Degradation of Multiply-Chlorinated Hydrocarbons on Cu (100)

Dechlorination and hydrodechlorination of multiply-chlorinated ethanes and propanes on a clean Cu-. (100) surface have been studied by Auger electron ...
0 downloads 0 Views 301KB Size
+

+

Langmuir 1997, 13, 229-242

229

Degradation of Multiply-Chlorinated Hydrocarbons on Cu(100) Michael X. Yang,* Sutapa Sarkar,† and Brian E. Bent‡ Department of Chemistry, Columbia University, New York, New York 10027

Simon R. Bare§ and Michael T. Holbrook The Dow Chemical Company, Midland, Michigan 48674 Received April 24, 1996. In Final Form: October 17, 1996X Dechlorination and hydrodechlorination of multiply-chlorinated ethanes and propanes on a clean Cu(100) surface have been studied by Auger electron spectroscopy, temperature-programmed desorption, and chemical displacement measurements. The rate-limiting step in degradation is dissociation of the first C-Cl bond in the molecule, and this process is more facile in CCl2 groups than in CCl groups. The activation energies for C-Cl bond scission on Cu(100) are 12-20% of the gas phase bond dissociation energies, and the extent of dissociation by the physisorbed molecules is a sensitive function of the relative rates of C-Cl bond scission and molecular desorption. The chlorinated hydrocarbon fragments generated on the surface by C-Cl bond cleavage undergo facile R- or β-chlorine elimination while all C-H bonds remain intact. β-Chlorine elimination dominates in cases where chlorine is present at both the R- and β-positions, and as a result of β-chlorine elimination, alkenes are generated and evolved to the gas phase. In cases where no β-Cl is present, R-chlorine elimination yields surface carbene intermediates, which readily couple to form longer chain alkenes. Surface hydrogen atoms readily scavenge these carbene intermediates to form alkyl groups. Upon thermal activation, these alkyls are converted to alkenes via β-hydride elimination or to alkanes via coupling with surface hydrogen. All processes subsequent to the initial dissociative adsorption of the chlorinated hydrocarbon occur with 100% selectivity, and all hydrocarbon products are evolved from the surface below 300 K. No carbon is detected on the surface after reaction, but Cl remains adsorbed up to 750 K where it is evolved as CuCl.

1. Introduction Chlorine derivatives are used extensively in products ranging from solvents, dyes, and lubricants to adhesives, additives, and detergents.1 Since these molecules are generally synthesized by chlorinating hydrocarbons in processes that often lead to overchlorination, it is important to design syntheses that minimize undesirable chlorinated byproducts and to develop methods to catalytically degrade or reformulate highly chlorinated materials.2 In these latter areas, the catalytic degradation of chlorinated hydrocarbons by transition metals has received considerable attention; however, systematic catalyst development has been hampered by a lack of detailed mechanistic understanding of the catalytic processes. In the current study, the focus is on identification and kinetic studies of elementary steps in the degradation of multiply chlorinated hydrocarbons on a copper surface. While copper is relatively inert in many catalytic processes, it is surprisingly effective for most of the elementary reactions which are required in catalytic dechlorination. For example, recent studies of carbon-halogen bond dissociation on single crystal surfaces have shown that copper promotes carbon-iodine bond dissociation at * To whom the correspondence should be addressed: Materials Sciences Division, Building 66, Lawrence Berkeley National Laboratory, Berkeley, CA 94720; tel, (510) 486-7463; fax, (510) 4864995; e-mail, [email protected]. † Present address: Department of Chemistry, University of California at Davis, Davis, CA 95616. ‡ Deceased. § Present address: UOP Research, 50 East Algonquin Rd., P.O. Box 5016, Des Plaines, IL 60017-5016. X Abstract published in Advance ACS Abstracts, January 1, 1997. (1) Hileman, B.; Long, J. R.; Kirschner, E. M. Chem. Eng. News 1994, 12. (2) European Patent, publication no: 0496446A1.

