摘要:

J . Chem. SOC., Faraday Trans. 1, 1985,81, 1037-1046 Equimolar Mixtures of Trivalent Metal Perchlorates as Constant-ionic-strength Media in Studies of Complex Formation in Dimethyl Sulphoxide Iron(m) and Aluminium(rr1) Thiocyanate Complex Formation BY DANUTA PUCHALSKA* AND DANUTA WOJCIK Department of Physical Chemistry, Technical University of Gdansk, 80-952 Gdansk, Poland Received 6th July, 1984 Equimolar mixtures of aluminium(m) and iron(m) perchlorates in dimethyl sulphoxide have been used as constant-ionic-strength media for studying the formation of complexes of the metal cations with the thiocyanate anion by spectrophotometric methods. Formation of the complexes has also been studied calorimetrically. The derived thermodynamic parameters are as follows : AG* = - 23.2 0.5 kJ mol-l, A H 0 = 5.04 & 0.04 kJ mol-l and A F = 95 & 1.8 J K-l mol-1 for FeSCN(DMSO)E+ and AGe = - 15.2 f 1.7 kJ mol-l, A H 0 = 15.42 f 0.1 1 kJ mol-1 and A P = 103f6 J K-l mol-1 for AlSCN(DMSO);+.The stability constants of ionic complexes are determined either in constant- ionic-strength media or in ionic solutions where the variation of the activity coefficients must be taken into account, Both procedures are unsatisfactory, especially for non-aqueous systems. The use of a supporting electrolyte cannot always ensure constancy of the activity coefficients.l? Moreover, little information is obtained as to what happens in pure solutions of the metal salts. The alternative approach, of investigating coordination equilibria under conditions of variable activity coefficient, provides the thermodynamic par- ameters of complex formation and allows the effect of solvent on the equilibria to be discu~sed.~ However, it suffers from inadequate knowledge of the concentration dependence of the activity coefficients, especially in non-aqueous systems.Therefore, the method proposed by Libui,,~ which enables the activity coefficients to be held constant, is of considerable interest. This method makes use of the fact that for some systems the activity coefficients are uniquely determined by the coordination states of the salts present, irrespective of the nature of the central metal atoms. In a series of papers it has been shown that aqueous equimolal divalent-transition-metal perchlorates and magnesium perchlorate behave as effective constant-ionic-strength media,g-fO as has also been found for transition-metal perchlorates in acetonitrilell and DMSO solution^.^ This is accounted for by the fact that the metal cations in these solvents occur in the form of octahedral hexasolvo-complexes.It is our intention to check experimentally the validity of the above conclusions for the trivalent metal perchlorates Fe(ClO,),, Al(ClO,),, Ga(ClO,), and In(ClO,), in DMSO. It has been found that in DMSO solutions the perchlorates contain the metal cations in the form of octahedral hexasolvo-complexes whose association with the non-coordinating anions varies within relatively narrow limits as the nature of the 10371038 COMPLEX FORMATION IN DIMETHYL SULPHOXIDE central metal atom changes.l2t l3 Therefore we expect that equimolar mixtures of these metal perchlorates would play the role of effectively constant-ionic-strength media in studies on the formation of their weak inorganic complexes.In the present paper we present results for the Fe'II-SCN-DMSO and A P - SCN-DMSO systems, for which the stability constants of the monothiocyanate complex-forming reaction M(DMSO);+ + SCN- e MSCN(DMSO);+ (1) have been determined. In order to obtain the thermodynamic parameters for the systems under investigation, the heats of complex formation for reaction (1) have been measured, and when combined with the standard free energies of complex formation they have enabled us to determine the standard entropy changes. EXPERIMENTAL Dimethyl sulphoxide, reagent grade, was dried over CaH, for several days and subsequently distilled through a column under reduced pressure. The product was further purified by repeated fractional crystallization in a vacuum rotary evaporator by a freezing process carried out under anhydrous conditions.The conductivity of the final product varied from 2 x to 3 x S cm-l. The DMSO-solvated metal perchlorates Fe(ClO,), - 6DMSO and A1(C104), .6DMSO were obtained from the hydrated metal perchlorates by repeated crystallizations from anhydrous DMSO. Reagent-grade potassium thiocyanate was recrystallized twice from anhydrous methanol and dried in vacuo at 60 "C. The stock solutions of Al(ClO,), and Fe(C104), were analysed for the metals using the 8-hydroxyquinoline gravimetric method. At least five determinations were performed in each case, the relative error not exceeding 0.1 % .The stock solution of KSCN was prepared by weight from the dried salt and the solvent. Solutions for spectrophotometric and calorimetric work were prepared by weight from the respective stock solutions. Final concentrations of the solutions were calculated taking into account the densities, which were determined independently. The preparation of the solutions and further manipulations were performed in a dry-box. Absorption spectra were determined using a Zeiss VSU-2P spectrophotorneter with a thermostatted cell compartment. The heats of formation of FeSCN2+ and A1SCN2+ solvo- complexes and the heats of dissolution of KSCN were determined using a reaction calorimeter described in ref.(14) and (15). In order to improve the experimental accuracy, the KSCN was added as a solution. RESULTS AND DISCUSSION EVALUATION OF THE SPECTROPHOTOMETRIC RESULTS FOR Fe(ClO,),-KSCN-DMSO The method requires a reliable value of the stability constant Fe/ll of the FeSCN2+ solvo-complex in the given medium. The results reported in the 1iteraturel6-l9 are not precise enough for our purpose, and thus the Fe(ClO,),-KSCN-DMSO system was studied spectrophotometrically in order to determine the thermodynamic stability constant of the FeSCN2+ solvo-complex. Fig. 1 shows the spectra of a series of solutions containing Fe(CIO,), at an approximately constant concentration of 0.12 mmol dm-, and KSCN at variable concentrations up to 1 : 9 mole ratio of thio- cyanate anion to ferric cation (curve 18).Included is a part of the c.t. band of the solvo-complex Fe(DMS0)3,+ with a maximum at 340 nm (curve 1). As is seen, addition of KSCN to a Fe(ClO,), solution results in an increase in the intensity of the c.t. band and the simultaneous appearance of a new band characteristic of the octahedralD. PUCHALSKA AND D. WOJCIK 1039 3500 3000 2500 2000 5 1500 1000 500 350 4 00 45 0 500 550 A/nm Fig. 1. Absorption spectra of Fe(ClO,), + KSCN solutions in DMSO at 25 "C. Concentrations (in mmol drnp3) of Fe(C10,), and KSCN are, respectively: 1, 0.123 and 0; 2, 0.135 and 0.006; 3, 0.131 and 0.009; 4, 0.123 and 0.020; 5, 0.121 and 0.041; 6, 0.134 and 0.059; 7, 0.123 and 0.071; 8, 0.121 and 0.086; 9, 0.1200 and 0.118; 10, 0.122 and 0.110; 11, 0.119 and 0.150; 12, 0.118 and 0.177; 13, 0.117 and 0.236; 14, 0.114 and 0.324; 15, 0.111 and 0.437; 16, 0.100 and 0.597; 17, 0.101 and 0.713; 18, 0.102 and 0.947.FeSCN(DMSO)i+ complex with a maximum at ca. 450nm. The linearity of the graphic test of Coleman et aL20 for the coexistence of two complexes in the series of solutions under investigation (performed at 410, 430, 440 and 470 nm) suggests that even at the very high KSCN concentrations the amount of 'higher' thiocyanate com- plexes appears to be negligible.1040 COMPLEX FORMATION IN DIMETHYL SULPHOXIDE 4000 3000 '1 2000 I I I I I I I I 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 C/mol dm-3 1000 Fig. 2. Dependence of the molar absorption coefficients of the FeSCN(DMSO)E+ complex on the concentration of Fe(ClO,), in DMSO solutions at 25 "C.The KSCN concentration is constant at 0.065 mmol dm-3. Evaluation of the spectrophotometric results in terms of equilibrium constants requires values for the molar absorption coefficients, E ~ , of the monothiocyanate complex. el is found from the absorption spectra of the series of solutions of constant KCSN concentration, while the concentration of Fe(C10,), was varied within the broad concentration range. It seems almost certain that, under these conditions, the observed spectral changes are due to the formation of the FeSCN2+ solvo-complex only, and at the high Fe(ClO,), concentrations all thiocyanate anions are involved in complex formation. On the assumption that only the Fe3+ and FeSCN2+ solvo- complexes are formed in the investigated solutions, the molar absorption coefficients el have been estimated taking into account the known molar absorption coefficient E , of the Fe3+ solvo-complex.The results are shown in fig. 2. As is seen, el remains almost constant within a broad range of [Fe(ClO,),]. We thus conclude that for selected wavelengths the molar absorption coefficients are: e!50 = 3917, e!60 = 3787 and etS0 = 3265. On the assumption that in the solutions under consideration only equilibrium (1) is set up, an attempt was made to derive the thermodynamic stability constant g from the spectrophotometric data. Thus the separate spectrophotometric investigations were carried out on a series of Fe(C10,),-KSCN-DMSO solutions, where the ratio of the thiocyanate anion to ferric cation did not exceed 1 .Accordingly is defined as where C, and C, are equilibrium concentrations of Fe3+ and FeSCN2+ solvo-complexes, & is