摘要:

Selective Adsorption of Methyl Esters of n-Fatty Acids at the Silica/Benzene and Silica/Carbon Tetrachloride Interface Part 1 .-Adsorption Isotherms BY A. K. MILLS? AND JOHN A. H0CKEY"T Chemistry Department, U.M.I.S.T., P.O. Box 88, Manchester, M60 1 QD Received 17th February, 1975 The adsorption isotherms of a series of n-fatty acid methyl esters at the silicalbenzene and silica/ carbon tetrachloride interfaces have been determined. For the benzene solutions both the selectivity and the limiting adsorption decrease with increasing chain length of the ester. Based on an analysis of literature data on the enthalpies and entropies of mixing of hydrocarbons in benzene, it is sug- gested that for the benzene solutions the adsorption limit is set by the presence at the interface of an adsorbed solvated layer.The limiting adsorption from benzene at 25°C corresponds, in terms of the interfacial molecular concentration of the ester, to approximately one third of the adsorbent's isolated surface silanol groups. Increasing the adsorption temperature from 25 to 50°C increases the limiting adsorption from the benzene solutions and it is suggested that this increase is due to a decrease in the solvation or " ordering " of the adsorbed layer as the temperature is raised. Re- moval of the hydrogen bonded silanol groups from the adsorbent's surface by prior heating produces no measurable change in the adsorption properties from either solvent and this is taken as support for the thesis that the adsorbent's free hydroxyl groups are the adsorption centres. The adsorption isotherms from carbon tetrachloride demonstrate a greater selectivity in the adsorption process than there is from the benzene solutions and also that the limiting adsorptions are close, in terms of the interfacial molecular concentration of the adsorbate, to the interfacial concentration of the adsorbent's free surface silanol groups.In a previous account of the selective adsorption of a series of n-fatty acids at the silica/benzene interface, it was suggested that the adsorption limit was con- trolled by the association of the solute in the bulk solution and not by the properties of the interfacial phase. Generally, such solution effects can be unfortunate since more information can be gained about the surface structure of the solid and the distribution of the adsorbate on it if the limit is controlled by these interfacial para- meters.As the aggregation observed in the fatty acid system was due to inter- carboxyl hydrogen bonding, in the present study we have used the methyl esters of some n-fatty acids. The object of the present work was to determine the factors that control selective adsorption in such systems and to use the results to obtain a more realistic model for the surface structure of the silica than has been possible hitherto. We have divided our account into three parts. The first describes the determination of the adsorption data, the second the measurement of the heats of adsorption from solution and the third the results of the i.r. spectroscopic studies carried out on the system.EXPERIMENTAL MATERIALS 500 g samples of each of the methyl esters were obtained from Fluka A.G. who stated that they were at least 99 % pure. They were stored under dry nitrogen in dark bottles which t present address : Unilever Research, Port Sunlight Laboratory, Port Sunlight, Wirral, Cheshire, L62 4XN. 2384A . K . MILLS AND J . A . HOCKEY 2385 were opened only for the minimum time necessary.' The corresponding radiotracer I4C labelled n-fatty acids (total activity per sample 21 0.1 mC) were obtained from the Radio- chemical Centre, Amersham and each dissolved in 3 cm3 of dry benzene. After dilution with 500 mg of the corresponding unlabelled acid (ex Fluka A.G. 2 99 % pure) the acids were converted into their methyl esters by reaction with CH2N2,3 generated in situ from p-tolylsulphonylmethylnitrosamide (Koch Eight).The solvent was evaporated and after confirmation by g.1.c. that quantitative conversion had been achieved the labelled samples were added to 250-cm3 portions of the corresponding unlabelled esters. Benzene was puri- fied by the method described previ~usly.~ Carbon tetrachloride (AnalaR grade) was dried by reflux over P205 and distilled from this reagent just prior to use. Water interferes with selective adsorption in systems of the present type and so each of the solvents and adsorbates was analysed for water by the Karl Fischer method. The moisture contents of the solvents were below the detection limit and each of the esters contained GO.03 % w/w. Infrared spectroscopic analysis of the esters also showed that they contained insignificant (GO.05 %) amounts of any other hydroxylic compounds (e.g.alcohols, fatty acids). The preparation of the silicas RA3, RA3/700 and RRA3/700 has been described previous- I Y . ~ Their surface hydroxylation, specific surface areas (RA3 = 170 m2 g-l, FL43/700 and RRA3/700 = 150 m2 g-' by B.E.T. method using N2 at 77 K) and heats of immersion have been studied in RA4/700 and RRA4/700 were prepared by similar methods to those referred to above, using a separate batch of Degussa Aerosil as starting material. The specific surface area, again as determined by the B.E.T. method using nitrogen at 77 K, was 154 m2 g-' for each of the series-4 adsorbents and their surface hydroxylation was essentially similar to that of their series-3 analogues.The unannealed silica (RA3) contained small volume of pores of molecular dimensions, some bulk hydroxyl groups and both hydro- gen bonded and non-hydrogen bonded surface hydroxyl groups. Annealing at 700°C in air for 48 €1 removed the bulk and hydrogen bonded surface hydroxyl groups and virtually all the micropores (e.g. RA4/700). Rehydroxylation of the annealed material in liquid water regenerates the hydrogen bonded surface hydroxyl groups but produces only insignificant bulk rehydroxylation and microporosity (e.g., RRA4/700). METHODS With the exception of the analytical method the experimental procedure used for the determination of the adsorption isotherms has been described previously in detail.' As before, the exclusion of moisture was of paramount importance.The initial solutions were prepared by a dilution procedure. About 10 g of the ester were transferred under dry N2 into a tared tube via its disassembled greaseless tap. After closure the tube was weighed, evacuated and filled with dry solvent from a reservoir. Further dilution in a similar manner gave the required solutions of known molality. About 1 g of the adsorbent powder was heated at 120°C in air to constant weight in a tared Pyrex bulb, which was then blown onto a vacuum line, evacuated at ambient temperature overnight at a pressure < 1.3 x N m-2 and sealed under vacuum. The bulb was loaded into a glass tube of about 40 cm3 capacity which was then sealed with a greaseless tap. The tube was evacuated, weighed and filled with about 25cm3 of solution.After it was reweighed the sample bulb was broken and the tube shaken overnight in a water bath controlled to )O.Ol"C. The tube was left immersed for the adsorbent to settle and a weighed quantity of the clear supernatant solution withdrawn for analysis. The solutions were counted in a 16-cm3 coated liquid source container (type N671A Ecko Electronics) by means of a liquid scintillation counter (N664A Ecko Electronics) coupled to an automatic scalar (N530F Ecko Electronics). For the benzene solutions about 2 cm3 of the active solution was weighed into a source container containing an excess ( - 6 cm3) of liquid scintillator solution (N641, Nuclear Enterprises Ltd.). For the most dilute solutions the time taken to record 30 OOO counts was taken as a measure of the activity and for the remainder 100 000 counts were recorded.Associated random errors were 0.6 % and 0.3 % respectively. Since CCI, is a very efficient scintillation quenching agent it was necessary to modify the procedure for the tetrahalide solutions. 2 cm3 of the supernatant CCI, solution were weighed into a 100-cm3 conical flask and the solvent removed at 0°C by evacuation with2386 a water pump. 5 cm3 of benzene were then weighed in and 2 cfn3 of the resulting solution counted as before. ADSORPTION AT SiOz/C6& AND sioz/cc& RESULTS AND DISCUSSION The excess interfacial concentration of component 1 (solute) was obtained by direct use of eqn (l), applicable to selective adsorption from dilute solution : where n; is the number of moles of solute adsorbed per grain of the adsorbent, m is the weight of the adsorbent in grams, o is the total weight of the solution phase and Ac is the solute molality change caused by adsorption.The adsorption isotherms are plotted as nt against the corresponding equilibrium solution molalities.y Fig. 1 shows the results obtained with the annealed adsorbents at 25°C. Both the selectivity of the adsorption process and the limiting adsorption decrease with increasing chain length of the ester. As it is known that saturated hydrocarbons do not displace benzene from hydroxylated silicas, it seems unlikely that the ester’s hydrocarbon chains are “ adsorbed ” onto the adsorbent surface.l0 It follows that there are two possible explanations for these effects. Either there is a change in the “ solvation ” of the ester’s alkyl chain as it passes from the solution into the interfacial phase or it carries with it into the interface a quantity of solvent sufficient to ‘‘ concentrate ” the equilibrium solution and so reduce the measured change in Ac.n: = mAc/103 rn (1) e m 0 rn m L 0 10 0 2 0 0 3 0 0 solution concentration x 103/mol kg-I (4 I v u v I00 2 0 0 30C 400 solution concentration x 103/mol kg-I RG. I.-Adsorption of methyl esters from benzene solutions onto (a) RRA3/700 at 25°C and (b) (b) RA31700. 0, methyl n-hexanoate ; (>, methyl n-decanoate ; 0, methyl n-tetradecanoate. The latter effect has been suggested by Wright and it is worthwhile to consider whether such a phenomenon could explain the chain-length dependence observed in the present work.From fig. l(a) it can be seen that the limiting value of n; (nS, Lim) for the c6 ester is of the order of 1.2 x Using eqn (l), for a system mol g-l.A . K . MILLS AND J . A . HOCKEY 2387 containing about 20 g of solution and 1 g of adsorbent, the typical experimental conditions in the present work, we find that the corresponding Ac is equal to 6 x mol kg-l. Similarly the decrease in Ac corresponding to a decrease in ni Lim of 2.6 x mol g-l, the order of the difference in n9, Lim between the c6 and C14 esters, is equal to 1.3 x mol kg-I. This corresponds to a change in the equilibrium solu- tion molality of about 1.3 o/o, since the limiting adsorption is reached at equilibrium molalities of about 10-1 mol kg-l. To produce such a change on the basis of Wright’s hypothesis * would require about 1.3 % of the solvent (3.25 x loA3 mol C6H6) to be associated with 9.6 x mol -CH2- in the interfacial phase.For ni = 1.2 x mol g-1 on a solid of specific surface area of 150 m2 g-I, it is easy to show that the molar volumes of C6H6 and -CH2-- will only allow for this solvent to solute ratio if the interfacial phase is a multimolecular layer or if it is of a form in which the solute is adsorbed directly onto the adsorbent but is separated from the equilibrium solution by a layer of pure solvent. As will be shown in a subsequent paper, spectro- scopic evidence precludes the first possibility and the second seems improbable. A more plausible explanation may be based on the possibility that the solvation of the hydrocarbon chain of the solute changes on adsorption.The observed effect in the prelimiting regions of fig. 1 is that the adsorption becomes less selective as the hydrocarbon chain length increases. That is to say, that the equilibrium constant that governs the partitioning of the ester between the solution (initial state) and inter- facial (final state) phases becomes smaller with increasing solute chain length. (i.e., 8 x 1.2 x Thus, where AGO is the standard free energy change on adsorption, E is the interfacial area and nCH2 is the number of CH, groups in the ester’s chain. Jones et aL9 have measured the heats of mixing of a range of n-paraffins with C6& and from their data it is possible to show that the molar enthalpy of dilution (AHd) of hydrocarbons in benzene is positive and increases with increasing hydrocarbon chain length, with a methylenic increment in AHd of about 100 J mol-l (of CH,) (mol kg-’)-l.It follows since by definition adsorption involves a concentration increase of the solute on pass- ing from the solution into the interface, that the enthalpy contribution to the free energy term in eqn (2) is small but negative. This requires the entropy contribution also to be negative, which may imply that the hydrocarbon chain of the solute is more ‘‘ ordered ” in the interfacial phase than in solution. This is an interesting result since it provides some evidence for the presence of an “ ordered ” hydrocarbon phase next to the solid’s surface. The limiting n: values shown in fig. 1 correspond to about 0.5 solute molecules nm-2.Close packed hydrocarbon chains would have a density about 8-9 times greater than this. The results of the infrared measurements and calorimetric studies carried out on this system lo confirm the expected result that the hydrocarbon chains of the solute do not displace the aromatic solvent from the interface and therefore are not lying flat on the surface. That is to say, the interfacial phase contains solvent and the most likely model for it is that of ester molecules hydrogen bonded via their head group to isolated surface silanol groups, the removal of the hydrogen bonded silanol groups does not affect the adsorption isotherm, with their solvated hydro- carbon chains forming an ordered layer, the structural coherence of which is respon- sible for the limiting values of n9, being lower than that corresponding to utilisatioh of all the isolated silanol groups as adsorption sites.The suggestion of such an ordered layer in the region of a solid surface is not2388 novel and similar effects have been demonstrated by Everett and Findenegg." Essentially, they have shown that the interfacial layer becomes less rigid the higher the experimental temperature is above the melting point of the liquid phase. In view of this we have also studied the selective adsorption of some of the esters at 50°C. The data shown in fig. 2 and 3 are less precise than those in fig. l(a), owing to the difficulty of sampling at these elevated temperatures at which the solvent is more volatile. In the time available we were only able to obtain a limited quantity of data at the higher temperature.However, it is quite clear that the n; Lim value is markedly increased. The increase is not a function of the solid phase, since the value of n; Lim corresponds to 1.25 0.05 molecules nm-2 for each of the solids. This is very close to the concentration of the isolated surface silanol groups. Also, as at 25"C, removing the hydrogen bonded surface groups has no effect on the adsorption. There is no positive evidence that adsorption in the prelimiting region is dependent on the ester's chain, although better data than are presented here would be necessary for such a conclusion to be certain. Nonetheless, it is apparent that raising the temperature markedly alters the composition and therefore the structure of the interface. 0 I00 2 0 0 3 0 0 4 00 5 0 0 solution concentration x 103/mol kg-' FIG.2.-Adsorption of methyl esters from benzenc solutions at 50°C. RA3/700 ; 0, methyl n-decanoate onto RRA3/700 ; @, methyl n-tetradecanoate onto RRA3/700. cb, methyl n-decanoate onto It is pertinent here to consider one other possible explanation for the effect of temperature on the system. It is well known that the presence of other polar com- pounds, especially water, in such systems markedly affects the adsorption equilibrium. For example, if the ester solutions contained water, then at 25°C the n; Lim value could correspond to a situation in which the water was very strongly adsorbed onto a fraction of the surface silanol groups, " blocking " them as possible adsorption sites for the ester.Raising the temperature to 50°C may be expected to have a rela- tively large effect on the water adsorption, so decreasing its equilibrium surface cover- age and enabling the fractional surface coverage of the ester to increase. Calculations based on the known moisture contents and the infrared studies lo show that this explanation is unsatisfactory.A . K. MILLS AND J . A . HOCKEY 2389 Fig. 4 considered along with fig. l(b) illustrates that the presence of micropores in the adsorbent enhances the adsorption and also reduces the dependence of ni Lim on chain length. Possibly, the micropores prevent the formation of the solvent/ hydrocarbon chain structure by imposing a spatial limitation around the chain.The Y 0 I00 2 0 0 3 0 0 4 0 0 solution concentration x 103/mol kg-l FIG. 3.-Adsorption of methyl decanoate from benzene solutions onto RRA4/700 at 25°C (e) and 50°C (0). results are interesting, since they show how the presence of pores of molecular dimen- sions can perturb the adsorption behaviour in such a way as to obscure the presence of other controlling parameters in such systems. !OO 2 0 0 3 0 0 4 0 0 solution concentration x 103/m01 kg-' FIG. 4.-Adsorption of methyl esters from benzene solution onto RA3 at 25°C. 0, (>, as fig. I . Fig. 5 shows the effect of temperature on the adsorption of lauric acid from benzene solutions onto RRA3/700. and the low value of ni Lim, which is very close to that recorded for the methyl esters at 25"C, was thought to be due to solute aggregation as described in the introduction.The present results clearly cast doubt on this explanation and therefore we have The results at 27°C have been reported previously2390 ADSORPTION AT SiOz/C6H6 AND SiO2/CCl4 determined the selective adsorption isotherm of this acid from benzene at 50°C. The limiting adsorption is higher at the higher temperature, although the increase is con- siderably less than that observed for the esters. Since for long-chain adsorbates it is reasonable to expect any effect due to the hydrocarbon chain to be independent of the head group, it seems likely that the increased fatty acid adsorption results from a system in which adsorption is " aggregation controlled " but the aggregation is more dependent on temperature than is the monomer adsorption.V 0 100 2 0 0 3 0 0 4 0 0 solution concentration x 103/mol kg-' FIG. 5.-Adsorption of lauric acid from benzene onto RA3 at 27°C (0) and 50°C (0). The adsorption isotherms for adsorption of the Clo and C14 esters onto the non- microporous RA4/700 and RRA4/700 from CC14 at 25°C are shown in fig. 6. Within the accuracy of the experimental data the adsorption is independent of the ester chain length and unaffected by the removal of the adsorbent's hydrogen bonded surface 2 2 2 X m - I I 0 5 0 100 150 solution concentration x 103/mol kg-' FIG. 6.-Adsorption of methyl esters from carbon tetrachloride solutions at 25°C onto RA4/700 (@) and RRA4/700 (0).A . R . MILLS AND J . A . HOCKEY 239 1 hydroxyl groups. The adsorption limits correspond closely to the concentration of the isolated surface silanol groups and presumably from this solvent they are set by the availability of these surface groups rather than by solvation phenomena.The adsorption limits observed with this system are too great to correspond to an inter- facial structure in which the hydrocarbon chains are parallel to the adsorbent surface. The prelimiting regions of the isotherms show that the selectivity which causes the surface to adsorb the ester more readily than it adsorbs the solvent is much greater for the ester/CCl, than for the ester/benzene system. This expected result occurs partly because the adsorbent's hydroxyl groups prefer the aromatic hydrocarbon to the apolar tetrahalide. We thank Unilever Limited for financial support of this work, for a personal grant to A. K. M. and also for permission to publish this work. C. G. Arrnistead, A. J. Tyler and J. A. Hockey, Trans. Faraday SOC., 1971, 67, 493. C. G. Armistead, A. J. Tyler and J. A. Hockey, Trans. Faraday. SOC, 1971, 67, 500. A. I. Vogel, Practical Organic Chemistry (Longman, London, 3rd edn.), p. 97. J. A. Hockey and €3. A. Pethica, Trans. Faraday Suc., 1962, 58,2017. A. J. Tyler, F. H. Hatnbleton and J. A. Hockey, J. Catalysis, 1969, 13, 35, 52. C. G. Armistead, A. J. Tyler, F. H. Harnbleton, S. A. Mitchell and J. A. Hockey, J. Phys. Chem., 1969,73,3947. ' J. J. Kipling, Adsorption from Solution of Nun-electrolytes (Academic Press, New York, 1965). E. H. M. Wright, Trans. Faraday SOC., 1965, 61, 1764. H. K. D. Jones, D. P. L. Poon, R. F. Lama and B. G. Y. Lu, Canad. J. Clzem. En@., 1967,45, 22. n. H. Everett and G. H. Findenegg, J. Chem. Thermodynamics, 1969, 1, 573, lo A. K. Mills and J. A. Hockey, J.C.S. Faraday I , 1975, 71,2392, 2398.

ISSN:0300-9599    

DOI:10.1039/F19757102384

出版商:RSC    

年代:1975

数据来源: RSC

摘要:

Selective Adsorption of Methyl Esters of n-Fatty Acids at the Silica/Benzene and Silica/Carbon Tetrachloride Interface Part 2.--Heats of Adsorption BY A. K. MILLS? AND JOHN A. HOCKEY*? Chemistry Department, U.M.I.S.T., P.O. Box 88, Manchester, M60 1 QD Received 17th February, 1975 The heats of adsorption of a series of n-fatty acid methyl esters at the silica/benzene and silica/ tetrachloride interfaces have been determined calorimetrically. The heats of solution of the esters have also been determined in CsH6 and CCl,. In all cases the heat of immersion increases as the solution concentration increases, until the limiting adsorption value is attained. Thereafter, the heat of immersion remains constant. The preferential molar heats of adsorption from both solutions are independent of the surface coverage of the adsorbate.The heats of adsorption at 25°C from the C6Hs solutions show no detectable dependence on the chain length of the ester. A similar lack of dependence on chain length is observed at 50°C, although there is a decrease in the value of the heats of immersion. This latter is in agreement with the hypothesis that there is a nett decrease in order of adsorbed layer as the adsorption temperature is increased. Essentially similar results were found with the CC14 solutions at 25"C, although the values of the preferential molar heats of adsorption were higher from this particular solvent. For both solvents removal of the H-bonded surface silanol groups by prior heating of the adsorbent caused no significant change in the mcasured heats of adsorption, which suggests that the adsorption centres are the isolated surface silanol groups of the adsorbent.The determination and interpretation of heats of adsorption from binary liquid solutions from which one of the components is selectively adsorbed have been con- sidered in detail by previous worker^.^'^ Possibly the most important result to emerge is that equations of the Clausius-Clapeyron type are not generally applicable to adsorption at the solid/liquid interface. The isosteric heat of adsorption ( A X ) of component 1 from binary solutions is given by an equation of the form : AH1 = --RT2( d F) In a: p,n? .n: , A where a: is the activity of component 1 in the bulk solutions which is in equilibrium with an interfacial phase containing ni and ni moles of components 1 and 2 respec- tively, adsorbed on a solid of area A . For this equation to be applicable must be equal to zero.Although apparently obvious, the point is worth restating because, as the work of Corkill et aI.4 and Everett and Findenegg has shown and the results reported in the previous paper have supported, the structuring of the interfacial phase may exert a controlling influence on its composition. In additior since Everett and Findenegg have also shown that the effect of temperature on such systems may be quite large, the continuing indiscriminate use of the Clausius- t present address : Unilever Research Port Sunlight Laboratory, Port Sunlight, Wirral, Cheshire L62 4XN. 2392A . K . MILLS AND J , A . HOCKEY 2393 Clapeyron equation, to obtain heats of adsorption from adsorption isotherms which represent the interfacial excess of only one component, will produce only misleading results.A better method is to determine the variation of the heat of immersion (h,) of the solid at constant temperature with ni. As Corkhill et aL5 have shown, it is possible from such measurements to obtain useful information about the heat of adsorption, although even in this case, as will be shown below, it is necessary to make certain simplifying assumptions. In the present work the features of particular interest were the dependence of the heat of adsorption on the chain length of the solute and temperature. EXPERIMENTAL The materials used in this work have been described in the preceding paper.6 The heats of immersion were determined in a differential ~alorimeter,~ with silica samples of about 1.5 g in sealed Pyrex bulbs.Dry conditions were maintained thro~ghout.~ To determine the heats of adsorption from solution it was necessary to know the heats of solution of the solutes as a function of ~oncentration.~ For convenience both calorimeters were loaded with a sealed bulb containing a known weight of outgassed ester and another containing a known weight of the adsorbent. Breaking the first bulb gave the heat of solution and about 300 cm3 of a solution of known molality, breaking the second gave the heat of immersion of the solid at a known surface coverage. The usual correction terms for bulb breaking etc. were applied to the measured heats, which were liberated within 5 min in all cases. RESULTS AND DISCUSSION Fig.1 summarises the heats of solution. In benzene, ester dissolution is endo- thermic and, over the molality range covered in our experiments, is virtually indepen- dent of concentration. This is dissimilar to the results obtained previously7 with fatty acids, and there is no evidence in fig. 1 of the type of solute aggregation observed with the acid s01utes.~ The increment per methylene group of the heats of solution in benzene is in good agreement with that calculated from the heats of mixing data presented by Jones et a2.l0 At 50°C the heat of solution of the Clo ester in C6H6 is slightly less than that obtained for the same solute at 25°C but again shows little variation with concentration. The heats of solution of the esters in carbon tetra- chloride are slightly exothermic and again virtually invariant with concentration. In contrast with our observations with the benzene solutions there is no variation with the ester's chain length.This result with the halide solvent is in agreement with that reported by Dacre and Benson who have shown that the heats of solution of alcohols in CC14 are virtually independent of chain length. (Although alcohols are associated by hydrogen bonding in CC14 solutions at the high concentrations used by Dacre and Benson the association phenomenon does not invalidate the comparison made here.) The relatively small absolute values of the heats of solution in each of the solvents means that in the present work the heat term corresponding to the dilution of the supernatant solution in the calorimeter on breaking the adsorbent bulb was negligible, G0.3 J when compared to the measured heats of immersion of about 20 J.Fig. 2 shows a typical set of results and demonstrates that the heat of immersion rises with the equilibrium concentration of the bulk solution until surface saturation is reached and thereafter it remains constant. The relation between the heat of immersion and ni at 25°C for the series-3 silicas and benzene solutions are shown in fig. 3. The heats of immersion in the pure solvent are in agreement with those reported previously and show the expected trend with temperature. The net excess heat of adsorption, i.e., the heat of immersion, is greatest for the microporous RA3 and least for the singly hydroxylated RA3/700, in agreement with the results of gas/2394 solid studies.Within the limits of experimental error there is no detectable depen- dence of the heat of immersion on the solute's chain length. This suggests that there is little enthalpy contribution to the free energy terms that describe the variation of ADSORPTION AT Si02/C6H6 AND SiOa/CC14 \ - 0.8 solution concentration x 103/mol kg-' solution ; x , Clo ester in CCI4 at 25°C ; 0, CI4 ester in CC14 at 25°C. FIG. 1 .-The heats of solution of the esters in benzene and carbon tetrachloride. , esters in benzene the selectivity of the adsorption process with chain length and it supports the are- rnent in the previous paper that this variation is the result of entropy effects.The linear relation between the heat of immersion and the surface excess of the selectively adsorbed component is also significant. A similar linearity has been observed by Corkill et aL4 and Armistead et aL7 in their studies of selective adsorption at the solid/liquid interface. The former workers have shown that the most plausible f a 0 1 170- * _ 30 40 50 do i o 80' solution concentration x 103/mol kg-' FIG. 2.-The heat of immersion of RA4/700 in CCI+ 3s a function of equilibrium solution molality.A . K. MILLS AND J . A . HOCKEY 2395 explanation for the linearity is that it results from the displacement of solvent mole- cules from the interface either by the solute or by a species (e.g., a solvated solute molecule) the composition of which is invariant with surface coverage.Under these conditions the rate of change of the heat of immersion with n: gives the preferential heat of adsorption7 On this basis, the data presented in fig. 3 imply that the heat 180- 170- RA31700 00 701 I 0 0.2 0.4 0 . 6 0.8 1.0 1.2 ns, x 104/m01g-1 FIG. 3.-The heats of immersion of RA3/700, RA3 and RRA3/700, in benzene solutions at 25°C as a function of H;. A, CI4 ester ; x , Clo ester ; 0, C6 ester. of adsorption of the esters from benzene solutions is independent of u;. This reflects either that the fall off with surface coverage of the heats of adsorption of the ester and the aromatic compound are such as to maintain a constant difference between them, or that the contribution to the preferential heat of adsorption from the inter- action of the ester's head group with the surface silanol groups does decrease with n:, but the effect is compensated by an exothermic term which increases with n;.' 0 .4 0.8 1.2 1.6 2.0 2.4 2.8 3.2 ~l x 10"/molg-l FIG. 4.-The heats of immersion of RRA4/700 in methyl ndecanoate solutions at 25°C f x ) and 50°C ( 0 ) as a function of n:. A similar pattern of results is shown by the data in fig. 4 and 5, which illustrate the variation of hi with n; for the Clo ester at 25 and 50°C from benzene solutions and at 25°C from carbon tetrachloride. The linearity in fig. 5 is less certain than for the other cases because the selectivity of the adsorption is so great that it is difficult2396 ADSORPTION AT Si02/CsH6 AND Si02/CC14 to interpolate the values of n; with any greater precision than that shown.The values obtained from the data in fig. 3 for the preferential heats of adsorption from benzene at 25°C are 33.8 kJ mol-1 for both RA3/700 and RRA3/700. This is clear evidence that the non-hydrogen bonded surface silanol groups are the sole adsorption centres. I801 ~~~ ~~~ 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 n; x 104/m01 g-1 FIG. 5.-The heats of immersion of RA4/700 (0, 0 ) and RRA4/700 ( x , 0) in CC14 solutions of methyl n-decanoate ( x , 0) and methyl n-tetradecanoate (0, a) at 25°C. The microporosity of RA3 is reflected in the higher value of the heat of adsorption of 43.5 kJ mol-l. With the series-4 silicas the Clo ester has a slightly higher heat of adsorption than with the corresponding series-3 silicas.The reason for the small difference is not clear. Of more significance is the fact that the heat of adsorption from benzene is temperature independent, although at equivalent ni values the heat of immersion decreases by about 15 mJ m-2 as the temperature is increased from 25 to 50°C. This latter variation implies that the interfacial phase has a net excess positive specific heat analogous to the systems described by Clint et al. and Everett and Findenegg.s There are two possible explanations of such an effect in the present work. Either, as Everett and Findenegg suggest, the interfacial phase is significantly ordered at the lower temperature and the excess specific heat term can be ascribed to a pseudo melting phenomenon, or the composition of the interface is different at 50°C to that at 25"C, at equal nS, values.Either explanation is in accord with the postulate in the preceding paper that the value of n; Lim is set at 25°C by a structured inter- facial phase, whereas at 50°C the limit is perhaps set by the number of adsorption sites. There is certainly no evidence in the present work nor in the literature that suggests that the structure of an outgassed silica surface changes significantly with temperature between 25 and 50°C. As stated above the heats of adsorption from benzene are of the order of 30-35 kJmol-l, and from CCl, solution the Clo and CI4 esters adsorb with a heat of 52.3 kJ mol-I on both the fully and singly hydroxylated adsorbents. Considering that these are preferential heats, the values seem to be quite large.As explained above the heat of adsorption is virtually invariant with chain length, suggesting that the heat represents the difference between the localised interaction of single surface silanol groups with the C=O group of the solute and with the aromatic solvent. Thus it might be expected that the preferential heats of adsorption would correspond closely to the difference in the heats of adsorption of such compounds at the gas/solid interface. However, no literature data are available for the methyl esters of n-fattyA . K . MlLLS AND J . A . HOCKEY 2397 acids, and so we attempted a calculation based on the correlation between the spectro- scopic “ shift ” of the perturbed single surface silanol and the heat of adsorption.’ Unfortunately, such a calculation was not possible, although the correlation predicts a positive linear relation between the spectroscopic shift (Av) to lower frequencies of the absorption band of the silanol and the differential heat of adsorption (Qaas) at fixed coverage (0) of a range of adsorbates, for any one adsorbate Qads decreases with 8 whilst Av increases with 8.The somewhat obvious contradiction lessens the predic- tive value of this corre1ation.l A better relation may exist between Av and the net excess integral heat of adsorption. We thank Unilever Limited for generous financial and technical support, for a personal award to A. K. M. and also for permission to publish this work. D. J. Crisp, J. Colloid Sci., 1956, 11, 356. D. H. Everett, Trans. Furuday SOC., 1964,60, 1803. T. L. Hill, J. Chem. Phys., 1950, 18, 246. J. M. Corkill, J. F. Goodman and J. R. Tate, Trans. Faraduy Soc., 1966, 62, 939. D. H. Everett and G. H. Findenegg, J. Chenz. Thermodynamics, 1969, 1, 573. A. K. Mills and J. A. Hockey, J.C.S. Furuduy I, 1975,71,2384. A. J. Tyler, J. A. G. Taylor, B. A. Pethica and J. A. Hockey, Trans. Faruduy SOC., 1971,67,483. A. K. Mills, PhD. nesis (University of Manchester, 1972). lo H. K. D. Jones, D. P. L. Poon, R. F. Lama and B. G. Y. Lu, Canad. J. Chem. Eng., 1967,45, 22. l1 B. Dacre and G. C. Benson, Canad. J. Chem., 1963,41,278. l2 J. H. Clint, J. S. Clunie, J. F. Goodman and J. R. Tate, Nature, 1969, 51, 223. l3 L. H. Little, Infrured Spectra of Adsorbed Species (Academic Press, London, 1966), p. 285-293. ’ C. G. Armistead, A. J. Tyler and J. A. Hockey, Trans. Furaduy SOC., 1971, 67, 500.

