ISSN:0300-9599    

DOI:10.1039/F198177FX001

出版商:RSC    

年代:1981

数据来源: RSC

ISSN:0300-9599    

DOI:10.1039/F198177BX003

出版商:RSC    

年代:1981

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I, 1981, 77, 9-13Excess Partial Molal Heat Capacities of Tetra-alkylAmmonium Bromides in Water -I- SulpholaneMixtures at 30 *CBY MAURIZIO CASTAGNOLO,*ANTONIO SACCO* ANDGIUSEPPE PETRELLA*Institute of Physical Chemistry, University of Bari, Via Amendola 173, 70126 Bari,ItalyReceived 13th November, 1979Enthalpies of solution of Bu,NBr and Am,NBr were measured at 25 and 35 *C in water+sulpholanemixtures within the composition range 0-20 mole percent sulpholane. Corresponding excess partial molalheat capacities, A P P , were calculated at 30 OC by the integral heat method. The results obtained have beendiscussed in terms of structural solute-solvent and solute-solute interactions.Recent studies on water+organic solvent mixtures have shown that the trends inexcess partial molal heat capacities, AC;, of structure-making solutes like tetra-n-butylammonium bromide (Bu,NBr) and tetra-n-amylammonium bromide(Am,NBr), as a function of the solvent composition give useful information onsolute-solvent and solute-solute structural interactions.We are interested in thebehaviour of electrolytes in the water + sulpholane (tetrahydrothiophene- 1,1 -dioxide)system. In a previous paper' enthalpies of solution at 30 OC of some electrolytes inthis solvent system were reported and ionic transfer enthalpies from water towater + sulpholane mixtures were calculated on the basis of the extrathermodynamicassumption: AAHO,(BPh;) = AAWS(Ph,P+). The results obtained were interpreted onthe basis of changes in water st,ructure caused by sulpholane.Following up thesestudies we have measured the enthalpies of solution of Bu,NBr and Am,NBr at 25and 35 OC in water+sulpholane mixtures in the composition range 0-20 mol %sulpholane where previous measurements1* had shown the existence of extrema forionic transport properties and for ionic transfer enthalpies. Excess partial molal heatcapacities at 30 O C have been calculated from enthalpies of solution by the integralheat m e t h ~ d . ~EXPERIMENTALWater + sulpholane mixtures weremade up by weight. Bu,NBr (Fluka puriss p.a.) and Am,NBr (K & K Laboratory) wererecrystallized three times from acetone+ether mixtures and dried in vacuo at 60 O C for 48 h.Tris(hydroxymethy1)aminomethane (THAM) (Fluka puriss p.a.) was recrystallized threetimes from water +methanol mixtures and vacuum dried at 80 O C for 48 h.Glass ampules werefilled with suitable weighed amounts of the salts and sealed by an oxy-propane flame. Enthalpiesof solution were measured by an LKB 8700-1 precision calorimetry system equipped with a100 cm3 reaction vessel. The experimental procedure for measurements has been previouslydescribed.' The calorimeter was tested by measuring the enthalpy of solution at 25 "C of THAMin excess 0.1000 mol dm-3 hydrochloric acid prepared by dilution of concentrated HC1 solution.AH, value at a THAM concentration of 5 g dm+ solvent was -71 10&3 cal mol-l, in goodagreement with the value of - 71 15 & 1 cal mol-l given by Prosen and Kilday.69Water and sulpholane were purified as describe10 THERMODYNAMICS I N WATER+SULPHOLANE MIXTURESTABLE ~.-ENTHALPIES OF SOLUTION AND HEAT CAPACITY DATA OF n-Bu,NBr AND n-Am,NBrIN WATER + SULPHOLANE MIXTURESmole fraction AHD,/cal mol-1 AHo,/cal mol-1 A P p at 30 "Csulpholane 25.00f0.03 O C 35.00f0.03 O C /cal K-l mol-lpp, at 30°C/cal K-l mol-10.000.020.040.070.100.200.000.020.040.070.100.20-2034+4-325f1750 f 21738f72351 f 33092 _+ 12900 f 73360+ 104792 f 45938 f 86484f 197060 f 7 1n-Bu,NBr1198f32054 f 22786 f 73223 f 163694 f 9n-Am,NBr3739 f 465377 f 246499 f 127362 f 47626 f 3578 10 f 29-236f 15 180f2152f 1130f 1105f287f260f3284 f 5202 f 4171 +2142f2114f775+ 10-3001 I ' , ,0 10 20mol % cosolvent296 f 3268 f 2246 2221 k 3203 f 2176+4FIG.1 .-AAPp of Bu,NBr (open symbols) and Am,NBr (filled symbols) in various aqueous binary solventmixtures at 30 O C . A, A, water+sulpholane; ., 0, water+t-butanol; 0, 0, water+dioxane; b ,water facetone; 0, water+ urea; v, water +ethylene glycolM. CASTAGNOLO, A. SACCO A N D G. PETRELLA 11RESULTSMeasurements of enthalpies of solution of Bu,NBr and Am,NBr in water+sulpholane mixtures have been carried out in the concentration range(0.5-1) x lop3 mol kg-l of solvent. In this concentration range any dependence ofsolution heat on the electrolyte concentration is within experimental error, so that theaverage of three or more measurements has been taken as solution enthalpy at infinitedilution, A P , .AHOs values at 25 and 35 "C for Bu,NBr and Am,NBr inwater + sulpholane mixtures in the solvent composition range 0.00-0.20 mole fractionsulpholane are reported in table 1 together with respective standard deviations, e A e .The table also shows excess partial molal heat capacities, ATp, for Bu,NBr andAm,NBr at 30 O C , calculated from A% by the integral heat method. The uncertainty,eAPp, in ATp values was obtained by the equation:Heat capacities of transfer, AAPp, for Bu,NBr and Am,NBr from water towater+sulpholane mixtures at 30 O C are reported in fig. 1 as a function of molepercent cosolvent, together with AAPp values at 30 O C for the same salts in otherwater + organic solvent mixtures, calculated from enthalpies of solution at 25 and35 O C by the integral heat rneth~d.~-lODISCUSSIONOur AHO,(Bu,NBr) value in water at 25 O C may be compared with the values of-2050f 15,11 -2012,12 -2036+7,13 -2020,14 -2210f20,15 -2070$3 l6 and- 2000 + 100 l7 cal mol-1 reported in the literature.The A P s value in water at 35 O Cfor the same salt is in good agreement with the value of - 260 f 25 cal mol-l reportedby Ahluwalia and co-workers.llFor Am,NBr, our AHO, value in water at 25 OC is more endothermic than the valuesof 640,l2 770 l4 and 796 & 39 la cal mol-1 reported in the literature. The value at 35 O Cin water may be favourably compared with the value of 3791 +43 cal mol-1 given byMohanty and Ahl~wa1ia.l~ The values in ref.(13), (14) arid (16) were obtained byextrapolation at infinite dilution.2 cal K-' mol-l for A P p ofBu,NBr at 30 OC, higher than the value of 173 f I l6 and in excellent agreement withvalues of 179 + 4 l1 and 176 1 .15 As regards A C p of Am,NBr in water at 30 O C , thehigh AHO, at 25 O C leads to a lower value than that of 300+8 reported by Mohantyand Ahluwalia.lsAHO, values in pure water lead to the value of 180From A p P values, it is possible to calculate the solute partial molal heat capacitiesPp, = APP+Cp (1)Ppz by the equationwhere C; is the heat capacity of the crystalline salt. In the case of Bu,NBr, a valueof 1 16.4 cal K-l mol-l may be calculated for Cp- at 30 *C by interpolation of Burnsand Verrall's l9 data. Pp, of Bu,NBr at the various solvent mixture compositions,calculated from this value, are reported in the last column of table 1.Our value of296 + 2 cal K-l mol-1 in pure water at 30 OC may be compared to the value of 288interpolated from data of Perron et aL20 Pp, of Am,NBr cannot be calculated becausethe Cp value is not found in the literature.From comparison of values of A P p in water with those in the literature it may bethought that the real uncertainty in the heat capacity data is greater than the standarddeviations reported in table 1. However, even when considering the possibl12 THERMODYNAMICS IN WATER+SULPHOLANE MIXTURESdependence of A P p on temperature in the range 25-35 OC, we maintain that theuncertainty is not more than 3%.As can be Seen in table 1, A P , values of Bu,NBr and Am,NBr at 25 and 35 OCbecome more endothermic as the percentage of sulpholane in the solvent mixtureincreases.This behaviour is different from that of other electrolytes whose solutionenthalpies in water+ sulpholane mixtures have been previously measured.l In fact, thesolution enthalpies of NaCl, NaBr, NaI, KCl, NaClO, and KClO, become more andmore exothermic in the composition range 0-0.3 mole fraction sulpholane. Solutionenthalpies of NaBPh, and Ph,PBr, on the other hand, show a sharp maximum atca. 2 mole percent sulpholane. The observed AH: trends may be explained by keepingin mind the properties of water + sulpholane mixtures. 22spectroscopic 23 and dielectric constant 24 measurements show that sulpholane alsobreaks down water structure in water-rich regions. Alkali halides and perchlorates arethought to be structure-breakers in water. Decreases in enthalpies of solution for theseelectrolytes in water + sulpholane mixtures may therefore be attributed to theirdecreased capacity as structure-breakers in mixtures with a lower degree of structurethan pure water. Walden product trends of anions in sulpholane water-rich mixtureslead to the same Bu,NBr and Am,NBr, unlike alkali halides andperchlorates, are strong structure-makers in water.25 Endothermic transfer of thesesalts from water to water + sulpholane mixtures would indicate, in agreement withArnett,2s that sulpholane increases water structure in water-rich mixtures.This is indisagreement with what has been said about ’the structural properties of these mixturesand about the behaviour of structure-breaking salts.A P , trends of Bu,NBr andAm,NBr in water + sulpholane are similar to those observed in all the water + non-electrolyte mixtures reported in the literature, apart from the fact that the co-solventincreases or decreases the water structure. A simple ‘cage’ model l2 can account forAW, trends of hydrophobic solutes in water + organic solvent mixtures quitesatisfactorily. The interpretation of the behaviour of NaBPh, and Ph,PBr seemsdifficult, as some authors think of these salts as structure-makers while others thinkof them as structure-breakers. Taking into account the structure-breaking propertiesof sulpholane, the A P , maximum observed for these salts should be caused by otherfactors than structural variations of the medium.Let us now consider heat-capacity data of Bu,NBr and Am,NBr.Ahluwalia andcoworkers have interpreted A P p of hydrophobic solutes in various aqueous binarysolvent mixtures in terms of modifications of water structure caused by the co-solvent.Thus the high A P p values of Bu,NBr and Am,NBr in water, attributed to thestructure-making ability of these hydrophobic solutes, should decrease in mixedaqueous solvents with a smaller degree of structure compared with that of water andincrease in mixed solvents with a higher degree of structure. A P p trends of Bu,NBrand Am,NBr in water-rich mixtures decrease in water +ethylene glyc01,~water + dioxane,’? 9 7 27 water + 2 o g 28 water + dimethylsulphoxide 29 andwater + morph~line,~~ while in water + t-butyl lo, 30 water + piperidine 27 andwater + acetone 7 9 29 they are characterized by maxima and minima, as can be seen infig. 1.In agreement with Ahluwalia, the maxima in AFp values of Bu,NBr andAm,NBr, observed in water-rich mixtures, would thus be attributed to the increasein structure of water determined by the presence of t-butyl alcohol, piperidine andacetone as co-solvents. In the case of water + sulpholane mixtures, the regular decreaseof A T p values would thus indicate that this solvent breaks the structure of water evenat very low concentrations.This approach, based on the degree of solvent structure, has recently been criticizedby Desnoyers et aL31 These authors suggest that trends of transfer functions oM.CASTAGNOLO, A. SACCO AND G. PETRELLA 13electrolytes from water to water + non-electrolyte mixtures may be explained on thebasis of electrolyte-non-electrolyte structural interactions. In the case of hydropho bicelectrolytes such as Bu,NBr and Am,NBr, the trend of heat capacities and of thevolumes of transfer from water to mixed aqueous solvents would be determined bythe degree of hydrophobic character of the co-solvent. Thus in the case of water + t-butyl alcohol, water + piperidine and water + acetone, A P p trends of Bu,NBr wouldreflect the existence of strong hydrophobic-hydrophobic interactions. Comparisonwith analogous curves for other co-solvents leads Desnoyers et al.to establish a scaleof hydrophobicity for non-electrolytes increasing in the following order :urea < acetamide 6 dioxane 6 dimethylsulphoxide < morpholine << piperazine < acetone < tetrahydropyrane < piperidine < t-butyl a1cohoL3lComparison between A P P trends of Bu,NBr and Am,NBr in water + sulpholaneand those in other aqueous solvent mixtures (fig. 1 ) would indicate that thehydrophobic character of sulpholane is very slight, so that this solvent would be placednext to urea and dioxane in this scale.M. Castagnolo, G. Petrella, M. Della Monica and A. Sacco, J. Solution Chem., 1979, 8, 501.G. Petrella, A. Sacco, M. Castagnolo, M. Della Monica and A. De Giglio, J. Solution Chem., 1977,6, 13.C. M. Criss and J.W. Cobble, J . Am. Chem. SOC., 1961, 83, 3223.M. Castagnolo, L. Jannelli, G . Petrella and A. Sacco, Z . Naturforsch., 1971, 26 a, 755.M. Castagnolo and G. Petrella, Electrochim. Acta, 1974, 19, 855.E. J. Prosen and M. V. Kilday, J. Res. Nat. Bur. Stand., Sect. A , 1973, 77, 581.R. K. Mohanty, T. S. Sarma, S. Subramanian and J. C . Ahluwalia, Trans. Faraday SOC., 1971, 67,305.T. S. Sarma and J. C. Ahluwalia, J. Phys. Chem., 1972, 76, 1366.R. K. Mohanty and J. C. Ahluwalia, J. Solution Chem., 1972, 1, 531.lo R. K. Mohanty, S. Sunder and J. C . Ahluwalia, J. Phys. Chem., 1972, 76, 2577.l1 T. S. Sarma, R. K. Mohanty and J. C. Ahluwalia, Trans. Faraday SOC., 1969, 65, 2333.l2 W. J. M. Heuvelsland, C. de Visser and G. Somsen, J. Phys. Chem., 1978, 82, 29.l3 R.B. Cassel and W.-Y. Wen, J. Phys. Chem., 1972, 76, 1369.C. V. Krishnan and H. L. Friedman, J. Phys. Chem., 1969, 73, 3934.l5 E. M. Arnett and J. J. Campion, J , Am. Chem. SOC., 1970, 92, 7097.l6 M. J. Mastroianni and C. M. Criss, J. Chem. Thermodyn., 1972, 4, 321.l7 R. Fuchs and P. Hagan, J . Phys. Chem., 1973, 77, 1797.R. K. Mohanty and J. C. Ahluwalia, J. Chem. Thermodyn., 1972, 4, 53.l9 T. A. Burns and R. E. Verrall, Thermochim. Acta, 1974, 9, 277.2o G. Perron, N. Desrosiers and J. C. Desnoyers, Can. J. Chem., 1976, 54, 2163.21 R. L. Benoit and G. Choux, Can. J. Chem., 1968, 46, 3215.22 D. D. McDonald, M. D. Smith and J. B. Hyne, Can. J. Chem., 1971, 49, 2818.23 0. Sciacovelli, Boll. Sci. Fac. Chim. Znd. Bologna, 1969, 18, 189.*4 0. Sciacovelli, L. Jannelli and A. Della Monica, Gazzetta, 1967, V, 1012.25 H. S. Frank and W.-Y. Wen, Discuss. Faraday Soc., 1957, 24, 133.26 E. M. Arnett, in Physico-Chemical Processes in Mixed Aqueous Solvents, ed. F. Franks (Heinemann,27 0. Kiyohara, G. Perron and J. E. Desnoyers, Can. J. Chem., 1975, 53, 2591.28 P. R. Philip, J. E. Desnoyers and A. Hade, Can. J . Chem., 1973, 51, 187.0. Kiyohara, G. Perron and J. E. Desnoyers, Can. J. Chem., 1975, 53, 3263.30 L. Avedikian, G. Perron and J. E. Desnoyers, J . Solution Chem., 1975, 4, 331.31 J. E. Desnoyers, 0. Kiyohara, G. Perron and L. Avedikian, Adv. Chem. Ser., 1976, 155, 274.London, 1967), p. 105.(PAPER 9/18 16

ISSN:0300-9599    

DOI:10.1039/F19817700009

出版商:RSC    

年代:1981

数据来源: RSC

摘要:

