from an oxygen atom balance. To convert the product gas concentrations to molar quantities for a given feed basis, the mole number change due to the chemical reactions was calculated using the measured N2 concentration. Since N2 is an inert in this system, the ratio of product gas to feed gas moles was inversely proportional to the ratio of product gas Nz concentration to feed gas N2 concentration. Individual species concentrations were measured with a reproducibility estimated to be &2%. Temperatures were monitored using thermocouples inserted from the front and the rear of the quartz tube in one of the center channels of the inert monolith immediately before or after the catalytic monolith. The reactor was operated at a steady-state temperature which is a function of the heat generated by the exothermic and endothermic reactions and the heat losses from the reactor. A simplified overall energy balance for the reactor can be represented by
-
In these expressions the bars over quantities represent averages with Fc - TO)the sensible temperature rise of the catalysts surface from szme gas temperature, To, to the average catalyst temperature, T,. The heat generation term, E(-A&)%,, is the
cp(
ATd
5.7 17.4
("C) 2800 815
C2H6/02 0.29 1.00
29.6
1061
2.00
A?P
(kJ/mol) -1429 -137 +136 -105 -53 -172 -131
Kq
(1200K)
K=?
(1200K)
>lo3'
4.2X 1P5 6.2
5.ox 108
6.9 X 10' 1.7 X 1 t 2 2.5 X l t 2
0.1 3.0 6.0
sum of the heats of reaction of both exothermic and endothermic reactions at some average conversion, Xi, for each reaction. At high flow rates, the reactor approaches the limit of adiabatic operation and the autothermal temperature approaches the adiabatic temperature. (9) Adiabatic temperature changes for several individual reactions are shown in Table I. Experiments were carried out with either air or 0 2 as the oxidant. In the experiments using 0 2 , N2 was also added as 20% of the feed as an internal GC calibration standard. Although the process in steady state is autothermal with feed gases at room temperature, heat was supplied initially to ignite the reaction. A mixture of ethane and air near the stoichiometric composition for production of synthesisgas was fed to the reactor, and the reactants were heated to the heterogeneous ignition temperatureof -230 OC. After light-off, theexternal heat source was removed,the reaction parameters wereadjusted to the desired conditions, and steady state was established (- 10 min) before analysis. Except where carbon deposition is noted, data shown were reproducible for time periods of at least several hours and on several catalyst samples. Resolts
Ethane in Air. In Figure 1, we show the carbon atom and hydrogen atom selectivities,the ethane conversion, and the reaction temperature for feed mixtures of ethane and air as a function of feed composition. The feed composition was varied while maintaining a fixed total flow of 5 slpm with room temperature feed. Table I lists the possible oxidation reactions for ethane. Below 6% C2H6 in air, we expect the products to be primarily COZand H2O with a temperature near 2500 O C . Experiments in this region were not done because of the danger of explosion. Between 6 and 17%,selectivitiesshould switch from C02 to CO and from H20 to H2, and the temperature should be much lower, near 800 O C . As seen in Figure 1, at 17.4% C2H6 in air, the syngas composition, we observe 40% selectivity to ethylene at 950 O C . This temperature is greater than the syngas ATad because the oxidativedehydrogenation reaction has a somewhat higher AT.d. This is extraordinary since thermodynamics predicts 100% conversionof ethane to C O and H2 with no significantbyproducts. In Figure 1, &,H, is seen to peak near a composition of 25% C2H6 in air (C2H6/02 = 1.7) with an optimal selectivity of 52% at 65% conversion of ethane. This composition lies between the stoichiometric compositions for syngas production (CzH6/02 = 1.0) and for oxidative dehydrogenation (C2H6/02 = 2.0). As the percentage of ethanein the feed increasesbeyond 2596, ethylene production remains high, but butylene is also formed by dimerization of C Z H ~thus , decreasing the apparent S c 2 ~ . Ethanebo,. In Figure 2, we show the selectivities,conversion, and reaction temperature for ethane oxidation in 0 2 as a function of feed composition. The feed composition was varied while maintaining 20% N2 in the stream and a fixed total flow of 4.5 SLPM with rmm temperature feed.