ISSN:0300-9599    

DOI:10.1039/F19757102392

出版商:RSC    

年代:1975

数据来源: RSC

摘要:

Selective Adsorption of Methyl Esters on n-Fatty Acids at the Silica/Benzene and Silica/Carbon Tetrachloride Interface Part 3.-Infrared Studies and the Structure of the Solid Surface BY A. K. MILLS? AND JOHN A. HOCKEY*-/- Chemistry Department, U.M.I.S.T., P.O. Box 88, Manchester, M60 1 QD Received 17th February, 1975 A cell for studying the infrared absorption spectrum of species adsorbed at the solid/liquid interface is described. It has been used to study the spectra of a series of methyl esters of n-fatty acids at the silica/C6H, and silica/CCl, interfaces. The results show that independent of whether the solid surface carries only isolated surface silanol groups or these and H-bonded silanol groups it is only the former species which act as adsorption centres. From each solvent the adsorbed mole- cules are held by H-bonding between the ester's carbonyl group and the surface silanol group.The absorption band of the perturbed surface silanol group is at 3450 cm-', suggesting that the H-bonding is of a Iocalised form and does not include a contribution from solvent molecules. Infrared spectroscopy has been widely used for studying adsorption at the gas/ solid interface, where the most common major experimental difficulty, the scattering of the incident beam, has been overcome by the use of pressed discs.I At the solid/ liquid interface scattering is much less of a problem, since the ratio of the refractive indices of the disperse and dispersed phases is closer to unity than for gas/solid systems. However, other experimental difficulties, particularly those of solvent evaporation, equilibration and solvent absorption, have prevented infrared spectroscopy being widely applied to solid/liquid systems.The object of our studies was not only to obtain information about the nature of the bonding at the surface but also to develop a simple cell that would enable semi-quantitative spectroscopic information to be obtained more quickly and easily than has been possible hitherto. EXPERIMENTAL The materials used have been described in Part 1.2 All the spectra were recorded on a Perkin-Elmer 125 grating infrared spectrophotometer with a spectral slit width of < 2 cm-l. The cell shown in fig. 1 is a derivative of a previous design used for gas/solid ~ysfern.~ To study solid/liquid systems one must be able to outgas the solid, admit the liquid and contact it with the solid and record the spectrum.Outgassing is generally achieved by mounting the sample in position A,3 the liquid is then admitted through the Quickfit joint and the cell tiIted to immerse the sampIe completely. After equilibration the cell was mounted in a horizontal ~ r a d l e . ~ The liquid drained to the reservoir, although sufficient was used to cover the bottom of the cell body to a depth of about 2 mm. Thus, in its recording position the sample disc (trapped between two infrared transparent plates) was always in contact with the bulk liquid and remained wet by capillary action. The problem of solvent evapora- tion was far less severe than had been cxpected. This was because there was relatively little i.r.scattering with pressed silica discs, a low-intensity source could thus be used and heating of the sample by the infrared beam was therefore minimised. A rather fortunate and unforseen advantage of the cell was that drying out of the sample, even to a degree so 7 present address : Unilever Research Port Sunlight Laboratory, Port Sunlight, Wirral, Cheshire L62 4XN. 2398A. K. MILLS AND J . A . HOCKEY 2399 slight as to be difficult to detect by eye, was immediately apparent in the spectrum, because drying caused enhanced scattering and this led to a noticeable increase in the baseline absorption. / / c r i l s --- -1 L- END WINDOW 1-1QL110 RESERVOIR Fro. 1.-The infrared cell. In the present work it was possible to use an even simpler arrangement by mounting a silica disc at position B (fig.1). This was possible because not only do annealed pressed Aerosil discs have considerable mechanical strength, they can also be freed of molecular water by evacuation at ambient temperatures. It is also fortunate that if silica is heated at 70O0C, so as to leave only isdlated surface silanol groups, it may be handled in the ambient atmosphere without significant rehydroxylation of the surface. Consequently, in the present work it was nct necessary to heat any of the samples in situ in the cell. Indeed, in practice, because of the inherently high transparency of pressed Aerosil discs and the concommittant low scattering we were obliged to use extremely thin pressed discs of a " density " of the order of only 10mgcm-2.At position B the thin capillary film formed between the disc and the cell window held the disc in a vertical position and also retained it in the wet state throughout. of the discs used was of the order of 0.5 cm3 gm-I, so that at the ni Lim values observed in the present work about 90 % of the solute within the disc was in the interfacial phase. For this reason we did not balance out the solution absorption with a reference cell, although with samples having a much smaller surface area/pore volume ratio a balancing cell would be fiecessary. Three general points are noteworthy. The cell was cemented together with Araldite and it was necessary to curd it for about one week in order to eliminate any possible contami- nation of the solutions by the resin components.The O-ring seals and the diaphragm of the greaseless tap (fig. 1) were either Neoprene or Viton A. These swelled in use when ben- zene or CC14 were used as solvents, but the solutions were not contaminated, provided that seals and diaphra,gn were extracted with the pure solvent until they no longer coloured a fresh solvent sample. All the solverit transfer processes were carried out in a dry box and solutions were prepared as described previously. The pore volume RESULTS AND DISCUSSION Fig. 2 summarises the results obtained for the adsorption of the C6 ester from CC14 solutions onto RA4/700. The narrow absorption band at 3750 cm-l observed with the outgassed oxide corresponds to surface silaiiol groups which are not hydrogen bonded. The absence of broader bands in the 3500-3700 cm-I region shows that the surface carries very few hydrogen bonded silanol groups." Immersion " of the disc in CC14 slightly broadens the absorption band 5 * and shifts the absorption maximum to 3690 cm-l, a relatively small shift which reflects the weakness of the interaction between CC14 and the free surface silanol groups. As the ester concentration in the solution is increased the intensity of the 3690 cm-I absorption band decreases and a new broad band appears at 3425 cin-l. The absorption spectrum of the ester in CC14 solution exhibits a single absorption band at 1740 cm-l, which is associated with the 1-762400 ADSORPTION AT SiOz/C6Hs AND SiOz/CC14 carbonyl group. The spectra shown in fig. 2 illustrate that the carbonyl absorption band has split into a doublet centred about 1725 cm-l as the ester is adsorbed.The absorption bands corresponding to the methoxy and methylene groups of the ester are unchanged by the adsorption process. Taken together these results suggest that the ester is adsorbed on the surface by a hydrogen bonding interaction between the carbonyl group on the ester and the free surface silanol groups. The shift of the free 4000 3 5 0 0 3000 2) ;/crn-' (4 00 L I 2 0 0 0 1700 ;/cm-' (b) 0 FIG. 2.(a) and (b)-The infrared spectrum of RA4/700 at 25°C. ( I ) in vacuum, (2) immersed in a solution of 2.5 x lo-* mol kg-l methyl n-hexanoate in carbon tetrachloride.A. K . MILLS AND J . A . HOCKEY 2401 silanol absorption maximum to 3425 cm-l is in agreement with the observations of Fontana ' who has studied the adsorption of polyfunctional esters onto silica.It is also interesting to note that the absorption frequency of isolated surface silanol groups hydrogen bonded to methyl esters at the gas/solid interface (fig. 3) also exhibits its absorption maximum at 3425 cm-l. This agreement strongly suggests that the hydrogen bonding interaction between the carbonyl group of the ester and the surface silanol is not significantly perturbed by the presence of the solvent. At limiting adsorption of ester there is very little residual absorption at 3690 cm-l. This is in agreement with the adsorption data, which show that the limiting surface concentra- tion of the ester corresponds to virtually complete saturation of the surface silanol groups by the ester at a molar ratio of 1/1.2 Similar results were obtained with RRA4/700, again at limiting adsorption there is virtually no residual absorption at 3690 cm-I, and the absorption band associated with the free surface silanol groups hydrogen bonded to the ester is at 3425 cm-l.;oc '. 3 5 0 C 3/cm-' 00 FIG. 3.-The infrared spectrum of RA4/700 at 25°C (1) in vacuum, (2) in equilibrium with methyl n-hexanoate vapour. With the ester/benzene solutions at 25"C, the absorption band corresponding to the perturbed surface silaiiol groups is again at 3425 cm-l, and the absorption band corresponding to the carbonyl of the solute is also similar to that observed with the halide solvent. The similarity between these spectroscopic features in the two systems is clear evidence that the hydrogen bonding interaction between the adsorbent's silanol groups and the solute is "localised ", with no solvent interference. The spectroscopic shift (A? = 150 cm-l) of the surface silanol in pure benzene is bigger than that observed in CC14 (A? = 50 cm-l) because the aromatic molecule interacts more strongly with the surface silanol than does the alphatic halide.* The spectra obtained with the ester/benzene solutions show that at equilibrium solution concentra- tions greater than those corresponding to n; Lim there is a large residual absorption band at 3600 cm-l, which corresponds to surface silanol groups hydrogen bonded to2402 ADSORPTION AT Si02/C6H6 AND SiOa/CC14 benzene.The position of this band shows that the limiting adsorption from benzene solutions is not due to interference from other more strongly adsorbing molecules in the system, e.g., water, alcohols, etc.Also, although it is not possible to calculate precisely what proportion of the surface silanol groups is perturbed by the ester, it is clear that this is between 30 and 40 %, which corresponds very closely with the adsorption limits obtained from the isotherms.2 It follows that, although the value of n; Lim is much less than the corresponding single surface silanol concentration, the selectively adsorbed ester is bonded directly to the surface silanol groups and the surface excess concentration of the solute is not present in a multimolecular layer. Fig. 4 shows the results obtained with the ester/bemzene system at a higher tem- perature.The experiment giving these results was only qualitative, as it was carried out by equilibrating the loaded cell in an oven at 50°C and then recording the spectrum immediately after removing the cell from the oven. Despite the obvious uncertainties, it was clear that the absorption intensity at 3425 cm-I is greater than that observed at 25"C, whereas the absorption at 3600 cm-I is less at 50°C than it is at 25°C. These results are in general agreement with those presented in Part 1 ,2 which show that ester adsorption is enhanced when the temperature is increased from 25 to 50°C. 4000 3 5 0 0 i/cm- FIG. 4.-The infrared spectrum of RA4/700 at -50°C immersed in a 30 x lo-* mol kg-' solution of methyl n-hexanoate in benzene. Fig. 5 shows the results obtained with rehydroxylated RRA4/700 and the ester/ CC14 solutions.Once again, as the ester solution concentration is increased the absorption band at 3690 cm-l decreases in intensity with a concomitant increase in the absorption intensity at 3425 cm-'. Clearly, even on the rehydroxylated silicas it is still the free surface silanol groups which are the adsorption sites. The infrared technique also enabled us to check two other important features of the system. First, prolonged evacuation at room temperature of an equilibrated sample led to complete removal of both solvent and solute and restored the sample to its initial state. Secondly, we were also able to carry out experiments on deuterated samples. In these experiments the surface It follows that adsorption is reversible.A .K . MILLS AND J . A . HOCKEY 2403 silanol groups of the adsorbent were completely exchanged in situ with D20 vapour at room temperature prior to the solvent being admitted. The resulting spectrum showed that only a small fraction of the surface OD groups were exchanged back into the hydrogen form. The same volume of solvent (i.e., approximately 25 cm3) was used in both the spectroscopic and the adsorption isotherm experiments, whereas only about 10 mg of the adsorbent was used in the spectroscopic work compared with about 1.5 g of adsorbent in the adsorption experiments. This result indicates quite clearly that the moisture content of the dry solvent, which could not be determined by the Karl-Fischer method, was insignificantly small in the adsorption experiments. 4 0 0 0 3 5 0 0 30CO 2 5 0 0 i/cm-l kg-' solution of methyl n-hexanoate in benzene.FIG. 5.-The infrared spectrum of RRA4/700 at 25°C (1) in vacuum, (2) immersed in a 2.5 x mol The infrared experiments provide clear evidence for the hypothesis that in the benzene systems the adsorption is not limited by spatial factors associated with the packing of the head group on the adsorbents surface, since with the CC14 systems virtually all the single surface silanol groups are bonded to ester head groups. This latter result is of interest in a wider context because it gives some indication of the distribution of the free silanol groups on the adsorbent surface. Annealed silica samples of the type used in the present work have a total surface hydroxyl group concentration of approximately 4.6 per 100 A2 of su~face.~ Also, in silica the silicon atoms have 4-fold coordination.Consequently, surface silicon atoms will be either 1, 2 or 3 coordinate to lattice oxygen atoms, with their remaining coordination posi- tions being occupied by surface hydroxyl groups under ambient conditions. Although there is positive evidence for surface silicon atoms carrying one or two surface hydroxyl groups there is little evidence of surface silicon atoms carrying three. It is known that the density of amorphous silicas is close to that of the cristobalite and tridymite crystalline polymorphs. These naturally occurring crystalline forms exhibit faces in which the silicon atoms are either 2- or 3-fold coordinate to lattice oxygens.For example, in the 111 (octahedral) face of P-cristobalite the surface silicon atoms are2404 ADSORPTION AT Si02/C6H6 AND Si02/CCJ4 3-coordinate, whereas in the 100, 010 and 001 (cubic) faces the surface silicon atoms are 2-c00rdinate.~* Thus the surfaces of amorphous silicas could correspond to a " mixture " of such crystal faces. Indeed, some previous workers have suggested this model.g However, the free surface silanol groups, which can exist only on the octa- hedral face of this model, have a density corresponding to only 21.6 A2/silanol group. The area occupied by the head group of a methyl ester hydrogen bonded through its carboxyl group is 32 A2. Since this is greater than 21.6 A2 and all the free silanol groups are available as adsorption sites, it follows that they must be more randomly distributed over the adsorbent surface than this model allows.Additional evidence which supports this point is available in the literature, since it is well known that quite bulky organic molecules such as ethers and ketones will also adsorb by hydrogen bonding onto all the free surface silanol groups. A better model for the structure of the surface of amorphous silicas may be derived by assuming that the bulk of the annealed solid corresponds approximately to the cristobalite or tridymite arrangement but that the surface corresponds to a semi-random termination of this lattice. Such a model gives the correct ratio of free to hydrogen-bonded silanol groups and also exhibits isolated geminal surface silanol sites in the proportions that previous studies of the system suggest. We thank Unilever Limited for support, a personal award to A. K. M. and for permission to publish this work. L. H. Little, Infrared Spectra of Adsorbed Species (Academic Press, London, 1966), chap. 2. A. K. Mills and J. A. Hockey, J.C.S. Farahy I, 1975,71,2384. C. G. Armistead, F. H. Hambleton, J. A. Hockey and J. W. Stockton, J. Sci. Instr., 1967, 44, 872. A. J. Tyler, F. H. Hambleton and J. A. Hockey, J. Catalysis, 1969, 13, 35. M. J. D. Low and M. Hasegawa, J. Colloid Interface Sci., 1968, 26, 95. ti D. M. Griffiths, K. Marshall and C. H. Rochester, J.C.S. Faraday I, 1974, 70,400. B. J. Fontana, J. Phys. Chem., 1966,44,872. * L. H. Little, Infrared Spectra of Adsorbed Species (Academic Press, London, 1966), p. 285-293. C. G. Armistead, F. H. Hambleton, S. A. Mitchell and J. A. Hockey, J. Phys. Chem., 1969, 73, 3947. A. F. Wells, Structural Inorganic Chemistry (Clarendon, Oxford, 2nd edn., 1950), p. 569. l1 C. Y . Cheung, PhD. Thesis (University of Manchester, 1974).

ISSN:0300-9599    

DOI:10.1039/F19757102398

出版商:RSC    

年代:1975

数据来源: RSC

摘要:

Thermal Decomposition of Isopropanol BY ANTONY B. TRENWITH Department of Inorganic Chemistry, School of Chemistry, The University, Newcastle-upon-Tyne NEI 7RU Received 17th February, 1975 The pyrolysis of isopropanol has been investigated at temperatures in the range 721-801 K and pressures between 10 and 100 Torr, using a KClcoated silica reaction vessel. Under these conditions the reactions occurring were essentially homogeneous and, for reaction times corresponding to < 30 % conversion, led to the formation of hydrogen and acetone, together with smaller amounts of water and propene ; traces of methane, ethane, ethylene, and acetaldehyde were also detected. Product against time curves showed a marked fall-off in rates of formation of hydrogen, acetone, water and propene with time.Both the former and the latter pair of products were formed at similar initial rates and by reactions which were first order with respect to isopropanol. Rate constants for the formation of hydrogen and acetone yielded the expression log(kHz/s-') = (14.0+0.6)-(57 700k 2000)/8 and for propene and water, the expression where 8 = 2.303 RT/cal mol-' (1 cal = 4.18 J). These results together with earlier observations can be satisfactorily interpreted on the basis of a free radical chain mechanism initiated by the splitting of a C-C bond in isopropanol, and yielding equal amounts of hydrogen and acetone and of propane and water in the propagating steps. A comparison of the results of this investigation with those obtained previously from the decom- position of 3-hydroxybut-1-ene strongly suggests that in the latter, the principal products butadiene and water are also formed by a free radical chain mechanism rather than a molecular process.log(kCsHs/S-') = (13.1 & 1.0)-(58 2005 3600)/8 In an earlier investigation,l 3-hydroxybut-1-ene was found on pyrolysis to yield butadiene and water by a reaction having an activation energy of 55.7 kcal mol-' and an experimental A factor which agreed well with that calculated on the assumption of a molecular reaction mechanism. Experimental evidence indicates that the de- hydration of tertiary alcohols occurs by a molecular mechanism and requires an activation energy in the region of 61 kcal mol-l, so that, by analogy with the trends observed for the dehydrochlorination of alkyl halides, an activation energy of approxi- mately 66 kcalmol-l would seem reasonable for the dehydration of a secondary alcohol although a value of 1-2 kcal mol-1 lower may apply to the dehydration of 3-hydroxybut-1 -ene because of resonance stabilisation of the transition state in this rea~tion.~ Thus, despite the agreement between the measured and calculated A factors, the experimental activation energy would seem unreasonably low for a reaction of this type.In view of the apparent discrepancy between the anticipated and the observed activation energy for the dehydration of 3-hydroxybut- 1 -ene and since information on the dehydration of secondary alcohols is very limited it was decided to re-examine the thermal decomposition of isopropanol. Two previous investigations of the pyrolysis of this compound have been reported.Barnard found the overall reaction, which was inhibited by nitric oxide, led to the formation of acetone and hydrogen as principal products together with smaller amounts of methane and water. A free 24052406 THERMAL DECOMPOSITION OF ISOPROPANOL radical mechanism was proposed to account for these observations, initiated by the splitting of the secondary C-H bond. reported that although the thermal decomposition of isopropanol alone is complex, in the presence of added nitric oxide it involves essentially the formation of propene and water by a homogeneous first order reaction having an activation energy of 64.5 kcal mol-1 and this was concluded to occur by a molecular mechanism. There is thus a measure of disagreement between the two investigations since Barnard on analytical evidence stated that the molecular elimination of water was of negligible importance in the overall process.More recently Maccoll and Thomas EXPERIMENTAL MATERIALS Isopropanol (B.D.H., AnalaR) was dried by allowing it to stand for several days over molecular sieve 5A. Subsequent analysis by gas chromatography showed it to contain 0.2 % by volume H20 as the only detectable impurity. This material was transferred to the vacuum apparatus and outgassed by repeated cooling to - 83°C and pumping. Nitric oxide from a cylinder (Matheson) was transferred to the vacuum apparatus, out- gassed by freezing and pumping and then fractionated between - 198 and - 16O"C, a middle fraction being stored in a sampling bulb.APPARATUS A N D PROCEDURE The apparatus was of the conventional static type. Two cylindrical silica reaction vessels were used, both 23 x 7 cm, one being packed with silica tubing. Surface/volume ratios were 0.7 cm-l for the unpacked and 4.9 cm-l for the packed vessel. A gas storage bulb and connecting tubing to the reaction vessel were heated to 60°C to prevent condensation. Reaction products were trasferred through traps cooled to - 125 and - 198°C to a Topler- McLeod gauge. Propene, collected in the trap at - 198°C and methane in the noncondens- able products were estimated by gas chromatographic analysis using a 1 m column of alumina, whilst water, acetone and unreacted isopropanol, which were retained in the trap at - 125°C were estimated using a 1 m column of Porapak type R.The packed and empty reaction vessels were cleaned by filling with hot concentrated HN03 and allowing to stand for 48 h. After draining out the HN03, the vessels were rinsed repeatedly with distilled water and dried under vacuum. KCI-coated surfaces were prepared using the procedure described by Horner and Style.' RESULTS Isopropanol decomposed in a clean reaction vessel at 766 K yielding at <30 % decomposition, water, propene and hydrogen together with traces of acetone and methane. Initially for a pressure of 44Torr of reactant, propene was formed at a rate of 0.36 Torr s-l. Repeated conditioning of the vessel with reactant and products led to a significant drop in this initial rate, and after a total reaction time of approxi- mately 40 h it had fallen to 2.7 x Torr s-l but even then reproducibility was not attained.In contrast, runs carried out in the KC1-coated vessel were' immediately reproducible and gave, under the same conditions as before, the significantly lower initial rate of propene formation of 1.13 x Torr s-l, which did not vary when air was admitted to the reaction vessel between runs. The variation in initial rates of formation of propene was even more marked when the packed reaction vessel was used, being 3.0 Torr s-1 in the clean vessel and 1.07 x Torr s-' in the KC1-coated vessel. From the close agreement between rates of formation of propene in the empty and packed KC1-coated vessels it was concluded that in the pyrolysis of iso- propanol with this surface, reactions leading to the formation of propene are essen-A .B . TRENWITH 2407 tially homogeneous. Yields of other products were also reproducible in the KCl- coated empty vessel except that small variations (- 10 %) in hydrogen yield resulted after prolonged pumping or if air was admitted to the vessel between runs. In the KCl-coated vessel, the relative proportions of products differed from those in the clean vessel since, for up to 30 % decomposition, hydrogen and acetone were now the major products formed together with smaller amounts of water and propene, methane was found only in trace amounts, and traces of ethylene, ethane and acetalde- hyde were also detected. The data obtained from analyses of products of pyrolysis of 44 Torr isopropanol at 766 K after various heating times are shown in fig.1. The curves show that initially d[H,]ldt and d[CH,COCH,]/dt were similar and this also applies to d[H20]/dt and d[C,H,]/dt. Further, for reaction times corresponding to <25 % conversion of reactant, the yields of hydrogen plus water were virtually equal to the change in concentration of isopropanol, so that in the early stages of the decomposition the processes occurring can reasonably be represented by the equations C3H70H + CH3COCH3 + H2 and C3H70H -+ C3& +H,O. The divergence of the hydrogen and acetone curves can be ascribed to the greater reactivity of acetone under the experimental conditions and that of the water and propene curves to the greater reactivity of the latter. The total volume of non-condensable gas gave a satisfactory measure of the amount of hydrogen formed, since methane, the only other constituent detectable by gas chromatography and mass spectrometry amounted to less than 1 % of this volume 10.0- 2 4 6 time/min FIG.1.-Product against times curves for the decomposition of 44 Torr isopropanol at 766 K : products : x , C3Hs ; 0, H20; A, CH3COCH3 ; 0, H, ; 0, C3H70H. The broken line is the C3H70H against time curve calculated for first order decomposition with rate constant = [k(H,) + k(c3 &)I.2408 THERMAL DECOMPOSITION OF ISOPROPANOL at 15 % decomposition and less than 4 % at 30 % decomposition. Similarly as the product collected in the trap at - 198°C was found on analysis to be virtually pure propene, its volume could be taken as a measure of the amount of this product formed.Subsequent kinetic measurements were therefore confined to the determinations of hydrogen and propene yields from volume measurements, since these could be obtained more accurately than by gas chromatographic analysis for other products. 745 755 766 766 766 766 766 777 787.5 801 43.6 43.7 10.9 21.9 43.8 87.5 43.8* 43.8 43.9 44.0 0.26 0.47 0.18 0.37 0.76 1.47 0.16 1.11 2.19 4.35 0.10 0.18 0.28 0.28 0.29 0.28 0.06 0.42 0.83 1.65 * + 10.5 Torr NO. The yields of hydrogen and of propene after various heating times were found for a range of initial pressures of isopropanol at 766 K and also for a fixed pressure of reactant at different temperatures between 721 K and 801 K. The initial rates of formation of these products were difficult to measure accurately because of the rapid fall in their rate of formation with time during the early stages of the reaction.Three TABLE 2.-KINETIC DATA FOR THE FORMATION OF H2 FROM C3H70H T/K po(C 3H sOH)/Torr (dH?/dt)o/Torr min-1 ~ ( H ~ ) x 103/~-1 721 733 741.5 754 766 766 766 766 775.5 43.4 43.5 43.6 43.7 10.9 21.9 43.8 87.5 43.9 0.86 1.84 2.85 4.51 2.29 4.47 9.83 18.8 15.6 0.32 0.70 1.09 1.73 3.50 3.40 3.70 3.59 5.92 methods were used to determine the initial slope of each curve drawn through the experimental points ; (i) a tangent was drawn at the origin by the " mirror " technique, (ii) a fourth degree polynomial was computed from data taken from the curve for reaction times of up to 4 min and the initial slope obtained from the value of its first derivative at zero time, (iii) the Lanczos equation was used to derive the slopes after various times and the initial slope obtained by extrapolation of a graph of l/slope against time.At all temperatures the results obtained by methods (ii) and (iii) agreed to within 2 %; those obtained by method (i) differed by 3 % at the lower temperatures but by 12 % at the higher temperatures. The kinetic data have been derived using initial rates obtained by method (ii) as this is the least subjective of the three procedures, being dependent only on the accuracy of the curve drawn through the experimentalA. B. TRENWITH 2409 points. Possible errors in slope arising from this cause are estimated to be roughly 10 % at the highest and 4 % at the lowest temperature.The results obtained are shown in tables 1 and 2. These indicate that at 766 IS a %fold variation in pressure caused a negligible variation in the first order rate constants for the formation of propene [k(C3H6)] and of hydrogen [k(H,)]. 1.25 1.30 1.35 1.40 103 KIT FIG. 2.-Arrhenius plot for the reactions C3H70H -+ CH3COCH3 + H2 (0), and C3H70H -+ C~&+HZO (a). The Arrhenius plots for k(C3H6) and k(H,) are shown in fig. 2 and yielded the equations lOg[k(C3H6)/S-l] = (13. I 1 .O) - (58 200 & 3600)/6 and log[k(H,)/s-l] = (14.0f0.6) - (57 700&2000)/8 where 8 = 2.303RT/cal mol-l and the quoted limits correspond to the standard error? The Arrhenius parameters for the formation of propene differ significantly from those given by Maccoll and Thomas for the molecular elimination of water and propene from isopropanol.Rate constants calculated from the two equations for a temperature of 766 K differ by an order of magnitude, the value from the equation given above being the greater. This significant difference can be ascribed to the effect of added nitric oxide on the rate of formation of propene, since as shown in table I, the addition of 10 Torr of this gas to 43.8 Torr of isopropanol, led at 766 K to a five-fold decrease in the rate of formation of propene. DISCUSSION The marked curvature in the product against time curves coupled with the inhibi- tion of the decomposition by added nitric oxide strongly suggests that all the products found in the decomposition of isopropanol are formed in a self-inhibited free radical2410 THERMAL DECOMPOSITION OF ISOPROPANOL chain mechanism.The scheme outlined below accounts for the major products and satisfies the kinetics deduced from initial rate measurements. CH3 + C3H70H + CH4 + CMe20H C3H70H 3 CH, +-CMeHOH (1) (2) CMeHOH -+ CH,CHO+-H (3) -H + C3H70H -+ H2 +CMe,OH OH + C3H70H -+ H2 +*CH,CMeHOH -CMe,OH -+ CH,COCH, +OH (5) CH,CMeHOH + C3H6 +*OH (6) -OH+C3H70H + .CMe2OH+H2O (7) (7A) (8) *OH + C3H70H 4 CH,CMeHOH + H20 OH + CMe20H + products. Values of 89 kcalmol-I and 93 kcal mol-l have been found 8* for D(CH,CHOH-H). Using the mean of these values and the heats of formation of ethanol and of isopropanol deduced from bond additivity data yields D(CH,-CMeHOH) = 82 kcal mol-I* This value is at least 8 kcal mol-1 lower than the dissociation energies of other bonds in the molecule so that reaction (1) is the most reasonable chain-initiating step. Inclusion of the somewhat unusual chain terminating step (8) is necessary to fit the observed overall order of unity for propene and for hydrogen formation. Appli- cation of the steady-state treatment to the proposed mechanism leads to the expression [=H]/[CMe,OH] = k5 /(k4[C3H70H]) for the relative concentrations of the radicals involved in the terminating step.Reaction (5) should have an activation energy of roughly 29.5 kcal mol-1 and an A factor lo close to 1014.1 s-l. These values together with the Arrhenius parameters reported l 5 for reaction (4) yield [.H]/[-CMe,OH) = 1.7 at 766 K and a reactant pressure of 44 Torr. By analogy with the rate constants for recombination of alkyl radicals, that for the alternative terminating step 2CMe2- QH -+ products, could be two orders of magnitude less than k8 so there is reasonable justification for accepting reaction (8) as the principal terminating step in the early stages of the decomposition.The occurrence of reaction (-5) would lead to the observed self-inhibition as would also the reactions *H + CHsCOCH3 + H2 +*CH,COCH, and -H+ C3H6 -+ C3H, or H2 +C,H5 (9) (10) as both lead to the formation of comparatively stable free radicals. Whilst reaction (-5) would cause no variation in the ratio d[H2]/d[C3H,], both (9) and (10) would lead to an increase in this ratio with reaction time. The experimental results indicate a slight decrease in d[H2]/d[C,H6] with time so that reaction (- 5) would appear to be the most likely inhibiting step.The relative reactivities of a primary hydrogen and the tertiary hydrogen atom of isopropanol have been found l1 to be in the ratio 1 : 25 at 455 K expressed on a per atom basis. The justification for including both (4) and (4A) and also (7) and (7A) in the reaction scheme is that by analogy with the more fully investigated abstraction reactions by methyl radicals,12 each of the above pairs of reactions should have similar activation energies so that the ratio of their rate constants could be of the same order of magnitude at 800 K as that found at 455 K, i.e., k,/k,, x k7/kTA FZ 4.A. B. TRENWITH 241 1 The heat of formation of the radical CMe,OH is given l 3 by AHf"[-CMe,OH] = - 27.1 kcal mol-1 from which the overall enthalpy change for reaction (5) becomes AH5 = 27.5 kcal mol-l.The heat of formation of the radical CH,CMeHOH has not been determined, but may be estimated by assuming D(CH,CH(OH)CH,-H) = D(CH3CH2CH2-H). Taking a value of 98.6 kcal mol-I for the latter l4 yields AH6 = 32.8 kcal niol-l. Reactions ( - 5) and ( - 6) should both require only a small activation energy so that the activation energies of the forward reactions should be only slightly greater than the respective enthalpy changes, making both reasonable reactions under the experimental conditions. Application of the steady-state treatment to reactions (1)-(8) yields (d[H21/dt)o = (kiks(k4 h~)/h)' [C~H~OHIO (1 1) (12) and assuming k7/(k7+k7A) = 1 The above equations indicate that the measured rate constants k(H,) and k(C3H6) cannot strictly be represented by Arrhenius expressions since they both involve the sum of two rate constants.On combining (1 1) and (12) we have (d[H,]/d[C,H,]), = 1 +(k4/k4A) which, from the measured rates of formation of hydrogen and propene at 766 K gives k4 = 12k4A, so that over the experimental temperature range any departure of the Arrhenius plot from linearity should be negligible. On the assumption that (k4 + k4A) x k4, eqn (1 1) leads to the expression E(H2) = $(El + E4 + E5 - Es) for the overall activation energy of reactions leading to the formation of hydrogen. Taking values of 82.0 kcal mol-1 for El and 6.4 kcal mol-1 for E4 l 5 and assuming E-5 = 2 kcal mol-1 so that E5 = 29.5 kcal mol-1 yields E(H,) = 59.0 kcal mol-1 in good agreement with the experimental figure of 57.7 kcal mol-l.In addition [E(H,) - E(C3H6)] should correspond roughly to (E4 - E4A) and since reactions (4) and (4A) would be expected to have very similar activation energies, the agreement between the two experimentally observed activation energies is readily explained. Thermochemical evidence therefore strongly supports the proposed free- radical mechanism and leads to the general conclusion that in the pyrolysis of iso- propanol, molecular elimination reactions are unimportant, and in the early stages of decomposition reactions (1) to (8) satisfactorily account for the experimental observations. On re-examining the data obtained from the pyrolysis of 3-hydroxybut-1-ene in the light of the above conclusion it is apparent that the previously suggested molecular mechanism for the formation of butadiene and water is unreasonable, a free radical mechanism being much more likely.The product against time curves for these two products again show a marked fall-off in their rates of formation with time and the inhibition in this system can be ascribed to the presence of various resonance stabilised free radicals. The occurrence of a free radical chain mechanism is also indicated by the close similarity in the two sets of Arrhenius parameters obtained for olefin and water formation from the two secondary alcohols. On the assumption that butadiene is formed from 3-hydroxybut-1-ene by a mechanism which is similar to that given above for propene formation, available thermochemical data [El. = 70.6 kcal mol-l,l AHSt = 31.6 kcal mol (estimated), E--5r and E4' assumed to be 2 kcal mol-1 and 7 kcal m01-l respectively] yields a value of 55 kcal mo1-l for E(C4H6), in excellent agreement with the measured value of 55.7 kcal mol-l. (d[C3H61/dt)0 = {klkSk4A2/(k4 + k4,)k8);c [C~H~OHIO- A. B. Trenwith, J.C.S. Paraday I, 1973, 69, 1737. H. E. O'Neal and S . W. Benson, J. Phys. Chem., 1967, 71,2903. P. J. Thomas, J. Chem. SOC., B 1967, 1238.2412 THERMAL DECOMPOSITION OF ISOPROPANOL J. A. 3arnard, Trans. Faraday SOC., 1960, 56, 72. A. Maccoll and P. J. Thomas, Prog. Reaction Kinetics, 1967, 4, 131. E. C. A. Horner and D. W. G. Style, Trans. Furuduy SOC., 1954,50, 1197. ’ J. C. Amphlett and E. Whittle, Trans. Faraday SOC., 1968, 64, 2130. A. M. Tarr and E. Whittle, Trans. Faraday SOC., 1964,60,2039 ; 1968,64,2139. Z . B. Alfassi and D. M. Golden, J. Phys. Chem., 1972,76,3314. lo S. W. Benson, Thermochemical Kinetics (Wiley, New York, 1968). l1 P. Gray and A. A. Herod, Trans. Faraday Soc., 1968, 64, 2723. l3 R. Walsh and S. W. Benson, J. Amer. Chem. Soc., 1966, 88, 3480. l4 D. M. Golden and S. W. Benson, Chenz. Rev., 1969, 69, 125. l5 E. Ratajczak and A. F. Trotman-Dickenson, Supplementary Tables of Bimolecular Gas Reac- l6 J. A. Kerr and A. C. Lloyd, Quart. Rev., 1968, 22, 549. P. Gray and A, A. Herod, Chem. Rev., 1971, 71,247. tions (Office of Scientific and Technical Information, London, 1971).

