J. Chem. SOC., Faraday Trans. 1, 1985, 81, 83-89 Adsorption and Conductivity Studies in Oxychlorination Catalysis Part 5.-Temperature-programmed Desorption BY PETER G. HALL,* PHILIP HEATON? AND DAVID R. ROSSEINSKY Department of Chemistry, University of Exeter, Stocker Road, Exeter EX4 4QD Received 15th March, 1984 The desorption of ethylene, 1,2-dichIoroethane (EDC), carbon dioxide and other adsorbates from CuCl, has been studied using temperature-programmed desorption (t.p.d.). A chemi- sorptive centre corresponding to an ethylene concentration of 5 x lo+' mol mP2 has an energy of activation for desorption (Ed) of 73 kJ mol-l. Another chemisorptive site has a value of 57 kJ mol-l for E d when the ethylene coverage is 10-6-10-7 mol mP2. These magnitudes are closely related to previously reported heats of chemisorption, suggesting adsorption to be only weakly activated.EDC is shown to desorb readily whereas CO, has three distinct values of E d (85, 94 and 97 kJ mol-l). An assumed value of 1013 s-l for the pre-exponential factor Ad is appropriate in Redhead's method of analysis ; the kinetics are consistent with unimolecular decomposition. In this paper we report results relating to the desorption of species from copper(I1) chloride; adsorption behaviour has been discussed in Parts 2 and 3. Temperature- programmed desorption is used to evaluate activation energies of desorption and surface coverages. These can be correlated with results obtained by gas-adsorption chromatography (g.a.c.) to complete the energetics of adsorption-desorption for ethylene oxychlorination.EXPERIMENTAL APPARATUS The temperature-programmed desorption technique consists of the following steps : ( 1) catalyst pretreatment, (2) preadsorption of the adsorbate, (3) evacuation after preadsorption to remove the physically adsorbed gas, (4) programmed desorption of the residual chemisorbed gas into the stream of a carrier gas, ( 5 ) detection of the desorbed gas in the carrier and (6) trapping and analysis of the desorbed gas to establish its identity. The apparatus was designed to allow these procedures except for step (3), i.e. there was no vacuum facility, and the experiment relied on the adsorbent being held at the boiling point of the adsorbate for sufficient time to remove physically adsorbed gas. Another requirement of the apparatus was that it should contain a cryogenic section where th.e adsorbent could be cooled down to ca.-200 "C and then the temperature increased in a controlled manner up to ca. 300 "C. This was achieved by the novel use of small aluminium arid brass blocks closely fitted around the sample tube which were cooled by immersing them in a Dewar of liquid nitrogen and then allowed to warm up inside a temperature-programmable oven. The apparatus is summarised schematically in fig. 1. In contrast to the g.a.c. experiments, in order to minimise the possibility of readsorption of t Present address : Johnson Matthey Research Centre, Blounts Court, Sonning Common, Reading. 8384 DESORPTION STUDIES IN OXYCHLORINATION CATALYSIS Fig. 1. Schematic diagram of t.p.d.apparatus. Key: A, adsorbate cylinder/regulator; C, catalyst; CR, chart recorder; He, helium cylinder/regulator ; I, injection head; K, katharometer ; MS, molecular sieve; R, restrictor; SV, switching valve; T, thermocouple/digital voltmeter/chart recorder; Z, oven. the desorbing materials' a short plug of copper(~i) chloride (ca. 1 cm long, 0.5 g) was used in the t.p.d. experiments. The catalyst was preheated at 250 "C for 1 h under helium to allow comparison to g.a.c. results. After cooling to room temperature, the exit side of the sample tube was connected directly to a katharometer detector and the apparatus was leak-tested. With the catalyst held at a low temperature T, by liquid nitrogen and by-passed by the carrier gas, a stream of adsorbate was introduced into the flow.All adsorbates were of standard laboratory grade as in Part 2. The pneumatic switching valve was then used to divert the helium+adsorbate stream over the catalyst. When the adsorbate was indicated by the detector to be in excess of that amount which the catalyst could adsorb, the adsorbate supply was closed off. For liquid