The Journal of Physical Chemistry, Vol. 97, No. 45, 1993 11817
Ethylene Formation by Dehydrogenation of Ethane
TO
a
4
0
8
80
3
74
c
.P
70
Total Hydrocarbons 60
'a
J
d
g
1 P
50
1
@t 30 20
IO I
100
1050
950
loo0
d
h
9ood
' 850
10
15
20
25
30
35
aPEthaneinAif Figure 1. Carbon selectivity (a), hydrogen selectivity (b), ethane
3
v
.-6
E!
8
I
I
I
I
I
I
I
I
I
I
1
I
90'
so
3
1100
ji
'
s
D
I
70'
9003
60' 50
850
3 1.4 1.6 1.8 2.0 800 C2H6 2"
pt
conversion, and reaction temperature (c) for a 45 ppi X 1 cm, 4.7 wt % foam monolith as a function of C2H6 in air fed at a total feed flow rate of 5 SLPM in an autothermal reactor at a pressure of 1.4 atm.
Figure 2. Carbon selectivity (a), hydrogen selectivity (b), ethane conversion, and reaction temperature (c) for an 80 ppi X 1 cm, 2.3 wt 9% Pt foam monolith as a function of C2H6 in 0 2 (20% N2)fed at a total flow rate of 4.5 SLPM in an autothermal reactor at a pressure of 1.4 atm.
The observed trends in the 0 2 experiments are similar to the trends observed in the air experiments (Figure 1). However, the selectivityto C2& and the CzH6 conversionare both significantly higher with 70% and 82%, respectively, at a CzH6/02 ratio of 1.7. The reaction temperature is also higher, illustrating the effect of reduced Nz diluent. Preheat. Figure 3 illustrates the effect of preheat on the selectivities, conversion, and reaction temperature. The C2H6/ 0 2 ratio was 1.62. The total flow rate was 5 and 4 SLPM for the air and 0 2 experiments, respectively. The ethyleneselectivities presented here without preheat differ from the optimum shown in Figures 1 and 2 because the C2H6/O2 ratio and flow rates were suboptimum in this case. For the air experiments, preheat greatly influenced both the selectivities and the conversion. At 400 OC preheat, S c a b increased by 40% while the ethane conversion increased to over 80%. The addition of preheat raises the autothermal reaction temperature and provides heat for some endothermic reactions including thermal dehydrogenation of C2H6 (listed in Table I) which would otherwiserequire heat from the exothermicoxidation of ethane to CO and H2. Figure 3 shows that as preheat is increased, less CO and more C2H4 are formed. Hydrogen production remains virtually the same since Hz is also a dehydrogenation product. In the 0 2 experiments, increased ethane conversion was observed, but the selectivities remained nearly constant over the
rangeof preheat used. It is important to note that the autothermal temperature is significantly higher in 02 than in air. Flow Rate. Figure 4 illustrates the effect of flow rate on the selectivities, conversion, and reaction temperature for the same catalyst sample at a feed composition of 25% C2H6 in air. It is seenthat higher flow rates produce both greater Scaband greater ethane conversion. At low flows, significantly more H2O and C02, complete oxidation products, are formed which consume the 0 2 and thus decrease the conversion of ethane. The H2O productiononlydrops slightlyat the higher flows. Wewereunable to observe a decrease in ethane conversion at flow rates up to 8 SLPM corresponding to a contact time of 10 ms. Rh .ndPd. In Figure 5 , we show the selectivities,conversion, and reaction temperature for ethane oxidation in 0 2 over a 4 wt % Rh catalyst as a function of feed composition at a fued total flow of 5 slpm with room temperature feed. In contrast to the results shown in Figure 2 for Pt, syngas production dominates on Rh. At a CzH6:O2 ratio of 1.62, Sea", is only 40% instead of nearly 70% for Pt. Also, significantly less CO2 is formed as required by the oxygen balance. This is important to the conversion of natural gas to synthesis gas. Rh has been shown to be an excellent catalyst for CH4 conversion to synthesis gas,15 and these results suggest that the presence of ethane should not interfere. A similar experiment was conducted using a 1.9 wt % Pd catalyst. Even at the syngas ratio (C2H6/02 = l.O), coke was
-
11818 The Journal of Physical Chemistry, Vol. 97, No. 45, 1993
Huff and Schmidt
lob A
-
E X *i3
1
940
(a
---*--:=- a 8.-
.,&:a
:----;--
80-
- 920
-9003 3 8803
i?
