Temperature-Programmed-Reaction Study on the Effect of Carbon

by temperature-programmed reaction (TPR). Gases were adsorbed at subambient temperatures and hydrogen was used as the reactive carrier gas during TPR...
0 downloads 0 Views 1MB Size
1700

I n d . Eng. Chem. Res. 1991,30, 1700-1707

Temperature-Programmed-Reaction Study on the Effect of Carbon Monoxide on the Acetylene Reaction over Pd/AI2O3 Yeung H. Park and Geoffrey L. Price* Department of Chemical Engineering, Louisiana State University, Baton Rouge, Louisiana 70803

The effect of CO on the hydrogenation of acetylene and ethylene on Pd/A1203has been studied by temperature-programmed reaction (TPR). Gases were adsorbed at subambient temperatures and hydrogen was used as the reactive carrier gas during TPR. When acetylene was adsorbed a t 223 K, ethylene, ethane, n-butane, n-hexane, benzene, cyclohexane, and other higher aliphatic oligomers were observed in the resultant spectra. Changes were observed in the TPR spectra when CO was adsorbed with acetylene, and the results were dependent upon the order of acetylene/CO adsorption. Preadsorbed CO decreased the oligomer yield, which is due to the suppression of initiation or propagation steps in the oligomerization reaction. Postadsorbed CO was not effective in reducing oligomerization except when the CO adsorption pressure approached the acetylene adsorption pressure. Both preadsorbed and postadsorbed CO blocked hydrogen adsorption a t single, isolated Pd sites, which resulted in a delay in hydrogenation of surface oligomers. Introduction The technique of temperature-programmed desorption (TPD) has been used widely to investigate the binding states and reaction sites on metal catalysts (Falcone and Schwarz, 19831, but its application to hydrocarbon adsorption on supported metal catalysts has been largely unsuccessful primarily because not all the adsorbed hydrocarbon can be recovered during thermal desorption due to decomposition and retention on the catalyst. In the case of TPD of ethylene on Pt/SiOz (Komers et al., 1969; Tsuchiya and Nakamura, 19771, for example, a self-hydrogenation product, ethane, and decomposition products, hydrogen and methane, were produced along with carbon deposits on the surface. Furthermore, significant effects of diffusion and readsorption have hindered the analysis of TPD spectra (Gorte, 1982). Recently, temperature-programmed reaction (TPR) has been developed, and this method has been especially successful when applied to adsorption of carbon monoxide, which partly undergoes decomposition into carbon and oxygen during the TPD procedure (Zagli et al., 1979). The adsorbed species can be recovered (though not in their original chemical form) from the surface by use of a hydrogen carrier gas. Studies on the behavior of these species thereby brought about a deeper understanding of the reaction mechanism of CO hydrogenation. TPR results applied to CO hydrogenation suggest that the surface species from hydrocarbon adsorption on supported metal catalysts might be recovered similarly. Knowledge on the interaction of acetylene and ethylene with Pd is important in understanding surface phenomena in the selective hydrogenation of acetylene, which is used in industry for the removal of trace acetylene from ethylene feedstocks. TPD studies on the interaction of acetylene and ethylene with metals have been performed but were limited to the study of the decomposition reactions or benzene formation from acetylene (Komers et al., 1969; Tsuchiya and Nakamura, 1977; Inoue et al., 1976; Gentle and Muetterties, 1983; Salmeron and Somorjai, 1982; Tysoe et al., 1983, 1986; Rucker et al., 1986). Carbon monoxide is routinely added in industrial processes to the raw ethylene stream to enhance the product selectivity, and many studies on the effect of CO in this reaction have appeared in the open literature (McGown et al., 1977,1978 Al-Ammar and Webb, 1978a,b, 1979; Weiss et al., 1984;

* To whom correspondence should be addressed.

Leviness et al., 1984). The effect of carbon monoxide on the adsorption of acetylene and ethylene was studied for Rh/SiOz by use of thermal desorption (Reid et al., 1973a,c), but stepwise rather than continuous desorption was used and mass spectra of eluting components were qualitatively analyzed. In this study, TPR was applied to the adsorption of acetylene and ethylene on Pd/Al2O3and Pd/SiOz to investigate the behavior of the surface species. Hydrocarbons were adsorbed at low temperatures to minimize decomposition during the adsorption procedure. TPR spectra of acetylene and ethylene in the presence of adsorbed carbon monoxide were obtained to examine the interaction of acetylene or ethylene and CO with Pd. Effects of carbon deposits on Pd on the oligomerization were also examined by acetylene TPR for comparison with the CO effect. Experimental Section Materials. A 1% Pd/Al2O3 catalyst was prepared by impregnation of Pd(NH&(NO,), on y-alumina (Linde 503, surface area 215 m2/g). A 0.2 M Pd solution was added to dried alumina with just enough distilled water to make a slurry. Water was boiled off on a hot plate, and then the catalyst was dried at 120 "C overnight and calcined at 300 OC for 4 h in a muffle furnace. Pd dispersion of the catalyst (after Hz reduction) was determined by chemisorption of carbon monoxide (using the difference of strongly bound and weakly bound CO) in a static adsorption apparatus and found to be 45.6%. Pd/SiOz @ioz surface area 266 m2/g) was prepared by the same procedure, and its dispersion was 22.4%. Hydrogen (99.999%) and helium (99.999%) from the gas cylinder were passed through a Supelpure (Supelco) purifier to remove trace oxygen. Acetylene (99.6%) was purified by bulb-to-bulb distillation using liquid nitrogen and stored in a flask connected to the batch reactor apparatus. Ethylene and carbon monoxide gases (CP grade) were used directly from the lecture bottle. Apparatus. The primary apparatus for this study is composed of two basic parts. One part is a batch recirculation loop for catalyst treatment and gas adsorption, and the other is a flow system for TPR (Figure 1). The batch reactor loop (690 mL), where the reactor is attached, is made of Pyrex glass. Adsorbates were circulated through the catalyst bed by means of a magnetically operated piston pump with check valve arrangement. The entire system could be evacuated by means of a rotary mechan-

