- Email: [email protected]
Dispersed Pd-Ag alloys for selective production of olefins from chlorinated alkanes
Studies in Surface Science and Catalysis 130 A. Corma, F.V. Melo, S. Mendioroz and J.L.G. Fierro (Editors) 9 2000 Elsevier Science B.V. All fights reserved.
2015
Dispersed Pd-Ag alloys for selective production of olefins from chlorinated alkanes B. Heinrichs a'*, J.-P. Schoebrechts b, and J.-P. Pirard a aLaboratoire de G6nie Chimique, B6a, Universit6 de Li6ge, B-4000 Li6ge, Belgium bLaboratoire Central, Solvay, S.A., Rue de Ransbeek, 310, B-1120 Brussels, Belgium A kinetic model is derived to describe experimental rates of selective hydrodechlorination of 1,2-dichlorethane into ethylene over a Pd-Ag/SiO2 catalyst. The role of each metal in Pd-Ag alloy particles is examined. A mechanism is proposed in which ethylene is produced by dissociative adsorption of 1,2-dichloroethane on silver which is subsequently dechlorinated thanks to recombination of adsorbed chlorine atoms with hydrogen dissociatively adsorbed on palladium. I. INTRODUCTION Hydrodechlorination (HDC) is an economically and environmentally attractive process for the recycling of chlorinated wastes [1]. HDC usually consists in the hydrogenolysis of the carbon-chlorine bond: --C-CI + H2 --> -C-H + HCI. In the case of chlorinated alkanes, this hydrogenolysis produces alkanes. However, Ito et al. [2] demonstrated the ability of bimetallic catalysts to selectively convert chlorinated alkanes into alkenes. In a previous work, we prepared highl3~ dispersed Pd-Ag/SiO2 sol-gel catalysts specifically designed for selective HDC of chlorinated alkanes into alkenes [3]. In order to understand the selectivity effect of the alloys, we try, in the present study, to identify the mechanism of the selective HDC of 1,2-dichloroethane over a Pd-Ag surface, the nature of the active sites involved in the process and the roles played by the two metals. This has been done through a kinetic study over a 1.9%Pd-3.7%Ag/SiO2 sol-gel catalyst. 2. EXPERIMENTAL The preparation and characterization of the catalyst, called X67, used in this study are described in detail in [3, 4]. The two reactions involved in the reaction scheme are the HDC of 1,2-dichloroethane into ethylene and the undesired hydrogenation of ethylene into ethane: CICH2-CH2C1 + H2 --~ CH2=CH2 + 2 HC1 HDC, rate r~ CH2=CH2 + H2 ---> CH3-CH3 hydrogenation, rate r2 The rates of those two reactions (rl and r2) are measured in an isothermal differential tubular reactor with an internal diameter of 0.8 cm. The reactor is fed with a gaseous mixture To whom correspondence shouldbe addressed. Phone: +32 4 366 35 05; Fax: +32 4 366 35 45; e-mail: [email protected]
2016 containing all components having an influence on r~ and/or r2, i.e. CICHe-CH2CI, HE, C2H4,and HCI diluted in He (C2I-k has no influence). The total pressure is fixed at 0.3 MPa and the total flowrate at 0.5 mmol s1. The temperature is successively fixed at 573, 596 and 647 K. The 6.2 cm high catalytic bed is composed of 0.25 g of catalyst pellets crushed and sieved between 250 and 500/~m. Prior to start the kinetic measurements, the catalyst is reduced in situ during 3 h under 0.025 mmol s~ flowing hydrogen at 623 K and atmospheric pressure. Rates rl and r2 are calculated from C2I-I4 and C2I-I6 concentrations in the effluent obtained from gas chromatography measurements. Kinetic measurements are performed according to a priori defined experimental designs. At each temperature (573, 596 and 647 K) r~ and re are measured in a four dimensions space of the partial pressures of CICH2-CH2CI (PD), H2 (PH), C2H4 (PE), and HCI (Pncl) in 40 peripheral points and the central point of a rotatable central composite design [5]. The experimental design is covered twice for each temperature in order to have the replicates which are necessary for the statistical tests of the various kinetic models. During the kinetic data acquisition, a deactivation of the catalyst is observed and the raw values of reaction rates have been corrected in order to all correspond to the same activity level. In this way, 480 data (3 temperatures x 2 rates x 40 points x 2 replicates) are available for parameter estimation. The absence of diffusional limitations was checked both experimentally (rate measurements with various pellet mean sizes) and theoretically (Weisz modulus calculation) [4, 6, 7]. The fitting of kinetic models was performed by minimizing the experimental chi-square function [5]. More details about kinetic data acquisition, data themselves, and fitting are available in [4]. 