a quotient of the activity coefficients and P1 is the equilibrium concentration quotient of reaction (1). Assuming that the Debye-Hiickel law will give a reasonableD. PUCHALSKA AND D. W~JCIK 1041 Table 1. Equilibrium concentration quotients Fe81 of the FeSCN(DMSO)t+ complex in DMSO at 25 "C 81 'Fe(C10& cx /mmol dm-3 /mmol dm-3 A = 450 nm A = 460 nm II = 480 nm ~ ~~ 0.050 72 0.053 80 0.062 56 0.092 25 0.117 60 0.123 00 0.130 52 0.134 66 0.136 04 0.142 58 0.150 01 0.172 83 0.218 90 0.234 50 0.278 58 0.308 70 0.352 61 0.376 50 0.407 59 0.458 90 0.461 33 0.565 54 0.673 80 0.779 01 0.062 20 0.009 97 0.063 15 0.064 64 0.032 79 0.064 64 0.008 81 0.006 44 0.124 87 0.007 34 0.021 09 0.017 22 0.034 30 0.018 19 0.037 61 0.063 80 0.041 96 0.058 90 0.263 99 0.01 1 60 0.051 55 0.062 35 0.090 95 0.091 30 9 840 9 163 10 174 9 722 8 780 9 546 9 302 8 521 7 128 5 640 7 999 7 259 8 228 7 974 8 071 9 091 8 082 7 410 6 447 6 008 7 805 6 818 5 153 5 626 10 168 9 208 10 415 10 044 9 003 9 708 9 883 8 850 7 575 7 821 8 354 7 812 7 948 8 292 8 661 9 334 8 565 7 869 6 783 5 981 8 305 9 480 5 793 7 352 9 157 9 028 8 632 7 567 8 044 8 922 9 369 6 410 7 159 7 051 7 163 6 486 7 193 6 943 7 098 6 599 6 037 5 128 5 635 6 100 5 504 4 308 4 988 - approximation to ionic activity coefficients atthese low concentrations, the corrected thermodynamic equilibrium constant should be Values of 1.1 141 and 0.4267 have been assumed for the constants A and B, respectively, for DMSO at 25 "C.The ion-size parameter ii has been estimated as 6.44 8, on the basis of the solvodynamic radii of the hexasolvated ferric cation and the un- solvated thiocyanate anion.l39 21 The equilibrium concentration C, of the FeSCN2+ solvo-complex were calculated from the measured absorption at 450,460 and 480 nm as where eo and e, are the molar absorption coefficients of the Fe3+ and FeSCN2+ solvo-complexes, respectively, E is the measured mean molar absorption coefficient of FeIII and C, is the total concentration of metal. The equilibrium concentration Co and [XI were derived from the material balance. The ionic strength for investigated systems was estimated as I = 6CM-3C,+Cx, where Cx is the total concentration1042 COMPLEX FORMATION IN DIMETHYL SULPHOXIDE I I I I I I 0.01 0.02 0.03 0.04 0.05 0.06 41 1 +&/I Fig.3. Dependence of the equilibrium concentration quotients ""Dl of the FeSCN(DMS0);' complex, derived at 450 nm, on the ionic strength of the solution. 1000 900 800 7 00 600 500 4 00 3 00 T 200 I 1 I I I I I 400 420 440 460 480 500 520 h/nm Fig. 4. Absorption spectra of the Fe(C10,), + Al(ClO,), + KSCN equimolar mixtures in DMSO solution at 25 "C. The KSCN concentration is constant at 0.163 mmol dm-3. The concentrations (in mmol drn-,) of Fe(C10,), and Al(ClO,), are, respectively: 1, 1.036 and 2.359; 2, 0.592 and 2.802; 3, 0.356 and 3.034; 4, 0.174 and 3,216. e is the mean molar absorption coefficient of Fe"'.D. PUCHALSKA AND D.WOJCIK 1043 of KSCN. Values of the equilibrium concentration quotient Fe/ll, derived at three wavelengths, 450, 460 and 480 nm, are listed in table 1. Fig. 3 shows a plot of the resulting values of log Fe/31 against dI/(l + B i d 0 determined at 450 nm. The Debye-Hiickel slope ( 6 4 is marked by the dashed line. The thermodynamic stability constants, FeE, at these wavelengths were found by linear least-squares extrapolation of log FeP1 against d I / ( 1 + B i d 0 plots to infinite dilution. Thus, the average value of Feg was estimated to be 11 700 f 1200 (least-squares standard deviation). This value is in good agreement with the value obtained by Wada and Yoshizawa16 although there is some disagreement over the molar absorption coefficient E , of the FeSCN(DMSO)t+ complex.At 460 nm, Wada and Yoshizawa16 report a value of 3160, Devia and Watts" report a value of 2970 and in this work we estimate a value of 3787. EVALUATION OF THE SPECTROPHOTOMETRIC RESULTS FOR Fe( ClO,),-Al( ClO,),-KSCN-DM SO MIXTURES In this work the stability constants of the AlSCN2+ solvo-complexes were determined using a series of solutions of two metal perchlorates Fe(CIO,), and Al(ClO,), with constant total concentration C,, to which a small amount of the complexing thiocyanate anion was added (C, 4 CM), so that it had a negligible effect on the activity coefficients. The spectral effects for one series of equimolar solutions are shown in fig. 4. Substitution of Al(C10,), for Fe(ClO,), in equimolar mixtures results in a marked increase in the mean molar absorption coefficient of FeIII, E.This indicates that changes in composition bring about an increase in the FeSCN2+ concentration. Thus, aluminium(II1) forms weaker thiocyanate complexes than iron(rI1). The equilibrium concentration of' free' ligand, which is a function of the composition of the investigated mixture, was determined from the spectrophotometrically measured equilibrium concentration of the FeSCN2+ solvo-complex. Evaluation of the spectro- photometric results uses the following relations : Taking into account the material balance for the anion and both metals, the 'medium' stability constant of the AlSCN2+ complex has been estimated as c, - [XI - cp *lP1 = (C*] - c, + [XI + Cp) [XI where C, and CAI are analytical concentrations of KSCN and Al(ClO,),, respectively.The equilibrium concentrations of the FeSCN2+ solvo-complex, Cp, were evaluated for each solution from the measured absorption. The equilibrium concentration of free ligand w] was derived as C p = Fy?,(C,, - cp) (7) where FePl is the 'medium' stability constant of the FeSCN2+ solvo-complex. The above relations are valid assuming that all the activity coefficients remain constant, while the relative amounts of the two metal perchlorates are varied. The Debye- Hiickel equation was used in the calculations. The ion size parameter, 6, used for the AlSCN2+ solvo-complex was the same as for FeSCN2+, as a result of the identical values of the limiting ionic conductivities of Fe(DMS0),3+ and A1(DMS0),3+.l3 In the1044 COMPLEX FORMATION IN DIMETHYL SULPHOXIDE Table 2.' Medium' stability constants of the FeSCN2+ and A1SCN2+ complexes in equimolar mixtures of both metal perchlorates of total concentration C , in DMSO at 25 "C C M FeD, /mol dm-3 Fepl "'81 "'81 0.000 61 0.001 10 0.001 93 0.002 06 0.003 27 0.003 39 0.008 04 0.008 99 0.014 21 5 280 4 218 3 259 3 159 2 460 2 422 1419 1321 968 232 182 122 143 96 I08 51 53 39 22.8 23.2 26.7 22.1 25.6 22.4 27.8 24.9 24.8 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 dI 1 Fig. 5. Dependence of the 'medium' stability constants "'8, of the AlSCN(DMSO);+ complex in DMSO solutions on the ionic strength of equimolar mixtures of Al"' and Fe"' perchlora tes. first approximation it was assumed that all the investigated electrolytes are completely dissociated and that the ionic strength was taken as I = 6C+ Cx, were C is the analytical concentration of both metals.In successive approximations the concentra- tions of the monothiocyanate complexes of AlI" and FeIrl were included in calculations of ionic strength. Using the above approach, convergent values of the 'medium' stability constants of the A1SCN2+ solvo-complex were obtained. The stability constants were derived by assuming that only MSCN2+-type solvo- complexes are formed, which is true under the conditions of a large excess of metal perchlorate. The average value of *I&, derived at 450nm, for a series of equimolar mixtures (each series consisting of at least four mixtures) are listed in table 2. The most important feature of table 2 is that the variation of t.he activity coefficients with theD.PUCHALSKA AND D. WOJCIK 1045 Table 3. Heats of complex formation of iron(II1) and aluminium(II1) cations as a function of the thiocyanate anion concentration in DMSO solutions at 25 "C FeSCN(DMSO);+ CFe/ 1 OP2 mol dm-3 0.928 0.927 0.927 0.925 0.925 C,/ 1 0-2 mol dm-3 0.1 14 0.242 0.460 0.631 0.705 AH/kJ mol-l 5.05 4.98 5.01 5.20 4.97 AlSCN(DMS0);' CAl/ mol dmP3 1.160 1.116 1.127 1.155 1.111 CX/1Op2 mol dm-3 0.117 0.194 0.282 0.722 1.178 AH/kJ mol-1 15.32 15.40 15.77 15.53 15.09 Table 4. Standard thermodynamic parameters for the FeSCN(DMSO);+ and AlSCN(DMSO);+ complexes AGe AHea A S e equilibrium /kJ mol-1 /kJ mol-1 /J K-l mol-I Fe3++ SCN-$ FeSCN2+ -23.2k0.5 5.04f0.04 95+ 1.8 Al3++SCN-+AlSCN2+ -15.2f1.7 15.42f0.11 103f6 a Average heats of complex formation.total concentration of the metal perchlorates is almost the same for both FeSCN2+ and A1SCN2+ solvo-complexes. This is because the ratios of their derived stability constants remain approximately constant when the total concentrations of the metal perchlorates, forming the ionic media, are varied. Thus the equimolar mixtures of FeIII and AllI1 perchlorates provide effectively constant-ionic-strength media for studies of complex formation. Fig. 5 shows the dependence of the stability constants of the AlSCN2+ solvo-complex on the ionic strength of the solutions. This variation is approximately the same as that arising from the Debye-Hiickel equation (the Debye-Huckel slope is marked by the dashed line). Thus, the stability constant of the A1SCN2+ solvo-complex at zero ionic strength is 460 150.CALORIMETRIC RESULTS In order to determine the heats of reaction (l), the effective heats of solutions containing the metal perchlorates (constant concentration) and potassium thiocyanate (variable concentration) were measured. The