0888-5885/91/2630-1700$02.50/00 1991 American Chemical Society

Ind. Eng. Chem. Res., Vol. 30, No. 8, 1991 1701 VACUUM RECIRCUUTION PUMP

CC DETECTOR

!1 _ - _MS _1

I I

I

Figure 1. Experimental apparatus.

i d pump. System pressures were monitored with digital readout diaphragm-type pressure gauges. A three-way stopcock arrangement allowed the reactor to be isolated from the remainder of the loop during adsorbate mixing. The reactor used in the TPR experiments was a coaxial quartz reactor with a 14-mm-0.d. jacket and 9-mm-0.d. inner tube. A quartz frit at the bottom of the inner tube supported the catalyst, and a removable thermowell (coaxially placed within the inner tube) allowed for catalyst charging. Reactants flowed downward through the jacket and then upward through the fritted disk, catalyst, and inner tube. A three-way stopcock arrangement allowed the reactor to be attached either to the recirculating batch loop or to a continuous flow of H2/He carrier gas. The effluent could be directed through the chromatographic column to the mass spectrometer (MS), or it could bypass the column and go directly to the MS. Both heating and cooling were required for TPR experiments, so a cylindrical copper block that has rod-type heaters imbedded in it for heating surrounded by a copper cooling coil was used. Compressed air, liquified in a liquid nitrogen bath, was used as the cooling medium. The block temperature was controlled by a digital temperature controller. Linear heating rates of 1-30 K/min were attainable. H2 and He carrier gases could be independently controlled and monitored with mass flow controllers and digital readout. TPR products were analyzed by a UTI lOOC quadrupole mass spectrometer. A jet separator roughed with an independent mechanical vacuum pump provided the interface to atmospheric pressure and removed the bulk of the H2and/or He carrier gases. A crw-pattem leak valve was used to split off carrier flow rates in excess of 30 cm3/min. The system could also be operated as a rough GC/MS system, which was used in product identification and calibration, by switching a chromatographic column in series with the jet separator by means of a six-port chromatographic sampling valve. For measurement of the reactor temperature during TPR experiments, a digital thermocouple module with RS-232 communications was used to monitor the reactor temperature. An IBM PS/2 Model 60 personal computer gathered raw mass spectral intensities from the MS and temperature measurements from the thermocouple module. Procedures. For TPR experiments, 50 mg of 200325-mesh catalyst particles were packed in the reactor and reduced at 773 K for 1 h in the recirculation system filled with 53.2 kPa of H2 and 53.2 kPa of He with a liquid

nitrogen trap in the circulating loop. After reduction, the catalyst was flushed with He gas at 773 K for 30 min to remove adsorbed H2, and it was then cooled to the adsorption temperature (normally 223 K) under He flow. In the standard experiment, a mixture of 73.2 Pa of acetylene in 106.4 kPa, of He was circulated over the catalyst at 223 K for 15 min. In preparing the adsorption mixture, C2Hz was prediluted in He so that very small C2H2partial pressure doses could be repeated precisely. After adsorption, the apparatus was switched to TPR mode and the catalyst was flushed with He for 30 min to remove weakly bound acetylene and self-hydrogenation products. The MS could be used to monitor this postadsorption flushing process if desired. The amount of acetylene, adsorption temperature, and flushing time could be varied from the standard conditions and will be noted as necessary. The catalyst was then further cooled to 173 K, a flow of 20 cm3/min H2 and 180 cm3/min He was established, and a linear temperature program at 5 K/min was initiated. Mass spectra were recorded every 20 s until the catalyst reached 673 K. Ethylene TPR was done under the same conditions as acetylene TPR except for the standard adsorption temperature (203 K) and the TPR starting temperature (153 K). These standard conditions for acetylene TPR experiments were determined after several exploratory TPR experiments at various conditions in an effort to get meaningful TPR spectra. The standard adsorption temperature (223 K)was chosen because lower temperatures (e.g., 203 K) resulted in heavy physisorption of acetylene which (as we will discuss later) lead to very high yields of ethane. Higher adsorption temperatures (e.g., 243 K)led to the loss of the adsorbate (as oligomers) during the postadsorption flush. A small amount of catalyst (50 mg) with a particle size of 200/325 mesh was adopted to facilitate catalyst flushing after the adsorption step. In TPR experiments designed to study CO effects, CO was adsorbed in the batch recirculation system, either before or after adsorption of the hydrocarbon. In the case of preadsorption, the catalyst was exposed to CO for 15 min at the same temperature as hydrocarbon adsorption and flushed briefly with He. Then, the hydrocarbon was adsorbed and the standard TPR procedures followed. For postadsorption studies, the order of adsorption was reversed. In an investigation on the effect of carbon deposits on the catalyst, surface carbon was deposited on Pd/A1203 by exposing the reduced catalyst to 73.2 Pa of acetylene at 673 K for 15 min. After the catalyst was flushed for 10 min with He, a standard TPR of acetylene followed. Calculations. In order to convert raw mass spectra into moles of each hydrocarbon species eluting at any given instant in a TPR or TPD experiment, we used the following procedure. We first determined the global composition of the TPR product by trapping all of the products from a complete TPR run and then injecting this sample into the analytical system in GC/MS mode. The MS fragmentation patterns were used to identify the components, and then a standard mixture of all significant componenta was prepared and injected in a known quantity to generate fragmentation patterns based on absolute ion intensities per mole of hydrocarbon. Carefully selected, characteristic ions of each component were monitored in this process. This allowed us to construct a matrix such that the columns represented the fragmentation patterns of individual components and the rows represented particular mass numbers. An experimentally determined mass spectrum at any instant is the product of the matrix and

1702 Ind. Eng. Chem. Res., Vol. 30, No. 8, 1991 "I

I10

mass 78

IO0 90

8 EO a, IO 4 80 VI

50 2 40 d

30 20

IO 0

150

250

350 450 550 Temp (K)

650

Figure 2. Raw TPR spectrum (maas spectral form) of C2Hz adsorbed on Pd/A1203at 223 K.