3. RESULTS 3.1. Empirical model: power rate laws Numerous kinetic equations derived from various sequences of elementary steps and from chosen rate-determining steps (rds) can be assumed so as to try to describe correctly the experimental rates. In order to restrict the number of plausible candidates among all those phenomenological models, the apparent reaction orders in CICH2-CH2CI, H2, C2H4, and HCI are first examined with an empirical model composed of two power rate law equations: W 1
rl
=
k l PD
X1
PH PY~ PHO z, ,. rE
W 2
=
X2
k2 PD PH P y: P H":C l
(1)
The 10 parameters of this model are estimated at each temperature. The fitted values as well as their 95% confidence interval [5] are given in Table 1. According to a chi-square test, the empirical model describes correctly the experimental data at all temperatures. No lack of fit is detected. The examination of the obtained values and their confidence intervals shows that rl is (i) nearly first order in CICH2-CH2CI at all temperatures (wl); (ii) positive and decreasing with temperature order in H2 (x~); (iii) negative and decreasing with temperature order in C2I-h (yl); (iv) negative and increasing with temperature order in HCI (Za); and r2 is (i) nearly independent of CICH2-CH2CI partial pressure (order ~ 0) (w2); (ii) between 1.5 and 2 order in H2 (x2); (iii) first or slightly lower order in C2H4 (y2); (iv) close to -1 order in HC1 (z2).
2017
Table 1 Empirical model Parameters
596 K
647 K
wl Xl Yl Zl
0.79 0.91 0.46 -0.11 -0.92
+ + • • +
0.76 0.18 0.16 0.11 0.22
3.3 0.96 0.25 -0.07 -0.67
+ + • • +
1.5 0.09 0.07 0.05 0.10
3.1 0.70 0.25 -0.47 -0.31
+ • • • •
2.8 0.17 0.15 0.11 0.19
k2 w2 x2 Y2 z2
9.6 -0.04 1.59 1.01 -0.91
+ • • + •
3.6 0.05 0.08 0.06 0.07
8.8 0.16 1.83 0.83 -1.09
• • • • •
8.1 0.12 0.21 0.12 0.19
12.1 -0.17 1.76 0.86 -0.89
• • • • •
7.6 0.08 0.15 0.09 0.12
kI
Note.
573 K
Umts of reaction rates are mmol kg~ s~ and umts of partial pressures are atm.
3.2. Phenomenological models In order to try to identify one or several models which are able to describe correctly the kinetic measurements, equations based on various sequences of elementary steps were successively fitted to the experimental data. A chi-square test as well as the examination of the compatibility of those equations with the apparent orders derived from the empirical model allowed to evaluate the quality of each considered phenomenological model. In this way, 20 models were examined. Those models are based on various choices concerning the ratedetermining steps (rds) and various hypothesis such as the presence of one or two types of active sites, molecular or dissociative adsorptions, surface reactions between adsorbed species only (Langmuir-Hinshelwood kinetics) or between adsorbed species and molecules in the gas phase (Eley-Rideal kinetics), and the type of sites on which each species is adsorbed when two types are considered. Among the 20 models considered, only one describes correctly the whole set of kinetic data according to the chi-square test. This model is based on the sequence (2) of elementary steps which involves two types of actives sites, sl and s2: HDC: (1) CICH2CH2CI + 2Sl ---> (2) CICH2CH2sl + Sl ~-> (3) C2I-I4Sl 0 (4) H2 + 2S2 <--> (5) Hs2 + Clsl <-> Hydrogenation: (3) C2H4+ Sl ~ (4) H2 + 2s2 r (6) C2I-I4sl+ Hs2 ~-~ (7) C2Hss~ + Hs2 -~ Exchange of adsorbed chlorine (8) Cls~ + s2 ~-~
CICH2CH2sl + ClSl C2I-I4Sl+ Clsl
rdsl
C284 + Sl
2Hs2 HCI + s2 + Sl (2) C2H4Sl
2Hs2 C2HsSl + S2 C2H6+ s~ + s2 atoms between s! and s2 CIs2 + Sl
rds2
2018 The rds for HDC is step (1) and the rds for hydrogenation is step (7). Steps (2) to (6) and step (8) are assumed to be in quasi-equilibrium. Step (8) which corresponds to a spillover of adsorbed chlorine atoms from s~ to s2 is not directly involved in HI)C or hydrogenation reactions. This step is however necessary in order to describe the apparent orders in HCI, Zl and z2 (Table 1), which evolve differently with temperature for HI)C, the rds of which involves two sites Sl, and for hydrogenation, the rds of which involves one site Sl and one site s2. The corresponding HDC and hydrogenation rate equations (r~ and r2 respectively) are: k~ PD
rl = 1 + A
PEPRc!