concentrations of the metal perchlorates were high in relation to that of KSCN. The excess of metal perchlorate eliminates the formation of ' higher' thiocyanate complexes. In order to improve the experimental accuracy, KSCN was added in the form of a solution. The molar heat of dilution of the KSCN solution from the initial to final volume determined in separate experiments was 1.85 kJ mol-1 for the investigated concentration range. The heats of complex formation were then calculated, taking into account the heats of dilution of the KSCN solutions.The equilibrium concentrations of the FeSCN2+ solvo-complex were calculated using the equilibrium constants determined spectrophotometrically, and for the A1SCN2+ solvo-complex from the straight line dependence presented in fig. 5. The results are listed in table 3. Inspection of table 3 shows that the heats of the1046 COMPLEX FORMATION IN DIMETHYL SULPHOXIDE complex-forming reaction of the FeSCN2+ and A1SCN2+ solvo-complexes are constant for the investigated concentration range of KSCN. The derived thermodynamic parameters for the FeSCN2+ and AlSCN2+ solvo- complexes at zero ionic strength are listed in table 4, from which it can be seen that the entropies of the complex-forming reactions are approximately the same within experimental error for both metals.It is probable that the main contributions to these high positive entropies correspond to release of solvent molecules during the complex-forming reactions. The close similarity of the entropies is an additional support of the assumption, mentioned earlier, concerning the coordination states of the perchlorates in DMSO solutions. The cation-solvent interactions are the same irrespective of the nature of the metal. Moreover, the complexes formed in the system under investigation have the same structure despite their differences in stabilities. We thank Dr R. Pastewski for helpful discussions. L. Johansson, Coord. Chem. Rev., 1974, 12, 241. 0. Benali-Baitich and E. I. Wending, J. Inorg. Nuci. Chem., 1975, 1219. W. LibuS, Muter. Sci., 1979, 5, 85. W. LibuS, XZZZ Znt. Con5 Coordination Chemistry (PWN, Warsaw, 1974), p. 241. W. LibuS, R. Pastewski and T. Sadowska, J. Chem. SOC., Faraday Trans. 1, 1982, 78, 377. Z. LibuS, J. Phys. Chem., 1970, 74, 947. ' Z. LibuS, Inorg. Chem., 1973, 12, 2972. * Z. LibuS and H. Tialowska, J. Solution Chem., 1975, 4, 101 1. Z. LibuS and G. Kowalewska, Pol. J . Chem., 1978, 52, 709. lo Z. LibuS and W. Maciejewski, Rocz. Chem., 1976, 50, 166. l1 W. LibuS and H. Strzelecki, Electrochim. Acta, 1972, 17, 577. l2 M. Pilarczyk and W. LibuS, Bull. Acad. Pol. Sci., Ser. Sci. Chim., 1974, 8, 717. l 3 W. LibuS, B. Chachulski, W. Grzybkowski, M. Pilarczyk and D. Puchalska, J. Solution Chem., 1981, l4 M. Mqcik and W. LibuS, Bull. Acad, Pol. Sci., Ser. Sci. Chim., 1976, 24, 477. l5 W. LibuS, M. Mqcik and W. Sulek, J. Solution Chem., 1977, 6, 865. l6 G. Wada and N. Yoshizawa, Bull. Chem. SOC. Jpn, 1971,44, 1018. D. H. Devia and D. W. Watts, Znorg. Chim. Acta, 1973, 4, 697. C. H. Langford and F. M. Chung, J. Am. Chem. SOC., 1968,90,4485. l9 B. Csiszar, V. Gutmann and E. Wychera, Monatsch. Chem., 1967, 98, 12. 2o J. S. Coleman, L. P. Varga and S. H. Mastin, Znorg. Chem., 1970, 9, 1015. 21 N-P. Yao and D. N. Bennion, J. Electrochem. SOC., 1971, 118, 1097. 10, 1. (PAPER 4/ 1 175)

ISSN:0300-9599    

DOI:10.1039/F19858101037

出版商:RSC    

年代:1985

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I, 1985, 81, 1047-1056 Binding and Decomposition of Oxovanadium(1v) Phthalocyanine, Tetraphenylporphyrin and Etioporphyrin on Hydrotreating Catalysts Studied by X-Ray Photoelectron and Ultraviolet-Visible Spectroscopies Relevance to Catalytic Demetallisation BY ~ I L I P C. H. MITCHELL* AND CARLOS E. SCOTT Department of Chemistry, The University, Whiteknights, Reading RG6 2AD AND JEAN-PIERRE BONNELLE AND JEAN G. GRIMBLOT Laboratoire de Catalyse Heterogene et Homogene, UA CNRS 402, Universite des Sciences et Techniques de Lille, 59655 Villeneuve d’Ascq Cedex, France Received 10th July, 1984 The interactions of oxovanadium(1v) phthalocyanine, VOPc, oxovanadium(1v) tetraphenyl- porphyrin, VOTPP, and oxovanadium(1v) etioporphyrin, VOETP, with alumina, silica-alum- ina, fluorided silica, Co,O,, MOO,, MoS,, MoO,/Al,O, and Co-Mo/Al,O, catalysts in the oxide and sulphide forms have been studied by X-ray photoelectron spectroscopy and u.v.-visible diffuse reflectance spectroscopy. Partial demetallisation of VOTPP and VOPc, but not VOETP, occurs on acidic (H+) supports.The V02+ thereby released is reduced in the X.P.S. spectrometer; porphyrin-bound vanadium is not reduced. Changes of core-electron binding energies of vanadium and nitrogen caused by interaction of the porphyrins with the supports is small, and there is no evidence of a strong interaction with the surface through vanadium or nitrogen. It is therefore suggested that the porphyrins are bound to the surfaces by a donor-acceptor, or charge-transfer, interaction through the IT system of the porphyrin ring and Brernsted- and/or Lewis-acid sites of the surfaces.This paper is one of a series about the binding of vanadium and nickel porphyrins and related compounds to catalyst surfaces. Our interest derives from problems caused by these compounds in the hydroprocessing of petroleum crudes. Some crude oils have a high metal content; for example the Boscan crude oil of Venezuela contains 0.02 mol dmP3 vanadium.2 Metal compounds also become concentrated in the heavy residues which remain after distillation of lighter petroleum fractions. They deactivate hydroprocessing catalysts, for example the Co-Mo/A1203 hydrodesulphurisation catalyst and the Si0,-A120, and zeolite hydrocracking catalysts, by binding to active sites and by blocking pores.The metal compounds in crude oils and residua, and also in oil shales, are vanadium and nickel p~rphyrins.~ A recent EXAFS has shown that vanadyl porphyrins are the major metal compounds in Boscan and Cabimas asphaltenes. The X.P.S. spectra of the asphaltenes led to the same conclusion: the vanadium photopeak corresponded to oxovanadium(1v) bound to four nitrogen^.^ The object of our work was to discover which atoms of the metalloporphyrins bind to the catalyst and to which surface sites. We used oxovanadium(rv) etioporphyrin because its structure is similar to the structure of the metalloporphyrins in petrole~m.~ 10471048 X.P.S. AND U.V.-VIS. STUDY OF v PORPHYRINS ON CATALYSTS We also used oxovanadiurn(1v) phthalocyanine and tetraphenylporphyrin in order to study the effect on properties and reactivity of changes in the porphyrin ring.We wished to observe any changes in the X.P.S. spectra of the vanadium compounds and/or the catalyst when the vanadium compounds were attached to the surface, and also to give an example of how X.P.S. can be used to study the deposition of vanadium on catalysts. EXPERIMENTAL MATERIALS Oxovanadium(rv) phthalocyanine (VOPc) and metal-free phthalocyanine (H,Pc) were purchased from K and K Laboratories Ltd. Oxovanadium(1v) tetraphenylporphyrin (VOTPP), was synthesised by the method of Adler et af. It was spectroscopically and analytically pure (found: C, 77.2; H, 4.20; N, 8.18% ; calc. for C,,H,,N,OV: C, 77.8; H, 4.15; N, 8.24%). Oxovanadium(1v) etioporphyrin (VOETP) was synthesised by the method of Erdman et ~ 1 .~ and was similarly purified (found: C, 70.1 ; H, 6.66; N, 10.2% ; calc. for C,,H,,N,OV: C, 70.7; H, 6.68; N, 10.3%). V,O, and cO@, were spectroscopically pure grade from Alpha. MOO, was pure sublimed oxide from Climax Molybdenum Co. MoS, and oxovanadium(1v) sulphate, VOSO, 5H,O, were from B.D.H. Oxovanadium(1v) acetylacetonate was prepared according to ref. (6). y-Al,O, (Ketjen; pore volume 1.00 mm3 g-l, surface area 259 m2 g-l), Co0-Mo03/A1,0, (COO, 4.1 % and MOO,, 11.6% ; pore volume 0.64 mm3 g-l, surface area 268 m2 g-l) and Ni0-MoO,/Al,O, (NiO, 3.33% and MOO,, 15.7% ; pore volume 0.52 mm3 g-l, surface area 276 m2 g-l) were kindly given by Akzo Chemie, Amsterdam. Silica (pore volume 1.2 mm3 g-l, surface area 300 m2 g-l) was from Davison. Fluorided silica was prepared as f01lows:~ Aqueous NH,F solution (10 cm3, 12%) was added to silica (10 g).The suspension was left to stand for 2 h, filtered, air-dried and calcined in an airflow at 973 K for 5 h. Catalysts and supports were ground to < 60 mesh for the adsorption experiments. Sulphided catalysts were prepared by a standard procedure (drying in nitrogen at 200 "C followed by H,S + H, in the ratio 1 : 10 for 0.5 h at 250 "C and 4 h at 300 "C). This procedure does not completely sulphide the Co and Mo oxides, as shown by the X.P.S. experiments (see below). The sulphided catalysts were handled under dry nitrogen. IMPREGNATION WITH VANADIUM COMPOUNDS SOXHLET EXTRACTION METHOD Our procedure was similar to that used by Boucher et af.for preparing supported phthalocyanine catalysts.8 The catalyst and dichloromethane were placed in the flask of a Soxhlet extraction apparatus and the vanadium phthalocyanine or porphyrin was placed in the extraction thimble. Dichloromethane condensed in the extraction thimble and dissolved some of the vanadium compound; this solution was then transferred to the flask containing the support. The concentration of the solution was limited by the low solubility of the vanadium compounds. We were unable to obtain samples with more than 