860
9
h
840
forming rapidly. Within 20 min, the catalyst was completely deactivatedand would not sustain reaction. The initial gas-phase products, however, had 16%selectivityto C2H4 and 55% selectivity to CO at this feed composition. These are close to the values seen on the Pt catalyst at the syngas ratio, except carbon deposition docs not cccur on the Pt catalyst.
To summarize these results, on Pt-coated monoliths up to 70% of the ethane is converted to ethylene at -900 OC at contact times of 1 P 2s in a steady-state reactor with nearly complete conversion of ethane. In contrast, Rh produces primarily syngas with no carbon buildup, while Pd rapidly deactivatesdue to carbon formation. These results demonstrate several surprising features: (1) a largedifference in reaction paths on the three metals, (2) formation of C2H4 at concentrations far beyond hydrocarbon equilibrium predictions, (3) the complete absence of carbon buildup (for Pt and Rh) for feed compositions within the region where solid graphite becomes thermodynamically stable under reaction conditions, (4) nearly completeconversion of the reactant alkane, and ( 5 ) nearly complete survival of the product olefin. In the following sections, we shall consider the reaction mechanisms responsible for these observations in terms of the reaction pathways involved and thermodynamic equilibrium in various processes. It is worth emphasizing the effect of reactant stoichiometry ( y s 1-4) on selectivities and conversions. First, the C2H6/02 ratio used for C2H4 formation is up to 7 times that required for completecombustion to COZand H20 (eq 1) and up to twice that requirod for CO and H2 (eq 2). In all cases all of the 0 2 is consumed, and therefore stoichiometryrequires specificrelations between products to satisfy balances on C, H, and 0 atoms. In all data shown these balances close completely.
-
0'
Reaction Mechanism. There are several experimental observations that must be accounted for in a reaction mechanism, (1) Coke formation is not observed on Pt or Rh but is on Pd. (2) Syngas production dominates near the stoichiometric composition for syngas in all cases. (3) Contrary to thermodynamic equilibrium which predicts carbon deposition, C2H4 production dominates near the stoichiometric composition for C2H4 production by oxidative dehydrogenation on Pt. (4) Ethylene production increases with temperature on Pt. ( 5 ) Conversion increases at higher flow rates. (6) Finally, H2O production increases and C2H4 production increases as 0 2 in the feed decreases. We believe that all of these observations can be accounted for by reaction steps shown in Figure 6, obviously with different rate coefficients on different metals. Reaction Initiation. A simple pyrolysis mechanism
would predict C2H4, C2H2, and CO as the primary gaseous products of sequential dehydrogenation, C-C bond cleavage, oxidation of C, and desorption of volatile products. This mechanism is unlikely for several reasons. The main one is that it is observed that 80%of C2H6 is consumed while on Pt most of the C2H4 survives. One would expect C2H4 to pyrolyze much faster than C2H6. We suggest that the most probable initiation step is reaction of C2H6 with adsorbed oxygen atoms, 0,, by oxidative dehydrogenation C2H6
+ 0,
-
C2H, + OH,
(12)
The Journal of Physical Chemistry, Vol. 97, No. 45, 1993 11819
Ethylene Formation by Dehydrogenation of Ethane
C2H4
I
co
J
h
1
I
I
I
I
I
7
a
I
1
100 I
-F I
I
9
2
3
I
4
1
5
I
6
I
7
I
8
Flow Rate (slpm) Figure 4. Carbon selectivity (a), hydrogen selectivity (b), ethane conversion, and reaction temperature (c) for a 45 ppi X 1 cm, 4.7 wt % Pt foam monolith as a function of reactant flow rate at a reactant composition of 25% ethane in air in an autothermal reactor at a pressure of I .4 atm.
or possibly with adsorbed hydroxyl
w' 0.9
I
I
I
I
I
I
I
I
I
I
I
I
I
1
1.1
1.2
1.3
1.4
1.5
1.6
C2H6
1.7
2'
Figure 5. Carbon selectivity (a), hydrogen selectivity (b), ethane conversion, and reaction temperature (c) for a 45 ppi X 1 cm, 4 wt 96 Rh foam monolith as a function of C2H6 in 0 2 (20% N2)fed at a total flow rate of 5 SLPM in an autothermal reactor at a pressure of 1.4 atm. C2H6 + Os
-
OH, + C2HS,,6C2H4,s
1."
