J. Chem. SOC., Faraday Trans. I, 1982, 78, 657-676 Chlorine-ca t a1 y sed Pyr ol y sis of 1,2-Dichloroe t hane Part 1 .-Experimental Results and Proposed Mechanism BY PHILIP G . ASHMORE,* JOHN W. GARDNER, ANTHONY J . OWEN, BARBARA SMITH AND PHILIP R . SUTTON Department of Chemistry, University of Manchester Institute of Science and Technology, P.O. Box 88, Manchester M60 1QD Received 30th December, 1980 Experimental studies of the pyrolysis and chlorination of l,2-C,H,C12 in the presence of small proportions of chlorine, or of chlorine plus nitric oxide, confirm that the main propagating steps between 520 and 620 K are 1 el+ C,H,CI,(DCE) + ezH3C12(k) + HCI 2 t,H,Cl, C,H,Cl(VC) + el -2 C1, + e2H3Clz A C,H,CI,(TCE) +el. The initial rates (d[VC]/dt), and (d[TCE]/dt), decrease together as the vessels age, but k,/k, = y remains constant at constant PDCE. y increases as PDCE is increased, or if inert gases are added, as expected from unimolecular behaviour of k,, and the Arrhenius parameters of y increase together as PDCE is increased.y(P) is evaluated for several ranges of PDCE at five temperatures to allow study of k i p ) by unimolecular theory in Part 2. The addition of VC lowers (d[VC]/dt), through reaction (-2), and the results are used to evaluate k-,/k, ; k-, is also pressure-dependent. In vessels with fresh surfaces, d[VC]/dt is proportional to PDCE xp:f* for low pel, and is independent of added inert gases; in aged vessels, the order in PDCE becomes 0.62, very close to the dependence of y(P) on PDCE. Calculations of [k], and [el], from the observed rates and rate constants point to an initiation step C12+S -,clS+el and termination by the reverse reaction in fresh-surfaced vessels, where S is a surface site; with aged surfaces, the results point to the same initiation step combined with termination by k + SCI + C,H,CI, + S especially at lower temperatures.The pyrolysis of 1,2-dichloroethane (DCE) to hydrogen chloride and vinyl chloride (VC) at temperatures between 670 and 770 K is thought to proceed by a radical chain mechanism with the propagation steps Cl + C,H,Cl, -+ C2H3C12 + HCl C,H,Cl, -+ C,H,Cl+ e l . The early investigations' 9 were interpreted in terms of homogeneous initiation and termination steps, but there are many difficulties with such an interpretation. The reaction rate generally increases in a run to a maximum rate, and widely different Arrhenius parameters have been reported for the overall rate constant.The differences seem to arise in part from changes to the vessel surface, suggesting that initiation 657658 CHLORINE-CATALYSED PYROLYSIS OF 1,2-C2H4C12 and/or termination may occur on the ~ u r f a c e s . ~ ? ~ Holbrook et aL5 also pointed out that the autocatalysis of the thermal decomposition might be caused by chlorine formed by surface reactions C,H,Cl, + surface(s) C,H,Cl,(s) C,H,Cl,(s) + C,H,(g) + Cl,(s) + els + Cl(g). Catalysis by ca. 1% of chlorine had been reported by Barton,, using a Pyrex flow system at 623 K, and by Takahashi et aL6 who used a stainless-steel reactor and proposed that the chlorine initiated the chains by the reaction metal + CI, + metal-Cl+ O(g).Huybrechts et al.' investigated the chlorine-sensitised photo-decomposition, and from the relative rates of formation of VC by reaction (2) and of 1,1,2-trichloroethane (TCE) by reaction (3) (3) Cl, + C2H3C12 -+ C,H,Cl, + el k 2 / k 3 = (d[VC1/dt) [C121/(d[TCEl/dt) they determined the ratio k2/k3 at five temperatures between 433 and 510 K. From the published* Arrhenius parameters for k , they found (1) However, a contemporary investigation by Gardner (J.W.G.) of the chlorine-catalysed thermal decomposition, at temperatures between 525 and 630 K, showed much lower values of k , than predicted by eqn (I), and also smaller Arrhenius parameters.@ In a short extension of J.W.G.'s work, Sutton (P.S.) showed that k, falls with total pressure as expected for a unimolecular rate constant at appropriate pressures,1o a possibility not considered in the photolysis studies.The effects of different pressures of DCE and of added inert gases upon k , have later been investigated in more detail by Smith (B.S.) and Owen (A.J.O.), who also investigated the inhibitory effects of VC through reaction log,,, k,/s-' = 14.33f0.47-(10710+630) K/4.576 T. Cl+ C,H,Cl + C2H3C12. (-2) (-2) This paper summarises the essential features of the dependence of the rates d[VC]/dt, d[TCE]/dt and - d[DCE]/dt on reaction conditions, and our evaluations of khP)/k3 = y ( P ) and kL%) at different temperatures and pressures. These results, and published data on k , and k,, are used to evaluate [Cl] and [C2H3C12] for different reaction conditions, and hence to provide information about the initiation and termination steps in fresh- and aged-surface vessels.In the following paperll the predictions of various theories of unimolecular fall-off are tested against A. J.O.'s extensive observations on the variations of kip) and kL%) with pDCE in order to obtain the best model for reaction (2). EXPERIMENTAL The cylindrical vessels were maintained at temperatures controlled by a Sirect controller and measured by a Pt/13% Rh-Pt thermocouple, in a tubular electric furnace (J.W.G., P.S.) or an air-thermostat with forced circulation (B.S., A.J.O.). Mixtures were made up in heated, blackened vessels and fed to the reaction vessel through heated tubing and greaseless stopcocks (Young's).Vessels A and B were of quartz with plane ends and s/v ratios 1.4 and 6.6 cm-l, respectively.B Pyrex vessel C, of dimensions similar to A, was used uncoated and,then with coatings of KCl, AgCl and Teflon to investigate the effects of different surfaces. Pressures wereASHMORE, GARDNER, OWEN, SMITH A N D SUTTON 659 measured (J.W.G., P.S.) by an all-glass transducer, developed and kindly supplied by Dr P. J. Thomas of I.C.I. Mond Division, or by an S.E. transducer (B.S., A.J.O.), and recorded on a Servoscribe chart recorder. The reactants and samples of products were purified as described el~ewhere.