P~K
2
PHCl + B + Cp + Ep P~fH] x/PH E E . --) (3) k2
r2
P~.PH
1 + A PEPHcl + B PHC1 4P.
A preliminary estimation of the parameters showed that C, D and E are not significantly different from zero which suggests that CICHzCH2sl (term with constant A) and Clsl (term with constant B) are the only species present in significant amount on sites sl and that Cls2 (term with constant F) is the only species present in significant amount on sites s2. The developed expressions of the significant parameters are given in Table 2. kl is equal to the intrinsic rate constant k'l of rds 1 whereas k2 represents a combination of the intrinsic rate constant k'2 of rds 2 with equilibrium constants Ki of steps (i) in quasi-equilibrium. Table 2 Phenomenological model Parameters k~ =k'~ k2 = k, 2 K 3 K 4
K6 A= B-
Values
1%1 E1
1.17 106 49.1
ko2
1.82 10 9 74.2
E2
K3K 5 K:~4
A0 AH ~
5.49 107
Ks ~4
B0 AH ~
1.54 10-6 -73.0
F0
3.42 107
F - KsK8 X~4
AH ~
68.6
68.8
Note. Units of reaction rates are mmol kg 1 s1 and units of partial pressures are atm.
A, B and F are combination of equilibrium constant Ki. Note that equations (3) are in agreement with the apparent reaction orders given in Table 1. Assuming that rate constants follow Arrhenius' law, and that equilibrium constants follow van 't Hoff's law, the temperature dependency of kl, k2, A, B and F is given by equation (4) where kol, ko2, Ao, B0 and F0 are preexponential factor, E1 and E2 are activation energies (J 0 0 0 mmoll), AH A, AH B et AH F (J mmolq) are combination of enthalpy changes of elementary steps in quasi-equilibrium and R = 8.3143 10-3 J mmolq K q. Fitted values of the parameters of model (3) are given in Table 2.
2019
9
4. DISCUSSION
Results of the kinetic study presented in section 3 suggest the presence of two types of active sites on the surface of the catalyst. The nature of those sites Sl and s2 is examined below. The rupture of the carbon-chlorine bond has been examined on various metals and published results show that Ag is much more active than Pd for this dissociation. For example, in a study on the dehalogenation of haloalkanes on SiO2 supported metals, Anju et al. [8] measured an activity for the dechlorination (without hydrogen) of 1,2-dichloroethane over pure silver which is 13 times higher than over pure palladium. In the sequence (2) of elementary steps, steps (1) and (2) describing dissociative adsorption of 1,2-dichloroethane with successive ruptures of the two C-CI bonds occur on sites sl. Literature results then suggest that sites Sl are Ag sites. The identification of Sl with Ag sites raises the question of the possibility for ethylene to adsorb on silver (steps (2), (3) and (6) in sequence (2)). Literature about ethylene epoxidation over silver is helpful to answer this question. Wachs and Kelemen [9] showed that above 200 K, C2H4 does not adsorb on clean silver and this absence of imeraction between C2H4 and Ag is mentioned by others authors. However the presence of adsorbed electronegative atoms such as chlorine atoms induces positively charged surface silver atoms, Ag ~*, which then become able to adsorb C2H4 [10]. The adsorption of C2I-I4 on sites Sl idemified with Ag sites in the sequence (2) is then physically acceptable. Step (4) of sequence (2) describes dissociative adsorption of H2 on sites s2. Dissociation of H2 is a well known phenomenon over Pd [11 ] and it is then plausible to assimilate s2 to Pd sites. Concerning ethylene hydrogenation, steps (3), (4), (6) and (7) in sequence (2) describe a mechanism in which hydrogen atoms adsorbed on s2 - Pd are added in two successive steps to ethylene adsorbed on s~ - Ag ~§ It is interesting to note that such a mechanism involving an adsorption of C2H4 on a positively charged metal and a non-competitive adsorption of H2 is analogous to the mechanism proposed by Derouane et al. [12] in a study of ethylene hydrogenation over a Cu/MgO catalyst. Finally some values of activation energies and enthalpy changes in Table 2 deserve a brief discussion. First, it is remarkable to note that the value of 49 J mmol "1 obtained for E~ which corresponds to the intrinsic activation energy of the dissociative adsorption of CICH2-CH2CI on Sl - Ag ~§ (step (1) in sequence (2)) is in agreement with the value of 53 J mmol ~ obtained by Anju et al. [8] for the same reaction on the same metal by a completely different technique. Second, it can be easily shown that the enthalpy change AH~ associated with the spillover of C1 from Sl - Ag ~§ to s2 - Pd (step (8) in sequence (2)) is equal to AH ~ - AH~ that is 142 J mmol 1. This result indicates that this spillover is an endothermic process which means that C1 adsorbed on silver is more stable than C1 adsorbed on palladium.