the vector representing the moles of each component eluting at that instant. In order to effectively solve for the mole vector, we set up a least-squares minimization of the difference between the actual spectrum and the computed spectrum. Nonnegativity constraints were imposed on all components of the mole vector. The calculational procedure allowed a quantitative analysis of the TPR spectra, but we did not attempt to calculate kinetic parameters for desorbed products (such as activation energies of desorption) due to virtually unavoidable experimental difficulties associated with the TPR experiment. Acetylene physisorption at 223 K was significant, and readsorption of this acetylene on the catalyst was very apparent as we will discuss. Our standard TPR conditions were checked for transport effects according to Gorte's criteria (1982), and there is a possibility of concentration gradients in the catalyst pellet even though other problems such as lag time in the catalyst bed, gas diffusion within the catalyst pore, and readsorption of TPR products (other than acetylene) were not probable. TPR experiments were therefore focused on the qualitative information on the effect of CO on the acetylene and ethylene reactions.

Results As an example of the calculation procedure used to determine moles of each species eluting at any instant, the TPR of acetylene in raw mass spectral output form is given in Figure 2. This figure shows the intensities of masses 30, 41, 43, and 78 as a function of temperature. Masses 14, 15, 26, 28, 56, 57, 58, 71, 84, 85, and 86 were also observed but are not shown in Figure 2 in the interest of simplicity. These masses were chosen for observation because they are characteristic of the molecular species that are produced in the TPR experiment. Acetylene, ethylene, ethane n-butane, n-hexane, benzene, and cyclohexane along with higher paraffins including n-octane and n-decane and traces of pentane, 2-methylpentane, 3methylpentane, methylcyclopentane, and heptane were determined to be the products of this TPR experiment. The n-octane, n-decane, and trace components were difficult to determine quantitatively and were therefore not used in the quantitation calculation. Figure 3 gives the results of the quantitation calculation, and integrated areas from Figure 3 are tabulated in Table I. These results contrast greatly with the TPD results also given in Table I. Almost 50% more total carbon in the form of various hydrocarbons and almost twice as much carbon as oligomers (C4+) was recovered in the TPR experiment versus the TPD experiment. In order to investigate effects associated with physically adsorbed acetylene, a series of experiments on Pd/A1203, Pd/Si02, A1203,and Si02were performed. Results given

150

250

350 450 550 Temp (K)

650

3 00-

150

250

350 450 550 Temp (K)

650

Figure 3. (a, b) TPR spectrum (molar form) of C2H2adsorbed on Pd/A120, at 223 K. Table I. Comparison of Acetylene TPR and TPD (73.2-Pa CzHlPartial Pressure, 15-min Exposure, 223 K, 50 mg of Pd/Alumina) TPR TPD ctmol/g T,K ctmol/g T,K acetylene 1.16 205 38.07 253 ethylene 7.78 213 ethane 34.82 215 n-butane 8.34 237 1.98O 304 n-hexane 2.99 288 benzene 0.70 325 7.94 334 0.88b 640 cyclohexane 7.56 340 total carbonc 194 132 =Sum of all C4 products; primarily butenes. *Broad peak, not complete by 673 K. cTotal carbon recovered, pmol/g.

in Table I1 show the effect of physisorbed acetylene, and we immediately note physisorbed acetylene in large quantities remaining on the alumina support even after extended flushing (>30 min) with He. The Si02support shows considerably less physisorbed C2H2than A120,,and this apparently translates to much smaller quantities of C2 product which are obtained from Pd/Si02 under TPR conditions than those obtained from Pd/A1203. Notice products from Pd/Si02 compare closely to that C& those from Pd/Ai20,. Also, compare the yield of C2speciea from A1203 (36.80 pmol/g as acetylene) to that from Pd/AI2O3 (38.42 pmol/g as ethylene and ethane) and compare C2species from Si02 (1.85 pmol/g) to those from Pd/Si02 (4.48 pmol/g). This comparison suggests that approximately 2 pmol/g C2products appear from material adsorbed on Pd and the remainder can be attributed to acetylene originally adsorbed on the support. In the case of Pd/A1203,the vast majority of the C2products therefore may be attributed to material originally adsorbed on the support. However, we find that ethane and ethylene elute from Pd/A1203during TPR at much lower temperatures (column 1) than acetylene from A1203 (column 3). Our

Ind. Eng. Chem. Res., Vol. 30,No. 8, 1991 1703 Table 11. Effect of Physisorbed Acetylene on TPR (146.3-Pa CzHz Partial Pressure, 15-min Exposure, 223 K, 100 mg of Catalyst) Pd/Al208 Pd/Si02 A1203 Si02 rmollg T,K rmollg T,K rmollg T,K rmollg T,K 251 36.80 242 1.85 acetylene 10.01 219 ethylene 28.41 222 4.48 248 ethane 4.24 242 4.21 254 n-butane 1.43 292 1.41 307 n-hexane 1.06 337 1.42 321 benzene 0.12' 573 0.49 523 344 cyclohexsme 5.67 350 2.25

'Broad peak, not complete by 573 K. Table 111. Effect of Adsorbate Pressure on TPR of Acetylene (15-min Exnosure. 223 K. 50 mg - of Pd/Aluminal P = 73.2 Pa Pc H = 20.0 Pa (PdF2H2 = 1/9.5) (Pd/d2H2 = 1/2.6) acetylene ethylene ethane n-butane n-hexane benzene cyclohexane C8+

1.16 7.78 34.82 8.34 2.99 0.70 0.88' 7.56 6.3 X loab

205 213 215 237 288 325 640 340 497

1.42 17.90 5.04 1.34

215 220 230 286

0.56 0.88' 1.77 1.3 X

490 673 338 505

1Pb

'Broad peak, not finished by 673 K. bIntegrated intensity of mass 43 characteristic of C8 and heavier paraffins.

interpretation of these phenomena is that, as physisorbed acetylene elutes slowly from the support during TPR (in the case of Pd/A1203),it readsorbs on vacant Pd sites that have been created by hydrogenation of acetylene and is hydrogenated to ethane and ethylene. As we will discuss shortly, there are single, isolated Pd sites which can accomodate hydrogen adsorption and supply hydrogen for this reaction. Since amounts of physisorbed ethylene and ethane on A 1 2 0 3 at 223 K determined by TPR in separate experiments were small compared to the amount of acetylene physisorbed on Al2O3, ethane and ethylene products from TPR over Pd/A1203 would elute much faster than as strongly acetylene since they are not adsorbed on A1203 as acetylene (acetylene probably undergoes many adsorption/desorption cycles before eluting) and therefore appear as product at temperatures lower than acetylene from

A1203. Though the C2 products in acetylene TPR appear to arise primarily from physisorbed acetylene, the C4and C6 oligomers are rather insensitive to the presence of physically adsorbed acetylene under the adsorption conditions that have been employed, and these oligomers were not observed during the postadsorption flushing process.