~ To check our 1 ,2-C2H,Cl,, Dr G. Martens kindly supplied some unstabilised material; both samples gave the same results. The organic compounds were identified and determined qlantitatively by gas chromatography, using a Pye Unicam model 104 dual-column instrument, flame ionisation detector and a temperature programming unit.The columns used and the calibrations are detailed elsewhere.g The product HC1 was determined by titrating samples from the reaction vessel and also by determining the pH of a solution prepared by a standard routine from the gas-chromatography sample. Pressures ofchlorine were determined with a single-beam photometer, using a monochromator (J.W.G.) or a Carl-Zeiss filter (B.S., A.J.O.) passing wavelengths c1os.- to 330 nm. The exit beam was monitored using an RCA 1P28 photomultiplier (J.W.G.) or an EM1 photomultiplier (B.S., A.J.O.), the output being fed to a Servoscribe recorder. Frequent calibration checks against chlorine pressures were carried out with both systems.In experiments using chlorine plus nitric oxide to catalyse the decomposition by the reaction NO + C1, -+ NOCl + e l the nitrosyl chloride formed (in small amounts, ca. l0-15% of the initial nitric oxide) was monitored9 using 250 nm radiation from the monochromator. RESULTS OVERALL REACTIONS A N D RATES Early work by J.W.G., confirmed by later checks, established that the primary products of the decomposition catalysed by chlorine were hydrogen chloride, vinyl chloride (VC) and 1,1,2-trichloroethane (TCE). The secondary products 1,l- dichloroethylene and cis- and trans- 1,2-dichloroethylene are formed by pyrolysis of the TCE. Typical results are shown in fig. 1 (a). A careful search was made for other C, and C, chloroalkanes and chloroalkenes, ethane, ethylene and methane; it can be asserted that none were present in concentrations > of typical initial DCE concentrations.It was also established by gas-chromatographic analysis that no detectable amounts of organic derivatives of NO were formed during the experiments with NO and Cl, present. High-molecular-weight products would not have been detected, but sample mass balances of observed reactants and products agreed within & 1 % at all extents of reaction in both vessels. Initial mixtures of DCE, C1, and NOCl gave much slower initial rates than with chlorine alone; initial mixtures of DCE, C1, and NO gave much faster rates, but these fell off rapidly as NOCl was formed. Later in each run with NO plus Cl,, however, the pressure increases were closely equal to the pressure of VC formed, as shown in fig.l(b). The amount of chlorine that disappeared was again closely equal to the amount of TCE f ~ r m e d . ~ The changes were the same in the fresh vessels A and B, as shown, and were found to be the same for pressures of NO between 0.5 and 5.0 T ~ r r . ~ * The absence of surface effects clearly indicates homogeneous initiation and homogeneous termination, probably by the reactions NO + C1, + NOCl + e l . At suitable points during various runs, determinations were made of k,/k, from (d[VC]/dt) [Cl,]/(d[TCE]/dt) and these are discussed in a later section. * 1 Torr = 101 325/760 N m-2.660 CHLORINE-CATALYSED PYROLYSIS OF 1,2-C,H4Cl, time/s LO 50 60 70 L O 30 t 20 --.3 5 a b \ 10 0 time/s ,d FIG. l.-(u) Correlations between pressure change (0) and VC formed (0) and between HCl formed (0) and VC plus TCE plus dichloroethylenes formed (+). Relationships between C1, lost (V), TCE formed (A) and total dichloroethylenes formed ( x). pDCE,O = 82.5 Torr at 584 K. Vessel B. (b) Correlations of pressure changes in vessel B (0) and in vessel A [a] with pvc formed (A) for initial pressurespcll = 10.8 Torr, pNO = 2.0 Torr and PDCE = 120.5 Torr (curve I) or 73.5 Torr (curve 11). 544 K. The curves are independent of pNO between 0.5 and 5.0 Torr. In the chlorine-catalysed reaction the overall initial reactions are and 1 ,2-C2H4Cl2 + CH2=CHC1 + HCl 1,2-C,H,Cl, + C1, -+ 1,l ,2-C2H3C13 + HCl. We have repeatedly confirmed by comparison between pressure increase, photometry of C1, and analyses of reactants and products, as illustrated in fig.l(a), that (dptotal/dt)o = (dp,,/dt), and - (dpClp/dt),, = (dpTCE/dt),, and hence that the initial rates could be determined satisfactorily from the continuous recording of total pressure (transducer) and chlorine partial pressure (photometer). The rates were corrected for the dead space;12 this is very important, because the corrections for dptotal/dt and - dpCl2/dr are in opposite senses.ASHMORE, GARDNER, OWEN, SMITH A N D SUTTON 66 1 INITIAL RATES (dp,,/dt)o After ca. 100 successive runs J.W.G. found the initial rates settled to nearly constant values for fixed initial pressures of DCE and C1, at fixed T. These rates, in the still-fresh vessels A (quartz, s/v = 1.4 cm-l) and B (quartz, s/v = 6.6 cm-l) were strictly proportional9 to PDCE,O with fixed pclz, o.Defining k’ by (d~vc/dt)o = ~’PDCE, o k’ was found to vary with pel,, as shown in fig. 2 for the fresh vessels A and B at / Y I I I I I P C l , l T O ~ 0 1.0 2.0 FIG. 2.