* C2H4
(on Pt)
C4H, C4HlO
(on Rh)
The early portion of the catalyst is predicted to be largely covered with 0,.For known kineticsof 0 2 interaction with Pt (dissociative adsorption, s r 0.01, Ea = 52 kcal/mol), a Langmuir isotherm predicts Bo 2: 0.99 monolayers at Po2 = 120 Torr (the initial 0 2 partial pressure) at Ts = 900 OC. This Bois sufficiently high that adsorbing C2H6 may always find 0, for reaction 12 rather than the clean surface, reactions 10 and 1 1. As soon as all 0 2 has been consumed, 00 falls to a very small value, and the surface probably becomes covered with a fraction of a monolayer of carbon. Now the hydrocarbons (primarily C2H4 and unreacted may become very unreactive and pass through the monolith without further decomposition. We believe that only an oxidative dehydrogenation initiation step explains the total reaction of C2Hb and the total survival of C2H4. Note also that the other product of this step is OH,, which sould rapidly hydrogenate to form H20; an increase in H20 production is observed to accompany increased C2H4 production as indicated in Figure 1. The increased H20 with decreased 0 2 argues that H20 formation is associated with the steps which produce C2H4, such as eqs 12 and 13. @-Hydrogen Elimination. Once adsorption has occurred
CS I
(on Pd)
H
Figure 6. Proposed surface reactions in ethane oxidation. At the right are indicated gaseous species produced, and in parenthesesare the metals for which those species are dominant.
(presumably as C~HS,),gaseous ethylene formation should only
be possible if the next step is elimination of an H atom on the carbon adjacent to the carbon-metal bond (a @-hydrogenatom). Thiswould produce C2H4, which can desorb. A carbon selectivity of 70%C2H4 requires that at least 7056 o f d e ethane must react through this channel on Pt. If there is no such preference for @-elimination(as is probable on Rh and Pd), elimination of the first a-hydrogen leads to adsorbed ethylidene (CHsCHs) and the second leads to ethylidyne (CH$2=), Once C2Hx,,species have formed (see Figure 6),the system must react further to yield either syngas or carbon deposition.
11820 The Journal of Physical Chemistry, Vol. 97, No. 45, 1993
The interaction of ethyl groups on several noble metals, especially Pt,21-24has been examined. The decomposition of the adsorbed groups was examined using temperature-programmed desorption (TPD) and x-ray photoelectron spectroscopy (XPS) in an ultrahigh-vacuum (UHV) system.21 The surface was exposed to C2HSI,which resulted in adsorbed C2HS fragments. It was shown that the first hydrogen is abstracted from the ethyl by ,%elimination to form adsorbed C2H4. This C2H4 may desorb or react on the surface to form ethylidyne(CHSCI). This species must eventually dissociate further to yield H, and C,; the H, can dimerize to H2 and desorb, and the Cs may either oxidize to CO and desorbor polymerize to form coke. However, in TPD studies, C2H4 was shown to desorb at temperatures below 170 K. It appears that, since the reaction is taking place at 1200 K, C2H4 desorbs immediately after being formed on Pt, thus averting ethy lidyne formation. From available data on C2Hs,Z1-"itappearsthat theselectivities of 8-hydrogen elimination proceed in the order of Cu = Au > Pt > Ni. This suggests that the more noble metals tend to favor &elimination rather than a-elimination and cracking to carbon. This indicates that it is reasonable that Pt should produce more C2H4 than Rh because Pt is the more noble, less reactive metal. It also seems unlikely that any addition processes such as H atom addition to hydrocarbon fragments or combination of hydrocarbon fragments such as
-
2CH2,s
-
(14)
C2H4
would be significant at high temperatures because coverages must be high for such bimolecular reactions to occur, and they are reduced entropy events. Since equilibrium predicts CH4 rather than C2H4 at these temperatures (see below), these pathways to C2H4 seem very unlikely. Steam Reforming of Ethylene. At low flow rates (Figure 4) the C2H4 selectivity decreaseswhile the CO selectivity increases. This is as expected from the steam reforming reaction of product ethylene
C2H4
+ 2H2O
-
2CO
+ 4H2
(15)
and the observed changes in these species are in fairly good agreement with this reaction being the dominant loss of C2H4 aside from dimerization to C4Hs (see below). Ethyleneshould only form in thevery early stagesof the reactor (perhaps the first millimeter) when 0 2 is present. It then can react only with H20, C02 (by C02 reforming), or itself (by dimerization) in the absence of decomposition to carbon. High flow rates (short contact times) enhance the yield of C2H4 by reducing the time for its further unimolecular or bimolecular reactions. For example, the observed losses of C2H4 at lower flow rates indicate that very low C2H4 selectivities should be observed at contact times > 0.1 s. Dimerization. The next most abundant hydrocarbon product after C2H4 is C4Hs (sometimes exceeded by CH4) over all metals. Butylene formation accompanies C2H4 and is typically 10% of the C2H4 (on a carbon atom basis). This must be formed by dimerization of C2 species.