'0

T= 255K

Therefore, the oligomers appear to represent the products that can be obtained from the species which were originally adsorbed on Pd. Table 111shows the effect of acetylene adsorbate pressure on the resultant TPR. When 20.0 Pa of acetylene was adsorbed instead of 73.2 Pa, higher aliphatic oligomers were produced at the expense of C4and CBaliphatic oligomers but cyclohexane was strongly attenuated. For reference purposes, the TPR results of CO (under the same conditions as acetylene TPR) are presented in Figure 4. A broad CO band extending from 225 to 375 K and methane band at 501 K are the primary features of this spectrum. The results of acetylene TPR with both preadsorbed and postadsorbed CO are summarized in Table IV. When 20.0 Pa of CO was preadsorbed (column 2), the yield of each oligomer was reduced by 2745% and all peaks were shifted to higher temperatures (except for the broad, high-temperature benzene peak). However, when the same amount of CO was adsorbed following acetylene (postadsorbed, column 31, aliphatic oligomer yields showed slight increases while large increases for benzene and cyclohexane were evident. Peak temperatures were again higher than the base case without CO. When the pressure of postadsorbed CO was increased to 73.2 Pa (column 41, the oli-

Table IV. Effect of CO on Acetylene TPR (73.2-Pa CIHz Partial Pressure, 15-min Exposure, 223 K, 50 mg of Pd/Alumina) preads CO postads CO postads CO no CO (20.0Pa) (20.0Pa) (73.2Pa) rmol/g T,K rmollg T,K /Jmol/g T,K rmoVg T,K methane 5.60 532 2.86 525 5.76 534 acetylene 1.16 205 34.94 253 30.44 247 28.98 252 ethylene 7.78 213 0.84 251 1.76 249 ethane 34.82 215 11.62 258 20.06 251 13.18 255 n-butane 8.34 237 5.24 265 8.68 258 6.10 262 n-hexane 2.99 288 2.10 291 3.63 287 2.37 290 benzene 0.70 325 0.16 330 3.55 335 1.42 340 0.88' 640 0.71' 639 1.12' 600 0.81' 610 cyclohexane 7.56 340 2.86 371 8.64 360 6.45 360

'Broad peak, not complete by 673 K.

1704 Ind. Eng. Chem. Res., Vol. 30, No. 8, 1991

f+

Acetylene

200 220 240 260 280 300 320 340

Temp (K) Figure 5. TPR spectrum of C2Hzwith preadsorbed CO on Pd/ A1,03 at 223 K.

Table V. Ethylene TPR and Effect of CO on Ethylene TPR (73.2-PaC2H,Partial Pressure, 15-min Exposure, 50 mg of Pd/Aluminab Tab 203 K Tab= 298 K no CO with COO no CO methane ethylene 8.0 ethane 24.0 173 5.0 n-butane n-hexane benzene 0.5b 648 C8+ 4.2 X lo'( 481 4.1 8.2 X IO6' 556 9.8

7.0 228 246 8.0 0.7 0.1 X X

421 182 256 364

lPC 492 3.4 X l V c 482 l V c 553 1.3 X loac 548

O20.'0 Pa of CO preadsorbed. *Not complete by 673 K. (Integrated intensity of mass 43 characteristic of CBand heavier paraffins.

gomer yields were reduced slightly and the peaks shifted to even higher temperatures. Large effects on C2products are also noted. For all the cases with CO, acetylene was the primary C2product, while ethane was the major product without CO. Peak temperatures for C2products were observed at 35-45 K higher when compared to the case without CO. The C2 and C4 peak shapes are interesting in the case of preadsorbed CO and are therefore included in Figure 5. Notice that ethane and butane bands begin abruptly on the acetylene tail. This contrasts to the case without CO (Figure 3) where ethane and butane peaks are more symmetrical and appear at nearly the same temperature as acetylene. TPR spectra of ethylene adsorbed at 203 and 298 K are given in Table V. When ethylene was adsorbed at 203 K, ethane was produced along with benzene and high oligomers. Adsorption at room temperature resulted in methane, butane, and hexane in addition to reduced amounts to ethane and high oligomers. Table V also includes the effect of CO on ethylene TPR. With preadsorbed CO, ethane and unreacted ethylene were produced at much higher temperature along with a reduction in higher oligomer yields. Similar experiments over a Pd/A120, catalyst that was poisoned by surface carbon were also performed, and the results are given in Table VI where a significant effect on the TPR spectrum of acetylene is noted. The oligomer yields decreased by roughly one-half to one-third and the peak temperatures shifted to higher values by 15-24 K with carbon deposition on the catalyst. Decreases in the yields were larger for the higher oligomers, while little change was observed for C2 products. The acetylene TPR spectrum for the fresh catalyst in Table VI shows a significant difference in the yield of cyclic products from that in Table 1. This problem is not asso-

Table VI. Effect of Carbon Deposits on Acetylene TPR (73.2-PaCIHaPartial Pressure, 15-min Exposure, 223 K, 50 mg of Pd/Alumina) C-deposited fresh Pd/A1208 Pd/A1209 pmol/g T,K pmol/g T,K acetylene 4.60 213 ethylene 10.95 216 7.34 214 ethane 37.19 221 44.78 219 n-butane 7.57 237 3.68 252 n-hexane 2.99 287 0.83 312 benzene 3.16 328 1.78 347 0.96" 629 cyclohexane 10.59 341 2.77 361 Broad peak, not complete by 673 K.