2020 5. CONCLUSIONS The kinetic study and the identification of active sites presented in this work suggest a mechanism in which the actual dechlorination of 1,2-dichloroethane into ethylene would occur on silver atoms present at the surface of Pd-Ag alloy particles dispersed in porous silica. Used alone, silver would deactivate quickly because of its recovery by chlorine atoms. Thanks to its hydrogen activation power by dissociative adsorption, palladium present at the surface of the alloy would supply hydrogen atoms for the regeneration of chlorinated silver into metallic silver. Note that the palladium surface would be partially covered with chlorine atoms due to their spillover from silver to palladium. The presence of hydrogen adsorbed on palladium causes the undesired hydrogenation of ethylene as well which leads to a loss of selectivity in the olefin. It is interesting to note that the mechanism of chlorination/hydrodechlorination of silver proposed here is similar to the mechanism of oxidation/reduction of the catalytic surface proposed by Mars and van Krevelen in their study of the oxidation of hydrocarbons [13]. However, an opposite phenomenon occurs in that case since the hydrocarbon is oxidized by reduction of the catalytic surface which is reoxidized by gaseous oxygen. REFERENCES
1. T.N. Kalnes and R.B. James, Environ. Prog., 7 (1988) 185. 2. L.N. Ito, A.D. Harley, M.T. Holbrook, D.D. Smith, C.B. Murchison and M.D. Cisneros, Processes for Converting Chlorinated Alkane Byproducts or Waste Products to Useful, Less Chlorinated Alkenes, International Patent Application WO 94/07827 (1994). 3. B. Heinrichs, P. Delhez, J.-P. Schoebrechts and J.-P. Pirard, J. Catal., 172 (1997) 322. 4. B. Heinrichs, L'hydrod6chloration s61ective du 1,2-dichloro6thane en 6thyl6ne sur des catalyseurs Pd-Ag/SiO2, Ph.D. thesis, Universit6 de Li6ge, Liege, 1999. 5. D.M. Himmelblau, Process Analysis by Statistical Methods, Wiley, New York, 1970. 6. C.N. Satterfield, Mass Transfer in Heterogeneous Catalysis, M.I.T. Press, Cambridge, 1970. 7. J. Villermaux, Genie de la Reaction Chimique - Conception et Fonctionnement des Reacteurs, Lavoisier, Paris, 1993. 8. Y. Anju, I. Mochida, H. Yamamoto, A. Kato and T. Seiyama, Bull. Chem. Soc. Jap., 45 (1972) 2319. 9. I.E. Wachs and S.R. Kelemen, in "Proceedings of the 7th International Congress on Catalysis, Tokyo, 1980", paper A48. 10. R. A. van Santen and H.P.C.E. Kuipers, Adv. Catal., 35 (1988) 265. 11. G. Bergeret and P. Gallezot, in "Handbook of Heterogeneous Catalysis", G. Ertl, H. Knozinger and J. Weitkamp (eds.), Vol. 2, p. 439, Wiley-VCH, Weinheim, 1997. 12. E.G. Derouane, J.-P. Pirard, G.A. L'Homme and E. Fabry-Volders, in "Catalysis: Heterogeneous and Homogeneous. Proceedings of the International Symposium on the Relations between Heterogeneous and Homogeneous Catalytic Phenomena, Brussels, Belgium, October 23-25, 1974", Elsevier, Amsterdam, 1975. 13. P. Mars and D.W. van Krevelen, Chem. Eng. Sci. Special Suppl., 3 (1954) 41.