ISSN:0300-9599    

DOI:10.1039/F19757102405

出版商:RSC    

年代:1975

数据来源: RSC

摘要:

Reaction of Oxygen Atoms with Methyl and Ethyl Nitrites BY J. A. DAVIDSON-~ and B. A. THRUSH* University of Cambridge, Department of Physical Chemistry, Lensfield Road, Cambridge CB2 1EP Received 12th March, 1975 The reactions of oxygen atoms with methyl nitrite and with ethyl nitrite have been studied in a discharge fiow system between 300 and 410 K. Initial attack is by abstraction of a hydrogen atom, for which the rate expressions are 1.4 x 1O'O exp(-21.8+2 kJ rnol-l/RT) and 2.6 x 1O'O exp(-20.3 2 2 kJ mol-'/RT) dm3 mol-' s-l. The activation energy for hydrogen abstraction is not reduced by the fact that the initially formed radical can rupture exothermically to yield an aldehyde and nitric oxide. Although the initial step in the pyrolyses of alkyl nitrites is well under~tood,"~ there is little information about the subsequent atomic and free radical reactions which are known to occur, and can also be studied in discharge flow systems.In particular, radical attack on an alkyl nitrite can yield initial products whose decomposition is exothermic. For instance, the abstraction of a hydrogen atom from methyl nitrite by an oxygen atom, which is studied here, is a highly exothermic process 0 + CH,ONO -+ OH + H2C0 +NO + 163 kJ mol-1 providing the reorganisation energy of the initially formed CH20N0 molecule into stable fragments is included. In contrast, the corresponding abstraction from a saturated hydrocarbon is almost thermoneutral (1) O+C2H, -+ OH+C2H5+17 kJmol-l. (2) An important question is whether the possibility of an exothermic reorganisation lowers the barrier to the initial abstraction step.EXPERIMENTAL The reaction between oxygen atoms and methyl or ethyl nitrite was studied in a low pressure fast flow system. The reactor was a 25 mm diameter Pyrex tube which passed through a copper-lined electric furnace 90 cm long. When using a Eurotherm temperature controller the combined temperature fluctuations plus variations along the furnace did not exceed 5 K at the highest temperatures used (410 K). The reactant could be introduced at different points along the flow tube by means of a greaseless moveable inlet system. The reagent was fed through narrow Teflon tubing into a narrow glass tube terminated by a fine jet ; this glass tube was slid along the flow system by frictional drive from a rotating shaft passing through a cone and socket joint.Reaction times of 0.02 to 1.0 s were readily obtained using a 450 dm3 min-I single stage rotary pump on the flow system. Total pressures between 0.1 and 2 kPa were measured with a calibrated silicone oil manometer. All gas flows were determined by measuring the pressure drop in a capillary flowmeter for a known total pressure; the calibration with the appropriate gas was based on, but did not assume the validity of Poiseuille's Law. t present address : Environmental Research Laboratories, N.O.A.A., Boulder, Colorado 80302, U.S.A. 24132414 O+MeONO AND EtONO Oxygen atoms were produced in two different ways. In the first, nitrogen was passed through a 200 W, 2450 MHz microwave discharge and the nitrogen atoms produced were titrated with nitric oxide according to the rapid stoichiometric reaction N+NO+ N2+O (3) to produce a known flow of oxygen atoms in a nitrogen carrier. The second method of generating atomic oxygen was to pass an argon carrier containing - 1 % of molecular oxygen through the microwave discharge.The discharge products then contained -0.5 % molecular oxygen but there was no evidence that this affected the course of subsequent reactions. An R.C.A. 1P28 photomultiplier cell fed from a stabilised supply was used to measure the air afterglow downstream of the furnace exit where the gases had completely cooled to ambient temperature. The oxygen atom concentrations at this point were determined by titration with nitrogen dioxide added through the mixing jet at the furnace exit.The stoichiometric reaction involved is O+NOz -+ NO+02 (4) and it can readily be seen that the maximum intensity of the air afterglow which obeys the relation I = Io[O][NO] (0 is observed when half the oxygen atoms have reacted. This provides an absolute calibration for the photomultiplier cell. It is then possible to use relation (i) to measure both the oxygen atom concentration remaining and the amount of nitric oxide formed in the reaction of oxygen atoms with a nitrogenous compound. If I is plotted against the flow of nitric oxide added just beyond the furnace a straight line is obtained, its slope giving the oxygen atom concentration and its intercept on the abscissa, the amount of nitric oxide formed. PREPARATION A N D PURIFICATION OF REAGENTS Methyl and ethyl nitrites were prepared by adding a mixture of the corresponding alcohol and sulphuric acid to aqueous sodium nitrite.The product was collected at 77 K. It was purified by pumping on the liquid at low temperature for several minutes, followed by distillation into a trap at 193 K through tubes packed with molecular sieve 5A and through soda lime. It was again distilled at low temperatures and was stored in blackened bulbs to prevent any photochemical decomposition. Nitric oxide from a Matheson cylinder was purified by distillation in vacuo from 90 to 77 K, and from 113 to 77 K through a trap at 113 K packed with molecular sieve and soda lime. Nitrogen dioxide was prepared by reacting unpurified nitric oxide with oxygen gas.It was condensed at 193 K and distilled in U ~ C U O until a pure white solid resulted. Oxygen and argon were purified by passage through two traps, at 193 K and atmospheric pressure, packed with glass wool foIlowed by a similar trap at 77 K and reduced pressure. Nitrogen was passed over hot copper turnings to remove oxygen and then passed through two traps at 77 K packed with glass wool. PROCEDURE Initially kinetic studies were made on the reaction of oxygen atoms with methyl nitrite and ethyl nitrite where the weakness of the 0-NO bond ensures that decomposition of parent molecule invariably yields nitric oxide or nitrogen dioxide. Any nitrogen dioxide formed would be reduced to nitric oxide in the rapid reaction O+NOz -+ NO+O2+ 192 kJ mol-'. (4) (k4 = 5 x lo9 dm3 mol-1 s-l) Thus measurement of the nitric oxide yield [NO]f as described in the previous section givesJ .A. DAVIDSON AND €3. A. THRUSH 2415 directly the amount of reactant removed. This divided into the measured consumption of oxygen atoms defines the reaction stoichiometry If, as in the reaction of oxygen atoms with hydrocarbons and with carbonyl compounds 6, ' the initial bimolecular reaction of an oxygen atom with a nitrite molecule (which has rate coefficient k) is followed by relatively rapid reactions of the fragments which may be with oxygen atoms but not with additional nitrite molecules, then the overall reaction is n O+RONO --+ NO+other products n = "olo-r~I)/FJolf. and the integrated rate expression is (ii) where nx = [O],-[O]. This equation was used to evaIuate the rate coefficients ; varying flows of reactant were added at a fixed distance upstream of the observation point (constant reaction time), the stoichiometry n and the oxygen atom consumption measured and k computed for each reactant flow.If the atoms or free radicals formed in this primary reaction sequence attack the nitrite molecules to a significant extent, additional nitric oxide is formed and the stoichiometry n is reduced, eqn (ii) then yields too high a value of k. RESULTS REACTION OF OXYGEN ATOMS WITH METHYL NITRITE The reaction between oxygen atoms and methyl nitrite is comparatively slow at room temperature and with the rather low concentrations of oxygen atoms obtained by titration of active nitrogen (< 5 x mol dm-3) large flows of methyl nitrite were needed to obtain reasonable extents of reaction.The observed stoichiometry fell with increasing nitrite flows due to reactions between the radicals formed and O0 '! L 8 12 16 20 methyl nitrite concentration x 10' mol dm-3 FIG. 1 .-Dependence of reaction stoichiometry and rate constant on methyl nitrite concentration. Reaction t h e 0.16 s at 353.1 K, [O], = 3.2 x lok7 rnol dm-3 ; A reaction time 0.15 s at 381.1 K, [010 = 2.5 x rnol dm-3 ; 0 reaction time 0.19 s at 296.1 K, [O], = 2.9 x lo-' mol dm-3 ; rate constant for run at 296.1 K. nitrite molecules. Raising the temperature accelerated both these reactions and the initial attack. The oxygen atom concentration was increased by a factor of ten by using oxygen in an argon carrier, and lower flows of methyl nitrite could be used.greatly decreasing these side reactions. Fig. 1 shows the stoichiometries then obtained241 6 O+MeONO AND EtONO plotted against the initial concentration of methyl nitrite for three different tempera- tures. All three plots clearly extrapolate to n = 5 at zero methyl nitrite concentration where its reaction with molecular fragments from the main reaction sequence must be negligible. The values for the rate constants computed from eqn (ii) setting n = 5 were inde- pendent of total pressure at a given temperature. Providing [O], > 5[CH30NO], the rate coefficients were independent of the initial concentrations, showing that the initial process is a bimolecular reaction of 0 and CH30N0. Two initial steps are plausible, direct hydrogen abstraction (1) or oxygen atom addition at the nitrogen atom followed by rupture of the weakest bond in the highly excited methyl nitrate molecule formed 0 + CH30N0 -+ OH + H2C0 + NO + 163 kJ mol-1 0 + CH30N0 -+ CH,ONOZ + 304 kJ mol-I where the initial addition is spin forbidden to form ground state methyl nitrate.These primary steps would be followed by the rapid secondary reactions -+ CH30+N02 + 136 kJ mol-I (5) O+OH -+ O2+H+71 kJ mol-1 (k6 = 3 x lo1* dm3 mol-1 s-l)* O+N02 -+ NO+02+192 kJ mol-I (6) (4) (7) 2 O+CH,ONO -+ H2CO+NO+02+H (1’) 4 O+CH30NO + H,CO+NO+2 02+H. (5’) The reaction of oxygen atoms with formaldehyde proceeds by the mechanism 0 + H2C0 -+ HCO + OH (8) O+HCO -+ OH+CO (9) and 0 + CH30 -+ OH + H2C0 + 334 kJ mol-l. The overall stoichiometries for these reaction sequences are therefore H+HCO -+ H,+CO (1 1) with k8 = 9 x lo7 dm3 mol-1 s-I at 300 K.With the rapid subsequent steps, its overall stoichiometry for oxygen atom removal is 2.8. Since reaction (8) is more than ten times faster than O+CH,ONO, it goes to virtual completion so that the overall stoichiometry is n = 5 if (1) is the initial step and n = 7 if (5) is. The former corresponds to the observed limiting stoichiometry at low methyl nitrite concentra- tions and the increase in nitric oxide production at higher nitrite concentrations can be attributed to decomposition of the nitrite by H and OH radicals produced in the above reaction sequence. Further evidence supporting (1) as the initial step comes from the Arrhenius para- meters for the initial step.If the limiting values of the rate constant k (with n = 5 ) at low nitrite concentration are plotted in Arrhenius form, the following rate expres- sion is obtained (fig. 2) : kl = 1 . 4 ~ 1O‘O exp(-21.8f2.0 kJ mol-’/RT) dm3 mol-I. s-I.J . A . DAVIDSON A N D B . A . THRUSH 2417 Herron and Huie’s formula for the rate of attack of oxygen atoms on alkanes gives k = 1.5 x 1O1O exp( -24.1 kJ mol-l/RT) dm3 mol-1 s-l for three primary hydrogen atoms in an alkane. Thus the observed Arrhenius parameters for oxygen atom attack on methyl nitrite are consistent with abstraction of primary hydrogen which is slightly more weakly bound than in an alkane [reaction (l)]. r 1 0 4 1 ~ FIG. 2.-Arrhenius plot of rate constants for oxygen atom attack. 0 Methyl nitrite with high oxygen atom concentrations ; 0 methyl nitrite with low oxygen atom concentrations ; A ethyl nitrite.The original data obtained at low oxygen atom concentrations were recalculated assuming that the true reaction stoichiometry is 5, the extra nitric oxide being produced by attack of molecular fragments (H, OH) on methyl nitrite. These data also gave a linear Arrhenius plot lying slightly below the previous points yielding kl = 1.4 x 1O1O exp(-23.2f2.0 kJ mol-l/RT) dm3 mol-1 s-I. This is nearly a factor of two less than the value found at high [O] since the OH and H formed in the atomic reactions may compete with 0 for the intermediate formaldehyde and for HCO, thereby reducing the number of oxygen atoms consumed per initial step. At room temperature the rates of the relevant reactions H + H,CO -+ H2 + HCO OH + H2C0 -+ H20 + HCO are k,, = 3.2 x lo7 and kI3 = 8 x lo9 dm3 mol-1 S - ~ , ~ * ? the former is a factor of three slower than (8), and (1 3) is three times slower than (6) which is the main sink for OH.However, with the much higher methyl nitrite flows used in the earlier experi- ments, these processes would be responsible for a significant fraction of the difference between data from the two regimes. The value from high oxygen atom concentra- tions is therefore preferred.241 8 O+MeONO AND EtONO REACTION OF OXYGEN ATOMS WITH ETHYL NITRITE The presence of an, additional CH2 group in ethyl nitrite increases the number of oxygen atoms removed per initial step and makes it difficult to distinguish the mechan- isms on this basis.However, its presence should affect the rate of the initial step 0 + EtONO + OH + CH3CH0 + NO + 184 kJ mol-1 (14) (15) -+ CH3 CH20NO; + 301 kJ mol-I -+ CH3 CH20 +NO2 + 129 kJ mol-l. Reaction (14) involves the abstraction of secondary hydrogen, and although there are no data on the C-H bond energies in alkyl nitrites, there is every reason to believe that this will be faster since the secondary C-H bond in ethyl nitrite is almost cer- tainly weaker than the primary C-H bond in methyl nitrite. The effect of the additional CH2 group on reaction (15), where the initial process is addition of an oxygen atom followed by break up of the excited nitrate molecule formed, would be to reduce the rate of the latter process by a factor of about 30, since the energetics of the reactions for the methyl and ethyl derivatives are very simi- lar, but the presence of an additional CH2 increases by 9 the number of vibrational degrees of freedom between which the vibrational energy has to be distributed.As the predicted lifetimes of these excited ethyl nitrate molecules lie in the range to s, the presence of the additional CH2 group might not lead to significant colli- sional stabilisation of the nitrate molecules at the total pressures used here. The reaction of ethyl nitrite with oxygen atonis was found to be significantly faster than the reaction of methyl nitrite. It was therefore possible to work with lower reagent concentrations. Nevertheless, the measured stoichiometries of nitric oxide production were strongly dependent on the amount of ethyl nitrite added and rose quite sharply to a limiting value close to 9 at zero reactant concentration, as shown in When the subsequent reactions (6), (4) and (16) are included, the overall stoichio- fig.3. metries became 0 + CH3CH20 -+ CH3CH0 + OH+ 360 kJ mol-' (16) (14') (15') 2 0 + CH3CH20NO + CHACHO +NO + 0 2 + H 4 O+CH3CH20N0 -+ CH3CHO+NO+2 O,+H. and The measured rate constant for the reaction of oxygen atoms with acetaldehyde ' is 3 x los dm3 mol-1 s-I at 300 K which is a factor of 30 faster than for 0 + EtONO at this temperature. It can therefore be assumed that the O+CH,CHO reaction goes to completion on the time scale of these experiments. The processes occurring under the present conditions would be ' 0 + CH3CH0 -+ OH + CH3C0 + 63 kJ mol-1 O+OH -+ O2+H+71 kJ mol-' 0 + CH3C0 -+ CH3 + C02 + 430 kJ mol-' 0 + CH3 -+ H2C0 + H + 286 kJ mol-' (17) (6) (1 8) (19) giving a stoichiometry of 4 for oxygen atom removal.Since the reaction of oxygen atoms with formaldehyde also goes virtually to completion under the present condi- tions and has a stoichiometry of 2.8 [see above and ref. (12)], the predicted overall stoichiometries are n = 9 if (14) is the initial step and rz = 11 if (15) is. Thus theJ . A . DAVIDSON AND B . A . THRUSH 2419 measured stoichiometry again favours abstraction as the primary step. The limiting values for k14 at low ethyl nitrite concentrations are plotted in fig. 2. It can be seen that reaction (14) is slightly more than three times as fast as (1) and appears to have a slightly lower activation energy. Because the A factor for reaction (1) is very close to that predicted by Herron and Huie's rate expression for primary alkanes,' their predicted value of 2.6 x 1Olo dm3 mol-1 s-1 for two secondary hydrogens in an alkane has been adopted for the intercept in fig.2. This value also gives the same ratio of A factors for oxygen atom attack on ethyl and methyl nitrite and on ethyl and methyl chloride lo (2.8). The line on fig. 2 corresponds to k12 = 2.6 x 1 O 1 O exp( -20.3 kJ mol-'/RT) dm3 rn01-~ s'l. DISCUSSION Herron and Huie have made careful and critical studies of the abstraction of hydrogen atoms from paraffin hydrocarbons by oxygen atoms. They found that the activation energies show a linear dependence on bond strength E, = 0.36 [D(C-H)-343 kJ mol-'I.Assuming that the C-H bonds in nitrites obey the same equation, this would give D(C-H) = 404 kJ mol-1 for methyl nitrite and D(C-H) = 400 kJ mol-l for the secondary hydrogen atoms in ethyl nitrite. The former is 5 kJ mol-' below that in ethane and the latter is within the range expected for secondary hydrogen atoms." It is therefore clear that the measured Arrhenius parameters for the attack of oxygen atoms on methyl and ethyl nitrite are those expected for hydrogen atom abstraction ; together with the measured overall reaction stoichiometries they provide convincing evidence of this initial step. r '0 I 2 3 L 5 6 7 8 9 1 0 1 1 1 2 ethyl nitrite concentration x lo8 mol dm-3 FIG. 3.-Dependence of reaction stoichiometry on ethyl nitrite concentration.A Reaction time 0.112 s at 293.1 K, [O], = 2.9 x lo-' rnol dm-3 ; 0 reaction time 0.092 s at 364.1 K, [O], = 2.5 x lo-' rnol dm-3 ; 0 reaction time 0.086 s at 379.1 K, [O], = 2.4 x lo-' mol dm-3. One not wholly unexpected feature is that the activation energies for 0 atom attack, and by implication the C-H bond strengths, are those for normal C-H bonds. This implies that the exothermic reorganisation and fragmentation of the initially formed free radical occurs after abstraction of the hydrogen atom. Reactions (1) and (14) can therefore be regarded as occurring in two steps, but it is unlikely that2420 O+MeONO AND EtONO the intermediate radical could be stabilised except possibly in matrices at low tempera- ture. 0 + CH30NO 4 OH + CH2ONO.CH20N0 + H2CO+N0. 0 + CM3CH20N0 4 OH + CH3CHON0. CHSCHONO + CH3CHO -t- NO. The rapid decrease in stoichiometry as the reactant concentration is increased shows that either or both H and OH which are produced in the primary reaction sequence react more rapidly with the parent molecule than do oxygen atoms. This is particularly striking for ethyl nitrite (fig. 3), but it is not unexpected, since hydrogen abstraction by H or OH is respectively 8 and 71 ItJmol-l more exothermic than abstraction by 0, although the steady state Concentration of OH is very low due to the rapidity of reaction (6). The most interesting result obtained is that the attack of oxygen atoms on methyl nitrite or ethyl nitrite is not accelerated by the fact that the radical formed by the initial abstraction of hydrogen can decompose exothermically to more stable pro- ducts. Thus the fact that energy may be released in the fragmentation of an unstable intermediate does not necessarily accelerate the reaction yielding that intermediate. The research reported here was sponsored in part by the United States Government. P. Gray, Proc. Roy. SOC. A , 1954, 221, 462; P. Gray and A. Williams, Chem. Rev., 1959, 59, 239. J. B. Levy, J. Amer. Chem. SOC., 1956,78,1780. L. Phillips, J. Chem. Soc., 1961, 3082. D. D. Davis, J. T. Herron and R. E. Huie, J. Chem. Phys., 1973,58, 530. J. M. Brown and B. A. Thrush, Trans. Furaday SOC., 1967, 63, 630. G. P. R. Mack and B. A. Thrush, J.C.S. Faraday I, 1973, 69,208. ’ G. P. R. Mack and B. A. Thrush, J.C.S. Faraday I, 1974,70, 178, 187. M. A. A. Clyne and B. A. Thrush, Proc. Roy. SOC. A, 1963,275,544. J. T. Herron and R. E. Huie, J. Phys. Chem., 1969, 73, 3327. lo J. T. Herron and R. E. Huie, J. Phys. Chem., 1969, 73, 1326. l1 J. A. Ken, Chem. Rev., 1966,66,465.

ISSN:0300-9599    

DOI:10.1039/F19757102413

出版商:RSC    

年代:1975

数据来源: RSC

摘要:

Infrared Spectroscopic Study of the Adsorption of Nitriles on Aluminium Oxide Fermi Resonance in Coordinated Acetonitrile BY HELMUT KNOEZINGER;~ AND HANS KRIETENBRINK Physikalisch-Chemisches Institut der Universitat Munchen, 8 Miinchen 2, Sophienstr. 11, West Germany Received 2nd April, 1975 The adsorption of acetonitrile CH3CN and CD3CN on to a 6-A1203 sample has been studied by i.r. spectroscopy. The nitrile is coordinately bonded to exposed A13+ ions or €€-bonded to surface hydroxyl groups. Fermi resonance between the V p fundamental and the (V3 + T4) combination in the two adsorbed species is discussed and the positions of the unperturbed vibrations are estimated. An assignment of the Fermi resonance doublet in the coordinated species is given. The C=N stretching vibration seems not to be very sensitive to energetic heterogeneities among the exposed cations.1.r. and Raman spectra of adsorbed t-butyl cyanide are also discussed. Pyridine has frequently been used as a probe for the i.r. detection of acid sites on oxide and mixed-oxide surfaces by i.r. spectroscopy. 1-4 However, the relatively large size of the pyridine molecule may cause difficulty and, due to its high basicity, pyridine does not interact specifically with surface acid sites.4 It would be desirable to use smaller, less basic molecules as probes. Nitriles have a promising functional group and are of low basicity. In addition, the C = N stretching band is in a very favour- able spectral range, in which the transmittance of the oxide samples is usually high and in which no other bands of the adsorbate appear.The C = N stretching vibra- tion is sensitive to the electron withdrawing power of metal ions to which the nitrile is co~rdinated.~ One might therefore expect to detect the acid strength distribution from wavenumber shifts of this band. Zecchina et aL6 have observed an influence of the acidity of chromium ions in a Cr203-Si02 surface on the position of the C=N stretching band. The adsorption of nitriles on oxide or mixed-oxide surfaces has been little studied and reports on theit adsorption on aluminia-containing surfaces are scarce. H- Bonding interactions are observed on silica surfaces,' while coordination adducts are formed on surfaces which expose incompletely coordinated cations. The adsorption of acetonitrile has been studied on zeo1ites,"l0 on Si02-A1203,11 on Al- and Zr- impregnated porous glass and silica surfaces l2 and on AlC13 l3 and ZnCI2.l4 Zecchina et aL6 recorded the spectra of propionitrile, benzonitrile and acrylonitde on Cr203-Si02.Benzonitrile adsorption has been studied on MgO, CaO and Be0 and on Zr02.16 Chapman and Hair l7 report chemical transformations of nitriles on aluminia-containing surfaces, which they describe as oxidation reactions. Chemi- cal transformations of acetonitrile on 6-A1203 have also been observed during the course of this work. A subsequent contribution will deal with these. Such reactions are reduced at low temperatures and at low nitrile pressures as well as by substitution in the nitrile. Under suitable conditions H-bonding and coordination occur and this paper deals particularly with the coordinative adsorption of acetonitrile. Fermi 24212422 FERMI RESONANCE I N COORDINATED ACETONITRILE resonance between the C = N stretching vibration V2 and the (V3 + 5,) combination vibration complicates the assignment of bands and the determination of wavenumber shifts on coordination of acetonitrile.An analysis of the Fermi resonance and assignments of the bands of weakly held and coordinated acetonitrile respectively are presented. EXPERIMENTAL Alumina was PllO C1 from Degussa, Hanau, Germany. It is mainly &phase which is produced by flame hydrolysis of AICl3. Its primary particle diameter is 5-30 nm and its B.E.T. surface area was (N2 adsorption) (105+ 10) m2 g-l. The oxide was heated in air at 800°C and was then rehydrated by atmospheric water vapour before use.Self supporting plates, 1 . 7 ~ 3.0 cm in size, were obtained by pressing the powder under a pressure of 5-9 MPa. The plates were approximately 0.02 cm thick and had a weight of (16+2) mg cm-2. Acetonitrile (fur Spektroskopie) was from Merck-Schuchardt, [2H3]acetonitrile from Aldrich-Europe and t-butyl cyanide from Fluka. All substances were p.A. grade and were dried over Linde molecular sieve 3A prior to use. The room temperature i.r. cell and its accessories have already been described else- where.'** l9 The sample temperature in this device cannot be measured easily, it is, how- ever, fairly high due to the high power of the incident i.r. beam. The light source is a globar. The temperature of a Si02 disc (which has a much higher transparency) was recently deter- mined to be about 80°C 2o and similar temperatures have been estimated indirectly for q-A1203 discs.2f Except at relatively low pressures (<4 mmHg), acetonitrile undergoes chemical transformations on A1203 under these conditions.A new version of the room temperature cell was therefore designed which allows external cooling. Fig. 1 shows a schematic diagram of the new cell. Owing to the poor heat exchange between the non- conducting oxide plate and the metal part of the cell at low pressures, a few mmHg of helium were introduced into the cell. Although the exact temperature of the plate cannot be mea- sured, it was effectively reduced when the cell was cooled to - 20°C, as shown by the absence of significant chemical reactions of acetonitrile.I IR. beam FIQ. 1.-The low temperature i.r. cell. (1) NaCl windows, (2) CaFz windows, (3) Viton O-rings, (4) sample position, (5) to heating zone and high vacuum line, (6) to vacuum line for prevention of condensation of atmospheric water on cold CaF2 windows, (7) coolant. b\J, brass, i?, macrolan.H. KNOEZINGER AND H . KRIETENBRINK 2423 The spectrometer was a Perkin-Elmer model 225. The spectra were recorded at a scan speed of 1.5-2.5 cm-' s-l. The resolution was better than 2 cm-' in the range 1000-2800 cm-I. The alumina plates were preheated in situ to any desired temperature up to 900°C in vacuum (< After cooling the sample to the cell temperature (prestabilized in the i.r. beam), adsorption experiments were carried out, starting at low pressures, and the spectra were recorded.mmHg). Desorption was carried out at elevated temperatures. RESULTS A N D DISCUSSION The CH,CN molecule belongs to the point group C3" and possesses twelve fundamental vibrations. V, to V4 vibrations belong to the totally symmetric species A , , whereas vibrations i j 5 to V8 are degenerate and belong to species E. These fundamental vibrations of CH,CN in the liquid state are summarized in table 1 and their assignments according to Venkateswarlu are given.22 The position of the (V3 +V4) combination band is also given, since this band undergoes Fermi resonance with the C=N stretching band V2. The relatively high intensity of the combination band is due to this resonance.TABLE 1.-I.R. BANDS OF LIQUID CHsCN AND THEIR ASSIGNMENTS description symmetric C-H stretching symmetric CH3 deformation antisymmetric CH3 deformation antisymmetric C-H stretching C = N stretching C-C stretching CH3 rocking C-C = N bending symmetry spec 1 e s wavenutnberlcm-1 2942 2254 1375 914 1442 3001 1047 380(R) 2292 On coordination of the nitrile group to a metal ion, the Vs vibration is shifted to higher wavenumbers as indicated by the spectra of other coordinated nitriles. In these compounds Fermi resonance does not occur. This wavenumber shift of the ij2 vibration is typically of the order 20-100 cm-l depending on the electron withdraw- ing power of the metal The increase in the wavenumber of the 7, vibration on coordination is due to a strengthening of the C = N bond by rehybridization of the nitrogen orbitals, which is connected with an increase of the strength of the a-bond as shown by Filimonov and Bystrov 2 5 and by Purcell and Drago.26 Due to the Fermi resonance between V2 and (V3 +q4) in CH,CN, the assignment of the bands in the range 2200 to 2350 cm-' for coordinated species is uncertain and no estimate of the position of the unperturbed bands in CH3CN coordination compounds has yet been given.ADSORPTION OF CH3CN ON 6-Al,03 Fig. 2 shows the spectrum between 2000 and 25OOci~1-~ for the adsorption of CH3CN on a 6-A1203 which had been pretreated at 420°C. Very similar spectra were obtained after higher pretreatment temperatures. The spectrum of liquid CH3CN is also shown for comparison, the ij2 band appearing at 2254 cm-' and the combination band (5, +V4) at 2292 cm-l.On adsorption of CH3CN at pressures of 4 mmHg on to the alumina disc, a band pair appears at 2300 and 2328 cm-'. The symmetric C-H stretching vibration produces a band at 2930 cm-l which is shifted 242424 FERMI RESONANCE IN COORDINATED ACETONITRILE to lower wavenumbers by 12 cm-I as compared with the liquid. The OH stretching bands of the surface hydroxyl groups which appear at 3792,3772,3728 and 3680 cm-' are practically unchanged. This indicates that CH3CN is interacting only with sites not involving protons, at the low pressures used. The significant shift of the Fermi resonance doublet to higher wavenumbers on adsorption suggests that a coordinate bond between the nitrile group and an incompletely coordinated A13+ ion in the surface is formed.25 0 Tjcrn-' FIG. 2.-1.r. spectra of CH,CN: (a) liquid, (b) 6-A1203 background, (c) adsorbed on 6-A1203 (pretreated at 420"C), 4 mmHg, cell temperature - 20°C ; (d) as (c), saturation vapour pressure at - 20°C ; (e) adsorbed on S-A1203 (pretreated at 300"C), saturation vapour pressure at - 20"C, cell temperature - 20°C. An increase of the CH3CN pressure to 75 mmHg leads to an increase of the intensity of the band at 2328 cm-l, the band at 2300 cm-l shifts to 2297 cm-l and increases in intensity and an additional band appears at the position of the ij2 mode of liquid CH3CN at 2253 cm-l. (A band at 2180 cm-l results from a chemical transformation of the CH3CN and will not be discussed here.) Simultaneously with the appearance of the band at 2253 cm-', the symmetric CH stretching band shifts to 2940 cm-l, the stretching bands of the surface hydroxyl groups become strongly perturbed and a broad band centred at 3550 cm-l is formed.These observations clearly indicate the formation of a second adsorbed species, which is held by an O-H - - N H-bond between a surface hydroxyl group and the nitrogen lone pair of the nitrile. Only small shifts to higher wavenumbers of the C =N stretching mode on H-bonding were observed by Low and Bartner,12 for CH3CN adsorbed on porous glass and silica. The formation of coordinated and H-bonded CHJCN species on the alumina surface thus seems to have been established. The assignment of the Fermi resonanceH. KNOEZINGER A N D H . KKIETENBRINK 2425 doublet at 2300 (2297) cm-' and 2328 cm-l, however, is not readily possible.In fact, information in the literature regarding the assignment of Fermi resonance doublets in coordination compounds of CH3CN, homogeneous as well as surface adducts, is confusing.24 As in the present case for the higher pressures, a triplet is often observed, e.g. on interaction of ZnCl, with CH3CN in the liquid phase or of Zn2+ ions with CH3CN in zeolites.* The various assignments of the bands in the triplet are summarized in table 2. Thus, an assignment of the present bands on the basis of literature data is not possible. An attempt to estimate the position of the unperturbed bands is therefore undertaken. TABLE 2.-cOMPARISON OF THE ASSIGNMENTS OF THE FEW1 RESONANCE BANDS IN ACETONITRILE/ Zn2+ ADDUCTS AND LIQUID ACETONITRILE compound highest position intermediate lowest position position ref.CH3CN-ZnC12 75- (i3+74)19 v'z (27) CH3CN-ZnC12 (Vg +v,)= (V3+74)'+75 V$ (28) (2CH CN)-ZnCl (V3 + 5 4 ) C 75 sl, (29) zeolite (73 +V,)C 5% sl, (8) CH3CN/Zn-Y Superscript c indicates coordination, superscript I indicates liquid or weakly adsorbed state. FERMI RESONANCE IN CH3CN Fermi resonance 30* 31 may arise from a transition from the vibrational ground state into a pair of energy levels which may be described by the wave functions $,, and \Clrn. These may be expressed as orthogonal combinations of the unperturbed wavefunctions : $n = a$: -t b$i (1) $,,, = b$,"-a$:. (2) Here $: describes the first excited vibrational state of a fundamental and $: that of a combination vibration of the same symmetry species.If strong resonance occurs, a doublet will result and since a E b, the intensity of the two constituents will be identical. As shown by H e r ~ b e r g , ~ ~ the secular equation for the perturbation can be solved if one assumes that only one of the unperturbed transitions leads to finice intensity. This approximation is very nearly valid since the intensity of a combination band is usually very weak as compared to that of a fundamental. The energy separation of the unperturbed states can then be calculated from that of the perturbed states accord- ing to eqn (4) Furthermore a2+b2 = 1. (3) P+1 p- 1' AEo = AE- (4) where p is the ratio of the actual intensities of the bands of the Fermi resonance doublet.31 Since the perturbation is symmetrical 30 the wavenumbers of the un- perturbed transitions can be written as 7: = +(fn+Tv,+Aio) ( 5 4 7: = HV, + 7,- AV).(5b) The bands in liquid CH3CN are found at 2254 cm-1 (7,) and 2293 cm-l (5, +V4) and2426 FERMI RESONANCE IN COORDINATED ACETONITRILE their intensity ratio as determined from fig. 2a is 4.00+0.06. It thus follows that the wavenumber separation of the unperturbed transitions is A G O = - (19.5 + 0.5) cm-I . The position of the bands for unperturbed transitions would therefore be observed at (V;)[ = (2261.5k0.5) cm-' and (i$+Vi)i = (2285.5f0.5) cm-l for liquid CH3CN. An analogous calculation can be carried out for the Fermi resonance doublet of coordinately adsorbed CH3CN on b-A1203 (fig. 2). The two bands observed at 2328 and 2300 cm-I are separated by AT = 28 cm-' and their intensity ratio is p = 1.14fO.10.The unperturbed bands of the doublet are then to be expected at (2316 f 1) and (2312f 1) crn-', their assignment, however, is still not clear. Depending on whether the higher or the lower wavenumber is assigned as the (S;)" mode of the coordinated CH3CN, wavenumber shifts on coordination of 54.5 cm-I or 50.5 cm-1 are calculated. The C-C stretching vibration (T4 mode) which contributes to the (V3 +Fa) combination is certainly affected by the coordination bond,23 the influence should, however, not be very strong since the vibration is mechanically only weakly coupled to the N-A1 stretching vibration through the V2 mode, although electronic perturbations may not be completely negligible.It may therefore be assumed that the lower wavenumber band at 2312 cm-1 can be assigned as the combination band (Go3 +Ti)", while the band at 2316 cm-l is due to the C=N stretching band, (V")" mode, of the coordinated CH3CN. This assignment leads to the result that the $02 mode is shifted to higher wavenumbers by 54.5 cm-l on coordination of CH3CN to incompletely coordinated A13+ in 6-A1203 pretreated at 400°C. TABLE 3.-ASSIGNMENT OF THE FERML RESONANCE BANDS OF ACETONITRILE ADSORBED ON &A1203 observed unperturbed positionlcm-1 position/cm- 1 assignment 2328 2316 25- 2300 2312 (T3 + V q Y 2297 2285.5 (T3+Vq)'* 2253 2261.5 75 CHBCN 2315 2259 CDjCN This assignment, which corresponds to that given by Evans and Lo 27 (see table 2), finds some support from the spectra obtained for adsorbed deuterated acetonitrile CD3CN.This molecule does not exhibit Fermi resonance. The spectra are shown in fig. 3. The band at 21 12 cm-l is the symmetric C-D stretching vibration. The G2 mode in liquid CD3CN absorbs at 2259 cm-l and shifts to 2315 cm-l, the shift being 56 cm-l. Bearing in mind the experimental resolution, which was better than 2 cm-l and the uncertainties in the estimated values of the unperturbed band positions for CH3CN, the higher estimated wavenumber shift of 54.5 cm-l for CH3CN comes much closer to that observed experimentally for CD,CN. The assignment for the Fermi resonance doublet, which is summarized in table 3, thus seems to be the most probable. If coordinated and H-bonded CH3CN are adsorbed simultaneously, the combination bands of both species very i:ear.ly coincide and lead to the central band of the observed triplet at 2297 cm- I .H.KNOEZINGER AND Id. RRIETENBRINK 2424 Since no Fermi resonance occurs with the CD3CN molecule, a possible dependence of the V2 mode on the surface coverage and on the activation temperature of the oxide should be detectable. Although CD3CN also undergoes chemical transformations on the surface, no shift of the ij2 mode seem to occur on variations of coverage or activation temperature. t-Butyl cyanide does not undergo chemical transformations at temperatures below 200°C and Fermi resonance is also absent. This molecule is therefore suitable fcr testing the sensitivity of the C E N stretching vibration towards the two parameters, surface coverage and activation temperature of the oxide.i1crn-l FIG. 3.-1.r. spectra of CDJCN adsorbed on 6-A1203 (pretreated at 420°C) ; cell temperature - 20°C. (a) background, (b) 4 mmHg, (c) saturation vapour pressure at -20°C. ADSORPTION OF t-BUTYL CYANIDE ON &A1203 The i.r. and Raman bands observed for liquid t-butyl cyanide have been assigned by C r o ~ d e r . ~ ~ The C=N stretching band appears at 2238 cm-1 in the infrared and at 2245 cm-l in the Ramaii spectrum. As shown in fig. 4, this band shifts to high wavenumbers on adsorpticn on 6-A1203, the band of the H-bonded species being located at the same position as in the liquid. The C=N stretching vibration of coordinately adsorbed t-butyl cyanide is reduced in intensity by desorpt ion at increas- ing temperatures, its position and half-width, however, remain fixed and are indepen- dent of the activation temperature of the oxide.(The small band 2090 cm-l, which occurs on desorpticn at 220°C on 6-AI2Q3, activated at 9WC, is due to the CN- ion. This indicates dissociative chemisorption of the nitrile at 220°C. This will be discussed in detail in a forthcoming paper.) Thus, the wavenumber shift of the C =N stretching vibration on coordination to A13+ ions seems to be independent of the surface coverage and activation temperature. The shift of AVcrN = 58 cm-1 is somewhat higher than that estimated for CW,CN and for CD3CN. This is due to the increased basicity of t-butyl cyanide. The bands are symmetric and fairly narrow (half-width 35 cm-l).The C = N stretching mode therefore does not show evidence of surface heterogeneity. This certainly does not indicate absence cf an energy distribution among the exposed A13+ ions in 6-A120,. The insensitivity of the CEEN stretching mode towards the heterogeneity of the exposed cations might be due to a cancellation of the different efiects of changes in the rehybridization and polarization of the C = N bond on coordi- nation onto A13+ ions of slightly differing electron withdrawing power. This inter-2428 FERMI RESONANCE I N COORDINATED ACETONITRILE pretation, however, is only tentative. It is interesting in this connection to note, that bands due to the adsorbed nitriles appear to be slightly broader than those for the liquid (see fig. 2). Since the rotational freedom of the CH3CN molecule perpendicu- lar t o the C3 axis which exists in the must be restricted in the adsorbed state, one would expect a partial removal of the band broadening on adsorption.Thus, as the opposite effect is observed, the slight increase of the band widths in the adsorbed state may be evidence for the existence of some energetic heterogeneity.* However, the sensitivity of the C EN stretching vibration towards energetic heterogeneity of the surface must be low, since the band position does not respond detectably to changes in surface coverage and pretreatment. I a3 m N s-4 00 2"JO 2 0 220 30 7lcrn-l FIG. 4.-1.r. spectra of t-butyl cyanide adsorbed on 6-A1203, normal temperature cell. (a) Activation temperature of 6-A1203, 300°C.(b) Activation temperature of 6-Al2O3, 900°C. A.T. = ambient cell temperature. The spectra are shifted along the ordinate ; the coverage decreases in the sequence (4 (b) from bottom to top. The numbers are desorption temperatures ("C). Finally, the Raman spectrum of t-butyl cyanide adsorbed on rpAl,O, (activated at 500°C) under its saturation vapour pressure is shown in fig. 5. The majority of the observed bands belong to the H-bonded species, the appearance of a weak band at 2286 cm-l, which is shifted by 52 cm-l with respect to the band of the H-bonded species at 2234 cm-l, indicates the formation of the coordinated species. A detailed discussion of the Raman spectra will not be given here, since the presently available information is not sufficiently large and the work is still in progress.There is one * This point was brought to our attention by one of the referees.H. KNOEZINGER AND H. KRIETENBRINK 2429 point that is worth mentioning at the present time. The relative intensities of the Raman bands of the H-bonded species with respect to that of the C r N stretching band differ from those observed in the liquid (see table 4). The methyl rocking vibration at 942 cm-', the relative intensity of which is increased by a factor of about .. . . . . . . . . . . . . - -+a . _- -. . . . -. . . . . . . .- . . -. . . .. - - . . ~ ..... 0; FIG. 5.-Raman spectra of t-butyl cyanide adsorbed on v-Alz03 (pretreated at 500°C) : Cary 82, argon ion laser ; exciting line 514, 5 nm (0.72 W), scan speed 0.25 cm-l s-l, pen period 50 s ; spectral slit width 0.3 cm-l, sensitivity 200 count s-' ; zero suppression.3 in the adsorbed state, is influenced particularly strongly. Also, the relative inten- sity of the antisymmetric CH3 stretching vibration is appreciably increased in the adsorbed state, while the other Raman bands remain practically unchanged. Analo- gous observations have been made for the adsorption of t-butyl cyanide on silica. Such results may possibly enable us to draw conclusions as to the orientation of the adsorbed molecule on the surface and to its interaction with the surface electric field. TABLE 4.-RELATIVE INTENSITIES (kfrc1) OF RAMAN BANDS OF t-BUTYL CYANIDE IN THE LIQUID AND ADSORBED STATE wavenumber/cm- 1 Arel (liquid) Arel (adsorbed) 2234 1 1 942 0.07 0.21 873 0.21 0.20 690 0.88 0.78 CONCLUSION On the basis of the estimated positions of the unperturbed vibrations of the Fermi resonance doublet of coordinated acetonitrile, the most probable assignment of the interacting vibrations has been given.The true wavenumber shift due to interaction2430 FERMI RESONANCE IN COORDINATED ACETONITRILE of the nitrile with an exposed cation could thus be calculated. The C=N stretching mode of nitriles d o s not seem to be very sensitive to the energetic heterogeneity of the alumina surface, although for chromia a dependence of the position of this vibra- tion on the surface acidity had been observed.6 Nitrites are therefore not useful as probes for the detection of an energy or acidity distribution on an alumina surface; they may, however, be useful as specific poisons in catalytic reactions under conditions where the nitrile does not undergo chemical transformation^.^ The authors gratefully acknowledge the financial support of this work by the Fonds der Chemischen Industrie and by the Deutsche Forschungsgemeinschaft. The i.r.spectrophotometer was placed at our disposal by the Stiftung Volkswagenwerk. The Raman spectrum was recorded by Dr. H. Jeziorowski. L. H. Little, Infrared Spectra of Adsorbed Species (Academic Press, London, New York, 1966). M. L. Hair, Infrared Spectroscopy in Surface Chemktry (Dekker, New York, 1967). A. V. Kiselev and V. I. Lygin, Infracrasnie Spectry Poverhnostnyh Soedinenii (Nauka, Moscow, 1972). H. Knoezinger, Adu. Catalysis, 1976, 25, in press. H. A. Brune and W.Zeil, 2. Naturforsch., 1961, 16a, 1251. A. Zecchina, E. Guglielminotti, S. Coluccia and E. Borello, J. Chem. SOC. A, 1969, 2196. H. Knoezinger, in Recent Progress in Hydrogen Bonds, ed. P. Schuster, G. Zundel and C. San- dorfy (North Holland, Amsterdam, 1975), in press. C. L. Angel1 and M. V. Howell, J. Phys. Chem., 1969,73,2551. H. Karge, Surface Sci., 1973, 40, 157. A. N. Ratov, A. A. Kubasov, K. V. Topchieva, E. N. Rosolovskaya and V. P. Kalinin, Kinetika i Kataliz, 1973, 14, 1024. L. M. Roev, V. N. Filimonov and A. N. Terenin, Optika i Spektroskopia, 1958, 4, 328. l2 M. J. D. Low and P. L. Bartner, Canad. J. Chem., 1970,48, 7. l3 A. Bertoluzza, G. B. Bonino, G. Fabbri and V. Lorenzelli, J. Chim. phys., 1966,63, 395. I4 G. B. Bonino, V. Lorenzelli and P. Nanni, Atti Accad. naz. Linwi, Rend. Classe Sci. fis. mat. nat., Ser. 8, 1967, 42, 152. N. E. Tretyakov and V. N. Filimonov, Kinetics and Catalysis, 1970,11, 815. I6 N. E. Tretyakov, E. Podmyakov, 0. M. Oranskaya and V. N. Filimonov, Russ. J. Phys. Chem., 1970,44,596. I7 J. D. Chapman and M. L. Hair, Proc. 3rd Int. Congr. Catalysis, 1964 (Amsterdam, 1965) 2, 1091. H. Knoezinger, H. Stolz, H. Buhl, G. Clement and W. Meye, Chemie-1ng.-Tech., 1970,42,548. l9 H. Knoezinger, Acta Cient. Yenezolana, 1973, 24, suppl. 2, 76. 2o H. Knoezinger and W. Stahlin, unpublished results. 21 A. Corado, A. Kiss, H. Knozinger and H. D. Muller, J. Catalysis, 1975, 37, 68. 22 P. Venkateswarlu, J. Chem. Phys., 1951, 19,293. 23 J. Reedijk, A. P. Zuur and W. L. Groeneveld, Recueil, 1967, 86, 1 127. 24 J. Yarwood, in Spectroscopy and Structure of Molecular Complexes (Plenum, London and New 2 5 V. N. Filimonov and D. S. Bystrov, Optika i Spektroskopia, 1962, 12,31. 26 K. F. Purcell and R. S . Drago, J. Amer. Chem. SOC., 1966, 88,919. 27 J. C. Evans and G. Y . S. Lo, Spectrochim. Acta, 1965,21, 1033. 28 C. C. Addison, D. W. Amos and D. Sutton, J. Chem. SOC. A, 1968,2285. 29 T. Kamo and M. Kimura, Bull. Chem. SOC. Japan, 1972,45,3309. 30 G. Herzberg, Molecular Spectra and Molecular Structure. II. Infrared and Raman Spectra of 31 J. Overend, in Infrared Spectroscopy and Molecular Structure, ed. M. Davies (Elsevier, Amster- 32 G. A. Crowder, J. Phys. Chem., 1971, 75, 2806. 33 J. E. Griffiths, J. Chem. Phys., 1973,59, 751. York, 1973) p. 141 ff. Polyatomic Molecules (van Nostrand-Reinhold, New York, 1945), p. 215 ff. dam, 1963), p. 350 ff.

ISSN:0300-9599    

DOI:10.1039/F19757102421

出版商:RSC    

年代:1975

数据来源: RSC

摘要:

BEBO Calculations Part 4.-Arrhenius Parameters and Kinetic Isotope Effects for the Reactions of CH3 and CF3 Radicals with H, and D, BY N. L. ARTHUR,* K. F. DONCHI AND J. A. MCDONELL Department of Physical Chemistry, La Trobe University, Bundoora, Victoria 3083, Australia. Received 26th September, 1974 The modified BEBO (bond energy-bond order) method, in which the Sat0 end-atom triplet repulsion term is replaced by a function fitted to the potential energy values of Hirschfelder and Linnett for the 3Z state of HZ, and the original BEBO method, have been used to calculate Arrhenius parameters and kinetic isotope effects for the reactions of CH3 and CF3 radicals with H2 and D2. The theoretical results are compared with average values deduced from a critical appraisal of the available experimental data.The semiempirical bond energy-bond order (BEBO) method of Johnston and Parr has been employed with some success in predicting activation energies and kinetic isotope effects for a number of gas phase hydrogen transfer reactions. In Parts 1 and 2 of this series,,. the three-point BEBO model was used to calculate activation energies and kinetic isotope effects for hydrogen abstraction from HCl, H2S, H,, SiH4 and NH,, by CH, and CF, radicals. For the NH, reactions, A factors and rate constants were also evaluated. In Part 3,s a modification of the BEBO method (hereafter referred to as BEBOA) was described, in which the Sat0 end-atom triplet repulsion term in the potential energy expression was replaced by a function fitted to the potential energy values proposed by Hirschfelder and Linnett Activation energies for several CH, and CF, radical reactions were calculated by both methods, and in almost every case the BEBOA model gave better agreement with the corresponding experimental value.In this paper we describe the application of both the BEBO and BEBOA models to the reactions of CH3 and CF, radicals with H2 and D2 ; Arrhenius parameters and kinetic isotope effects have been calculated, and the results are compared with the available experimental data. for the 'X state of HZ. THEORY The theory of the BEBO model and its use in the calculation of activated complex properties, Arrhenius parameters, tunnelling corrections and kinetic isotope effects, has been described in Parts 1 and 2. The same methods and notation are used in this paper, but some changes in procedure have been made. All bond dissociation energies at 298 K have been corrected to 0 K by means of the relation Dg = Di98 - RT(n/2+ 1) where n is the difference between the number of vibrational degrees of freedom in the 1-77 243 12432 BEBO C A L CU LA TI 0 N S molecule and its dissociation products.Spectroscopic dissociation energies were then obtained from where E,,, is the zero-point energy. is related to the potential energy of activation, V*, by D, = Dg+EZpe For the linear three-atom model of the activated complex, the activation energy E, E = V * + (6: + 20: - 8; + 39T,R)RT where 0,' and 202 are the corrections for the stretching and two degenerate bending vibrational degrees of freedom of the activated complex, 8,' is the correction for the single vibrational degree of freedom of the reactants, and 30T.R is the correction for the difference between the number of translational and rotational degrees of freedom of the reactants and activated complex.It was pointed out in Part 1 that for a full- atom model of the activated complex, the t9T,R term is larger than that of the three- atom model; for reactions involving CH3 and CF3 radicals it is 50T,R for a linear substrate and 66T,R for a non-linear substrate. Thus the larger model predicts an increase in E of about 4 kJ niol-I for CH3 and CF3 attack on H, or D,, and in this work the 0T.R term has been evaluated from the full-atom model. In calculating the A factor for CH3 and CF3 reactions by the method outlined in Part 2, the H - C-H(F) bending force constant in the activated complex was taken to be equal to that in the stable molecule CH4 or CF3H.This is unlikely to be the case, and in this paper the suggestion of Sharp and J~hnston,~ and Shapiro and Weston,8 that the bending force constant be assigned half its value in the correspond- ing stable molecule, has been adopted. In the calculation of B3, the 23 factor for a linear three-atom system, values of the statistical factor B, and the electronic multiplicity factor Be are required. For the reactions discussed here, B, = 2 and Be = 1. Since the reactions are of the type XY(1inear) + Z(non1inear) -+ [XYZ] * (nonlinear) the Jacobian factor Y used in the calculation of the B factor for the N-atom systcni has the value 1/4z Because XY is linear there are no new rotational interactions resulting from the formation of the activated complex, and therefore the vibrational amplitude factor 2, is 1.END-ATOM TRIPLET REPULSION TERMS The construction of the end-atom triplet repulsion energy terms used in the BEBO has been described in Part 3. The BEBO triplet repulsion energy term 3s given by and BEBOA potential energy expressions for a linear triatomic system [X Y * * 21 V,, = 0.25 Dxz { [(nm)o*2606@ e-BAR + 112 - 1) where Dxz is the spectroscopic dissociation energy of the XZ bond, p is the corres- ponding Morse parameter, n and m are the bond orders of X - - Y and Y - - - Z respec- tively, and AR is related to the single bond distances RXY, RYZ and Rxz, by AR = Rxu+RYZ-RXZ.The BEBOA term is given by V,: = 5.873 D xz [(nm)o.26068 e-BAR' 1 1.747 x {PAR' - ln[(nm)0'2 606P] f lS5 where AR' = RXY+ RYZ, and all other terms have the same meaning as before.N. L . ARTHUR, K . F . DONCHI AND J . A . MCUONELL 2433 ACTIVATED COMPLEX FORCE CONSTANTS Replacement of Ytr in the BEBO potential energy expression by V,: also involves modifying the expressions used to calculate the force constants of the activated com- plex. Fp, the force constant along the reaction path, becomes Fp = -0.1022/[(1/n*)* +(1/rn*)2][p(p- 1)Dxy(n*)p-2+q(q- 1)Dyz(m*)4-2 + 2-0.47s X y(n*n~*)"~~~y-~((1.7472~- 1.525Z)[2n*mq - (1 - 2n*)"( 1.7471, - l)] + 1.525~(1-2~~*)~( 1.7472-0.525))] where X = 5.873 DxZ(2B)1*747, Z = PAR'-yln(n*m*), B = 0.5 e-BAR' and y = 0.26068.F,, the force constant for small displacements perpendicular to the reaction path, is given by F, = l/[(n*)' +(rn*)2](Fxy(rz*)3 +Fyz(m*)3 +[(5.873/2)Fxz Fp and Fa are transformed from Cartesian coordinates to valence bond coordinates giving the force constants F'xf;, F& and F,&, which, together with the bending force constant, x (flR2z)-0*475 e-1-747PRz][ - 1.747PRxf(3.05 - 1.747PR&)+0.8006]). F$ = - (5.8 73 /2>R& R&F,Z(PR&)-' * exp[ - 1 .7478RgZ]( 1.525 - 1.7478R&) are used to calculate the activated complex vibrational frequencies. RESULTS AND DISCUSSION The values of the molecular properties used in these calculations are listed in table 1, and the activated complex properties predicted by the BEBQ and BEBOA methods are given in table 2.TABLE 1 .-INPUT DATA FOR BEBO AND BEBOA CALCULATIONS H-H D-D H-CH3 D-CH3 H-CF3 @/kT mol-i 432.00 a Dgg8/kJ mol-1 DJkJ mol-l 458.1 Rlpm 74.13 FIN cm-l 5.733 Fp/10-18 J w/crn-l 4395 j P 1.052 4 p/108 cm-I 439.53 a 436.0 458.1 447.3 74.14 109.3 5.767f 5.0 u 0.28 3118j 2917 1.052 1.098 1.771 445.2 447.3 457.3 109.5 109.8 5.0 5.0 0.28 0.40 2200 3036 1.099 1.104 1.820 1.823 D-CFJ 457.3 = 109.8 5.0 0.40 2261 1.104 1.847 a B. de B. Darwent, Bond Dissociation Energies in Simple Molecules, 1970, NSRDS-NBS 31 ; b J. C . Amphlett and E. Whittle, Trans. Faraday Soc., 1968, 64, 2130 ; C it is assumed that De(R-D) = De(R--H) ; d Chem. Soc. Spec. Publ., 1958,ll; Chem. SOC. Spec. Publ., 1965,18 ; e T. L. Cottrell, The Strengths of Chemical Bonds (Butterworth, London, 2nd edn., 1950) ; f calculated from F = 5888.3 pw2 ; 9 6.Herzberg, Molecular Spectra and Molecular Structure. II. Infrared and Raman Spectra ofPolyutomic Molecules (Van Nostrand, Princeton, New Jersey, 1945) ; h half the value given by M. J. Kurylo, G. A. Hollinden and R. B. Timmons, J. Chem. Phys., 1970, 52, 1773; ihalf the value given by C. L. Kibby and R. E. Weston, Jr., ref. (9) ; i G. Herzberg, Molecular Spectra and Molecular Structure. I. Spectra of Diatomic Molecules (Van Nostrand, Princeton, New Jersey, 1950) ; k T. Shimanouchi, Tables of Molecular Vibrurional Frequencies, Part 1, 1967, NSRDS-NBS 6 ; Part 3, 1968, NSRDS-NRS 17 ; Part 6, .I. Php. Chem. ReJ Dam, 1973, 2, 121.2434 BEBO C A L CU LA T I 0 N S In Parts 1 and 2, 8;ft and r t , the high temperature tunnelling corrections derived from Bell's truncated parabola, were shown to be satisfactory approximations only when u* < 7z where u* = hcco,/kT and is the reaction coordinate vibrational fre- quency.For most systems this restriction limits the usefulness of the high tempera- ture approximation to the upper end of the usual experimental temperature range, and therefore only 8" and r*, the tunnelling corrections calculated from the unsymmetrical Eckart potential energy function, are included in this paper. TABLE 2.-ACTNATED COMPLEX PROPERTIES BEBO n* 0.478 R & l w 126.2 F$/N cm-1 0.956 F&/N cm-1 I .295 F&/Ncm 1 1.890 ~ , f /lO-lS J rad-2 0.054 w,lcm-l 1537 wblcm-1 649 wilcm-1 1784i Rxfylpm 93.4 BEBOA 0.476 93.5 126.2 0.990 1.360 1.937 0.054 1560 648 17831' BEBO 0.478 93.4 126.4 0.956 1.291 1.881 0.050 1134 443 1270i BEBOA 0.475 93.5 126.3 0.982 1.356 I .894 0.045 1144 423 12451' BEBO 0.517 91.3 128.8 1.253 0.986 1.919 0.049 1552 624 17221' BEBOA 0.519 91.2 128.9 1.316 1.019 I .930 0.044 1575 593 16761' BEBO 0.517 91.3 128.8 1.256 0.989 1.917 0.047 1109 433 1219i ~~ BEBOA 0.520 91.2 128.9 1.319 1.019 1.915 0.041 1123 403 11761' The number of significant figures used in tables 2-4 might at first sight appear to be unrealistically large.It is our experience, however, that to truncate these values would make it unnecessarily difficult for others to repeat our calculations. ARRHENIUS PARAMETERS The terms contributing to the activation energies and pre-exponential factors are listed in table 3, and the rate constants, calculated with and without tunnelling correc- tions in the range 350-600 K, are presented in table 4.The potential energy of V* ORT E (8-t O*)RT E (tunl) log B log B(tun1) log A log A(tun1) log r log r* BEBO BEBOA 62.29 51.68 57.20 46.71 45.85 36.95 1.38 1.37 12.56 12.55 0.64 0.60 13.20 13.15 11.97 11.97 11.29 11.44 -5.10 -4.91 - 16.44 - 14.73 -ARRHENI.US PARAMETERS CH3+D2 CF3+H2 BEBO BEBOA BEBO BEBOA 60.53 48.54 51.17 44.69 59.59 47.56 51.98 39.40 -6.85 -6.34 -15.18 -13.00 53.68 42.20 41.98 31.69 0.88 0.88 1.39 i.39 11.95 11.99 12.35 12.39 0.33 0.31 0.58 0.51 12.28 12.31 12.93 12.90 11.84 11.88 11.74 11.78 11.49 11.57 11.17 11.39 -0.93 -0.99 -5.19 -5.28 BEBO BEBOA 56.25 43.27 -1.11 -1.19 55.14 42.08 -6.45 -5.82 49.81 37.45 0.89 0.89 11.72 11.78 0.30 0.28 12.03 12.06 11.59 11.64 11.28 11.39 a Calculated at 450 K.Preexponential terms in cm3 mol-1 s-1 and energy terms in kJ mol-1. activation, V*, and the activation energy, E, have been previously calculated by the BEBO method for CH3 and CF3 attack on H2, but not on D2. For CH3 + H2, the values of V* obtained were 63,2 56.9 * and 57.6,3 kJ mol-', and the values of E at 500 K were 54.8 and 49.5 kJ mol-l. Our calculations yield Yli: = 62.30 kJ mol-I, and E = 55.12 kJ mol-'. For CF3 +H2, values of 59,2 52.55 and 53.56 kJ rnol-l, have been calculated, and the results for E a t 500 K were 53.40 and 45.56 kJ mol-', which can be compared with our values of V* = 57.15 and E = 52.89 kJ md-l.N. L . ARTHUR, K . F . DONCHI AND J . A . MCDONELL 2435 A factor and rate constant calculations via the BEBO method have not previously been reported for these reactions, although Johnston has graphed, but not tabulated rate constants for CH3 + H2.T / K 3 50 400 450 500 550 600 350 400 450 500 550 600 TABLE 4.-vARIATION OF RATE CONSTANTS WITH TEMPERATURE a CH3SHz CH3+D2 CFs+Hz CF3+D2 BEBO BEBOA BEBO BEBOA BEBO BEBOA BEBO BEBOA without tunnelling correction 3.41 4.97 2.93 4.76 3.96 5.87 3.34 5.35 4.48 5.85 4.04 5.65 4.94 6.61 4.37 6.13 5.33 6.55 4.93 6.36 5.71 7.20 5.19 6.76 6.02 7.12 5.65 6.94 6.34 7.69 5.86 7.28 6.60 7.60 6.25 7.43 6.87 8.10 6.42 7.71 7.09 8.00 6.76 7.84 7.32 8.45 6.90 8.09 with tunnelling correction 4.50 5.94 3.50 5.29 4.93 6.66 3.86 5.81 5.30 6.60 4.47 6.05 5.68 7.24 4.76 6.48 5.97 7.15 5.26 6.67 6.29 7.71 5.50 7.04 6.54 7.61 5.92 7.19 6.82 8.11 6.11 7.51 7.03 8.01 6.47 7.64 7.26 8.45 6.62 7.91 7.45 8.35 6.95 8.02 7.65 8.75 7.07 8.25 a k in cm3 mol-' s-l units.From the variety of the Vf and E values given above it can be seen that differences in input data lead to substantial differences in the results of BEBO calculations. The quantitative effect of varying some of the molecular parameters invoived was discussed in some detail in Parts 1 and 2. In this work, particular care has been taken in the selection of the data required for the calculations, and the most reliable, and usually the most recent, values have been used. Because of the considerable spread of the available Arrhenius parameters for the reactions of CH3, CD3 and CF3 radicals with €3, and D,, it was necessary to under- take a critical evaluation of the experimental data before a comparison could be made with the results of the BEBO and BEBOA calculations.In most of the experimental work considered here, rate constants for abstraction were measured relative to those for radical recombination. Rate constants for CH, and CF3 recombination have been measured many times, originally by the rotating sector technique,l0* and the radical buffer method of Hiatt and Benson.I7V Workers in the field of H-abstraction reactions have, however, been reluctant to adopt new values for these important rate constants, and we have therefore retained the values of Shepp lo for CH3 (1013-34 cm3 md-I s-l) and Ayscough l 1 for CF3 (1013*36 cm3 mol-1 s-'). Since these values differ only slightly from the averages of those reported recently (1013-55 for CH3 and 1012.86 for CF3), future changes in the accepted values are unlikely to affect the conclusions reached in this paper.After the studies involving radical sources other than acetone and L2H6]acetone, and the data (obtained at only two temperatures) of McNesby et ciZ.,19 had been eliminated, there remained six sets of data for CH,/CD3 + H2 and six for CH3/CD3 + D2. Fortunately, the original rate constant against temperature data were available in each case, but some rate constants had been measured relative to the rate of radical recombination, others to the rate of abstraction from the radical source. In several and most recently by kinetic spectroscopy2436 BEBO C A L CU LA TI ON S studies, Arrhenius parameters for abstraction from the radical source were not measured independently in the same system, and in recalculating the original data, we have used average values deduced from the compilations of Trotman-Dickenson and Milne,20 and Gray, Herod and Jones.21 These values were : CH, + CH3COCH3, A = cm3 mol-I s-I, E = 40.29 kJ mol-1 ; CD,+CD,COCD,, A = J011a6’ cm3 mol-1 s-I, E = 47.99 kJ mol-l.Inspection of the combined data showed that the four points at 566-571 K of Majury and Steacie 22 deviated widely from the remainder of the data, and therefore they were neglected. Some of the data of Davison and Burton 2 3 involved consider- able decomposition of the substrate; the suggestion of Wijnen and Steaciz 24 was adopted whereby those data for which decomposition was greater than 10 % were omitted.Arrhenius parameters for each set of data, and all of the data combined, were evaluated by means of a conventional least squares procedure. This simple statistical treatment of the data follows the usual practice in experimental hydrogen abstraction work, but it almost certainly underestimates the error limits associated with the Arrhenius parameters; more realistic systematic error limits are closer to +2 kJ mol-1 in E and k0.3 in log A . The values obtained are fisted in table 5, the errors quoted being standard deviations. Individual data points are depicted in fig. 1 ; for the sake of clarity, only the data of Shapiro and Weston, and Majury and Steacie, have been included. TABLE 5.-EXPERIMENTAL ARRHENIUS PARAMETERS Hz D2 radical log(A/cm3 mol s-1) E/kJ mol-1 log(A/cmJ mol s-1) E/kJ mol-1 ref.CH3 11.39f0.08 CD3 11.67f0.08 CH3 11.993.0.03 CH3 1 1.65 f 0.07 CD3 11.76-& 0.03 CH3 11.863. 0.14 CHS/CD3 1 1.73 f 0.06 CF3 11.97$- 0.08 CF3 11.90f0.20 CF3 1 1.95 & 0.36 41.1 5 f 0.74 42.76k0.76 46.26k0.32 42.77f0.64 43.58f0.29 44.46k 1.31 43.51 f0.54 40.84k0.61 40.29f 1.48 39.71 f 2.66 11.67k0.07 11.44f 0.13 1 1.84+ 0.07 11.83k0.03 11.56k0.04 11.91 & 0.10 1 1.77 f 0.06 12.06f 0.07 11.47f0.10 11.86k0.12 48.95+ 0.61 45.35+ 1.16 49.78 & 0.65 49.48f 0.29 46.06k0.33 55.442 1.09 49.01 +0.58 46.75+ 0.54 42.65 + 0.81 45.35k0.94 22 22 25 ’ 8 8 23 mean 9 26 mean a Reactions studied were CH3 + Dz and CD3 + HZ. There have been three determinations of the Arrhenius parameters for CF3 + H2 and two for CF3 + D2.As can be seen from fig. 2, the results of Ayscough and Polanyi 26 and Kibby and Weston are in good agreement at temperatures below 500 K, but above this temperature the latter’s data curve steeply upward. The data of Fagarash, Moin and Ocheret’ko 27 for CF3 + H2 also deviate considerably from the line established by the lower temperature points, and their data, together with the higher temperature values of Kibby and Weston, were ignored in evaluating the mean Arrhenius parameters listed in table 5. The calculated Arrhenius parameters are compared with the experimental data in table 6 and in fig. 1 and 2. If the BEBO and BEBOA methods are compared solely in terms of the success of their predictions of activation energies, the BEBOA model without correction for tunnelling, is clearly superior for all reactions except CH3 + H1, for which the BEBO(tun1) value of 45.86 kJ mol-l differs only slightly from the BEBOA value of 46.69 kJ mol-l. The differences between the BEBOA and experi- mental activation energies lie between 0.31 kJ mol-1 for CF3 + H2 and 3.27 kJ mol-1N.L . ARTHUR, K . F . DONCHI AND J. A . MCDONELL 2437 for CF3 + D2, and, except for CH3 + H2, all the theoretical activation energies are lower than the corresponding experimental values. TABLE 6.-COMPARISON OF ARRHENIUS PARAMETERS AT 450 K expt.(mcan) log(A/cm3 mol-I s-I) 11.73 EIkJ mol-1 43.51 log(k/cm3 mol-' s-l) 6.68 1 og( A /cm3 mol- s- ) 11.77 E/kJ mo1-1 49.01 log(k/cm3 mo1-1 s-l) 6.08 log(A/cm3 mo1-' s-l) 11.95 E/kJ mol-' 39.71 log(k/cm3 rno1-I s-l) 7.34 log(A/cm3 mol-' s-l) 11.86 log(k/cm3 mol-1 s-l) 6.60 Elk3 mo1-1 45.35 BEBO CH3+H2 1 1.97 57.20 5.33 CH3 + D2 I I .84 59.59 4.93 CF3 + HZ 11.74 51.98 5.71 CF3 + D2 11.59 55.14 5.19 BEBOA 11.97 46.71 6.55 11 .S5 47.56 6.36 11.78 39.40 7.20 11.64 42.08 6.76 BEBO(tun1) 11.29 45.85 5.94 11.49 53.68 5.26 11.17 41.98 6.29 11.28 49.81 5.50 BEBOA(tun1) 11.44 36.95 7.15 11.57 42.20 6.67 11.39 31.69 7.71 11.39 37.45 7.04 1.4 t :e 2.2 2.6 lo3 KIT FIG.1 .-Arrhenius plot of calculated and experimental rate constants for CH3 /CD3 + H2 and CP13/CD3 + Dz. Curve A, H2, BEBOA(tun1) ; curve B, D2, BEBOA(tun1) ; curve C, Hz, BEBOA ; curve D, D2, BEBOA ; curve E, H2, BEBO(tun1) ; curve F, H2, BEBO ; curve G, D2, BEBO(tun1) ; curve H, D2, BEBO.Experimental points : 0, CH3+Hz or D2, a, CD3+H2 or Dz, ref. (8); 0, CH3+H2 or Dz, A, CD3fHz or D2, ref. (22) ; in each case the upper points refer to H2 and the Iower to Dz.2438 BEBO C A L CU LATT ONS There is little to choose between the two models in their ability to predict A factors. Agreement between the calculated and mean experimental values is good ; A factors for the CH, reactions are slightly higher, while those for the CF, reactions are a little lower, than the corresponding experimental values. I I I I 2 . 0 2.4 2.8 103 KIT FIG. 2.-Arrhenius plot of calculated and experimental rate constants for CF3 + Hz and CF3 + D2. Curve A, H2, BEBOA(tun1) ; curve 8, H2, BEBOA ; curve C, D2, BEBOA(tun1) ; curve D, DZ, BEBOA ; curve E, H2, BEBO(tun1) ; curve F, H2, BEBO ; curve G, Dz, BEBO(tun1) ; curve H, D2, BEBO.Experimental points for CF3+H2 or D2 : 0, ref. (9); 0, ref. (26); A, ref. (27). In each case the upper points refer to H, and the lower to D,. From fig. 1 and 2 it can be seen that BEBOA is much the better of the two models in predicting rate constants. This follows from the good correlation between the BEBOA and experimental activation energies. When tunnelling corrections are included, the BEBOA curves are shifted upwards and away from the experimental points. It is interesting to compare the relative reactivities of CH, and CF, radicals in these reactions with those deduced from the results of the BEBO and BEBOA calcula- tions. As might be anticipated for non-polar molecules, both H, and D, are more susceptible to CF3 than to CH, attack, the ratio of rate constants, kCF3/kCH3, being 4.6 for H2 and 3.3 for D2.Again the BEBOA model offers the better correlation; the values of kCF3/kCH3 are 4.5 for H, and 2.5 for D,, compared with the BEBO values of 2.4 and 1.8. The inclusion of tunnelling corrections results in lower values of k,,,/k,,"j and hence poor agieement with the experimental rate constant ratios. KINETIC ISOTOPE EFFECTS Kinetic isotope effects, kH/kD, calculated from both the BEBO and BEBOA models, are compared with the experimental data in table 7 and fig. 3. The alterna- tive method of calculating kinetic isotope effects, in which q is set equal to p and varied until E equals E(expt.), was also tried. The results of this procedure differed negligibly from the values given in table 7.Included in the table are isotope effectsN . L . ARTHUR, K . F . DONCHI AND J . A . MCDONELL 2439 calculated from the simple model which assumes that only the zero-point energy of the stretching vibration of the XY bond is lost, i.e., kH/kD(zpe) = exp[(Ezpe(H) - Ezpe(D))/kTl. The values of kH/kD(expt.) in the table were evaluated from the mean Arrhenius para- meters in table 5, and are plotted as dotted lines. The individual points in the figure refer to the results of competitive experiments in which k,/kD values were measured directly. T/ 1c 350 400 450 500 550 600 3 50 400 450 500 550 600 k H / kD(eXpt.) 5.99 4.74 3.95 3.41 3.02 2.74 8.57 6.72 5.57 4.79 4.23 3.82 TABLE 7.-KINETIC ISOTOPE EFFECTS k d k n kH/kD(tud) ~ - _ _ _ _ krr/k&Pe) 13.80 9.94 7.70 6.28 5.3 1 4.62 13.80 9.94 7.70 6.28 5.31 4.62 BEBO CH3 + Hz 6.15 5.19 4.51 4.02 3.64 3.34 CF3 + Hz 5.97 5.07 4.43 3.96 3.60 3.31 BEBOA 6.09 5.14 4.47 3.99 3.61 3.32 6.03 5.10 4.45 3.97 3.61 3.32 BEBO 19.94 12.63 9.07 7.06 5.80 4.95 16.88 11.27 8.36 6.65 5.55 4.78 BEBOA 15.73 10.93 8.27 6.64 5.57 4.81 12.63 9.37 7.40 6.11 5.23 4.59 The difference between the lower (CH3 + Hz /D2) and upper (CF3 + H2 /D2) experi- mental curves stems from very small variations in A factor ratios, AH/AD, and activa- tion energy differences, ED-EH : for the CH, reactions, &/AD = 0.91 and E D - E H = 5.48 kJ mol-', and for the CF, reactions, A H / A D = 1.23 and ED -EH = 5.65 kJ mol-'.These results are equal, within the error liinits usually found in experiments of this kind, and the discrepancy between the curves illustrates the inadequacy of the avail- able experimental data for the evaluation of accurate kinetic isotope effects.It should be pointed out, however, that the kH/kD curve for CF3 + HJD, deduced from the Arrhenius parameters of Kibby and We~ton,~ coincides almost exactly with the mean curve for CH, + H2/D2, and the difference between the two mean experimental curves is the result of the contribution made to the mean Arrhenius parameters for CF3 + D, by the data of Ayscough and P ~ l a n y i , ' ~ data which by themselves yield the rather unlikely values, AH/& = 2.69 and E D - E H = 2.34 kJ mol-'. On this analysis, we regard the lower experimental curve as the more reliable of the two, and it is noticeable that it is surrounded by the bulk of the individual experimental points.It can be seen from table 7 that the kinetic isotope effects (without tunnelling corrections) predicted by the BEBO and BEBOA methods are almost identical, and, as might be expected, there is very little difference between the values for the reactions involving CH, and CF3 radicals. These four sets of kH/kD values have therefore been represented by a single curve in fig. 3. By contrast, there are substantial differences between the kH/kD(tunl) values for the two models and the two radicals : the BEBOA values are lower than those deduced from the BEBO model, and the CF, radical results are lower than those for CH, radical attack. The theoretical isotope effect2440 BEBO C A L C UL A TI ONS curve without tunnelling is much closer to the recommended experimental curve and the individual experimental points than are any of the curves in which tunnelling corrections are included.The k,/k,,(zpe) curve is much higher than the experimental results, and the BEBO and BEBOA models are clearly superior to the simple model in predicting kinetic isotope effects. 12.0 - x 8 . 0 - 1.8 2.2 2.6 3.0 103 KIT FIG. 3.--Kinetic isotope effects as a function of temperature. Curve A, CH3, BEBO(tun1) ; curve B, CF3, BEBO(tun1) ; curve C, CH3, BEBOA(tun1) ; curve D, zpe ; curve E, CF3, BEBOA(tun1) ; curve F, CF3(mean expt); curve G, CH3/CF3, BEBO/BEBOA; curve H, CH3 (mean expt.). Experimental points: 0, CH3, ref. (8); 0, CD3, ref.(8); @, CF3, ref. (9); A, CH,, ref. (23); V, CF3, ref. (28). CONCLUSIONS The variation of the BEBO model described here provides a much more satisfactory correlation with the experimental Arrhenius parameters for the reactions of CH, and CF, radicals with H2 and D2 than the original BEBO method ; kinetic isotope effects are equally well predicted by either method. On the basis of the three-point BEBOA model, quantum mechanical tunnelling does not appear to play a significant part in these reactions. We thank Dr. J. R. Christie of La Trobe for helpful discussion and J. A. McDonell is indebted to La Trobe University for a research scholarship. H. S. Johnston and C. Parr, J. Amer. Chem. SOC., 1963,85,2544. H. S. Johnston, Gas Phase Reaction Rate Theory (Ronald, New York, 1966).N. L. Arthur and J. A. McDonnell, J. Chem Phys., 1972, 56, 3100. N. L. Arthur and J. A. McDonell, J. Chem. Phys., 1972, 57, 3228. N. L. Arthur, K. F. Donchi and J. A. McDonell, J. Chem. Phys., 1975,62,1585. J. 0. Hirschfelder, C. F. Curtis and R. B. Bird, Molecular Theory of Gases and Liquids (Wiley, New York, 1954). ' T. E. Sharp and H. S. Johnston, J. Chem. Phys., 1962,37, 1541. * J. S. Shapiro and R. E. Weston, Jr., J. Phys. Chem., 1972,76, 1669.N. L . ARTHUR, K . F . DONCHI AND J . A . MCDONELL 244 1 C. L. Kibby and R. E. Weston, Jr., J. Chem. Phys., 1968,49,4825. P . B. Ayscough, J. Chem. Phys., 1956,24,994. lo A. Shepp, J. Chem. Phys., 1956,24,939. l2 N. Basco, D. G. L. James and R. D. Stuart, Int. J. Chem. Kinetics, 1970,2,215. l 3 F . K.Truby and J. K. Rice, Int. J. Chem. Kinetics, 1973, 5,721. l4 A. M. Bass and A. H. Laufer, Int. J. Chem. Kinetics, 1973, 5, 1053. l5 T. Ogawa, G. A. Carlson and G. C. Pimentel, J. Phys. Chem., 1970,74,2090. l6 N. Basco and F. G. M. Hathorn, Chem. Phys. Letters, 1971, 8, 291. l7 R. Hiatt and S. W. Benson, J. Amer. Chem. SOC., 1972, 94, 25. lS R. Hiatt and S. W. Benson, Int. J. Chem. Kinetics, 1972,4,479. l9 J. R. McNesby, A. S. Gordon and S. R. Smith, J. Amer. Chem. Soc., 1956,78, 1287. 2o A. F. Trotman-Dickenson and G. S. Milne, Tables of Bimolecular Gas Reactions, 1967, NSRDS- 21 P. Gray, A. A. Herod and A. Jones, Chem. Reu., 1971,71,247. 22 T. G. Majury and E. W. R. Steacie, Canud. J. Chem., 1952,30, 800. 23 S. Davison and M. Burton, J. Amer. Chem. Soc., 1952, 74, 2307.24 M. H. J. Wijnen and E. W. R. Steacie, Disc. Faraday SOC., 1953,14, 118. 25 E. Whittle and E. W. R. Steacie, J. Chem. Phys., 1953, 21, 993. 26 P. B. Ayscough and J. C. Polanyi, Trans. Faraduy SOC., 1956, 52, 960. 27 M. B. Fagarash, F. B. Moin and V. I. Ocheret’ko, Kinetics and Catalysis, 1968, 9, 762. 28 G. 0. Pritchard, H. 0. Pritchard, H. I. Schiff and A. F. Trotman-Dickenson, Trans. Faraday NBS 9. Soc., 1956, 52, 849. BEBO Calculations Part 4.