temperatures as low as 140 K.3 In fact, in the case of CH3I, the C-I bond dissociates more readily on copper than on Pt,4 a common hydrocarbon catalyst that is generally considered much more “active” than copper in bond scission processes. Copper is also unusually effective in the coupling of hydrocarbon fragments5 and in the liberation of hydrocarbon products (even unsaturated ones).6 Finally, since copper does not form a bulk carbide, one might expect copper-based catalysts to be relatively resistant to coking. For these reasons, we have previously suggested the utility of copper for catalytic dehalogenation.7 Despite the importance of chlorine removal from organic compounds, most prior studies of carbon-halogen bond scission on single crystal metal surfaces under vacuum conditions have focused on bromine- and iodine-containing organics. The reason for this is that chlorinated molecules have proven more difficult to study. The reactivity of alkyl halides with metals decreases in the order of iodides > bromides > chlorides, consistent with the order of gas phase carbon-halogen bond energies: ∼55 kcal/mol for C-I, ∼70 kcal/mol for C-Br, and ∼85 kcal/mol for the C-Cl bond. In most cases, chlorinated hydrocarbons desorb from metal surfaces at a faster rate than they dissociate.6 Thus, while heating a monolayer of bromides and iodides generally leads to dissociation, heating a monolayer of chlorides generally leads to desorption without reaction. To circumvent this low reactivity and to study the reactions of chlorinated molecules, a number (3) Lin, J.-L.; Bent, B. E. J. Phys. Chem. 1993, 97, 9713. (4) Henderson, M. A.; Mitchell, G. E.; White, J. M. Surf. Sci. 1987, 184, L325. (5) Chiang, C.-M.; Wentzlaff, T. H.; Bent, B. E. J. Phys. Chem. 1992, 96, 1836. (6) Lin, J.-L.; Bent, B. E. J. Phys. Chem. 1992, 96, 8529. (7) Lin, J.-L.; Bent, B. E. In Catalytic Control of Air Pollution; ACS Symposium Series 495, American Chemical Society: Washington, DC, 1992.

+

230

+

Langmuir, Vol. 13, No. 2, 1997

of approaches have been demonstrated. These include high-pressure gas exposures to achieve measurable dissociation rates despite the low reaction probability per gas/surface collision,8 photon-9-11 or electron-induced8,10,12 cleavage of C-Cl bonds on the surface at temperatures below where the parent molecule desorbs, the addition of bromine or iodine to the chlorinated molecule to promote reaction and covalent attachment to the surface prior to desorption,13 and increasing the molecular weight of the chlorinated hydrocarbons to slow the rate of desorption so that dissociation becomes the favored pathway in a single monolayer experiment.6 In the current studies, it is these latter two approaches (addition of reactive halogens and increasing the heat of adsorption) that are applied to study the degradation of multiply chlorinated ethanes and propanes on a Cu(100) surface. Specifically, the larger heat of adsorption for multiply chlorinated hydrocarbons compared with monochloro compounds means that in many cases dissociation competes favorably with desorption and can be studied in a single monolayer experiment. In the case of the dichloroethanes, however, the presence of two Cl atoms is insufficient to promote dissociation, and so, in the studies here, one of the chlorines has been replaced with bromine to initiate the decomposition process. The questions addressed in these studies include the following: Is the rate of C-Cl bond dissociation affected by the presence of other chlorine atoms in the molecule? After cleavage of the first C-Cl bond, what are the preferred reactions for chlorine-containing alkyl groups, and is hydrogen or chlorine elimination favored? What are the relative rates of R- and β-elimination for chlorine and hydrogen? In order to address these issues, the studies have been carried out on clean copper single crystal surfaces in ultrahigh vacuum (UHV). The ability to achieve atomically clean and structurally uniform surfaces allows us to isolate and identify the elementary steps in the thermal degradation of these compounds. As presented in the following sections, our results indicate that C-Cl bond cleavage is more facile in -CCl2 groups than in -CCl groups and that, in all cases, scission of the first C-Cl bond is followed by rapid R- or β-chlorine elimination. In the presence of hydrogen atoms on the surface, hydrodechlorination of these compounds is also observed. 2. Experimental Section Details of the UHV system and the experimental procedures for sample cleaning and data acquisition have been described previously.5 The Cu(100) crystal (Monocrystals Inc., 99.999%) can be cooled with liquid nitrogen to 100 K and resistively heated to 1000 K. Surface temperatures were measured by a chromelalumel thermocouple with the junction wedged into a 2 mm deep hole drilled into the side of the crystal. Temperature-programmed desorption/reaction (TPD/R) experiments were conducted using a shielded and differentially pumped mass spectrometer. The sample was held ∼3 mm away from the aperture to the mass spectrometer, in order to collect signal from the center of the surface. Unless specified, an electron impact ionization energy of 70 eV was used in the experiments. The linear heating rate was 3 K/s, and the data points were collected every 100 ms. The peak areas of TPD/R curves were obtained by an integration with respect to time. They are indicative of the total desorption yields. (8) Walter, W. K.; Jones, R. G. Surf. Sci. 1992, 264, 391. (9) Grassian, V. H.; Pimentel, G. C. J. Chem. Phys. 1988, 88, 4484. (10) Zhou, X.-L.; White, J. M. J. Chem. Phys. 1990, 92, 5612. (11) Jo, S. K.; Zhu, X.-Y.; Lennon, D.; White, J. M. Surf. Sci. 1991, 241, 231. (12) Zhou, X.-L.; Blass, P. M.; Koel, B. E.; White, J. M. Surf. Sci. 1992, 271, 452. (13) Kadodwala, M.; Jones, R. G. J. Vac. Sci. Technol., A 1993, 11, 2019.