2015
Dispersed Pd-Ag alloys for selective production of olefins from chlorinated alkanes B. Heinrichs a'*, J.-P. Schoebrechts b, and J.-P. Pirard a aLaboratoire de G6nie Chimique, B6a, Universit6 de Li6ge, B-4000 Li6ge, Belgium bLaboratoire Central, Solvay, S.A., Rue de Ransbeek, 310, B-1120 Brussels, Belgium A kinetic model is derived to describe experimental rates of selective hydrodechlorination of 1,2-dichlorethane into ethylene over a Pd-Ag/SiO2 catalyst. The role of each metal in Pd-Ag alloy particles is examined. A mechanism is proposed in which ethylene is produced by dissociative adsorption of 1,2-dichloroethane on silver which is subsequently dechlorinated thanks to recombination of adsorbed chlorine atoms with hydrogen dissociatively adsorbed on palladium. I. INTRODUCTION Hydrodechlorination (HDC) is an economically and environmentally attractive process for the recycling of chlorinated wastes [1]. HDC usually consists in the hydrogenolysis of the carbon-chlorine bond: --C-CI + H2 --> -C-H + HCI. In the case of chlorinated alkanes, this hydrogenolysis produces alkanes. However, Ito et al. [2] demonstrated the ability of bimetallic catalysts to selectively convert chlorinated alkanes into alkenes. In a previous work, we prepared highl3~ dispersed Pd-Ag/SiO2 sol-gel catalysts specifically designed for selective HDC of chlorinated alkanes into alkenes [3]. In order to understand the selectivity effect of the alloys, we try, in the present study, to identify the mechanism of the selective HDC of 1,2-dichloroethane over a Pd-Ag surface, the nature of the active sites involved in the process and the roles played by the two metals. This has been done through a kinetic study over a 1.9%Pd-3.7%Ag/SiO2 sol-gel catalyst. 2. EXPERIMENTAL The preparation and characterization of the catalyst, called X67, used in this study are described in detail in [3, 4]. The two reactions involved in the reaction scheme are the HDC of 1,2-dichloroethane into ethylene and the undesired hydrogenation of ethylene into ethane: CICH2-CH2C1 + H2 --~ CH2=CH2 + 2 HC1 HDC, rate r~ CH2=CH2 + H2 ---> CH3-CH3 hydrogenation, rate r2 The rates of those two reactions (rl and r2) are measured in an isothermal differential tubular reactor with an internal diameter of 0.8 cm. The reactor is fed with a gaseous mixture To whom correspondence shouldbe addressed. Phone: +32 4 366 35 05; Fax: +32 4 366 35 45; e-mail: [email protected]
2016 containing all components having an influence on r~ and/or r2, i.e. CICHe-CH2CI, HE, C2H4,and HCI diluted in He (C2I-k has no influence). The total pressure is fixed at 0.3 MPa and the total flowrate at 0.5 mmol s1. The temperature is successively fixed at 573, 596 and 647 K. The 6.2 cm high catalytic bed is composed of 0.25 g of catalyst pellets crushed and sieved between 250 and 500/~m. Prior to start the kinetic measurements, the catalyst is reduced in situ during 3 h under 0.025 mmol s~ flowing hydrogen at 623 K and atmospheric pressure. Rates rl and r2 are calculated from C2I-I4 and C2I-I6 concentrations in the effluent obtained from gas chromatography measurements. Kinetic measurements are performed according to a priori defined experimental designs. At each temperature (573, 596 and 647 K) r~ and re are measured in a four dimensions space of the partial pressures of CICH2-CH2CI (PD), H2 (PH), C2H4 (PE), and HCI (Pncl) in 40 peripheral points and the central point of a rotatable central composite design [5]. The experimental design is covered twice for each temperature in order to have the replicates which are necessary for the statistical tests of the various kinetic models. During the kinetic data acquisition, a deactivation of the catalyst is observed and the raw values of reaction rates have been corrected in order to all correspond to the same activity level. In this way, 480 data (3 temperatures x 2 rates x 40 points x 2 replicates) are available for parameter estimation. The absence of diffusional limitations was checked both experimentally (rate measurements with various pellet mean sizes) and theoretically (Weisz modulus calculation) [4, 6, 7]. The fitting of kinetic models was performed by minimizing the experimental chi-square function [5]. More details about kinetic data acquisition, data themselves, and fitting are available in [4]. 3. RESULTS 3.1. Empirical model: power rate laws Numerous kinetic equations derived from various sequences of elementary steps and from chosen rate-determining steps (rds) can be assumed so as to try to describe correctly the experimental rates. In order to restrict the number of plausible candidates among all those phenomenological models, the apparent reaction orders in CICH2-CH2CI, H2, C2H4, and HCI are first examined with an empirical model composed of two power rate law equations: W 1