0.1 % vanadium, which is near the limit of detection using X.P.S. ROTARY EVAPORATION IN VACLIO We used saturated solutions of vanadium phthalocyanine and porphyrins in CH,Cl, and sufficient volumes (ca. 1 dm-3 for 5 g catalyst) to provide 0.5 wt % vanadium in the catalysts.The solvent was removed in vucuo using a rotary evaporator. X.P.S. MEASUREMENTS X.p. spectra were recorded at Lille using two instruments: AEI ES 200B and Leybold LH-S10 spectrometers, both with A1 anodes (Av = 1486.6 eV, 300 W). The working pressures were ca. Torr for the Leybold instrument.? The powdered Torr for the AEI instrument and t 1 Torr = 101 325/760 Pa.W wl Table 1. Binding energiesa of carbon, nitrogenb and vanadiumb from X.P.S. binding energy, b.e./eV (and f.w. h.m. /eV) H,Pc VOPC VOTPP VOETP supportC c 1s N 1s c 1s N 1s V2p3,, c 1s N 1s V2p3,, C 1s N IS V 2p3/2 ~~ none A1203 MOO, - MoS, - Co-Mo/Al,O, oxide - part sulphided - 399.2 401 .O 399.8 399.0 (1.6) (1.6) (3.5)’ (1.6) 285.0 (1-8) d d 9 516.5 514.2 (1.9) (1.9) 285.0 285.0 285.0 285.0 ( 2 W (2-3) (2.4) (1.6) 285.0 (2.7) 399.8 399.0 (3.5)’ (1.6) 9 9 d 5 16.4 514.0 5 16.4 (2.3) (2.8) (1.9) 9 514.3 (1.9) 398.7 516.6 (1.5) (1-9) - (5 1 6.O)e - d a Relative to C 1s = 285.0eV; numbers in parentheses are full widths at half maximum (f.w.h.m.). For a Boscan aspheltene, V 2p3,, = 516.0eV V peak covered by a strong oxygen N 1s and N 1s = 398.6 and 400 eV. satellite peak. overlaps with 3p,,, of Mo in support. Concentrations of compounds on the supports corresponded to 0.5 wt % V. Weak shoulder. f The large f.w.h.m. for the compounds on A120, is due to a charging effect; for A1 2p f.w.h.m. = 2.8 eV. -! > w - 0 P \o1050 X.P.S. AND U.V.-VIS. STUDY OF V PORPHYRINS ON CATALYSTS I 1 514.2 \ ! \ 514.3 i16.4 516.5 h/\ / 4 00 395 binding energy/eV 5 20 510 binding energy/eV Fig.1. For legend see opposite.P. c . H. MITCHELL, c . E. SCOTT, J-P. BONNELLE AND J. G. GRIMBLOT 1051 samples were pressed on indium foil. The C 1s peak from adventitious grease was taken as reference at 285 eV when calculating binding energies (E,,). For Al,O, and Al,O,-based catalysts we were able to correlate C 1s and A1 2p binding energies. For a C 1s binding energy of 285 eV, the A1 binding energies were consistently 74.8 f 0.2 eV. Thus the precision of the binding energies, based on C Is, is ca. kO.2 eV. Peak areas were obtained by planimetry after linear background subtraction. For the supported vanadium compounds the spectra were accumulated for ca. 16 h using the AEI instrument.ULTRAVIOLET-VISIBLE MEASUREMENTS The spectra of the compounds in solution were recorded using a Unicam SP 1800 spectrophotometer and the reflectance spectra of powdered solids relative to a magnesium oxide standard were recorded using a Beckmann Acta M IV spectrophotometer. RESULTS AND DISCUSSION X-RAY PHOTOELECTRON SPECTROSCOPY In the X.P.S. work we had two objectives: (a) to assess the usefulness of the techniques for studying vanadium deposited on catalysts (namely, the nature of the species, the possibility of distinguishing different species and changes of binding energy caused by interactions of the porphyrins with the catalysts) and (b) to interpret shifts in binding energies in terms of the nature of the interaction of the porphyrins with the surfaces.At this stage of the work we have worked with just one concentration of vanadium on the catalyst, 0.5 wt %, and our interest has been centred on binding-energy shifts. Binding energies from X.P.S. are listed in table 1 and selected spectra are shown in fig. 1. It is evident from table 1 that we shall sometimes be discussing quite small changes in binding energy and it is necessary therefore to establish the precision of our measurements. We start by discussing the calibration of the two spectrometers and then interpret the binding energy shifts, in particular the large decrease of binding energy sometimes observed for vanadium. Our general conclusions are that acidic supports decompose VOTPP at least partly and that vanadium released from the porphyrin is reduced in the spectrometer, and that VOEPP, which is more stable to acid and is not reduced on acidic supports, does not decompose.CALIBRATION OF THE SPECTROMETER For the C 1s peak we used that from grease contaminant in the spectrometer and assumed a binding energy of 285 eV. For the alumina-supported materials the C 1s and A1 2p binding energies were consistent, but for pure compounds two problems Fig. 1. Spectra from X.P.S. study. Intensities (counts s-l) plotted against binding energies (ev). (A) C 1s spectra: (a) H,Pc, (b) VOPc, (c) VOTPP and ( d ) VOETP (on the high-energy side we see a small satellite peak labelled sat). (B) N 1s spectra: (a) H,Pc, (b) VOPc, (c) VOPc/Al,O,, after accumulation, ( d ) VOPc/Al,O,, normal scan, not accumulated [the intensity scale is not the same as (c)], (e) VOETP, cf) VOTPP and ( g ) VOTPP/Co,O,.(C) V 2p,,, and 0 1s satellite spectra (the 0 1s peak excited by the satellite A1 K , , , radiation is on the high-energy side and is prominent for high oxygen samples such as oxides; the 0 1s peak excited by A1 Ka5.6 is on the low-energy side, as is the Mo 3s peak excited by A1 Kal.,, which is prominent for high Mo samples, MOO, and MoS,): (a) VOTPP, (6) VOPc, (c) VOETP, ( d ) VOPc/MoS,, (e) Mo/A1,0,, part sulphided, ( j ) VOPc/Co-Mo/Al,O,, oxide form and ( k ) Co0-Mo03/A1,0,, VOETP/MoS,, cf) VOTPP/MoS,, ( g ) VOTPP/Co,O,, (h) VOTPP/MoO,, (i) VOTPP/Co- not containing V, showing the 0 1s K,,., and Ka5 background. 35-21052 X.P.S. AND U.V.-VIS. STUDY OF v PORPHYRINS ON CATALYSTS arose.First, the amount of carbon contaminant deposited on pure compounds was less than the amount deposited on supported substances because the spectral accumulation was less for the pure compounds and consequently the carbon peak was weaker. Secondly, for a compound containing carbon with a binding energy different from that of contaminant carbon a problem of assignment arises. We cannot simply assign the strongest C 1s peak to contaminant carbon. For example, for VOPc and H, Pc we see two carbon peaks with intensity ratio 1/3, equal to the atomic ratio of the two types of carbon in these molecules, i.e. carbon bound to nitrogen and carbon of the aromatic ring. Therefore, there was little carbon contamination. In agreement with this, the carbon intensities were unchanged after the samples had been in the spectrometer for 3 h.Thus we assign the more intense peak to aromatic carbon with a binding energy of 285 eV, the same as the contaminant carbon. We thus obtain binding energies for N 1s in agreement with those reported for H,Pcg and VOPC;~~ the values for VOTPP are similar to those reported for N 1s using other metal tetraphenylporphyrins. lo EFFECT OF CONTAMINANT CARBON IN THE SPECTROMETER We wished to discover if the contaminant had any effect on the catalysts or absorbed compounds so we compared results from the AEI and the Leybold spectrometers, the carbon contamination being less in the Leybold spectrometer. The following materials had similar binding energies on the two instruments (values in parentheses are the largest discrepancies in binding energies): VOPc/MoS, (0.2 eV), VOPc (0.6 eV) and MoS, (0.8 eV).For these three samples the mean deviation for each binding energy was f0.2 eV. We thus conclude that the two instruments give very similar results. Note, however, that we were not able to compare spectra of the oxide-supported substances because spectral accumulation was performed using the AEI instrument only. There is no doubt that during spectral accumulation contamination with carbon increases. For VOPc on CoMo/Al,O, the C Is intensity (in arbitrary units) increased from an initial value of 800 to 950 after overnight accumulation and for CoMo/Al,O, the intensity increased from 530 to 700. On the evidence we have so far, it seems that contaminant carbon does not affect the measurement of the binding energies.ASSIGNMENTS Assignments and binding energies are given in table 1. Problems arise because of overlapping peaks. On molybdenum-containing catalysts we cannot assign N 1s because it is coincident with Mo 3p3,,. Similarly S 2s and Mo 3d are coincident. V 2p3,2 in VOETP on oxide catalysts is hidden by a satellite of 0 1s (but can be seen in pure VO compounds where the 0 Is peak is less intense). We can assign V 2p,,, in supported VOPc and VOTPP because the shift of the peak resulting from reduction of vanadium (see below) is sufficient for the V and 0 peaks to be resolved. Typical spectra which illustrate these points are shown in fig. 1. The compounds H,Pc, VOPc, VOTPP and VOETP have similar spectra, as we would expect from their similar structures.The N/V intensity ratio was, as expected, greater for VOPc (8) than for VOETP and VOTPP (4). BINDING-ENERGY SHIFTS The major change is the decrease of vanadium binding energy on certain supports; this is futher discussed below. There are also slight changes in the N 1s binding energies. For H,Pc (see table 1) and for other metal-free porphyrins there are two N 1s peaks at ca. 399 and 401 eV. These correspond to unprotonated and protonated nitrogens. For VOTPP we see only one N 1s peak, as we would expect, since the four nitrogens are equivalent. For VOPc we see two N 1s peaks since we still have two typesP. c. H. MITCHELL, c . E. SCOTT, J-P. BONNELLE AND J. G. GRIMBLOT 1053 of