Yang et al. Auger electron spectroscopy (AES) spectra were acquired in the first derivative mode, and the intensities refer to peak-topeak heights. Assuming that the Auger peaks have normal Gaussian profiles and that the angular distribution of the ejected Auger electrons is independent of the adsorbed molecule and coverage, then these peak-to-peak heights are proportional to surface coverage. To account for variations in beam current, the characteristic Cl Auger peak at 181 eV has been normalized by the 60 and 920 eV transitions for the substrate copper atoms. To minimize the effects of electron stimulated desorption of chlorine from the surface,14 spectra were acquired with an current to ground at the crystal of tertiary with the difference being comparable to the difference in gas phase C-Br bond energies.30 All of these general observations indicate, as observed, that 1,1-dichloro compounds dissociate more readily that 1,2-dichlorides and that 2,2-dichlorides dissociate more readily than 1,1dichlorides. A final important point to make concerning dissociative adsorption of chloroalkanes concerns the markedly different extents of dissociation for the various compounds as indicated by the results summarized in column 1 of Table 1. While direct dissociative adsorption channels are known for some systems,29,43 the majority of thermal dissociative adsorption processes appear to occur through as a two-step process: formation of surface-molecule complex followed by dissociative adsorption. As a result, the extent of dissociative adsorption reflects a competition between desorption and dissociation.6 Small changes in either desorption or dissociation rate can have dramatic effects on the dissociative adsorption rate. Previous examples of this effect for halogenated hydrocarbons are documented in refs 6 and 44. For example, in the case of monochloroalkanes, it is found that there is no detectable dissociation for monolayers of 1-chloroalkanes up to five (40) CRC Handbook of Chemistry and Physics, 71st ed.; CRC Handbook of Chemistry and Physics: West Palm Beach, FL, 1991. (41) Luo, Y. R.; Bensen, S. W. J. Phys. Chem. 1989, 93, 3304. (42) Holmes, J. L.; Lossing, F. P. J. Am. Chem. Soc. 1988, 110, 7343. (43) Ceyer, S. T. Annu. Rev. Phys. Chem. 1988, 39, 479. (44) Bent, B. E.; Nuzzo, R. G.; Zegarski, B. R.; Dubois, L. H. J. Am. Chem. Soc. 1991, 113, 1137.