rl
=
k l PD
X1
PH PY~ PHO z, ,. rE
W 2
=
X2
k2 PD PH P y: P H":C l
(1)
The 10 parameters of this model are estimated at each temperature. The fitted values as well as their 95% confidence interval [5] are given in Table 1. According to a chi-square test, the empirical model describes correctly the experimental data at all temperatures. No lack of fit is detected. The examination of the obtained values and their confidence intervals shows that rl is (i) nearly first order in CICH2-CH2CI at all temperatures (wl); (ii) positive and decreasing with temperature order in H2 (x~); (iii) negative and decreasing with temperature order in C2I-h (yl); (iv) negative and increasing with temperature order in HCI (Za); and r2 is (i) nearly independent of CICH2-CH2CI partial pressure (order ~ 0) (w2); (ii) between 1.5 and 2 order in H2 (x2); (iii) first or slightly lower order in C2H4 (y2); (iv) close to -1 order in HC1 (z2).
2017
Table 1 Empirical model Parameters
596 K
647 K
wl Xl Yl Zl
0.79 0.91 0.46 -0.11 -0.92
+ + • • +
0.76 0.18 0.16 0.11 0.22
3.3 0.96 0.25 -0.07 -0.67
+ + • • +
1.5 0.09 0.07 0.05 0.10
3.1 0.70 0.25 -0.47 -0.31
+ • • • •
2.8 0.17 0.15 0.11 0.19
k2 w2 x2 Y2 z2
9.6 -0.04 1.59 1.01 -0.91
+ • • + •
3.6 0.05 0.08 0.06 0.07
8.8 0.16 1.83 0.83 -1.09
• • • • •
8.1 0.12 0.21 0.12 0.19
12.1 -0.17 1.76 0.86 -0.89
• • • • •
7.6 0.08 0.15 0.09 0.12
kI
Note.
573 K
Umts of reaction rates are mmol kg~ s~ and umts of partial pressures are atm.
3.2. Phenomenological models In order to try to identify one or several models which are able to describe correctly the kinetic measurements, equations based on various sequences of elementary steps were successively fitted to the experimental data. A chi-square test as well as the examination of the compatibility of those equations with the apparent orders derived from the empirical model allowed to evaluate the quality of each considered phenomenological model. In this way, 20 models were examined. Those models are based on various choices concerning the ratedetermining steps (rds) and various hypothesis such as the presence of one or two types of active sites, molecular or dissociative adsorptions, surface reactions between adsorbed species only (Langmuir-Hinshelwood kinetics) or between adsorbed species and molecules in the gas phase (Eley-Rideal kinetics), and the type of sites on which each species is adsorbed when two types are considered. Among the 20 models considered, only one describes correctly the whole set of kinetic data according to the chi-square test. This model is based on the sequence (2) of elementary steps which involves two types of actives sites, sl and s2: HDC: (1) CICH2CH2CI + 2Sl ---> (2) CICH2CH2sl + Sl ~-> (3) C2I-I4Sl 0 (4) H2 + 2S2 <--> (5) Hs2 + Clsl <-> Hydrogenation: (3) C2H4+ Sl ~ (4) H2 + 2s2 r (6) C2I-I4sl+ Hs2 ~-~ (7) C2Hss~ + Hs2 -~ Exchange of adsorbed chlorine (8) Cls~ + s2 ~-~
CICH2CH2sl + ClSl C2I-I4Sl+ Clsl
rdsl
C284 + Sl
2Hs2 HCI + s2 + Sl (2) C2H4Sl
2Hs2 C2HsSl + S2 C2H6+ s~ + s2 atoms between s! and s2 CIs2 + Sl
rds2
2018 The rds for HDC is step (1) and the rds for hydrogenation is step (7). Steps (2) to (6) and step (8) are assumed to be in quasi-equilibrium. Step (8) which corresponds to a spillover of adsorbed chlorine atoms from s~ to s2 is not directly involved in HI)C or hydrogenation reactions. This step is however necessary in order to describe the apparent orders in HCI, Zl and z2 (Table 1), which evolve differently with temperature for HI)C, the rds of which involves two sites Sl, and for hydrogenation, the rds of which involves one site Sl and one site s2. The corresponding HDC and hydrogenation rate equations (r~ and r2 respectively) are: k~ PD
rl = 1 + A
PEPRc!