adsorbates the switching valve was not necessary, a syringe being used to inject into the carrier-gas stream over the catalyst. The sample temperature was maintained at until the detector response had returned to close to the baseline. (No vacuum facility was available to remove totally all physically adsorbed material.) The temperature of the catalyst was then raised at a linear rate p using the metal block and programmable oven.The natural heating rates for each metal block had to be determined in separate experiments so that the ramping of the oven could be matched to ensure a continuation of the initial nearly linear rate. To check the identity of desorbing species the katharometer was replaced by a glass column of 15% MS 550 silicone on Chromosorb W. AW-DMCS at 80 "C with a flame ionisation detector. This column separated ethylene from all chlorination products, and for the heating rates generally used allowed sampling every 10 "C. RESULTS AND DISCUSSION It was found that the different metal blocks gave a range in/3 of 4-12 K min-l, which could be extended to 80 K min-l by removing the block completely, but at higher rates there was considerable error in reading T, and B.The apparatus has therefore provided a more suitable heating rate than the very rapid exponential rate previously used2 for eliminating thermal gradients in the catalyst.P. G. HALL, P. HEATON AND D. R. ROSSEINSKY 85 time c- c temperature Fig. 2. Desorption spectrum for ethylene from copper(r1) chloride. Table 1. Activation energies for the desorption of C,H, from CuC1, 1 2b 3 4 5 6 7 8 9 10 187.85 218.45 215.95 206.7 216.2 21 1.2 212.15 187.65 213.15 194 0.1307 0.3436 0.5464 0.4902 0.472 1 1.2165 0.1910 0.0538 0.1286 0.0538 53 59.8 58.2 55.9 58.7 55.7 59 54 60 55.8 246.7 0.0555 71.5 263.15 0.2469 73 C - - - - - - 257.15 0.4386 75 254.15 0.1626 71.4 257.1 0.1032 73.6 - - - a Runs 1-6 on same sample of CuC1,; runs 7-10 on another sample. Run 2 shown in fig.2. - Denotes peak not clearly distinguishable. RESULTS FOR THE DESORPTION OF ETHYLENE FROM COPPER(II) CHLORIDE For a constant heating rate, the position of peak maxima were independent of the preadsorption temperature T,. A value of -90 "C for T, was chosen to avoid the inherent problems of lower temperatures with physisorbed and liquid ethylene (b.p. = - 104 "C), which appears generally as a peak at ca. -92 "C. Maintaining the sample at -90 "C for 10 min caused the loss of the majority of weakly held species. Thus the desorption chromatogram shown in fig. 2 does not start with a flat baseline. The main features of the spectrum are a major peak at ca. - 60 "C and a smaller peak shoulder at ca. - 10 "C. Sampling by g.c. at ca. 10 "C intervals showed that all desorbing species in this temperature region were pure ethylene, i.e.there was no86 DESORPTION STUDIES IN OXYCHLORINATION CATALYSIS L l I l I 1 I I I I 1 I 1 I I I 1 I l 1 I I I I I l I I I I I I TPC Fig. 3. Desorption spectrum from copper(rr) chloride. -80 -60 -40 -20 0 20 40 60 80 dissociative adsorption. However, an important observation revealed by sampling at high sensitivity is that the concentration of ethylene never reaches zero but reduces to approximately one-seventieth of the value at the main peak maximum. The actual heating curve also appears above the peaks, a twin channel recorder being used. Each desorption was continued up to 120 "C and then the temperature was reduced to T, again and the process repeated for different heating rates; therefore a sample was only preheated to 250 "C once, before any ethylene adsorption.The value of 120 "C as an upper limit was used because of the complicating effects of chemical reaction and disproportionation of the cupric chloride above 150 "C, which was detected by the katharometer as a continuous increase in baseline. The results for a series of heating rates are given in table 1, with activation energies for desorption Ed calculated by Redhead's method3 with an assumed value of 1013 s-l for the pre-exponential factor Ad. Since E d values show a random variation with p, the assumption for Ad is valid. The values of the two activation energies for desorption are 57k4 and 73+2 kJ mol-l. This compares with 