J. Chem. Soc., Faraday Trans. 1, 198 1, 77, 1530Analysis of the Factors Affecting Selectivity in the PartialOxidation of Benzene to Maleic AnhydridePart 1 .-Detailed Kinetics of Maleic Anhydride AdsorptionBY JEAN LUCAS, DANIEL VANDERVELL AND KENNETH C . WAUGH*The Corporate Laboratory, I.C.I., P.O. Box 1 1 , The Heath, Runcorn.Cheshire WA7 4QEReceived 2nd November, 1979The adsorption of maleic anhydride on a vanadium pentoxide-molybdenum trioxide catalyst has beenstudied using the transient techniques of temperature programmed desorption and gas adsorptionchromatography. The desorption spectra obtained over a range of adsorption temperatures show two peakshaving maxima at between 220 and 250 OC and 340 and 350 O C . Solution of the Redhead equation for anassumed value of 1013 s-' for the desorption A-factor at the peak maxima gave desorption energies ofbetween 138 and 148 kJmol-1 and 171 and 181 kJ mol-'. In spite of the high dosages used(ca.lo7 Langmuirs), the surface coverages never exceeded 6 x 10l2 sites (molecules) cm-* with the numberof high energy sites remaining constant at 3 x lo1* sites cm+ over the large range of adsorptiontemperatures; it is inferred that these sites are a defect property of the mixed oxide. The gas adsorptionchromatographic peak shapes are also analysed in terms of two adsorption bonds from which the heatsof adsorption of 79.9 and 104.2 kJ mol-l are derived. The adsorption of maleic anhydride on the mixedoxide catalyst is therefore seen to be activated, having adsorption activation energies of 63 and 71 kJ mol-'.These not inconsiderable heats of adsorption and desorption activation energies imply the possibility ofproduct inhibition in the selective oxidation of benzene to maleic anhydride.In the partial oxidation of hydrocarbons identification of the factors whichdetermine selectivity has proved elusive.Since selectivity is the result of competitiverate processes, a kinetic study of such systems would be expected to be illuminatingand indeed there have been many such analyses undertaken from this point of view.lThe results of these, however, have proved to be ambiguous.In the oxidation of benzene to maleic anhydride, for example, the kinetics andmechanism are normally described by reactions (1)-(3) :(1)(2)(3)C,H, + 40, -+ C4H203 + co + c0,+ 2H,OC,H, + 6 0 , -+ 3CO + 3 c 0 2 + 3H20.C4H203 -I- 2 0 , -P 2C0 + 2C02 + H20For benzene concentrations below 5 mol % and oxygen concentrations above20 mol % the rates of reactions (1)-(3) were found2 to be given byandk3[ c, H 63'.respectively.v3 = [C4H,03]0*74' (4)1MALEIC ANHYDRIDE ADSORPTION KINETICS 16It was concluded from these results that both the selective and unselectiveoxidations of benzene were inhibited by the maleic anhydride. This resulted from itsbeing more strongly adsorbed and therefore caused the surface to be covered (to anundetermined extent) preferentially by the product.The activation energies E, and E3 were also obtained in this study and were foundto be 94.5 and 154.7 kJ mol-l, respectively.In a separate study3 the activation energy for the oxidation of maleic anhydride,E,, was found to be 52.3 kJ mol-l, completing an apparently paradoxical picture ofa successful partial oxidation, in which the catalyst surface is covered to the greaterextent by the product whose activation energy for further oxidation is considerablylower than that of the selective reaction of the reactant to that product.It is generally accepted, however, that successful partial oxidations occur only whenthe product is table,^ inferring that E, of the above mechanism should be the greatest.Putting it in more detailed form, Ge~main,~ using his rake mechanism, has shown thatselectivity to any given compound derives from the relative rates of desorption andfurther surface oxidation.These rates, however, have not been determined. It is thepurpose of this series of papers to determine the detailed kinetics of adsorption,desorption and surface reaction of reactants and products in the partial oxidation ofbenzene to maleic anhydride with the aim thereby of specifying the nature of thesurface reactions of the adsorbed reactants and products and hence to gain insightsinto the selective reaction pathway.This paper is concerned with determining the energetics of the reversible chemi-sorption of maleic anhydride on the supported vanadium pentoxide-molybdenumtrioxide catalyst using the transient techniques of temperature programmeddesorptions and gas adsorption chromatography7 which allow the evaluation ofactivation energies of desorption, surface coverages and heats of adsorption.The following paper8 examines benzene adsorption and desorption kinetics and alsodetermines the energetics of the surface oxidation of an adsorbed benzene moleculeto form an adsorbed maleic anhydride molecule.Taken in combination, these papers allow explicit statements about productinhibition in this reaction, and also allow inferences to be made about the nature ofthe selective reactive pathway.A subsequent paperg will show the relevance of theelementary kinetics derived from these transient experiments for the overall reactionat steady state. In it, the temperature dependence of the rates of oxidation of maleicanhydride to carbon monoxide, carbon dioxide and water and of benzene to maleicanhydride, carbon monoxide, carbon dioxide and water derived from tubular reactormeasurements will be reproduced in a model in which the overall reactions areexpressed in their component elementary steps, the Arrhenius parameters of most ofthese having been obtained by transient experiments.EXPERIMENTALCATALYST PREPARATIONThe catalyst was prepared by dissolving ammonium metavandate (15.95 g) and ammoniummolybdate (5.06 g) in concentrated hydrochloric acid (80 cm3) at ca.70 "C. The a-Al,O, support(1 10 g, 60-80 mesh) was added and the solution was evaporated to dryness overnight at 120 O C .It was then calcined in 5% oxygen for 3 h at 400 O C . This produced a 12% w/w coating of thevanadium molybdenum mixture (3: 1 molar V,O,: MOO,) on the support. The surface area ofthis catalyst was 0.558 m2 g-lJ.LUCAS, D. VANDERVELL A N D K. C. W A U G H 17MATERIALSThe maleic anhydride was supplied by B.D.H. Chemicals and was > 98% pure, having amelting point in the range 52-55 OC. The gases, oxygen and helium, were supplied by BritishOxygen Co. Their specified purities were, oxygen: oxygen 99.7%, CO, < 2 xCO < 1 x to2 x (the oxygen content was measured in our laboratory at 1.5 x in a Herschoxygen meter, Englehard Industries Ltd), CO nil, CO, nil, N, < 10-21j/o.hydrocarbons as CH, < 2 x helium: helium 99.5”/,, 0, 1 xTEMPERATURE PROGRAMMED DESORPTION (T.P.D.) AND GAS ADSORPTIONCHROMATOGRAPHY (G.A.C.) EXPERIMENTSThe apparatus used to produce the desorption spectra (t.p.d.) and the adsorption isotherms(g.a.c.) of maleic anhydride on the vanadium-molybdenum catalyst is shown diagrammaticallyin fig.1. A constant stream of helium containing a fixed partial pressure of maleic anhydridewas obtained by bubbling the dry helium through liquid maleic anhydride maintained at 77 O Cby refluxing carbon tetrachloride. This mixture (1.5% maleic anhydride in He) was then passedthrough a sample loop (0.83 cm3) of the pneumatic switching valve. The maleic anhydride +- Hemixture was injected on to the column (0.4 cm i.d.) of catalyst by switching the valve (P.s.v.)and sweeping the contents of the loop on to the catalyst. The pneumatic sample valve, sampleloop and all of the piping were maintained at 130 O C .FIG.1 .-The temperature programmed desorption and gas adsorption chromatography apparatus.C = catalyst; D = detector (flame ionization or mass spectrometer); MS = molecular sieve; NV = needlevalve; PR = pressure regulator; R = rotameter; S = saturator; SV = switching valve.EX PER IM E N T A L PROCEDURE T.P.D.FLAME IONIZATION DETECTORTo minimise the possibility of re-adsorption of the desorbing material,s a short plug ofcatalyst (ca. 1 cm long: 0.499 g catalyst) was used in the temperature programmed desorptionexperiments. Prior to the injection of the maleic anhydride, the catalyst was pre-treated in astream of oxygen (50 cm3 min-l, 350 OC, 30 min). The temperature was then lowered under aflow of dry helium (50 cm3 min-l) to that at which the maleic anhydride was to be adsorbed.After injection of the maleic anhydride, the weakly adsorbed material was removed in thehelium stream, temperature programming (normally at a heating rate of 25 K min-l) beingstarted only after the detector had returned to the baseline.(Blank experiments, i.e. with thecatalyst removed, showed that the peaks obtained did not derive from desorption from the wallsof the tube nor from the glass wool used to hold the catalyst plug in place.18 MALEIC ANHYDRIDE ADSORPTION KINETICSMagnetA m p l i f i e rPyc S e r i e s 104RecorderGas Chromatography Oven Vacuums y s t e mFIG. 2.-The temperature programmed desorption apparatus employing the mass spectrometer detector.MASS SPECTROMETER DETECTORTo ensure that the peaks obtained in the desorption spectra were maleic anhydride, sometemperature programmed desorption experiments were carried out using an Edwards 60° massspectrometer detector.The apparatus is shown in fig. 2. The technique involved in its use differedslightly from that using the flame ionization detector in that the catalyst (0.210 g) was pre-treatedin an atmosphere of oxygen at 350 O C for 30 min after which the temperature was lowered toambient and the system was evacuated with valves V,, V,, V, and V, open and valve V, closed.Valves V, and V, were then closed and the maleic anhydride (ca. g) was injected into thetube containing the catalyst through the septum by means of a solids syringe. Valve V, wasclosed and the temperature was then raised to the value at which the adsorption was to bestudied (normally between 80 and 100 O C in these experiments) and maintained there for 30 min.Valves V, and V, were then opened and temperature programming was commenced when thepressure in the system had reached Torr (1 Torr = 133 N m-2), valve V, first having beenclosed so that all the desorbing material passed into the ionization chamber of the massspectrometer.The desorbing maleic anhydride was followed on the 26 and the 54 m/e peaks.Because of the considerable length of low conductance piping (ca. 40 cm long, 0.3 cm id.)connecting the column containing the catalyst to the ionization chamber of the massspectrometer and as a result of the design of that ionization chamber (a solid repeller electrodebarred direct entry of the desorbing material), the desorption peaks obtained in theseexperiments were distorted and poorly resolved. Therefore they were not used for anyquantitative measure of desorption activation energies but were used, rather, in a confirmatorysense bracketing the temperature regime in which desorption of a given compound would occur.The flame ionization detector, however, was connected to the catalyst plug by a short length(ca.1 cm, 0.16 cm id.) of tubing so that distortions of the desorption peak shapes in transportfrom the catalyst plug to the detector would be minimised.EXPERIMENTAL PROCEDURE: G.A.C.In contrast to the temperature programmed desorption experiments where only a short plugof catalyst is used to prevent re-adsorption, in the gas adsorption chromatographic experimentsa long column of catalyst is used to maximise the number of re-adsorptions of the maleicanhydride.The apparatus is therefore the same as for t.p.d. experiments (fig. 1) with theexceptions that the column length is now 20 cm (7.504 g catalyst) and that the experiment isconducted isothermally.The catalyst pre-treatment was the same as for the t.p.d. experiments in that prior to theinjection of the maleic anhydride (by sweeping out the sample loop of the pneumatic switchinJ. LUCAS, D. VANDERVELL AND K. C. WAUGH 19valvej oxygen was passed over it (50 cm3 min-l, 350 O C , 30 min). The trailing edge of the elutedpeak was logged by a PDP 11 computer and transferred to punched paper tape. In theseexperiments only a flame ionization detector was used; in separate experiments, however, theeluted material was confirmed to be maleic anhydride by sweeping it onto a column of poly-2,2-dimethyl propane-1, 3-succinate [lo% on embacel (60-80 mesh)].This column was capableof separating benzene, acetic acid, acrylic acid, p-benzoquinone, maleic anhydride and phenol.RESULTS AND DISCUSSIONTEMPERATURE PROGRAMMED DESORPTION EXPERIMENTSThe desorption spectra obtained on temperature programming the vanadium-molybdenum catalyst on which maleic anhydride had been pre-adsorbed are shownin fig. 3. These spectra obtained using the flame ionization detector show two peaks,the temperatures of whose maxima lie in the range 220-240 O C and at 350 OC.Usingthe mass spectrometer detector, no mass fragments other than those attributable tomaleic anhydride were observed in this temperature regime, confirming that they aresolely maleic anhydride.The different spectra [fig. 3 (a)-(g)] are the result of varying the temperature at which1 1 I I I150 200 250 300 350temperature / O cFIG. 3.-Desorption spectra of maleic anhydride, pre-adsorbed at different temperatures, from thevanadium-molybdenum catalyst. Adsorption temperatures: (a) 135, (b) 150, (c) 160, ( d ) 170, (e) 180(f) 190 and (g) 200 OC. (The spectra are displaced vertically for clarity.TABLE 1 .-ADSORBATE COVERAGES, DESORPTION PEAK MAXIMA AND DESORPTION ACTIVATIONdesorption activationtemperature of peak amount adsorbed/1012 energies fromadsorption maximum/OC molecules cm-2 /kJ mol-temperature -1°C peak 1 peak 2 peak 1 peak 2 total peak 1135135a1351501 50a*1601 60a*1 70"1 70b1801 80a190"190200200a22522322723622724924625424 133835 13 5034834335 135 13 5034 13403433383453393462.592.132.270.820.860.450.540.430.380.240.230.170.143.442.143.082.462.8 12.8 13.743.842.592.492.733.202.883.523.746.024.275.353.283.673.264.284.272.972.732.963.373.023.523.74139.3138.1139.3141.8139.3145.6143.5147.7143.5a These spectra are shown in fig.3. Computer simulations shown in figJ. LUCAS, D. VANDERVELL A N D K.C. WAUGH 21the maleic anhydride was adsorbed. As outlined in the experimental section, afterinjection, the maleic anhydride adsorbed on the walls etc. was removed in a streamof helium and temperature programming was commenced when the detector wasdeemed to have returned to baseline, i.e. when the rate of desorption at thattemperature was negligibly small. This meant that the time of the helium purge variedas a function of temperature, ranging from ca. 45min at the 2OOOC adsorptiontemperature to over 1 h at the 135 O C adsorption temperature. (At any one adsorptiontemperature though, the purge time remained constant,) The variations in the totalamounts adsorbed on changing temperature (table 1) can be attributed to this since,as shall be explained in detail later, on increasing the adsorption temperatures thepurge time becomes comparable to, and then greater than, the desorption half-life ofthe lower temperature peak, causing its depopulation.The higher temperature peakremains roughly constant over the complete temperature range at (3.03 f 0.51) x 10l2sites (molecules adsorbed) cm-2, the large standard deviation probably arising fromintegrating the peaks simply by dropping a vertical at the minimum. (More sophis-ticated forms of deconvolution were not used since they involved assumptions aboutthe peak shapes and therefore about the desorption kinetics.) Nevertheless, at anyone adsorption temperature (with the exception of one point at 135 *C and one at170 "C) the total amount adsorbed is within 5% of the mean value, as is the ratioof the low temperature peak to the high temperature peak.The integral of the peaks shown in fig.3 correspond to near constant total coverage[(3.78 _+ 0.48) x 10l2 molecules ~ m - ~ ] because of which, and because of the extremelylow coverage, it is impossible to ascribe the two maxima to coverage-dependentrepulsions between the adsorbate molecules, an explanation used to describe theadsorption kinetics of carbon monoxide on single crystal platinum. lo We havetherefore taken them to indicate discrete activation energies of desorption, which couldderive from different configurations of adsorption of the maleic anhydride, or fromsites on the surface of the catalyst having different heats of adsorption.The activationenergies to desorption (listed together with the coverages of these different states intable 1) are obtained by the standard method by solution of the Polanyi-Wignerequation for each peak maximum,6* 11, i.e. rate of desorption equalsgiving on solution at the maximumexp (- E,/RT,) = 0 Ed ARTA Bwhere Ed is the activation energy of desorption, A is the Arrhenius pre-exponentialterm for desorption (= 1013 s-l, assumed), C, is the number of adsorbed molecules,T, is the temperature of the peak maximum (K), is the heating rate (0.42 K s-1)and R is the gas constant (= 8.314 J K-I mol-l).The desorption activation energies obtained by this method (1 38-148 kJ mol-l and171-181 kJ mol-l) are large and imply the possibility of product inhibition in thepartial oxidation of benzene to maleic anhydride.This will occur if the maleicanhydride is adsorbed on the sites on which benzene adsorbs and reacts and if thedesorption activation energies of benzene are less than those of maleic anhydride. Acorollary then is, by the principle of microscopic reversibility, that the desorptionactivation energies ( i x . peak maxima temperatures) of maleic anhydride formed bysurface oxidation of adsorbed benzene must be identical to those found here byadsorption of maleic anhydride (see following paper)22 MALEIC ANHYDRIDE ADSORPTION KINETICSThese large desorption energies of the adsorbed maleic anhydride also suggest thatthe molecule is immobile on the surface, inferring that, in the steady-state reactingsituation in which benzene is being oxidised to maleic anhydride, these fixed productmolecules could inhibit the surface migration of chemisorbed oxygen species (shouldthis be the oxidation mechanism) to the chemisorbed benzene.[Eqn (1) shows thatat least four molecular oxygen species or eight atomic species are required for thetransformation of benzene to maleic anhydride, suggesting that some oxygen migrationis required.] If the maleic anhydride is adsorbed at an anion vacancy then not onlywill the migration be inhibited, but the total oxygen uptake will also be reduced bythis strong adsorption of the maleic anhydride.However, the activation energies so obtained depend totally on the assumed valueof 1013 s-l for the desorption pre-exponential term, which ideally should be determinedexperimentally, or at least should be justified.Independent determination of theactivation energies and of the pre-exponential terms by heating rate variation requiresfor accuracy a two order of magnitude change in heating rate;12 this is inapplicableto plugs of catalyst since at the high heating rates required (should they be achievable)large temperature gradients would be induced in the plug.Madix,13 using his desorption rate isotherms, has shown that the desorption peakshape contains the information from which the A and E values may be obtainedindependently. The height of the desorption peak is proportional to the rate ofdesorption at that temperature, while the area of the peak from that temperature tocomplete desorption is proportional to the coverage at that rate of desorption.Afundamental requirement of the method, though, is that the peak shape relates solelyto the kinetics of desorption and is not distorted by diffusion or re-adsorption.Using a variant of the technique we analysed the following maleic anhydridedesorption peaks: (i) the low temperature peak obtained at the adsorption temperatureof 135 "C and (ii) the single peak obtained at the adsorption temperature of 200 O C .For a first-order desorption process the amount desorbed at the peak maximum is(1 - l/e), i.e. 63%, and since the amount desorbed on both peaks was > 50%(second-order) but < 63 %, the process was regarded as first-order. Distortions causedby bulk gas-phase diffusion would tend to produce a more nearly symmetric peakshape. A plot of ln(peak height/remaining area) against 1/T gives the desorptionactivation energy, while the intercept (with appropriate calibration constants) givesthe A-factor.The values obtained by this method are 64 kJ mol-l and 1.1 x lo5 s-lfor the low-temperature maleic anhydride peak ( Tmax = 225 "C) and 89 kJ mol-1 and3.2 x lo5 s-l for the high-temperature maleic anhydride peak (Tmax x 350 "C).Disappointingly, these desorption activation energies are ca. half those obtainedpreviously while the pre-exponential terms are ca. eight orders of magnitude lowerthan the assumed 1013 s-l. Nevertheless, the desorption energies, 64 and 89 kJ mol-l,are still high, maintaining the idea of a fixed adsorbate.The pre-exponential values of ca.lo5 s-l appear to be too small for the followingreasons. On the basis of simple transition state theory where the desorption A-factoris given bykTh A = - exp (AS*/R) (7)an A-factor value of 1.1 x lo5 s-l would correspond to a loss in entropy of 36.5 eu(1 53 J K-l mol-l) on moving from the thermalised adsorbed state to the desorptiontransition state.This large loss in entropy seems implausible. For desorption of oxygen from the firsttransition series oxides, Germain14 regards an immobile adsorbed molecule anJ . LUCAS, D. VANDERVELL A N D K. C. WAUGH 23transition state as 'not unreasonable' and therefore sets the entropy change to zeroand the pre-exponential to 1013 s-l.Since the energies involved in maleic anhydridedesorption are large (regardless of the value of the pre-exponential) the adsorbed stateis very probably immobile and, with the concept of a fixed transition state beingacceptable for oxygen desorption from transition metal oxides where many of thedesorption energies are lower than those obtained for maleic anhydride desorption,it is reasonable that the transition state for maleic anhydride should also be immobileand that the A-factor should have a value of 1013 s-l. (A mobile transition state wouldrequire a higher value of the desorption A-factor due to the increase in entropy inmoving from the fixed adsorbed state; this would result in yet higher predictions ofthe desorption activation energies.) These spurious desorption activation energies andlow A-factors obtained by what is, in effect, line shape analysis probably derive fromdistortions of the desorption peak shape by bulk gas-phase diffusion in the heliumstream.The minimal extent to which the peaks require to be distorted can be seenby inspection of fig. 4, 5 and 6, which compare the experimental line shapes for theadsorption temperatures of 150, 160 and 170 O C with those obtained by optimisingthe desorption activation energies of a two site heteroenergetic model for fixed A-factorsof 1013 s-l and a point source of catalyst. (The optimised desorption activationenergies so obtained are listed in table 1 and are virtually identical to those obtainedby solution to the peak maximum temperature.) However, it appears to be anunavoidable consequence of carrying out temperature programmed desorptionexperiments using an inert carrier gas that bulk diffusion in the gas phase will causedistortions of the desorption line shapes, necessitating the use of 1013 s-l for thedesorption A-factor and solution of eqn (6) to obtain desorption activation energies.The low energies and A-factors obtained above are not heats of adsorption resultingfrom multiple re-adsorptions in the catalyst plug.Germain14 has shown in the worst*O OOri a -00116 00114 00112 00- 5 .3 2 10.00-u3 8 00,6 00 I0 00 2 00 4 00 6 00 8 00 10 00 I2 00 14 00 16 00 18 00 2 0 00 22 00 24 00 26 00 28 OCchart distance X lo-'FIG. 4.--Computer simulation of the desorption spectra for a pre-adsorption temperature of 150 O C .Crosses = experimental line shapes.Solid line = predicted line shapes for the desorption activation energies141.8 and 177.4 kJ mol-l24 MALEIC ANHYDRIDE ADSORPTION KINETICSIIchart distance X lo-'FIG. 5.4omputer simulation of the desorption spectra for a pre-adsorption temperature of 160 OC.Crosses = experimental line shapes. Solid line = predicted line shapes for the desorption activation energies146.9 and 176.6 kJ mo1-I.r 20 0018 ooIl 6 O 0 I r6 : .I14 00~ ' 2 0014 8 0012 10 oocU6 001* / J 0 0 Z - L I - 1 1 L A d L000 2 0 0 4 0 0 6 0 0 8 0 0 1000 1 2 0 0 1 4 0 0 1 6 0 0 1 8 0 0 2 0 0 0 2 2 0 0 2 4 0 0 2 6 0 0 2 8 0 0 3 0 0 0chart distance X lo-'FIG.6.-Computer simulation of the desorption spectra for a pre-adsorption temperature of 170 O C .Crosses = experimental line shapes. Solid line = predicted line shapes for the desorption activation energies144.4 and 174.5 kJ mol-IJ. LUCAS, D. VANDERVELL AND K. C. WAUGH 25case of Knudsen diffusion in 1000 A pores that the length corresponding to unitprobability of re-adsorption for a low activation energy of adsorption ofca. 30 kJ mol-1 is several centimetres. Here the pore length is ca. cm, the mean porediameter is ca. lo4 A and the re-adsorption activation energy is probably much greaterthan 30 kJ mol-1 (see later). Re-adsorption in the 1 cm plug of catalyst is thereforeconsidered unlikely. Furthermore, were 1.1 x lo5 and 64 kJ mol-l to correspond tothe pre-exponential and the heat of adsorption of an equilibrium constant, the valueof 1.1 x lo5 for the pre-exponential means that the adsorption process would resultin an increase in entropy of 23 eu (97 J K-l mol-l), an impossible condition.Corroborative evidence both as to the energetically heterogeneous nature of themaleic anhydride desorption and to the spurious nature of the lo5 s-l desorptionA-factors can be found in calculations of the desorption half-lives of the two statesat the different adsorption temperatures, using the rate constants comprising (i) thelo5 s-l and (ii) the 1013 s-l A-factors.Comparison of the change with adsorptiontemperature in the numbers of molecules adsorbed on these states (table 1) with thepredicted values of the desorption half-lives (remembering that, in the experiment,after adsorption at any temperature, the system was purged with helium for at least45 min before temperature programming) allows conclusions to be made about thevalidity of the values of the desorption rate constants and about the energeticallyheterogeneous nature of the adsorption.TABLE 2.-DEPENDENCE OF THE DESORPTION HALF-LIVES ( t i ) ON ADSORPTION TEMPERATUREFOR THE TWO DESORPTION RATE CONSTANTSk (low-temperature peak,T, = 225 "C)k (high-temperature peak,T, z 350 "C)adsorption1°Ctemperature 1 .1 x 105 S-1 1013 s-1 3.2 x 105 s-1 1013 S-1 e-8g OOO/RT e-l72OOO/RT e-t34 OOOlRT e-140 OOO/RT135 17 min 16.5 h 151 h 1.8 x lo5 h160 5.6 min 1.5 h 33 h 9.6 x lo3 h170 3.7 min 40 min 19 h 3.3 x lo3 h200 74 s 3.4 min 4.1 h 170 hR = 8.3 J K-l mol-1.The predicted values for the desorption half-lives for both peaks and for both setsof rate constants for each peak for the adsorption temperatures 135, 160, 170and 200 OC are listed in table 2.From it, it can be seen that regardless of the valueof the desorption rate constant [3.2 x lo5 s-l exp (-69000/R1") or1013 s-l exp (- 172000/RT), R = 8.3 J K-l mol-'1 the desorption half-life of the high-temperature peak (Tmax x 350 "C) far exceeds the 45 min purge time and so thecoverage of this peak will be unaffected by desorption during the helium purge.Inspection of table 1 shows that the total number of molecules adsorbed into this stateis virtually constant at (3.03 k0.51) x 10l2 molecules cm-2 for all adsorptiontemperatures and at dosages (ca.lo7 Langmuir) more than sufficient to saturate thesurface, suggesting that this number is an intrinsic surface property of the mixed oxidein the temperature range 130-200 OC in helium containing 15 ppm of oxygen; thesesites might well be some form of anion vacancy.For the low-temperature peak (T, x 225 "C) using the desorption rate constant,1.1 x lo5 s-l exp (- 64000/RT) (R = 8.3 J K-l mol-l), at all adsorption temperatures2 FAR 26 MALEIC ANHYDRIDE ADSORPTION KINETICSthe predicted desorption half-life is considerably shorter than the time of helium purgeso that, were it correct, this peak would not be observed. Using the desorption rateconstant, 1013 s-l exp (- 140000/RT)(R = 8.36 J K-l mol-l), thepredicteddesorptionhalf-lives change from 3.4 min at an adsorption temperature of 200 O C to 16.5 h atthe adsorption temperature of 135 "C, showing why this peak should not be seen atthe 2OOOC adsorption temperature and giving added weight to the argument thatlOI3 s-l is a good approximation to the desorption A-factor.[The predicted desorptionhalf-life at the adsorption temperature of 170 O C is of the order of the helium purgetime, inferring that the original number of sites should be double this value, i.e. shouldbe 0.8 x 10l2 sites (molecules adsorbed) cm-2. This should be the same as that recordedat 135 O C where the predicted desorption half-life is 16.5 h. This is not the case andso some significant desorption must be occurring at 135 O C which could be accountedfor if this low energy state comprised an unresolved lower-energy site, suggested bythe lack of fit to the low-temperature rise on the basis of a single (142 kJ mol-l)desorption activation energy, fig.4,5 and 6. This would also contribute to the spuriousvalue 1 . 1 x lo5 s-l obtained for the desorption A-factor by line shape analysis of thispeak.]GAS ADSORPTION CHROMATOGRAPHY EXPERIMENTS:MALEIC ANHYDRIDE ADSORPTIONThe temperature dependence of the peaks eluted after the injection of essentiallya &function of maleic anhydride is shown in fig. 7. At 270 O C and below, 100% ofthe injected maleic anhydride was reversibly adsorbed; at 290 O C , 90% was reversiblyadsorbed. At higher temperatures, however, increasing amounts of the maleicanhydride were oxidised by the catalyst, so that these peaks were not analysed forheats of adsorption as the increasing number of anion defects induced in the catalystcould influence the heat of adsorption.In equilibrium gas adsorption chromatography each point along the trailing edgeof the peaks (fig.7) is the differential of the number of moles adsorbed with respectto the gas-phase concentration,' i.e. the differential of the isotherm, so that a fit canbe made to these trailing edges for a given form of the isotherm. Giddings and Eyring15have proposed that peak tailing in gas-solid chromatography derives from two (ormore) different kinds of adsorption site. One of these is the 'normal' adsorption siteleading to rapid molecular exchange and is responsible for the main chromatographic'*t'iretention time/s-\ ~-78LOFIG.7.-Temperature dependence of the maleic anhydride gas adsorption chromatographic peak shapeson the vanadium-molybdenum catalyst. Adsorption temperatures: (a) 230, (b) 240, (c) 260, (d) 270, (e) 280and cf) 290 "CJ. LUCAS, D. VANDERVELL A N D K. C. WAUGH 27effect. The other (or others) is a ‘tail-producing, site which is relatively scarce but hasa slow desorption rate; molecules are adsorbed only infrequently on these sites, butonce adsorbed are strongly held and are released only after the bulk of the zone haspassed, thereby increasing the tail.Also, since the temperature programmed desorption experiments had shown theexistence of at least two distinct bonds to the surface, the trailing edges of the gasadsorption peaks were analysed in terms of a two-site Langmuir adsorptionisotherm.16 The expression of this isnumber of moles adsorbed = o +(1 -a)where o is the number of moles of adsorption sites, K, is the equilibrium constantfor site 1 = 10A1 exp-AHJRT, K2 is the equilibrium constant for site2 = 10Az exp-AH2/RT, a is the fraction of the surface on which the heat ofadsorption is -AH, and C is the equilibrium concentration (mol ~ m - ~ ) of theadsorbate.Although eqn (8) contains six variables, this is the minimum number of parametersrequired to describe a two-site Langmuir isotherm.Addition of another site, as appearsto be suggested by a close analysis of the temperature programmed desorption spectra,would lead to an over-definition of the isotherm.The temperature dependence of the peak trailing edges were therefore ‘fitted’ byvarying the six parameters A,, A,, AH,, AH,, a and o of the differential of the isotherm,again using a weighted Gauss-Newton least-squares program.The differential of theisotherm is expressed as:retention time to a given concentration, C = -where F is the carrier gas volume flow rate. (Due to the low surface area and theconsequent wide pores of the catalyst, intra-particle diffusion can be ignored.)The ‘best fit’ parameters are listed in table 3 and a comparison between thepredicted and experimental trailing edges is shown in fig. 8. The crosses are theexperimental points; only every fifth point recorded is plotted but three separate setsTABLE 3.-BEST FIT PARAMETERS FOR A TWO-SITE LANGMUIR ADSORPTION MODELfraction of surface maximum predictedheat of adsorption pre-exponential subtending a heat of combined coverage/kJ mol-l terms/cm3 mol-l adsorption of AHl /sites-AH1 79.9f0.5 A , -0.06k0.4 0.987 +_ 0.007 2.5k2.1 x 1013-AH2 104.2k0.5 A , -0.29k0.22PARAMETER CORRELATION MATRIXaa 1 .o 1 OAllOAl 0.996 1 .o -AHl10.4% 0.710 -0.724 -0.311 1 .o -AH2-AH, 0.458 0.393 1 .o 1 OAz-AH2 0.744 0.755 0.366 -0.996 1 .o 0.0.0.999 -0.995 -0.478 0.720 -0.757 1 .o2-280 400 360 32v) Y’2 0 28$ 0 2 4E%a, 0 169 r:. .? 0 20 -e3.f:% .- 0 1 *f0.080 040 02MALEIC ANHYDRIDE ADSORPTION KINETICSretention distance/ lo-’ sFIG.8.-Comparison of predicted and experimental trailing edges of the gas adsorption peaks for a two-site adsorption model. Crosses = experimental points. Solid line = prediction for the heats of adsorption79.9 and 104.2 kJ mol-l, the former bond occurring on 98.704 of the adsorbing surface which itself ismaximally only 2.5 x lo1* sites cm-’. Adsorption temperatures: (a) 230, (b) 240, (c) 260, ( d ) 270 and( e ) 280 O C .of experimental data are plotted for each temperature. The fit is well withinexperimental error, the lack of correspondence at the maxima of the low-temperaturepeaks (230, 240 “C) being the result of Fick’s law diffusional effects which alwayssuperimpose on the thermodynamics at the maxima; the peak spreading at themaximum derives from large concentration gradients.While the correlation matrix (also shown in table 3) shows a significant correlation,i.e.greater than 0.7, between most of the parameters of the model, the heat ofadsorption found on the majority of the surface sites, AHl, is not correlated to anyof the parameters, conferring reasonable confidence in the value obtained. Thisparameter determines the temperature dependence of the isotherms; the values of theremaining five parameters are those which, without constraints, minimise the sum ofthe squares of the residuals around this energy.The adsorption isotherms of fig. 9 are obtained by ‘stripwi~e’~~ integration of thegas adsorption peaks (fig. 7). The experimental isotherms are curved (crossed points).They might be interpreted in coverage-dependent isosteric heats of adsorption,resulting from boundary layer limitations on adsorption on a semiconductor.Theyare, however, described, within experimental error, in a two-site model, the values ofwhich are given in table 3; the basis for the model derives from the temperatureprogrammed desorption experiments. The curvature of the isotherm derives from thesmall number of high energy sites which, depending on the temperature, are filled atlow concentration of the adsorbateJ. LUCAS, D. VANDERVELL A N D K. C. WAUGH 290.00 0.10 0.20 0.33 0.40 0.50 0.60 0.70gas phase concentration/ 1 0-9 mol ~ r n - ~FIG. 9.-Adsorption isotherms of maleic anhydride on the vanadium-molybdenum catalyst.Crosses =experimental points. Solid line = prediction on the basis of the two-site Langmuir model. Adsorptiontemperatures: (a) 230, (b) 240, (c) 260, (d) 270 and (e) 280 OC.1 = 176.6FIG. 10.-Potential energy diagram for maleic anhydride adsorption on the vanadium-molybdenumcatalyst. All numbers on the diagram are in units of kJ mol-I30 MALEIC ANHYDRIDE ADSORPTION KINETICSThe crude adsorption potential energy diagram for maleic anhydride adsorptionon the vanadium-molybdenum catalyst (fig. 10) can be obtained by combining theresults of the temperature programmed desorption experiments with those of the gasadsorption chromatography ; the higher heat of adsorption has been associated withthe larger activation ecergy to desorption on the expectation that the stronger bondwould have the higher energy to desorption.What then emerges is that, incontradistinction to adsorption on metals, the adsorption of maleic anhydride on thismixed oxide catalyst is activated, having considerable energy barriers of 63 and71 kJ mol-l.Finally, the predicted maximum total number of adsorption sites listed in table 3,2.6 x 1013 sites (i.e. molecules adsorbed) cm-2, is in rough accord with the numberobtained in the temperature programmed desorption experiments, 6 x 10l2 sites cm-2.The prediction as to the maximum number of low energy sites, 2.56 x 1013 sites cm-2,is greater than the maximum obtained in the t.p.d. experiments, 2.6 x 10l2 sites cm-2,suggesting that in the t.p.d. experiments some desorption was taking place even atthe lowest adsorption temperature. It appears, though, that 1013 sites cm-2, whichcould be anion vacancies, is an intrinsic property of this mixed oxide in thetemperature range 130-300 *C in helium containing 15 ppm oxygen. In air, in thesame temperature range, their number would be expected to diminish.We thank Dr I. B. Parker and Mr M. L. Harris for writing the fitting programs.J. E. Germain, Catalytic Conversion of Hydrocarbons (Academic Press, New York, 1969), chap. 5, p.256 and references therein.I. I. Ioffe and A. G. Lyubarskii, Kinet. Katal., 1962, 3, 261.I. I. Ioffe and A. G. Lyubarskii, Kinet. Katal., 1963, 4, 294.D. A. Dowden, in Surface Science (Int. Atomic Energy Agency, Vienna, 1975), vol. 2, p. 215.J. E. Germain, Intra-Science Chem. Rep., 1972, 6, 101.R. J. Cvetanovic and Y. Amenomiya, in Catalysis Reviews, ed. H. Heinemann (Marcel Dekker, N.Y.,1972), vol. 6, p. 21.K. C. Waugh, J . Chromatogr., 1978, 155, 83.J. Lucas, D. Vandervell and K. C. Waugh, J. Chem. SOC. Faraday Trans. 1,1981, 77, 31.J. Lueas, D. Vandervell and K. C. Waugh, J. Chem. SOC., Faraday Trans. I , in press.lo R. A. Shigeishi and D. A. King, Surf. Sci., 1976, 58, 379.l1 P. A. Redhead, Trans. Faraday SOC., 1961, 57, 641.l2 P. A. Redhead, Vacuum, 1962, 12, 203.l3 J. L. Falconer and R. J. Madix, J. Catal., 1977, 48, 262.l4 B. Halpern and J. E. Germain, J. Catal., 1975, 37, 44.l5 J. C. Giddings and H. Eyring, J. Phys. Chem., 1955, 59, 416.l6 I. Langmuir, J. Am. Chem. SOC., 1918, 40, 1361.l7 A. V. Kiselev and Y. I. Yashin, Gas Adsorpfion Chromatography (Plenum Press, New York, 1972),chap. 4, p. 104.(PAPER 9/1767