ciated with run to run variability but is due to a change in the prediluted acetylene/helium mixture. Similar changes in cyclic oligomer yields were also observed in the preliminary experiments coupled with small changes in the acetylene adsorbate pressure. The extremely low partial pressures of adsorbates used in these experiments (20-200 Pa or 0.15-1.5 " H g ) are difficult to reproduce accurately, but as long as the same dilution mixture was used, run to run reproducibility was observed. Therefore, in each table presented in this paper where we make column to column comparisons, the same dilution mixture was used. Discussion Interaction of C2H2and C2H4 with Pd i n the Absence of CO. When acetylene is adsorbed on Pd at the base conditions of 223 K, the surface species were recovered primarily as aliphatic oligomers (n-butane, nhexane) and cyclic oligomers (benzene, cyclohexane) upon TPR as in Table I. C2 products are believed to be the result of acetylene physisorbed on A1203 as discussed previously. Aliphatic oligomers are observed in addition to benzene which has been observed by other researchers using TPD (Tysoeet al., 1983,1986; Rucker et al., 1986). Cyclohexane observed in our TPR is believed to be the hydrogenated form of the benzene product. Our results are similar to those from a previous thermal desorption study (Reid et al., 1973c) on the existence of C, and C6 products from Rh/silica. Aliphatic oligomers are also similar to products observed from the direct hydrogenation of acetylene (Bond and Wells, 1965). Surface oligomerization is probably completed during adsorption of acetylene rather than during the TPR experiment since acetylene interactions with Pd would be limited by the surface oligomers which had already formed during adsorption and because acetylene which could undergo further oligomerization was only available in the initial stages of TPR during desorption of the physisorbed acetylene. The products from the interaction of this acetylene with Pd were C2 species, not oligomers as we have already discussed. Other researchers have also observed oligomer formation on Pd at low temperatures. In a TPD study by Tysoe et al. (1983), benzene was formed from acetylene on P d [ l l l ] when adsorbed at 195 K, and benzene elution began at 195 K and reached a maximum at 230 K (which is comparable to our adsorption temperature of 223 K) in the resultant TPD spectrum. Our results also demonstrated that oligomers can be formed directly on Pd during adsorption at 223 K because masses 43 and 78 were observed initially while monitoring the effluent during the postadsorption flush and these are the result of benzene and C4 paraffins and olefins. Acetylene TPR spectra were found to depend upon the adsorption conditions, of which the partial pressure of

Ind. Eng. Chem. Res., Vol. 30, No. 8, 1991 1705 acetylene during adsorption was the most significant variable affecting the product distribution among the oligomers. When the acetylene pressure was low (20.0 Pa), higher oligomers were produced in preference to C4 and C6oligomers (Table In). Monolayer coverage on Pd might not be obtained in the case of 20.0 Pa acetylene pressure even though the acetylene pressure corresponded to C2H2/Pd= 2.6 because a significant fraction of the acetylene was physisorbed on the support. When the surface coverage is less than a full monolayer, surface oligomers can form without steric hindrance from neighboring acetylene molecules, which contrasts with the case of the fully covered surface where the surface is sterically hindered. Sheridan (1945) and McGown et d. (1978) have discussed the packing arrangements of acetylene at high and low surface coverages. According to their theories, high surface coverages favor smaller oligomers in a greater overall abundance than low surface coverages; these theories compare favorably with our observations. Therefore, the partial pressure effect on oligomer formation can be traced to steric effects on the surface. Oligomerization of acetylene to aliphatics apparently is initiated by a free radical which is formed either from partial hydrogenation of acetylene (Sheridan, 1945)or from dissociation of acetylene (Margitfalvi et al., 1981; Leviness et al., 1984). This free radical reacts in series with associatively adsorbed acetylene to form oligomers. We have observed some acetylene decomposition upon adsorption at 223 K as evidenced by the formation of a small amount of ethylene which was observed in the effluent during the postadsorption catalyst flush. This ethylene must be produced via self-hydrogenation of acetylene during adsorption. Thus, a source of radicals from both dissociated acetylene and vinyl radicals is readily apparent. TPR bands for benzene and cyclohexane (which are thought to come from the same common intermediate) were much broader and had larger areas than n-hexane (which has same number of carbons) even though the formation of the cyclic oligomers is more difficult due to a restricted configuration of molecules. This suggests that sites for the production of c6 cyclic oligomers are distinct from sites for aliphatic oligomers. In fact, if we compare the changes in yields of cyclohexane and hexane for different adsorption pressures (see Table III), we note that the cyclohexane yield goes up by a factor of 4.3 while the hexane yield goes up by a factor of 2.3 when the acetylene adsorption pressure is increased from 20 to 73 Pa. This observation is further evidence that cyclic oligomers form on distinct sites and that these sites are more difficult to fill completely than aliphatic oligomer sites. Some possibilities concerning the exact identification of these sites may be extracted from the open literature. In acetylene hydrogenation over a Ni/pumice catalyst, Sheridan (1945) suggested that the sites for aliphatic oligomers were Ni[ 1101, where polymerization of adsorbed acetylene is allowed without the rupture of C-Ni bonds due to wider spacings of the metallic sites. P d [ l l l ] was found to be the most active among low Miller index planes in benzene formation in several studies (Tysoe et al., 1983, 1986; Rucker et al., 1986). Inoue et ai. (1976) suggested the stepped surfaces around Pd[lll] terraces as the active sites for benzene formation. We have no direct evidence to discriminate among these possibilities. A desorption behavior different from aliphatic oligomers has also been observed for cyclic compounds. The yield of cyclic oligomers was similar for both TPR and TPD while aliphatic oligomers were not observed in TPD (Table I). Cyclohexane was the primary cyclic product in acety-