73+ 1 and 83 k 1 kJ mol 1-1 reported2 for higher exponential heating rates.After calibration the area of the major peak corresponded to a surface concentration I' of ca. 10-6-10-7 mol m-2. The uncertainty in r arises from the resolution of the peak from shoulders (see dotted line in fig. 2) and the unknown extent of tailing. The amount of C2H, desorbing with E d = 73 kJ mo1-l was equal to a surface concentration of ca. 5 x mol m-2. RESULTS FOR THE DESORPTION OF 172-DICHLOROETHANE FROM COPPER(I1) CHLORIDE For preadsorption temperatures below the melting point (-35 "C) of 1,2- dichloroethane (EDC) it was found that the extensive desorption on melting, even for a 1 mm3 dose size, decreased the sensitivity of the experiment to an unacceptable level. Choosing T, to be -20 or 20 "C overcome this problem, but a longer length of time than for ethylene was required for physically adsorbed material to be desorbed; 30 min was allowed.The effect of heating was to cause a slow steady increase in the amount of EDC desorbing up to its boiling point (83 "C) and then a slow decrease with no other maximum before heating ceased at 150 "C. Sampling at ca. 20 "C intervals revealed that only EDC was desorbing. In one experiment, the solid was held at T, = 20 "C for 4 h to allow all weakly adsorbed species to desorb; on heating, no EDC was seen. The conclusion is that EDC is only physisorbed on copper(r1) chloride and readily desorbs. RESULTS FOR THE DESORPTION OF CARBON DIOXIDE FROM COPPER@) CHLORIDE = -90 "C, a desorption spectrum was obtained from the katharometer detector as shown in fig. 3. In preliminary experiments using ethylene as adsorbate forTable 2.Activation energies for the desorption of CO, from CuCl, 1 2 3 4 6 7 8 9 10 11 12 13 294 290 294 295 316 319 302 310.5 309 304 302 28 1 280 0.0575 0.0555 0.0529 0.0779 0.1035 0.0983 0.2131 0.23 15 0.23 15 0.1267 0.2584 0.1004 0.0992 85 84 85.5 85 90.5 91.5 84.5 87 86 86.5 84 80 80 33 1 33 1 332.5 331.5 333 333.5 326.5 0.0667 0.06 17 0.0667 0.08 17 0.1 138 0.1075 0.1945 96 96 96.5 95.5 95 95.5 92 - 338 321.5 3 20 318 - 0.258 0.2564 0.0952 0.09 16 - 94.5 90 92 91.5 a - - 339.5 344.5 345 342 347 345 347 342 332 332 - - 0.0641 0.093 1 0.1263 0.2203 0.254 0.265 0.273 0.2564 0.0980 0.0926 - 98.5 99 98.5 96 97 96.5 97 95.5 95.5 95.5 - a Denotes peak not clearly distinguishable. Run 5 spectrum shown in fig.3. 9 z b P w88 DESORPTION STUDIES IN OXYCHLORINATION CATALYSIS Comparison with the spectrum from an f.i.d. indicated that all peaks above 0 "C were inorganic. The peaks at ca. 40, 60 and 75 "C were identified by control experiments as being due to the desorption of carbon dixoide. The spectra obtained for the desorption of CO, from CuCl, were independent of the presence of C2H4 (and vice versa) and the value of T,, provided that T, was below the sublimation temperature (- 78 "C) of carbon dioxide. The results for a series of heating rates are given in table 2, with activation energies for desorption calculated by Redhead's method3 with an assumed value of 1013 s-l for Ad. The table shows it is not necessarily possible to distinguish all 3 peaks on each run.The average values of the activation energies for desorption were 85, 94 and 97 kJ mol-l. There was no apparent dependence of Ed on /3 so the assumption of 1013 s-l for Ad was justified. No calibration for the amount of CO, adsorbed was carried out, but it was a factor of 10, greater than the amount of C,H, (fig. 3). The magnitudes of Ed indicate non-dissociative adsorption of CO,. RESULTS FOR THE DESORPTION OF OTHER ADSORBATES In order to identify the desorption peak at ca. 10 "C, oxygen and water vapour adsorbates were studied. Owing to the proximity of the melting point, water was unsuitable for t.p.d. in the temperature range of interest. Oxygen preadsorbed on CuCl, gave no peaks, although CO, impurities in the oxygen gave a weak background of peaks between 40 and 80 "C.Dosing of copper(]) chloride with oxygen at - 196 "C gave small peaks at - 135 and 3 "C in one run, and for another sample peaks at - 145 "C (B = 0.2688 K s-l) and 6.5 "C (p = 0.2268 K s-l) were obtained. When copper(r) chloride was dosed with CO, at -90 "C a single peak at 3.5 "C (B = 0.2315 K s-l) was obtained. Thus the unidentified peak could be due to the desorption of CO, or 0, from CuCl present by disproportionation during pretreatment. On heterogeneous surfaces Ad will vary, and consequently kd also varies. Distortion of curves attributed to varying Ed is often due to varying kd. An interesting analysis of factors that may influence Ad for first-order kinetics is given by Peter~nann.