-The experimental rate constants k’l,4 (0) in fresh vessel A (s/v = 1.4 cm-l) and k’6.6 (0) in fresh vessel B (s/v = 6.6 anF1) against pel, at 584 K. The solid curves is based on point + with k’6.6 proportional to PEf2. 584 K. In vessel A, k’l.4 appeared to reach a limiting value as pc12,0 increased to moderate values. In vessel B, however, a limiting rate was not achieved even at much higher values ofpClz, o. Indeed, in the lower range ofpClz, o, the rates were proportional to ptf:,, o.This is illustrated in fig. 2, where the line for k’6.6 is based on point + with k’6.6 OC Podfz, 0’ In the aged vessel B (i.e. after ca. 500 runs) the initial rates remained proportional to over a wide range, as shown in fig. 3 for three pressures of DCE. Moreover, in other vessels with different surface coatings the rates with fixed PDCE were found by A.J.O. to be proportional top$fz, or to a slightly higher order. Thus the behaviour of fresh vessel A seems in retrospect to be anomalous. As vessel B aged (i.e. during the experiments of J.W.G. and B.S.) several significant changes in the kinetics of the decomposition occurred. In the first place the rates for chosen conditions of T, pDCE and pclz decreased significantly as shown in table 1.Secondly, the order of reaction in pDCE decreased significantly, falling well below unity in later experiments by B.S.; from the slopes in fig. 3 it was deduced that in the aged vessel B (d~vcldt), = Q(~ci,, (PDCE, o / T ~ r r ) ~ . ~ ~662 CHLORINE-CATALYSED PYROLYSIS OF 1,2-C,H4C1, 0.9 / / / Y 4 0.3- s I 0 1.0 2.0 3.0 (pa,/Torr)t’So / I L .O A / I 1, 5.0 FIG. 3.-Plots of (dp,,/dt), against (pc1,)g.6 in aged vessel B with pDCE,o equal to x , 35; A, 65 and 0, 135 Torr. 572 K. TABLE l.--(dp,,/dt),/Torr s-l IN FRESH AND AGED VESSEL B AT 572 K PClZ, ,/Tom state of B 35 65 135 1 .o fresh 0.37 0.70 1.43 aged 0.10 0.13 0.19 2.0 fresh 0.53 0.98 2.02 aged 0.14 0.19 0.29 where B is a specific rate. Thirdly, at a fixed chlorine pressure (1 .O Torr) the activation energies for k’1.4 (vessel A, fresh), k’6.6 (vessel B, fresh) and for /I (vessel B, aged) were found to be 27, 32 and 47 kcal mol-l, respectively.Fourthly, while investigating the effect of adding inert gases J.W.G. found that large pressures of added argon had no effect on (dp,,/dt),, whereas B.S. found that adding inert gases such as CO,, CCl,, cyclo-C4F, and C,Fl, all increased (dp,,/dt),, the increases being larger for the gases with larger molecules. Table 2 illustrates the change for adding CO, to 41 Torr of DCE at 532 K (the significance of yexp is explained in the next section of this paper). Fifthly, the order in chlorine in fresh vessel A was not simple, as the rates reached limiting values at moderate pcl,,o; in the fresh vessel B it was 0.5 at low values of pc12, ,, but not simple at higher pressures; in aged vessel B, it was 0.5 over a wide range of pressures; in other vessels, with different surface coatings, it was 0.5 or (at some temperatures) slightly higher.ASHMORE, GARDNER, OWEN, SMITH AND SUTTON 663 TABLE 2.-&LATIVE MAGNITUDE OF (dpvc/dt), ON ADDING co, TO 41 TORR OF DCE (VESSEL B, 532 K) relative magnitude relative magnitude PCO*/PDCE dPvc/dt Yexp 0 1 .oo 1.80 2.60 1 .oo 1.39 1.63 1.80 1 .oo 1.41 1.61 1.74 In summary, it appears that the rate (dpvc/dt)o depends on the state of the vessel as well as on T, pDCE,O and pClz,,.STUDIES OF THE PROPAGATING STEPS (2) AND (3) I N THE CHLORINE-CATALYSED DECOMPOSITION If VC and TCE are formed predominantly by steps (1)-(3), where R = CH,ClCHCl, C~+DCE -, R+Hc~ (1) R+VC+Cl (2) R+C12 + TCE+Cl (3) and the reverse of reaction (2) is ignored while [VC] is small then where yexp is calculated from the rates and pclo.J.W.G. measured the rates (dptOtal/dt) and - (dpClz/dt) and pc12 at various points during each of many runs. On plotting (d&,,,,/dt) x pcle against - (dp,,Jdt) it was found that points for runs in fresh vessels A and B at the same temperature and the samep,,., fitted the same straight line, as exemplified in fig. 4 forp,,,, = 24 Torr. These results suggested that yexp was independent of the vessel and of the percentage of chlorine, and did represent the ratio k2/k3 of the rate constants of two homogeneous reactions. It was thought at that time that yexp was also independent ofp,,, (which was usually in the range 40-90 Torr) so the values were averaged at each temperature.The averages for five temperatures are shown on an Arrhenius plot in fig. 5, with the weightings of the averages in brackets, together with four values derived from the experiments9 on the homogeneous catalysis of the decomposition by chlorine + nitric oxide. Line I11 in fig. 5 was fitted by a least-squares procedure to the chlorine-catalysed results. Regression analysis of the individual experimental variances shows the 95 % confidence limits on E2 - E3 and on log[(A2/A,)/dm3 mol-'1 are as shown in eqn (111) k,/s-l 17 100 & 350 4.576 T/K ' = (3.57 0.13)- loglo k3/dm3 mol-l s-l These results seem to confirm that the principal reactions forming VC and TCE in the chlorine-catalysed pyrolysis are the homogeneous reactions (2) and (3).However, doubt was cast on these results or their interpretation when it became known that Huybrechts et a1.' had studied the apparently similar chlorine-sensitised photo-664 CHLORINE-CATALYSED PYROLYSIS OF l,2-C2H,C1, ;7 I/" XO I 0 1 2 3 4 5 6 7 6 9 -(dpa,/dt)/Torr min-' FIG. 4.