nitrogen, the four nitrogens bound to V and the four meso nitrogens.Thus X.P.S. can distinguish between metal-bound and metal-free heterocyclic nitrogen. That we see two N 1s peaks for Boscan asphaltene (see table 1) tells us that there are two types of nitrogen, one metal-bound and one metal-free. This conclusion agrees with the analysis which gives an N/V ratio of ca. 8. When H,TPP is protonated to H,TPP2+ the lower-energy 1s peak disappears and the higher-energy peak increases in intensity, a consequence of the nitrogens being more p0sitive.l' A similar effect, a general increase of N 1s binding energy, is observed when H,TPP is absorbed by montmorillonite. We were interested in whether such an effect occurs when the metal porphyrins are adsorbed by the supports used in this work. An increase of N 1s binding energy would indicate an increase of positive charge through interaction with the surface, i.e.charge transfer from the metalloporphyrin to the surface (as suggested by electron spin resonance data).', There is some indication from table 1 that N 1s binding energies, where we were able to observe them, are higher for the compounds supported on oxides, e.g. VOPc at 399eV and VOPc/Al,O, at 399.8 eV, but the shifts were small. We conclude the X.P.S. will certainly tell us whether in a particular material we have bound and free nitrogen, as with the Boscan asphaltene, but that subtle charge-transfer interactions with surfaces are probably too small to be resolved by X.P.S. The spectra of VOPc and VOTPP change significantly when the compounds are supported : the V binding energy for VOPc and VOTPP on the CoMo/Al,O, catalyst is 2.5 eV less than for the unsupported compounds.This decrease is unlikely to be due to some experimental artefact or charging effect since the Mo, Co and A1 spectra are the same for the CoMo/Al,O, catalyst and for the catalysts doped with VOPc and VOTPP. The size of the decrease of the vanadium binding energy (2.5 eV) suggests that the Vtv has been reduced in the spectrometer. For example, the binding energy for vanadium in VItl trisacetylacetonate is 514.2 eV,I3 which is close to our value for VOPc and VOTPP on CoMo/Al,O,. For VOETP on CoMo/Al,O, there was no signal near 514 eV and so reduction of VOETP has apparently not occurred, unlike for VOTPP on the same catalyst. It is possible that some vanadium remains as vanadium(1v) but we were unable to see it by X.P.S.because of the overlapping oxygen 1s satellite peak at 516-517 eV. On MoS,, VOTPP and VOPc have the same vanadium binding energies as the: pure compounds; vanadium(1v) is not reduced in the spectrometer. The core-electron binding energies of Mo and S were not changed by the vanadium porphyrin but the binding energy of the lowest valence band decreased by 0.8 eV. Our X.P.S. results reveal differences in the chemistry of VOETP and VOTPP on the supports. Such differences need to be recognised in studies of demetallisation. VOETP, which is structurally similar to petroporphyrin~,~ is acceptable in model studies but VOTPP is not. We must now enquire into the reason for this difference. It is unlikely that the reducibility of vanadium in the two porphyrins is intrinsically different. The clue is provided not by the X.P.S.work but by our work on the diffuse reflectance spectra of the supported porphyrins. ULTRAVIOLET-VISIBLE SPECTROSCOPY Spectra are shown in fig. 2. The u.v.-vis. spectra of the metal-free porphyrins H,TPP and H,ETP on alumina are very like the CH,Cl, solution spectra. We see an intense band, the soret band, near 400 nm, and a succession of less intense bands extending to 650 nm. Thus we conclude that the porphyrins are dispersed on alumina as neutral molecules. The spectra of the porphyrins on the acidic supports, fluorided silica and silica-X.P.S. AND U.V.-VIS. STUDY OF v PORPHYRINS ON CATALYSTS ( b ) 523 , ,&,I 620 300 500 70 0 1 I I I 00 500 700 A/nm 300 500 700 h/nm Fig.2. Ultraviolet-visible reflectance spectra : absorbance (increasing vertically) against wave- length (Ilnm). (A) Metal-free etioporphyrin on (a) alumina and (b)cation in CH,Cl, + CF,CO,H. (B) Oxovanadium(rv) etioporphyrin on (a) alumina, (b) fluorided silica and (c) Ni-Mo/Al,O, catalyst in the oxide form; note that the three spectra are similar and in particular there is no band of protonated porphyrin (523 nm). (C) Metal-free tetraphenylporphyrin on (a) alumina and (b) cation in CH,Cl, + CF,CO,H. (D) Oxovanadium(1v) tetraphenylporphyrin on (a) alumina, (b) fluorided silica and (c) Ni-Mo/Al,O, catalyst in the oxide form; note the additional band near 650-660 nm in the two lower spectra, showing that some of the tetraphenylporphyrin is in the metal-free protonated form.P.C. H. MITCHELL, c. E. SCOTT, J-P. BONNELLE AND J. G. GRIMBLOT 1055 alumina, are similar to the spectra of the protonated porphyrins. For the species H, TPP2+ and H, ETP2+ we see new, more intense bands at 655 and 523 nm and a shift of the soret band by ca. + 15 nm. Therefore, the major species on the acidic supports are protonated porphyrins. The spectra of VOTPP and VOETP on alumina (fig. 2 ) are typical metalloporphyrin spectra. However, on fluorided silica, which is a strongly acidic support, VOTPP has a new band at 655 nm. Similarly, we see a new 650 nm band in the spectrum of VOTPP on Ni-Mo/Al,O,, but it is less intense than on fluorided silica. A band also appeared at 655 nm when a solution of VOTPP in dichloromethane was acidified with trifluoroacetic acid.The 650-655 nm band is clearly due the the protonated species H, TPP2+. Thus acid decomposes VOTPP and the liberated metal-free porphyrin becomes pro tonated. VOETP behaved differently from VOTPP. The spectrum of VOETP on fluorided silica and Ni-Mo/Al,O, was the same as that on alumina. In particular there was not a new band near 525 nm (H,ETP2+). Also, the spectrum of VOETP in CH,Cl, did not change when the solution was acidified with trifluoracetic acid. Clearly there is little, if any, decomposition of VOETP on acidic supports. DECOMPOSITION OF VANADYL PORPHYRINS We can now suggest why the binding energy of VOTPP, but not that of VOETP, decreases in X.P.S. The u.v.-vis. spectra show that VOTPP, unlike VOETP, decom- poses on acidic supports.One product, clearly identified in the u.v.-vis. spectrum, is protonated, metal-free porphyrin. The surface reaction is a proton-assisted demetallation : VOTPP + 4HO- + V02+ + H,TPP2+ + 402-. We suggest that the vanadium(1v) species released from the porphyrin is reduced in X.P.S. We see reduced vanadium because reduction shifts the V binding energy away from the 0 1s satellite peak. The spectrum of any undecomposed VOTPP will still be hidden by the 0 1s satellite peak. The idea that vanadium(1v) released by decomposition of the vanadyl porphyrin is reduced in X.P.S., and not the vanadyl porphyrin itself, is supported by a number of observations. There is no decomposition of VOETP, even on acidic supports, and no peak of reduced vanadium in the spectrum of supported VOETP.On MOO,, an acidic oxide, VOTPP decomposes and reduced vanadium is seen in X.P.S., but on Co,O, and MoS, there is no decomposition or reduction. Finally, a general feature of vanadium porphyrins is that those with an oxidation state of less than four are not easy to prepare; redox processes apparently occur only at the periphery of the porphyrin ring.1°-12 The phthalocyanine, VOPc, also undergoes reduction in X.P.S. For VOPc we would expect preferential protonation at the meso nitrogens.12 VOPc is regarded as more stable than VOTPP but is known to undergo slow demetallation in strong sulphuric acid.14 CONCLUSIONS On acidic (H+) supports VOTPP undergoes at least partial demetallation. The u.v.-vis. diffuse reflectance spectrum of the protonated species H4TPP2+ has been studied.VOPc experiences rather less demetallation and VOETP none at all. In the X.P.S. experiments that part of the vanadium which was released by demetallation of the porphyrins on acidic supports was reduced in the spectrometer by adventitious grease. Porphyrin-bound VO was apparently not reduced.1056 X.P.S. AND U.V.