ISSN:0300-9599    

DOI:10.1039/F19817700015

出版商:RSC    

年代:1981

数据来源: RSC

摘要:

J. Chem. Soc., Faraday Trans. I, 1981, 77, 31-48Analysis of the Factors Affecting Selectivity in the PartialOxidation of Benzene to Maleic AnhydridePart 2.-Detailed Kinetics of Benzene Adsorption and Surface ReactionBY JEAN LUCAS, DANIEL VANDERVELL AND KENNETH C. WAUGH*The Corporate Laboratory, I.C.I., P.O. Box 1 1 , The Heath, Runcorn,Cheshire WA7 4QEReceived 2nd November, 1979The kinetics of the oxidation of benzene to maleic anhydride over a vanadium pentoxide-molybdenumtrioxide catalyst have been studied using a combination of transient techniques. Temperature programmeddesorption (t.p.d.) has shown (i) the existence of a weakly-bound molecular oxygen species having adesorption activation energy of between 105 and 113 kJ mol-l and (ii) that the adsorption of benzene isheteroenergetic having three sets of desorption activation energies of between 97.9 and 104.2 kJ mol-l, 107.9and 119.2 kJ mol-l and 130.5 and 133.5 kJ mol-l.(As before, these desorption energies are obtained bysolution of the Redhead equation for the peak maximum temperature for an assumed 1013 s-l A-factor.)In spite of the high benzene dosages used in these studies (ca. 3 x 1Olo Langmuir, 1 Langmuir =1.33 x N m-2 s) the total coverages by these sites never exceeded 4 x 10" cm-2, suggesting that theyare an intrinsic defect property of the catalyst.Temperature programmed reaction spectroscopy (t.p.r.s.) has shown that (i) the overall oxidation ofbenzene to maleic anhydride is rate limited in the desorption of the product, the desorption activationenergies of the reactively produced maleic anhydride (in the ranges 143.5-148.1 kJ mol-' and166.1-169.9 kJ mol-l) being identical to those obtained for pre-adsorbed maleic anhydride, (ii) theactivation energy for the surface oxidation of the adsorbed benzene to adsorbed maleic anhydride is low,having a value of 3 1.4 kJ mol-', (iii) the reactively produced maleic anhydride occupies the site upon whichits precursor, benzene, was chemisorbed, inhibiting further benzene adsorption on that site and (iv) fromthe population distribution of the adsorbed maleic anhydride on the energetically heterogeneous surface,it was formed at the benzene adsorption temperature (50 "C).The immobility of the adsorbed maleic anhydride suggests that the oxygen is transported to the adsorbedbenzene, which in combination with the low surface oxidation activation energy appears to preclude theinvolvement of lattice oxygen in this step, inferring that the oxidant is a mobile chemisorbed species, possiblythe weakly bound molecular state seen in the t.p.d.experiments.Gas adsorption chromatography (g.a.c.) shows the heats of adsorption of benzene to be low, 58.6 and69.5 kJ mol-l, the latter heat being a composite of two heats which cannot be resolved. The maximumnumber of sites available for benzene adsorption is only 2.7 x 10l2 cnP2 of which 97.5% have a heat ofadsorption of 58.6 kJ mol-l. Benzene adsorption is therefore weakly activated.The previous paper' has shown that maleic anhydride is extremely strongly andenergetically heterogeneously held to the surface of a vanadium pentoxide-molybdenum trioxide catalyst (3: 1) normally used in the oxidation of benzene tomaleic anhydride.This would result in the overall oxidation of benzene to maleicanhydride being rate limited by the need to desorb the product only if the maleicanhydride produced on the surface by benzene oxidation had the same desorptionenergies as the maleic anhydride adsorbed from the gas phase, these energiesconstituting the maximum energy barrier in the overall reaction pathway.It is the purpose of this paper to analyse this proposition by evaluating separatelythe kinetics of benzene adsorption, desorption and surface reaction, i.e. to enumeratethe rate constants of Germain's2 rake mechanism, using a combination of transienttechniques, viz.temperature programmed de~orption,~ temperature programmedreaction spectroscopy4 and gas adsorption chr~matography.~332 BENZENE PARTIAL OXIDATION KINETICSAdditionally, the mechanism of the selective oxidation is examined. Questions asto the nature of the oxidant (chemisorbed or lattice) and whether the reactanthydrocarbon is mobile on the surface are addressed.EXPERIMENTALThe method of preparation of the catalyst, a supported 3 : 1 molar V,O,: MOO, catalyst, hasbeen described in the previous paper. The principles of the techniques applied, temperatureprogrammed desorption (t.p.d.) and gas adsorption chromatography (g.a.c.) have also beendescribed there and will be elaborated on here only where they differ from the previouspublication.MATERIALSThe benzene was AnalaR grade supplied by Hopkin and Williams (Essex, England).Itsspecification was that not less than 95% of it boiled in the range 79.5-80.5 OC and its impuritieswere: sulphur-containing compounds 3 x thiophene 2 x water 5 x Thegases, oxygen and helium, were supplied by British Oxygen. Their purities have been defined.'EXPERIMENTAL PROCEDURE: TEMPERATURE PROGRAMMED DESORPTIONAs before,l both a mass spectrometer and a flame ionization detector were used in thetemperature programmed desorption experiments and they were identical to those describedearlier.'MASS SPECTROMETER DETECTORBENZENE DESORPTION SPECTRA. The technique involved has been described' except thathere the adsorbate (benzene) was introduced by a liquid syringe (0.1 mm3) giving a dosage ofca.10" Langmuir. The benzene desorption spectra of fig. 1 were obtained by following the 78mass number peak for adsorption temperatures of 60 [curve (a)], 75 [curve (b)] and 90 OC [curve(41 * As explained,' the mass spectrometer detector was used only for the semi-quantitativecorrelation of chemical identity of a desorbing species with temperature, i.e. to estimate upperlimits on the temperatures in the desorption spectra, above which one could predict that a givencompound would have been completely desorbed from the catalyst surface, or below whichthe desorption of another compound would not have begun; this is important later whenassigning chemical identity using the temperatures of peak maxima in desorption spectraobtained employing the flame ionization detector.MALEIC ANHYDRIDE DESORPTION SPECTRA FROM BENZENE ADSORPTION.The desorp-tion spectrum of maleic anhydride deriving from benzene adsorption on the catalyst at 150 O Cis shown in fig. 2(a). The catalyst pre-treatment in oxygen and the technique employed forbenzene adsorption were identical to that described above. The spectrum was obtained byfollowing the 54 mass number peak on the mass spectrometer, i.e. the CH=CH-CO fragment,the largest fragment in the maleic anhydride cracking pattern.Surprisingly the better defined maleic anhydride spectrum [fig. 2 (b)] was obtained after havingadsorbed the benzene at 70 "C following the 26 mass-to-charge ratio peak (the second largestpeak in the maleic anhydride cracking pattern on the mass spectrometer).OXYGEN DESORPTION SPECTRA.In obtainirig the oxygen desorption spectra of fig. 3 thecatalyst had been pre-treated in the method adopted as standard here, i.e. by holding it underone atmosphere of oxygen at 350 OC for 30 min prior to cooling to the adsorption temperature;this latter was room temperature and the catalyst was maintained at that condition for a further30 min under the one atmosphere oxygen pressure. The system was then evacuated toTorr, after which the programmed temperature increase was begun, following mass number32 on the mass spectrometer detectorJ. LUCAS, D. VANDERVELL AND K. C. WAUGH 33FLAME IONIZATION DETECTORThe technique involved and the apparatus used in obtaining the desorption spectra using theflame ionization detector have been described before.' As in the desorption spectra taken usingthe mass spectrometer detector, prior to the injection of benzene, the catalyst was pre-treatedwith oxygen, in this case a flow of oxygen (50 cm3 min-l, 30 min, i.e.4 x 10l2 Langrnuir,350°C). The temperature of the catalyst was then reduced to the adsorption value under astream of helium.A constant stream of benzene in helium was prepared by bubbling the helium through thebenzene contained in two saturators, in tandem, maintained at 10 OC and the catalyst was dosedwith the benzene + helium mixture in one of two ways: (a) as a pulse and (b) as a long squarewave.(a) The pulse was obtained by sweeping out the sample loop (0.372 cm3) of the pneumaticswitching valve through which the benzene + helium (6%) stream had flowed.In this way thedesorption spectra shown in fig. 4 were obtained for adsorption temperatures of 50 [curves (a)],60 [curves (b)], 70 [curves (c)], 80 [curves (41, 90 [curves (e)] and 100 OC [curve (f)]. Table 1lists the experimental conditions of carrier-gas flow rate, duration of helium purge for theremoval of loosely held and gas-phase benzene, weight of catalyst and the temperatureprogramming rate. Blank experiments, i.e. catalyst removed, showed that all of the peaks listedin tables 1-3 derived from the catalyst alone and not from the walls of the tubing nor fromthe glass wool.(b) The long square wave was obtained simply by using the pneumatic valve to switch flowsfrom helium to the helium + benzene mixture.This technique allowed variations to be madein the dosage of the adsorbate. Additionally, on changing the adsorption temperature, theduration of flow of carrier gas to remove the gas-phase and loosely held benzene was changedso as to leave detectable quantities of adsorbate on the surface at high adsorption temperatures.Typical desorption spectra for an adsorption temperature of 50 OC are shown in fig. 5[(a)and (b)], curve 2 of fig. 5 (a) was taken directly after curve 1, the catalyst having been maintainedunder oxygen at 350 OC for 30 min between benzene dosages ( 3 x 1Olo Langmuir). Thedesorption spectrum [fig.5(b)] was taken directly after the spectrum described by line 2 offig. 5(a) without any intervening oxidation step and with a slightly higher heating rate. Fig. 6shows the spectra obtained for benzene adsorption at 75 [curve (a)], 90 [curve (b)] and 100 OC[curve (c)]. The experimental conditions which produced these spectra (fig. 5 and 6) are listedin table 2.The desorption spectrum shown in fig. 7 is typical of a number of such in which in additionto varying the dosage, by changing the length of time of benzene flow over the catalyst, a furtherperiod of stopped flow in which the catalyst was immersed in a static benzene + helium mixturewas included just before purging in the helium flow. A complete listing of these experimentalconditions is given in table 3.EXPERIMENTAL PROCEDURE: GAS ADSORPTION CHROMATOGRAPHYAs in the previous paper,l the application of gas adsorption chromatography entails the useof a long column of adsorbent (8.339 g of the mixed oxide catalyst) and the experiment wasconducted isothermally by pulsing a &function of benzene onto the column [by sweeping outthe sample loop (0.372 cm3) of the pneumatic sample valve] and logging the trailing edge ofthe eluted peak (the differential of the adsorption isotherm) on a PDP 1 1 computer.The catalystwas pre-treated in oxygen (50 cm3 min-l, 1 atm, 30 min, 350 "C) before the temperature wasreduced to the desired value in a helium flow. The gas adsorption chromatographic peak shapesshown in fig. 8 for the temperature range 80-140 OC were obtained in this way.Above 140 OCthe eluted benzene peak fell to less than 100% of the injected material and so no attempt wasmade to obtain the benzene adsorption isotherms above this temperature when reaction couldaffect the eluted line shapes and hence the heats of adsorption34 BENZENE PARTIAL OXIDATION KINETICSRESULTS AND DISCUSSIONTEMPERATURE PROGRAMMED DESORPTIONMASS SPECTROMETER DETECTORBENZENE DESORPTION. As explained in the previous paper,' because of thedistortions in the peak shapes and the unquantifiable delays in their arrival at theionization section of the mass spectrometer detector, resulting from having connectedthe catalyst to the detector by low conductance piping, the spectra obtained using thismachine could only be used to give an upper bound in temperature for a givendesorbing species, above which no further desorption would occur.Fig. 1 thereforeshows that benzene desorption is complete at 200 O C for an adsorption temperatureof 60 OC and by 250 *C for adsorption temperatures of 75 and 90 OC. The peak shapes,however, are not typical of first-order de~orption,~ having high temperature tails. Thisand the fact that the temperatures of the peak maxima increase with increasingadsorption temperature could be taken to infer an energetically heterogeneous surfacefor the benzene adsorption; this will be elaborated upon later.I I I I I100 150 200 250temperature /"CFIG. 1.-The desorption spectra of benzene pre-adsorbed at different temperatures on the vanadium-molybdenum catalyst, following the 78 mass-to-charge peak on the mass spectrometer.Adsorptiontemperature: (a) 60, (b) 75 and (c) 90 OC.MALEIC ANHYDRIDE DESORPTION/BENZENE PRE-ADSORPTION. Fig. 2(a) and(b) are the maleic anhydride desorption spectra which result from the adsorption ofbenzene on the catalyst at 150 and 70 O C , respectively. The better defined spectrumof fig. 2(b) is obtained following the 26 mass number peak, the C H S H fragment.Combination of fig. 1 and 2 show that in the desorption spectra obtained after benzeneadsorption using a flame ionization detector (fig. 4-7), the desorption peaks obtainedup to 250 O C derive from benzene desorption and those from 250 OC upward derivefrom maleic anhydride desorption.OXYGEN DESORPTION.The desorption spectra of fig. 3 are obtained on the massspectrometer detector following the 32 mass number peak after pre-treating thecatalyst with oxygen (1 atm) for 30 min at 350 "C and a further 30 min at roomtemperature. Although the oxygen dosage was constant, the surface coverage justbefore programmed temperature increase began was variable, depending on theconstancy of the system's pumping speed ; temperature programming was begun whenthe whole system had been evacuated to TorrJ. LUCAS, D. VANDERVELL A N D K. C. WAUGH 35Nevertheless, the movement of the peak maxima to higher temperatures withincreasing coverage and the high temperature tail of the desorption peak suggest thatthe surface is energetically heterogeneous for oxygen desorption.The temperaturesof the maxima are 105, 110, 123 and 133 OC, suggesting a weakly-bound molecularspecies seen previously on some of the oxides of the first transition series but not onV205.6 The inclusion of MOO, to its limit of solubility in V,O, results in a catalystwith a maximum V4+ cation concentration which produces maximum selectivity inthe oxidation of benzene to maleic anhydride.2 Kasansky and co-workerss haveshown that molecular oxygen adsorption occurs at the V4+ cation; our observationof a weakly-bound molecular oxygen state on the mixed oxide which is not observedon the minimally defected V2056 is consistent with these findings.ha, z2 1 5 QI I I 1 I2 00 250 3 0 0 3 5 0 4 0 0temperature /"C% I2 00 2 5 0 3 0 0 350temperature / "C40 0FIG.2.-The desorption spectra of maleic anhydride from the vanadium-molybdenum catalyst resultingfrom prior adsorption of benzene (1 mm3); (a) mass 54, the CH=CH-CO fragment, followed on the massspectrometer, benzene adsorption temperature 150 OC. The curves show the reproducibility of theexperiments and are displaced vertically for clarity; (6) mass 26 followed on the mass spectrometer, benzeneadsorption temperature 70 OC.For the arguments outlined in the previous paper1 which now correspond exactlywith those of Halpern and Germain,6 we obtain the desorption activation energiesof these weakly-bound oxygen species by solution of the Redhead equation36 BENZENE PARTIAL OXIDATION KINETICSwhere E is the desorption activation energy (J mol-l), A is the desorption A-factor(s-l), R is the gas constant (J K-l mol-l), /? is the heating rate (K s-l), and T, is thetemperature of the peak maximum (K) for first-order desorption. For these peakmaximum temperatures desorption activation energies of 105.4, 106.3, 1 10.5 and112.6 kJ mol-l, respectively, are obtained for an assumed value of 1013 s-l for thedesorption A-factor.1 I I5 0 150 2 5 0ternperaturel'cFIG. 3 .4 x y g e n desorption spectra from the vanadium-molybdenum catalyst after an oxygen pre-treatmentof 1 . 4 ~ 1OI2 Langmuir at 350 "C followed by 1.4 x 10l2 Langmuir at room temperature. The 32mass-to-charge ratio peak was followed on the mass spectrometer detector. The baseline for each curveis displaced for clarity.The larger amounts adsorbed derive from pumping for decreasing time lengths priorto desorption.FLAME IONIZATION DETECTORFig. 4-7 are the desorption spectra obtained using the flame ionization detector.Having confirmed using the mass spectrometer detector that up to 250 O C thedesorbing peaks are solely benzene, we can affirm that the spectra shown in fig. 4 derivesolely from benzene desorption. Only one peak maximum temperature is observed,increasing from 103 O C for an adsorption temperature of 50 O C to 121 O C for anadsorption temperature of 100 OC. Although the benzene dosages are not inconsiderable(ca. 2 x lo7 Langmuir) the coverages are extremely small, ranging from a maximumvalue of 8.4 x lo1* molecules cm-2 (ca.of a monolayer) at 50 O C to 2.5 x lo9molecules cm-2 (ca. of a monolayer) at 100 O C . Solution of eqn (1) for the peakmaximum gives the desorption energies listed in table 1. These range from104.6 kJ mol-l for the adsorption temperature of 50 OC (coverage 8.4 x 1O1O moleculescm-2) to 109.2 kJ mol-l for the adsorption temperature of 100 OC (coverage 2.5 x lo9molecules cm-2)J. LUCAS, D. VANDERVELL AND K. C. WAUGH 37l ~ ~ l ~ 60 80 100 120 140temperature/ "CFIG. 4.-Benzene desorption spectra obtained on the flame ionization detector following benzeneadsorption (7 x lo6 Langmuir dosage) at (a) 50, (b) 60, (c) 70, (6) 80, (e) 90 and cf) 100 O C .50 100 150 200 25050 100 150 200 250temperature / "CFIG. 5.-Benzene desorption spectra obtained on the flame ionization detector following 3 x 1Olo Langmuirbenzene dosage at 50 O C (a) heating rate 0.42 K s-l, (b) heating rate 0.5 K s-l.On the figures (a) and (b)refer to benzene and maleic anhydride, respectively38 BENZENE PARTIAL OXIDATION KINETICS1350~75 100 150 200 250 300temperature/ "CFIG. 6.-Desorption spectra obtained from benzene adsorption (3 x 1Olo Langmuir dosage) at (a) 75, (h)90 and (c) 100°C, flame ionization detector. The peaks are (1) benzene, (2) benzene and (3) maleicanhydride.temperature/ "CFIG. 7.-Desorption spectra obtained from benzene adsorption (6 x 1Olo Langmuir dosage) at 50 OC - flameionization detector. The peaks are (a) benzene, (6) and (c) maleic anhydride.TABLE 1 .-BENZENE COVERAGES AND DESORPTION ACTIVATION ENERGIEStotal numberadsorption of molecules desorptiontemperature adsorbed coverage/ 1 0lo T,,, activation energy,I"c 1014 molecules cm-2 1°C E,/kJ mol-l~ ~~50 2.34 8.40 103 104.650 2.35 8.44 103 104.66060707080809090901001.60 5.751.36 4.881.13 4.060.90 3.230.69 2.480.64 2.300.22 0.790.24 0.861051051101101141141171180.290.0705.005.006.306.307.507.508.808.81.04 117 108.80.25 121 109.2Weight of catalyst = 0.499 g; surface area of catalyst = 0.558 m2 g-l; rate of temperatureincrease = 0.42 K s-l; total amount of benzene injected in the pulse = 5.33 x lo1' molecules(ca.2 x lo7 L); helium flow rate = 0.83 cm3 s-l; duration of helium purge prior to programmedtemperature increase = 1 h; oxygen pre-treatment = 1.4 x 10l2 Langmuir at 350 OCTABLE 2.-cOVERAGES AND ACTIVATION ENERGIES OF DESORPTION FOR BENZENE AND MALEIC ANHYDRIDEcoverages/ 1011 molecules cm-2benzene peak 4absorption dosage calculated T*,Y./OCtemperature /lolo peak 1 peak 2 peak 3 as maleic/OC Langmuir (calculated as benzene) anhydride peak 1 peak 2 peak 3 peak50" 2.7 0.54 3.36 0.50 1.87 80 123 203 32350" 2.7 1.22 1.40 0.64 3.48 80 115 195 33050b 2.7 1.26 1.22 0.21 3.30 77 122 205 34050 0.07 0.16 1.66 - 1.8 83 115 - 31750 0.09 0.19 0.46 - 0.6 83 135 - 31950 0.14 0.15 0.70 - 3.6 85 155 - 32275 0.27 0.09 0.11 - 2.30 85 130 - 33375 1.4 0.27 0.51 0.15 7.15 87 136 - 33575" 2.7 0.48 1.70 2.31 6.28 87 135 312 33090 0.27 0.34 - - 4.5 94 - - 33590 0.14 0.12 0.59 - 7.6 96 140 - 33890" 2.7 0.19 0.76 - 11.1 96 145 - 333100" 2.7 0.11 0.27 -- 14.8 102 152 - 333Weight of catalyst = 0.499 g; surface area of catalyst = 0.558 m2 g-l; rate of temperature increase = 0.42helium flow rate = 0.83 cm3 s-l; duration of helium purge prior to programmed temperature increase: 6045 min at adsorption temperature of 75 O C , 45 min at adsorption temperature of 90 O C , 45 min at adsorptionpretreatment = 1.4 x 10l2 Langmuir at 350 O C , except (").fig. 8(b), no oxygen pretreatment,purge; fig. 9.a fig. 8(a)TABLE 3 .