~ An evaluation of readsorption may be gained by the effect on Tm of the flow rate F.However, T, is approximately proportional to -log Fso a large variation is required. Despite these assumptions Ed values in the present work were obtained to a reproducibility of S%, indicating the useful nature of the simple Redhead analysis method. The important parameter is T,; an error of + 5 % gives an error of + 5 % in Ed. A variation of f 5% in gives an error of only 0.1 % in Ed. CONCLUSIONS The desorption spectrum for ethylene from copper@) chloride, shown in fig. 2, clearly indicates that besides physically adsorbed C2H4 there are two other types of chemisorbed ethylene, neither being dissociatively adsorbed. The amount of material adsorbed remained roughly constant for different preadsorption temperatures, and thus the two peaks cannot be attributed to coverage-dependent repulsions between adsorbed molecules.One of these adsorption sites has a surface concentration r of ca. 5 x mol rn-, and an Ed value of 73 kJ mol-l. The results in Part 2 show that qst at this region of r is 65 kJ mol-l. The surface concentration of the other site is ca. 10-6-10-7 mol m-,. Table 10 in Part 2 clearly shows that qst is nearly constant at 44 kJ mol-l for this region of r. Thus t.p.d. and g.a.c. lead to the conclusion that other than physically adsorbing sites there are two types of sites, as also observed by Par~ley,~ which have energetics summarised by fig. 4. The values of the activationP. G. HALL, P. HEATON AND D. R. ROSSEINSKY 89 65 i‘ k J mol-’ 4 1 4 3 k J mol-’ I I I 57 k J mol-’ Fig. 4. Adsorption-desorption energy diagram for ethylene on copper(I1) chloride.energy for adsorption Ei are 8 and 13 kJ mol-l, the low value of which indicates that surface diffusion is not necessary for desorption, but the non-zero values confirm that the C2H4 adsorption is chemisorption. These conclusions ignore the broad width of the major desorption peak and the observation that the concentration of desorbing C2H, did not reach zero below 300 “C. The broadness indicates surface heterogeneity, with a range of higher energy sites, the surface concentration of which could not be determined. These higher-energy sites could be due to edges, dislocations and other surface imperfections, the extremely low surface concentration of which generally render them of little catalytic interest.using T, and the peak width at half-maximum only, gave Ed = 71.5 kJ mol-1 for the higher-energy site of C,H, on CuCl,, which is in very good agreement to 73 kJ mol-1 obtained by Redhead’s analy~is.~ However, for the other sites it gave Ed = 15 kJ mol-l, which is too low because of the peak shape being broadened by surface heterogeneity. Results for EDC indicate that it is only physically adsorbed and desorbs readily. Carbon dioxide has a high activation energy (ca. 100 kJ mol-l) for desorption when preadsorbed on CuC1, below its sublimation temperature. When dosed at higher temperatures such as those used in g.a.c. it readily desorbs. An alternative Financial assistance for this work was provided by an S.E.R.C. - Imperial Chemical Industries PLC (Runcorn) CASE award. We also thank Dr R. A. Hann (I.C.I.) and Dr T. Tribbeck (I.C.I.) for helpful discussions and advice. R. J. Cretanovic and Y. Amenomiya, in Catalysis Reviews, ed. H. Heinemann (Marcel Dekker, New York, 1972), vol. 6, p. 21. P. G. Hall, M. Parsley, D. R. Rosseinsky, R. A. Hann and K. C. Waugh, J. Chem. SOC., Faraday Trans. 1, 1983, 79, 343. P. A. Redhead, Vacuum, 1962, 12, 203; Trans. Faraday SOC., 1961, 57, 641. L. A. Petermann, Nuovo Cimento, Suppl., 1967, 5, 364. M. Parsley, Ph. D. Thesis (Exeter University, 1979). L. D. Schmidt, Catal. Rec. Sci. Eng., 1974, 9, 1 1 5. ’ D. Edwards Jr, Surf. Sci., 1975, 54, 1 . (PAPER 4/42 1 )