-VIS. STUDY OF v PORPHYRINS ON CATALYSTS Changes of N 1s binding energies were too small to be interpreted by N binding to the catalysts. The N 1s X.P.S. did, however, reveal different types of nitrogen; for example, the nitrogen bound to V and the meso nitrogen of VOPc. Changes of core-electron binding energies of V and N caused by interactions of the vanadyl porphyrins with catalysts and supports are small, no more than a few-tenths of an electronvolt and hardly outside the experimental precision.Any idea that the porphyrins bind strongly through vanadium or nitrogen is not supported by our X.P.S. study. E1sewhere14 we have suggested that the porphyrins are bound to the catalyst by a donor-acceptor, or charge-transfer, interaction, the delocalised n system of the porphyrin ring being the donor and the Brarnsted- and/or Lewis-acid function of the catalyst being the acceptor. Some support for this idea is given by our observation of a 1 eV decrease of the binding energy of the valence band of MoS, caused by interaction with vanadyl porphyrins. P. C. H. Mitchell and J. A. Valero, React. Kinet. Catal. Lett., 1982,20, 219; Znorg. Chim. Acta, 1983, 77, 179. E. Edlerin, in Znt. Symp. Vanadium and other Metals in Petroleum, ed. G. Kapo (Maracaibo, Venezuela, 1973), vol. 111-C. (a) T. F. Yen, The Role of Trace Metals in Petroleum (Ann Arbcr Science Press, Ann Arbor, 1975); (b) A. Ekstrom, C. J. R. Fookes, T. Hambley, H. J. Loeh, S. A. Miller and J. C. Taylor, Nature (London), 1983, 306, 173; (c) C. Berthe, J. F. Muller, D. Cagniant, J. Grimblot, J. P. Bonnelle, J. Goulon, R. Guilard, J. L. Poncet, J. C. Escalier and B. Neff, Nouv. J. Chim., 1985, in press. * A. D. Adler, F. R. Longo, F. Kampasand J. Kim, J. Znorg. Nucl. Chem., 1970,32,2443; E. C. Johnson and D. Dolphin, Znorg. Synth., 1980, 20, 143. J. G. Erdman, N. W. K. Ramsey and W. E. Harrison, J. Am. Chem. Soc., 1956,78, 5844. R. A. Rowe and M. M. Jones, Znorg. Synth., 1957, 5, 113. L. J. Boucher, N. L. Hoky and B. H. Davis, in New Approaches to Coal, ed. B. D. Blanstein, B. C. Bockrath and S. Friedman, ACS Symp. Ser. no. 169 (American Chemical Society, Washington DC, 1981). (a) G. V. Ovedraogo, D. Benlian and L. Porte, J. Phys. Chem., 1980, 73, 642; (6) M. V. Zeller and R. G. Hayes, J. Am. Chem. Soc., 1973,95,3855; (c) Y. Niwa, H. Kobayashi and T. Tsuchiya, Znorg. Chem., 1974,13,2891; ( d ) Y. Niwa, H. Kobayashi and T. Tsuchiya, J. Chem. Phys., 1974,60, 1799. P. Canneson, M. I. Cruz and H. van Damme, Deu. Sedimentol., 1978, 27, 217. ’ I. D. Chapman and M. L. Hair, J. Catal., 1963, 2, 145. lo K. M. Kadish and L. R. Shive, Znorg. Chem., 1982, 21, 3623. l2 P. C. H. Mitchell and C. E. Scott, to be published. l3 J-M. Mouchot, Thbe (Universite de Metz, 1978). 14 P. C. H. Mitchell and C. E. Scott, ZX Zberoamerican Symp. Catal., Lisbon, 1984, to be published. (PAPER 4/ 1 197)

ISSN:0300-9599    

DOI:10.1039/F19858101047

出版商:RSC    

年代:1985

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. 1, 1985,81, 1057-1069 Kinetics of the Oxidation of 2-Hydroxy- 2-methylpropanoic Acid by Silver(r1) Ions Complexed with 2,2'-Bipyridine in Aqueous Nitrate Media BY MALCOLM P. HEYWARD AND CECIL F. WELLS* Department of Chemistry, University of Birmingham, Edgbaston, P.O. Box 363, Birmingham B 15 2TT Received 1st August, 1984 Stopped-flow traces show that the oxidation of 2-hydroxy-2-methylpropanoic acid (hmpa) by [Ag(bipy),12+ proceeds in two consecutive reactions. Both are found to be first order in [Ag'I] and first order in [hmpa]. The first rapid reaction is ascribed to complex formation between Ag" and hmpa and the second to a slower redox step. A mechanism is proposed to account for the observed orders in [AgII], [hmpa] and [H+] for each reaction and values for the enthalpies and entropies of activation are determined. To investigate the effect on the redox kinetics of oxidatively inert species close to the cation, these are compared with the transition-state parameters for the oxidation of hmpa by aqua-metal cations and for the oxidation of other substrates by metal cations complexed with 2,2'-bipyridine.Following our spectrophotometric examination1 of the complexed species of Ag'l which exist in aqueous nitric acid containing 2,2'-bipyridine, we have investigated the kinetics of the oxidation of hydrogen peroxide,2 propan-2-o13 and pinaco14 by such complexed AgII ions. With the latter two substrates, the kinetics of the oxidation show3' that intermediate complexes with the substrate ligand are involved, but with hydrogen peroxide such intermediate complexes were not detected in the oxidation.2 Intermediate cation-substrate complexes have been detected in the oxidation of 2-hydroxy-2-methylpropanoic acid (hmpa) by aqua-cations such as Ceaq6 and VZq,' and the importance of chelation in determining the rate of oxidation has been di~cussed.~ We wish, therefore, to compare AgII complexed with 2,2'-bipyridine with these aqua-cations in the reaction with hmpa.Some kinetic work has also been recorded for the oxidation of iminodiacetic acid and N-methyliminodiacetic acid by these complexes involving AgI1 and 2,2'-bip~ridine.~ EXPERIMENTAL MATERIALS The [Ag(bipy),12+ ion was prepared in aqueous nitric acid by the anodic oxidation of AgI under nitrogen with sufficient 2,2'-bipyridine added to provide an excess of 4.0 x mol dm-3 after complexing of the AgII with two 2,2'-bipyridine molecule~.~-~ 2-Hydroxyl-2-methyl- propanoic acid was recrystallized from benzene and dried under vacuum: solutions in water were prepared by weight.All other materials were as described previ~usly.l-~* PROCEDURES Concentrations of the species of AgI' complexed with 2,2'-bipyridine were determined spectrophotometrically using the recorded variation of extinction coefficient with acidity.' The product of the oxidation, acetone, was removed by high-vacuum distillation in a closed system 10571058 OXIDATION BY SILVER(II) IONS IN NITRATE MEDIA with the reaction mixture first neutralized by adding KOH:3-5, the concentration of acetone in the distillate was determined spectrophotometrically as the anion of the 2,4-dini trop hen ylhydrazone. Rates were followed spectrophotometrically at 450 nm using the thermostatted cell com- partment of the Unicam SP800 spectrophotometer with a Honeywell chart recorder or of a Unicam SP6-500 with digital display or a Durrum-Gibson stopped-flow spectrophotometer coupled with a Tektronix oscilloscope with a storage screen.The excess 2,2'-bipyridine over [Ag(bipy),12+ was always maintained at 4.0 x as in previous investigations involving this complex. -4 RESULTS AND DISCUSSION STOICHIOMETRY OF THE OXIDATION Owing to the low rate of the oxidation when [AgII] = [hmpa] w lop4 mol drn-,, it is only possible to determine values for I A[AgI1] \/I ACacetone] I with [hmpa] $ [AgII] w loF4 mol drn-,.The acetone, distilled off such mixtures after neutralizing the nitric acid with KOH, was determined spectrophotometrically and the values for IAIAglI]l/lA[acetone]I are given in table 1 for initial [AgII] w 2.5 x mol drn-,. Table 1 shows that this ratio is not influenced by the presence or absence of molecular oxygen in the reaction mixture. The mean value of I A[AgII] ]/I A[acetone] I is 1.92 0.16 under conditions where the loss of [AgII] due to reaction with water and acetone is < 1 % of the observed decrease. The stoichiometry is represented by 2AgI' + (CH,),COHCOOH -+ 2Ag' + (CH,),CO + H& + CO,. VARIATION OF RATE WITH [hmpa] AND [H+] AT 25 "C The result of the stopped-flow experiment in fig. 1 shows that the decrease of [AgII] for the reaction of ca.2 x mol dm-, of [Ag(bipy),12+ with excess [bipy] = 4.00 x lo-, mol dm-, with ionic strength adjusted to 1.00 mol dm-, by the addition of NaNO, proceeds through two consecutive reactions. The decrease in optical density at 450 nm was followed on the stopped-flow spectrophotometer over a long period by repeated manual triggering at one minute intervals after the start of the reaction and shows a rapid decrease followed by a much slower decrease. The slower decrease is first order in [AgII] and the slope of the log plot agrees well (fig. 1) with the slope of a similar plot for the same reaction determined using a conventional spec tropho t ometer . Values of the first-order rate constants for the slow decay were determined from linear log plots using conventional spectrophotometry for [hmpa] in the range (3.0-1 5.0) x lo-, mol dm-, for acid concentrations ranging from 0.1 to 1 .O mol drn-,.At any particular acid concentration it was found that the first-order rate constant varied linearly with [hmpa], passing through the origin, as shown in fig. 2. Values for the second-order rate constant k determined from the slopes using the least-squares procedure are given in table 2. It was shown that the first-order rate constant is unaffected by replacing molecular oxygen in the mixture by molecular nitrogen, replacing the added NaNO, by LiNO, to maintain the ionic strength at 1 .OO mol dm-, or replacing the added NaNO, partially with added AgNO,, for the same [hmpa] and With initial [hmpa] w 25.0 x 10 mol dm-, the fast initial reaction accounts for all the loss of optical density at 450 nm and this decrease can be followed using stopped-flow spectrophotometry.Linear log plots were always found and fig. 3 shows that the first-order rate constant varies linearly with [hmpa] with the plot passing W+l*M. P. HEYWARD AND C. F. WELLS 1059 Table 1. Values of I A[Ag"] \/I A[acetone] 1 at various [HClO,] for an ionic strength of 1.0 mol dmF3 at 20 "C [HClOJ [hmpa] initial [AgI'] 1 AIAgl'] 1 /mol dm-3 /mol dmP3 /lo-, mol dm-3 1 A[acetone] 1 0.10 0.05 2.20 2.06 0.10 0.10 2.58 2.03" 1 .oo 0.05 2.50 1.89 1 .oo 0.05 2.50 1.80" 1 .oo 0.10 2.50 1.84 a Under nitrogen. 