-Plots for 578 and 561 K which show (dptotal/df)pC1, is proportional to - (dp,,,/dt), independently of the vessel. Vessel A, s/v = 1.4 cm-' ( x ) ; vessel B, s/v = 6.6 cm-' (0); pDCE.0 = 24 Torr. -2.5 7 -3.0 E 'c1 - E 1 n * -3.5 2 Y 0 - - 3 -4.0 -4.5 I I I I I lo3 K I T M 1.5 1.6 1.7 1.8 1.9 2.0 2.1 2.2 2.3 FIG. 5.-Arrhenius plots of log (k,/k,) for chlorine catalysed (0 with weightings in brackets) and nitric oxide+chlorine catalysed (+) decompositions (line 111) and for eqn (IV), (VI) and (VII).ASHMORE, GARDNER, OWEN, SMITH A N D SUTTON 665 decomposition of 1,2-dichloroethane over the temperature range 433-5 10 K and found k,/s-l 19970f 580 4.576 T/K .= 5.58 f 0.27 - loglo k3/dm3 mol-l s-l Eqn (IV) gave considerably higher values of k,/k, than eqn (111) as shown in fig. 5. In searching for reasons for the discrepancies, it was considered that they could not be attributed to the onset of reactions other than (2) or (3) at our higher temperature!, nor to other effects of the photolysing beam such as enhanced decomposition of R. In view of the unimolecular nature of reaction (2), the most likely explanation of our lower results at lower total pressures than those of Huybrechts et al. was the unimolecular fall-off of k , at pressures around 10-100 Torr.The apparently similar reaction c2H5 + C,H, + H shows fall-off in that (total) pressure range.13 In a short project, P.S. found a clear dependence of k, on pressure of DCE by exploring a wider pressure range and lower temperatures than used by J.W.G., as reported in ref. (10). Those experiments were extended by B.S., who found further evidence for unimolecular behaviour of reaction (2) in experiments with selected pressures of DCE at temperatures between 520 and 575 K in vessel B. Typical fall-off curves were found for yexp, illustrated by curve (1) through the points 0 in fig. 6 I/- 0 I 0 LO I I I 80 120 160 PDCE + pco2/Torr FIG. &-Relative values of y as pDCE, is increased (0, curve 1) and as CO, is added to 41 Torr of DCE (0, mean of 9 runs, curve 2).532 K. Vessel B, intermediate age. which shows yexp relative to the value forp,,, = 41 Torr. Fig. 6 also shows the effects on yrelative of adding CO, to 41 Torr of DCE at 532 K as points 0, each being the mean of 9 runs. The increases in yrelative as CO, was added are shown in table 2, together with relative changes in (dp,,/dt). The increases in yrelative closely parallel the increases in dp,,/dt; there was very little increase in dpTCE/dt. Inspection of the rate equations for reactions (2) and (3) shows that these results can only be explained666 CHLORINE-CATALYSED PYROLYSIS OF 1,2-C2H,C1, by changes in k,, and not by changes in [R], as CO, is added. B.S. also observed that adding inert gases such as cyclo-C,F, and C,F,, increased yexp to a greater extent than did CO,, bringing the fall-off curves close to that of DCE itself.These 'inert gas' effects are in keeping with unimolecular rate theory. Chlorine itself would be expected to have less effect than CO,; in addition, the proportion of chlorine used was rarely > 10% of the total pressure, and all later experiments (A.J.O.) at fixedp,,, showed that yexp was independent of pclo in those proportions. A.J.O. obtained extensive results with aged vessel B and also with a Pyrex vessel C , of dimensions close to those of A, when uncoated (CP) and when coated with Teflon (CT). While running-in the vessels CP and CT, A.J.O. observed that the overall rates fell but yexp remained constant as shown in fig. 7.Table 3 summarises data at 572 K 1 . 2 1 . a 0 - 0.8 I 2 y" \ a 0.6 0.4 100 4 b t- 80 --.. 60 1 4 0 12 16 number of admission FIG. 7.-Changes in kexp = (d In PDCE/df)O (0) and constancy of yexp (A) with successive admissions in a fresh Teflon-coated vessel, s/v = 1.4 cm-l, 572 K. pDCE.0 x 62f2 Torr, 7.8-8.1 % C1,. for the three vessels after running-in, and shows that yexp is substantially independent of the particular vessel. Over the higher pressure range the results are a reasonable fit to the empirical equation (V) log,,(y/Torr) = 0.64+0.61 log,, (pDcE/Torr). The last line of table 3 gives the predicted values of y for the mean pressures in vessel B. The dependence of y UponpDCE can only be attributed to changes in k , withp,,,, k, K (PDCE/Torr)o'gl. so It is significant that this proportionality correlates very closely with the equation established for (dpvc/dt), in aged vessel B, viz.(dpVC/dt)O = k2[k10 = B(pcl,/Torr)~'5((pDCE/Torr),o.s2.TABLE 3.