Yang et al.

carbons long, but a six-carbon chain shows some dissociation and dissociation predominates for a seven-carbon chain. The reason for this effect is that increasing the alkyl chain length increases the heat of adsorption (i.e., decreases the rate of desorption) without significantly affecting the rate of C-Cl dissociation. Dissociation thus becomes competitive with desorption for increasing chain length. A similar effect on the rate of desorption probably accounts for the fact that 1,2-dichloroethane in the studies here shows no detectable dissociation on Cu(100) but ∼25% of a 1,2 dichloropropane monolayer dissociates. Note that 1,2-dichloroethane desorbs from Cu(100) with a TPR peak temperature of 180 K while monolayer desorption of 1,2dichloropropane occurs with peak temperatures of 195200 K. This 15-20 K difference in peak temperature corresponds to about 1 kcal/mol difference in the heat of adsorption of these compounds.35 Assuming no barrier for molecular adsorption, the heat of adsorption corresponds to the activation energy for desorption, and a 1 kcal/mol change in the activation energy for desorption in this temperature range corresponds to about an order of magnitude change in rate. Thus, the added methyl group in the 1,2-dichloropropanes compared with the 1,2dichloroethanes decreases the rate of desorption at any given temperature by about an order of magnitude. In addition, based on the discussion above on primary and secondary C-Cl bond cleavage, one would expect methyl substitution to 1,2-dichloroethane lowers the activation energy for C-Cl dissociation. Both of these effects (decreased rate of molecular desorption and increased rate of C-Cl dissociation) will increase the extent of dissociative adsorption. In applying these results to understand catalyst activities for dechlorination at higher temperatures and pressures, it is important to note that dissociative adsorption via a two-step process (molecular adsorption followed by C-Cl dissociation) generally shows a negative apparent activation energy with respect to the temperature of the surface because of the temperature dependence of the preequilibrium between the gas phase and molecularly adsorbed state. On the other hand, increasing the temperature of the molecules incident on the surface tends to favor dissociation.43 The net result of these competing effects is difficult to predict. In the case of 1,2-dichloroethane dissociation on polycrystalline copper surfaces at room temperature and ∼1 Torr pressure, the apparent activation energy of 21 kcal/mol8,32 is much larger than the activation energy expected for C-Cl bond dissociation on the basis of the results in Table 1. This comparison illustrates that activation energies for elementary steps in monolayer dissociative adsorption reactions should be extrapolated to higher temperatures and pressures with caution and vice versa. 4.2. H and Cl Elimination from Adsorbed Chloroalkyl Groups. Carbon-chlorine bond scission in adsorbed chloroalkanes produces adsorbed chlorine atoms plus alkyl or chloroalkyl groups on the surface. The experimental results regarding the relative rates of H and Cl elimination from alkyl and chloroalkyl groups on Cu(100) are summarized in the last two columns of Table 1. In the case of monochloroalkanes, C-Cl dissociation produces an alkyl group devoid of Cl atoms, and the favored decomposition process in this case is β-hydride elimination. The activation energy for this β-hydride elimination process in the case of ethyl groups on a Cu(100) surface is 14.5 ( 2 kcal/mol, but this value varies from 9 to 21 kcal/mol depending on the structure of the alkyl group.17,45 (45) Forbes, J. G.; Gellman, A. J. J. Am. Chem. Soc. 1993, 115, 6277.

+

+

Chlorinated Hydrocarbons on Cu(100)