0 -0.5 0 400 800 t l s Fig. 1. Comparison of the decrease in absorbance at 450 nm for the reaction of 2 x lo-, mol dm-3 Ag(bipy)E+ with 5 x mol dm-3 hmpa at an ionic strength of 1.0 mol dm-3, [H+] = 0.20 mol dm-3 and with excess [bipyH+] = 4 x lop3 mol dm-3 at 25 "C using conven- tional spectrophotometry ( x ) and stopped-flow spectrophotometry (0).through the origin. Values for the second-order rate constants k' determined from the slopes using the least-squares method with [hmpa] in the range 0.05-0.20 mol dm-3 at each acidity, except for [H+] = 0.1 mol dm-3 where the range was 0.025- 0.20 mol dm-3, are given in table 3. VARIATION OF RATE CONSTANTS WITH TEMPERATURE The reaction was also found to take place in two stages at other temperatures. The slow step could be followed at 11.5, 19.4 and 30.0 "C and the reaction was found to be first order in [Ag"] and first order in [hmpa] at constant acidity.Values for the second-order rate constant k calculated from the slopes of the plots of the pseudo-1060 OXIDATION BY SILVER(II) IONS IN NITRATE MEDIA 20 0 10 20 [ hmpa I / 1 0-3 mol dm-' Fig. 2. Plot of the pseudo-first-order rate constant k, against [hmpa] for the slow oxidation of hmpa by AgIIbipy complexes at an ionic strength of 1.00 mol dm-3 and with excess [bipyH+] = 4.0 x mol dm-3 at 25 "C using the following [H+] (mol drnp3): x , 0.10; I, 0.30; A, 0.50; 0, 0.80; 0, 1.00. Table 2. Values of the second-order rate constant k and a derived function for the redox step at an ionic strength of 1.00 mol dm-3 with excess [bipyH+] = 4.00 x mol dm-3 ~ ~~ ~~ ~ [H+l k k(1 +K,h)(I +K',h) T/"C /mol dm-3 /dm3 mol-l s-l /dm3 mol-l s-' 11.5 11.5 11.5 11.5 11.5 19.4 19.4 19.4 19.4 19.4 25.0 25.0 25.0 25.0 25.0 30.0 30.0 30.0 30.0 30.0 0.100 0.300 0.500 0.800 1 .oo 0.100 0.300 0.500 0.800 1 .oo 0.100 0.300 0.500 0.800 1 .oo 0.100 0.300 0.500 0.800 1 .oo 0.039 f 0.001 0.1 20 f 0.0 1 4 0.167 f 0.008 0.255 f 0.028 0.256 f 0.027 0.116f0.001 0.237 f 0.006 0.369 f 0.008 0.40 f 0.01 0.56 f 0.02 0.248_+0.010 0.49 f 0.03 0.95 f 0.03 1.24 f 0.05 2.06 k 0.14 0.48+0.01 0.98 f 0.05 1.43 & 0.13 2.55 kO.11 3.01 f0.04 0.042 & 0.001 0.148 f 0.01 7 0.233 f 0.01 1 0.42 f 0.05 0.47 & 0.05 0.126f0.001 0.299 & 0.008 0.53 k 0.01 0.70 f 0.02 1.07 f 0.03 0.271 +0.011 0.63 f 0.04 1.40 & 0.04 2.25 f 0.09 4.I2 0.28 0.53f0.01 1.27 f 0.07 2.16 0.20 4.7 +_ 02 6.2 0.08M. P. HEYWARD AND C. F. WELLS 1061 3 2 d I \ - 0 1 0 0.1 0.2 [hmpa] /mol dm-3 Fig.3. Plot of the psuedo-first-order rate constant kh against [hmpa] for the rapid initial reaction between hmpa and AgIIbipy complexes at an ionic strength of 1.00 mol dm-3 and with excess [bipyH+] = 4.0 x mol dm-3 at 25 "C using the following [H+] (mol dm-3): ., 0.10; 0, 0.20; A, 3.30; 0, 0.50; x , 0.80; T7, 1.00. Table 3. Values of the second-order rate constant k' and a derived function for the rapid initial reaction at an ionic strength of 1.00 mol dm-3 with excess [bipyH+] = 4.00 x mol dm-3 H+ k k(l+K,h)(l+K',h) T/"C /mol dm-3 /dm3 mol-l s-' /dm3 mol-l s-l 25.0 25.0 25.0 25.0 25.0 25.0 10.6 10.6 10.6 10.6 38.0 38.0 38.0 38.0 0.100 0.200 0.300 0.500 0.800 1 .oo 0.100 0.300 0.500 1 .oo 0.100 0.300 0.500 1 .oo 17.0 + 0.6 10.1 f0.5 8.9 f 0.3 7.0 t 0.3 5.2 _+ 0.4 5.2 f 0.2 3.2kO.l 1.8f0.1 1.7 f 0.1 1.2f0.2 73+ 13 35.5 + 1.9 23.2 f 2.3 23.0 + 0.3 18.6 f 0.7 12.0 f 0.6 11.4k0.4 10.4f0.4 9.3 f0.7 10.4 & 0.4 3.4 +O.1 2.2kO.l 2.4 f 0.1 2.2 f 0.4 82f 15 48.3 f 2.6 37.4+ 3.7 52.4 f 0.71062 OXIDATION BY SILVER(II) IONS IN NITRATE MEDIA first-order rate constant against [hmpa] using the least-squares procedure are collected in table 2. The initial rapid stage was followed at 10.6 and 38.0 "C using the stopped-flow technique. Linear log plots were always found and the pseudo-first-order rate constant varied directly with [hmpa] passing through the origin at constant acidity. Table 3 contains values for the second-order rate constant calculated using the least-squares procedure.MECHANISM OF THE RAPID INITIAL REACTION It appears likely that the rapid fall in optical density observed when mixing the reactants, as illustrated in fig. 1, can be ascribed to the formation of a complex between AgII and hmpa in a similar manner to the intermediate complex found in the oxidation of hmpa by Ceiz6 and V&.l0 The following processes may be involved: K h Ag(bipy)i+ + H+ + Ag(bipy),+ + bipy H+ K , hmpa + H+ hmpa H+ klf Ag(bipy)i+ + hmpa $ Ag(bipy)i+ hmpa Ag(bipy)i+ + hmpa + Ag(bipy),+ hmpa + bipy Ag(bipy),+ + hmpa + Ag(bipy)"+ hmpa k4f Ag(bipy)2+ + hmpa + Ag2+ hmpa + bipy. kib k2f kzb k3f k3b kib (3) (4) (7) If it is assumed that the proton transfer equilibria reactions (2) and (3) are very rapid compared with reactions (4)-(7), and if one can neglect the back reaction in our conditions because either k,, < k l f , k,, < kZf, k,, 6 k,f and k4, < k4f or the concen- tration of the complexes containing hmpa is kept low by removal in the subsequent redox step, the rate of change in the optical density is given by where h = [HCIO,IT, the subscript T indicates the total concentration of that species, K;I = K,/[bipyH+] and it is assumed that the contributions of AgII complexed with hmpa to [AgII], are small.Eqn (8) complies with the observed first order in each of [AgII], and [hmpa],, and equating the observed second-order rate constant k with the complex term involving rate and equilibrium constants in eqn (8) gives k'( 1 + K;1 h) (1 + Kc h) = kif + k2f+ (k3f +k4f) (9) Table 3 shows that values for the left-hand side of eqn (9) calculated using the experimental values for k' and K;1 together with5 K, = 0.1 dm3 mol-l are approxi- mately constant at each temperature in the range [H+], = 0.2-1.0 mol drn-,.This suggests that k,, and k4f contribute little to eqn (9) in this range of acidities. Although these results do not permit one to distinguish between klf and k,f in eqn (9), where intermediate complexes have been detected3? for other substrates X oxidized by Ag(bipy)i+, the dominant intermediate species is Ag(bipy)X2+, suggesting that reactionM. P. HEYWARD AND C. F. WELLS 1063 ( 5 ) may dominate here. A plot of the logarithm of the mean value of the left-hand side of eqn (9) in the range [H+IT = 0.2-1.0 mol dmP3 against the reciprocal of the absolute temperature gives a straight line, and application of the least-squares procedure gives AH* = 78 f 3 kJ mol-1 and AS* = 34 The rise in the value of the left-hand side of eqn (9) at [H+] = 0.1 mol dmP3 is the reverse of the change expected if k3f and k,, become significant.It is more likely that this rise at low [H+IT is due to the participation of anionic forms of hmpa, which has an acid dissociation constant K, z 1 x lop4 mol dm-3 in this temperature range.'l 1 1 J K-l mol-l. MECHANISM OF THE REDOX REACTION Following our investigation of the kinetics of the oxidation of pinaco14 by Ag(bipy)i+, it is necessary to include all possibilities of AgI' complexed with 2,2'-bipyridine and hmpa as being capable of being involved in the mechanism in the following rate-determining steps : (10) k Ag(bipy)2+hmpa -4 Agl + bipy H + radical (11) k Ag(bipy)2+hmpa- 1 Ag' + bipy + radical Ag2+hmpa 3 AgI + radical + H+ (12) k, Ag2+hmpa- -+ AgI + radical (13) (14) k' Ag(bipy)i+hmpa -3 Ag' + bipy H+ +radical + bipy (1 5 ) k' Ag(bipy)z+hmpa- 1 Agl + 2bipy + radical with unprotonated 2,2-bipyridine being rapidly protonated in these acidic solutions.The rate-determining reactions (1 OF( 15) will be preceded by the following rapidly established equilibria : B Ag(bipy)2+ + hmpa Ag(bipy)2+hmpa Ir Ag(bipy)i+ + hmpa $ Ag(bipy);+hmpa (17) K , Ag(bipy)2+hmpa $ Ag(bipy)2+hmpa-+ H+ (18) K2 Ag(bipy)2+hmpa + H+ e Ag2+hmpa + bipyH+ (19) K3 Ag2+hmpa Ag2+hmpa- + H+ K : Ag(bipy)i+hmpa Ag(bipy),hmpa- + H+ (21) K6 Ag(bipy)i+hmpa + H+ Ag(bipy)2+hmpa + bipyH+.(22)1064 OXIDATION BY SILVER(II) IONS IN NITRATE MEDIA The radical formed in the rate-determining steps will react very rapidly with more AgI1 species to produce Ag' and to produce a ~ e t o n e . ~ ~ 6 9 lo The rate of loss of AgII will then be given by (23) [2QKL(k, h + k , K1+ k3 K2 h2 + k4 K2 K3 h) + 2p'(k; + ki K; h-l)] [Ag*']T [hmpa], (1 + Kk h) (1 + Kc h) + [hmpa], [QKk(h + KI + K2 h2 + K2 K3 h) +p'( 1 + K; h-l)] ' If the second term in the denominator is much smaller than the first, the observed second-order rate constant is given by k( 1 + K;I h) ( 1 + K, h) = 2QKk( kl h + k , Kl + k3 K2 h2 + k4 K, K, h) + 2p'(k; + ki K; h-l). (24) Fig. 4 shows that plots of the left-hand side of eqn (24) using the experimental values of k and K;1 with K, = 0.1 dm3 mol-1 against h2 are linear with intercepts on the ordinate.Eqn (24) suggests that the slopes of these plots are 2pKL k3 K, with intercepts of 2Qk2 KL K , + 2p'k;. Note that to produce the term in eqn (24) in h2 it is essential to include the possibility that Ag(bipy)i+ and Ag(bipy),' arising in reaction (2) are both involved in rapid pre-equilibria with hmpa. If only Ag(bipy)i+ is involved in such a pre- equilibrium, as suggested might be the case from the analysis of the kinetics of the rapid initial change in optical density, by combining the equilibrium (25) B" Ag(bipy)i+ + hmpa -+ Ag(bipy),+hmpa + bipy with equilibria reactions (1 7)-(2 l), the equation analogous to eqn (24) is k( 1 + K;1 h) (1 + Kc h) = 2/?"