lene TPR while benzene was the main product in TPD. These observations are consistent with a mechanism suggested by Tysoe et al. (1983) that involves an initial formation of a 1,3,Bhexatriene species followed by closure to cyclohexadiene. Formation of 1,3,5-hexatriene occurs by a series of bimolecular reactions. Cyclohexadiene can be desorbed as benzene (after Pd-induced dehydrogenation) during TPD or as cyclohexane via hydrogenation during TPR. There is no net consumption of hydrogen in the formation of benzene, so its formation is allowed even in the absence of hydrogen (TPD). The TPR spectra of ethylene and acetylene are very different particularly in that the degree of ethylene oligomerization is very small compared to the degree of acetylene oligomerization (see Tables I and V). When we monitored the reactor effluent during the postadsorption reactor flush following ethylene adsorption, the catalyst flushed quickly indicating that very little weakly bound material was present and therefore physisorption of ethylene was minor. However, a significant amount of ethane was detected during flushing, and it follows that ethylene undergoes self-hydrogenation extensively. When ethylene was adsorbed at 203 K, ethane, benzene, and high oligomers were detected upon subsequent TPR, but the TPR spectrum following 298 K adsorption was drastically different where methane, butane, and hexane were observed, the ethane and high oligomer yields were strongly attenuated, and benzene was not detected (Table V). Butane and hexane probably originate from dissociatively adsorbed ethylene, which is favored a t the higher adsorption temperature. Oligomerization of ethylene contrasts to the case of acetylene, where C4 and c6 oligomers are formed extensively from adsorption at 223 K and the mechanism of ethylene oligomerization is clearly different from that of acetylene. The relative difficulty of ethylene oligomerization seems to come partly from its double bond as was suggested by Sheridan (1945). When ethylene is adsorbed associatively, each carbon atom shares one electron with the metal atom so the molecule has no unshared electron pair that can interact with another adsorbed molecule. This is unlike acetylene, which remains unsaturated upon adsorption on a metal and is reactive toward other hydrocarbons on the surface so that polymerization can propagate. Our results are consistant with the report of Stephens (1958), which requires residues of ethylene self-hydrogenation on Ni for the formation of C4oligomers. Webb (1978) also suggested that C4 species arise from random polymerization of dissociatively adsorbed ethylene residues. The methane product from TPR following ethylene adsorption at 298 K appears to be related to a distinct surface species. Ethylene is known to form a stable ethylidyne species (=C-CH3)when it is adsorbed at room temperature on Pd (Beebe et al., 1985; Beebe and Yates, 1986) and was observed to decompose at 450 K. Other researchers (Komers et al., 1969; Tsuchiya and Nakamura, 1977) have observed methane from TPD of ethylene over Pt at 493 K, and Salmeron and Somorjai (1982) observed decomposition of ethylidyne into C and CH surface species at 493 K. The formation and hydrogenation of ethylidyne is known to be a very slow process below 353 K (Beebe et al., 1985). The source of methane from the TPR of ethylene adsorbed at 298 K is likely this species, which undergoes decomposition instead of hydrogenation due to ita low reactivity. Ethylidyne apparently cannot be formed at 203 K, so no methane was observed in the resultant TPR.

1706 Ind. Eng. Chem. Res., Vol. 30, No. 8, 1991

CO Effect on the Interaction of CzHzand CzH4 with Pd. The TPR spectrum of CO (Figure 4) yields both CO and CHI, indicating that CO adsorbs on Pd in two states composed of weakly adsorbed CO which desorbs molecularly at low temperatures and strongly adsorbed CO which reacts with Hz and desorbs as methane at higher temperatures. In the TPR spectrum of acetylene with CO (Table IV), the existence of strongly adsorbed carbon monoxide on the catalyst surface is observed as a methane peak around 532 K, a slightly higher temperature than in the TPR spectrum of pure CO. This methane peak was always observed in acetylene/CO experiments regardless of whether CO was preadsorbed or postadsorbed, which indicates that there are particular sites where only CO can adsorb. These sites may be single isolated sites that cannot accommodate acetylene. A global effect of CO on the adsorption of acetylene and ethylene that has been observed in the TPR experiments is the suppression of self-hydrogenation. When CO was adsorbed before acetylene, no ethane or ethylene was observed in the effluent during the postadsorption flush. This effect has also been observed on Rh/SiOz (Reid et al., 1973a,c). The reduction in self-hydrogenation is probably caused by CO blockage of single sites which accommodate the hydrogen product of dissociative chemisorption of hydrocarbons, and therefore, no hydrogen is available for self-hydrogenation. The product distribution of Cz products from TPR of acetylene with adsorbed CO (Table IV) indicates how CO affects the hydrogenation of acetylene. During the initial stages of TPR, physisorbed acetylene desorbs from the support and, when no CO is present, readsorbs on available metal sites for hydrogenation to ethane (as discussed above), which is observed as the primary product. When CO was adsorbed, unconverted acetylene was the predominant product in the early stages of TPR. Acetylene eluting from the support could not be converted to ethane because single metal sites, which normally are available to supply hydrogen for the reaction, have been blocked by

co.

The effect of CO on oligomerizationis shown in the TPR spectra of acetylene, which are summarized in Table IV. The amount and the order of CO adsorption strongly influenced the results. When 20.0 Pa of CO was preadsorbed, a decrease in the yield of all oligomers was observed. However, postadsorption of the same amount of CO resulted in an increase in the oligomer yield particularly with respect to c6 cyclic products. When the CO pressure was increased to 73.2 Pa (the same as the acetylene pressure), postadsorbed CO caused a slight decrease in the oligomer yield. The effect of preadsorbed CO on oligomerization is caused by CO blockage of acetylene adsorption and suppression of acetylene dissociation. Acetylene probably cannot displace preadsorbed CO completely. In the case of Rh/SiOz only 40% of the preadsorbed CO could be displaced by acetylene (Reid et al., 1973b). CO left on the surface would be especially effective in suppressing oligomerization because oligomerization requires nearby adsorbed acetylene molecules to proceed. A reduced dissociation of acetylene could also prevent formation of free radicals which would destroy initiation and result in suppression of oligomerization. The relative importance of each effect in suppressing oligomerization probably depends on the pressure of CO and acetylene. Suppression of acetylene dissociation would be the main effect when the CO pressure is much lower than the acetylene pressure, since CO may not be able to compete for acetylene sites