-COVERAGES AND ACTIVATION ENERGIES OF DESORPTION FOR BENZENE AND MALEIC ANHYDRIDEcoverages/l 0l1 molecules cmP2duration stopped flowbenzene of helium time peak 1 peak 2 peak 3 Tm,x/"dosage/ 1O1O purge /min (calculated (calculated as maleicLangmuir /min (see text) as benzene) anhydride) peak 1 peak5.04.25.83.233.53.25.96.75.9"5.96060757575756075757501010101010202022.5203.073.900.831.024.284.701.781.312.621.6713.38.89.74.323.18.911.823.312.713.88.28.26.13.82.43.414.74.27.18.9100 246103 257107 250100 24795 24099 24397 245101 243103 245106 245Weightofcatalyst = 0.4729 g;surfaceareaofcatalyst = 0.558 m2 g-l; rateoftemperatureincrease = 0.42oxygen pretreatment = 1.4 x loi2 Langmuir at 350 O C ; benzene adsorption temperature = 50 O C .a figJ. LUCAS, D. VANDERVELL AND K . C. WAUGH 41The larger coverages (ca.10l2 molecules benzene adsorbed cmP2) listed in table 2 areobtained for much larger benzene dosages (ca. lo3 times those of table 1 , fig. 4). Typicaldesorption spectra shown in fig. 5 are obtained for an adsorption temperature of 50 OCand a benzene dosage of 2.7 x 1 O 1 O Langmuir. These larger coverages allow resolutionof the benzene desorption spectra (fig. 4) into three separate desorption peak maxima.Curve 1 of fig. 5(a) is obtained after the first adsorption of benzene on thevanadium-molybdenum catalyst which had been pre-treated with oxygen (1 0I2Langmuir at 350 OC). Curve 2 [fig. 5(a)] was obtained from the same plug of catalystas curve 1 directly after it, having treated the catalyst with oxygen (10l2 Langmuir,350 "C) before adsorbing the benzene (in helium) at 50 OC.Whereas the temperaturesof the three peak maxima, ascribed to benzene desorption, remain the same (table 3),the distribution of benzene molecules on these sites under the same conditions ofdosage and adsorption temperature changes markedly and must therefore be afunction of the surface oxidation state of the catalyst. On all subsequent benzeneadsorptions, at 50 OC, after the first, the two lower energy sites were about equal inmagnitude (ca. lo1' molecules cm-2 each, or lo1' sites cm-2). This change in thedistribution in energies of the sites for benzene adsorption could relate to the catalystphase change which occurs during catalyst activation, since the desorption peaksobtained at temperatures above 25OOC (i.e.maleic anhydride) on the second andsubsequent injections at 50 OC exceeded those of the first benzene injection.The desorption spectra for the same benzene dosages but for the higher adsorptiontemperatures of 75, 90 and 100 OC, are shown in fig. 6, curves (a), ( b ) and (c),respectively. Several points are worthy of comment: (i) For the same dosage, as theadsorption temperature is increased so the benzene coverage decreases, dropping froma total coverage of 3 x loll molecules cm-2 at 50 OC, to 2.2 x 10" molecules ern+ at75 OC, to 1 .O x loll molecules cm-2 at 90 OC and to 3.8 x 1 O 1 O molecules cm-2 at 100 O C .(ii) Only two peak maxima are observed in the temperature regime ( < 250 "C) ascribedto benzene desorption. The highest energy site at around 200 OC has disappeared.(iii)The distribution of the benzene molecules adsorbed on these two remaining sites showsa shift to the higher energy site upon increasing the adsorption temperature. (iv) Theamount of maleic anhydride desorbed (for the same benzene dosage) increases as theadsorption temperature is increased, rising from 3.3 x loll molecules cm-2 at 50 OC,to 8.6 x lo1' molecules cm-2 at 75 OC, to 1 1 . 1 x 10" molecules cm-2 at 90 OC and to14.8 x 10I1 molecules cm-2 at 100 OC. (v) The total of all desorbing species (i.e. maleicanhydride plus benzene) increases upon increasing the adsorption temperature, risingfrom 6.3 x loll molecules cm-2 at 50 OC to 15.2 x 10" molecules cm-2 at 100 OC.The above information can be used to calculate the activation energy for anadsorbed benzene molecule to be transformed into an adsorbed maleic anhydridemolecule. Using the heats of adsorption obtained from the gas adsorption chromat-ography experiments (see later) and the benzene dosages used (2.7 x 1 O 1 O Langmuir)the fractional coverage by benzene of all of the adsorption states at all adsorptiontemperatures is found to be unity: the reaction is, under these conditions oftemperature and benzene partial pressure, zero order in the benzene gas-phaseconcentration.[Using the same heats of adsorption at the normal benzene oxidationtemperature (35OOC) the coverage of the low energy site (see gas adsorptionchromatography section) changes from 10 to 50% in moving from a 1 % benzene feedto a 5% feed, while that of the high energy site changes from 50 to 80%.It is thereforepredicted that the reaction will be first-order in benzene under these conditions, asfound by Ioffe and Lyubarskii, see previous paper.]The previous paper also shows that (i) the rate of oxidation of the adsorbed maleicanhydride is negligible at these adsorption temperatures (detectable rates of malei42 BENZENE PARTIAL OXIDATION KINETICSanhydride oxidation were found to occur only at temperatures in excess of 290 "C),(ii) the desorption half-lives of the adsorbed maleic anhydride states far exceed thehelium purge time at these adsorption temperatures and (iii) the maximum amountof maleic anhydride formed (1.5 x 10l2 molecules cm-2) at the 100 O C adsorptiontemperature is much less than the predicted saturation coverage of 2.6 x 1013 moleculescm-2.Therefore the amount of maleic anhydride formed on the surface for the samebenzene exposure times (i.e. dosages) will be a function solely of the activation energyfor the surface oxidation of the adsorbed benzene molecule. A plot of the logarithmof the maleic anhydride coverage against 1/T from the results listed in table 2 for thesame benzene dosage (2.7 x 1 O 1 O Langmuir) gives an activation energy for the overallsurface oxidation of adsorbed benzene to adsorbed maleic anhydride of3 1.4 6.3 kJ mol-l. (The drop in benzene coverages upon increasing the adsorptiontemperature and the increase in coverage of the higher energy peak relative to the lowerenergy peak stems from the benzene desorption half-lives on these states becomingcomparable to and then less than the helium purge time, as explained in the previouspaper.)Unfortunately, it is impossible from these experiments to determine directly whetherthe oxygen derives from the catalyst or is chemisorbed from the gas phase, since theoxygen impurity in the helium (15 ppm) results in an oxygen dosage of 6.8 x lo6Langmuir at the benzene dosage of 2.7 x 1O1O Langmuir.We can conclude, though,that in the oxidation of benzene the surface oxidation reactions have a surprisinglysmall overall activation energy. Furthermore, in the overall reaction pathway frombenzene to maleic anhydride the highest energy barriers are those for the desorptionof the product and these constitute the rate limiting step.As has been shown in resolving the separate peaks in the benzene desorptionspectrum, resolution of the reactively formed maleic anhydride spectrum is also afunction of coverage. Using the stopped flow technique we have extended the exposureof the catalyst to benzene (and oxygen).The results of these experiments in terms ofbenzene coverages, maleic anhydride coverages, temperatures of peak maxima anddesorption energies for a fixed benzene adsorption temperature of 50 O C as a functionof dosage are listed in table 3. A typical desorption spectrum is shown in fig. 7.It can be seen from the figure and from the table that the maleic anhydridedesorption spectrum is now clearly resolved at the coverages produced (> 10l2molecules cm-2) by the longer exposures.Two distinct peak maxima are observed inthe maleic anhydride spectrum, having peak maxima in the temperature ranges 243-250and 320-337 O C , values which are identical to those obtained by the temperatureprogrammed desorption of pre-adsorbed maleic anhydride.l The distribution of thereactively formed maleic anhydride between the two sites is more heavily weightedto the lower energy site than is observed by pre-adsorbing the maleic anhydride atdifferent adsorption temperatures. Indeed, even for the lowest maleic anhydrideadsorption temperature (1 30 "C) which shows the highest population on the lowerenergy site, the ratio of the amounts of maleic anhydride on the two sites wasapproximately 2:3 (lower to higher)' in sharp contrast to the results listed in table3 where in the reactively formed case the ratio varies from approximately 1 : 1 to 10: 1,depending on coverage.It can be concluded then that the maleic anhydride was formed on the surface ata temperature < 130 O C and probably at the benzene adsorption temperature (50 "C).[While it is tempting to suggest that the temperature of formation of the maleicanhydride by surface oxidation of the benzene might coincide with the temperatureof the maximum in the oxygen desorption spectrum (ca.100 "C), the fact that theamount of maleic anhydride formed is a function both of temperature and of the lengtJ. LUCAS, D. VANDERVELL AND K . C. WAUGH 43of time of dosage rather than the amount of adsorbate striking the surface (dosagesof as much as 3 x loll Langmuir would be expected to be more than sufficient tosaturate the surface) strongly infers that the long exposure is required to overcomethe time constant of the surface oxidation reaction and therefore that the maleicanhydride is being formed at the adsorption temperature.]Interestingly, while the benzene coverages in these * stopped flow ’ experiments aresimilar to those listed in table 2, suggesting that resolution of the benzene spectrashould have been possible, none of the benzene spectra obtained showed more thanone peak maximum.If these sites of higher desorption activation energy have higherheats of adsorption, as will be shown in the gas adsorption chromatographyexperiments, then even though the fractional coverage of all of the sites on which thebenzene adsorbs is unity, the residence times of the benzene molecules on these higherenergy sites will be longer, giving them a greater probability of surface oxidation. Theyare therefore more likely to be depleted by reaction to maleic anhydride. Thisprobability argument holds only if the adsorbed benzene is reacting with the adsorbedoxygen species (possibly a molecular anion 0;) whose surface population is also low,since reaction of the adsorbed benzene molecules with the abundant lattice surfaceoxygen for an overall surface oxidation activation energy of 3 1.4 kJ mol-l would notdiscriminate between the benzene adsorption sites of different heats of adsorption.Even if the benzene is adsorbed on a propitious configuration of lattice anions similarto those proposed in butene oxidation by Matsuura and S ~ h u i t , ~ e.g.such that theanions were at the para positions in the ring, at least two oxygen atoms must betransported to this chemisorbed species to form maleic anhydride and water. It is thistransport and addition activation energy which is low (3 1.4 kJ mol-l), suggesting thatthe oxidising species is the weakly bound molecular species seen in our temperatureprogrammed desorption experiments. Additionally, while the different heats ofadsorption could induce differently perturbed electronic states of the adsorbedbenzene molecules, predisposing the more strongly held to surface oxidation bylowering the surface activation energy, the argument concerning the involvement ofadsorbed molecular oxygen remains unaltered.The total loss of these sites to benzene re-adsorption at dosages which are morethan sufficient to re-populate them can be interpreted in a variety of plausible waysamong which the following two are the most reasonable. (i) Maleic anhydride ischemisorbed on them, inferring a n-bond to the surface, and during reaction oxygentransports on the surface to the chemisorbed benzene.(ii) The maleic anhydride isclosely associated with them, on a neighbouring site, say, effectively blocking themoff to benzene adsorption; this infers a flipover of the adsorbed benzene to achemisorbed oxygen with the resultant maleic anhydride being bonded to the catalystthrough an oxygen atom and a ‘dual’ site is then necessary for maleic anhydrideformation on the surface. Since more than one oxygen molecule is required totransform the adsorbed benzene to adsorbed maleic anhydride, more than one flip-overwould also be required with the constraint that the peripatetic maleic precursorremains sufficiently close to the original site to block off benzene adsorption.Themodel of the fixed hydrocarbon and the mobile oxygen molecules is therefore moreacceptable.GAS ADSORPTION CHROMATOGRAPHYIoffe’s inferencelo from steady-state rate measurements that in the oxidation ofbenzene to maleic anhydride the product is more strongly held on the surface thanthe reactant is borne out by the results of the gas adsorption chromatographyexperiments.Fig. 8 shows the temperature dependence of the elution times of pulsesof benzene which are swept isothermally over the catalyst. The retention times of th44 BENZENE PARTIAL OXIDATION KINETICSbenzene pulses are shorter than those of maleic anhydride1 in spite of having useda greater amount of adsorbent (8.3999 g for benzene, 7.504 g for maleic anhydride)and in spite of having carried out the experiments in a lower temperature range(80-140 "C for benzene, 230-290 O C for maleic anhydride). Qualitatively, then, one cansay that the benzene has lower heats of adsorption on the vanadium-molybdenumcatalyst than maleic anhydride. (The upper limit of 140 O C was chosen since ab-ove60 120 180 240 300 360 420 480retention time/sFIG.&-Temperature dependence of the benzene gas adsorption chromatographic peak shapes at (a) 80,(b) 90, (c) 100, (d) 110, (e) 120, (f) 130 and (s) 140OC.0 . 3 60 . 3 23 0 . 2 832Y .-31Z 0 . 2 4- 0 . 2 023 0 . 1 60 . 1 22200 D50 . 0 8 20 -040 -0c0.00 8-00 16*00 32-00 48.00 54.00retention distance/lO sI I80.00FIG. 9.-Comparison of the predicted and experimental trailing edges of the gas adsorption peak shapesfor a two site adsorption model. Crosses = experimental points. Solid line = prediction for the heats ofadsorption of 58.6 and 69.5 kJ mol-I the former bond occurring on 97.5% of the adsorbing surface whichitself maximally is only 2.7 x lo1* sites Adsorption temperatures: (a) 80, (b) 90, (c) 100, (d) 110,(e) 120, (f) 130 and (g) 140 OCJ.LUCAS, D. VANDERVELL A N D K. C. WAUGH 45that temperature < 100% of the injected benzene was eluted, aftcr which, ontemperature programming, maleic anhydride was desorbed showing two peak maximaat around 250 and 350 O C . The analysis is on the basis of a reversible adsorption onlyand is therefore not complicated by reaction possibly inducing other adsorption sites.)A more quantitative picture is obtained by analysis of the trailing edge of the gasadsorption peak shapes, the differential of the isotherm^,^ in terms of a given formof the isotherm. The skewed peak shapes and the curved isotherms derived from theirintegration (fig. 10) are indicative of a coverage dependent heat of adsorption or ofan energetically heterogeneous adsorption.The temperature programmed desorptionexperiments (table 2, fig. 5 and 6) have shown the benzene+atalyst interaction to beenergetically heterogeneous so this was the model chosen for simulation of the peaktrailing edges.0.08 0 . 2 4 0.40 0 . 5 6 0 * 7 2 I * 0 4gas phase concentration/10-9 mol cm-3FIG. 10.-Adsorption isotherms derived from fig. 9 for benzene on the vanadium-molybdenum catalyst.Crosses = experimental points. Solid line = prediction on the basis of the two-site Langmuir model.Adsorption temperatures: (a) 80, (6) 90, (c) 100, ( d ) 110, (e) 120 (f) 130 and (g) 140 OC.The fit to the trailing edges and to the isotherms is shown in fig. 9 and 10,respectively. This is achieved by minimising the sum of the squares of the differencesbetween experiment and prediction by changing the values of the six variables of atwo-site Langmuir model:ll -+ ( l - q l +K2c) K2c 1 number of moles adsorbed = owhere o is the number of moles of adsorption sites, Kl is the equilibrium constantfor site 1 = 10Alexp(-AH,/RT), K2 is the equilibrium constant for site2 = 10A2exp(-AH2/RT), a is the fraction of the surface on which the heat o46 BENZENE PARTIAL OXIDATION KINETICSadsorption is -AH, and C is the equilibrium concentration (mol ~ m - ~ ) of theadsorbate.The values of the 'best fit' parameters are listed in table 4 together with theparameter correlation matrix.The latter shows the parameters a, CT and the twoequilibrium constants K , and K , to be independent variables (with the expected highcorrelation between the pre-exponential terms and the heats of adsorption whichcomprise the equilibrium constant), giving reasonable confidence in the valuesobtained.All other indicators, e.g. the experimental correction to the standarddeviation and the value of the sum of the squares of the residuals, showed the fit tobe within experimental error.TABLE 4.-' BEST FIT ' PARAMETERS FOR A TWO-SITE LANGMUIR ADSORPTION MODELfraction of thepre-exponential adsorbing surface maximum predictedheat of adsorption terms subtending a heat of combined coverage/kJ rno1-l /cm3 mol-1 adsorption of -AH, /sites cm-z- AHl 58.6 f 1.3 A , 0.68k0.02 0.975 & 0.0009 0.273 f 0.005 x 1013-AHz 69.5 & 1.3 A , 0.68f0.2PARAMETER CORRELATION MATRIXcca 1 .o 10Al10Al 0.427 1.0 - AHl1 O*z 0.342 -0.228 0.081 1 .o -AH2-AHl -0.034 -0.814 1.0 lOAz- AH, 0.439 0.254 -0.056 -0.993 1 .o crcr 0.601 -0.245 -0.359 0.191 -0.269 1 .oAlthough the benzene desorption spectra (fig.5) showed three desorption maxima,since the fit to the peak trailing edges is within experimental error, inclusion of anothersite, a three-site model, would lead to an over definition of the line shapes. It istherefore impossible to obtain the expected three heats of adsorption from the gasadsorption peak shapes.Closer examination of the best fit values predicted by the two-site model revealsthat only 2.5% of the adsorbing surface (which is itself only 2.7 x 10l2 sites cm-2) isassociated with the higher heat of adsorption, i.e.that the predicted maximumcoverage by the sum of the higher energy sites is only 7 x 1O1O molecules cm-2.Inspection of table 2 shows this to be an underestimate, there being ca. 2-3 times thisamount of adsorbate on the higher energy sites at an adsorption temperature of 50 OCand comparable amounts to this value for adsorption temperatures of 75 and 90 O C .The two-site model is therefore deficient requiring an increase in the fraction of siteshaving a higher energy, a reduction in that energy and the inclusion of a yet higherenergy site of very small population.The higher energy site of the two-site Langmuir model therefore is a composite oftwo energies, the value for its heat of adsorption being high.This overestimate of theenergy, however, will not be too great since it is apparent from table 2 that the highestenergy site is smaller in number than the intermediate energy one by at least a factorof 3, further remembering that the heat of adsorption must be greater thaJ. LUCAS, D. VANDERVELL AND K. C. WAUGH 4758.6 kJ mol-l, the heat of the lowest energy site, and yet lower than the overestimateof 69.5 kJ mol-l.Combining the two heats of adsorption with the two lower activation energies todesorption we obtain the result that the activation energy to adsorption on the lowestenergy site is 41.8 kJ mol-l while that on the intermediate site has a minimum valueof between 38.5 and 49.8 kJ mol-l.When these results are compared with those obtained for maleic anhydride1 theyshow the benzene to have lower activation energies to adsorption, 41.8 andw 46.0 kJ mol-1 compared with 62.7 and 71.1 kJ mol-1 for maleic anhydride, but tohave lower heats of adsorption, 58.6 and 69.5 kJ mo1-1 compared with 79.5 and104.5 kJ mol-l.Therefore in steady-state reaction with a 1 % benzene feed at normalreaction temperatures (ca. 350 "C) and finite conversions, the majority of the adsorptionsites will be covered with maleic anhydride, the high temperature of the reaction beingrequired to maintain a sufficient rate of maleic anhydride desorption, and thereforeof reaction, by allowing benzene to adsorb on the vacated sites. [The differencebetween the predicted maximum number of sites for benzene adsorption (2.7 x 10l2sites cm-2) and for maleic anhydride adsorption (2.6 x 1013 sites cmP2) as determinedby gas adsorption chromatography is probably due to the maleic anhydride isothermshaving been obtained in a higher temperature range (230-290°C) than those ofbenzene (80-140 "C).At the higher temperature it is probable that a higher surfacedefect concentration, the likely site for hydrocarbon adsorption, would exist.]The picture that emerges then is of benzene being relatively easily but weaklyadsorbed onto the catalyst surface on which it is relatively short-lived, remaining there,in the absence of reaction, for between lo-' (higher energy site) and lop8 s (lowerenergy site) at steady-state reaction temperatures (350 "C).The surface coverageby these adsorption sites is small, being ca. 3 x loll sites cm-2 in the temperature range80-140 "C, increasing to 3 x 10l2 sites cm-2 at 230-290 "C and at steady-state reactiontemperatures of 350°C it would not be expected that the upper value would beexceeded greatly.The activation energy for the overall oxidation of adsorbed benzene to adsorbedmaleic anhydride is low (3 1.4 kJ mol-l) which, since the process requires the transportof at least three oxygen atoms to the adsorbed hydrocarbon, mitigates against latticeoxygen being the oxidant. While the surface coverages of the adsorbed benzene andadsorbed oxygen (molecular) will be low, < 10l2 molecules cm-2, at steady-statereaction temperatures, the low value of the activation energy for the surface oxidationprocess between them affords measurable reaction rates.The maleic anhydride produced, however, is strongly and heterogeneously held,having desorption energies of between 138 and 146.4 kJ mol-l and between 17 1.4 and175.6 kJ mol-l; it appears to be held on the site on which the benzene, from whichit was formed, was adsorbed; benzene adsorption on these sites is thereby inhibited.The earlier inference that it is oxygen transport to the adsorbed hydrocarbon ratherthan vice versa is therefore taken as proved. Nevertheless, these high activationenergies for the desorption of the reactively formed maleic anhydride, virtuallyidentical in value to those obtained for the desorption of pre-adsorbed maleicanhydride, show explicitly that the benzene oxidation is rate limited in the desorptionof the product, maleic anhydride.J. Lucas, D. Vandervell and K. C. Waugh, J. Chem. SOC., Faraday Trans. 1, 1981, 77, 15.J. E. Germain, Catalytic Conversion of Hydrocarbons (Academic Press, New York, 1969), chap. 5, p.259.R. J. Cvetonovic and Y. Amemomiya, in Catalysis Reviews, ed. H. Heinemann (Marcel Dekker, NewYork, 1972), vol. 6, p. 2148 BENZENE PARTIAL OXIDATION KINETICSI. E. Wachs and R. J. Madix, J. Catal., 1978, 53, 208.K. C. Waugh, J . Chromatog., 1978, 155, 83.B. Halpern and J. E. Germain, J . Catal., 1975, 37, 44.P. A. Redhead, Vacuum, 1962, 12, 203.* V. A. Shvets, M. E. Sarichev and V. B. Kasansky, J . Catal., 1968, 11, 378.I. Matsuura and G. C. A. Schuit, J . Cataf., 1971, 20, 19.lo 1. I. Ioffe and A. G. Lyubarskii, Kinet. Kataf., 1962, 3, 261.I. Langmuir, J. Am. Chem. Soc., 1918, 40, 1361.(PAPER 9/ 1768