C2H4 + C2H4
-
C4H8
(16)
Thedimerization reaction (eq 16) occurs at high C2H6/02 ratios. The yield of CzH4isof course reduced by the dimerizationreaction in favor of C4Hs. Therefore, one should actually add the C4Hs selectivity to the C2H4 selectivity to obtain the actual selectivity of formation of C2H4, and this gives an optimum C2H4 selectivity of 74% in 0 2 . Ethylene and C2H6 evidently do not combine efficiently to form C4Hlo C2H6
+ C2H4
-
C4H10
(17)
because the alkane C ~ H IisOalways much less abundant than
Huff and Schmidt
C4&. Presumably, the first step of this mechanism (eq 12) is so fast that C2H6 and C2H4 are never present together in the amountsneeded for this reaction. The formationof C4Hs requires adsorption and reaction of C2H4. Ethylene must therefore not be totally unreactive in that it can undergo the dimerization reaction but not cracking. Steps on Pt. The primary surface reaction sequence on the Pt catalyst must be the upper channel shown in Figure 6, because ethylene is the dominant product. The OH, formed by oxidative dehydrogenation (eq 4) then combines with H, (or perhaps abstracts an H atom from C2&) to form H20 and desorb. Along with oxidative dehydrogenation,partial oxidation to syngas and thermaldehydrogenationare also contributing. At higher reaction temperatures (obtained either by the addition of preheat or by the use of 0 2 rather than air), the thermal dehydrogenation reaction proceeds to a greater extent, increasingthe overall C2H4 yield. The higher temperature reduces the amount of heat that must be provided by exothermic reactions, thus increasing the possible yield of ethylene. The conversion increases at higher flow rates because short contact times favor partial oxidation products. At long contact times, the reactants remain on the catalyst surface long enough to produce complete combustion products, CO2 and H20. This quickly consumes all of the 0 2 so oxidative dehydrogenation is greatly reduced; thus some C2Ha remains unreacted. When there isless 0 2 in the feed, Le., at highC2H6compositions, oxidative dehydrogenation is more important than syngas production. This increases the production of H20 relative to Hz, and there is more H20 produced on Pt than on Rh or Pd. The increased C02 relative to CO production can be explained by the equilibrium of the Boudouard reaction (Table I). At these compositions, the reaction temperature is lower which favors the disproportionation of CO to C, and C02. The carbon is then steam reformed. Steps on Rh. The primary surface reaction sequence on Rh catalyst is the production of CO and H2. Surfacehydroxyl groups are known to be much less stable on Rh than on Pt,l5 and this could slow reaction 12 on Rh, thus explaining the lower H20 production. This resulting preference for Hz production on the Rh surface leads the C2Hs,s to experience rapid H abstractions, resulting in C, which is oxygenated and desorbs as CO. This is the second channel in Figure 6. Sreps on Pd. In the experiments on Pd, both syngas and ethylene production were initially observed in quantities comparable to those seen on Pt. However, the route to carbon deposition was very active, which led to rapid deactivation. Carbon buildup evidently blocks further oxidative dehydrogenation (eq 12), and all reactions eventually shut off. Pd may in fact form gaseous C2H4 within the catalyst, but it decomposes to C, too rapidly to escape. As discussed in the following section, reactions on Pd evidently go to chemical equilibrium which produces carbon. Equilibrium. Solid Carbon. In Figure 7a, we show the compositionsand temperaturesatwhichcokeformation (graphite) is predicted at thermodynamic equilibrium for C2H6 and air mixtures and for C2H6 and 0 2 mixtures. The compositions and temperatures obtained in the air and 0 2 experiments on Pt are also included in Figure 7a. Since most of the experiments were conducted in the region where graphite is thermodynamically predicted, most of the results shown have produced rapid carbon deposition if thermodynamic equilibria were attained. In Figure 7b, we show the gaseous selectivities and the percent carbon deposition predicted at thermodynamic equilibrium at 1200 K as a function of percent ethane in air. It is seen that at 25% C2H6 in air, where C2H4 is peaking experimentally, thermodynamics predicts negligible C2H4 production and 35% of the ethane should be converted to graphite. This was not experimentally observed when either Pt or Rh was used; both
The Journal of Physical Chemistry, Vol. 97, No. 45, 1993 11821
Ethylene Formation by Dehydrogenation of Ethane
Formation
Famanon
. 8
I
0.5
I
Exp't
5.