effectively. However, CO could still occupy single sites which cannot accommodate acetylene and block these sites, which normally would accept the hydrogen product of acetylene dissociation. All resultant spectra of acetylene TPR modified with CO yielded shifts of products to higher temperatures (Table IV). This effect is probably due to a limited hydrogen supply during TPR caused by hydrogen site blockage by CO. Unsaturated products are difficult to desorb, so a delay in hydrogenation results in the observed shifts. The shift was also observed for the low-temperature benzene peak, which does not require hydrogen for its desorption. The mechanism for benzene formation discussed above may explain this trend. CO blocks the vacant sites necessary to accommodate hydrogen which comes from the benzene precursor during the dehydrogenation step so that the desorption of benzene is delayed. The observed temperature shift was more severe for ethane and butane products than for higher oligomers. Hydrogenation of surface oligomers apparently is most affected in the early stages of TPR when the hydrogen supply is more strongly suppressed by CO. As the temperature is increased, weakly bound CO desorbs, creating sites for the hydrogen supply. Hydrogenation and subsequent desorption of surface species result. The existence of weakly bound CO has been demonstrated by TPR experiments of CO without acetylene and is still possible in the presence of acetylene. Blyholder (1964) suggested that CO adsorbs on corner and edge sites strongly and weakly on plane sites. Single plane sites may exist as geometrically isolated sites surrounded by multiply adsorbed acetylene molecules. Results on the effects of CO are comparable, in some ways, to the effects of carbon deposits where decreases in oligomer yields and shifts in peak temperatures were observed (Table VI). Since the source of carbon deposited on Pd was acetylene, hydrocarbon sites active for acetylene adsorption should be blocked effectively. The results reflect this expectation, and a geometric effect on oligomerization is clearly shown in the product distribution. The butane yield decreased by half while the yields of n-hexane and c6 cyclic compounds were cut to one-third. Higher oligomers were suppressed to a greater extent because oligomerization requires adjacent adsorbed species for propagation. However, carbon deposition had little effect on Cz products and acetylene was not the major Cz product as was observed for CO poisoning. Peak temperatures for Cz products showed virtually no shift while significant shifts were noted for oligomers. Since carbon deposits are probably located on sites normally associated with acetylene adsorption, which are multiple sites, single sites would be left unaffected and these sites can promote hydrogenation particularly of Czspecies. Also, hydrogenation of oligomer precursors requires more hydrogen than Cz species so geometric effects of carbon should be more evident as observed. Thus, carbon deposition has little effect on Cz formation in contrast to CO, which can block single sites for hydrogen adsorption. One of the physical differences in the poisoning effects associated with surface carbon versus carbon monoxide is due to the fact that CO adsorption can be reversed by raising the temperature while carbon blockage of sites is virtually irreversible and cannot be removed during TPR. Carbon deposits are therefore still effective at high temperatures during TPR, an this results in upscale temperature shifts for high oligomers that are even larger than shifts caused by CO.

Ind. Eng. Chem. Res., Vol. 30, No. 8, 1991 1707 When the effects of CO on acetylene adsorption are compared to the effects of carbon deposits as discussed above, the existence of single, isolated Pd sites for Hz adsorption that can be blocked by CO becomes even more apparent. Our earlier studies (Park and Price, 1991)found from deuterium tracer experiments that the effect of CO on acetylene hydrogenation is due to the blockage of hydrogen adsorption by CO. The results in this paper further amplify these findings and have helped to identify the affected hydrogen adsorption sites as single, isolated sites. The effect of CO on ethylene adsorption is similar to the effect on acetylene particularly in the suppression of hydrogenation. When CO was preadsorbed on Pd prior to TPR of ethylene adsorbed at 203 K, unconverted ethylene was observed in the effluent, which was not observed in the case of pure ethylene. Ethylene could not undergo hydrogenation to ethane due to CO blocking the hydrogen supply as in the case of acetylene TPR. The decrease in the yield of high oligomers with preadsorbed CO is probably due to CO suppressing decomposition of ethylene, which precedes oligomerization of ethylene.

Conclusions The primary effects of CO on the hydrogenation of acetylene are the abilities of CO to block hydrogen adsorption particularly at isolated Pd sites and to slow dissociative adsorption of hydrocarbons by blocking sites that would normally accommodate the hydrogen product from the dissociative adsorption. Geometric effects of CO and coke deposits are also apparent in attenuating the production of high oligomers. Distinct Pd sites are operable, including isolated sites where Hzor CO can adsorb and sites that are rather specific for forming cyclic oligomers. Acknowledgment We thank the Exxon Foundation for financial support. Literature Cited Al-Ammar, A. S.; Webb, G. Hydrogenation of Acetylene over Supported Metal Catalysts. Part 1. Adsorption of [“C] Acetylene and [“C] Ethylene on Silica Supported Rhodium, Iridium and Palladium and Alumina Supported Palladium. J. Chem. SOC., Faraday Trans. 1 1978a, 74, 195. Al-Ammar, A. S.; Webb, G . Hydrogenation of Acetylene over Supported Metal Catalysts. Part 2. [“C] Tracer Study of Deactivation Phenomena. J.Chem. Soc., Faraday Trans.1 1978b, 74,657. Al-Ammar, A. S.; Webb, G. Hydrogenation of Acetylene over Supported Metal Catalysts. Part 3. [I4C]Tracer Studies of the Effects of Added Ethylene and Carbon Monoxide on the Reaction Catalyzed by Silica-Supported Palladium, Rhodium, and Iridium. J. Chem. SOC.,Faraday Trans. 1 1979, 75, 1900. Beebe, T. P., Jr.; Yates, J. T. An In Situ Infrared Spectroscopic Investigation of the Role of Ethylidyne in the Ethylene Hydrogenation Reaction on Pd/Al2O9. J. Am. Chem. SOC.1986, 108, 663. Beebe, T. P., Jr.; Albert, M. R.; Yates, J. T. Infrared Spectroscopic Observation and Characterization of Surface Ethylidyne on Supported Palladium on Alumina. J. Catal. 1985, 96,1. Blyholder, G. Molecular Orbital View of Chemisorbed Carbon Monoxide. J. Phys. Chem. 1964,68, 2772. Bond, G. C.; Wells, P. B. The Hydrogenation of Acetylene I1 The Reaction of Acetylene by Alumina-Supported Palladium. J. Catal. 1965, 5, 65. Falconer, J. L.; Schwarz, J. A. Temperature-Programmed Desorption and Reaction Applications to Supported Catalysts, Catal. Reo. -Sci. Eng. 1983, 25 (2), 141.