P~K
2
PHCl + B + Cp + Ep P~fH] x/PH E E . --) (3) k2
r2
P~.PH
1 + A PEPHcl + B PHC1 4P.
A preliminary estimation of the parameters showed that C, D and E are not significantly different from zero which suggests that CICHzCH2sl (term with constant A) and Clsl (term with constant B) are the only species present in significant amount on sites sl and that Cls2 (term with constant F) is the only species present in significant amount on sites s2. The developed expressions of the significant parameters are given in Table 2. kl is equal to the intrinsic rate constant k'l of rds 1 whereas k2 represents a combination of the intrinsic rate constant k'2 of rds 2 with equilibrium constants Ki of steps (i) in quasi-equilibrium. Table 2 Phenomenological model Parameters k~ =k'~ k2 = k, 2 K 3 K 4
K6 A= B-
Values
1%1 E1
1.17 106 49.1
ko2
1.82 10 9 74.2
E2
K3K 5 K:~4
A0 AH ~
5.49 107
Ks ~4
B0 AH ~
1.54 10-6 -73.0
F0
3.42 107
F - KsK8 X~4
AH ~
68.6
68.8
Note. Units of reaction rates are mmol kg 1 s1 and units of partial pressures are atm.
A, B and F are combination of equilibrium constant Ki. Note that equations (3) are in agreement with the apparent reaction orders given in Table 1. Assuming that rate constants follow Arrhenius' law, and that equilibrium constants follow van 't Hoff's law, the temperature dependency of kl, k2, A, B and F is given by equation (4) where kol, ko2, Ao, B0 and F0 are preexponential factor, E1 and E2 are activation energies (J 0 0 0 mmoll), AH A, AH B et AH F (J mmolq) are combination of enthalpy changes of elementary steps in quasi-equilibrium and R = 8.3143 10-3 J mmolq K q. Fitted values of the parameters of model (3) are given in Table 2.
2019
9
4. DISCUSSION
Results of the kinetic study presented in section 3 suggest the presence of two types of active sites on the surface of the catalyst. The nature of those sites Sl and s2 is examined below. The rupture of the carbon-chlorine bond has been examined on various metals and published results show that Ag is much more active than Pd for this dissociation. For example, in a study on the dehalogenation of haloalkanes on SiO2 supported metals, Anju et al. [8] measured an activity for the dechlorination (without hydrogen) of 1,2-dichloroethane over pure silver which is 13 times higher than over pure palladium. In the sequence (2) of elementary steps, steps (1) and (2) describing dissociative adsorption of 1,2-dichloroethane with successive ruptures of the two C-CI bonds occur on sites sl. Literature results then suggest that sites Sl are Ag sites. The identification of Sl with Ag sites raises the question of the possibility for ethylene to adsorb on silver (steps (2), (3) and (6) in sequence (2)). Literature about ethylene epoxidation over silver is helpful to answer this question. Wachs and Kelemen [9] showed that above 200 K, C2H4 does not adsorb on clean silver and this absence of imeraction between C2H4 and Ag is mentioned by others authors. However the presence of adsorbed electronegative atoms such as chlorine atoms induces positively charged surface silver atoms, Ag ~*, which then become able to adsorb C2H4 [10]. The adsorption of C2I-I4 on sites Sl idemified with Ag sites in the sequence (2) is then physically acceptable. Step (4) of sequence (2) describes dissociative adsorption of H2 on sites s2. Dissociation of H2 is a well known phenomenon over Pd [11 ] and it is then plausible to assimilate s2 to Pd sites. Concerning ethylene hydrogenation, steps (3), (4), (6) and (7) in sequence (2) describe a mechanism in which hydrogen atoms adsorbed on s2 - Pd are added in two successive steps to ethylene adsorbed on s~ - Ag ~§ It is interesting to note that such a mechanism involving an adsorption of C2H4 on a positively charged metal and a non-competitive adsorption of H2 is analogous to the mechanism proposed by Derouane et al. [12] in a study of ethylene hydrogenation over a Cu/MgO catalyst. Finally some values of activation energies and enthalpy changes in Table 2 deserve a brief discussion. First, it is remarkable to note that the value of 49 J mmol "1 obtained for E~ which corresponds to the intrinsic activation energy of the dissociative adsorption of CICH2-CH2CI on Sl - Ag ~§ (step (1) in sequence (2)) is in agreement with the value of 53 J mmol ~ obtained by Anju et al. [8] for the same reaction on the same metal by a completely different technique. Second, it can be easily shown that the enthalpy change AH~ associated with the spillover of C1 from Sl - Ag ~§ to s2 - Pd (step (8) in sequence (2)) is equal to AH ~ - AH~ that is 142 J mmol 1. This result indicates that this spillover is an endothermic process which means that C1 adsorbed on silver is more stable than C1 adsorbed on palladium.