I
1.5
.O
Figure 7. Calculated coke (graphite) forming regimes (a) at thermodynamicequilibriumas a function of temperature and ethane composition in air and in 0 2 . Also shown in (a) are measured temperatures from Figures 1 and 2. Thermodynamicequilibriumgas phase selectivitiesand conversion of C2& to coke as a function of ethane composition at 1200 K (b). In part (c), carbon formation is not allowed.
systems appeared to exhibit steady states which were stable for many hours with no detectable deactivation. It should be noted that continued formation of even traces of solid carbon on the catalyst must completely shut off the reaction very quickly. In a typical experiment, - 5 g of carbonlmin was fed to the reactor. If only 1% of this formed carbon, then the catalyst would build up 10% of its initial weight in graphite in 3 min. Even if only 1 W of the carbon fed formed solid carbon, a 10%buildupwould m r in 5 h. Our observation of deactivation of Pd requires that no more than 0.2% of the ethane fed forms solid carbon. Thus, we conclude that there must be no solid carbon formation in steady state on Pt or Rh. This of course does not imply that there is no carbon on the catalysts but that a true steady state is attained where carbon is being gasified at exactly the same rate as it is being deposited. The thermodynamic data shown in Figure 7 for carbon deposition apply only to the graphitic form of carbon. Carbon deposition has been studied and discussed extensively on PtlGts and on other metal surfaces?O and 'coke" on catalysts has been shown to have variable properties in terms of hydrogen content and ease of removal which depend strongly on the conditions of its deposition. Both graphitic and amorphous forms of carbon
-
have been identified on Pt. It has been shown that the presence of C2H4 and C2H6 may lead to major deviations in the behavior of carbon deposition.20 Hydrocarbon Equilibrium. Since carbon deposition is not experimentally o k e d , thermodynamicequilibrium calculations were alsoperformed where solid carbon formationwas not allowed by requiring its activity to be less than unity. Figure 7c presents the resulting predicted selectivities at 1200 K as a function of percent ethane in air. Even with this limitation, thermodynamics does not predict the observed products. At thermodynamic equilibrium, lo6 too large) by cracking or pyrolysis of C2H4. Thus, we conclude that on Pt the ethylene cracking must not reach equilibrium and that the CO2 and HzO partial pressures are high enough to suppress graphite formation by CO disproportionation and reverse steam reforming. We suggest therefore, that, on Pt, sufficient gaseous C2H4 forms to effectively reduce the amount of carbon in the C, HI and 0 mixture because gaseous C2Hd does not participate in solid carbon equilibrium reactions. On Pd equilibrium in all reactions evidently occurs with heavy carbon buildup, while Rh is an intermediate situation. On Pd the primary reaction (see Figure 6) may in fact be formation of C2H4, but either this molecule decomposes further to carbon without gaseous desorption or gaseous C2H4 more easily adsorbs dissociatively on Pd than on
Pt. coaclusioos Ethane reacts rapidly with 0 2 in excess hydrocarbon to produce predominantly ethylene and CO and Hz at atmospheric pressure over monoliths coated with Pt at residence times on the order of milliseconds. The different metals lead to very different product distributions, with Rh producing more syngas and Pd deactivating due to carbon deposition. Syngas production is consistent with chemical equilibrium while CzH4 is not. These results show that ethane oxidation in a very complicated reactor (atmospheric pressure, very fast reactions, mass transfer limited, very large heat generation) can in fact be explained very simply as proceeding by a simple sequence of elementary s t e p
11822 The Journal of Physical Chemistry, Vol. 97,No. 45, 1993
(Figure 6) which are in agreement with processes suggested by surface science experiments performed under ultrahigh-vacuum experiments on clean surfaces. Production of C2H4 at 170% selectivity and 280% conversion of C2H6 in a rapid autothermal process yields more ethylene from ethane than the current industrial processes. On Rh, synthesis gas production is the dominant process, suggesting that the recently described process for conversion of CH4 to syngas15 should be adaptable to natural gas containing CzHs. The high selectivitiesto specificproducts (CO and H2 or CzH4) are strong evidence that very simple reaction pathways dominate, and formation of C2H4 and complete reaction of CzHa strongly argue that the process in initiated by oxidative dehydrogenation which is followed by &hydrogen elimination and that these steps account for up to 70% of the reaction pathways on Pt. We have also examined reactions of CsHs, n-C4Hlo, i-C4H10, and other alkanes in this reactor. These also produce high yields of olefins on Pt, with C2H4 a dominant product. This indicates that hydrocarbon cracking is significant for larger alkanes on R. This is further evidence for the reactivity of groups attached to the carbon atom adjacent to the alkyl bond (the 8-carbon).