Langmuir, Vol. 13, No. 2, 1997 241

On the basis of propylene formation from 1,2-dichloropropane and butene formation from 1,1-dichloroethane, we can conclude that both R- and β-Cl elimination are more facile than β-H elimination. Upper limits of 9 kcal/ mol for the activation energies of these processes are provided by the chlorobromoalkane studies as described in sections 3.3 and 3.4. This small value for the activation energy for R-Cl elimination is particularly interesting since, in the case of H elimination from alkyls on copper surfaces, the activation energy for R-H elimination from methyl was found to be almost twice that of β-H elimination from ethyl.5 The results of the studies here also allow us to compare the elimination rates of Cl at different positions in the alkyl group. On the basis of the 100% selectivity for conversion of 1,1,2-trichloroethane to vinyl chloride, we can conclude (as discussed in section 3.5) that the rate of β-Cl elimination is significantly faster than the rate of R-Cl elimination. 4.3. The Other Elementary Steps in Hydrodechlorination Reactions. H2 Adsorption on and Desorption from Copper Surfaces. While the kinetics of hydrogen adsorption on and desorption from copper surfaces have not been explicitly addressed in the studies here, they have been extensively investigated in the literature. Two conclusions from these studies which are relevant for catalytic dehalogenation are as follows: 1. Dissociative adsorption of hydrogen on copper surfaces shows an Arrhenius activation energy of 12 ( 2 kcal/mol28,29,46,47 with a pre-exponential factor of 0.006 s-1.28 2. Recombinative desorption of H atoms on copper surfaces to produce H2 has an activation energy of 18 ( 4 kcal/mol27 with a prefactor of ∼0.0003 (H atom/cm2)-1 s-1 at low coverages (ref 27, Cu(111) data). These values indicate that the probability for dissociative adsorption of H2 from a room temperature gas sample is ∼2 × 10-11 per collision and that recombinative desorption of hydrogen begins to occur at a significant rate from copper surfaces near room temperature. The net result is that for low flux conditions, the surface hydrogen coverage on copper above room temperature is negligible. Significant surface coverages are possible, however, at higher pressures, and using the fact that the rate of H recombination equals the rate of H2 dissociation at steady state one can determine that the expected steady state, surface coverage of hydrogen on a clean copper surface is given by: 3

θequil,H ) xFluxH2 e(1.5×10 )/T where

FluxH2 ) 2.5 × 1022

P molecules/(cm2 s) xT

where P is the hydrogen pressure (in Torr) and T is the temperature (in Kelvin). Because of uncertainties in the assumed temperature and coverage independence of the kinetic parameters, the surface coverages from these equations are expected to be only an order of magnitude and they will be most accurate in the limit of zero coverage. Within this uncertainty, however, it can be calculated that for 1 atm of H2 at 300 °C the surface coverage of H is ∼5 × 1013 atoms/cm2, which is to be compared with the surface copper atom density of 1 × 1015 atoms/cm2. (46) Vo¨lter, J.; Jungnickel, H.; Riena¨cker, G. Z. Anorg. Chem. 1968, 360, 300. (47) Campbell, J. M.; Campbell, C. T. Surf. Sci. 1992, 259, 1.

Scheme 1. Hydrodechlorination of 2,2-Dichloropropane on a Clean Cu(100) Surface

* obtained from molecular bean and ambient hydrogen adsorption studies.28,29,46,47 # obtained from thermal desorption studies.27

Hydrogen Addition to Hydrocarbon Fragments on Copper. The rate of hydrogenation of surface hydrocarbon fragments on copper depends on the surface coverage of hydrogen, but for a given H coverage, the relative rates of addition are carbene . alkyl, with the latter having a rate (for high surface H coverages) comparable to that for β-H elimination from alkyl groups. These conclusions are based on the results from the studies of 2,2-dichloropropane in the presence and absence of surface hydrogen. As summarized in Scheme 1, in the absence of H, 2,2dichloropropane loses both Cl atoms at 140 K, and the presumed dimethylcarbene intermediates couple promptly to form 2,3-dimethylbutene. By contrast, when this compound is coadsorbed with hydrogen, no 2,3-dimethyl2-butene is produced. Instead, propene and propane are evolved. The suppression of the dimethylbutene yield implies that the dimethylcarbene intermediates are rapidly scavenged by surface hydrogen to form 2-propyl groups on the surface at 140 K. A surface reaction temperature of 140 K corresponds to an activation energy of ∼8 kcal/mol. It is the upper limit for the alkyl group formation from carbene hydrogenation, since the carbene formation is likely the rate-limiting step in the process. Studies presented elsewhere show that β-elimination from 2-propyl groups occurs at 215-225 K to evolve propene (as is observed here). The propane evolved concurrently with propene results from competing hydrogenation of the 2-propyl groups. We therefore conclude that 2-propyl hydrogenates at ∼195 K, corresponding to an activation energy of 11.8 kcal/mol. Desorption of Hydrocarbon Products. Stable hydrocarbon molecules (alkanes, alkenes, and alkynes) interact weakly with copper in comparison with other transition metals, and in the case of alkenes and alkanes, molecular desorption occurs significantly below room temperature. Thus, unlike Pt, Rh, Ir, Ni, Ru, Pd, and other transition metals where ethylene decomposes (eventually) to carbon and hydrogen (refs 48-52 and references therein), ethylene desorbs from copper intact. The implication for the studies here and for catalytic dehalogenation is that hydrocarbon product poisoning of the surface/catalyst is not a significant issue. The exception in this regard is acetylene which bonds strongly enough to copper surfaces (48) Bent, B. E.; Mate, C. M.; Kao, C.-T.; Slavin, A. J.; Somorjai, G. A. J. Phys. Chem. 1988, 92, 4720. (49) Carter, E. A.; Koel, B. E. Surf. Sci. 1990, 226, 339. (50) Hills, M. M.; Parmeter, J. E.; Mullins, C. B.; Weinberg, W. H. J. Am. Chem. Soc. 1986, 108, 3554. (51) Zeara, F.; Hall, R. B. Surf. Sci. 1987, 180, 1. (52) Stuve, E. M.; Madix, R. J. J. Phys. Chem. 1985, 89, 105.