-Arrhenius Parameters and Kinetic Isotope Effects for the Reactions of CH3 and CF3 Radicals with H, and D, BY N. L. ARTHUR,* K. F. DONCHI AND J. A. MCDONELL Department of Physical Chemistry, La Trobe University, Bundoora, Victoria 3083, Australia. Received 26th September, 1974 The modified BEBO (bond energy-bond order) method, in which the Sat0 end-atom triplet repulsion term is replaced by a function fitted to the potential energy values of Hirschfelder and Linnett for the 3Z state of HZ, and the original BEBO method, have been used to calculate Arrhenius parameters and kinetic isotope effects for the reactions of CH3 and CF3 radicals with H2 and D2.The theoretical results are compared with average values deduced from a critical appraisal of the available experimental data. The semiempirical bond energy-bond order (BEBO) method of Johnston and Parr has been employed with some success in predicting activation energies and kinetic isotope effects for a number of gas phase hydrogen transfer reactions. In Parts 1 and 2 of this series,,. the three-point BEBO model was used to calculate activation energies and kinetic isotope effects for hydrogen abstraction from HCl, H2S, H,, SiH4 and NH,, by CH, and CF, radicals.For the NH, reactions, A factors and rate constants were also evaluated. In Part 3,s a modification of the BEBO method (hereafter referred to as BEBOA) was described, in which the Sat0 end-atom triplet repulsion term in the potential energy expression was replaced by a function fitted to the potential energy values proposed by Hirschfelder and Linnett Activation energies for several CH, and CF, radical reactions were calculated by both methods, and in almost every case the BEBOA model gave better agreement with the corresponding experimental value. In this paper we describe the application of both the BEBO and BEBOA models to the reactions of CH3 and CF, radicals with H2 and D2 ; Arrhenius parameters and kinetic isotope effects have been calculated, and the results are compared with the available experimental data.for the 'X state of HZ. THEORY The theory of the BEBO model and its use in the calculation of activated complex properties, Arrhenius parameters, tunnelling corrections and kinetic isotope effects, has been described in Parts 1 and 2. The same methods and notation are used in this paper, but some changes in procedure have been made. All bond dissociation energies at 298 K have been corrected to 0 K by means of the relation Dg = Di98 - RT(n/2+ 1) where n is the difference between the number of vibrational degrees of freedom in the 1-77 243 12432 BEBO C A L CU LA TI 0 N S molecule and its dissociation products.Spectroscopic dissociation energies were then obtained from where E,,, is the zero-point energy. is related to the potential energy of activation, V*, by D, = Dg+EZpe For the linear three-atom model of the activated complex, the activation energy E, E = V * + (6: + 20: - 8; + 39T,R)RT where 0,' and 202 are the corrections for the stretching and two degenerate bending vibrational degrees of freedom of the activated complex, 8,' is the correction for the single vibrational degree of freedom of the reactants, and 30T.R is the correction for the difference between the number of translational and rotational degrees of freedom of the reactants and activated complex. It was pointed out in Part 1 that for a full- atom model of the activated complex, the t9T,R term is larger than that of the three- atom model; for reactions involving CH3 and CF3 radicals it is 50T,R for a linear substrate and 66T,R for a non-linear substrate.Thus the larger model predicts an increase in E of about 4 kJ niol-I for CH3 and CF3 attack on H, or D,, and in this work the 0T.R term has been evaluated from the full-atom model. In calculating the A factor for CH3 and CF3 reactions by the method outlined in Part 2, the H - C-H(F) bending force constant in the activated complex was taken to be equal to that in the stable molecule CH4 or CF3H. This is unlikely to be the case, and in this paper the suggestion of Sharp and J~hnston,~ and Shapiro and Weston,8 that the bending force constant be assigned half its value in the correspond- ing stable molecule, has been adopted.In the calculation of B3, the 23 factor for a linear three-atom system, values of the statistical factor B, and the electronic multiplicity factor Be are required. For the reactions discussed here, B, = 2 and Be = 1. Since the reactions are of the type XY(1inear) + Z(non1inear) -+ [XYZ] * (nonlinear) the Jacobian factor Y used in the calculation of the B factor for the N-atom systcni has the value 1/4z Because XY is linear there are no new rotational interactions resulting from the formation of the activated complex, and therefore the vibrational amplitude factor 2, is 1. END-ATOM TRIPLET REPULSION TERMS The construction of the end-atom triplet repulsion energy terms used in the BEBO has been described in Part 3.The BEBO triplet repulsion energy term 3s given by and BEBOA potential energy expressions for a linear triatomic system [X Y * * 21 V,, = 0.25 Dxz { [(nm)o*2606@ e-BAR + 112 - 1) where Dxz is the spectroscopic dissociation energy of the XZ bond, p is the corres- ponding Morse parameter, n and m are the bond orders of X - - Y and Y - - - Z respec- tively, and AR is related to the single bond distances RXY, RYZ and Rxz, by AR = Rxu+RYZ-RXZ. The BEBOA term is given by V,: = 5.873 D xz [(nm)o.26068 e-BAR' 1 1.747 x {PAR' - ln[(nm)0'2 606P] f lS5 where AR' = RXY+ RYZ, and all other terms have the same meaning as before.N. L . ARTHUR, K . F . DONCHI AND J . A . MCUONELL 2433 ACTIVATED COMPLEX FORCE CONSTANTS Replacement of Ytr in the BEBO potential energy expression by V,: also involves modifying the expressions used to calculate the force constants of the activated com- plex. Fp, the force constant along the reaction path, becomes Fp = -0.1022/[(1/n*)* +(1/rn*)2][p(p- 1)Dxy(n*)p-2+q(q- 1)Dyz(m*)4-2 + 2-0.47s X y(n*n~*)"~~~y-~((1.7472~- 1.525Z)[2n*mq - (1 - 2n*)"( 1.7471, - l)] + 1.525~(1-2~~*)~( 1.7472-0.525))] where X = 5.873 DxZ(2B)1*747, Z = PAR'-yln(n*m*), B = 0.5 e-BAR' and y = 0.26068.F,, the force constant for small displacements perpendicular to the reaction path, is given by F, = l/[(n*)' +(rn*)2](Fxy(rz*)3 +Fyz(m*)3 +[(5.873/2)Fxz Fp and Fa are transformed from Cartesian coordinates to valence bond coordinates giving the force constants F'xf;, F& and F,&, which, together with the bending force constant, x (flR2z)-0*475 e-1-747PRz][ - 1.747PRxf(3.05 - 1.747PR&)+0.8006]).F$ = - (5.8 73 /2>R& R&F,Z(PR&)-' * exp[ - 1 .7478RgZ]( 1.525 - 1.7478R&) are used to calculate the activated complex vibrational frequencies. RESULTS AND DISCUSSION The values of the molecular properties used in these calculations are listed in table 1, and the activated complex properties predicted by the BEBQ and BEBOA methods are given in table 2. TABLE 1 .-INPUT DATA FOR BEBO AND BEBOA CALCULATIONS H-H D-D H-CH3 D-CH3 H-CF3 @/kT mol-i 432.00 a Dgg8/kJ mol-1 DJkJ mol-l 458.1 Rlpm 74.13 FIN cm-l 5.733 Fp/10-18 J w/crn-l 4395 j P 1.052 4 p/108 cm-I 439.53 a 436.0 458.1 447.3 74.14 109.3 5.767f 5.0 u 0.28 3118j 2917 1.052 1.098 1.771 445.2 447.3 457.3 109.5 109.8 5.0 5.0 0.28 0.40 2200 3036 1.099 1.104 1.820 1.823 D-CFJ 457.3 = 109.8 5.0 0.40 2261 1.104 1.847 a B.de B. Darwent, Bond Dissociation Energies in Simple Molecules, 1970, NSRDS-NBS 31 ; b J. C . Amphlett and E. Whittle, Trans. Faraday Soc., 1968, 64, 2130 ; C it is assumed that De(R-D) = De(R--H) ; d Chem. Soc. Spec. Publ., 1958,ll; Chem. SOC. Spec. Publ., 1965,18 ; e T. L. Cottrell, The Strengths of Chemical Bonds (Butterworth, London, 2nd edn., 1950) ; f calculated from F = 5888.3 pw2 ; 9 6. Herzberg, Molecular Spectra and Molecular Structure. II. Infrared and Raman Spectra ofPolyutomic Molecules (Van Nostrand, Princeton, New Jersey, 1945) ; h half the value given by M. J. Kurylo, G. A. Hollinden and R. B. Timmons, J. Chem. Phys., 1970, 52, 1773; ihalf the value given by C.L. Kibby and R. E. Weston, Jr., ref. (9) ; i G. Herzberg, Molecular Spectra and Molecular Structure. I. Spectra of Diatomic Molecules (Van Nostrand, Princeton, New Jersey, 1950) ; k T. Shimanouchi, Tables of Molecular Vibrurional Frequencies, Part 1, 1967, NSRDS-NBS 6 ; Part 3, 1968, NSRDS-NRS 17 ; Part 6, .I. Php. Chem. ReJ Dam, 1973, 2, 121.2434 BEBO C A L CU LA T I 0 N S In Parts 1 and 2, 8;ft and r t , the high temperature tunnelling corrections derived from Bell's truncated parabola, were shown to be satisfactory approximations only when u* < 7z where u* = hcco,/kT and is the reaction coordinate vibrational fre- quency. For most systems this restriction limits the usefulness of the high tempera- ture approximation to the upper end of the usual experimental temperature range, and therefore only 8" and r*, the tunnelling corrections calculated from the unsymmetrical Eckart potential energy function, are included in this paper.TABLE 2.-ACTNATED COMPLEX PROPERTIES BEBO n* 0.478 R & l w 126.2 F$/N cm-1 0.956 F&/N cm-1 I .295 F&/Ncm 1 1.890 ~ , f /lO-lS J rad-2 0.054 w,lcm-l 1537 wblcm-1 649 wilcm-1 1784i Rxfylpm 93.4 BEBOA 0.476 93.5 126.2 0.990 1.360 1.937 0.054 1560 648 17831' BEBO 0.478 93.4 126.4 0.956 1.291 1.881 0.050 1134 443 1270i BEBOA 0.475 93.5 126.3 0.982 1.356 I .894 0.045 1144 423 12451' BEBO 0.517 91.3 128.8 1.253 0.986 1.919 0.049 1552 624 17221' BEBOA 0.519 91.2 128.9 1.316 1.019 I .930 0.044 1575 593 16761' BEBO 0.517 91.3 128.8 1.256 0.989 1.917 0.047 1109 433 1219i ~~ BEBOA 0.520 91.2 128.9 1.319 1.019 1.915 0.041 1123 403 11761' The number of significant figures used in tables 2-4 might at first sight appear to be unrealistically large.It is our experience, however, that to truncate these values would make it unnecessarily difficult for others to repeat our calculations. ARRHENIUS PARAMETERS The terms contributing to the activation energies and pre-exponential factors are listed in table 3, and the rate constants, calculated with and without tunnelling correc- tions in the range 350-600 K, are presented in table 4. The potential energy of V* ORT E (8-t O*)RT E (tunl) log B log B(tun1) log A log A(tun1) log r log r* BEBO BEBOA 62.29 51.68 57.20 46.71 45.85 36.95 1.38 1.37 12.56 12.55 0.64 0.60 13.20 13.15 11.97 11.97 11.29 11.44 -5.10 -4.91 - 16.44 - 14.73 -ARRHENI.US PARAMETERS CH3+D2 CF3+H2 BEBO BEBOA BEBO BEBOA 60.53 48.54 51.17 44.69 59.59 47.56 51.98 39.40 -6.85 -6.34 -15.18 -13.00 53.68 42.20 41.98 31.69 0.88 0.88 1.39 i.39 11.95 11.99 12.35 12.39 0.33 0.31 0.58 0.51 12.28 12.31 12.93 12.90 11.84 11.88 11.74 11.78 11.49 11.57 11.17 11.39 -0.93 -0.99 -5.19 -5.28 BEBO BEBOA 56.25 43.27 -1.11 -1.19 55.14 42.08 -6.45 -5.82 49.81 37.45 0.89 0.89 11.72 11.78 0.30 0.28 12.03 12.06 11.59 11.64 11.28 11.39 a Calculated at 450 K.Preexponential terms in cm3 mol-1 s-1 and energy terms in kJ mol-1. activation, V*, and the activation energy, E, have been previously calculated by the BEBO method for CH3 and CF3 attack on H2, but not on D2. For CH3 + H2, the values of V* obtained were 63,2 56.9 * and 57.6,3 kJ mol-', and the values of E at 500 K were 54.8 and 49.5 kJ mol-l.Our calculations yield Yli: = 62.30 kJ mol-I, and E = 55.12 kJ mol-'. For CF3 +H2, values of 59,2 52.55 and 53.56 kJ rnol-l, have been calculated, and the results for E a t 500 K were 53.40 and 45.56 kJ mol-', which can be compared with our values of V* = 57.15 and E = 52.89 kJ md-l.N. L . ARTHUR, K . F . DONCHI AND J . A . MCDONELL 2435 A factor and rate constant calculations via the BEBO method have not previously been reported for these reactions, although Johnston has graphed, but not tabulated rate constants for CH3 + H2. T / K 3 50 400 450 500 550 600 350 400 450 500 550 600 TABLE 4.-vARIATION OF RATE CONSTANTS WITH TEMPERATURE a CH3SHz CH3+D2 CFs+Hz CF3+D2 BEBO BEBOA BEBO BEBOA BEBO BEBOA BEBO BEBOA without tunnelling correction 3.41 4.97 2.93 4.76 3.96 5.87 3.34 5.35 4.48 5.85 4.04 5.65 4.94 6.61 4.37 6.13 5.33 6.55 4.93 6.36 5.71 7.20 5.19 6.76 6.02 7.12 5.65 6.94 6.34 7.69 5.86 7.28 6.60 7.60 6.25 7.43 6.87 8.10 6.42 7.71 7.09 8.00 6.76 7.84 7.32 8.45 6.90 8.09 with tunnelling correction 4.50 5.94 3.50 5.29 4.93 6.66 3.86 5.81 5.30 6.60 4.47 6.05 5.68 7.24 4.76 6.48 5.97 7.15 5.26 6.67 6.29 7.71 5.50 7.04 6.54 7.61 5.92 7.19 6.82 8.11 6.11 7.51 7.03 8.01 6.47 7.64 7.26 8.45 6.62 7.91 7.45 8.35 6.95 8.02 7.65 8.75 7.07 8.25 a k in cm3 mol-' s-l units.From the variety of the Vf and E values given above it can be seen that differences in input data lead to substantial differences in the results of BEBO calculations.The quantitative effect of varying some of the molecular parameters invoived was discussed in some detail in Parts 1 and 2. In this work, particular care has been taken in the selection of the data required for the calculations, and the most reliable, and usually the most recent, values have been used. Because of the considerable spread of the available Arrhenius parameters for the reactions of CH3, CD3 and CF3 radicals with €3, and D,, it was necessary to under- take a critical evaluation of the experimental data before a comparison could be made with the results of the BEBO and BEBOA calculations. In most of the experimental work considered here, rate constants for abstraction were measured relative to those for radical recombination. Rate constants for CH, and CF3 recombination have been measured many times, originally by the rotating sector technique,l0* and the radical buffer method of Hiatt and Benson.I7V Workers in the field of H-abstraction reactions have, however, been reluctant to adopt new values for these important rate constants, and we have therefore retained the values of Shepp lo for CH3 (1013-34 cm3 md-I s-l) and Ayscough l 1 for CF3 (1013*36 cm3 mol-1 s-').Since these values differ only slightly from the averages of those reported recently (1013-55 for CH3 and 1012.86 for CF3), future changes in the accepted values are unlikely to affect the conclusions reached in this paper. After the studies involving radical sources other than acetone and L2H6]acetone, and the data (obtained at only two temperatures) of McNesby et ciZ.,19 had been eliminated, there remained six sets of data for CH,/CD3 + H2 and six for CH3/CD3 + D2.Fortunately, the original rate constant against temperature data were available in each case, but some rate constants had been measured relative to the rate of radical recombination, others to the rate of abstraction from the radical source. In several and most recently by kinetic spectroscopy2436 BEBO C A L CU LA TI ON S studies, Arrhenius parameters for abstraction from the radical source were not measured independently in the same system, and in recalculating the original data, we have used average values deduced from the compilations of Trotman-Dickenson and Milne,20 and Gray, Herod and Jones.21 These values were : CH, + CH3COCH3, A = cm3 mol-I s-I, E = 40.29 kJ mol-1 ; CD,+CD,COCD,, A = J011a6’ cm3 mol-1 s-I, E = 47.99 kJ mol-l.Inspection of the combined data showed that the four points at 566-571 K of Majury and Steacie 22 deviated widely from the remainder of the data, and therefore they were neglected. Some of the data of Davison and Burton 2 3 involved consider- able decomposition of the substrate; the suggestion of Wijnen and Steaciz 24 was adopted whereby those data for which decomposition was greater than 10 % were omitted. Arrhenius parameters for each set of data, and all of the data combined, were evaluated by means of a conventional least squares procedure. This simple statistical treatment of the data follows the usual practice in experimental hydrogen abstraction work, but it almost certainly underestimates the error limits associated with the Arrhenius parameters; more realistic systematic error limits are closer to +2 kJ mol-1 in E and k0.3 in log A .The values obtained are fisted in table 5, the errors quoted being standard deviations. Individual data points are depicted in fig. 1 ; for the sake of clarity, only the data of Shapiro and Weston, and Majury and Steacie, have been included. TABLE 5.-EXPERIMENTAL ARRHENIUS PARAMETERS Hz D2 radical log(A/cm3 mol s-1) E/kJ mol-1 log(A/cmJ mol s-1) E/kJ mol-1 ref. CH3 11.39f0.08 CD3 11.67f0.08 CH3 11.993.0.03 CH3 1 1.65 f 0.07 CD3 11.76-& 0.03 CH3 11.863. 0.14 CHS/CD3 1 1.73 f 0.06 CF3 11.97$- 0.08 CF3 11.90f0.20 CF3 1 1.95 & 0.36 41.1 5 f 0.74 42.76k0.76 46.26k0.32 42.77f0.64 43.58f0.29 44.46k 1.31 43.51 f0.54 40.84k0.61 40.29f 1.48 39.71 f 2.66 11.67k0.07 11.44f 0.13 1 1.84+ 0.07 11.83k0.03 11.56k0.04 11.91 & 0.10 1 1.77 f 0.06 12.06f 0.07 11.47f0.10 11.86k0.12 48.95+ 0.61 45.35+ 1.16 49.78 & 0.65 49.48f 0.29 46.06k0.33 55.442 1.09 49.01 +0.58 46.75+ 0.54 42.65 + 0.81 45.35k0.94 22 22 25 ’ 8 8 23 mean 9 26 mean a Reactions studied were CH3 + Dz and CD3 + HZ.There have been three determinations of the Arrhenius parameters for CF3 + H2 and two for CF3 + D2. As can be seen from fig. 2, the results of Ayscough and Polanyi 26 and Kibby and Weston are in good agreement at temperatures below 500 K, but above this temperature the latter’s data curve steeply upward. The data of Fagarash, Moin and Ocheret’ko 27 for CF3 + H2 also deviate considerably from the line established by the lower temperature points, and their data, together with the higher temperature values of Kibby and Weston, were ignored in evaluating the mean Arrhenius parameters listed in table 5.The calculated Arrhenius parameters are compared with the experimental data in table 6 and in fig. 1 and 2. If the BEBO and BEBOA methods are compared solely in terms of the success of their predictions of activation energies, the BEBOA model without correction for tunnelling, is clearly superior for all reactions except CH3 + H1, for which the BEBO(tun1) value of 45.86 kJ mol-l differs only slightly from the BEBOA value of 46.69 kJ mol-l. The differences between the BEBOA and experi- mental activation energies lie between 0.31 kJ mol-1 for CF3 + H2 and 3.27 kJ mol-1N.L . ARTHUR, K . F . DONCHI AND J. A . MCDONELL 2437 for CF3 + D2, and, except for CH3 + H2, all the theoretical activation energies are lower than the corresponding experimental values. TABLE 6.-COMPARISON OF ARRHENIUS PARAMETERS AT 450 K expt.(mcan) log(A/cm3 mol-I s-I) 11.73 EIkJ mol-1 43.51 log(k/cm3 mol-' s-l) 6.68 1 og( A /cm3 mol- s- ) 11.77 E/kJ mo1-1 49.01 log(k/cm3 mo1-1 s-l) 6.08 log(A/cm3 mo1-' s-l) 11.95 E/kJ mol-' 39.71 log(k/cm3 rno1-I s-l) 7.34 log(A/cm3 mol-' s-l) 11.86 log(k/cm3 mol-1 s-l) 6.60 Elk3 mo1-1 45.35 BEBO CH3+H2 1 1.97 57.20 5.33 CH3 + D2 I I .84 59.59 4.93 CF3 + HZ 11.74 51.98 5.71 CF3 + D2 11.59 55.14 5.19 BEBOA 11.97 46.71 6.55 11 .S5 47.56 6.36 11.78 39.40 7.20 11.64 42.08 6.76 BEBO(tun1) 11.29 45.85 5.94 11.49 53.68 5.26 11.17 41.98 6.29 11.28 49.81 5.50 BEBOA(tun1) 11.44 36.95 7.15 11.57 42.20 6.67 11.39 31.69 7.71 11.39 37.45 7.04 1.4 t :e 2.2 2.6 lo3 KIT FIG.1 .-Arrhenius plot of calculated and experimental rate constants for CH3 /CD3 + H2 and CP13/CD3 + Dz. Curve A, H2, BEBOA(tun1) ; curve B, D2, BEBOA(tun1) ; curve C, Hz, BEBOA ; curve D, D2, BEBOA ; curve E, H2, BEBO(tun1) ; curve F, H2, BEBO ; curve G, D2, BEBO(tun1) ; curve H, D2, BEBO. Experimental points : 0, CH3+Hz or D2, a, CD3+H2 or Dz, ref. (8); 0, CH3+H2 or Dz, A, CD3fHz or D2, ref. (22) ; in each case the upper points refer to H2 and the Iower to Dz.2438 BEBO C A L CU LATT ONS There is little to choose between the two models in their ability to predict A factors. Agreement between the calculated and mean experimental values is good ; A factors for the CH, reactions are slightly higher, while those for the CF, reactions are a little lower, than the corresponding experimental values.I I I I 2 . 0 2.4 2.8 103 KIT FIG. 2.-Arrhenius plot of calculated and experimental rate constants for CF3 + Hz and CF3 + D2. Curve A, H2, BEBOA(tun1) ; curve 8, H2, BEBOA ; curve C, D2, BEBOA(tun1) ; curve D, DZ, BEBOA ; curve E, H2, BEBO(tun1) ; curve F, H2, BEBO ; curve G, Dz, BEBO(tun1) ; curve H, D2, BEBO. Experimental points for CF3+H2 or D2 : 0, ref. (9); 0, ref. (26); A, ref. (27). In each case the upper points refer to H, and the lower to D,. From fig. 1 and 2 it can be seen that BEBOA is much the better of the two models in predicting rate constants.This follows from the good correlation between the BEBOA and experimental activation energies. When tunnelling corrections are included, the BEBOA curves are shifted upwards and away from the experimental points. It is interesting to compare the relative reactivities of CH, and CF, radicals in these reactions with those deduced from the results of the BEBO and BEBOA calcula- tions. As might be anticipated for non-polar molecules, both H, and D, are more susceptible to CF3 than to CH, attack, the ratio of rate constants, kCF3/kCH3, being 4.6 for H2 and 3.3 for D2. Again the BEBOA model offers the better correlation; the values of kCF3/kCH3 are 4.5 for H, and 2.5 for D,, compared with the BEBO values of 2.4 and 1.8. The inclusion of tunnelling corrections results in lower values of k,,,/k,,"j and hence poor agieement with the experimental rate constant ratios.KINETIC ISOTOPE EFFECTS Kinetic isotope effects, kH/kD, calculated from both the BEBO and BEBOA models, are compared with the experimental data in table 7 and fig. 3. The alterna- tive method of calculating kinetic isotope effects, in which q is set equal to p and varied until E equals E(expt.), was also tried. The results of this procedure differed negligibly from the values given in table 7. Included in the table are isotope effectsN . L . ARTHUR, K . F . DONCHI AND J . A . MCDONELL 2439 calculated from the simple model which assumes that only the zero-point energy of the stretching vibration of the XY bond is lost, i.e., kH/kD(zpe) = exp[(Ezpe(H) - Ezpe(D))/kTl.The values of kH/kD(expt.) in the table were evaluated from the mean Arrhenius para- meters in table 5, and are plotted as dotted lines. The individual points in the figure refer to the results of competitive experiments in which k,/kD values were measured directly. T/ 1c 350 400 450 500 550 600 3 50 400 450 500 550 600 k H / kD(eXpt.) 5.99 4.74 3.95 3.41 3.02 2.74 8.57 6.72 5.57 4.79 4.23 3.82 TABLE 7.-KINETIC ISOTOPE EFFECTS k d k n kH/kD(tud) ~ - _ _ _ _ krr/k&Pe) 13.80 9.94 7.70 6.28 5.3 1 4.62 13.80 9.94 7.70 6.28 5.31 4.62 BEBO CH3 + Hz 6.15 5.19 4.51 4.02 3.64 3.34 CF3 + Hz 5.97 5.07 4.43 3.96 3.60 3.31 BEBOA 6.09 5.14 4.47 3.99 3.61 3.32 6.03 5.10 4.45 3.97 3.61 3.32 BEBO 19.94 12.63 9.07 7.06 5.80 4.95 16.88 11.27 8.36 6.65 5.55 4.78 BEBOA 15.73 10.93 8.27 6.64 5.57 4.81 12.63 9.37 7.40 6.11 5.23 4.59 The difference between the lower (CH3 + Hz /D2) and upper (CF3 + H2 /D2) experi- mental curves stems from very small variations in A factor ratios, AH/AD, and activa- tion energy differences, ED-EH : for the CH, reactions, &/AD = 0.91 and E D - E H = 5.48 kJ mol-', and for the CF, reactions, A H / A D = 1.23 and ED -EH = 5.65 kJ mol-'.These results are equal, within the error liinits usually found in experiments of this kind, and the discrepancy between the curves illustrates the inadequacy of the avail- able experimental data for the evaluation of accurate kinetic isotope effects. It should be pointed out, however, that the kH/kD curve for CF3 + HJD, deduced from the Arrhenius parameters of Kibby and We~ton,~ coincides almost exactly with the mean curve for CH, + H2/D2, and the difference between the two mean experimental curves is the result of the contribution made to the mean Arrhenius parameters for CF3 + D, by the data of Ayscough and P ~ l a n y i , ' ~ data which by themselves yield the rather unlikely values, AH/& = 2.69 and E D - E H = 2.34 kJ mol-'.On this analysis, we regard the lower experimental curve as the more reliable of the two, and it is noticeable that it is surrounded by the bulk of the individual experimental points. It can be seen from table 7 that the kinetic isotope effects (without tunnelling corrections) predicted by the BEBO and BEBOA methods are almost identical, and, as might be expected, there is very little difference between the values for the reactions involving CH, and CF3 radicals. These four sets of kH/kD values have therefore been represented by a single curve in fig.3. By contrast, there are substantial differences between the kH/kD(tunl) values for the two models and the two radicals : the BEBOA values are lower than those deduced from the BEBO model, and the CF, radical results are lower than those for CH, radical attack. The theoretical isotope effect2440 BEBO C A L C UL A TI ONS curve without tunnelling is much closer to the recommended experimental curve and the individual experimental points than are any of the curves in which tunnelling corrections are included. The k,/k,,(zpe) curve is much higher than the experimental results, and the BEBO and BEBOA models are clearly superior to the simple model in predicting kinetic isotope effects.12.0 - x 8 . 0 - 1.8 2.2 2.6 3.0 103 KIT FIG. 3.--Kinetic isotope effects as a function of temperature. Curve A, CH3, BEBO(tun1) ; curve B, CF3, BEBO(tun1) ; curve C, CH3, BEBOA(tun1) ; curve D, zpe ; curve E, CF3, BEBOA(tun1) ; curve F, CF3(mean expt); curve G, CH3/CF3, BEBO/BEBOA; curve H, CH3 (mean expt.). Experimental points: 0, CH3, ref. (8); 0, CD3, ref. (8); @, CF3, ref. (9); A, CH,, ref. (23); V, CF3, ref. (28). CONCLUSIONS The variation of the BEBO model described here provides a much more satisfactory correlation with the experimental Arrhenius parameters for the reactions of CH, and CF, radicals with H2 and D2 than the original BEBO method ; kinetic isotope effects are equally well predicted by either method. On the basis of the three-point BEBOA model, quantum mechanical tunnelling does not appear to play a significant part in these reactions. We thank Dr. J. R. Christie of La Trobe for helpful discussion and J. A. McDonell is indebted to La Trobe University for a research scholarship. H. S. Johnston and C. Parr, J. Amer. Chem. SOC., 1963,85,2544. H. S. Johnston, Gas Phase Reaction Rate Theory (Ronald, New York, 1966). N. L. Arthur and J. A. McDonnell, J. Chem Phys., 1972, 56, 3100. N. L. Arthur and J. A. McDonell, J. Chem. Phys., 1972, 57, 3228. N. L. Arthur, K. F. Donchi and J. A. McDonell, J. Chem. Phys., 1975,62,1585. J. 0. Hirschfelder, C. F. Curtis and R. B. Bird, Molecular Theory of Gases and Liquids (Wiley, New York, 1954). ' T. E. Sharp and H. S. Johnston, J. Chem. Phys., 1962,37, 1541. * J. S. Shapiro and R. E. Weston, Jr., J. Phys. Chem., 1972,76, 1669.N. L . ARTHUR, K . F . DONCHI AND J . A . MCDONELL 244 1 C. L. Kibby and R. E. Weston, Jr., J. Chem. Phys., 1968,49,4825. P . B. Ayscough, J. Chem. Phys., 1956,24,994. lo A. Shepp, J. Chem. Phys., 1956,24,939. l2 N. Basco, D. G. L. James and R. D. Stuart, Int. J. Chem. Kinetics, 1970,2,215. l 3 F . K. Truby and J. K. Rice, Int. J. Chem. Kinetics, 1973, 5,721. l4 A. M. Bass and A. H. Laufer, Int. J. Chem. Kinetics, 1973, 5, 1053. l5 T. Ogawa, G. A. Carlson and G. C. Pimentel, J. Phys. Chem., 1970,74,2090. l6 N. Basco and F. G. M. Hathorn, Chem. Phys. Letters, 1971, 8, 291. l7 R. Hiatt and S. W. Benson, J. Amer. Chem. SOC., 1972, 94, 25. lS R. Hiatt and S. W. Benson, Int. J. Chem. Kinetics, 1972,4,479. l9 J. R. McNesby, A. S. Gordon and S. R. Smith, J. Amer. Chem. Soc., 1956,78, 1287. 2o A. F. Trotman-Dickenson and G. S. Milne, Tables of Bimolecular Gas Reactions, 1967, NSRDS- 21 P. Gray, A. A. Herod and A. Jones, Chem. Reu., 1971,71,247. 22 T. G. Majury and E. W. R. Steacie, Canud. J. Chem., 1952,30, 800. 23 S. Davison and M. Burton, J. Amer. Chem. Soc., 1952, 74, 2307. 24 M. H. J. Wijnen and E. W. R. Steacie, Disc. Faraday SOC., 1953,14, 118. 25 E. Whittle and E. W. R. Steacie, J. Chem. Phys., 1953, 21, 993. 26 P. B. Ayscough and J. C. Polanyi, Trans. Faraduy SOC., 1956, 52, 960. 27 M. B. Fagarash, F. B. Moin and V. I. Ocheret’ko, Kinetics and Catalysis, 1968, 9, 762. 28 G. 0. Pritchard, H. 0. Pritchard, H. I. Schiff and A. F. Trotman-Dickenson, Trans. Faraday NBS 9. Soc., 1956, 52, 849.