References and Notes (1) Kolts, J. H.; Guillory, J. P. EP Patent 0205765 A2, 1986. (2) Hickman, D. A.; Schmidt, L. D. J . Catal. 1992, 138, 267-282. (3) Satterfield, C. N. Heterogeneous Catalysis in Industrial Practice, 2nd ed.; McGraw-Hill: New York, 1991.
Huff and Schmidt (4) Song, Y.; Velcnyi, L. J.; Leff, A. A.; Kliewer, W.R.;Metcalfe, J. E. In Novel Production Methods for Ethylene, Light Hydrocarbons, and Aromatics, Albright, L. F., Crynes, B. L., Nowak, S.,Eds.; Marcel Dekkar: New York, 1992;p 319. ( 5 ) Le Bars, J.; Vedrine, J. C.; Auroux, A. Appl. Coral. A 1992, 88, 179-195. (6) Kennedy, E. M.; Cant, N. W. Appl. Catal. 1991, 75, 321. (7) Moral-, E.; Lunsford, J. H. J . Catal. 1989, 118, 255-265. (8) Mendelovici, L.; Lunsford, J. H. J. Catal. 1985, 94,37-50. (9) Erdohelyi, A.; Solymosi, F. J. Catal. 1991, 129, 497-510. (10) Erdohelyi, A.; Solymosi, F. J. Catal. 1990. 123. 31-42. (1 1) Oyama,-S. T.;Middlebrook, A. M.; Somorjai, G. A. J . Phys. Chem. 1990,94,5029-5033. (12) Le Bars, J.; Vendrine, J. C.; Auroux, A.; Pommier, B.; Pajonk, G. M. J . Phys. Chem. 1992, 96,2217-2221. (13) Michalakos, P. M.; Kung, M. C.; Jahan, I.; Kung, H . H. J. Catal. 1993, 140, 226. (14) Burch, R.;Crabb, E. M. Appl. Catal. 1993, 97,49-65 (15) Hickman, D.A.; Schmidt, L. D. Science 1993, 259, 343. (16) Lang, B. Surf.Sci. 1975, 53, 317-329. (17) Wu,N. L.; Philip, J. J . Catal. 1988,113, 383-397. (18) Gallezot, P.; Leclerq, C.; Barbier, J.; Marecot, P . J . Catal. 1989, 116, 16&170. (19) Chang, T.S.;Rodriguez, N. M.; Baker, R.T. K. J . Catal. 1990,123, 486-495. (20) Bartholomew, C. H. Catal. Rev.-Sci. Eng. 1982, 24, (l), 67-112. (21) Zaera, F. Surf.Sci. 1989, 219, 453466. (22) Zaera, F. J . Am. Chem. SOC.1989, 111, 8744-8745. (23) Jenks, C. J.; Chiang, C.-M.; Bent, B; E. J. Am. Chem. Soc. 1991, 113, 63084309. (24) Bent, B. E.; Nuzzo, R. G.; Zegarski, B. R.;Dubois, L. H. J. Am. Chem. SOC.1991, 113, 1137-1142.