-MEAN VALUES p , 7 (WITH STANDARD ERRORS OF MEANS AND NUMBER OF RUNS) OF PDcE, YExp FOR DIFFERENT VESSELS AT 572 K quartz B p f S.E./Torr s/v = 6.6 cm-l jjf S.E./Torr no. of runs Pyrex C (CP) p f S.E./Torr s/v = 1.4 cm-l jjf S.E./Torr no. of runs Teflon C (CT) p f S.E./Torr s/v = 1.4 cm-l jjf S.E./Torr no. of runs prediction from eqn (V) of y for p values in aged vessel B 9.02 f 0.08 13.1 k0.49 19 9.05 f 0.13 12.6 f 1.20 4 8.97 f 0 . 16 12.3 f 0.89 6 22.8 f 0.33 29.1 f 0.78 18 23.4 f 0.35 32.8 f 1.20 4 23.3 f 0.25 32.6 f 2.93 8 30.1 36.5 f 0.29 40.1 f 0.48 19 36.4 & 0.58 41 .Of 1.07 3 36.7 f 0.3 1 44.7f 1.80 9 40.2 63.4 f 0.61 57.4 f 1.12 24 65.0 f 0.47 63.1 f 1.76 8 64.3 f 0.3 1 60.0 f 0.63 27 56.4 10 1.2 & 0.58 75.4 f 1.29 23 103.0 f 2.23 73.0 & 3.58 4 104.0& 1.11 70.0k3.12 8 75.2 TABLE 4.-MEAN VALUES p, 7 (WITH STANDARD ERRORS OF MEANS AND NUMBERS OF RUNS) OF PDCE, YExxp IN VESSEL B AT THE TEMPERATURES INDICATED 572 K p f S.E./Torr jjk S.E./Torr no.of runs 560 K p+_ S.E./Torr 7f S.E./Torr no. of runs 547 K p & S.E./Torr jjk S.E./Torr no. of runs 534 K p & S.E./Torr jjfS.E./Torr no. of runs 521 K pfS.E./Torr jjf S.E./Torr no. of runs 9.0 f 0.08 13.1 k0.49 19 9.2f0.13 12.5 f 0.54 5 9.3 f 0.33 8.9 f0.89 4 9.4f 0.12 5.3 f 0.16 6 9.6 f 0.46 3.6 f 0.36 5 22.8 & 0.33 29.1 f 0.78 18 24.5 f 0.66 26.1 f 0.92 8 22.5 k0.38 18.4f 0.83 4 23.0 f 0.26 11.9 f 0.41 6 18.6 f 0.25 5.2 f0.18 9 36.5 f 0.29 40.1 f 0.48 19 36.8 f 0.48 31.4f 1.40 7 36.8 f 0.76 24.7f 1.13 5 35.6 f 0.78 16.0f0.37 6 27.3 f 0.46 8.1 f 0.40 11 63.4f 0.61 57.4f 1.12 24 64.3 f 0.96 44.1 f 1.73 6 65.0 f 1.26 32.4 f 1.74 4 65.2 f 1.35 23.1 f 0.33 6 55.0 f 0.38 1 1.6 f 0.25 11 101.2 f 0.58 75.4 f 1.29 23 100.4f 1.51 54.0 f 2.00 6 97.3 f 0.90 38.3 & 1.58 6 100.2+_ 1.35 27.7 f 0.92 6 91.8 f0.77 16.8 f 0.62 11 134.6 f 0.90 87.2 f 1.80 11 138.3 +_ 1.82 63.8 f 1.55 4 - 131.7 f2.35 30.4 f 1.05 4 134.8 f 1.57 18.9 f 0.24 4 z X c3 4 0668 CHLORINE-CATALYSED PYROLYSIS OF l,2-C2H,Cl, The correlation means that [R], is independent of pDCE, in aged vessel B; this point is discussed in detail in a later section of this paper.Table 4 summarises the results from many runs in the aged vessel B, and shows the mean values of yig; for groups of pressures at five different temperatures. In Part 211 of this series RRKM calculations are described that identify the model of unimolecular fall-off that best fits these mean 7, p values, and the appropriate fall-off 90 80- 70 - 8 60- a3 * 50- 40- 30 - *O- b $ t- n5 I * 1Q - PLXE/Tofl FIG. 8.-Mean-values y' of yipb at 0, 572; 0, 560; A, 547; A, 534 and 0, 521 K with the number of runs which determined each mean. The curves show the RRKM predictions for model G (see Part 211 of this series). curves are shown in fig. 8. From these curves the variations of k2/k, with temperature was found to depend on the pressure range according to eqn (VI) and (VII) 16500 (100 Torr) 4.58 T/K = 2.86- 4.58 l5 550 T/K (25 Torr).= 3.65- k2/s1 loglo ( k3/drn3 mol-l s-l These equations are plotted in fig. 5 , and fit very satisfactorily around line I11 which represents J.W.G.'s mean values for the range 40-90 Torr. It therefore appears certain that the experimental results shown by lines I11 and IV are in keeping with predictions from RRKM theory illustrated by lines VI and VII. In Part 211 the high-pressure parameters A? and E p of kp are evaluated. EFFECTS OF THE PRODUCT vc ON THE RATE OF THE CATALYSED DECOMPOSITIONS J.W.G. showedg that the addition of VC to mixtures of DCE and C1, reduced the initial rate of decomposition, because reaction (-2) is then important. A full stationary-state treatment of reactions (l), (2), (- 2) and (3) shows that expression (11)ASHMORE, GARDNER, OWEN, SMITH AND SUTTON 669 for yexp is modified to give yvc in the presence of VC: vc dptotaddt yvc = ( -dpclz/dt) x p c l z (VIII) where a = k-,[VC]/k,[DCE].Thus if values of k,/k, are known it is possible to calculate k-,/k, for selected points in any run. Using this method J.W.G. found k-,/k, = 0.75 at 630 K. A.J.O. investigated this effect in greater detail in order to study the temperature variation of k J k , and to see whether k-, was pressure-dependent, as would be 1.61 . , 0 ( b ) 15 .5 10 b V >O + 1 5 0 0 , 0.4 0.8 1.2 1.6 2.0 2.4 @VC/PDCE) I 1 I I I I 1 I L 5 10 15 20 25 30 35 40 45 Pvc, 0 /Torn FIG. 9.-(a) Experimental values (0) of y r c dotted against pvc, with (pDCE +pvc)o = 65 Torr, at 520 K.The curves are calculated from eqn (VIII) for k _ , / k , = 1 .O, 1.3 or 1.6. (b) Plots of k _ , / k , at 520 K calculated from eqn (VIII) assuming yvc = 15.0 Torr for pDcE,o = 65 Torr, pVc,, = 0. The line is the least- mean-squares fit. Vessel B. expected by analogy with the pressure dependence of k,. Fig. 9(a) shows the experimental values of y y c determined from plotted against pvc, for runs at 520 K with pDCE, +pvc, kept constant at 65 Torr. The three lines show y y c calculated from eqn (VIII) assuming (a) k-,/k, has values 1 .O, 1.3 or 1.6 and calculating a from appropriate values of pvc, PDCE and pClz, and (b) k,/k, is constant and has the value for pvc = 0, which would only be true if DCE670 CHLORINE-CATALYSED PYROLYSIS OF 1,2-C2H4C12 and VC have equal efficiencies in energising or de-energising the radical t2H3Cl2.The best