+

242

+

Langmuir, Vol. 13, No. 2, 1997

that some carbon deposition is observed.53,54 In the studies here, no acetylene was formed, and the Cu(100) surface remained carbon-free even after multiple adsorption/ desorption cycles. Chlorine Removal from Copper. The same features that make copper surfaces effective at dechlorination, i.e., strong copper-chlorine bonds, also make copper a potentially poor catalyst because copper binds chlorine too strongly to readily regenerate clean surface sites for catalysis. While the Cl-Cu(100) bond strength is unknown, the CuCl evolution temperature and the CuCl bond energy suggest a value between 53 and 86 kcal/mol (see section 4.1). In the dechlorination studies presented here where the surface Cl coverage is in the monolayer regime, all Cl is thermally evolved as CuCl at temperatures above 750 K. It is interesting to note, however, that for the higher Cl coverages (which can be achieved by Cl2 adsorption), the Cl penetrates the Cu(100) surface to form a CuCl film on Cu(100), and in this case, chlorine is evolved from the surface as Cu3Cl3 at a temperature of ∼470 K.31 This process stops, however, when the Cl coverage is depleted to the monolayer regime at which point heating above 750 K is required to remove the monolayer. Preliminary studies suggest that this situation is not changed dramatically by the addition of H atoms to the surface; for coadsorbed monolayers of H and Cl, H2 is evolved at ∼300 K prior to formation and evolution of measurable amounts of HCl. (53) Outka, D. A.; Friend, C. M.; Jorgensen, S.; Madix, R. J. J. Am. Chem. Soc. 1983, 105, 3468. (54) Avery, N. R. J. Am. Chem. Soc. 1985, 107, 6711.

Yang et al.

5. Summary Dechlorination and hydrodechlorination of (1,1- and 1,2-) dichloroethanes and (2,2- and 1,2-) dichloropropanes have been achieved in ultrahigh vacuum (UHV) on a clean Cu(100) surface, and the surface reaction mechanisms have been determined. The rate-limiting step in each case is dissociation of the first C-Cl bond. Increasing the degree of chlorination at any given carbon atom decreases the activation energy for dissociation. Dissociation can also be promoted by slowing down the competing desorption channel via increased carbon chain length. Both of the 1,2-dichloroalkanes undergo 1,2-chlorine elimination to yield the corresponding alkene. By comparison, 1,1- (or 2,2-) chlorinated alkanes undergo 1,1- and 2,2chlorine elimination to generate carbene intermediates, which undergo facile dimerization reactions to form longer alkenes. In the presence of surface hydrogen atoms, the surface carbene intermediates are converted to surface alkyl groups, which subsequently undergo β-hydride elimination and hydrogenation reactions to form alkenes and alkanes of the original chain length. If both 1,1- and 1,2-chlorine elimination are possible, 1,2-chlorine elimination dominates over 1,1-chlorine elimination. As a result, vinyl chloride is the sole hydrocarbon product from the reaction of 1,1,2-trichloroethane on Cu(100). Acknowledgment. We are grateful to the Dow Chemical Company and National Science Foundation (Grant CHE-93-18625) for financial support of this work. LA960404Y