ISSN:0300-9599    

DOI:10.1039/F19757102431

出版商:RSC    

年代:1975

数据来源: RSC

摘要:

BEBO Calculations Part 5.-Arrhenius Parameters and Kinetic Isotope Effects for the Reactions of CzHS and C2F5 Radicals with H2 and D, BY N. L. ARTHUR," K. F. DONCMI AND J. A. MCDONELL Department of Physical Chemistry, La Trobe University, Bundoora, Victoria 3083, Australia Received 18th October, 1974 The BEBO and BEBOA methods have been used to calculate Arrhenius parameters and kinetic isotope effects for the reactions of C2HS and CzF5 radicals with H2 and D2, and the results of these calculations are compared with the available experimental data. In Part 4 of this series, the modified BEBO method (BEBOA) was shown to be superior to the original method in predicting activation energies and A factors for the reactions of CH3 and CF3 radicals with H, and DZ, but there was little to choose between them in the calculation of kinetic isotope effects. BEBO calculations of Arrhenius parameters and kinetic isotope effects for the hydrogen transfer reactions of small radicals have mainly been concerned with CH3 and CF3 radicals,"* and the earlier papers in the series,l* 9 9 followed this theme.In this paper we have extended our previous work by studying the reactions of C2H5 and C2F5 radicals with H2 and D2. Both the BEBO and BEBOA methods have been used to calculate Arrhenius parameters and kinetic isotope effects, and the results are compared with the available experimental data. RESULTS AND DISCUSSION The molecular properties which comprise the input data in these calculations are presented in table 1. For reactions of the type XY(1inear) + Z(non1inear) -+ [XYZ] * (nonlinear) the thermal correction to the activation energy for the translational and rotational degrees of freedom of the system is 50T,R, and the Jacobian factor has the value 1/4n.There are no new internal rotational interactions resulting from the formation of the activated complex, and therefore the vibrational amplitude factor 2, is 1. For all reactions studied the statistical factor B, = 2, and the electronic multiplicity factor In the formation of the activated complex for the C2H5 reactions there are two H-C.. .H and one C-C.. .H, new bending vibrations, which each make a contribu- tion to log B of (2nkT/F)+ where Pis half the value of the force constant of the corres- ponding vibration in ethane. In this case the total contribution of the three vibrations was evaluated from (2nkT/Ff)* where 3'' is the geometric mean of the individual I; values.For the CzF5 reactions, two F-C...H and one C-C...H vibrations are involved; the F-C...H force constant was taken to be equal to half its value in CF3H ; the C-C.. .H force constant was taken from ethane as before, 2442 Be = 1.'N. L . ARTHUR, K. E. DONCHE AND J . A . MCDONELL TABLE l.-INPUT DATA FOR BEBQ AND BEBOA CALCULATIONS @/kJ mol-I D,298/kJ mol-I DJkJ mol-1 R/Pm FIN cm-l F~/10-~* J rad-2 O/ClIl--' P 9 p/108 cm-I H-H a D-D H-C~HS 432.00 439.53 410.0 458.1 458.1 421.3 74.13 74.14 109.5 5.733 5.767 5.35 Q 0.28 4395 3118 2954 1.052 1.052 1.091 1.848 D-C~HS 421.3 109.5f 5.35 0.28 2181 j 1.091 1.859 H-CzFs 431.0 C 442.7 l09.8f 5.35 0.37 2990 1.100 1.825 2443 D-CzFs 442.7 109.8 f 5.35 0.37 2248 1.100 1.871 =The input data for H2 and D2 were taken from Part 4 ; b J.A. Kerr, Chem. Rev., 1966, 66, 465 ; C J. E. Bassett and E. Whittle, J.C.S. Faraday I, 1972, 68, 492 ; dit is assumed that De(R-D) = De(R-H) ; e J. Chao, R. C. Wilhoit and B. J. Zwolinski, J. Phys. Chem. Ref. Data, 1973, 2,427 ; f Chem. SOC. Spec. Publ., 1958, 11 ; Chem. SOC. Spec. Publ., 1965, 18 ; Q G. E. Hansen and D. M. Dennison, J. Chem. Phys., 1952, 20, 313 ; h half the values of G. E. Hansen and D. M. Dennison, see also Results and Discussion ; i half the values of C. L. Kibby and R. E. Weston, Jr., ref. ( 5 ) ; j L. R. Posey, Jr. and E. F. Barker, J. Chem. Phys., 1949, 17, 183 ; k P. Klaboe and J. R. Nielsen, J. Chem. Phys., 1961, 34, 1820; lcalculated on the basis of reduced mass from the frequency of CFJD taken from T.Shimmuchi, Tables of Molecular Vibrational Frequencies, Part 6, J. Phys. Chem. Ref. Data, 1973, 2, 121. ARRHENIUS PARAMETERS The activated complex properties obtained by the BEBO and BEBOA methods are given in table 2, and the terms contributing to the activation energies and pre- exponential factors are listed in table 3. Rate constants, calculated in the range 350-600 K, are presented in table 4. The potential energy of activation, V*, for TABLE 2.-ACTIVATED COMPLEX PROPERTIES GHs+Hz CzHs+Dz C2FsfHz C2Fs+Dz BEBO BEBOA BEBO BEBOA BEBO BEBOA BEBO BEBOA 0.367 100.3 121.4 0.393 2.379 1.129 0.050 1525 604 16361 0.350 101.5 120.7 0.366 2.589 1.680 0.043 1558 554 1531i 0.366 100.3 121.4 0.392 2.385 1.726 0.049 1112 424 1156i 0.348 101.6 120.6 0.363 2.600 1.670 0.041 1137 385 1074i 0.460 94.6 125.9 0.885 1.501 1.920 0.053 1477 637 1766i 0.456 94.6 125.7 0.901 1.587 1.926 0.047 1488 602 1724i 0.460 94.4 125.8 0.883 1.501 1.910 0.049 1050 434 12433 0.454 94.7 125.6 0.889 1.593 1.889 0.040 1052 394 1194i C2H5 +H, has previously been calculated by J~hnston.~ The value he obtained (75.3 kJ mol-') is slightly higher than ours (70.9 kJmol-l) because of the different input data used.There have, however, been no previous calculations of activation energies, A factors, and kinetic isotope effects, for the reactions of C2H5 and C2F5 radicals with H2 and D2. There have been three experimental studies of the reaction C2H5+H2,11-13 two ~ O ~ C ~ H ~ + D , , ' ~ ~ 14andtwoeachforC2F,+H2,15* andC,F5+D,.16* l7 Ineach case (except for ref.(1 3) which contained only one rate constant value for H2 and one for D,) we have recalculated the Arrhenius parameters from the original data by the usual least squares method, and where appropriate we have combined the data and2444 BEBO CALCULATIONS TABLE 3 .-hRHENIUS PARAMETERS" C2H5+H2 CzHsfDz C2Fs+% BEBO BEBOA BEBO BEBOA BEBO BEBOA V* ORT E (e+ e * ) m E(tun1) log r log B iog r* log B(tunl) log A log A(tun1) 70.89 65.41 55.07 1.41 12.65 0.56 13.21 12.01 11.37 - 5.48 - 15.82 59.32 70.52 53.70 69.38 45.45 64.42 1.41 0.89 12.72 11.99 0.47 0.28 13.19 12.27 12.07 11.86 11.58 11.56 -5.62 -1.14 -13.87 -6.10 58.78 - 1.20 57.58 -5.33 53.45 0.89 12.07 0.24 12.31 11.93 11.69 62.45 50.35 56.93 44.65 45.79 35.61 1.41 1.42 12.42 12.47 0.63 0.56 13.05 13.03 11.78 11.81 11.11 11.32 -5.52 -5.70 - 16.66 - 14.74 BEBO BEBOA 60.80 47.83 59.39 46.24 53.74 41.33 0.91 0.92 11.80 11.90 0.32 0.29 12.12 12.19 11.64 11.71 11.30 11.43 -1.40 -1.59 -7.06 -6.51 0 Calculated at 450 K.A factor terms in units of cm3 mol-1 s-1 and energy terms in kJ mol-1. TABLE 4.-vARIATION OF RATE CONSTANTS (k/Cm3 m0l-l S-') WITH TEMPERATURE CZH5 + H2 C ~ H S + D ~ CZF5+ HZ CzFs+D2 T/K BEBO BEBOA BEBO BEBOA BEBO BEBOA BEBO BEBOA without tunnelling correction 350 2.23 4.03 1.49 3.32 3.26 5.13 2.76 4.80 400 3.45 5.04 2.78 4.40 4.32 5.96 3.87 5.66 450 4.42 5.83 3.81 5.25 5.17 6.63 4.75 6.35 500 5.21 6.49 4.64 5.95 5.86 7.18 5.47 6.91 550 5.86 7.03 5.34 6.53 6.43 7.63 6.07 7.39 600 6.42 7.49 5.93 7.03 6.92 8.03 6.58 7.80 with tunnelling correction 350 3.21 4.83 1.96 3.72 4.33 6.03 3.30 5.28 400 4.17 5.64 3.14 4.70 5.13 6.66 4.27 6.02 450 4.98 6.31 4.08 5.48 5.80 7.19 5.06 6.64 500 5.65 6.87 4.86 6.14 6.36 7.64 5.72 7.15 550 6.23 7.35 5.52 6.69 6.85 8.02 6.28 7.58 600 6.72 7.76 6.08 7.16 7.27 8.35 6.76 7.96 TABLE s.-EXPERIMENTAL ARRHENIUS PARAMETERS log(Alcrn3 mol-1 s-1) E/kJ mol-1 log(klcm3 mol-1 s-1) C2HS + H2 1 1.78 fi0.16 49.12+ 1.37 6.08 11.91 kO.11 50.13+ 1.01 6.09 12.60+0.06 57.14fi0.64 5.97 C2HS+D2 1 1.92+ 0.16 55.12k 1.45 5.52 12.48 & 0.10 12.82+ 0.21 12.62fi 0.64 C2FS+ H2 50.45k0.56 5.85 51.61 f 2.00 6.83 51.O4+ 5.69 6.70 ref.11 12 mean 14 15 16 mean C2FS + D2 12.095 0.17 52.48+ 1.54 6.00 16 12.9550.58 57.55 + 5.18 6.27 17 12.225 0.43 52.224 3.84 6.1 G meanN.L . ARTHUR, K . F . DONCHI AND J . A . MCDONELL 2445 obtained mean values. The Arrhenius parameters obtained are listed in table 5, the errors quoted being standard deviations. The reference reaction in each of these studies was the corresponding radical recombination reaction. The rate constant for the combination of C2H5 radicals is at present the subject of debate : the “ high ’’ value ( lo1 3*4 cm3 mol-1 s-I) obtained by the rotatingsector technique l 8 is beingchallenged bythe “low ”value (1011.6cm3m01-1 s-I) given by the radical buffer method,lg* 2o and the analysis of the pyrolysis of ethane and n-butane.21* 22 As the new value has yet to gain acceptance by gas kineticists, we have chosen to retain the old value.If, however, the new value is subsequently confirmed, the conclusions reached in this paper would have to be modified. The recombination of C2F5 radicals has not yet been studied, and we have therefore chosen the same value for the rate constant as was used in Part 4 for CF3 recombination (1013*36 cm3 mol-1 s-l). I I 1 I 103 KIT 2.0 2.4 2.8 FIG. 1.-Arrhenius plot of calculated and experimental rate constants for C2H5 + H2 and CzH5 + Df. Curve A, H2, BEBOA(tun1) ; curve B, HZ, BEBOA ; curve C, D2, BEBOA(tun1) ; curve D, Df, BEBOA ; curve E, H2, BEBO(tun1) ; curve F, Hz, BEBO ; curve G, DL, BEBO(tun1) ; curve H, D2, BEBO. Experimental points: a, C2H5fH2, ref. (11); 0, CfHJ+Hz, ref. (12); 0, C2H5+D2, ref. (14). From fig. 1 it can be seen that the two sets of data for C2H5 + H, are in good agreement, the discrepancy between their Arrhenius parameters being due to the limited temperature range covered by the experiments of Reid and Le Roy.The Arrhenius parameters recommended for C2H5 + D2 have been evaluated after omitting the three lowest temperature points, which deviate widely from the line of best fit (the two lowest temperature points in fig. 1 and another point not included in the figure). The two sets of data for C,F5+H2 form two nearly parallel Arrhenius plots. Because the higher set of points is predominantly at the upper end, and the lower set at the lower end, of the temperature range in fig. 2, the Arrhenius parameters obtained by combining all of the data (log A = 13.19, E = 55.98 kJ rnol -I) differ significantly from those of the individual studies.An alternative procedure was therefore devised2446 H 1.8 2.2 2.6 103 KIT FIG. 2.-Arrhenius plot of calculated and experimental rate constants for C2F5 + H2 and C2FS +D Curve A, H2, BEBOA(tun1) ; curve B, D2, BEBOA(tun1) ; curve C , H2, BEBOA ; curve D, D: BEBOA ; curve E, H2, BEBO(tun1) ; curve F, H2, BEBO ; curve G, Dz, BEBO(tun1) ; curve H, D; BEBO. Experimental points: 0, C2F5+Hz, ref. (15); 0, CzFs+Hz, ref. (16); CD, C2F5fD; ref. (17); 0, CzF5+D2, ref. (16). in which the activation energy was taken to be the mean of the original values, ant logA was calculated by the method of least squares. Of the two investigations o C2F5 + D,, the data of Pritchard et al. l 7 show poor reproducibility and yield ai TABLE 6.-COMPARISON OF ARRHENIUS PARAMETERS (AT 450 K) log(A/cm3 mol-1 s-l) E/kJ mol-1 log(k/cm3 mol-1 s-l) log(A/cm3 mol-1 s-l) E/kJ mol log(k/cm3 mol-l s-l) Iog(A/cm rno1-I s-l) E/kJ mol-' iog(k/cm3 mol-l s-l) log(A/cm3 mol-1 s-l) E/kJ mol-' log(k/cm3 mol-1 s-l) expt .(mean) 11.91 50.13 6.09 1 1.92 55.12 5.52 12.62 51.04 6.70 12.22 52.22 6.16 BEBO C2Hs + H2 12.01 65.41 4.42 C2H5 + Dz 11.86 69.38 3.81 C2F5f Hz 11.78 56.93 5.17 CzF5+D2 11.64 59.39 4.75 BEBOA 12.07 53.70 5.83 11.93 57.58 5.25 11.81 44.65 6.63 11.71 46.24 6.35 BEBO(tun1) BEBOA(tun1) 11.37 11.58 55.07 45.45 4.98 6.31 11.56 11.69 64.42 53.45 4.08 5.48 11.21 11.32 45.79 35.61 5.80 7.19 11.30 11 -43 53.74 41.33 5.06 6.64N. L . ARTHUK, K . F . DONCHI AND J . A . MCDONELL 2447 unusually high A factor.Nevertheless we have combined their data with those of Pritchard and Foote to obtain mean Arrhenius parameters. The Arrhenius parameters and rate constants calculated by both the BEBO and BEBOA methods are compared with the corresponding experimental quantities in table 6 and fig. 1 and 2. It can be seen that for all four reactions BEBOA predicts rate constants in much better agreement with experimental values than BEBO. When the tunnelling correction is added, the BEBOA curve becomes higher than the data points in all cases except C2H5 +D2, and the slope of the curve decreases significantly corresponding with a lowering of activation energy. In all cases the BEBO curves are much lower than the experimental results, and the inclusion of tunnelling correc- tions is not sufficient to improve the correlation.Table 6 shows that there is little difference between the A factors obtained by BEBO and BEBOA. This is due to the similarity of the activated complex properties predicted by the two methods. Agreement between the BEBOA and mean experi- mental A factors is good for the C2H5 reactions, but for the C2F5 reactions, for which the experimental parameters are suspect, the calculated A factors are low by a factor of around 5. When tunnelling corrections are included, the A factors are lowered substantially in all cases, thereby weakening the correlation with the mean experi- mental values. The activation energies given by BEBOA, with and without tunnelling corrections, are 10-13 kJ mol-1 lower than the corresponding BEBO values.Comparison of the calculated and experimental activation energies reveals a similar trend to that shown by the A factors : the BEBOA activation energies agree well with the experimental results for the C2H5 reactions, but they are low by around 6 kJ mol-1 for the C2F5 reactions. Again, this may be due to the doubtful experimental Arrhenius parameters for the C2F5 reactions : if the A factors for these reactions are set equal to 11.9 (the mean experimental value for the C,H, reactions) the experimental activation energies become 45 kJ mol-1 for C2F5 + H, and 49 kJ mol-' for C2F5 + D2, values which are very close to those predicted by the BEBOA method. KINETIC ISOTOPE EFFECTS Kinetic isotope effects, kH/kD, calculated from both the BEBO and BEBOA models, are compalied with the values given by the simple model (AH/AD = 1 and ED-& = Ezpe(H)-Ezpe(D)), and the experimental data, in table 7 and fig.3. The results of the alterfiative procedure for evaluating kH/kD from activated complex properties betting q equal to p and varying it until E equals E(expt.)] were almost identical to those obtained by the conventional method, and they have therefore been omitted. The values of kH/kD(expt.) in the table were evaluated from the mean Arrhenius parameters in table 5, and are plotted as dotted lines in fig. 3. From fig. 3 it can be seen that the mean experimental kinetic isotope effects of the C2H5 and C2F5 reactions are similar at 450 K but diverge substantially at the ends of the temperature Pange being considered. This is due to significant differences in A factor ratios, AH/A,, and activation energy differences, ED-EH, for the reactions of the two radicals.For C2H5, A,/AD = 0.98 and ED-& = 4.99 kJ mol-I and for C2F5, AH/AD = 2.5 afid ED-& = 1.18 kJ mol-I. The C2H5 data are in accord with the overall pattern established by most CH, and CF, reactions : 2 3 the ratio of A factors is close to unity and ED-E,, is between 4 and 8 kJ mol-I. The CzF5 data, on the other hand, with their high AH/AD value and low ED-& value, must be regarded with some reserve; in particular both the A factor and activation energy for C2F5 + H, appear unreasonably high.2448 BEBO CALCULATIONS Table 7 and fig. 3 show that there are only small differences between kH/kD values without tunnelling corrections predicted by the BEBO and BEBOA models, but the C2F5 curves are slightly lower than those for C2H5.Because of the similarity of the BEBO and BEBOA values for each radical, they have been represented by a single curve in the figure. Just as for the reaction of CH3 and CF, with H, (Part 4), the TABLE 7.-KINETIC ISOTOPE EFFECTS krr/kD TlK 350 400 450 500 550 600 350 400 450 500 550 600 kH/kD(expt.) 5.43 4.38 3.71 3.25 2.91 2.66 3.77 3.58 3.44 3.34 3.25 3.18 k d k D ( Z P C ) 13.80 9.94 7.70 6.28 5.31 4.62 13.80 9.94 7.70 6.28 5.31 4.62 BEBO C2&+ H z 6.34 5.32 4.62 4.10 3.71 3.40 C2F5 + H2 6.16 5.21 4.54 4.05 3.67 3.37 BEBOA 6.46 5.40 4.67 4.1 3 3.73 3.42 6.28 5.29 4.59 4.09 3.70 3.39 BEBO 20.17 12.39 8.82 6.86 5.65 4.84 20.17 12.75 9.16 7.13 5.86 5.00 BEBOA 15.91 10.65 7.96 6.37 5.34 4.63 15.57 10.89 8.27 6.66 5.60 4.85 103 KIT FIG.3.--Kinetic isotope effects as a function of temperature. Curve A, C2F5, BEBO(tun1) ; curve B, C2H5, BEBO(tun1) : curve C, C2HS, BEBOA(tun1) ; curve D, C2F5, BEBOA(tun1) ; curve E, zpe; curve F, C2H5, BERO/BEBOA ; curve G, C2F5, BEBO/BEBOA ; curvc H, C2Hs (mean expt.) ; curve 1, C2F5 (mean expt.).N. L. ARTHUR, K . F. DONCHI A N D J . A . MCDONELL 2449 BEBOA(tun1) curves are somewhat lower than the BEBO(tun1) results, but both are much higher than those without tunnelling corrections included. Again, the kH/kD (tunl) values are nearly the same for the C2H5 and C2F5 radicals. The calculated kinetic isotope effect curves which agree best with the experimental data are those not containing tunnelling corrections.The curve calculated on the basis of the zero-point energy difference of H2 and D2 is much higher than the experi- mental curves, and as with CH3 + H2 and CF3 + H2, the simple model is markedly inferior to the BEBO and BEBOA models in predicting kinetic isotope effects. CONCLUSIONS The BEBOA method is superior to the BEBO method in the prediction of Arrhenius parameters for the reactions of C2H5 and C2F5 radicals with H2 and D2 ; the kinetic isotope effects derived from both models are almost identical. BEBOA calculations indicate that tunnelling plays a minimal role in these reactions. J. A. M. is indebted to La Trobe University for a research scholarship. I N. L. Arthur, K. F. Donchi and J. A. McDonell, J.C.S. Faraday I, 1975,71,2431.H. S. Johnston and C. Parr, J. Amer. Chem. Soc., 1963, 85, 2544. H. S. Johnston, Gas Phase Reaction Rate Theory (Ronald, New York, 1966). T. E. Sharp and H. S. Johnston, J. Chem. Phys., 1962, 37, 1541. C. L. Kibby and R. E. Weston, Jr., J. Chem. Phys., 1968, 49,4825. J. S. Shapiro and R. E. Weston, Jr., J. Phys. Chem., 1972, 76, 1669. ' 0. P. Strausz, E. Jakubowski, H. S. Sandhu and H. E. Gunning, J. Chem. Phys., 1969,51,552. E. Jakubowski, H. S. Sandhu, H. E. Gunning and 0. P. Strausz, J. Chem. Phys., 1970, 52, 4242. N. L. Arthur and J. A. McDonell, J. Chem. Phys., 1972,56,3100. lo N. L. Arthur and J. A. McDonell, J. Chem. Phys., 1972,57,3228. I1 P. J. Boddy and E. W. R. Steacie, Canad. J. Chem., 1961, 39, 13. I 2 L. E. Reid and D. J. Le Roy, Canad. J.Chem., 1968,46,3275. l3 R. R. Baldwin, R. W. Walker and D. H. Langford, Trans. Faraday SOC, 1969, 65, 2116. I4 M. H. J. Wijnen and E. W. R. Steacie, J. Chem. Phys., 1952,20, 205. S. J. W. Price and K. 0. Kutschke, Canad. J. Chem., 1960,38,2128. l6 G. 0. Pritchard and J. K. Foote, J. Phys. Chem., 1964, 68, 1016. G. 0. Pritchard, J. R. Dacey, W. C. Kent and C. R. Simonds, Canad. J. Chem., 1966,44, 171. 'I3 A. Shepp and K. 0. Kutschke, J. Chem. Phys., 1957,26,1020. R. Hiatt and S. W. Benson, Int. J. Chein. Kinetics, 1972, 4, 151. 2o R. Hiatt and S. W. Benson, J. Amer. Chem. SOC., 1972, 94, 6886. 21 P. D. Pacey and J. H. Purnell, Int. J. Chem. Kinetics, 1972, 4, 657. 22 D. G. Hughes, R. M. Marshall and J. H. Purnell, J.C.S. Furaday I, 1974, 70, 594. 23 P. Gray, A.A. Herod and A. Jones, Chem. Rev., 1971, 71, 247. BEBO Calculations Part 5.-Arrhenius Parameters and Kinetic Isotope Effects for the Reactions of CzHS and C2F5 Radicals with H2 and D, BY N. L. ARTHUR," K. F. DONCMI AND J. A. MCDONELL Department of Physical Chemistry, La Trobe University, Bundoora, Victoria 3083, Australia Received 18th October, 1974 The BEBO and BEBOA methods have been used to calculate Arrhenius parameters and kinetic isotope effects for the reactions of C2HS and CzF5 radicals with H2 and D2, and the results of these calculations are compared with the available experimental data. In Part 4 of this series, the modified BEBO method (BEBOA) was shown to be superior to the original method in predicting activation energies and A factors for the reactions of CH3 and CF3 radicals with H, and DZ, but there was little to choose between them in the calculation of kinetic isotope effects.BEBO calculations of Arrhenius parameters and kinetic isotope effects for the hydrogen transfer reactions of small radicals have mainly been concerned with CH3 and CF3 radicals,"* and the earlier papers in the series,l* 9 9 followed this theme. In this paper we have extended our previous work by studying the reactions of C2H5 and C2F5 radicals with H2 and D2. Both the BEBO and BEBOA methods have been used to calculate Arrhenius parameters and kinetic isotope effects, and the results are compared with the available experimental data. RESULTS AND DISCUSSION The molecular properties which comprise the input data in these calculations are presented in table 1.For reactions of the type XY(1inear) + Z(non1inear) -+ [XYZ] * (nonlinear) the thermal correction to the activation energy for the translational and rotational degrees of freedom of the system is 50T,R, and the Jacobian factor has the value 1/4n. There are no new internal rotational interactions resulting from the formation of the activated complex, and therefore the vibrational amplitude factor 2, is 1. For all reactions studied the statistical factor B, = 2, and the electronic multiplicity factor In the formation of the activated complex for the C2H5 reactions there are two H-C.. .H and one C-C.. .H, new bending vibrations, which each make a contribu- tion to log B of (2nkT/F)+ where Pis half the value of the force constant of the corres- ponding vibration in ethane.In this case the total contribution of the three vibrations was evaluated from (2nkT/Ff)* where 3'' is the geometric mean of the individual I; values. For the CzF5 reactions, two F-C...H and one C-C...H vibrations are involved; the F-C...H force constant was taken to be equal to half its value in CF3H ; the C-C.. .H force constant was taken from ethane as before, 2442 Be = 1.'N. L . ARTHUR, K. E. DONCHE AND J . A . MCDONELL TABLE l.-INPUT DATA FOR BEBQ AND BEBOA CALCULATIONS @/kJ mol-I D,298/kJ mol-I DJkJ mol-1 R/Pm FIN cm-l F~/10-~* J rad-2 O/ClIl--' P 9 p/108 cm-I H-H a D-D H-C~HS 432.00 439.53 410.0 458.1 458.1 421.3 74.13 74.14 109.5 5.733 5.767 5.35 Q 0.28 4395 3118 2954 1.052 1.052 1.091 1.848 D-C~HS 421.3 109.5f 5.35 0.28 2181 j 1.091 1.859 H-CzFs 431.0 C 442.7 l09.8f 5.35 0.37 2990 1.100 1.825 2443 D-CzFs 442.7 109.8 f 5.35 0.37 2248 1.100 1.871 =The input data for H2 and D2 were taken from Part 4 ; b J.A. Kerr, Chem. Rev., 1966, 66, 465 ; C J. E. Bassett and E. Whittle, J.C.S. Faraday I, 1972, 68, 492 ; dit is assumed that De(R-D) = De(R-H) ; e J. Chao, R. C. Wilhoit and B. J. Zwolinski, J. Phys. Chem. Ref. Data, 1973, 2,427 ; f Chem. SOC. Spec. Publ., 1958, 11 ; Chem. SOC. Spec. Publ., 1965, 18 ; Q G. E. Hansen and D. M. Dennison, J. Chem. Phys., 1952, 20, 313 ; h half the values of G. E. Hansen and D. M. Dennison, see also Results and Discussion ; i half the values of C. L. Kibby and R. E. Weston, Jr., ref. ( 5 ) ; j L. R. Posey, Jr. and E. F. Barker, J.Chem. Phys., 1949, 17, 183 ; k P. Klaboe and J. R. Nielsen, J. Chem. Phys., 1961, 34, 1820; lcalculated on the basis of reduced mass from the frequency of CFJD taken from T. Shimmuchi, Tables of Molecular Vibrational Frequencies, Part 6, J. Phys. Chem. Ref. Data, 1973, 2, 121. ARRHENIUS PARAMETERS The activated complex properties obtained by the BEBO and BEBOA methods are given in table 2, and the terms contributing to the activation energies and pre- exponential factors are listed in table 3. Rate constants, calculated in the range 350-600 K, are presented in table 4. The potential energy of activation, V*, for TABLE 2.-ACTIVATED COMPLEX PROPERTIES GHs+Hz CzHs+Dz C2FsfHz C2Fs+Dz BEBO BEBOA BEBO BEBOA BEBO BEBOA BEBO BEBOA 0.367 100.3 121.4 0.393 2.379 1.129 0.050 1525 604 16361 0.350 101.5 120.7 0.366 2.589 1.680 0.043 1558 554 1531i 0.366 100.3 121.4 0.392 2.385 1.726 0.049 1112 424 1156i 0.348 101.6 120.6 0.363 2.600 1.670 0.041 1137 385 1074i 0.460 94.6 125.9 0.885 1.501 1.920 0.053 1477 637 1766i 0.456 94.6 125.7 0.901 1.587 1.926 0.047 1488 602 1724i 0.460 94.4 125.8 0.883 1.501 1.910 0.049 1050 434 12433 0.454 94.7 125.6 0.889 1.593 1.889 0.040 1052 394 1194i C2H5 +H, has previously been calculated by J~hnston.~ The value he obtained (75.3 kJ mol-') is slightly higher than ours (70.9 kJmol-l) because of the different input data used.There have, however, been no previous calculations of activation energies, A factors, and kinetic isotope effects, for the reactions of C2H5 and C2F5 radicals with H2 and D2. There have been three experimental studies of the reaction C2H5+H2,11-13 two ~ O ~ C ~ H ~ + D , , ' ~ ~ 14andtwoeachforC2F,+H2,15* andC,F5+D,.16* l7 Ineach case (except for ref.(1 3) which contained only one rate constant value for H2 and one for D,) we have recalculated the Arrhenius parameters from the original data by the usual least squares method, and where appropriate we have combined the data and2444 BEBO CALCULATIONS TABLE 3 .-hRHENIUS PARAMETERS" C2H5+H2 CzHsfDz C2Fs+% BEBO BEBOA BEBO BEBOA BEBO BEBOA V* ORT E (e+ e * ) m E(tun1) log r log B iog r* log B(tunl) log A log A(tun1) 70.89 65.41 55.07 1.41 12.65 0.56 13.21 12.01 11.37 - 5.48 - 15.82 59.32 70.52 53.70 69.38 45.45 64.42 1.41 0.89 12.72 11.99 0.47 0.28 13.19 12.27 12.07 11.86 11.58 11.56 -5.62 -1.14 -13.87 -6.10 58.78 - 1.20 57.58 -5.33 53.45 0.89 12.07 0.24 12.31 11.93 11.69 62.45 50.35 56.93 44.65 45.79 35.61 1.41 1.42 12.42 12.47 0.63 0.56 13.05 13.03 11.78 11.81 11.11 11.32 -5.52 -5.70 - 16.66 - 14.74 BEBO BEBOA 60.80 47.83 59.39 46.24 53.74 41.33 0.91 0.92 11.80 11.90 0.32 0.29 12.12 12.19 11.64 11.71 11.30 11.43 -1.40 -1.59 -7.06 -6.51 0 Calculated at 450 K. A factor terms in units of cm3 mol-1 s-1 and energy terms in kJ mol-1.TABLE 4.-vARIATION OF RATE CONSTANTS (k/Cm3 m0l-l S-') WITH TEMPERATURE CZH5 + H2 C ~ H S + D ~ CZF5+ HZ CzFs+D2 T/K BEBO BEBOA BEBO BEBOA BEBO BEBOA BEBO BEBOA without tunnelling correction 350 2.23 4.03 1.49 3.32 3.26 5.13 2.76 4.80 400 3.45 5.04 2.78 4.40 4.32 5.96 3.87 5.66 450 4.42 5.83 3.81 5.25 5.17 6.63 4.75 6.35 500 5.21 6.49 4.64 5.95 5.86 7.18 5.47 6.91 550 5.86 7.03 5.34 6.53 6.43 7.63 6.07 7.39 600 6.42 7.49 5.93 7.03 6.92 8.03 6.58 7.80 with tunnelling correction 350 3.21 4.83 1.96 3.72 4.33 6.03 3.30 5.28 400 4.17 5.64 3.14 4.70 5.13 6.66 4.27 6.02 450 4.98 6.31 4.08 5.48 5.80 7.19 5.06 6.64 500 5.65 6.87 4.86 6.14 6.36 7.64 5.72 7.15 550 6.23 7.35 5.52 6.69 6.85 8.02 6.28 7.58 600 6.72 7.76 6.08 7.16 7.27 8.35 6.76 7.96 TABLE s.-EXPERIMENTAL ARRHENIUS PARAMETERS log(Alcrn3 mol-1 s-1) E/kJ mol-1 log(klcm3 mol-1 s-1) C2HS + H2 1 1.78 fi0.16 49.12+ 1.37 6.08 11.91 kO.11 50.13+ 1.01 6.09 12.60+0.06 57.14fi0.64 5.97 C2HS+D2 1 1.92+ 0.16 55.12k 1.45 5.52 12.48 & 0.10 12.82+ 0.21 12.62fi 0.64 C2FS+ H2 50.45k0.56 5.85 51.61 f 2.00 6.83 51.O4+ 5.69 6.70 ref.11 12 mean 14 15 16 mean C2FS + D2 12.095 0.17 52.48+ 1.54 6.00 16 12.9550.58 57.55 + 5.18 6.27 17 12.225 0.43 52.224 3.84 6.1 G meanN.L . ARTHUR, K . F . DONCHI AND J . A . MCDONELL 2445 obtained mean values. The Arrhenius parameters obtained are listed in table 5, the errors quoted being standard deviations. The reference reaction in each of these studies was the corresponding radical recombination reaction. The rate constant for the combination of C2H5 radicals is at present the subject of debate : the “ high ’’ value ( lo1 3*4 cm3 mol-1 s-I) obtained by the rotatingsector technique l 8 is beingchallenged bythe “low ”value (1011.6cm3m01-1 s-I) given by the radical buffer method,lg* 2o and the analysis of the pyrolysis of ethane and n-butane.21* 22 As the new value has yet to gain acceptance by gas kineticists, we have chosen to retain the old value. If, however, the new value is subsequently confirmed, the conclusions reached in this paper would have to be modified.The recombination of C2F5 radicals has not yet been studied, and we have therefore chosen the same value for the rate constant as was used in Part 4 for CF3 recombination (1013*36 cm3 mol-1 s-l). I I 1 I 103 KIT 2.0 2.4 2.8 FIG. 1.-Arrhenius plot of calculated and experimental rate constants for C2H5 + H2 and CzH5 + Df. Curve A, H2, BEBOA(tun1) ; curve B, HZ, BEBOA ; curve C, D2, BEBOA(tun1) ; curve D, Df, BEBOA ; curve E, H2, BEBO(tun1) ; curve F, Hz, BEBO ; curve G, DL, BEBO(tun1) ; curve H, D2, BEBO. Experimental points: a, C2H5fH2, ref. (11); 0, CfHJ+Hz, ref.(12); 0, C2H5+D2, ref. (14). From fig. 1 it can be seen that the two sets of data for C2H5 + H, are in good agreement, the discrepancy between their Arrhenius parameters being due to the limited temperature range covered by the experiments of Reid and Le Roy. The Arrhenius parameters recommended for C2H5 + D2 have been evaluated after omitting the three lowest temperature points, which deviate widely from the line of best fit (the two lowest temperature points in fig. 1 and another point not included in the figure). The two sets of data for C,F5+H2 form two nearly parallel Arrhenius plots. Because the higher set of points is predominantly at the upper end, and the lower set at the lower end, of the temperature range in fig. 2, the Arrhenius parameters obtained by combining all of the data (log A = 13.19, E = 55.98 kJ rnol -I) differ significantly from those of the individual studies.An alternative procedure was therefore devised2446 H 1.8 2.2 2.6 103 KIT FIG. 2.-Arrhenius plot of calculated and experimental rate constants for C2F5 + H2 and C2FS +D Curve A, H2, BEBOA(tun1) ; curve B, D2, BEBOA(tun1) ; curve C , H2, BEBOA ; curve D, D: BEBOA ; curve E, H2, BEBO(tun1) ; curve F, H2, BEBO ; curve G, Dz, BEBO(tun1) ; curve H, D; BEBO. Experimental points: 0, C2F5+Hz, ref. (15); 0, CzFs+Hz, ref. (16); CD, C2F5fD; ref. (17); 0, CzF5+D2, ref. (16). in which the activation energy was taken to be the mean of the original values, ant logA was calculated by the method of least squares. Of the two investigations o C2F5 + D,, the data of Pritchard et al. l 7 show poor reproducibility and yield ai TABLE 6.-COMPARISON OF ARRHENIUS PARAMETERS (AT 450 K) log(A/cm3 mol-1 s-l) E/kJ mol-1 log(k/cm3 mol-1 s-l) log(A/cm3 mol-1 s-l) E/kJ mol log(k/cm3 mol-l s-l) Iog(A/cm rno1-I s-l) E/kJ mol-' iog(k/cm3 mol-l s-l) log(A/cm3 mol-1 s-l) E/kJ mol-' log(k/cm3 mol-1 s-l) expt .(mean) 11.91 50.13 6.09 1 1.92 55.12 5.52 12.62 51.04 6.70 12.22 52.22 6.16 BEBO C2Hs + H2 12.01 65.41 4.42 C2H5 + Dz 11.86 69.38 3.81 C2F5f Hz 11.78 56.93 5.17 CzF5+D2 11.64 59.39 4.75 BEBOA 12.07 53.70 5.83 11.93 57.58 5.25 11.81 44.65 6.63 11.71 46.24 6.35 BEBO(tun1) BEBOA(tun1) 11.37 11.58 55.07 45.45 4.98 6.31 11.56 11.69 64.42 53.45 4.08 5.48 11.21 11.32 45.79 35.61 5.80 7.19 11.30 11 -43 53.74 41.33 5.06 6.64N. L .ARTHUK, K . F . DONCHI AND J . A . MCDONELL 2447 unusually high A factor. Nevertheless we have combined their data with those of Pritchard and Foote to obtain mean Arrhenius parameters. The Arrhenius parameters and rate constants calculated by both the BEBO and BEBOA methods are compared with the corresponding experimental quantities in table 6 and fig. 1 and 2. It can be seen that for all four reactions BEBOA predicts rate constants in much better agreement with experimental values than BEBO. When the tunnelling correction is added, the BEBOA curve becomes higher than the data points in all cases except C2H5 +D2, and the slope of the curve decreases significantly corresponding with a lowering of activation energy. In all cases the BEBO curves are much lower than the experimental results, and the inclusion of tunnelling correc- tions is not sufficient to improve the correlation. Table 6 shows that there is little difference between the A factors obtained by BEBO and BEBOA.This is due to the similarity of the activated complex properties predicted by the two methods. Agreement between the BEBOA and mean experi- mental A factors is good for the C2H5 reactions, but for the C2F5 reactions, for which the experimental parameters are suspect, the calculated A factors are low by a factor of around 5. When tunnelling corrections are included, the A factors are lowered substantially in all cases, thereby weakening the correlation with the mean experi- mental values. The activation energies given by BEBOA, with and without tunnelling corrections, are 10-13 kJ mol-1 lower than the corresponding BEBO values.Comparison of the calculated and experimental activation energies reveals a similar trend to that shown by the A factors : the BEBOA activation energies agree well with the experimental results for the C2H5 reactions, but they are low by around 6 kJ mol-1 for the C2F5 reactions. Again, this may be due to the doubtful experimental Arrhenius parameters for the C2F5 reactions : if the A factors for these reactions are set equal to 11.9 (the mean experimental value for the C,H, reactions) the experimental activation energies become 45 kJ mol-1 for C2F5 + H, and 49 kJ mol-' for C2F5 + D2, values which are very close to those predicted by the BEBOA method. KINETIC ISOTOPE EFFECTS Kinetic isotope effects, kH/kD, calculated from both the BEBO and BEBOA models, are compalied with the values given by the simple model (AH/AD = 1 and ED-& = Ezpe(H)-Ezpe(D)), and the experimental data, in table 7 and fig.3. The results of the alterfiative procedure for evaluating kH/kD from activated complex properties betting q equal to p and varying it until E equals E(expt.)] were almost identical to those obtained by the conventional method, and they have therefore been omitted. The values of kH/kD(expt.) in the table were evaluated from the mean Arrhenius parameters in table 5, and are plotted as dotted lines in fig. 3. From fig. 3 it can be seen that the mean experimental kinetic isotope effects of the C2H5 and C2F5 reactions are similar at 450 K but diverge substantially at the ends of the temperature Pange being considered. This is due to significant differences in A factor ratios, AH/A,, and activation energy differences, ED-EH, for the reactions of the two radicals.For C2H5, A,/AD = 0.98 and ED-& = 4.99 kJ mol-I and for C2F5, AH/AD = 2.5 afid ED-& = 1.18 kJ mol-I. The C2H5 data are in accord with the overall pattern established by most CH, and CF, reactions : 2 3 the ratio of A factors is close to unity and ED-E,, is between 4 and 8 kJ mol-I. The CzF5 data, on the other hand, with their high AH/AD value and low ED-& value, must be regarded with some reserve; in particular both the A factor and activation energy for C2F5 + H, appear unreasonably high.2448 BEBO CALCULATIONS Table 7 and fig. 3 show that there are only small differences between kH/kD values without tunnelling corrections predicted by the BEBO and BEBOA models, but the C2F5 curves are slightly lower than those for C2H5.Because of the similarity of the BEBO and BEBOA values for each radical, they have been represented by a single curve in the figure. Just as for the reaction of CH3 and CF, with H, (Part 4), the TABLE 7.-KINETIC ISOTOPE EFFECTS krr/kD TlK 350 400 450 500 550 600 350 400 450 500 550 600 kH/kD(expt.) 5.43 4.38 3.71 3.25 2.91 2.66 3.77 3.58 3.44 3.34 3.25 3.18 k d k D ( Z P C ) 13.80 9.94 7.70 6.28 5.31 4.62 13.80 9.94 7.70 6.28 5.31 4.62 BEBO C2&+ H z 6.34 5.32 4.62 4.10 3.71 3.40 C2F5 + H2 6.16 5.21 4.54 4.05 3.67 3.37 BEBOA 6.46 5.40 4.67 4.1 3 3.73 3.42 6.28 5.29 4.59 4.09 3.70 3.39 BEBO 20.17 12.39 8.82 6.86 5.65 4.84 20.17 12.75 9.16 7.13 5.86 5.00 BEBOA 15.91 10.65 7.96 6.37 5.34 4.63 15.57 10.89 8.27 6.66 5.60 4.85 103 KIT FIG.3.--Kinetic isotope effects as a function of temperature. Curve A, C2F5, BEBO(tun1) ; curve B, C2H5, BEBO(tun1) : curve C, C2HS, BEBOA(tun1) ; curve D, C2F5, BEBOA(tun1) ; curve E, zpe; curve F, C2H5, BERO/BEBOA ; curve G, C2F5, BEBO/BEBOA ; curvc H, C2Hs (mean expt.) ; curve 1, C2F5 (mean expt.).N. L. ARTHUR, K . F. DONCHI A N D J . A . MCDONELL 2449 BEBOA(tun1) curves are somewhat lower than the BEBO(tun1) results, but both are much higher than those without tunnelling corrections included. Again, the kH/kD (tunl) values are nearly the same for the C2H5 and C2F5 radicals. The calculated kinetic isotope effect curves which agree best with the experimental data are those not containing tunnelling corrections.The curve calculated on the basis of the zero-point energy difference of H2 and D2 is much higher than the experi- mental curves, and as with CH3 + H2 and CF3 + H2, the simple model is markedly inferior to the BEBO and BEBOA models in predicting kinetic isotope effects. CONCLUSIONS The BEBOA method is superior to the BEBO method in the prediction of Arrhenius parameters for the reactions of C2H5 and C2F5 radicals with H2 and D2 ; the kinetic isotope effects derived from both models are almost identical. BEBOA calculations indicate that tunnelling plays a minimal role in these reactions. J. A. M. is indebted to La Trobe University for a research scholarship. I N. L. Arthur, K. F. Donchi and J. A. McDonell, J.C.S. Faraday I, 1975,71,2431. H. S. Johnston and C. Parr, J. Amer. Chem. Soc., 1963, 85, 2544. H. S. Johnston, Gas Phase Reaction Rate Theory (Ronald, New York, 1966). T. E. Sharp and H. S. Johnston, J. Chem. Phys., 1962, 37, 1541. C. L. Kibby and R. E. Weston, Jr., J. Chem. Phys., 1968, 49,4825. J. S. Shapiro and R. E. Weston, Jr., J. Phys. Chem., 1972, 76, 1669. ' 0. P. Strausz, E. Jakubowski, H. S. Sandhu and H. E. Gunning, J. Chem. Phys., 1969,51,552. E. Jakubowski, H. S. Sandhu, H. E. Gunning and 0. P. Strausz, J. Chem. Phys., 1970, 52, 4242. N. L. Arthur and J. A. McDonell, J. Chem. Phys., 1972,56,3100. lo N. L. Arthur and J. A. McDonell, J. Chem. Phys., 1972,57,3228. I1 P. J. Boddy and E. W. R. Steacie, Canad. J. Chem., 1961, 39, 13. I 2 L. E. Reid and D. J. Le Roy, Canad. J. Chem., 1968,46,3275. l3 R. R. Baldwin, R. W. Walker and D. H. Langford, Trans. Faraday SOC, 1969, 65, 2116. I4 M. H. J. Wijnen and E. W. R. Steacie, J. Chem. Phys., 1952,20, 205. S. J. W. Price and K. 0. Kutschke, Canad. J. Chem., 1960,38,2128. l6 G. 0. Pritchard and J. K. Foote, J. Phys. Chem., 1964, 68, 1016. G. 0. Pritchard, J. R. Dacey, W. C. Kent and C. R. Simonds, Canad. J. Chem., 1966,44, 171. 'I3 A. Shepp and K. 0. Kutschke, J. Chem. Phys., 1957,26,1020. R. Hiatt and S. W. Benson, Int. J. Chein. Kinetics, 1972, 4, 151. 2o R. Hiatt and S. W. Benson, J. Amer. Chem. SOC., 1972, 94, 6886. 21 P. D. Pacey and J. H. Purnell, Int. J. Chem. Kinetics, 1972, 4, 657. 22 D. G. Hughes, R. M. Marshall and J. H. Purnell, J.C.S. Furaday I, 1974, 70, 594. 23 P. Gray, A. A. Herod and A. Jones, Chem. Rev., 1971, 71, 247.

ISSN:0300-9599    

DOI:10.1039/F19757102442

出版商:RSC    

年代:1975

数据来源: RSC

摘要:

Homogeneous Gas Phase Pyrolysis of N-Isopropylacetamide BY ALLAN MACCOLL AND SURJIT S. NAGRA-f The Christopher Ingold Laboratories, 20 Gordon Street, London WClH OAJ Received 28th November, 1974 The gas phase pyrolysis of N-isopropylacetamide (IPA) has been studied in a static system at temperatures between 430 and 480°C. The decomposition occurs by a unimolecular mechanism to yield propene and acetamide. This was accompanied by a bimolecular decomposition to give isopropylamine, propene, acetic acid and acetonitrile. The latter reaction was also cataIysed by the acetic acid formed. The rate constants for the unimolecular (kl), bimolecular (k2) and the catalysed reaction (k,) are defined respectively by the Arrhenius equations k , = 1012*07*0.33 exp [(-54 OOO+ 1100) cal mol-'/RT] s-I kz = 1013.75f0.85 exp [( - 42 200 k 2900) cal mol-l IRT] cm3 mokl s-I k3 = 1013*66*0-30 exp [(-35 SOOi-990) cal mol-'/RT] cm3 mol-l s-l where R = 1.987 cal mol-I K-l, 1 caI = 4.187 J.The relative rates of unimoIecular elimination of acetamide in the N-substituted acetamides are compared with those for the elimination of carboxylic acid from the esters. The greater mesomeric effect in the amide than in the ester group is responsible for a relatively smaller effect of cc-methyl- substitution in the N-alkyl substituted acetamides. The acetic acid catalysed decomposition of N- alkyf substituted acetamides is similar to that of the catalysed decomposition of acetamide. A number of workers 1-3 have suggested that the pyrolysis of N-alkylamides is similar to the pyrolysis of esters, the product being an olefin and the parent amide.Hurd and Blunk ' 9 proposed it six-centred transition state for the pyrolysis of esters. Similarities between substitutent effects 6 v on the rate of thermal decomposition of alkyl halides and esters in the gas phase and certain S,l and El reactions in solution suggested that ionic structures, formulated as a separation of a carboxylate and a carbonium ion (in esters) make an appreciable contribution to the transitional state. The quasi-heterolytic hypothesis was proposed to explain the effect of substituents for molecular elimination reactions in the gas phase. It was hoped that the kinetic studies of the thermal decomposition of N-alkyl substituted acetamides might throw some more light on the polar mechanism proposed for other molecular elimination reactions.The homogeneous and unimolecular gas phase decomposition of N-t- butylacetamide l o between 385 and 460°C has already been described. To investigate the effect of a-methyl substitution on the rate of the elimination reaction, we have studied the thermal decomposition of N-isopropylacetamide (IPA). EXPERIMENTAL The gaseous materials used in this work were commercial samples of purity greater than 99 %. IPA was prepared by the reactions of acetyl chloride with isopropylamine below 0°C. The constant boiling middle fraction (b.p. 95°C at 13 mmHg) was retained for use. The purity of IPA was checked by g.1.c. and the mass spectrum confirmed its identity. Most of the analyses were performed by gas chromatography using a Perkin Elmer F11 instru- ment with a flame ionization detector.The gaseous hydrocarbons were quantitatively t present address : Chemistry Dept., University of Calgary, Alberta, Canada. 2450A . MACCOLL AND S. S . NAGRA 245 1 estimated with isobutene as an internal standard on a 4m d u m n packed with 80-100 Chromosorb P containing 20 % w/w bis-(2-methoxyyethyl) adipate. Quantitative analysis of IPA was performed using acetophenone as an internal standard on a 2 m column packed with 80-100 Chromosorb W containing 8 % w/w Antarox. A commercial (B.D.H.) sample of N-ethylacetamide was fractionated under reduced pressure in an atmosphere of nitrogen. The constant boiling (94°C at 4mmHg pressure) fraction was used. The analysis of this sample showed less than 0.2 % detectable impurities.Propionitrile was employed as an internal standard for the estimation of acetonitrile on a 2.5 m column packed with 60-80 G cell containing 30 % w/w dinonyl phthalate. In each case the internal standard was intro- duced into the reaction vessel and its pressure measured. This was then condensed into a trap at liquid nitmgen temperature. At the termination of the run the reaction mixture was condensed into the same trap. The gaseous products of the mixture were separated from solids and liquids by distillation from the trap at -80°C to a trap at liquid nitrogen temperature. In all cases the internal standards were normally left in the reaction vessel for 1-2 min and no detectable decomposition was observed even for 5 min.The following calibration factors (f) defined by (mol), - (peak area), (subscript c stands for compound ; (mol), f(peak area), s for standard) -- were obtained : IPA 2.39, acetonitrile 1.53 ; propene 1.22. The presence of isopropylamine acetate in the products of pyrolysis was confirmed by i.r. spectroscopy, mass spectrometry and mixed m.p. measurements, the comparison being made with an authentic sample. The quantitative estimation of isopropylamine acetate and ammonium acetate in the pyrolgsate was by use of the Conway microdiffusion method." The reaction vessel was a Pyrex cyIinder of about 250 cm3 capacity and about 5 cm internal diameter. The packed vessel was similar but it contained three concentric glass cylinders giving a surface to volume ratio of about 5.0 cm-l.An optical diaphragm gauge connected directly to the reaction vessel was used to measure the pressure by a null deflection method. The lower limit of pressure change detectable with a typical gauge of this kind was about 0.2mmHg. The reaction vessel was enclosed ih an aluminium block furnace. The sample inlet system and gauge were maintained at approximately 200°C by wrapping with electrothermal heating tape. The temperature of the furnace was controlled by a Sunvic resistance thermometer (type R.T.2) unit. The temperature of the reaction vessel was measured to +O.l"C by a thermocouple platinum/platinuin - 13 % rhodium constructed from wire calibrated at the National Physical Laboratories, Teddington. The kinetic runs for IPA were carried out with the addition of inhibitors.In a normal run, the pressure of the inhibitor used (isobutene) was about one and a half times the pressure of isopropyl acetamide. This was more than sufficient to suppress the radical side reactions i the methane formed in a typical inhibited run was only about 5 % of the propene formed, RESULTS The products of the pyrolysis of N-isopropylacetamide in a self conditioned reac- tion vessel between 430 and 485°C were identified as (a) acetamide, acetonitrile, acetic acid, isopropylamine acetate and ammonium acetate, (b) propene and traces of Cz hydrocarbons, propane, n-butane, but- 1 -ene, cis- and trans-but-2-ene and (c) methane. carbon monoxide and hydrogen as non-condensible gases at liquid nitrogen tem- perature. In an overnight run the ratio of final to initial pressure was found to be close to 3.There was almost three times as much methane as propene in the products. When the gas phase pyrolysis of IPA was conducted in the presence of inhibitors (isobutene, cyclohexene l 1 and toluene 12) the initial rate was found to be considerably less (7.0 x s-l) than the uninhibited rate (30.0 x s-l). The pressure against time curves for IPA (195.0 mmHg) and IPA (1 82.0 mmHg) in the presence of 155.0 mmHg of toluene (85 % of Po, the initial pressure of IPA) at 442.2"C are shown in fig. 1A2452 HOMOGENEOUS GAS PHASE PYROLYSIS OF N-ISOPROPYLACETAMIDE and B respectively. Analysis also showed that the products of the inhibited runs contained much less methane than those of the uninhibited ones (table 1).1700--- 8 16 24 32 time/min FIG. I.-Pressure against time curves : A, for isopropylacetamide (195.0 mmHg) ; B, for isopropyl- acetamide (182.0 mmHg) + 85 % toluene inhibitor ; C, for isopropylacetamide (186.0 mmHg)+ 90 % isobutene inhibitor + acetic acid (53.2 mmHg). TABLE 1 .-PRODUCTS OF THE DECOMPOSITION REACTION (T = 442+ 2°C) % in the gaseous products relative to propene inhibitor cyclohexene cyclohexene cyclohexene cyclohexene isobutene isobutene isobutene isobutene is0 butene toluene (P,/PO)X loo* 20 43 128 150 10 30 83 148 210 85 PO(IPA)/ =Hgt 241.2 212.8 130.0 120.5 131.7 149.5 160.7 112.3 108.0 195.0 CH4 18.0 13.5 7.5 8.0 46.0 24.0 38.0 16.0 10.5 18.0 C2 hydro- carbons 4.5 4.0 10.0 12.0 2.3 4.0 2.3 3.5 2.2 3.4 propane 4.0 4.5 4.0 3.8 3 .O 2.2 3 .O 2.8 1.7 2.9 * Pi is the inhibitor pressure.t 1 mmHg = 133 N m-'. As cyclohexene decomposes at the temperature of pyrolysis of IPA, isobutene was employed as the inhibitor. From the equations Vo = k,PE log Vo = log k,+n log Po where Yo is the initial rate, Po the initial pressure, k the rate constant and n the order of decomposition, the plot of log Yo against logPo yielded 1.33 as the order of decomposition of IPA. The dependence of initial rate coefficient upon the initial pressure of IPA at 462.6"C is shown in fig. 2. A straight line with both a positive intercept and slope was obtained. It was concluded from this observation that there was a bimolecular component present in the decomposition of IPA.A . MACCOLL AND S . S . NAGRA 2453 By applying the equation (where k l and k, are the first and second order rate constants respectively), the plot of initial rate constant against Po (fig.2) yielded an intercept (k,) and slope (k2). Vo/Po = k , +kzPo 0 200 300 Po(IPA)jminHg Fro. 2.-Dependence of initial rate constant with initial pressure of isopropylacetamide at 462.6"C. The pressure against time profiles for propene, acetonitrile, isopropylamine, acetic acid, remaining IPA, methane and ammonia were obtained at 442.2"C in the presence of isobutene as the inhibitor (as shown in fig. 3). Acetamide analysis was found to be somewhat inconsistent. In the presence of inhibitor, propene and acetamide are believed to be produced from a unimolecular decomposition of IPA. 15C 100 $.? E 8 5 0 n "0 6 0 120 I80 timejmin FIG.3.-Reaction profile for the pyrolysis of IPA at 442.2"C. 0 , IPA ; D , CHJCN ; 0, C3H6 ; x , CH3COOH ; 0, PriNHz ; 0, NH3 ; A, CH4.2454 HOMOGENEOUS GAS PIIASE PYROLYSIS OF N-ISOPROPYLACETAMIDE To account for the very early production of isopropylamine, acetonitrile and acetic acid, a bimolecular decomposition of IPA is proposed. This also accounts for the small bimolecular component as found in fig. 2. The decomposition reactions are represented by eqn (1) and (2) respectively k i CH,CONHCH(CH3)2 -4 CH,CH=CH, +CH,CONH2 (1) CH,CH=CH,. (2) k2 2CH3CONHCH(CH3)2 -4 (CH3)2CHNH2 + CH3CN + CH3COOH + These reactions account for the observed stoichiometry and the products of the pyrolysis of IPA up to about 30 % decomposition. After this, the amounts of acetonitrile, acetic acid and isopropylamine are higher than expected from these reactions.These observations, together with the sigmoid nature of the pressure against time curves, were found to be due to catalysis by the acetic acid formed in the pyrolysis. The pressure against time curves for a normal run of IPA and in the presence of acetic acid are shown in fig. 1A and C respectively. The formation of isopropylamine and propene in the presence and in the absence of acetic acid at 442.2"C is compared in table 2. This reaction of acetic acid and IPA is similar to the acetic acid catalysed decomposition of N-t-butylacetamide O and is represented by eqn (3) k2 CH3CONHCH(CH3)3 + CH3COOH --+ (CH3)2CHNH2 + CH3COOCOCH3. (3) However, acetic anhydride yield acetic acid and ketene.decomposes instantaneously at these temperatures to Further reactions of ketene with acetamide l o to form TABLE 2.-FORMATION OF ISOPKOPYLAMINE AND PROPENE IN THE PRESENCE AND ABSENCE OF ACETIC ACID (T = 442.2"C) time/ min 20 160.0 - 11.0 4.7 20 160.0 50.5 32.0 24.5 40 160.0 I 21 .s 13.5 40 160.0 50.3 41 .O 31.0 acetonitrile and acetic acid and the reactions of ketene with IPA to form propene, acetonitrile and acetic acid take place. Also it has been shown previously lo* l 4 that the acetamide formed, starts to decompose at these temperatures by a bimolecular reaction and by an autocatalylic process to form acetonitrile, acetic acid and ammonia. The rate constant (k3) for the acetic acid catalysed decomposition of IPA was calcu- lated from the slope of the plots of Vo/P,-k2Po against initial concentration of acetic acid : The variation of rate constants kl, k2 and k3 with temperature are shown in table 3, together with the Arrhenius equations obtained by least squares treatment.The packed reaction vessel gave essentially the same results as that of the unpacked vessel. Therefore, the decomposition is believed to be homogeneous. When N-ethylacetamide was pyro- lysed at 442.2"C, the analysis showed the major products to be ethylene, carbon monoxide, methane, hydrogen, acetamide and acetonitrile. Traces of propane, Vo/Po - k,Po = k , + k3[CH3COOH]o. The errors expressed are the 95 % confidence limits.A . MACCOLL AND S . S . NAGRA 2455 ethane, butane, butene and hydrogen cyanide were also observed. Even in the presence of isobutene inhibitor (at twice the initial pressure of N-ethylacetamide), as much methane as 15 % of the amount of ethylene was present in the pyrolysis of N-ethylacetamide. A rough value of 2 .2 ~ s-I at 442.2"C was obtained for the initial first order rate constant of decomposition of N-ethylacetamide into ethylene and acetaniide in the presence of inhibitor. TABLE 3.-EFFECT OF TEMPERATURE ON THE RATE CONSTANTS k,(first order rate constant) T/"C 431.6 442.1 452.2" 462.6 472.4 452.1 1 05k /s-l 2.1 4.0 6.4 10.2 17.4 28.0 k l = 1012.07*0.33 exp[(-54 003+ 1113) cal ~ O I - ~ / R T ] s-' k,(second order rate constant) 7'1°C 431.6 442.1 452.2* 462.6 472.4 482.1 kL/cm3 mol-' s-' 4.84 6.47 11.30 16.06 23 -26 35.33 k2 = 1013~75k0.85 exp[(-42 225f2856) cal rn0I-~/R2'] cm" mol-' s-l k,(rate constant for the acetic acid catalysed reaction) T/"C 421.3 432.0 442.1 452.2' 462.6 472.3 10-2k3/~m3 rno1-I s-l 2.60 3.70 5.26 7.51 1 1.06 15.02 k , = 1013.66+0.30 exp[(- 35 800+ 990) cal rn~l-~/RT] cm3 mol-' s-l * packed vessel runs.DISCUSSION The homogeneous gas phase elimination of isobutene and acetalnide in the has been suggested as proceeding through a six- pyrolysis of N-t-butyl-acetamide centred transition state : I C H 3 -c +\ NH - - H x, \ C H2 The same reasoning in favour of such a transition state may be applied for the horno- geneous inhibited first order decomposition of IPA to yield propene and acetamide. Some evidence for a cyclic transition state for the pyrolysis of N-alkylamides comes from the work of Rye.I5 He found that NN-dimethylacetamide which lacked B- hydrogen atoms did not undergo ready decomposition.All these observations suggest that an analogy can be drawn between the pyrolyses of esters and N-alkyl- acetamides which support previous suggstions. A comparison of the relative rates of pyrolysis of esters and N-alkylacetamides is shown in table 4.2456 HOMOGENEOUS G A S PHASE PYROLYSIS OF N-ISOPROPYLACETAMIDE The effect of a-methylation in N-alkylacetamides is much less than that in esters. This is interpreted in terms of greater mesomeric effect (+ M) in the amide group as compared with that in the ester group,16 and is the result of the lone pair of electrons being more readily available for conjugation on the nitrogen atom than on the oxygen atom. The inductive effect of the a-methyl substitutent will result in the delocalisation of the charge over the whole of the ainide group; whereas in the esters, the charge will become localised on the oxygen atom to some extent and hence this will be reflected in the higher polarity (heterolysis) of the alkylcarbon-oxygen bond in esters than the alkyl carbon-nitrogen bond in the N-substituted acetamides.Therefore, it TABLE 4.--RELATIVE RATE CONSTANTS OF PYROLYSIS k,(RX)/k,(EtX) compound ethyl isopropyl t-butyl ref. formate (400°C) I 20 720 9 acetate (41 1°C) 1 26 1600 9 acetamide (442OC) 1 2 20 this work is concluded that the degree of polarity in the transition state of the N-alkylacetamides is much less than that in the esters. The smaller effect of a-methyl substitution was also observed in the hydrogen bromide catalysed decomposition of aliphatic amines l7 as compared with that in the hydrogen bromide catalysed dehydration of aliphatic alcohols.From this it may be concluded that the inductive effect of substituents in the gas phase elimination reactions of the nitrogen analogues of esters and alcohols is small. The bimolecular component in the pyrolysis of IPA is similar to the pyrolysis of tlcetamide.14 Therefore a mechanism similar to that proposed for the bimolecular decomposition of acetamide may be considered. The observed products (acetonitrile, acetic acid, propene and isopropylamine) and the second order nature of the de- composition are explained in terms of bimolecular deamination CH3 \ CH(CH& / C ” 3 \ Such a carbonyl attack would be facilitated by the dipole-dipole attraction of the two carbonyl groups, and the lone pair electrons on the leaving NHCH(CH& group could approach the hydrogen atom on the other nitrogen atom, forming a bond with it to produce isopropylamine.The substituted imide (N-isopropyldiacetamide) formed will decompose rapidly in a unimolecular step to yield propene, acetonitrile and acetic acid. Catalysis by carboxylic acids does not appear to have been previously reported for gas phase reactions. Aspden l4 found that the decomposition of acetamide was autocatalysed by the acetic acid formed. In the gase phase pyrolysis of N-t-butyl- acetamide (NBA) lo and N-isopropylacetamide (IPA), the acetic acid formed also catalysed the formation of the corresponding amine.The rate constants for these catalysed decompositions were expressed by the Arrhenius equations :A . MACCOLL A N D S . S . NAGRA 2457 k,(acetamide) = 1011*32 exp(-30 100 cal rnol-l/RT) cm3 mol-i s-l k,(NBA) k,(IPA) - 1013*66 exp(-35 800 cal mol-l/RT) cm3 mol-1 s-' = 1013.56 exp( - 35 000 cal mol-l/RT) cm3 mol-1 s-I. The relative rates for the acetic acid catalysed deamination of acetamide and N-alkyl substituted acetamides at 4420°C are listed in table 5. TABLE S.-RELATIVE RATES OF CATALYSIS BY ACETIC ACID AT 442.0"c compound 106 kZIs-1 mmHg-1 relative to acetamide acetamide 3.5 1 .o IPA 11.8 3.3 NBA 17.6 5.0 It is clear that although the effect of N-alkyl substitution on the catalysed deamination reaction is not marked, it is in the right direction.The observed catalysis by acetic acid (by analogy with the acetic acid catalysed decomposition of acetamide) can also be interpreted in terms of the carbonyl attack as follows 0 H3, 4 4 - H \ c - 0, CH3-C 0 J fcy --- 0 + RNW2 CH3- C - N--H / CHg- C k II \ 0 R 0 The subsequent unimolecular decomposition of acetic anhydride l 3 is very rapid. It was observed previously lo that ketene reacts with N-alkylacetamide to form olefin, acetonitrile and acetic acid. Therefore the steps in the catalysed reaction are CH3CONHR + CH3COOH -+ RNH, + CH3COOCOCH3 CH3COOCOCH3 -+ CHzCO + CHSCOOH CHzCO + CH3CONHR + olefin + CH3CH + CH3COOH which lead to the observed kinetic form and stoichiometry of the acetic acid catalysed decomposition, i.e., rate proportional to the product of the first power of the con- centration of acid and amide.This scheme also accounts for the products derived from the two molecules of amide. One of us (S. S. N.) acknowledges the receipt of a studentship from I.C.I. Fibres Division and both are grateful to Dr. R. A. Ross and Dr. A. G. Loudon for helpful discussions. H. E. Baumgarten, R. A. Setterquist and R. E. Allan, J. Amer. Chem. SOC., 1958,80,4588. W. J. Bailey and C. N. Bird, J. Org. Chem., 1958,23,996. C. H. Delury and R. W. Kind, Chem. Rev., 1960, 60,431. C. D. Hurd and F. H. Blunk, J. Amer. Chem. Soc., 1938,60,2419. A. T. Blades, Canad. J. Chem., 1954, 52, 336. A. Maccoll and V. R. Stimson, J. Chem. SOC., 1958,2838. ' A. Maccoll, Theoretical Organic Chemistry, Kekule Symposium (Buttenvorth, London, 1958), p. 230. C. K. Ingold, Structure and Mechanism in Organic Chemistry (Cornell University Press, New York, 1953).2458 HOMOGENEOUS GAS PHASE PYROLYSIS OF N-ISOPROPYLACETAMIDE A. Maccoll and P. J. Thomas, Progr. Reactiole Kinetics, 1967, 4, 119. A. Maccoll and S. S. Nagra, J.C.S. Faraday I, 1973, 69, 1108. 1 1 J. H. S. Green, G. D. Harden, A. Maccoll and P. J. Thomas, J. Cizem. Phys., 21, 178, 1953, l 2 H. 0. Pritchard, R. 6. Sowden and A. F. Trotman-Dickenson, J. Chem. Soc., 1954, 546. l3 M. Szwarc and J. Muraswski, Darts. Faraday SOC., 1951, 47, 269. l4 J. Aspden, A. Maccoll and R. A. Ross, Trans. Faraday SOC., 1968,64,965. Is R. T. B. Rye, P1i.D. Thesis (University of London, 1960). l6 ref. (8), 2nd edn., 1969, p. 91. A. Maccoll and S. S. Nagra, J.C.S. Perkin IZ, 1974, 1099.

ISSN:0300-9599    

DOI:10.1039/F19757102450

出版商:RSC    

年代:1975

数据来源: RSC

摘要:

Transient Diffusion through a Membrane Separating Finite and Semi-infinite Volumes BY JAMES A. BARRIE::;: H. GARTH SPENCER? AND ALEXANDER QUIG Department of Chemistry, Imperial College of Science and Technology, London SW7 2AY Received 13th February, 1975 The transient diffusion of penetrant through a membrane separating a finite from a semi- infinite volume is investigated. Solutions of the diffusion equation reduce for long times to relatively simple expressions from which diffusion constants, solubilities and permeabilities may be calculated. The procedures developed are applied to the diffusion of carbon dioxide and propane through a silicone rubber membrane. Permeation experiments with a membrane separating two semi-infinite volumes have been used extensively to determine diffusioa coefficients and permeabilities of gases, vapours and solutes in a variety of membrane systems.’ For liquid diffusion measurements the diaphragm cell, in which two finite volumes are separated by a porous disc, is often used.2 Measurement of the rate of sorption of penetrant from either a semi-infinite or finite volume by a solid of known geometry is also widely used for the determination of diffusion coefficients.2* The system in which a membrane separates finite and semi-infinite volumes has not been used to the same extent for the measurement of transport parameters.Jenkins et aL4 transposed zn earlier solution for heat flow ’ to describe transport of penetrant from the semi-infinite to the finite volume through a membrane, initially at zero concentration, and developed procedures for determining the diffusion co- efficient. In this paper we present a solution for more general initial and boundary conditions and examine the potential of the system for the measurement of transport parameters.It is assumed throughout that the diffusion coefficient D is constant and that solution follows Henry’s law. A procedure is developed for obtaining diffusion coefficients, permeabilities and solubilities and is applied to the diffusion of carbon dioxide and of propane through a silicone rubber membrane. THE DIFFUSION EQUATIONS Consider a membrane of cross sectional area A and of thickness I separating penetrant at pressure p in the finite volume Y from penetrant at a constant pressure pc as shown in fig. 1 . In general p # p c and either may be zero ; the pressure in V as FIG.1 .-Schematic diagram of the system studied. t on kave froni the Department of Chemistry, Clemson University, Clemson, S.C.29631, U S A . 24592460 DIFFUSION a function of the time is required. are as follows : PC ac ax2 at D - = - THROUGH MEMBRANES The diffusion equation and boundary conditions C(L 0) = CrJ; C(x, 0) = ci; C(0, t) = c, (2) ac i ac(i, t) D(z)x=l = - - H d t ' (3) The partition coefficient K for the system is defined as K = C/Cg where C and Cg denote the concentration of penetrant in the membrane and in the volume Vrespec- tively when the two phases are in equilibrium ; the corresponding ratio of the amount of penetrant in the membrane to that in the volume V is given by H = KAZ/V.Eqn (3) relates the change in the concentration in the volume Y to the rate at which penetrant diffuses into the face at x = 1. The concentration Ct is uniform throughout the membrane. Using the Laplace transformation method the solution for C can be expressed as C(X, t) = c,+ (4) O0 [(Ci - C,)(Hz+ Q,') sin Qn + ( C , - Ci)HQn] sin(Qnx/l) exp( - DQ;t/l2) (H+ H2+ Q,")Q, sin Qn 2c n = l where the Q,, are the non-zero, positive roots of Q tan Q = H. ( 5 ) The case of a membrane with spherical geometry is considered in the Appendix. For illustrative purposes we consider five special cases of the initial and boundary concentrations designated I, . . . , V, which are readily accessible experimentally. Of these I to IV correspond to a uniform initial concentration of penetrant C, in the membrane whilst V is an example of a corresponding non-uniform initial concentra- tion.In what follows, px with X = I, . . . V denotes the pressure of penetrant in the volume V. (I) The membrane is initially free of penetrant ; the initial pressure in Y is p o and a constant zero pressure is maintained on the other side of the membrane so that Ci = C, = 0; Co # 0 and from eqn (4) * N sin(Q,x/2) exp( - DQ;t/l2) n = l (H+H2+Qi) sin Qn * c = 2c, c This result is also obtained by transformation of the corresponding solution for heat flow in a slab.5 Evaluating C at x = I and defining the Henry's law solubility constant as CT = C/p one obtains for the pressure of penetrant in the volume V r f = f A,'exp(-DQit/i2) (7) PO n = l where (8) 2 cos Qn sin Qn - 2H A,' = H+H2+Qi - Q,+cos Q, sin Q,' (11) The membrane is first equilibrated with penetrant to establish a uniform concentration throughout ; the pressure on the surface x = 0 is reduced to zero andJ .A . BARRIE, H . G . SPENCER AND A . QUIG 246 1 maintained constant so that Ci = Co # 0; Cc = 0. Proceeding as in case (I) one obtains a, (9) -- P I ' - A:' exp(-DQit/Z2) Po n = l where (10) - 2 sin Q,, 2(H2+ Q,") sin Qn (H+H2+Qi)Q, Q,+cos Qn sin Qll* (111) The membrane is equilibrated with penetrant as in (11) ; the pressure on the surface x = 0 is maintained constant whilst that in V is reduced to zero initially so that Ci = C, # 0 ; Co = 0 and the pressure p"' in Y is given by - A:' = (11) P"' O3 I - - = A,f exp( -DQit/Z"). Pc n = l Thus the fuiiction (1 -pllr/pc) decays with time at exactly the same rate as the function P'/Po- (IV) The membrane is initially free and a constant pressure is maintained C, # 0 and and the function (1-pxV/pc) decays at reported on previ~usly.~.of penetrant ; the initial pressure in V is zero on the surface x = 0, so that Ci = Co = 0; the same rate as p1I/p0. This case has been Eqn (7), (9), (1 1) and (12) may be represented by the single equation co .fX(p) = c exp( - DQ:t/z2) 1 1 = l where and Also S"(d = PXlP0 for X = 1, 11 f X ( p ) = (1 -p"p,) for x = 111, IV. A' = A"' and A" = A". (13) For large t the first term of the summation dominates andfX(p) decays exponentially with time so that Further if H is sufficiently small, tan Q1 E el, QI 2 H and eqn (14) becomes where In[f"(p)] = In A : - D Q i t / l Z .ln[fx(p)] g In AT- DHt/12 (1 4) (15) Finally we consider a special case of a non-uniform initial concentration Ci in the membrane at t = 0 ; (V) c; = -Co; x c, = 0. 12462 DIFFUSION THROUGH MEMBRANES A linear concentration gradient is established in the membrane at t = 0. Experi- mentally this may be achieved by attaching a large buffer volume to V. Once a steady state of flow has been achieved the buffer volume is then isolated at t = 0. The Laplace transformation method then gives and - PV = * A: exp(-DQ;t/l2) Po n = l where For large t the first term dominates giving as before simple exponential decay. from eqn (9, Q: < H, then in general AT > A 1' and will be close to unity. Since DETERMINATION OF TRANSPORT PARAMETERS TIME-LAGS In the early stages of the experiments, I-V, the decay of the function f X ( p ) is not a simple exponential and terms with n 2 contribute to the summation in eqn (13).One may define a time-lag characterizing the establishment of the region of simple exponential decay. Eqn (14) may be rewritten as In[f"(p)] = -T(t I - 0x1 where Ox, the time-lag, is given by For X = I and 111 and For X = II and IV and l2 liin ex = - - Qi-0 30' lim 8" = 12/6D Qi-0 as for the conventional permeation time-lag. The dependence of (In AF)/Q; on Q, is not severe as shown in fig. 2, and the limiting expressions may be used with an error of less than 1 "/o for Ql values as large as 0.2. In the same way, a time-lag 0vJ. A. BkRZIILE, €4. G. SPENCER AND A . QUIG 2463 I -0.31, t (a1 I 0.6 I 0.2 I 0.L Q I t/min FIG.3.-Plots of -logpx/po against t. (0) logpI/po ; (0) logpII/po. may be defined with a limiting value of zero and which, in general, will be small and negative. Time-lag differences may also be defined, for example, and Iim AO'*ll = 1*/20. Qi-0 1-782464 DIFFUSION THROUGH MEMBRANES In fig. 3, both type I and I1 experiments for propane in silicone rubber are shown and are discussed later. DIFFUSION COEFFICIENTS A N D PERMEABILITIES From eqn (14) one obtains for the slope of the linear region of the semilogarithmic plot which when combined with either of the time-lag equations (22) or (24) yields both D and Ql by iterative procedures. Alternatively if both type I and type I1 measure- ments are made then D and Ql are obtained more directly from eqn (28) and (26).Eqn (5) is then used to determine H from which the solubility constant CT may be obtained since H = aAZTp,/ VT, where p , and T, are respectively the standard pressure and temperature; this relation follows since the partition coefficient K may be expressed as K = oTp,/T,. The permeability F is then given by I" = DS and all three parameters are therefore evaluated from a single experiment. For many practical purposes Hand hence Q , will be small enough for the limiting expressions (23), (25) and (27) to apply with sufficient accuracy and D may be deter- mined directly from the time-lag or time-lag differences. In this region also when H is small enough to justify the approximation Qf = H then eqn (15) applies and d DKA DaATp, - In[fx(p)] z -- 1: -~ dt vz - VlT ' The quantity - VlK d pa = --- ATp, dt 1.L-f x(P>I is obtained directly from the slope of the linear region of the semilogarithmic plot and may be used as an approximation to the true permeability P r= Do.From eqn (9, (14) and (30) For H < 0.03, Pa and P differ by less than 1 % and for N ,< 0.15 they differ by less than 5 %. Finally although the concentration gradient in the membrane in these experiments is changing with time it may be shown that at large t, (dC"/ax),= = cos Ql(aCx/~x),= so that the ratio of the gradients at the two faces is independent of time. As Ql -+ 0, the gradients at x = 0 and x = I tend to become equal and the concentration gradient throughout the membrane tends to become linear. EXPERIMENTAL Carbon dioxide (grade X, British Oxygen Company) and propane (99.9 mole % ; National Physical Laboratories) were used without further purification.A lightly crosslinked mern- brane of polydimethylsiloxane was prepared by Dow Corning, U.K. Permeabilities for both gases were determined by experiments of type I and I1 (V finite) and by the conventional steady-state permeation procedure (V-+ GO). For propane only, time-lags were measured by procedures I and I1 and by the conventional permeation time-lag method. For type I and I1 experiments diffusion cells were constructed from two glass flanges in one of whichJ . A . BARRIE, H . G . SPENCER AND A . QUIG 2465 was mounted a copper gauze support. The membrane was sealed between the two flanges with Silastic 732 RTV (Dow Corning) and the outer edges of the membrane and glass flanges coated with Araldite epoxy resin.The area for diffusion was A = 5.4 cm' and the thickness I = 0.185 cm. The cells formed part of a conventional vacuum apparatus and on the un- supported side of the membrane two volumes, V, were available. Two cells were used with volumes, V, 72.4 and 36.9 cm3 in the one case and 66.2 and 30.7 cm3 in the other. The gas pressure in Y was measured with a calibrated pressure transducer (type 4-327 0-10 p.s.i. ; Bell and Howell) and the temperature of the system maintained at 30+ 0.2"C in an air thermo- stat. For conventional permeation and time-lag measurements the cell at 30+O.l0C was attached to a system with large V (+a) and the pressure on the downstream side of the membrane was recorded as a function of time using a micromanometer (" Baratron ", type 170M-6A ; M.K.S.Instruments.). RESULTS AND DISCUSSION Measurements were made in the range ofp/po, I to 0.3 and plots of ln[fx(p)] against t were linear except for the initial curvature associated with short times. Values of Pa, with different values of V, are given for both gases in table 1. As expected from eqn (31), Pa tends to decrease as V decreases corresponding to H and Q, increasing. Values of H were calculated from known values of CT determined from measurements of P and D by the conventional permeation time-lag method ; these were for propane 8.61 and for carbon dioxide 1.59 [cm3(s.t.p.) C M - ~ atm-'1. Q1 values were then obtained from eqn (5) and corrected values of P from eqn (31).Values of P,, the conventional steady-state permeability, are included for comparison. For carbon dioxide, the relatively low value of Q ensures that H is small and Ha is not significantly different from P . The higher value of cr for propane results in a significant, if not large, correction to Pa. Allowing for the various errors in the different methods of measurement, the agreement between H and F , is regarded as satisfactory. TABLE 1.--PERMEABILITIES* OF C 0 2 AND C3Hs BY TYPE I AND STEADY-STATE PERMEATION METHODS penetrant Yjcm3 pax 105 H P X 105 P, x 105 co2 72.4 3.3 0.023 3.3 72.4 3.3 0.023 3.3 3.2 36.9 3.3 0.023 3.3 C3HB 36.9 72.4 30.7 66.2 30.7 66.2 66.2 66.2 30.7 4.8 5.1 4.5 4.9 4.5 4.9 5.1 4.9 4.7 0.246 0.125 0.306 0.142 0.306 0.142 0.142 0.142 0.306 5.2 5.3 5.0 5.2 5.0 4.9 5.2 5.3 5.2 5.1 * Units of permeability : ciii' (s.t.p.) cm cm-* s-' atm-'.In the event that independent values of G are not available, D and el, as indicated previously, can be obtained from a single experiment, e.g., type I, or from a combina- tion of two experiments, e.g., types I and 11. Thus from an analysis of the semi- logarithmic plot ofy*/po in fig. 3 one obtains Ql = 0.357 and D = 6.2 x lo-'* m2 s-l. With V = 66.2 cm3 and H = 0.133 a value of 8.34 cm3(s.t.p.) ~ m - ~ atm-l is obtained for Q and hence P = 5.2 cm3(s.t.p.) cin cm-2 s -' atm-' which compares favour- ably with the corresponding value in table I .2466 DIFFUSION THROUGH MEMBRANES Time-lags for propane of types I and I1 are shown in fig.3 and are given in table 2 along with the conventional permeation time-lag. From these the corresponding D values were obtained by the iterative procedures discussed above and are also given in table 2 ; for comparison approximate values of the diffusion coefficient, calculated using the corresponding limiting expressions and denoted by D,, are included. Again the agreement between the values of D is regarded as satisfactory and for this system the differences between the D and D, values are within the experimental error. TABLE 2.-DIFFUSION COEFFICIENTS FOR C3Hs BY TYPE I, TYPE I1 AND CONVENTIONAL TIME-LAG PROCEDURES method 10381s 1 O'OD/mZ s- 1 1010D,/m2 s - l 81 - 1 . 9 2 6 . 2 [eqn (21)] 5.9 [eqn (23)] 011 0.99 5.7 [eqn (21)] 5.8 [eqn (25)] 811- 81 2 .9 1 5.9 [eqn (2611 5.9 [eqn (27)1 conventional 1 .oo 5.7 5.7 These measurements demonstrate that the pressure-decay of a gas (or the con- centration decay of a solute) in a finite-volume can be used to determine experimentally the transport parameters D, H and 0 for the penetrant in the membrane provided that D and 0 are both constant. High vacuum requirements are not so stringent as for the conventional permeation time-lag procedure, however, problems arising from leaks and " edge effects " are common to both. Finally the system provides a useful model for estimating concentration changes in cubic or spherical containers with permeable walls and has been used to examine the long-term storage performance of plastic containers.' APPENDIX Consider a volume V bound by a hollow sphere of thickness (b-a), and internal surface area A(= 47~2~).The diffusion equation is a2c 2ac 1 ac -+--= -- dr2 r dr D at C(r, 0) = Ci; and the initial and boundary conditions are C(a, 0) = Co; C(b, t ) = C, dC(a, t ) 1 X ( a , t ) ar hD at -= -- where h = KA/V. Using the Laplace Transformation method [(Co - Ci)Q,a sin Q, -(Cc - Ci)Hb] sin X 4 " r n = l C(r, t ) = cc+- c 2HQ, + 4Q, sin2 Q, - H sin 2Q, exp [ - (b-a)2 ""1. The Q,, are non-zero positive roots of and H = h(b-a).J . A . BARRIE, H . G. SPENCER AND A . QUIG 2467 Introducing Q = C/p at r = a one obtains wherepi = Ci/a. analogous problem in heat condu~tion.~ may be represented by For the special case Ci = C o ; C, = 0, eqn (A4) reduces to the solution given for the As for the plane sheet the special cases (I) to (IV) f "(p) = f A: exp[ - DQit/(b - a)2] n = l where thefX(p) are as defined in the text and 4Qn sin2 Q, 2HQn -I- 4Qn sin2 Q,, - H sin 2Q,, A! = = and 4bH sin Qn a[2HQn -k 4Q, sin2 Q,, - H sin 2Qn]' A!' = A:' = For large t, f x(p) = AT exp[ - DQ;t/(b - a12] and the decay of the functionfx(p) becomes simple exponential. In the limit b -+ a, eqn (A5) reduces to ( 5 ) and the Qn are the same as for the slab. In the limit of large a eqn (A5) again reduces to ( 5 ) and in addition eqn (A8) and (A9) become identical to (8) and (30) and the problem for the hollow sphere reduces to that for the slab. As for the slab one may define time-lags 6 X for X = I, 11, III and IV and time-lag differences A@, e.g. This work was carried out with the support of Procurement Executive, Ministry of Defence. J. Crank and G. S. Park, Di'usion in Polymers, ed. J. Crank and G. S. Park (Academic Press, New York, 1969). R. Mills and L. A. Woolf, The Diaphragm Cell (Australian National University Press, 1968). J. Crank, Mathematics of Diflusion (Clarendon, Oxford, 1956), pp. 128, 350. R. C. L. Jenkins, P. M. Nelson and L. Spirer, Trans. Furuday Suc., 1970, 66, 1391. H. S. Carslaw and J. C. Jaeger, Condz4ction of Heat in Solids (Clarendon, Oxford 1959), p. 128, 129. R. M. Barrer, J. A. Barrie and M. G. Rogers, Trans. F'vaduy Soc., 1962, 58, 2473. ' J. A. Barrie, A. Quig and H. G. Spencer, J. Appl. Polymer Sci., in press.

ISSN:0300-9599    

DOI:10.1039/F19757102459

出版商:RSC    

年代:1975

数据来源: RSC