fit is for k-,/kl x 1.3, but the lines do not curve sufficiently. This may be because k 2 / k 3 changes as pvc, OIPDCE, o changes. A practical way round the difficulty is to use the same basic data and eqn (VIII) to calculate a and hence k-,/k, point by point, and to investigate whether the ratio changes systematically with pvc, O/pDCE, ,. Fig. 9(b) shows the plot of k-,/kl against pvc, O/pDCE, (with pDCE, , +pvc, , = 65 Torr) with the line fitted by least-mean- squares procedures. Similar systematic variation was found at other pressures (100, 35 and 25 Torr) at 520 K, and also from experiments at 544, 570 and 590 K. For comparison of results we take the intercept values of k-,/k,, so that the only energising/de-energising species is DCE.TABLE VALUES OF k-,/k, AND 0 = log,, (k-,/dm3 mo1-l s1). (21, IS THE MEAN OF THE 0 AT EACH TEMPERATURE. ~ PDCEITOrr x 25 x 35 x 65 x 100 0 m 595 K k-,/k, 0.79 0 9.56 570 K k-,/k, 0.76 0 9.49 544K k-,/k, 0.91 % 9.51 520K k-,/k, 1 .oo 0 9.50 0.87 9.60 0.8 1 9.52 0.89 9.50 1.15 9.56 1.12 9.71 1.45 9.77 1.41 9.70 1.35 9.63 1.66 - 9.88 9.69 1.74 - 9.85 9.66 1.90 9.83 9.64 1.75 - 9.74 9.61 - Table 5 shows the intercept (pvc = 0) values of k-,/kl, and the corresponding values of 0 = log,, (k-,/dm3 mol-l s-l), using the published data14 for k,. At all four temperatures the values of k-,/k, fall with decrease in pressure, showing that k-, is in a fall-off region, just as k , is, in the pressure range investigated.These results are examined in more detail in Part 2.11 DISCUSSION OVERALL MECHANISMS OF THE CATALYSED DECOMPOSITIONS The experimental results confirm that the propagating steps in the early stages of the catalysed decompositions are (1)-(3), with reaction ( - 2) affecting the rates later in each run or when VC is added to initial mixtures. The changes in rates and rate laws in vessel B as it aged, and differences from vessel to vessel, point to changes and differences in the control of the chain-centre concentrations by the initiation and termination steps. Many overall rate laws, derived from different combinations of heterogeneous and homogeneous initiation and termination steps, were tested against the experimental results but did not prove very discriminating. Examination of chain-centre concentrations calculated from measured rates and the rate constants k,, kip) and k3 has revealed much more about the nature of the initiation and termination steps under different reaction conditions.ASHMORE, GARDNER, OWEN, SMITH AND SUTTON 67 1 CHAIN-CENTRE CONCENTRATIONS BASED ON RATE MEASUREMENTS THE RE LA T I VE CON C E N TR A TIONS [R],/[el], These depend solely on the propagating reactions (1)-(3) in the initial stages when VC is low and reaction (- 2) can be neglected.For steady rates they will adjust to give irrespective of the initiation and termination reactions. Taking the published date for k, and k, and using the data for kip) reported in Part 2,11 the ratio takes the value TABLE TH THE RATIO [kl0/[tl], FOR VARIOUS REACTION CONDITIONS 520 572 629 2 5 2 5 2 5 T/K PClZ, o/Torr p,,,, o/Torr 35 43 34 14 13 4.7 4.6 65 57 48 18 17 5.8 5.7 135 88 77 25 24 7.6 7.5 shown in table 6.It is therefore likely that termination reactions involving R will be more im ortant at lower temperatures, lower pel, and higher pDCE; and those in- volving 8 1 will be relatively more important at higher temperatures and lower pDCE. ABSOLUTE VALUES OF [el], AND [R], IN VESSEL B [el], was calculated from - (d[DCE]/dt),/k,[DCE], with1* log,,(k,/dm3 mol-l s-l) = 10.80 -(3100 K/4.576 T). Fig. 10 shows how [el], depends on [Cl,], at 572 K. In the fresh vessel [el], is proportional to [Cl,]o*5, is independent of [DCE],, and lies very close to [Cl],, = (&[c12]o)o*5. In the aged vessel [el], depends on a power > 0.5, and is smaller with higher values of [DCE],.Similar differences were found at other temperatures. log,,(k,/dm3 mol-l s-l) = 8.76-(920 K/4.576 T ) ; check calculations from (d[VC]/dt),/kip) confirmed the values. The left-hand lines in fig. 11 show that in the fresh vessel at 572 K [R], is proportional to [C1,]8.5 but is larger for higher values of [DCE],. The points show experimental results in the aged vessel at 572 K; [k], is then closely proportional to [C1,]8.5 (more accurately it fits a slightly higher power) and is clearly independent of [DCE], and the values are much lower than in the fresh vessel. Similar results were found at other temperatures. [R], was calculated from (d[TCE]/dt),/k,[Cl,], with8 DEDUCTIONS ABOUT INITIATION AND TERMINATION REACTIONS IN VESSEL B FRESH VESSEL B The simplest explanation of the observations summarised in fig.10 and 11 is that the initiation and termination reactions in fresh vessel B controlled [el],, especially at higher temperatures where [cl]/[R] is higher.672 CHLORINE - c AT ALY SED PYROLYSIS OF 1 ,2-C2H,Cl, FIG. 10.-Plots of [el],, and [el], against (pcl /T~rr)$.~ for the fresh vessel B (0) at all PDCE,O and for the aged vessel B at pDCE.0 equal to x , 35; A, 65 and 0, 135 Torr, 572 K. 36r 6' 18 16 I4 I2 10 8 6 4 2 FIG. 11.