ISSN:0300-9599    

DOI:10.1039/F19817700031

出版商:RSC    

年代:1981

数据来源: RSC

摘要:

J . Chem. SOC., Faraday Trans. I , 1981, 77, 49-61Measurement of Surface Tension Using a Vertical ConeBY ZANJETA UGARCIC, DARSHAN K. VOHRA, EZATT ATTEYA ANDSTANLEY HARTLAND*Swiss Federal Institute of Technology, Department of Industrial and EngineeringChemistry, Universitatsstrasse 6 , CH-8092 Zurich, SwitzerlandReceived 30th November, 1979The surface tension may be determined from the maximum force,fma,g exerted on a vertical cone asit is pulled from a liquid interface using= gAPj UmaJF'rnaxPin which F,,, is a unique function of the cone angle when the liquid wets the surface of the cone. Valuesof F,,, are calculated from a force balance on the cone, taking into account the known shape of the liquidinterface. The surface tension of several liquids has been obtained using quartz cones having half-anglesof 30' and 45' and good agreement obtained with the accepted values.Capillary phenomena associated with the introduction of a solid body into aninterface between two immiscible fluids have been the subject of numerous studies.The calculation of surface tension from the maximum force exerted on a body as itis withdrawn from an interface forms the basis of the well-known Wilhelmyand Du Nouy ring4?j techniques.In recent years many theoretical and a fewexperimental treatments have been reported for a variety of axisymmetrical solids, e.g.the sphere6-13 and vertical cylinder.14-19 A recent paper by Padday20 presents resultsfor stainless-steel cones. Another paper21 discusses the equilibrium and stability ofcones but is mainly concerned with their vertical displacement relative to theundisturbed interface.The cone has the advantage that its shape is completely definedby the apex angle, which is dimensionless. Other shapes, such as the sphere andcylinder, are defined by their radius, which must be made dimensionless by a suitablecombination of physical properties, including the surface tension. This complicatesthe application of the theoretical analysis as the dimensionless radius varies with theliquid system, in addition to the measured force.In the present paper a rigorous theory is developed for the measurement of thesurface tension of an interface using a vertical cone. Using quartz glass cones it isconfirmed that this theoretical model can be combined with measurements of themaximum force to obtain accurate values of the surface tension, without the need forempirical correction factors.THEORETICALConsider the cone of density ps at a fluid/liquid interface of tension 0 as shown infig.1. A volume vh is immersed in the heavy fluid of density P h and a volume v1 inthe lighter fluid of density pl. The radius of the contact circle where the interface meetsthe surface of the cone is x and its height above the undisturbed liquid surface is z.At the contact circle the inclination of the interface to the horizontal is &The upward forces acting on the cone are the forcefg on the supporting wire andthe buoyancy forces (vh P h g + v1 p1 g ) . The downward forces are the weight of the cone450 MEASUREMENT OF SURFACE TENSION~~~FIG.1 .-Notation for vertical cone suspended at a fluid/liquid interface.(vh + vl)psg plus the vertical component of the surface tension force acting at thecontact circle, 2nxosin@,. In addition there is a force nx2z(ph-pl)g acting down-wards at the contact circle which arises because the buoyancy terms as written abovedo not correctly represent the integral of the hydrostatic pressure in the light andheavy liquids over the free surfaces of the cone.Equating the upward and downward forces when the cone is at equilibrium gives:f g + vh Ph g+ v1 P1 g = (vh + vl)psg + 2nxa sin $c + nx2z(ph -pl)g- (1)The value of q5c is given by q5c = 90+a-j?in which a is the cone half-angle and j? the contact angle at the solid surface measuredthrough the liquid.If the balance measuring the forcefg is tared in the upper phase (vh + v,)(p, -pl) = 0and eqn (1) reduces tofg = 2nxa sin #c + (nx2 z - vh)(ph -pl)g (3)(4)X = x d and Z = z d (5)and a dimensionless force I; = fgd/a (6)I; = 2nX sin @c + nPZ - n P / 3 tan a.(7)If a cone which is first submerged is slowly withdrawn from the interface the forceacting on it gradually increases until a maximum forcef,,, is reached. As the coneis further withdrawn the force first decreases and then the cone breaks away from theinterface. At the position of maximum force, eqn (7) may be re-written as:in which vh is the volume of that part of the cone immersed in the heavy liquid givenby:vh = nx3/3 tana.Defining dimensionless lengths :where c = @h-pl)g/o, eqn (3) becomesFmax = 2nXc sin @c + nGZc - n%/3 tan a (8)where Fmax is the value of F corresponding tofmax, Xc is the dimensionless contactradius and Zc the dimensionless meniscus height when the maximum force is reached.The shape of an external meniscus has been determined by Huh and Scriven22 anZ .UGARCIC, D . K . VOHRA, E . ATTEYA A N D S . HARTLAND 51TABLE CORRESPONDING VALUES OF a, #, X,, Z , AND Fmax, FOR VERTICAL CONES WITHp = ooa ZC01234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914001.98470 x 10-12.932 30 x 10-13.72890 x 10-l4.45302 x 10-15.13386~ 10-15.84273 x 10-l6.488 84 x 10-l7.12081 x 10-l7.74230 x 10-l8.35593 x 10-19.09325 x 10-l9.71348 x 10-l1.033 131.094 771.156 391.238 931.302 531.366381.430 561.495 141.589061.656 531.724 661.793 521.863 191.970 362.043 842.118412.194172.27 1 222.393612.475 572.559 192.644 582.731 832.872002.965 673.061 753.160413.261 843.423 263.533 103.646443.763 533.884 644.072 674.205 204.342 864.486044.635 2004.971 98 x 10-16.22004 x 10-17.04646 x 10-l7.671 24 x 10-l8.17387 x 10-l8.93622 x 10-l9.31886 x 10-19.65207 x 10-l9.946 14 x 10-l1.020 851.085 701.108 361.129011.147 941.165 381.225471.241 181.255811.269 471.282 281.338 701.350491.361 611.372 141.382 121.435001.444311.453 171.461 641.469 751.519001.526 621.533941.540 991.547 771.593251.599681.605 901.611921.617751.659 341.664 901.670 301.675 551.680 671.718 271,723 171.727 951.732621.737 1908.39341 x 10-l1.253 231.61 1501.946 632.270 632.622 092.949 183.277 763.609 273.944744.378 074.739925.108 965.485 685.870 546.447 306.872 647.308 947.756 758.216658.994 549.510771.00428 x 101.059 15 x 101.11579~ 101.22052 x 101.28492 x 101.351 64 x 101.42083 x 101.49265 x 101.633 13 x 101.71578 x 101.80193 x 101.891 85 x 101.98581 x 102.17405 x 102.28345 x 102.39831 x 102.51906 x 102.64,621 x 102.89958 x 103.04944 x 103.20805 x 103.37622 x 103.55483 x 103.89991 x 104.1 1322 x 104.341 13 x 104.585 11 x 104.84685 x 152 MEASUREMENT OF SURFACE TENSIONTABLE 1 .--continued515253545556575859606162636465666768697071727374757677787980818283848586878889901411421431 441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791804.858 235.023 445.196 385.377715.568 165.840046.054256.280 576.520216.774517.1 19667.41 1427.123428.058 038.417918.883 189.307 659.769 221.02731 x 101.08257 x 101.151 25 x 101.21921 x 101.29498 x 101.38001 x 101.476 18 x 101.59356 x 101.72049 x 101.86833 x 102.04276 x 102.251 79 x 102.51420 x 102.83368 x 103.24406 x 103.79059 x 104.55531 x 105.707 14 x 107.61844 x 101.14396 x lo22.30067 x lo2001.770731.775 131.779431.783 661.787 8 11.8 17 231.821 241.825 181.829051.832 871.858 151.861 851.865 501.869 101.872 651.893 781.897241.900 681.904051.907401.924 391.927 661.930901.934121.937 3 11.950 201.953 321.956431.959511.962 581.971 421.974441.977 451.980441.983411.988 281.99 1 231.994 161.997 102.000 00~5.32711 x 105.64454 x 105.98744 x 106.35873 x 106.76181 x 107.45403 x 107.95235 x 108.49789 x 109.09700 x 109.75706 x 101.08102 x lo21.16475 x lo21.25801 x lo21.36234 x lo21.47955 x lo21.653 55 x lo21.80783 x lo21.983 80 x lo22.18601 x lo22.41959 x lo22.74639 x lo23.072 17 x lo23.45785 x lo23.91899 x lo24.476 52 x lo25.23477 x lo26.09531 x lo27.182 16 x lo28.581 73 x lo21.04259 x lo31.30372 x lo31.65679 x lo32.17291 x lo32.96989 x lo34.29460 x lo36.758 19 x lo31.206 58 x lo42.72646 x lo41.09523 x lo5coextensively tabulated by Hartland and H a r t l e ~ .~ ~ The dimensionless meniscus height2, is a unique function of the dimensionless contact radius Xc and meniscusinclination q5,.For a given cone half-angle a and contact-angle p the value of q5, isdetermined by eqn (2) and the value of Z , is a function of X, only. It follows thatthe value of Fmax in eqn (8) is a function of only a and p. For given values of a and/3 it is possible to find the value of X, which gives Fmax by integrating the equationsgoverning the shape of an annular meniscus to obtain related values of Xc, 2, and& Maximization of the right-hand side of eqn (7) was carried out simultaneouslywith the numerical integration using a modified Coggin maximization ~ubroutine.~Z . UGARCIC, D. K . VOHRA, E . ATTEYA AND S . HARTLAND 53The error criteria for terminating maximization procedures were set such that theirreduction by a factor of 10 did not change the calculated value of Fmax by more thanone in the sixth significant figure.Eqn (6) may be re-written in terms of the maximum force as:from which 0 may be obtained from experimental measurements of fmax, knowingthe density difference Ap = ph -pl.The value of F,,, is only a function of the half-coneangle a and contact angle a. For the normal case in which the liquid wets the surfaceof the cone (so that = 0), Fmax is a unique function of a. Methods based on othershapes such as cylinders and spheres involve the determination of the dimensionlessradius which itself involves the physical properties and hence changes with thefluid/liquid system.The volume of revolution of the external meniscus, um, between the horizontal plane0 and z is given byso eqn (1) may be re-written asvm(ph - P I ) g = 2nx sin +c + ?Ix2z(ph - pJ gfg = (vh+vl?@s-Pl)g+ Vm(Ph-pl)g-vh(Ph-pl)g*(10)(1 1)The first term on the right-hand side of this equation represents the net weight of thecone in the light phase and the difference between the second and third terms is thevolume of revolution of the external meniscus, less the volume of the submerged partof the cone.It is immaterial whether or not the apex of the cone penetrates the levelof the undisturbed interface. When the balance is tared in the light phase the first termdisappears and eqn (1 1) becomes:(12)A similar equation to eqn (9) has been obtained by Padday20 in which fmax is replacedfg = (vm - vh>@h - P I ) g.by AP(vm - vh)max.EXPERIMENTALLIQUIDSThe four organic liquids benzene, toluene, ethanol and n-hexane of Fluka analytical gradewere used without further purification. The water was triple-distilled, had a maximumconductivity of 0.8 x lob6 i2-l cm-l and was stored in a thoroughly steam-cleaned Pyrexcontainer for < 24 h before use.The literature values of surface tension for these liquids weretaken from data given in ref. (25) and (26). The densities of benzene, toluene and n-hexane werecalculated for temperatures other than 20 OC using constants taken from ref. (26); the densitiesof water and ethanol were taken directly from ref. (27) for the measured temperature.APPARATUSThe apparatus shown in fig.2 used for measuring the maximum force of the fluid interfaceon the cone was similar to that used by Furlong and Ha~t1and.l~ The force exerted on the conewas measured with a Mettler HE 20 precision balance which displayed the force in digital formto +O.OOOl g whilst it was being simultaneously recorded on chart paper for ready detectionof the approaching maximum value. The balance was also connected to a Hewlet Packard97 S calculator through an interface. The calculator was programmed to calculate the surfacetension from the measured value of the maximum force. The quartz cones were carefully cutand polished with diamond tools in the workshops of Wisag in Zurich. Each cone was hungfrom a central hook attached to a plate which could be adjusted by three vertical screws toensure the top surface of the cone was horizontal.The cone angles were measured with 54 MEASUREMENT OF SURFACE TENSIONTABLE 2 .-COMPARISON OF SURFACE TENSION MEASUREMENTS WITH LITERATURE VALUESg = 9.80663 m s - ~ in Zurichtemperature 0 (measured 0 (literature deviationsystem 1°C value)/mN m-’ value)/mN m-l /mN m-lwaterethanoltoluenebenzenen-hexanewaterethanoltoluenebenzenen-hexane20.320.320.321.520.420.220.220.221 .o20.820.620.620.520.520.420.419.819.819.819.820.920.920.920.920.220.220.920.321.421.521.221 .o20.720.521.221.320.520.520.520.4cone angle a = 29O 54.5’72.5672.6272.4872.3222.2522.2822.3022.2328.4328.4128.4428.4628.8328.7828.8128.7818.3018.4818.4818.46cone angle a = 44O 57.5’72.4772.3872.3072.2822.2922.3022.1822.928.2728.2428.2828.2728.7828.7528.7228.7218.3218.3418.3518.3672.7072.7072.7072.5122.2322.2522.2522.2528.4128.4328.4628.4628.8028.8028.8228.8218.4218.4218.4218.4272.6072.6072.6072.6022.2522.2522.1822.2428.3828.3728.4028.4228.7828.8128.7228.7118.3518.3518.3518.36-0.14- 0.08- 0.22-0.19 + 0.02 + 0.03 + 0.05- 0.02+O.Ol- 0.03- 0.02 +o.oo+ 0.03- 0.02-0.01- 0.04-0.12 + 0.06 + 0.06 + 0.04-0.13- 0.22- 0.30- 0.32 + 0.04 + 0.050.00 + 0.05-0.11-0.13-0.12-0.150.00- 0.060.00+0.01- 0.03-0.010.000.0Z .UGARCIC, D . K . VOHRA, E . ATTEYA AND S . HARTLAND 551 I r iFIG. 2.-Experimental arrangement for measuring surface tension by withdrawing a vertical cone from theinterface: A, cone; B, resistance thermometer; C, hydraulic lift.precision better than 0.5 min using a Nikon profile projector. The liquid was placed in a glasscontainer of 13 cm diameter standing on a platform which could be raised or lowered by ahydraulic lift. The temperature was measured continuously with a Fluka 2180A digitalthermometer accurate to & 0.01 OC. To minimize air currents and evaporation losses the liquidvessel was enclosed in a Perspex box. The vibrations were damped by standing the entireapparatus on a system of polystyrene foam, 100 kg steel plates and air filled rubber tubes.PROCEDUREThe quartz cones and Pyrex glass cell were soaked in concentrated nitric acid for 1 h, treatedin an ultrasonic bath, copiously rinsed with triple-distilled water and dried in air at 60 O C .Theliquid was placed in the container and its position adjusted so that the cone tip rested on theinterface at the centre of the vessel when suspended from the balance. When measuring thesurface tension of benzene, toluene and n-hexane the system was left overnight to establishequilibrium. However, to avoid contamination the experiments were performed immediatelyafter adding water to the cell. This was also the procedure with ethyl alcohol which absorbswater vapour from the surrounding air.19 The balance was tared with the cone in air.The conewas dipped into the liquid and the platform slowly lowered at a speed of CQ. 1 cm h-l. Themaximum value of the force and surface tension were measured several times for each liquid56 MEASUREMENT OF S U R F A C E T E N S I O NRESULTS A N D DISCUSSIONThe variations in surface tension force (2xXsin q+), bouyancy force(71x2- x P / 3 tan a) and the total force F defined by eqn (7) for a = 45O and p = Ooare plotted in fig. 3. The tension force (b) decreases as the contact radius decreases(i.e. as the cone is pulled out of the liquid). The buoyancy force (c) first increases andthen decreases. The total force (a) shows a maximum at X z 3.9.400 1 2 3 4 5 6contact radius XFIG.3.-Variation with contact radius X of force components acting on a vertical cone with half-anglea = 45' suspended at a fluid/liquid interface which wets the cone, so /3 = 0: (a) total, (b) surface tension,(c) buoyancy.For cone half-angles varying from 1 to 90°, tables have been prepared giving thevalues of $,, Xc, Zc and Fmax for different angles, p, between 0 and 90'. The valuesfor b = 0 are given in table 1 at 1' steps in a. Fig. 4 shows that Fmax increases witha, but for a given cone angle decreases with increase in contact angle and approachesinfinity when the cone angle approaches 90°, i.e. when a cone reduces to a plate ofinfinite size.Padday20 gives values of Fmax, Xc and 2, for cone half-angles between 0 and 75'in 5' steps for the case of zero contact-angle.His values of Fmax usually agree to 6figures with our values, even though the corresponding values of X, and 2, sometimesdiffer in the fifth or sixth figure. [Because of the form of eqn (8) the value of Fmaxis not sensitive to small errors in X , and Z,.] However, there are serious discrepanciesin the third and fourth figures in some values of Fmax. These can usually be tracedto incorrect substitution of Padday's values of X , and 2, into eqn (8). For example,when a is loo our value of Fmax is 3.94474 and Padday's is 3.99434, whereas the valuecalculated from his values of Xc and Zc using his equation is 3.944735, which agreesexactly with our value. The corresponding values for a = 50" are 48.4685 (our table),48.48652 (Padday's table) and 48.46846 (calculated from Padday's values of Xc and2,).When a = 40' there is a small difference in the sixth figure, the values being26.4621, 26.46244 and 26.462 13, respectively. However, when a = 1' there is anunexplained discrepancy in the third figure, our value of Fmax being 0.839341 andPadday's 0.837904.Fig. 5 gives the value of the dimensionless contact radius of the cone at the positioZ . UGARCIC, D . K . VOHRA, E . ATTEYA AND S . HARTLAND 575.04.03.02.0x z 4 1.0Do40.0-1 .o-2.0-3.0 I0 15 30 45 60 75 90cone half-angle a/"FIG. 4.-Variation of maximum force with cone angle at different contact angles 8: (a) 0'; (b) 30'; (c) 60';( d ) 90'.of maximum applied force as a function of cone angle a and contact angle p.Thesecurves indicate the necessary cone dimensions for a given liquid.In fig. 6 the error in the value of the surface tension given by assuming p = 0 isshown as a function of cone angle a for different values of p. The error alwaysdecreases with increase in cone angle and for low values of B the error is small, beingca. 1 % when jl= 10'.In fig. 7 the error in values of a for errors of +0.1', k0.5' and f 1 .Oo in the coneangle is shown as a function of the true cone angle. The error passes through a broadminimum for these situations when the cone angle lies between 25' and 50'. On thisbasis the cone angles selected for the present investigation were 30' and 45'. Thesurface tensions of the liquids in table 2 were measured as described above using quartzcones with half angles of 29' 54.5' and 44' 57.5'.For these cone angles a maximizationprogram was run separately to obtain Fmax. The calculations are based on theassumption that the liquids used perfectly wet the quartz surface. The density of thelight fluid is taken to be that of air. Repeating the calculations, assuming the air tobe saturated with the vapour of the liquid (which slightly increases the density of thelight fluid), only usually affected the surface tension in the fourth figure. The resultsare summarized in table 2, together with the comparable literature values.In the case of water the measured values of surface tension were ca. 0.2 mN m-llower than the literature values for both the cones.Similar low values of the surfacetension of water were also reported by Padday14 using a vertical cylinder, since water,having a high surface energy, quickly adsorps atmospheric contaminants. Themeasured values for hexane with the 30' cone and for toluene with the 45' cone differfrom the literature values by ca. f 0.1 mN m-l. For the other liquids the differenceCAR I 58 MEASUREMENT OF SURFACE TENSION80706050 -CI0.ir2 4033Ccl.d0 15 30 45 60 75 90cone half-angle a/"FIG. 5.-variation of contact radius with cone angle at different contact angles p : (u) Oo; (b) 30°; (c) 60;( d ) 90°.is always < 0.1 mN m-l for both cones, confirming the validity of the theoreticalmodel and the precision of the experimental technique.Differentiating eqn (9) enables the fractional error in o to be estimated from thefractional errors in g , Ap,fmax and Fmax.We will assume that the error in g is verysmall and that the error in Ap arises from a temperature deviation of kO.1 'C. Theerror in Ap then follows from the known temperature coefficients of the liquids witha maximum value of T 0.0 13 % for n-hexane. Assuming the error in the measured valueOffma, is fO.OOO1 g gives maximum errors for hexane of k0.214 and +0.091% forthe 30' and 45' cones, respectively. (More typical values for both benzene and toluenebeing f 0.126 and 0.053 % for the 30' and 45' cones, respectively.) The main errorin Fmax arises from the error in the measurement of the cone half angle a which isassumed to be 1'.The values of dFma,/da obtained from table 1 are 1.0615 and2.6185 deg-l for the 30' and 45* cones, respectively, so the error in Fmax is 0.12% forboth cones. Multiplying the fractional errors in Ap,fmax and Fmax by the factors $,and $, respectively, before adding the absolute values shows that the error in thepredicted values of 0 lies between 0.10% and 0.23% for the 30' cone and between0.07% and 0.15 % for the 45' cone. The absolute error in 0 is ca. 0.05 and 0.03 mN m-lfor the 30' and 45' cones, respectively, which correspond, in order of magnitude, tothe deviations within the measured values listed in table 2Z . UGARCIC, D . K . VOHRA, E . ATTEYA AND S . HARTLAND 590X2cone angle a/"FIG. 6.-Error in the value of the surface tension calculated assuming = 0 as a function of a for actualvalues of contact angle p : (a) = 10'; (b) = 20'; (c) = 30°; ( d ) = 40'; (e) = 50'; cf> = 60'; (g) = 70';(h) = 80'; (i) = 90".CONCLUSIONS(1) The maximum dimensionless force Fmax exerted on a vertical cone at an infiniteinterface is calculated from a force balance on a control surface around the cone andthe known shape of the interface.(2) Values of Fmax are tabulated as a function ofthe cone half-angle a and contact angle fl which the liquid makes with the surfaceof the cone, together with the dimensionless contact radius Xc. (3) The surface tensionmay then be obtained by measuring the maximum forcef,,,g exerted on a verticalcone at a liquid interface using:0 = gA&max/Fmax)'knowing the values of a and j? and the density difference between the liquid and thefluid above it.(4) The surface tension of several liquids was measured in this way usingquartz cones wetted by the liquids (so p = 0) with half-angles a of 30' and 4 5 O , andgood agreement obtained with the accepted literature values.3-60 MEASUREMENT OF SURFACE TENSIONcone angle a/"FIG. 7.-Error in the value of the surface tension for small errors in the measurement of cone angle asa function of true cone angle. % error = (Aa)/a(a); Aa = a(a)-a(a_+Aa). (a) Aa = 1.0, Aa-ve; (b)Aa = - 1.0, da+ve; (c) Aa = -0.5, Aa+ve; ( d ) Aa = +0.5, Aa-ve; (e) Aa = kO.1, Ao;;:.APPENDIXNOTATIONconstant characterizing physical propertiesbalance readingforce applied to conedimensionless forceacceleration due to gravityvolumedimensionless volumehorizontal radius measured from vertical axis of conedimensionless xvertical dimension measured upwards from undisturbed liquid leveldimensionless Z .UGARCIC, D . K . VOHRA, E . ATTEYA AND S . HARTLAND 61aB@AP = Ph -P1P0h1SCmaxGREEK SYMBOLShalf-angle of cone apexcontact angle at surface of cone measured through liquidangular inclination of interface to horizontaldensityinterfacial tensiondensity differenceSUBSCRIPTSheavy (lower) phaselight (upper) phasesolid conecontact circle of liquid on conemaximum valueL. Wilhelmy, Ann. Phys. (Leipzig), 1863, 119, 177.D. N. Furlong and S. Hartland, J. Colloid Interface Sci., 1979, 71, 301.J.Burri and S. Hartland, Colloid Polym. Sci., 1977, 255, 576,L. Du Nouy, J. Gen. Physiol., 1919, 1, 521.W. D. Harkins and H. F. Jordan, J. Am. Chem. Soc., 1930,52, 1751.A. D. Scheludko and A. D. Nikolov, Colloid Polym. Sci., 1975, 253, 396.C. Huh and S. G. Mason, Can. J. Chem., 1976, 54, 969.S. Hartland and J. Burri, Chem. Eng. J., 1976, 11, 7.A. V. Rapacchietta and A. W. Neumann, J. Colloid Interface Sci., 1977, 59, 555.lo E. A. Boucher and H. J. Kent, Proc. R. Soc. London, Ser. A, 1977, 356, 61.l1 E. A. Boucher and H. J. Kent, J. Chem. Soc., Faraday Trans. I , 1978, 74, 846.l2 E. Bayramli, C. Huh and S. G. Mason, Can. J. Chem., 1978, 56, 818.l3 J. Burri and S. Hartland, Tenside Detergents, 1976, 13, 18.l4 J. F. Padday, A. R. Pitt and R. M. Pashley, J. Chem. Soc., Faraday Trans. I , 1975,71, 1919.l5 A. L. Clarke, J. Phys. E., 1976, 9, 592.l6 E. A. Boucher and M. J. B. Evans, J. Phys. E, 1977, 10, 306.l7 E. A. Boucher and H. J. Kent, Proc. R. Soc. London, Ser. A , 1977, 356, 61.la E. A. Boucher, Proc. R. SOC. London, Ser. A, 1978, 358, 519.l9 D. N. Furlong and S. Hartland, J. Chem. Soc., Faraday Trans. I , 1980, 76, 457, 467.2o J. F. Padday, J. Chem. Soc., Faraday Trans. I , 1979,75, 2827.21 E. A. Boucher and H. J. Kent, J. Chem. Soc., Faraday Trans. I , 1977, 73, 1882.22 C. Huh and L. E. Scriven, J. Colloid Interface Sci., 1969, 30, 32323 S. Hartland and R. W. Hartley, Axisymmetric FluidlLiquid Interfaces (Elsevier, Amsterdam, 1976),24 G. F. Coggin, Univariate Search Methods, Central Instrument Laboratory Research Note 64/1125 J. Timmermans, Physiochemical Constants of Pure Organic Compounds (Elsevier, Amsterdam, 1965).26 Handbook of Chemistry and Physics, ed. R. C. Weast (Chemical Rubber Co., Cleveland, Ohio,27 International Critical Tables (McGraw-Hill, New York, 1928).p. 256.(I.C.I., London, 1964).1975-1 976).(PAPER 9/1905