2020 5. CONCLUSIONS The kinetic study and the identification of active sites presented in this work suggest a mechanism in which the actual dechlorination of 1,2-dichloroethane into ethylene would occur on silver atoms present at the surface of Pd-Ag alloy particles dispersed in porous silica. Used alone, silver would deactivate quickly because of its recovery by chlorine atoms. Thanks to its hydrogen activation power by dissociative adsorption, palladium present at the surface of the alloy would supply hydrogen atoms for the regeneration of chlorinated silver into metallic silver. Note that the palladium surface would be partially covered with chlorine atoms due to their spillover from silver to palladium. The presence of hydrogen adsorbed on palladium causes the undesired hydrogenation of ethylene as well which leads to a loss of selectivity in the olefin. It is interesting to note that the mechanism of chlorination/hydrodechlorination of silver proposed here is similar to the mechanism of oxidation/reduction of the catalytic surface proposed by Mars and van Krevelen in their study of the oxidation of hydrocarbons [13]. However, an opposite phenomenon occurs in that case since the hydrocarbon is oxidized by reduction of the catalytic surface which is reoxidized by gaseous oxygen. REFERENCES
1. T.N. Kalnes and R.B. James, Environ. Prog., 7 (1988) 185. 2. L.N. Ito, A.D. Harley, M.T. Holbrook, D.D. Smith, C.B. Murchison and M.D. Cisneros, Processes for Converting Chlorinated Alkane Byproducts or Waste Products to Useful, Less Chlorinated Alkenes, International Patent Application WO 94/07827 (1994). 3. B. Heinrichs, P. Delhez, J.-P. Schoebrechts and J.-P. Pirard, J. Catal., 172 (1997) 322. 4. B. Heinrichs, L'hydrod6chloration s61ective du 1,2-dichloro6thane en 6thyl6ne sur des catalyseurs Pd-Ag/SiO2, Ph.D. thesis, Universit6 de Li6ge, Liege, 1999. 5. D.M. Himmelblau, Process Analysis by Statistical Methods, Wiley, New York, 1970. 6. C.N. Satterfield, Mass Transfer in Heterogeneous Catalysis, M.I.T. Press, Cambridge, 1970. 7. J. Villermaux, Genie de la Reaction Chimique - Conception et Fonctionnement des Reacteurs, Lavoisier, Paris, 1993. 8. Y. Anju, I. Mochida, H. Yamamoto, A. Kato and T. Seiyama, Bull. Chem. Soc. Jap., 45 (1972) 2319. 9. I.E. Wachs and S.R. Kelemen, in "Proceedings of the 7th International Congress on Catalysis, Tokyo, 1980", paper A48. 10. R. A. van Santen and H.P.C.E. Kuipers, Adv. Catal., 35 (1988) 265. 11. G. Bergeret and P. Gallezot, in "Handbook of Heterogeneous Catalysis", G. Ertl, H. Knozinger and J. Weitkamp (eds.), Vol. 2, p. 439, Wiley-VCH, Weinheim, 1997. 12. E.G. Derouane, J.-P. Pirard, G.A. L'Homme and E. Fabry-Volders, in "Catalysis: Heterogeneous and Homogeneous. Proceedings of the International Symposium on the Relations between Heterogeneous and Homogeneous Catalytic Phenomena, Brussels, Belgium, October 23-25, 1974", Elsevier, Amsterdam, 1975. 13. P. Mars and D.W. van Krevelen, Chem. Eng. Sci. Special Suppl., 3 (1954) 41.