-Plots of [k], against (pclo/Torr)~.5 at 572 K for the fresh vessel withpDCE,o x x , 35; A, 65 and 0, 135 Torr (left-hand axis); and for the aged vessel B with pDCE.0 x +, 25; x , 35; A, 65 and 0, 135 Torr (right-hand axis).ASHMORE, GARDNER, OWEN, SMITH AND SUTTON 673 Since [ello is proportional to [C1,]8.5 and close to [el],, = d(K,,[Cl,],), the gas-phase dissociation and recombination steps (d) and (r) must be examined as candidates for the initiation and termination reactions d r C1, + M e 2el+ M, There are several objections to this assignment. It has been shown15 previously that the attainment of 90% of [el],, when C1, is admitted to vessels at 650 K would take more than 100 s by homogeneous dissociation of Cl,(qcl, = 20 Torr, ptotal = 760 Torr).It would take longer with ptotal z 100 Torr, especially at lower temperatures, but no induction periods were observed in the catalysed pyrolysis. Reaction (r) is much slower than the termination step (t) t c l + R + RCl (9 for which the rateconstant has been estimatedls as dm3 mol-1 s-l; experimentally measured values for e l + other chloroethylene radicals are lower ( 1010.9 dm3 mol-l s-l for CHCl,CCl, and 10ll.O dm3 mol-l s-l for C,C15).A recent experimental meas~rementl~ of the rate constant for path (r) suggests a maximum value of 5 x lo9 dm6 rnol-, s-l at 572 K. Combining these rate constants with typical values of [ell0 and [k], for fresh-surfaced vessels at 572 K shows that reaction (t) is > lo4 times as effective as reaction (r) for all reactant pressures used. Similarly, calculations from published data1* on reaction (d) show that it produces chlorine atoms at a much slower rate than they would be removed by reaction (t) were it operative under those conditions.We therefore reject reactions (d) and (r) as effective initiation and termination reactions. Another pair of reactions that could result in [el], = [el],, are reactions of chlorine molecules and atoms with surface sites S such as the pair (i,s) that would be balanced at true equilibrium (from the principle of microscopic reversibility) i c1, + s * SCl +el. s Collision theory shows that surface removal of el in vessel B would have to have an efficiency &(el) > to compete with reaction (t) at the calculated levels of [el] and of [A]. The lowest measured19 limit of &(el) is ca. on freshly acid-washed walls (conventionally described as 'poisoned' for the removal of el). After heating to 100 OC, the efficiency rises to ca. 3 x on salt-coated or flamed Pyrex surfaces.It therefore seems likely that the surface removal can compete successfully with gas-termination steps in the vessel B under the conditions used by J.W.G., although it is unlikely that it completely overwhelms step (t). There is a good deal of evidence from previous studies of temperature distribution in vessels where thermal chlorinations are taking place that the reaction takes place close to the walls,,O with chains initiated and terminated on the walls. If we assume that [el], M (KD[C1,]o)o.5 through the reversible pair (i, s) then for low [Cl,], in fresh B and is higher (> This describes the observed kinetics. It also suggests the overall activation energy674 CHLORINE-CATALYSED PYROLYSIS OF 1,2-C2H,C1, should be close to El +AH/2 where AH is the enthalpy change of the reaction c12 (8) f 2 a (g).As El = 3 kcal mol-1 and AH = 58 kcal mol-l, the predicted overall activation energy is 3 + 29 = 32 kcal mol-l. J.W.G.'s resultsg gave 33 kcal mol-1 for the fresh vessel B. Stationary-state treatment also gives k [DCEI, [ell0 k l [ ~ ~ ~ l , (G [cl,~,)~ * 50 k2 x [RIo = k , + k3[C1,], because k, % k,[Cl,], under the experimental conditions. If this is correct, [R],/[C1,]8.50 should be proportional to [DCE]8.38 because experi- mentally over this pressure range it was found that k , was proportional to [DCE];as2. The slopes of the left-hand lines in fig. 11 should then be proportional to [DCE];-38. They are closely so, as shown in table 7. TABLE 7.-DEPENDENCE OF [R], ON PDCE (aged vessel B, 572 K) 135 65 35 22.5 16.8 13.2 3.49 3.44 3.42 Adding inert gases would increase k,, but if [el], is fixed [k], must fall proportionately from the above equation.Hence (d[VC]/dt) = k,[k], remains constant, as found9 experimentally for the fresh vessel B. AGED VESSEL B In the aged vessel B, the experimental evidence shows that at the same temperature the rates are lower than in the fresh vessel and centre concentrations are correspondingly lower by factors of 4-8. This further favours a first-order surface termination against a second-order gas-phase termination. In the aged vessel the addition of inert gases increases (d[VC]/dt), (table 2) but not (d[TCE]/dt),. The increase in (d[VC]/dt), = k,[R], is entirely attributable to the increase in k,, as [k], = (d[TCE]/dt),/k,[Cl,], remains constant.This points to control of [k] by the initiating and terminating steps in the aged vessel B. Calculations of the rates of the gas-phase termination reactionslG and R+R+R2 R + e l + RCl show that in the aged vessel B surface removal of R should be more effective provided ~ ( k ) b 2 x which is very likely. The surface removal of k by the reaction SCl+R-+ RCl+S (s') combined with the initiation step (i) and the equation 6 = K[Cl,]0.5/( 1 + K[C1,]0-5)ASHMORE, GARDNER, OWEN, SMITH A N D SUTTON 675 where K is the Langmuir coefficient, for the fraction 8 of sites occupied by C1 leads [R], = ki[Cl,]~.5/k,, K. to the prediction Thus [R], is proportional to [C1,]8.5 and is independent of [DCE], as required by the experimental results in fig. 11. If [R], is controlled, then by eqn (IX) [el], has to follow Recalling that k , cc [DCE]o.S2 and that k3[C1,], < k , when [Cl,], is low, the expression for [el],, correctly describes the shape of the curves through the experimental points x , A, and 0 in fig.10, and their slopes at fixed [Cl,], are closely proportional to [DCE]-o.38. Other assumptions about surface termination reactions, such as removal of R by empty sites S, do not lead to the correct relationships for [R], and [el],. Accepting path (i) as the initiating step and path (s') as the terminating step, (d [ VC] / d t)$P) = k gP) ki [ Cl ,] 8.5 / k, I K and the overall activation energy is given by Eoverall = B2P) + Ei - E,, - AH' where AH', the enthalpy change of the reaction s + 3c1, e SCl is equal to +D(cl-cl) -D(s-cl). Ei is not less than the enthalpy change of the reaction (i) which is D~cl-cl)-D~s-cl). E, is probably small.Therefore Eoverall w EP) + D(cl-cl) - D(s-cl) - P ( c l - C l ) + D(s-cl) w EP) + iD(c1-a)- Now D(cl-cl) M 58 kcal mol-1 and EgP) w 17 kcal mol-l at p w 100 Torr (see eqn (VI), with E3 w 0.9 kcal mol-l]. Therefore Eoverall M 46 kcal mol-l. This predicted value compares very favourably with the experimental value for the aged vessel B of 47 kcal mob1 (A.J.O.). CONCLUSIONS FOR VESSEL B We therefore have self-consistent and reasonably quantitative interpretations of the experimental results for the chlorine-catalysed pyrolysis in the fresh-surface vessel B and in the aged vessel B, on the basis that the chain-termination reaction in the fresh vessel is essentially removal of el on the surface, while in aged B it is removal of k on the surface.This change-over is helped, undoubtedly, by the relatively higher values of [k]/[el] at the lower temperatures used for many of the experiments in the aged vessel. Unfortunately, we lack information about the ageing process that might point to chemical reasons for the aged quartz surface to favour reaction (s') with k rather than (s) with e l , and the fresh quartz surface to favour reaction with el rather than with R.676 CHLORINE-CA TALY SED PYROLYSIS OF 1 ,2-C,H4Cl, INITIATION A N D TERMINATION REACTIONS IN OTHER VES.SELS It is not possible to provide a satisfactory explanation of the limiting rate shown in fig. 2 for the quartz vessel A, s/v = 1.4 cm-l, as pclp is increased. Inclusion of the gas-phase termination step (t) as well as the surface termination, which would be reduced at the lower s/v, makes the rate less dependent on pclz than one-half order, and this effect increases as pclz and the radical concentrations increase.However, detailed analysis shows that including (t) makes the order in pDCE lower than the observed unity. In the uncoated Pyrex vessel C , with s/v = 1.4 cm-l, [el] and [R] behaved like the results found for fresh vessel B, with some evidence for gas-phase as well as surface termination. After coating with Teflon the concentrations fell to even lower levels than in aged vessel B. This suggests that coating with Teflon, and to a less extent ageing in B, might reduce the rate of initiation through blocking chemisorption sites for Cl,, while retaining the ability of the walls to adsorb or remove R. D. H. R. Barton, J. Chem. SOC., 1949, 148. D. H. R. Barton and K. E. Howlett, J. Chem. SOC., 1949, 155. K. E. Howlett, Trans. Faraday SOC., 1952, 48, 25. G. A. Kapralova and N. N. Semenov, Russ. J. Phys. Chem. (Engl. Transl.), 1963, 37, 35, 156, 258. K. A. Holbrook, R. W. Walker and W. R. Watson, J. Chem. SOC. B, 1968, 1089. K. A. Holbrook, R. W. Walker and W. R. Watson, J. Chem. SOC. B, 1971, 577. T. Takahashi, T. Abe, Y. Migkoshi and S. Asano, Kogyo Kagaka Zasshi, 1968, 71, 504. 1972, 81, 65. F. S. Dainton, D. A. Lomax and M. Weston, Trans. Faraday SOC., 1962, 58, 308. J. W. Gardner, Thesis (University of Manchester, 1975). (This includes a full summary of earlier work on the catalysed and uncatalysed decompositions). lo P. G. Ashmore, J. W. Gardner and P. Sutton, Chem. Wetenschap, Belgische Chemische Industrie, June 1973, p. 11 (abstract of paper, Third International Symposium on Gas Kinetics, Brussels, 1973). l1 P. G. Ashmore, A. J. Owen and P. J. Robinson, J. Chem. SOC., Faraday Trans. 1, 1982, 78, 677. l2 P. J. Robinson, Trans. Faraday Soc., 1965, 61, 1655. l3 P. J. Robinson and K. A. Holbrook, Unimolecular Reactions (Wiley Interscience, New York, 1972), l4 C. Cillien, P. Goldfinger, G. Huybrechts and G. Martens, Trans. Faraday SOC., 1967, 63, 1631. l5 S. W. Benson and J. H. Buss, J. Chem Phys., 1957, 27, 301. l7 M. A. A. Clyne and D. H. Stedman, Trans. Faraday SOC., 1968, 64, 2698. la R. A. Carabetta and H. N. Palmer, J. Chem. Phys., 1967, 46, 1333. l9 P. G. Ashmore, A. J. Parker and D. E. Stearne, Trans. Faraday SOC., 1971, 67, 3081. 2o G. A. Kapralova and N. N. Semenov, Russ. J. Phys. Chem. (Engl. Transl.), 1963, 37, 35, 156, 258. 'I G. Huybrechts, J. Katihabwa, G. Martens, M. Nejszaten and J. Olbregts, Bull. SOC. Chim. Belg., pp. 262-263. G. Chiltz, P. Goldfinger, G. Huybrechts, G. Martens and G. Verbeke, Chem. Rev., 1963, 63, 355. (PAPER O/ 1986)