Gentle, T. M.; Muetterties, E. L. Acetylene, Ethylene, and Arene Chemistry of Palladium Surfaces. J.Phys. Chem. 1983,87,2469. Gorte, R. J. Design Parameters for Temperature Programmed Desorption from Porous Catalysts. J. Catal 1982, 75, 164. Inoue, Y.; Kojima, I.; Moriki, S.; Yasumori, I. Template Effect of Adsorbate Acetylene on Catalysis by Palladium Surface. Proc. Znt. Congr. Catal., 6th 1976, 1, 139. Komers, R.; Amenomiya, Y.; Cvetanovic, R. J. Study of Chemisorption and Hydrogenation of Ethylene on Platinum by Temperature Programmed Desorption. J. Catal. 1969, 15, 293. Leviness, S.; Nair, V.; Weiss, A. H.; Schay, Z.;Guczi, L. Acetylene Hydrogenation Selectivity Control on PdCu/AlzOa Catalysts. J. Mol. Catal. 1984,25, 131. Margitfalvi, J.; Guczi, L.; Weias, A. H. Reaction of Acetylene during Hydrogenation over Pd Black Catalyst. J. Catal. 1981, 72,185. McGown, W. T., Kemball, C.; Whan, D. A,; Surrel, M. S. Hydrogenation of Acetylene in Excess Ethylene on an Alumina Supported Palladium Catalyst in a Static System. J. Chem. SOC., Faraday Trans. 1 1977, 73,632. McGown, W. T.; Kemball, C.; Whan, D. A. Hydrogenation of Acetylene in Excess Ethylene on an Alumina-Supported Palladium Catalyst at Atmospheric Pressure in a Spinning Basket Reactor. J. Catal. 1978,51, 173. Park, Y. H.; Price, G. L. The Effect of Carbon Monoxide on the Selective Hydrogenation of Acetylene over Pd/A1209. Znd. Eng. Chem. Res. 1991, in press. Reid, J. U.; Thomson, S. J.; Webb, G. Radiochemical Studies of Chemisorption and Catalysis XI. A Mass-Spectrometric Study of the Interaction of Ethylene, Carbon Monoxide, and Hydrogen with a Rhodium-Silica Catalyst. J. Catal. 1973a, 29, 433. Reid, J. U.;Thomson, S. J.; Webb, G. Radiochemical Studies of Chemisorption and Catalysis XII. A Study of the Adsorption of [ C y Acetylene, [C14]ethylene,and [C14]Carbon Monoxide on a Silica- and Alumina-Supported Rhodium Catalyst. J. Catal. 1973b,30, 372. Reid, J. U.;Thomson, S. J.; Webb, G. Radiochemical Studies of Chemisorption and Catalysis XIII. A Mass-Spectrometric Study of the Interaction of Acetylene, and Carbon Monoxide with a Rhodium-on-Silica Catalyst. J. Catal. 1973c, 30, 378. Rucker, T. G.; Logan, M. A.; Gentle, T. M.; Muetterties, E. L.; Somorjai, G. A. Conversion of Acetylene to Benzene over Palladium Single-Crystal Surfaces I. The Low-Pressure Stoichiometric and High-pressure Catalytic Reactions. J.Phys. Chem. 1986,90,2703. Salmeron, M.; Somorjai, G. A. Desorption, Decomposition, and Deuterium Exchange Reaction of Unsaturated Hydrocarbons (Ethylene, Acetylene, Propylene, and Butenes) on the Pt(ll1) Crystal Face. J. Phys. Chem. 1982,86, 341. Sheridan, J. The Metal Catalyzed Reaction between Acetylene and Hydrogen Part 11, Further Experiments with Nickel Catalysts. J. Chem. SOC.1945, 133. Stephens, S. T. Chemisorption and Surface Reaction of Ethylene on Evaporated Palladium Films. J. Phys. Chem. 1958,62, 714. Tsuchiya, S.; Nakamura, M. Study of Metal Catalysts by Temperature Programmed Desorption I. Chemisorption of Ethylene on Silica-Supported Platinum. J. Catal. 1977, 50, 1. Tysoe, W. T.; Nyberg, G. L.; Lambert, R. M. Photoelectron Spectroscopy and Heterogeneous Catalysis: Benzene and Ethylene from Acetylene on Palladium(ll1). Surf. Sci. 1983, 135, 128. Tysoe, W. T.; Nyberg, G. L.; Lambert, R. M. Selective Hydrogenation of Acetylene over Palladium in Ultra High Vacuum. J. Phys. Chem. 1986,90,3188. Webb, G. Catalytic Hydrogenation. In Comprehensioe Chemical Kinetics; Bamford, Tippen, Eds.; Elsevier: Amsterdam, 1978;Vol. 20, p 1. Weiss, A. H.; Leviness, S.; Nair, V.; Guczi, L.; Sarkany, A.; Schay, Z. The Effect of Pd Dispersion in Acetylene Selective Hydrogenation. Znt. Congr. Catal., 8th 1984, 591. Zagli, A. E.; Falconer, J. L.; Keenan, C. A. Methanation on Supported Nickel Catalysts Using Temperature Programmed Heating. J. Catal. 1979, 56, 453.

Receiued for review February 6, 1991 Accepted April 23, 1991