ISSN:0300-9599    

DOI:10.1039/F19817700049

出版商:RSC    

年代:1981

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. 1, 1981, 77, 63-12Mixed Dimers of Chloroacetic Acids and DecanoicAcid in BenzeneBY YUKIO FUJII AND YOHOKI KAWACHIFaculty OP Engineering, Gifu University, Kagamihara, Gifu, 504 JapanAND MOTOHARU TANAKA*Laboratory of Analytical Chemistry, Faculty of Science, Nagoya University,Nagoya, 464 JapanReceived 21st January, 1980The formation of mixed dimers of acetic, monochloroacetic, dichloroacetic and trichloroacetic acids withdecanoic acid has been studied by partition and solubility measurements of water in benzene. Thedimerization constants of these acids decrease with increasing acidity of the acids, while the reverse is foundfor the hydration constant of the acid monomer monohydrate. The formation constants of the mixed dirnershave been determined for the following acid pairs : acetic-decanoic, monochloroacetic-decanoic,dichloroacetic-decanoic and trichloroacetic-decanoic. The formation constant is just as expectedstatistically for all these acid pairs.Linear free energy relationships have been observed between theformation of dimeric acids and acid monomer monohydrates and the dissociation of acids in water.There have been numerous investigations of the dimerization of acetic acid and itschloro derivatives.l-* The partition method has been widely used for the determinationof dimerization constants of organic acids in solvents which are slightly miscible with~ a t e r . ~ - l ~However, since the organic phase in a partition system is equilibrated with theaqueous phase, the dimerization constants are always different from those inanhydrous s01vents.l~ From our previous studies it may be concluded that the waterpresent in the organic phase forms hydrates with the acid monomers and dimers.Most previous work has been devoted to the dimerization of the same carboxylicacid, while some authors have demonstrated the formation of a mixed dimer betweentwo different kinds of carboxylic acids by i.r.absorption,14 partiti~n,~ solubility15 andvapour pressurels measurements. Affsprung et aZ.14 showed that the mixed dimer ofacetic-trichloroacetic acids is more stable than expected statistically. On the otherhand, Sekine et aZe9 claimed that the mixed dimer of two different carboxylic acidsgives the statistically expected formation constant.In this paper we report the hydration and dimerization of acetic acid and itschlorinated derivatives and discuss the formation and hydration of mixed dimers ofthese acids with decanoic acid.EXPERIMENTALREAGENTSAcetic, dichloroacetic and decanoic acids of C.P. grade were distilled twice under reducedpressure. Monochloroacetic and trichloroacetic acids were purified by fractional freezingfollowed by recrystallization from anhydrous benzene. All these carboxylic acids were titratedwith standard sodium hydroxide. The equivalent weight agreed with the theoretical molecular664 A C I D DIMER FORMATION I N BENZENEweight to within 0.2%. The stock solutions of these acids were prepared by dilution withanhydrous benzene.Benzene of G.R.grade was washed successively with dilute sodium hydroxide solution, waterand concentrated sulphuric acid and finally five times with distilled water. Anhydrous benzenewas prepared from the benzene saturated with water. It may be dried by refluxing overanhydrous calcium chloride, following by distillation over sodium.After precipitation of heavy metal impurities as hydroxides, sodium perchlorate of C.P. gradewas purified by recrystallization from water.PARTITIONAll experiments were carried out in a room thermostatted at 25.0 & 0.5 "C. 15 cm3 each ofthe aqueous and organic phases were equilibrated in a 50 cm3 centrifuge tube immersed in abath thermostatted at 25.0 & 0.2 OC. The ionic strength of the aqueous phase was adjusted to1 .O mol dm-3 with (Na, H) ClO, or H(A, ClO,), where A is carboxylate ion.The concentrationsof carboxylic acid in the two phases were determined by the potentiometric titration with0.1 mol dm-3 carbonate-free sodium hydroxide solution. The variation of liquid junctionpotentials originating from (Na/H) and (ClO,/A) exchanges were taken into account.HYDRATIONThe solubility of water in benzene was determined as follows. A 50 cm3 centrifuge tube wasused as equilibration bottle, into which was inserted a 10 cm3 small tube. A solution of knownpartial pressure (known activity) of water was placed in the small tube and pure or acid-containingbenzene was placed in the outer centrifuge tube. The equilibrater was sealed with vinyl filmand immersed in a water bath thermostatted at 25.0k0.2 OC for at least 2 days.The acidconcentration in the organic solvent was determined as in the partition experiment. The waterconcentration in benzene was measured coulometrically with a Hiranuma model AQ- 1aquacounter.RESULTS A N D DISCUSSIONEQUILIBRIUM TREATMENTLet the carboxylic acid be denoted by HA and its partition constant anddimerization constant by K,,,, and K2,HA, respectively. The subscripts o and wrefer to the benzene and water phases, respectively. The following two equilibria arerelevant to the partition of the acid HA:HA, HA0 KD, HA = [HAlo/[HAl, (1)2HA0 * o K2, HA = [(HA)210/[HA]t. (2)With hydrated species of the acid being denoted by HA - (H,O), and (HA), - (H,O),,the hydration equilibria are written as:The total equilibrium concentrations of the acid monomer and dimer in the organicphase equilibrated with water, [HA'], and [(HA);],, respectively, are given as:The distribution ratio D of the acid HA is related to the conditional dimerizatioY .FUJII, Y . KAWACHI AND M . TANAKA 65constant K 2 , HA = [(HA)&,/[HA']E and the conditional partition constant(7)[HA],, the concentration of unionized carboxylic acid in the aqueous phase, is given= CHA - IH'l (8)KL, H A = [HA'],/[HA], as follows:= Kb, H A ( 1 + 2Kb, HA KL, HAbywhere CHA denotes the total concentration of acid. CHA is determined by potentio-metric titration with standard sodium hydroxide. [H+] is determined independentlyby pH measurements. Under the present experimental conditions, i.e.[HA], < 1 mol dm-3, dimerization of the acids is negligible in the aqueous phase."In fig.1 the value of log D is plotted against log [HA],. Values of &, HA and K2.,HA,04, 2 -1.0-2.0II - 1-0 0log [HA],FIG. 1.-Distribution of acetic acid and chloroacetic acids between benzene and 0 mol dm-3 (Na, H)C10,or H(A, CIO,) solution as a function of the acid concentration in the aqueous phase. AA, acetic acid; MCA;monochloroacetic acid ; DCA, dichloroacetic acid ; TCA, trichloroacetic acid.TABLE 1 .-CONDITIONAL PARTITION AND DIMERIZATION CONSTANTS OF SOME CARBOXYLIC ACIDSacids logKD,HA log&, HA log&, mix logK,,- 1, decanoic acid18 2.70 2.42 4.922, acetic acid - 2.04 2.10 2.54 4.763, monochloroacetic acid - 1.77 1.29 2.36 2.874, dichloroacetic acid - 1.15 0.74 2.23 1.305, trichloroacetic acid - 0.30 0.23 1.98 0.666 ACID DIMER FORMATION IN BENZENEevaluated by fitting the plot with normalized curve y = log( 1 + p ) against x = log p ,are collected in table 1.The formation of mixed dimer was studied in benzene saturated with water.Twocarboxylic acids, HA and HB, form a mixed dimer HA-HB in the organic phase:HA, + HB, HA HB, K,, mix. (9)HA; + HB; HA - HBb K2, mix. (10)In wet benzene we can write the following equilibrium:When the concentration of HA is low compared with that of HB, i.e.CHB: o/C.HA, > 20, we may assume no dimeric species (HA),, in benzene. Then thedistnbution ratio of HA is expressed asD = ([HA’],+[HA.HB’],)/[HA],= KD, H A ( l + K 2 , mix[HB’lo) (1 1)-where K2,mix is the conditional mixed dimerization constant.The plot of logD-logKD, HA against log[HB’], is given in fig. 2. These plots were fitted with the normal-ized curve described above. The curves in fig. 2 are theoretical curves calculatedwith constants given in table 1.- 2.5 - 2.0 -1.5log IHB‘loFIG. 2.-Determination of the conditional mixed dimerization constant. Labelling as in fig. 1 .Hydration numbers and hydration constants were determined as follows. Measure-ments of water solubility revealed that water is monomeric in such solvents asbenzene and carbon tetrach10ride.l~ In order to control the monomeric waterconcentration, we have used the isopiestic equilibration method,20 in which water isdistributed by vapour contact between a phase of known water activity and thebenzene solution.The total concentration of the acid and water are given as follows:CHA, 0 = [HA10 + WWH20),10 + 2[(HA)210 + 2E[(HA),(H2O)nI0CH*O, 0 = [H2010 + ~~[HA(H20),10 + W(HA)2(H,O)nIo.(12)( 1 3 Y .FUJII, Y . KAWACHI A N D M . TANAKA 670.6o 0.4B0”0.2TCA JI ~~~0 0 2 0.4 0.6HA, oFIG. 3.-Solubility of water in benzene as a function of the total concentration of the acid at constant watermonomer concentration, [H,O], = 0.0345. Labelling as in fig. 1.1.00.5%DD -0-0.51 .o0.5xM - 00.51 I- 2.0 -1.5log [H,OI,FIG. 4.-Determination of the hydration number of the monomeric and dimeric trichloroacetic acid inbenzene68 ACID DIMER FORMATION IN BENZENEThe relation of the solubility of water to the acid concentration in benzene is givenin fig.3. Substitution of eqn (1)-(6) and (12) into (13) gives(CH20, 0 - [ ~ 2 ~ 1 0 ) iHA'1,' = zm81m, HA[H201?(1 +zplm, HAIHzol?)-l+K 2 , HAznb2n, HA [HzO1,"(l +zpim, HA[H201?)-2[HA'lo (14)where m and n are hydration numbers and plm, HA and 82,, HA are hydration constantsfor monomer HA and dimer ( H A ) , , respectively. The plots of the left-hand side ofeqn (14) against [HA'], should be linear, with an interceptFrom the plot of logarithmic values of X and Y against the logarithmic concentrationof water, we obtain values of the hydration numbers m and n, the hydration constantspl,, HA and /?2n, HA and the dimerization constant K2, HA.The plots for trichloroaceticacid are shown in fig. 4. Constants thus determined for several acids are summarizedin table 2.TABLE 2.-DIMERIZATION AND HYDRATION CONSTANTS OF SOME CARBOXYLIC ACIDS1, decanoic acid 2.55 - 0.78 -2, acetic acid 2.41 2.86 1.13 -3, monochloroacetic acid 1.63 2.53 1.33 1 .oo4, dichloroacetic acid 1 .oo 2.04 1.51 2.365, trichloroacetic acid 0.48 1.79 1.78 3.261, decanoic acid2, acetic acid3, monochloroacetic acid4, dichloroacetic acid5, trichloroacetic acid- - 0.03 1.241.600.89 - 2.57 -1.19 3.37- 3.81 6.85 2.47- - -- -The hydration constant of the mixed dimer was also determined from the relationbetween water solubility and the acid concentration in the organic solvent. Theformation of HA - HB - (H,O), is written as follows :HA - HB, + xH20, + HA * HB - (H20)z,0 (15)where x denotes the hydration number of the mixed dimer.The equilibrium constantfor reaction (1 5) is defined asD z , mix = [HA * HB (H2O),Io/"HA HBlo[H2OlE* (16)The total concentration of water in benzene is given by the followingY . FUJII, Y . KAWACHI AND M. TANAKA 69where m' and n' are the hydration numbers of monomeric and dimeric decanoic acids,HB and (HA),, respectively. With values of the hydration and dimerization constantsof HA and HB, the second to the fifth terms on the right-hand side of eqn (17) arereadily calculated. The hydration number x was then evaluated by plottinglogXx[HA-HB(H,O),], against log[H,O],. The best fit was found for x = 1.Theequilibrium constant is given in table 2.From consideration of the hydration equilibria, the real mixed dimerization andthe apparent dimerization constants calculated from the partition measurement arerelated to each other as follows:K2,mix = Ki, mix(l +xpim, HA[H201p) (1+811, HB[H2010) (1+82,rnix[H2010)-'*(18)HYDRATION OF THE ACID MONOMERS A N D DIMERSThe following equilibria are relevant to the hydration of carboxylic acids inbenzene :HA, + H,O, + HA - H,O,HA, + 2H20, $ HA - (H20),,(HA),, 0 + H2Oo (HA), * H2Oo(HA),, 0 +2H,OO * (HA), * (H,O),,O(HA),, 0 +4H200 + (HA), * (H20)4, 0-Bell and Arnold21 have made a cryoscopic study of trichloroacetic acid in wet benzeneand found that HA.H,O and (HA),.(H,O), are important species. The hydratedspecies obtained with acetic acid are in accord with the results reported by Christianet aL2, except for the dimer dihydrate which we found.On the basis of our results,the fundamental types of hydrated species in benzene are the monomer monohydrateand the dimer dihydrate. The dimer dihydrate and the dimer tetrahydrate seem to beformed as a result of the association of the monomer monohydrate and the monomerdi hydra te, respectively .DIMERIZATION AND HYDRATIONRelatively few quantitative distribution studies have been made for chlorinatedacetic acids between benzene and water since these dissociate strongly into ions inaqueous solution, while the distribution of acetic acid has been widely studied bynumerous investigaters. The apparent dimerization constant at 25 *C from distributionstudies of acetic acid between water and benzene is generally in the range 130-1 50.6v lo, 23Christian et ~ 1 .~ ~ have attempted to correct for the hydration effect in theirdistribution data and found that the K, values are a factor of 2-3 greater than theKt values. Zaugg et ~ 1 . ~ ~ have investigated the dimerization constant of acetic acidby a calorimetric method in anhydrous benzene. They report a value of K, = 270. Weobtained a corrected value of K2 = 260 30 for the uncorrected value K2 = 1 10 20.The corrected value K, is in agreement with that obtained by Zaugg et aZ.Steigman et aL3 have studied the dimerization of chlorinated acetic acids in benezeneby an i.r. method. They found the K2 values of 48.427.1 and 6.7 for monochloroacetic,dichloroacetic and trichloroacetic acids, respectively.Our values are lower than thesevalues. As pointed out by Satchel1 et a1.,26 serious discrepancies were found amongthe K, values for these chlorinated acids when determined by an i.r. method.'. 2 $ 26The K, value of decanoic acid, K, = 350f30, was checked by vapour-pressureosmometry in rigorously dried benzene, CHBO, < 5 x low4. The value, K, = 390 & 5070 ACID DIMER FORMATION IN BENZENEfrom vapour-pressure osmometry agrees with that from partition-water solubilitymeasurements.Taft et aLZ7 have established linear free-energy relationships in the formation ofhydrogen-bonded complexes of various OH reference acids with a variety of protonacceptors.The hydration constants of pyridine bases and water in benzene have beencorrelated with the protonation constant of the bases in aqueous solution.28The formation of a 1 : 1 complex between an acid HA and a base B is written asfollows :KHA.B HA,+B, S HA-B,and the formation of HA in water is expressed asThe following two Taft equations [eqn (21) and (22)] may be written for the formationof hydrogen-bonded complexes and the formation of the acid in water, respectively,where KHA. B(KHA) and KGA. B(KO,A) refer to the formation of the hydrogen-bondedcomplexes (HA) of substituted and non-substituted carboxylic acids, respectively. p;"and p: are constants reflecting the susceptibility of a given reaction series to polarsubstituents for the association in benzene and the acid formation in water, respectively,and o* is a polar substituent constant relative to the standard CH, group.Asillustrated in fig. 5, the hydration constant and the dimerization constant are relatedlinearly with the polar substituent constant.Combining eqn (21) and (22) we obtain the relationP* log KHA. = -+ logKHA +constant.PZ2GbD310\ TTA /i DCA-1 0 1 2 3U*FIG. 5.-Correlations of hydration and dimerization constants with polar substituent constantsY . F U J I I , Y . KAWACHI A N D M . TANAKA 71From our experimental results we have found the following relation in the chloro-substituted acetic acid systems:log/?,, = -0.2210gK,A+ 1.89. (24)The correlation is obviously good: the stronger the acid, the more extensive thehydration.The increasing order of dimerization constants of carboxylic acid in benzene is :trichloroacetic < dichloroadetic < monochloroacetic < acetic < decanoic acid.Thevalue of K , for the dimerization of a fatty acid, RCOOH, is little affected by differentalkyl groups,ll while chlorine substitution results in a considerable decrease in K,. Thisconfirms the previous results that there is a nearly linear logarithmic relationshipbetween the dimerization and the acid dissociation in water. From our experimentalresults, we have found the following relationlogK, = O.4710gKHA+0.31. (25)MIXED DIMERIZATIONTwo acid dimers, (HA), and (HB),, reacting with each other, yield a mixed dimerHA-HB: K(HA),, , + (HB),, , S 2HA * HB,.K = G, mix(&, HA &, HB)-~*(26)(27)Suppose that in an equilibrium mixture of (HA),, (HB), and HA * HB, the distributionof HA and HB over the dimer sites is random, then the statistical equilibrium constantwould be K = 4.29 Hence, the statistical mixed dimerization constant K:,mix is(28)expressed asThe ratios of the constants statistically calculated to those experimentally deter-mined K2,mix/K~,mix are 1.3, 1.4, 0.9 for the acid pairs: acetic-decanoic,monochloroacetic-decanoic, dichloroacetic-decanoic and trichloroacetic-decanoic,respectively. Thus we may conclude that the formation constant of these mixed dimersis as expected statistically.For a given acid HB (decanoic acid in our case), we havefrom eqn (27) the following linear free energy relationship:The equilibrium constant K is given byK:, mix = 2 (K2, HA K2, HB)O*~*logK,, mix = 0.51 logK,, ,,+constant.(29)logK,, = 0.2410gKH,+constant. (30)Substitution of eqn (25) into (29) leads to the following:Comparison of eqn (25) and (30) reveals that logK,,,, is twice as sensitive aslog K2,rnjx to the acidity of HA. Eqn (24), (25) and (28)-(30) should be useful inpredicting the hydration and dimerization constants of carboxylic acids in benzene.H. A. Pohl, M. E. Hobbs and P. M. Gross, J . Chem. Phys., 1941,9, 408.Y. Nagai and 0. Shimamura, Bull. Chem. SOC. Jpn, 1962, 35, 132.J. Steigman and W. Cronkright, Spectrochim. Acta, Part A , 1970, 26, 1805.N. S. Zaugg, L. E. Trejo and E. M. Woolley, Thermochim.Acta, 1973, 6, 293.N. Kolossowsky and F. Kulikov, 2. Phys. Chem., Abt. A , 1934, 169, 459.M. Davies, P. Jones, D. Patnaik and E. A. Moelwyn-Hughes, J . Chem. SOC., 1951, 124972 A C I D DIMER FORMATION IN BENZENE7891011121314151617181920212223242526272820C. P. Brown and A. R. Mathieson, J. Phys. Chem., 1954,58, 1057.M. Davies and H. E. Hallam, J. Chem. Educ., 1956, 33, 322.T. Sekine, M. Isayama, S. Yamaguchi and H. Moriya, Bull. Chem. Soc. Jpn, 1967,40, 27.I. Kojima, M. Yoshida and M. Tanaka, J. Znorg. Nucl. Chem., 1970, 32, 987.Y. Fujii and M. Tanaka, J. Chem. Soc., Faraday Trans. 1, 1977, 73, 788.Y. Fujii, K. Sobue and M. Tanaka, J. Chem. SOC., Faraday Trans. I , 1978, 74, 1467.L. KuEa and E. Hogfeldt, Acta Chem. Scand., 1967,21, 1017.H. E. Msprung, S. D. Christian and A. M. Melnick, Spectrochim. Acta, 1964, 20, 285.B. W. Szyskowsky, 2. Phys. Chem., 1927, 131, 175.S. D. Christian, J. Phys. Chem., 1957, 61, 1441.M. Davies and D. M. L. Griffiths, 2. Phys. Chem. (N.F.), 1954, 2, 353.M. Tanaka and T. Niinomi, J. Znorg. Nucl. Chem., 1965, 27, 431.S. D. Christian, H. E. Affsprung and J. R. Johnson, J. Chem. Soc., 1963, 1896.S. D. Christian, H. E. Affsprung, J. R. Johnson and J. D. Worley, J. Chem. Educ., 1963, 40, 419.R. P. Bell and M. H. M. Arnold, J. Chem. Soc., 1935, 1432.S. D. Christian, A. A. Taha and B. W. Gash, Quart. Rev. Chem. Soc., 1970, 24, 20.B. A. Moelwyn-Hughes, J. Chem. Soc., 1940, 850.S. D. Christian, H. E. Affsprung and S. A. Taylor, J. Phys. Chem., 1963, 67, 187.N. S. Zaugg, S. P. Steed and E. M. Woolley, Thermochim. Acta, 1972, 3, 349.D. P. N. Satchel1 and J. L. Wardell, Trans. Faraday Soc., 1965, 61, 1199.R. W. Taft, J. Am. Chem. SOC., 1953, 75, 4231; R. W. Taft, D. Gurka, L. Joris, P. von R. Schlayerand J. W. Rakshys, J. Am. Chem. Soc., 1959, 91, 4801.K. Hirose and M. Tanaka, Bull. Chem. SOC. Jpn, 1976, 49, 623.R. Powler and E. A. Guggenheim, Statistical Thermodynamics (Cambridge University Press, 1965),p. 167.(PAPER 0/114

ISSN:0300-9599    

DOI:10.1039/F19817700063

出版商:RSC    

年代:1981

数据来源: RSC

摘要:

J . Chem. SOC., Furuday Trans. I, 1981, 77, 73-79Effects of Oxygen, Nitrogen Dioxide andTrifluoroborane on Photoconductivity of Perylene andPhthalocyanine Single CrystalsBY ROBERT L. VAN EWYK, ALAN V. CHADWICK AND JOHN D. WRIGHT*University Chemical Laboratory, University of Kent, Canterbury, Kent CT2 7NHReceived 28th January, 1980Exposure of single crystals of zinc and lead phthalocyanine to low pressures (1-10 Pa) of NO, and BF,enhances the photoconduction by factors of cu. 100 and 10, respectively, and the photoconduction actionspectra change to resemble the absorption spectrum, rather than its inverse as in vucuo. In air, similarchanges in action spectra, but smaller changes in magnitude, occur. Additional slow increases for leadphthalocyanine in air are consistent with oxygen diffusing into the crystals.Photocurrent is proportionalto (light intensity)n, with n = 0.54-0.62 in vucuo, 0.34-0.40 in air and 0.20-0.27 in NO,. Charge carriergeneration mechanisms involving exciton dissociation at surface sites occupied by adsorbed gas moleculesare discussed. Singlet photoconduction in perylene crystals is inhibited by low pressures of NO, and BF,,and spectroscopic evidence is presented for reaction of NO, with surface layers of perylene, formingnitroperylene. A new photoconduction response at higher NO, pressures (los Pa), centred at 16200 cm-l,is consistent with carrier generation following excitation of a surface perylene-NO, charge-transfer complex.This is reversible on evacuation and of similar magnitude to the vacuum photoconduction.The different behaviours of perylene and phthalocyanines are discussed in terms of the greater mobilityof electronic excitation energy in phthalocyanines.Photoconduction, although more sensitive to low gasconcentrations than semiconduction, has less potential for gas detection due to the poor reversibility ofthe effects.In a previous paper1 we have reported the effects of a range of gases, particularlyNO,, BF, and 0,, on the semiconductivity of several molecular crystals. The presentstudy is concerned with the effect of these gases on photoconductivity in phthalocyanineand perylene single crystals. Two main objectives of the study were to compare thesensitivity of semiconduction and photoconduction parameters to low gas concen-trations and to examine the potential of the technique of studying changes in the actionspectrum of photoconduction in various ambient gases as a probe for revealing thenature of the adsorbed species.Phthalocyanine and perylene are particularly usefulas a complementary pair of materials for this study, since optical excitation energyis mobile in phthalocyanine as excitons, whereas the dimeric structure of perylenefacilitates charge resonance which severely restricts the transfer of electronic excitationenergy in the crystal., This means that in phthalocyanines even light absorbedrelatively far from the surface may yield excitons capable of diffusing to the surface,whereas in perylene any influence of surface-adsorbed species on the photoconductionaction spectrum must be due to optical excitation of species at or very near the surfaceof the crystal.Although there have been many reports of effects of oxygen on thephotoconductivity of organic solids, e.g. ref. (3)-( 12), the only report of investigationsof NO, effects is a brief note by Compton and Waddington13 on the effects of thisgas on anthracene photoconduction.774 PHOTOCONDUCTIVITY I N PERYLENE AND PHTHALOCYANINE CRYSTALSEXPERIMENTALZinc and lead phthalocyanines and perylene were purified by repeated entrainer sublimationin a flow of oxygen-free nitrogen and single crystals were obtained from the final sublimation.Steady state d.c. 2-probe photoconductivity measurements were made in various ambient gasenvironments using apparatus described previous1y.l Silver-paint electrodes were attached toopposite ends of needle crystals and well-developed exposed crystal faces between the electrodeswere illuminated.In all cases, effects due to illumination of the electrode<rystal interfaces werenegligible, since masking these regions from the incident light produced no significant differencefrom results obtained without the masking.RESULTS AND DISCUSSIONPHTHALOCYANINESFig. 1 and 2 show the photoconduction action spectra for lead and zinc phthalo-cyanines in a vacuum of Pa after heating to 520 K, and in air, NO, and BF, atvarious pressures, together with the diffuse reflectance spectra of the materials in air.The action spectra in vucuo are approximately the inverse of the absorption spectra.12 16 20 24energy/ 1 O3 cm-'FIG.1.-Photoconduction action spectra at room temperature (a) in vacuum Pa), (b) air (lo5 Pa),(c) NO, (7 Pa) and (d) BF, (1.3 Pa), and (e) diffuse reflectance spectrum for lead phthalocyanine.Photocurrent scale units are A m-* for vacuum, lop5 A m-, for air and lop4 A m-* for NO, and BF,.This is consistent with the conclusion of Popovic and Sharp,14 based on pulsedphotoconductivity studies of metal-free phthalocyanine films, that carrier generationis a bulk phenomenon. In air, NO, and BF, the photoconduction is generally enhancedover the entire spectral range studied (12000-25000 cm-l) and shows a closerresemblance to the absorption spectrum. This suggests that adsorbed electronacceptor gases enhance exciton dissociation to charge carriers at the surface.Theenhancement is not restricted to light absorbed close to the surface, as excitons areable to diffuse to the crystal surface from the bulkR. L. VAN EWYK, A. V. CHADWICK AND J . D . WRIGHT 75In air, the enhancement is much smaller than in NO, and BF, and is readily reversedby heating and evacuation. For lead phthalocyanine, this increase occurs slowly over200 h and is consistent with exciton dissociation at oxygen diffused into the bulk ofthe crystal. Such diffusion also affects semiconductionlq l5 and long-wavelengthphotoconductionll in lead phthalocyanine.16 18 24energy/103 cm-'FIG. 2.-Photoconduction action spectra at room temperature (a) vacuum (lo-' Pa), (b) air ( lo5 Pa) and(c) NO, (13 Pa), and (d) diffuse reflectance spectrum for zinc phthalocyanine.Photocurrent scale units areA rn-, for vacuum, A rnP2 for air and A m-* (right-hand scale) for NO,.NO, and BF, produce much larger and more rapid increases in photoconduction(fig. 1 and 2). These increases are less readily reversed and for low gas pressures areca. two orders of magnitude larger than the effects on semiconduction.l For example,the dark resistivity of zinc phthalocyanine increases to its original vacuum valuefollowing evacuation to Pa at 520 K for 12 h after exposure to NO,, whereas thephotocurrent remains enhanced by a factor of 100 after this treatment. However,exposure to ammonia (1 Pa) removes this residual effect within 10 s and this reversalis not affected by subsequent evacuation of the ammonia.Similarly, the effects oftrifluoroborane are only reversed on treatment with ammonia, but in this case theeffects return on evacuation of the ammonia.Increased photoconductivity is accompanied by a decrease in the power law (n) ofthe dependence of photocurrent on light intensity. For photon fluxes between 1013and 1015 cm-, s-l, the values of n are 0.54-0.62 in vacuo, 0.34-0.40 in air and 0.20-0.27in NO,. Heilmeier and Harrisonl6 analysed the intensity dependence of photocurrentin phthalocyanines and showed that, in addition to n = a for the case of low carriergeneration efficiency with bimolecular carrier recombination near the surface, a valueof n = + is possible for high carrier generation efficiency if diffusion of carriers fromsurface to bulk is important.Our results in air agree with the latter case, as alsoreported16 for metal-free phthalocyanine in air. The results in vacuo are closer to th76 PHOTOCONDUCTIVITY IN PERYLENE AND PHTHALOCYANINE CRYSTALSformer case, as expected for lower carrier generation efficiency. In NO, the values ofn are < &. The derivation of the Q power dependence assumes constant carrier mobility.In practice, at high carrier densities such as in the presence of NO,, mobility is likelyto decline as carrier density increases, yielding values of n < Q.The low-power laws of dependence of photocurrent on light intensity imply thatthe enhancement of photocurrent by absorbed gases is due to increased carriergeneration rate rather than reduced carrier recombination.The observation ofenhanced photoconduction in conditions where the effects of gases on semiconductionare completely reversedl implies the existence of strongly bound ionised states at themost active sites on the crystal surface. Energy transfer to these states from excitonswould induce dissociation to charge carriers, although thermal dissociation would beinsignificant. Further quantitative treatment of the results in terms of this mechanismis impossible in the absence of any currently available technique for characterisingstrongly bound ionised states present at low density on molecular crystal surfaces.However, the mechanism is also consistent with the irreversibility of the BF, effectsand the fact that the enhancement is less for BF, than for NO,.Charge transfer toBF, involves a localised o orbital on boron and leads to a strong localised electrostaticattraction between the negatively charged adsorbed gas molecule and a positive holeat the active site on the crystal surface. Hence, dissociation following exciton energytransfer is less probable and the effects are less reversible. Alternative mechanisms notinvolving strongly bound ionised states cannot account for these observations.PERY LENEFig. 3 shows the photoconduction action spectra for a perylene crystal at Pabefore exposure to NO, and under pressures of 5 and lo3 Pa NO,. Oxygen did notaffect the photoconductivity of perylene. The response in vacuo is similar to thatreported by Mulder, and corresponds closely to the singlet absorption spectrum ofperylene.Even low NO, pressures ( 5 Pa) inhibit this response, and the inhibition isnot reversible by heating in vacuo or treatment with ammonia, although theseconditions do reverse the effect of NO, on semiconductivity of perylene. This suggestsa reaction between adsorbed NO, and surface perylene molecules, yielding a surfacelayer of molecules which inhibit singlet photoconduction while not affecting semi-conduction. Spectroscopic evidence for such reaction was sought, by grinding a fewmg of perylene with a few g of KBr, measuring the diffuse reflectance spectrum ofthe sample (essentially KBr particles covered with a thin surface layer of perylene),exposing the sample to NO, and measuring the spectrum of the product. Thetreatment with NO, led to a rapid change in the colour of the sample, from yellowto dull reddish brown, and the spectral changes are shown in fig.4. The absorptionmaximum at ca. 21 500 cm-l agrees well with the value of 21 460 cm-l reported for3-nitr0pery1ene.l~ Ristagno and Shinela report that 3-nitroperylene can be convenientlyprepared by reaction of the NO; ion with the perylene cation. Adsorption of NO,on to the perylene crystal surface, followed by charge transfer, could readily producesuch ions (this is the mechanism which we have proposed for the effect of NO, onsemiconduction of donor crystals1). The irreversible inhibition of singlet photocon-duction may therefore be the result of less favourable energy transfer to, ordissociation of excited states at, the surface layer of nitroperylene.Under higher NO,pressures ( lo3 Pa) a new region of photoconduction response with reproducible finestructure is observed at lower energy (maximum at 16200 cm-l) of comparablemagnitude to the original singlet photoconduction, and reversible on evacuation. Thisnew response is in the region where a charge-transfer absorption band of an NO,complex with perylene or 3-nitroperylene should occur. Fig. 5 is a plot of first charge-transfer band energy against electron affinity for several electron acceptors complexeR. L. VAN EWYK, A. V. CHADWICK AND J . D. WRIGHT 7714 18 22energy/l03 cm-'FIG. 3.-Photoconduction action spectra for perylene single crystals at room temperature (a) vacuum(lop4 Pa) and NO, [(b) 5 Pa and (c) lo3 Pa].energy/ 1 O3 cm-IFIG.4.-Diffuse reflectance spectra of perylene adsorbed on potassium bromide, before (a) and after (b)treatment with NO,. (The Kubelka-Munk function isK (I-R,)' -S 2R, 'where K and S are absorption and scattering coefficients, respectively, and R , is the absolute reflectivityof the sample.78 PHOTOCONDUCTIVITY I N PERYLENE AND PHTHALOCYANINE CRYSTALSwith perylene. For a complex of perylene with NO, (electron affinity 2.2 eV)199 2o thistransition is predicted to occur at 16 100 cm-l. For a 3-nitroperylene-N02 complex,the transition would be at slightly higher energy due to the electron withdrawingsubstituent to the donor molecule. Furthermore, the interval of 750 cm-l in the finestructure corresponds to the frequency of the bending mode of Such vibronicfine structure has been observed in the crystal spectra of several molecular complexes.22The new response is thus consistent with charge generation involving separation ofthe ion pair formed in the first charge-transfer excited state of this complex.The initialstep of this separation may involve migration of the positive charge from a surfacenitroperylene molecule to a neighbouring perylene molecule, a process expected to bethermodynamically favourable as nitroperylene will have a rather higher ionisationpotential than perylene. The only other possible origin of a new low-energy absorptionin the presence of excess NO, is a singlet-triplet transition enhanced by paramagneticNO,, but this occurs at 12500 cm-l for perylene,, and not at 16200 cm-l.2.82.6% ---CI h .* 5 2.4E: E Y0 *0 2.22.0I I 1 I10 12 14 16 18energy of first charge-transfer transition/ 1 O3 cm-'FIG.5.-Relationship between electron affinity and first charge-transfer transition energy for perylenecomplexed with tetracyanoethylene (l), tetracyanoquinodimethane (2), chloranil (3), fluoranil (4), pyro-mellitic dianhydride (5) and 1,3,5-trinitrobenzene (6). The dotted line indicates the expected transitionenergy for the complex of perylene with NO, (electron affinity 2.2 eV).Exposure of a fresh crystal of perylene to trifluoroborane (5 Pa) also inhibited thesinglet photoconduction irreversibly.Higher trifluoroborane pressures produced nonew photoconduction response below 25 000 cm-l, as expected, since extrapolationof fig. 5 predicts that the first charge-transfer band of the complex of perylene withBF, (electron affinity 0.65 eV)24 should occur near 29000 cm-l. As with NO,, the effectof BF, on photoconduction was irreversible. At present we know of no satisfactoryexplanation of this irreversibility. The reaction of perylene in dichloromethane withBF,, forming the perylene +BF; ion pair, is completely reversible on removal of BF,.25The enhancement of semiconduction of perylene by trifluoroborane, arising from thesame ion pairs, is also reversible by treatment with ammonia.' The magnitude of thisenhancement ( lo8) corresponds to complete surface coverage by ion pairs, eacR.L. VAN EWYK, A. V. CHADWICK A N D J . D . WRIGHT 79capable of producing charge carriers. The concentration of any residual very stronglybound ionised states following treatment with NH, must therefore be small, andinsufficient to account for the inhibition in terms of quenching of singlet excited states26by the radical ions.CONCLUSIONSThe photoconduction of phthalocyanines and perylene is more sensitive than thesemiconduction to low concentrations of nitrogen dioxide and trifluoroborane.Photoconduction in phthalocyanines is enhanced, while that in perylene is inhibitedunder these conditions. These effects are much more difficult to reverse than effectsof the same gases on semiconductivity and this limits the use of photoconductivityin gas detecting devices.Higher concentrations of nitrogen dioxide induce a newresponse in the photoconduction action spectrum of perylene, whose energy and finestructure are consistent with a charge-transfer transition of a perylene-nitrogen-dioxidecomplex. The differences between effects observed for perylene and phthalocyaninearise from the fact that electronic excitation is mobile in phthalocyanine crystals butmuch less so in perylene crystals. Probable mechanisms for the effects observed havebeen discussed but the absence of suitable experimental techniques for establishingthe nature of adsorbed species on the surfaces of molecular crystals remains a seriousobstacle to detailed interpretations.We thank the Health and Safety Executive and the S.R.C.for the award of a CASEstudentship (to R. L. van E.).R. L. van Ewyk, A. V. Chadwick and J. D. Wright, J. Chem. SOC., Faraday Trans. I , 1980,76,2194.B. J. Mulder, Rec. Trav. Chim., 1965, 84, 713.A. T. Vartanyan, Dokl. Akad. Nauk SSSR, 1950,71, 641.A. C. Chynoweth, J. Chem. Phys., 1954, 22, 1029.A. Bree, D. J. Carswell and L. E. Lyons, J. Chem. SOC., 1955, 1729, 1735.A. Bree and L. E. Lyons, J. Chem. SOC., 1960, $179.J. A. Bornmann, J. Chem. Phys., 1958, 27, 604.P. Day and R. J. P. Williams, J. Chem. Phys., 1962, 37, 567.P. Day, G. Scregg and R. J. P. Williams, Nature, 1963, 197, 589.l o P. Day and R. J. P. Williams, J. Chem. Phys., 1965, 42, 4049.H. Yasunaga, K. Kasai and K. Takeya, J. Phys. SOC. Jpn, 1979, 46, 839.l2 P. Day and M. G. Price, J. Chem. SOC. A, 1969, 236.l 3 D. M. J. Compton and T. C. Waddington, J. Chem. Phys., 1956, 25, 1075.l4 Z. D. Popovic and J. H. Sharp, J. Chem. Phys., 1977, 66, 5076.l5 H. Yasunaga, K. Kojima, H. Yohda and K. Takeya, J. Phys. SOC. Jpn, 1974, 37, 1024.l6 G. H. Heilmeier and S. E. Harrison, J. Appl. Phys., 1963, 34, 2732.l7 H. Hopff and H. R. Schweizer, Helv. Chim. Acta, 1959, 42, 2315.C. V. Ristagno and H. J. Shine, J. Am. Chem. SOC., 1971, 93, 181 1l9 C. Lifshitz, B. M. Hughes and T. D. Tiernan, Chem. Phys. Lett., 1970, 7, 469.2o J. Berkowitz, W. A. Chupka and D. Gutman, J. Chem. Phys., 1971, 55, 2733.22 H. Kuroda, T. Kunii, S. Hiroma and H. Akamatu, J. Mol. Spectrosc., 1967, 22, 60.23 P. S. Egel and B. M. Monroe, Adv. Photochem., 1970, 8, 302.24 H. A. Skinner, Proc. Con$ Univ. Coll. North Staflordshire, 1952, p. 28.25 W. I. Aalsberg, G. J. Hoijtink, E. L. Mackor and W. P. Weijland, J. Chem. SOC., 1959, 3055.26 G. J. Hoytink, Acc. Chem. Res., 1969, 2, 114.E. T. Arakawa and A. H. Nielsen, J. Mol. Spectrosc., 1958, 2, 413.(PAPER 0/159

ISSN:0300-9599    

DOI:10.1039/F19817700073

出版商:RSC    

年代:1981

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I, 1981, 77, 81-87Enhancing Action of SO, on the CarboniogenicIsomerization of Butene over La-X ZeoliteBY KIYOSHI OTSUKA,* YOSHIHISA TAKIZAWA AND AKIRA MORIKAWADepartment of Chemical Engineering, Tokyo Institute of Technology, Ookayama,Meguro-ku, Tokyo 152, JapanReceived 30th January, 1980The S0,-enhanced isomerization of cis-but-Zene over La-X zeolite has been characterised. The effectof pretreatment temperature of the zeolite on its activity for the double-bond shift of the olefin has suggestedthat the SO, adsorbed on the dehydroxylated sites of the zeolite enhance the carboniogenic activity of thezeolite. Infrared spectroscopic studies have shown that no new acidic hydroxyl groups are formed by theadsorption of SO,. The enhanced production of pyridinium ions on the zeolite after the addition of SO,suggests that the enhanced activity of the zeolite can be ascribed to an increase in activity of the acidichydroxyl groups caused by the electron-withdrawing effect of SO,.The enhancing action of SO, on the carboniogenic activities of X- or Y-type zeoliteshas been reported previo~sly.l-~ In the case of alkaline-earth-exchanged X zeolites,SO, increases the number of acidic hydroxyl groups which produce the ca.3640 cm-li.r. band, thus producing the favourable action of SO,. The Bronsted acid sites areformed through the reaction of SO, with water molecules coordinated to thealkaline-earth cations or with basic hydroxyl groups bonded to the cations.3* * On theother hand, in the case of hydrogen-Y zeolite, a different mechanism for SO, actionhas been proposed.The enhanced activity of the zeolite has been ascribed to anincrease in activity of the acidic hydroxyl groups caused by the inductive effect of SO,adsorbed on dehydroxylated sites.'For trivalent-cation-exchanged zeolites, such as rare-earth zeolites, both the casesdescribed above may occur because the zeolites contain a large number of the Bronstedsites together with water and basic hydroxyl groups attached to the trivalent cations.The water molecules and basic hydroxyl groups may react with SO, by similar reactionschemes to those suggested in the case of alkaline-earth-exchanged X zeolites,increasing the number of Bronsted sites. Hence, in this work we intended to clarifywhich type of SO,-action, the type proposed for the divalent-cation-exchanged zeolitesor that for hydrogen zeolite, is operative in the case of rare-earth-exchanged X zeolites.Kinetic studies of the SO,-enhanced isomerization of cis-but-2-ene over La-X zeolitehave been carried out.The changes in the i.r. bands caused by the acidic hydroxylgroups and the pyridinium ions after adsorption of SO, have been examined to tryto determine the nature of the enhancing action of SO,.EXPERIMENTALMATERIALSThe sample La-X zeolite was prepared by conventional ion exchange of an Na-X zeolite(Linde, 13X). The ion exchange was carried out at 343 K by repeated exchanges (6 times in7 days) with a 0.1 mol dm-3 LaCl, solution. The zeolite was then washed by ion-exchanged8a2 ISOMERIZATION OF BUTENE OVER La-X ZEOLITEwater until no chloride ion was detected.The degree of La exchange of the parent Na-X wasdetermined by the flame-photometric analysis of aluminium and residual sodium. The degreeof exchange was 99%. The B.E.T. surface area of the zeolite pretreated at 673 K was 705 m2 g-l.The reagent SO, gas (Matheson, anhydrous grade) and cis-but-2-ene (Phillips Petroleum, highpurity) were purified five times at 208 K by trap-to-trap distillation. The pyridine (Wako PureChemical Co., extra-pure grade) was dehydrated by adding degassed 13X molecular sieve.PROCEDUREThe apparatus employed for the isomerization of cis-but-2-ene was a static gas-circulationsystem with a volume of 290 cm3.For the infrared spectroscopic measurements of the hydroxylgroups in the zeolite and of the pyridine adsorbed, a self-supporting wafer of the sample zeolitewas prepared by pressing its powder (ca. 12 mg) in 2 cm diameter stainless-steel discs at2.7 ton crn-,. The wafer was then placed into the i.r. cell for heat treatment and spectroscopicmeasurements.Before each experiment an La-X sample was degassed in a vacuum with increasingtemperature (6 K min-l) to a required value between 523 and 973 K. The evacuation wasmaintained at that temperature for 3 h. Following the adsorption of SO, at 298 K for 30 min,the isomerization was begun at the same temperature by feeding and circulating cis-but-2-enegas (1.6 x mol) at a pressure of 1.30 k0.03 x lo4 Pa through the catalyst bed.Thepreadsorbed SO, did not desorb during the progress of the isomerization under the experimentalconditions applied in this work.RESULTSSO2-ENHANCED ISOMER IZ ATIONS OF CiS-BUT-2-ENEThe kinetic curves obtained for the system consisting of La-X and cis-but-2-eneshowed rapid initial formation of trans-but-2-ene and but-1-ene during the first fewminutes followed by a more steady formation of the products with time. Since theresults at times < 180 s are uncertain because of experimental difficulties in theanalysis of the products by gas chromatography, the average quantities of trans-but-2-ene and of but-1-ene formed per s, & and respectively, which werecalculated from the amounts of the two isomers formed during the first 420 s, weretentatively chosen as a measure of the initial activity of the zeolite.The SO,-enhancedactivities, and are the differences in initial activities between thoseobtained in the presence and absence of SO,. The reproducibility of all data was withinThe activities of La-X for the two isomerizations are plotted in fig. 1 as functionsof the quantity of SO, adsorbed on the zeolite pretreated at different temperatures(773-923 K). The effects of SO, concentration on activities for the two isomerizationsare different, especially for the zeolites pretreated at 773 and 873 K. Moreover, as seenfor the three zeolites shown in fig. 1 the enhanced activity for cis-trans isomerizationat high concentrations of SO, is unusually greater than that for double-bond shift.Fig.2 shows the SO,-enhanced activities for the two reactions, AAc.t and AAc.l,as functions of pretreatment temperature for the zeolite. The enhancing action of SO,on the cis-trans conversion can be seen at all the temperatures examined. On the otherhand, for the double-bond shift the favourable action of SO, is seen when the zeoliteshave been degassed at temperatures > 773 K; the reaction is retarded by SO, forzeolites pretreated at lower temperatures. The effects of nitric oxide on the tworeactions are very different ; it poisons the cis-trans conversion effectively but doesnot retard the double-bond migration.In order to determine the contribution of residual water in the SO, gas to theenhanced catalytic activity the effect of pure water on the activity has been examined- +4%K.OTSUKA, Y. TAKIZAWA A N D A. MORIKAWA 83FIG. 1.-Effect of SO, concentration of the activities of La-X for the isomerizations at 298 K: (A, 0, O),Ac-l; (A, ., o), &. Pretreatment temperatures of La-X: A, A, 773; 0, D, 873; 0, 0, 923 K.-'t- 2500 600 700 800 900 1000degassing temp/KFIG. 2.-Effects of pretreatment temperature on the S0,-enhanced or S0,-reduced isomerization. Thequantity of adsorbed SO, was 1.0 x mol g-l, 0, &; 0 , &84 ISOMERIZATION OF BUTENE OVER La-X ZEOLITE1 I0 2 4[H,O (ads.)l/lO4mol g-’FIG. 3.-Effect of water on the catalytic activity of La-X pretreated at 923 K. 0, Ac-,; 0, &.for La-X pretreated at 923 K (fig. 3). The effect of SO, concentration (from fig.1)is also indicated by the dotted curve in fig. 3 using the same scale as that applied foradsorbed water. The results show that the enhancing ability of water is comparablewith that of SO,. However, since the mass spectroscopic analysis has indicated thatimpurity water contained in the SO, was < 1 %, the contribution of impurity water inthe SO, can be neglected in the results of fig. 1 and 2.When the rates from the slopes ofthe kinetic curves at 420 and 3600 s were appliedinstead of the activities shown in fig. 1-3, the shapes of the curves did not changequalitatively. Therefore, we conclude that the activity calculated from the quantitiesof products formed in the first 420 s of reaction can be considered as a measure ofthe intrinsic activity of the catalyst being investigated.EFFECTS OF so, ON THE ACIDIC HYDROXYL GROUPS AND PYRIDINIUM IONSFig.4 shows the series of changes in intensities of the infrared absorption bandswhen successive doses of SO, are added to the surface, or after the evacuation of SO,.For the sample of La-X pretreated at 673 K, the intensities of the bands at 3640 and3600 cm-l due to acidic hydroxyl g r o ~ p s ~ - ~ and to hydroxyl groups characteristic ofa dealuminated La-exchanged faujasite8 both decrease after the addition of SO, andthere is a corresponding increase in absorbance in the lower frequency range between3400 and 3500 cm-l. These changes suggest a hydrogen-bonding interaction of thehydroxyl groups with adsorbed SO,.9 The decrease in intensity of the bondcharacteristic of acidic hydroxyl groups (3640 cm-l) caused by hydrogen bonding withSO, is also seen for the zeolite pretreated at 923 K [fig.4(b), curves (1)-(3)]. Theprocedure of degassing the adsorbed SO, at 473 K restores the intensities of thehydroxyl group bands to levels close to those of the original bands, but never beyondthe original intensities, for zeolites pretreated at both 673 and 923 K [spectra 4(a),curve (9, and 4(b), curve (5)]. In the case of the zeolite pretreated at 923 K, the banK. OTSUKA, Y. TAKIZAWA AND A. MORIKAWA( a )85T I '3600 3520 cm-'i?3640 cm-'FIG. 4.-Changes in the i.r. spectra of OH groups after adsorption or desorption of SO,. The SO,-adsorptionand the measurements of the spectra were carried out at 298 K.(a) Spectra for the 673 K-pretreated La-X:(1) before the addition of SO,, (2) after the adsorption of 2.8 x mol g-' SO,, (3) 8.9 x lop4 mol g-'SO,, (4) after evacuation at 298 K for 25 min, (5) after evacuation at 473 K for 25 min. (b) Spectrafor the 923 K-pretreated La-X: (1) before adsorption of SO,, (2) 2.2 x loF4 mol g-' SO,, (3)9.1 x lop4 mol g-l SO,, (4) after evacuation at 298 K for 25 min, (5) after evacuation at 473 K for 25 min.due to adsorbed SO, (1 325 cm-l) remained after evacuation at 473 K for 25 min. Onthe other hand, for La-X pretreated at 673 K no adsorbed SO, remained afterevacuation. The results in fig. 4 indicate that there are no hydroxyl groups formedafter the adsorption of SO,.The enhancing effect of SO, on the formation of pyridinium ions has been examinedat 298 K by infrared spectroscopic measurements.Fig. 5 shows the effect of SO, onthe absorbance of the band at 1545 cm-l due to the pyridinium ion. The spectra wereobtained at 298 K after the addition of 1.6 x mol g-l pyridine and then after theaddition of SO,86 ISOMERIZATION OF BUTENE OVER La-X ZEOLITEh-i 0. Emd md W2 I8 ii % 0 0 5 10[SO,(ads.)l /lo4 mol g-'FIG. 5.-Effect of SO, on the formation of pyridinium ion at 298 K: 0, 673 K-pretreated La-X; 0,923 K-pretreated La-X.DISCUSSIONMECHANISMS OF THE TWO ISOMERIZATIONSThe different reaction mechanisms for the two SO,-enhanced isomerizations overthe cation-exchanged X- or Y-type zeolites have been proposed previously.'* Thedifferent effects of SO, concentration on the two isomerizations, as demonstrated infig.1 at high concentration of SO,, show that the proposed mechanisms are alsoapplicable to the present case. The cis-trans isomerization at high concentration ofSO, in fig. 1 can be explained by the previously proposed mechanism assuming theformation of sulphone complex from SO, and butenes. The specific cis-transconversion proceeds at the site of the complex through the addition and eliminationof but-2-enes. The selective poisoning by nitric oxide supports the above assumption.'and at low concentrationsof SO, [(0.43-2.4) x mol g-'1 lie between 1.9 and 3.9 (fig. 1). The ratios do notdiffer much from the value obtained in the absence of SO, (Ac-t/A,.l = 2.7). The resultsshow that the cis-trans isomerization observed at low concentrations of SO, can beexplained by the same mechanism as that for double-bond migration. At highconcentrations of SO,, however, the cis-trans conversion proceeds mainly via thesulphone complex as a reaction intermediate.A detailed discussion of the reactionmechanism and of the active sites for cis-trans isomerization may be obtainedelsewhere.'* Hence, the following discussion is limited to the enhancing action ondouble-bond migration.A sec-butyl carbenium ion has generally been postulated as a common reactionintermediate in double-bond isomerization over cation-exchanged zeolites. lo TheSO,-enhanced double-bond migration over La-X in the present work can tentativelybe explained by a carbenium ion-type mechanism.lV 2*For the 923 K-pretreated La-X, the ratios oK.OTSUKA, Y. TAKIZAWA A N D A. MORIKAWA 87ENHANCING ACTION OF so,For the divalent-cation-exchanged zeolites, such as Zn-Xll or Ca-X,12 the rate ofthe S0,-enhanced double-bond shift decreased sharply with a rise in the degassingtemperature for the zeolites above 673 K. The decrease in the SO, effect can beascribed to the dehydration of coordinated water on the exchangeable cations or todehydroxylation of the basic hydroxyl groups during the degassing pretreatment athigh temperatures.In contrast to the divalent-cation-exchanged zeolites, the enhancing effect ofSO, on hydrogen-Y zeolites has been observed only when the zeolites had been pre-treated at temperatures > 723 K.l The effect increased on increasing the degassingtemperature.' These observations have suggested that the sites formed after dehydr-oxylation of the zeolite are essential for the enhancing action of SO, on hydrogen-Y.lFor La-X, the fact that the enhancing effect of SO, on the double-bond isomerizationemerges only when the catalyst is pretreated at temperatures > 773 K (fig.2) impliesthat the enhancing mechanism of SO, is of the same type as that proposed fordehydroxylated hydrogen-Y zeo1ite.l It is believed that dehydroxylation of hydrogen-Yzeolites occurs at degassing temperatures > 773 K7,l3 The SO, adsorbed in thevicinity of the electronegative sites generated through dehydroxylati~n~? l3 wouldwithdraw a negative charge from the surface, increasing the acid strength of theresidual Bronsted sites by an inductive effect.l The chemisorbed SO, observed by thei.r.measurements for La-X pretreated at 923 K may be the species adsorbed on thesesites. The i.r. spectroscopic studies did not show any newly formed acidic hydroxylgroups after the addition of SO, (fig. 4) but showed an increase in the reactivity ofthe group with pyridine for the zeolite pretreated at 923 K (fig. 5). For La-X pretreatedat 673 K neither the enhancing action of SO, on the rate of double-bond shift northe SO,-enhanced formation of pyridinium ions was observed. These results supportthe above consideration that the enhancing mechanism of SO, for La-X is similar tothat proposed for hydrogen-Y.The apparent activation energy for double-bondmigration, calculated from an Arrhenius plot of the reaction rates at 420 s for the923 K-pretreated La-X, decreased from 29.3 (without SO,) to 18.4 kJ mol-l (with1 .O x mol g-l of SO,). This fact also supports the idea that SO, does not increasethe number of active sites but enhances the activity of the Bronsted acid sites.The poisoning effect of SO, observed when La-X had been degassed at temperatures< 773 K (fig. 2) must be attributed to the adsorption of SO, on the Bronsted sitesthrough a hydrogen bond, blocking the approach of the butene molecules. Thedecrease in the rate of double-bond shift on increasing the quantity of adsorbed SO,at high SO, concentrations, as is shown in fig. 1, can also be ascribed to thedeactivation of Bronsted sites by the adsorption of SO, on these sites. However, atlow concentrations of SO,, the negative effect of SO, can be neglected because SO,would adsorb preferentially on the dehydroxylated sites.K. Otsuka and A. Morikawa, J. Chem. SOC., Faraday Trans. 1, 1980, 76, 1196.2 K. Otsuka and A. Morikawa, J . Catal., 1979, 56, 88.3 Y. Ishinaga, K. Otsuka and A. Morikawa, Bull. Chem. SOC. Jpn, 1979, 52, 933.4 K. Otsuka, Y. Wada, K. Tanaka and A. Morikawa, Bull. Chem. SOC. Jpn, 1979, 52, 3443.J. W. Ward, Adv. Chem. Ser., 1971, 101, 380.J. W. Ward, J. Phys. Chem., 1968, 72, 2689,421 1 .C. L. Angel1 and P. C. Schaffer, J . Phys. Chem., 1965, 69, 3463.* J. Scherzer and J. L. Bass, J. Phys. Chem., 1975, 79, 1200.A. V. Deo, I. G. Dalla Lana and H. W. Habgood, J. Catal., 1971, 21, 270.lo P. A. Jacobs, Carboniogenic Activity of Zeoliles (Elsevier, New York, 1977).l1 K. Otsuka, R. Oouchi and A. Morikawa, J. Catal., 1977, 50, 379.l2 K. Otsuka, unpublished data.l3 J. B. Uytterhoeven, L. G. Christner and W. K. Hall, J . Phys. Chem., 1965, 69, 2117.(PAPER O/ 175

ISSN:0300-9599    

DOI:10.1039/F19817700081

出版商:RSC    

年代:1981

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I, 1981, 77, 89-100Free-radical Addition to OlefinsPart 26,--Kinetics of the Addition of Trifluoromethyl Radicals to Acetylene andSubstituted Acetylenes'BY AMR EL SOWENI, JOHN M. TEDDER* AND JOHN C. WALTONDepartment of Chemistry, The University, St Andrews, Fife KY 16 9STReceived 7th February, 1980Absolute Arrhenius parameters have been determined for the addition of trifluoromethyl radicals toacetylene, propyne, but-2-yne, 1 , l , 1-trifluoropropyne and hexafluorobut-2-yne by a competitive methodusing the previously determined Arrhenius parameters for the addition of trifluoromethyl radicals toethylene as standard.I44HCECHHECCH,HCGCCF, 4CH,CGCCH,4CF,CECH4CF,ECCF,1 5.005 0.30 8.70 f 0.255.2b 3.45 f0.28b 8.63 +0.08b0.49 5.10 & 0.26 8.44+ 0.132.0 4.36 0.30 8.68 k 0.160.043 6.61 k0.33 8.13 k 0.180.068 5.86 f 0.38 7.95 + 0.061 cal = 4.1845 J.* Arrow shows site of attack; sum of attack at both sites in propyne.Both the activation energies and the A-factors are larger than those for the addition of trifluoromethylradicals to similarly substituted olefins.As a result of these opposing effects the rates of addition are similar.The first studies of the addition of trifluoromethyl radicals to acetylenes were thoseof Haszeldine.2f The initial paper reported the thermal and photochemical additionof trifluoromethyl iodide to acetylene itself and the second paper reported thephotochemical addition of trifluoromethyl iodide to 1 , 1,l -trifluoropropyne yieldingl,l, 1,4,4,4-hexafluor0-2-iodobut-2-ene together with some telomer. No kinetic datawere reported.In their important compilation of kinetic data of gas-phase additionreactions Kerr and Parsonage4 record only four determinations of the rate of additionof methyl radicals to acetylenes, three to acetylene itself5-' and one to propyne.* Onlyone paper dealing with trifluoromethyl radical addition to acetylenes is reported, andthis work by Szwarc and co-worker~~ describes the determination of the overall rateof addition of trifluoromethyl radicals to acetylene, propyne and but-2-yne. The workwas put on an absolute scale by competition with hydrogen abstraction from2,3-dimethyl butane.Since the compilation of kinetic data4 was complete Kerr has re-examined thekinetics of the addition of methyl radicals to ethylene and acetylene.'* The early worksuggested that at temperatures around 150 O C the rates were similar.The recent work4 89 FAR 90 FREE-RADICAL ADDITION TO OLEFINSconfirms this but shows that both Arrhenius parameters are larger for the additionto acetylene (log kCaHl = 8.78 & 0.2 - 7.7 f 1.5/2.303 RT dm3 mol s-l) than ethylene(log kCzHI = 8.22 f 0.5 - 7.3 f 1.0/2.303 RT dm3 mol s-l). Cationic addition to acety-lenes in solution is well established as proceeding substantially slower than additionto similarly substituted olefins (Br+ adds to RCH-CHR’ anything from 20 to 5 x lo4faster than to RCrCR’).ll Taking these results together it would be expected thatelectrophilic radicals like trifluoromethyl would add more readily to olefins than toacetylenes. Qualitative evidence that this is so comes from the researches of Kharaschwho found that trichloromethyl radicals add faster to an alk-1-ene than to analk-l-yne.12 The only quantitative data are those of Szwarc and co-workers referredto above.The Arrhenius parameters quoted by Szwarc are based on only threetemperatures and are unlikely to be very acc~rate,~ more especially as the competitivemethod these workers were using has been que~ti0ned.l~ However, it has been ourfinding that although Arrhenius parameters differ substantially from Szwarc’s we haveinvariably found similar relative reactivity.Apart from Szwarc’s work there are no kinetic data for the addition of trifluoro-methyl radicals to acetylenes and confirmation of the high pre-exponential termswould be of considerable importance.In addition, this work is intended to be anextension of our study of directive effects in free radical addition rea~ti0ns.l~ Thereare, as far as we are aware, no kinetic data for the addition of radicals to specific sitesin unsymmetrical acetylenes and only two studies of the stereochemistry of the adductsformed by the addition of radicals to acetylenes.EXPERIMENTAL AND RESULTSThe experimental method has been described in previous papers.15 The light sourcewas a 100 W medium-pressure mercury arc. Acetylene, obtained from an ordinarycommercial cylinder, was bubbled through sulphuric acid to remove traces of acetone,and then passed through a tower of sodium hydroxide.The gas used in the experimentsshowed only a single peak in the gas chromatogram. The substituted acetylenes werecommercial products which, apart from trap-to-trap distillation, were used withoutfurther purification. The trifluoromethyl iodide was supplied by Bristol Organics andwas stored in the dark.ACETYLENEIn a preliminary preparative experiment trifluoromethyl iodide and acetylene werephotolysed at 150 O C for 2 h. The g.1.c. showed two product peaks which massspectrometry showed to have the same mass (C3H,F31 m/e = 222 M+). A further seriesof preparative runs were then completed in order to distinguish between the isomersby n.m.r. spectroscopy. Only the major adduct was obtained in sufficient quantity forspectroscopic examination.This showed two doublets of quartets in the lH n.m.r.,JH{*)H(,) = 15 Hz which is consistent with structure 1,16 and the minor product musttherefore have structure 11.6 = 6.74 ppm, JCF3H(I) = 6 Hz; JH(2)H(1) = 15 Hz; 6 = 7.32 ppm, JCF3H(2) = 2 HzAMR EL SOUENI, J. M. TEDDER A N D J. C. W A L T O N 91TABLE 1 .-ADDITION OF TRIFLUOROMETHYL RADICALS TO ACETYLENE[CF31] = 3.3 x rnol dm-3, [C2H2] = 4.5 x mol dm-3.32334536739341 14374.13.353.02.01.351.20.0760.0870.100.1 10.120.13A least-squares plot of log II/I against 103K/T gave a straight line of gradientA series of runs in which the acetylene concentration was varied from 4.3 xA second series of experiments was completed in which ethylene was an additional- 0.29 f 0.02 and intercept - 0.20 k 0.05.to 17.3 xreactant.mol dm-3 showed the ratio of the 2 and E isomers to be constant.TABLE 2,-cOMPETITIVE ADDITION OF TRIFLUOROMETHYL RADICALS TO ETHYLENE AND ACETYLENE[CF31] = 4.4 x 10-3 mol dm-3, [C,H,] = 1.3 x mol dm-3, [C,H,] = 3.9 x lop4 mol dm-3.35 1 1.1 0.069 0.0063 0.07537 1 0.3 0.075 0.0078 0.084393 0.2 0.093 0.0102 0.10241 7 0.15 0.108 0.01 32 0.120423 0.11 0.1 14 0.0144 0.126453 0.07 0.126 0.0171 0.141A least-squares plot of log ([E-C3H2F31]/[C3H,F31]) against lo3 K/Tgave a gradientof -0.43k0.025 and an intercept of 0.58f0.06.A similar plot of log[Z-C,H2F,I]/[C3H4F31] against lo3 K/Tgave a gradient of - 0.73 f 0.025 and an interceptof 0.38 _+ 0.06.A third plot of logZ[C,H2F,I]/[C,H,I] against lo3 K/T gave a gradientof -0.46kO.025 and an intercept of 0.7k0.06.A further series of experiments was carried out in which the acetylene concentrationwas varied while all the other variables (light, temperature and CF,I) were keptconstant. Within experimental error the product ratio [E-C3H2F,I]/[Z-C,H2F31]remained constant.4-92 FREE-RADICAL ADDITION TO OLEFINSPROPYNEThere are four expected adducts from the additions of trifluoromethyl iodide topropyne :H CH3 CF3 CH3 I CH3\ / \ / \ / \ / c=c c=c c-c c-c/ \ / \ / \ / \CH3I CF3 I H CF3 H CF3I11 IV V VIPreparative g.1.c. separated two product peaks both of which had parent ions in themass spectrum (m/e = 236) corresponding to the adducts.The lH n.m.r. spectrum ofthe major peak consisted of a doublet of quartets at 6 = 2.67 ppm (JCFICHI = 2.1 Hz;JH, CH3 = 1.55 Hz) and a quartet of quartets at 6 = 6.36 ppm ( J C H , , = 1.5 Hz;,ICF3, = 7.7 Hz). In addition there were two low-intensity peaks at 6 = 1.97 ppm and6 = 7.28 ppm. The ratio of the high and low intensity lines was 15: 1. The 19F n.m.r.spectrum consisted of a doublet of quartets at q5 = 59.17 ppm (JCH3, CF3 = 2.1 Hz;JCF,, = 7.7 Hz) together with a low-intensity doublet of quartets q5 = 69.17 ppm(JCHICF3 = 0.3 Hz; JH,CF3 = 1.7 Hz). The spectra correspond to a mixture of I11(main product) and V. The lH n.m.r. spectrum of the smaller peak in the gaschromatogram consisted of a doublet of quartets at 6 = 2.69 ppm (JCH,, CFI = 2.2 Hz;J C H , , = 1.6 Hz) together with a quartet of quartets at 6 = 6.24 ppm(JCF3, = 7.2 Hz; JCHIH = 1.5).The 19F n.m.r. consisted of a doublet of quartets atq5 = 61.16 ppm (.ICH,, CF3 = 2.2 Hz; JCF3, = 7.2 Hz). These spectra correspond toIV. Isomer VI if formed was present in too small a proportion to be identified.TABLE 3 .-COMPETITIVE ADDITION OF TRIFLWOROMETHYL RADICALS TO ETHYLENE AND PROPYNE[CF31] = 3.26 x mol dm-3, [CH,C'CH] = 8.6 x[CH,=CH,] = 3.9 x lo-* mol dm-3.mol dm-3,435 0.07 0.46 0.047 0.51427 0.13 0.45 0.043 0.49405 0.16 0.43 0.033 0.46385 0.3 0.44 0.037 0.47365 0.45 0.42 0.028 0.44383 1.3 0.39 0.022 0.413 14 2.3 0.38 0.023 0.40A least-squares plot of log([Z-C4H4F,I]/[C3H4F3I]) against 1 O3 K/ T gave a gradientof -0.095_+0.009 and an intercept of 0.22k0.02; a similar plot of log([&C4H4F,I]/[C,H4F,I]) gave a slope of - 0.37 f 0.06 and an intercept of - 0.17 0.15;a third plot of log@[C4H4F,I]/[C,H,F31]) gave a slope of -0.12+0.01 and anintercept of 0.33 f 0.03 and a fourth plot of log([E-C4H,F31]/[Z-C4H4F,I]) gave a slopeof -0.25 f 0.06 and an intercept of - 0.43 k 0.15AMR EL SOUENI, J.M. TEDDER AND J. C. WALTON 93BU T-2-Y NEIn preliminary experiments in which trifluoromethyl iodide was photolysed in thepresence of ethylene and but-2-yne, three adduct products were detected, two isomerswith m/e = 250 M+(C,H,F,I) and the ethylene adduct. The isomers of m/e = 250 wereseparated by preparative g.1.c. The lH n.m.r. spectrum of the major product had abroad quartet at 6 = 2.12 ppm (JCH,(a), CH,(b) = 1.6 Hz), together with a complex setof lines at 6 = 2.82 ppm which could be assigned to the splitting of the CH,(a) protonsby those of CH,(b) and the fluorines of the CF, [the coupling constant between thefluorine atoms and the protons of CH,(b) is small].The leF n.m.r. spectrum consistedVII VIIIof a quartet at q5 = 58.16 ppm (JCH,(a), CF3 = 2.3 Hz). The lH n.m.r. spectrum of theminor isomer consisted of a broad quartet, 6 = 1.91 ppm (JCH,(a), CH,(b) = 1.2 Hz),together with a complex set of lines at 6 = 2.7 ppm. The 19F n.m.r. spectrum consistedof a quartet at q5 = 62.95 ppm (JCH,(a),CF, = 2.5 Hz). These n.m.r. spectra, whencompared with those of the isomers for the addition to propyne, lead to assigning themajor isomer, structure VII and the minor product structure VIII.TABLE 4.-cOMPETITIVE ADDITION OF TRIFLUOROMETHYL RADICALS TO ETHYLENEAND TO BUT-2-YNE[CF31] = 4.35 x mol dm-3, [CH,CZCCH,] = 8.64 x mol dm-3,[CH,=CH,] = 3.9 x lod4 mol dm-3.327 2.4 0.18 0.03 I 0.21336 2.1 0.21 0.041 0.25360 1.4 0.23 0.043 0.28383 1.15 0.25 0.054 0.30408 1 .o 0.29 0.054 0.34436 0.4 0.34 0.072 0.41479 0.1 0.37 0.086 0.45A least-squares plot of log([Z-C,H,F,I]/[C,H4F31]) against l O3 K/T gave a gradientof -0.31+0.02 and an intercept of 0.57k0.05; a similar plot of log([&C5H6F31]/[C3H4F31]) had a slope of - 0.39 _+ 0.05 and an intercept of 0.07 _+ 0.14; athird plot of log (C[C,H,F,I]/[C,H,F,I]) had a gradient of -0.32k0.02 and anintercept of 0.68 k 0.06.The fourth plot (log [E-C,H,F,I]/log[Z-C,H,F,I]) had agradient of -0.09 k0.04 and an intercept of - 0.47 & 0.1194 FREE-RADICAL A D D I T I O N TO OLEFINS1 , 1 , 1 -TRIF LUOROPROPY NEIn preliminary experiments trifluoromethyl iodide was photolysed in the presenceof 1 , 1 , 1 -trifluoropropyne and four adduct products were identified. Three wereisomers, m/e 290 (M+ C,HF,I) and the fourth with a higher mass m/e 384 (M+C,H,F,I). The lH n.m.r. spectrum of the first isomer to be eluted, the major isomer,consisted of a quartet of quartets at 6 = 7.18 ppm (JCF3(a), = 1.4 Hz;JCF,(b), = 6.5 Hz). The 19F n.m.r. spectrum consisted of two sets of double quartetsJCF,(a), = 1.4 Hz). These spectra are consistent with structure IX.at # = 62m34 ppm (JH, CF3(b) = 6*5 Hz; JCF3(a), CF,(b) = le4 Hz; JCF,(b), CF,(a) = 1.4 Hz;IX X XIThe lH n.m.r.spectrum of the second product eluted consisted of a quartet at6 = 8.24 ppm (JCF,(a), = 1.5 Hz) and the 19F n.m.r. spectrum consisted of a quartetof doublets at # = 64.7 ppm ( J H , CF,(a) = 1.5 Hz; JCF,(a), CF,(b) = 6.5 Hz) and aquartet at $ = 61.9 ppm (JCF3, CF3(b) = 6.7 Hz). These spectra are consistent withstructure X. The lH n.m.r. spectrum of the third product eluted consisted of a quartetat 6 = 6.89 ppm (JcF3(a,, = 8.1 Hz). The 19F n.m.r. spectrum consisted of a doubletof quartets at # = 59.1 ppm (JCF3(a), CF,(b) = 11.2 Hz; JH, CF3(a) = 8.1 Hz) and aquartet at $ = 60.9 ppm (JCF3(a), CF3(b) = 11.2 Hz). These spectra are consistent withstructure XI.The lH n.m.r. spectrum of the remaining product with the highermolecular weight consisted of a quartet at 6 = 6.46 ppm (JCF3(a), Hi = 7.8 Hz) and abroad peak at 6 = 7.2 ppm. The 19F n.m.r. spectrum had a doublet at # = 62.1 ppmand two singlets at # = 66.97 and 67.28 ppm. These two spectra correspond to structureXII./ \H2 IXI1In a series of runs at different temperatures trifluoroiodomethane and l , l , l -trifluoropropyne were photolysed in the temperature range 56- 186 OC. The productsconsisted of the E- and 2-isomers of CF,CH=CICF, and (CF,),C=CHI; the telomerC,H,F,I was only present in the high conversion runs used for structural analysis.A least-squares plot of log([E-CF,CH=CICF,]/[Z-CF3CH=CICF3]) against103K/Tgave a line of gradient -0.16k0.01 and an intercept of -0.43k0.03; asimilar plot of log{[(CF3),C=CHI]/Z[CF3CH=CICF3]] against 1 O3 K/ T gave agradient of -0.34k0.05 and an intercept of -0.3kO.l.A second set of lowconversion competitive runs was completed using trifluoroiodomethane (43.5 x l 0-4mol dm-3), l,l,l,-trifluoropropyne (13.0 x lo-, mol dm-3) and ethylene (3.9 x lodAMR EL SOUENI, J. M. TEDDER AND J. C. WALTON 95TABLE 5.-ADDITION OF TRIFLUOROIODOMETHANE TO 1,1,1 -TRIFLUOROPROPYNE[CF,I] = 32.6 x mol drn-,, [CF,C=CH] = 8.6 x mol drn-,.[(CF,),C=CHI] [E-CF,CH=CICF,][Z-CF,CH=CICFJ [Z-CF,CH=CICF,]32936039 141 8459421.100.270.150.0530.0610.0770.0970.100.120.140.140.160.16mol drn-,). The consumption of ethylene was kept below 5% and the temperaturevaried over a 150 degree range.Three products were obtained in sufficient quantitiesfor accurate analysis, CF,CH,CH,I ; 2-CF,CH=CICF, and C6H5F61. The lattercompound was identified by its mass spectrum as the cross telomerCF,CH=C(CF,)CH,CH,I but it was not possible to determine its stereochemistry.TABLE 6.-COMPETITIVE ADDITION OF TRIFLUOROIODOMETHANE TO ETHYLENE AND1,1,1 -TRIFLUOROPROPYNET/K [Z-CF,CH=CICF,] [C6H, ~ 6 1 1[CF,CH,CH,I] [CF,CH,CH,I]3183383483603934254474733.42.21.451.150.400.160.080.020.0440.0530.0570.0600.0790.0860.100.120.0100.0120.0150.02 10.0160.0140.0220.020A least-squares plot of log({[Z-CF,CH=CICF,] + [c,H5F,I]}/([cF,cH,cH2~]/[CH,=CH,]/[CF,C=CH]}) against 1 O3 K / T gave a gradient of - 0.36 0.02 and anintercept of - 0.11 & 0.06. By combining the results from tables 5 and 6 the absoluteArrhenius parameters for the addition of trifluoromethyl radicals to both ends of1 , 1 , 1 -trifluoropropyne could be calculated using the parameters we have previouslydetermined for eth~1ene.l~ (These are given in table 10.)HEXAF LUOROBUT-2-Y NEWhen it was attempted to study the addition of trifluoromethyl radicals tohexafluorobut-2-yne competitively with ethylene, the consumption of ethylene wasrelatively so rapid that it was impossible to use unintegrated rate expressions. Insteadtherefore hexafluorobut-2-yne was studied competitively with trifluoroethylene.Trifluoroiodomethane (3.8 1 x lo-, mol dm-3), hexafluorobut-2-yne (5.85 xmol drn-,) and trifluoroethylene (5.85 x mol drn-,) were photolysed togetherfor 2.3 h.Three products were isolated, the two simple adducts CF,CHFCF,Iand (CF,),C=CICF, together with the one to one telomer(CF,),C=C(CF,)CHFCF,I. A series of runs of varying time and temperature wasthen completed96 FREE-RADICAL ADDITION TO OLEFINSTABLE 7.-cOMPETITIVE ADDITION OF TRIFLUOROIODOMETHANE TO TRIFLUOROETHY LENE ANDHEXAFLUOROBUT-2-Y NE[(CF,),C=CICF,] [C,HFI,Il[CF,CHFCF,I] [CF,CHFCF,I]3273433673833934334734.303.352.502.302.101.100.140.630.640.760.770.790.840.99-0.0620.0460.0380.0550.0500.050A least-squares plot of 1og(([(CF3),C=C1CF,] + [C,HF,,I]}/{[CF3CHFCF21][CHF=CF,]/[CF,C~CF,]}) against lo3 K/T gave a line with a gradient- 0.23 0.02 and an intercept of 0.50 f 0.06.Using our previous data for the relativerate of addition of trifluoromethyl radicals to trifluoroethylene and our absoluteArrhenius parameters for ethylene, absolute parameters could be calculated.DISCUSSIONThe mechanism of the photochemical addition of trifluoromethyl iodide to alkynesis exactly similar to the analogous addition to alkenesCFJ + CF; + I - (1)CF,+A + CF,A* (2)(3 a) CF,A* + CFJ + CF,AI + CF, 0 .The initiation step is followed by the chain-carrying steps (2) and (3), in which A standsfor alkyne. In many of the present experiments the alkyne was reacted competitivelywith an alkene (E).CF;+E+CF,E* (2 4(3 4 CF,E- + CF,I --+ CF,EI + CF, .The chains are long, and making the normal steady state assumptions we have forsmall conversions :where the subscripts f and i stand for final and initial, respectively. In the present workE was either ethylene itself or trifluoroethylene for both of which the value ofArrhenius parameters A,, and E,, are known2 so that the corresponding Arrheniusparameters A,, and E,, could be determined and listed in table 8.The addition of trifluoromethyl iodide to an alkyne can yield two geometric isomers( E and 2) and, if the alkyne is unsymmetric, this will lead to four isomers in all.However, with propyne only three isomers were identified.1 , 1 , 1 -Trifluoropropyneyielded the expected three isomers (there is only one isomer when the CF; radicaladdition occurs at the CF,-end).When reacted by itself, 1, 1,l-trifluoropropyne yieldeAMR EL SOUENI, J. M. TEDDER A N D J. C. W A L T O N 97TABLE 8.-ABSOLUTE ARRHENIUS PARAMETERS AND RELATIVE RATES FOR THE ADDITION OFCF; RADICALS TO ALKYNES AND ALKENESacetylenesak;/k:zHZlog A; E (1 64 "C)H-CEC-H 8.70 5.00 1H-C=C-CH, 8.63 3.45 5.2H-C'C-CF, 8.44 5.10 0.49CH,-C=C-CH, 8.68 4.36 2.0CF,-C=C--H 8.13 6.61 0.043CF,-C=C-CF, 7.95 5.86 0.068ratios C=C/C-C olefinseH-CH=CH-H 8.00 2.85 1 -0.70 2.15 2.4H-CH=CH-CF, 7.76 3.10 0.43 -0.68 2.00 1.95CH,-CH=C(CH,), 7.48 0.90 2.9 - 1.20 3.46 3.26CF,-CH=CH-H 7.74 5.76 0.18 -0.42 0.85 9.0H-CH=CH-CH, 7.92 1.98 2.3 -0.71 1.47 1 .o- -_ - - CF,--CH=CH--CF, - -a 1 : 1 telomer CF,CH=C(CF,)CH=CICF, (this was originally observed byHaszeldine), and when reacted competitively with ethylene it yielded the cross-telomerCF,CH=C(CF,)CH,CH,I. These telomeric products can probably be attributed tothe great reactivity displayed by a vinylic radical in which the a-position carries atrifluoromethyl group (=C-CF,).Exactly similar telomeric products were observedwhen hexafluorobut-2-yne was reacted competitively with trifluoroethylene. Again thehigh reactivity of the a-trifluoromethyl vinyl radical is observed. We can representthe telomer formation with 1 , I , 1 -trifluoropropyne as follows :-CF, + CF,C-CH --+ CF,CH=&.F, (4)either with itself:CF, CF,or with ethylene: II CF I ICF,CH=CCF, + CH,=CH, + CF,CH=CCH,CH,* CF,CH=CCH,CH,I.(6)(7)The reaction of hexafluorobut-2-yne with trifluoroethylene is similar:CF, + CF,C=CCF, + (CF,),C=&F,CF3 CF3CFJ I I(CF,),C=cCF, + CHF-CF, 4 (CF,),C=CCHFCF, * (CF,),C=CCHFCF,I.(898 FREE-RADICAL ADDITION TO OLEFINSDiadducts in which a second CF,I molecule adds across the double bond in theprimary adduct were not observed in any of these experiments.The Arrhenius parameters for the addition of trifluoromethyl radicals to alkynesare compared with the parameters for similarly substituted alkenes in table 8.Thistable also presents the ratio of rate constants for the two types of addition. A verystriking feature is the close similarity of the rate of constants of addition at 164 O C .Attack occurs faster at the alkene than at the alkyne, but the difference in rate is nevergreater than a factor 9, although within the alkyne series the rates vary by more thanan order of magnitude.This shows that substituents have very similar effects on thereactivity of both double and triple bonds and establish that free-radical addition toalkynes and alkenes must be a very similar process. The electrophilic trifluoromethylradicals add faster to the 1-positions in both propyne and propene and slower to theequivalent position in both 1 ,1, 1 -trifluoropropyne and 1 ,l,l -trifluoropropene than tothe corresponding sites in acetylene and ethylene, respectively. The 2-positions in allfour unsaturated molecules are less rapidly attacked, the 2-positions in trifluoropro-pyne and trifluoropropene being particularly unreactive.In spite of the very similar relative rates of addition at 164 O C , the Arrheniusparameters for addition to the alkynes are substantially different from the parametersfor addition to similar sites in the alkenes.The pre-exponential terms are withoutexception larger for addition to the alkynes than for addition to the alkenes, by a factorof between 2.5 and 5. This increased A-factor is offset by an appreciably largeractivation energy for the additions to the alkynes. These results are very satisfactoryconfirmation of theoretical expectation. The large A-factors for the alkyne additionsare in accordance with simple transition-state theory. Ethylene has a rotation aboutthe carbon5arbon axis which is lacking in acetylene and the entropy of acetylene istherefore considerably less than that of ethylene.In the transition state this distinctiondisappears and hence the overall entropy of activation A S is less in the addition ofa radical to an acetylene. Alternatively we can use Benson’s group additivity termsto determine the total entropy changes. The calculated A s 0 values for the additionof CH, and CF, to acetylene to form the propenyl and the 3,3,3-trifluoropropenylradicals are found to be - 25.8 and - 34.6 cal mol-l K-l, respectively, referred to thestandard state of 1 mol drn-,. Similarly the As0 values for the addition of the sameradicals to ethylene to form propyl and 3,3,3-trifluoropropyl radicals are found to be- 30.0 and - 35.3 cal mol-l K-l.These total entropy changes should parallel theentropies of activation and confirm that higher A-factors for radical addition toacetylenes are to be expected.The activation energies for the addition of trifluoromethyl radicals are withoutexception substantially larger for addition to the alkynes than the correspondingalkenes. This is consistent with the electrophilic character of the trifluoromethylradical and the known reluctance of electrophiles to add to acetylenes.The only previous study involving trifluoromethyl radicals where both geometricisomers have been reported in a single qualitative experiment involves the additionof trifluoromethyl iodide to propyne.16 There are studies of the addition of sulphurradicals to acetylenes (SF,* and CH,S*), in which both geometric isomers have beenisolated.The addition of thiyl radicals to both olefins and acetylenes is known to bereversible, but Heiba and Dessau made the important observation that the additionof CH,S* radicals to acetylenes was much less reversible than the correspondingaddition to 01efins.l’ They also observed that the resultant vinyl radical was much morereactive than the corresponding alkyl radical. Szwarc has provided good evidence thatthe addition of trifluoromethyl radicals to alkenes is not reversible, so the additionto alkynes is even less likely to be reversible.lAMR EL SOUENI, J. M. TEDDER A N D J. C. W A L T O N 99The interconversion of E and 2 vinyl radicals is known to be very rapid even at- 180 O C in and, although the size of substituents affects the rate ofisomerisation, it is probable that at the temperatures of the present work it was veryrapid indeed.If we assume that the rate of iodine abstraction from iodotrifluoro-methane is fast and similar for the two vinyl radicals, the proportions of the two adductgeometric isomers will be governedgeometric isomersby the equilibrium between the two vinyl radicalIn the Experimental section we report the variation of the ratio of the geometric isomeradducts with temperature and hence we can calculate the enthalpy and entropydifference between the vinyl radical geometric isomers. Addition to acetylene ispredominantly trans, while addition to propyne, but-2-yne and 1,l , l -trifluoropropyneis predominantly cis.Table 9 shows that this difference can be attributed to the relativeTABLE 9.-RATIO OF THE GEOMETRIC ISOMER ADDUCTS AND A F AND FOR THECORRESPONDING VINYL RADICALSalkyne2 and E vinyl radicalsproduct ratioZ : E, AH”I A PI160 OC kcal mol-l cal mol-l K-’CH=CHCH,C=CHCH,C=CCH,CF,C=CH- 0.13 - 1.339 . 8 1.154.8 0.426.2 0.74.1.02.02.22.0stabilities of the vinyl radicals. These results can be rationalised in terms ofintramolecular repulsion. If it is assumed that the single electron (like a lone pair)occupies more space than a hydrogen atom but less space than methyl or atrifluoromethyl group (i.e. the effective size of groups is H < loneelectron 4 CH, < CF,) then we would expect the E-isomer to predominate foracetylene itself but that the 2-isomer would predominate for all the other additions.Similarly the difference in enthalpy between the two radicals derived from but-2-yneis the smallest difference of the series, as required by the intramolecular hypothesis.Part 25, J.M. Tedder, J. C. Walton and L. L. T. Vertommen, J . Chem. SOC., Faraday Trans. I , 1979,75, 1040.R. N. Haszeldine, J. Chem. SOC., 1950, 3037.R. N. Haszeldine, J. Chem. SOC., 1952,2504; K. Leedham and R. N. Haszeldine, J . Chem. SOC., 1954,1634.J. A. Ken- and M. J. Parsonage, Evaluated Kinetic Data on Gas Phase Addition Reactions, (Butter-worths, London, 1972).L. Mandelcorn and E. W. R. Steacie, Can. J. Chem., 1954, 32,4741 00 FREE-RADICAL A D D I T I O N TO OLEFINSL. C. Landers and D. H. Volman, J. Am. Chem. SOC., 1957, 79, 2996.J. A. Garcia-Dominguez and A. F. Trotman-Dickenson, J. Chem. SOC., 1962, 940.R. R. Getty, J. A. Kerr and A. F. Trotman-Dickenson, J. Chem. SOC. A , 1967, 1360.@ G. E. Owen, J. M. Pearson and M. Szwarc, Trans. Faraday SOC., 1965, 61, 1722.lo R. M. Holt and J. A. Kerr, Int. J. Chem. Kinet., 1977, 9, 185.l1 P. W. Robertson, W. E. Dasent, R. M. Milburn and W. H. Oliver J. Chem. SOC., 1950, 1628.l2 M. S. Kharasch, J. J. Jerome and W. H. Urry, J. Org. Chem., 1950, 15, 966.l3 S. E. Braslavsky, F. Casas and 0. Cifuentes, J. Chem. SOC. B, 1970, 1059.l4 J. M. Tedder and J. C. Walton, Acc. Chem. Res., 1976,9, 183; Adv. Phys. Org. Chem., 1978, 16, 51l5 H. C. Low, J. M. Tedder and J. C. Walton, J. Chem. SOC. Faraday Trans. 1, 1976, 72, 1300.l6 L. P. Anderson, W. J. Feast and W. K. R. Musgrave, J. Chem. SOC. C, 1969, 214.l7 E. I. Heiba and R. M. Dessau, J. Org. Chem., 1967, 32, 3837.J. M. Pearson and M. Szwarc, Trans. Furaday Soc., 1964, 60, 553.Is R. W. Fessenden and R. H. Schuler, J. Chem. Phys., 1963, 39, 2147.(PAPER 0/227

ISSN:0300-9599    

DOI:10.1039/F19817700089

出版商:RSC    

年代:1981

数据来源: RSC