(k, + k, Kl h-l+ k, K , h + k, K , K,) + 2Q'(k; + k;lK; h-') (26) and it is not possible to explain the observed linear variation of the left-hand side with h2.We must therefore conclude that, although the equilibrium constants for equilibria involving Ag(bipy),+ and hmpa are too low in magnitude to show experimentally in eqn (9), they do, nevertheless, operate as pathways towards the redox steps. The values of Qk, K, derived from the slopes of fig. 4 are collected in table 4 . A plot of logQK, k, against the reciprocal of the absolute temperature is linear and the overall values for AH* and AS* obtained using the least-squares procedure are given in table5. A plot of the logarithm of the intercepts of fig. 4 against the reciprocal of absolute temperature is also linear, suggesting strongly that only one of the terms in 2Qk,KkKl+2Q'k; is significant.As in the oxidations of other substrates by Ag(bipy)E+, Ag(bipy),+ is the bipyridine-silver (11) complex which is oxidatively active and not Ag(bipy)i+; it is assumed, by analogy, that p'k; % Qk, K;I K,, and values for QK, k, are also included in table 4. The overall values of AH* and AS* derived from the application of the least-squares procedure to the linear plot of logQK, k , against the reciprocal of absolute temperature are given in table 5. COMPARISON WITH OTHER CATION + SUBSTRATE REDOX SYSTEMS INFLUENCE OF OXIDATIVELY INERT SPECIES ADJACENT TO THE CATION The overall values of AH* and AS* for the oxidation of hmpa by Ag(bipy)i+ are compared in table 5 with AH* and AS* for the oxidations of hmpa by other metal ions and with those for the oxidation of other organic substrates by Ag(bipy)i+ : the reactions are represented in table 5 by the intermediate complex involved.Initial inspection shows that for a variation in E, of between 1 and 2 V for the cations,M. P. HEYWARD AND C. F. WELLS 1065 " " a . - h2/mo12 dm" Fig. 4. Plot of the left-hand side of eqn (24) against h2 for the slow oxidative step between hmpa and AgI'bipy complexes at an ionic strength of 1.00 mol dm-3 and with excess [bipyH+] = 4.0 x mol dmP3 at the following temperatures ("C): x , 11.5; 0, 19.4; A, 25.0; 0, 30.0. Table 4. Values of k,PK, and k2PK, for the oxidation of hmpa by Ag(bipy)2f and A&: at an ionic strength of 1.00 mol dm-3 with excess [bipyH+] = 4.00 x lop3 mol dmV3 k3 B K 2 T/"C /dm6 mol-2 s-l k2 BKl /s- 11.5 0.32 +_ 0.05 0.073 f0.029 19.4 0.58 & 0.07 0.135 f 0.039 25.0 2.23 k0.21 0.170+0.113 30.0 3.27 f 0.18 0.386 k 0.097 although the values of AH* for any species of 2-hydroxy-2-methylpropanoic acid range only between 58 and 95 kJ mol-l, the values of AS* vary from -64 to 4 8 1 1 K-1 mnl-1 hiit fnr the nnid2tinn nf iin-inni7ed hmna AH* ranwc nnlv between 78 and 95 kJ mol-1 whilst AS* still varies between- -52 and +8l J K-l mol-l.To explain this wide range of AS* values it is necessary to look at the variation of AS* involving two extreme cases: the oxidations of substrate by Ni(bipy),3+, where little disturbance of the coordination sphere round Ni3+ occurs, and oxidations by Ceti, where there is considerable disturbance of the coordination sphere around Ce4+.For oxidations by Ni(bipy)i+ there are two major contributions to AS* arising from configurational changes in the solvent: first, AS,* has positive values because of water oriented around Nl(Dipy);T being released to tne DUIK solvent on tne cnange 3 + -+ L + in the transition state; secondly, AS: is negative because of restrictions imposed on1066 OXIDATION BY SILVER(II) IONS IN NITRATE MEDIA Table 5. Comparison of overall values of AH* and AS* for the oxidation of hmpa by Ag(bipy)2+ and Ag:; with those for the oxidation of other organic ligands by the same ions and with those for the oxidation of hmpa by other aqua-cations A f c v AS,*, intermediate complex /kJ mol-1 /J K-' mol-l ref. Ag( bipy)z+ hmpa- AgZfhmpa Ag(bipy)2+pin Ag2+pin Ag(bipy)2+PriOH Mn3+hmpa VOlhmpa V(OHi+)hmpa Ce4+hmpa- 58+6 9 5 k 8 49.5 & 0.7 45.3 f 3.3 88+5 93 7 8 k 5 82+7 64 -64+39 32 k 56 81 + 5 18+ 12 13& 15 81 -41 & 15 -52&20 22 this work this work 4 4 3 5 10 10 6 solvent molecules if a proton is released to the solvent.Thus for the oxidations of the uni-negative ions Br- and N; by Ni(bipy):+, where contributions to AS* other than AS: are likely to be similar, AS* is ca. O,l2>l3 but for H202, where a proton is released to the solvent in the transition state and a negative AS: is also involved, AS* has a large negative value.14 In contrast to oxidations by Ni(bipy)i+, oxidations with Ceit involve considerable movement of the molecules contiguous with the metal ion, in this case water molecules; this adds to the possibility of a variation in the entropy of activation.As a consequence of the larger size of Ce4+ compared with metal ions of the first transitional series, the coordination number for solvent molecules around Ce4+ is greater than six and these loosely packed solvent molecules will have considerable freedom of movement. With Ce4+, therefore, there is a strong tendency to form complexes involving the partial removal of these solvating molecules and for all substrates, HA, oxidation proceeds through an intermediate complex as in rapid + radical + Hiq. (27) Values of AHpre and AS,,, for the formation of the complex in the rapid pre-equilibrium and values of AH* and AS* for the slow redox step are given in table 6 for a range of HA. If ASc represents the entropy change resulting from the disturbance of oriented water molecules solvating Ce4+, the positive ASc deriving from the release of such water molecules by the complexation of HA with Cei; will contribute to AS,,,.During the process initial state -+ transition state for the subsequent redox step involving CeHAii, the Ce4+.-.HA distance probably contracts to facilitate electron transfer, with further release of oriented water giving a positive contribution AS,* to AS*, to which AS,* and AS; must also contribute. Table 5 shows that the larger substrates have the smaller values for AS,,, because of their low ability to penetrate the aqua-sheath in the initial state, but their larger positive values of AS* arise from the larger positive AS,* which results from the greater contraction in the Ce4+-.-HA distance to produce the transition state required for electron transfer : correspondingly, the smaller substrates penetrate the aqua-sheath further in the initial state, producing a large positive AS,,,, and therefore require a smaller contraction of Ce4+...HA to achieve the transition state for electron transfer with a smaller AS,* leading to a negative AS*.For the diols other than pinacol, where only values of AH* and AS*M. P. HEYWARD AND C. F. WELLS 1067 Table 6. Comparison of values of AH,,,, AS,,,, AH* and AS* for the oxidation of HA by Ceti via reaction (27) substrate, HA AHpre Aspre AH* A S /kJ mol-l /J K-' mol-l /kJ mol-l /J K-l mol-l ref. ~~ (CH,),CHOH CH,CHOHCH,CH, /CH,-CH, CH, )CHOH - \CH,-CH, (CH,),COH(CH,),COH CH,OHCH,CH,OH C:H,CHOHCH,CH,OH CH,OHCH,CH,CH,OH CH,OHCH,CH,CH,CH,OH CH,OHCH,CH,CH,CH,CH,OH CH,CH(OCH,)CH,CH,OH 67 243 86 60 225 73 5.9 5.9 127 34.3 156 94 - 73 64 73 123 - 115 85 - - - - - - - - - - - 7.5 15 - 48 16 145 16 67 17 29.3 18 - 16.7 18 39.3 18 192 18 171 18 67 18 are available, AS* gets larger as the carbon chain gets more complex: presumably the more complex substrates do not penetrate the aqua-sheath so well in the initial state, therefore requiring a greater contraction in Ce4+...HA to achieve the transition state for electron transfer, which results in a larger positive AS* from the large AS,*.The oxidation of some substrates by Ceti proceed by the alternative pathway Ceki + H A a q 2 H & + CeAii %Ce3+ + radical involving the loss of a proton from HA in the initial state: values of AH,,, and AS,,, for the pre-equilibrium and of AH* and AS* for the redox step are given in table 7.Now AS, for proton release contributes to AS,,,, which is large and negative for the carboxylic acids : presumably the counterbalancing positive ASc is only small arising from the diffuse negative charge on -COO-, producing little penetration of the aqua-sheath of Ce4+ in an ion-pair type complex CetiAiq. However, large positive values of AS* are found for these substrates in table 7, owing to the considerable contraction in the Ce4+.-.A- distance needed in the transition state to facilitate the electron transfer producing a large positive AS,* from the attendant release of oriented water. However, with the proton released with the electron in the oxidation of the C-H bond in A- merely transferring to the -COO- group, resulting in little or no influence from AS:, [ASpre[ < [AS*I, leaving the overall entropy of activation, AS& = AS,,,+AS*, lying in the range 23-44 J K-l mol-l.The alcohols, with their more concentrated charge on -0-, penetrate farther into the aqua-sheath of Ceii and produce a large positive AS,,,, with a large positive ASc counterbalancing the negative ASp: because of the large penetration in the initial state, only a small contraction in Ce4+**.A- is needed to produce the transition state for electron transfer and the low values of AS,* result in smaller values of AS* than found with the carboxylate anions. The energetics for the oxidation of the > C-H bonds are relatively uninfluenced by these considerations as the values of AH* lie in a close range over the whole series of carboxylic acids and alcohols.Table 5 shows for the oxidation of undissociated hmpa a close parallel for Ag2+hmpa and Mn3+hmpa. The complete kinetic analysis for the latter5 shows that the high oxidation rate of hmpa and pinacol by Mnii compared with those for simple1068 OXIDATION BY SILVER(II) IONS IN NITRATE MEDIA Table 7. Comparison of values of AH,,,, ASpre, AH* and AS* for the oxidation of HA by Ceti via reaction (28) substrate, HA ~~ AH,,, AS,,, AH* A S /kJ mol-l /J K-' mol-l /kJ mo1-I /J K-l mo1-l ref. CH,OHCOOH CH,CHOHCOOH (CH,),COHCOOH PhCHOHCOOH (CH,),COH(CH,),COH (CH,),CHOH CH,CHOHCH,CH, /CH,-CH,\ CH2, /CHOH \cH,-cH,/ I CH,-COOH HO-CH,-COOH - 40 - 107 121 - 32 - 76 105 - 54 - 149 119 - 40 - 96 98 87 112 32 131 122 - - - - 81 307 98 151 113 172 121 41 75 111 35 40 6 6 6 6 17 15 16 16 19 alcohols and carboxylic acids oxidized by Mnii arises from the high positive ASc produced by the chelation process in the pre-equilibrium: it is likely that a similar explanation holds for the positive AS,*, for Ag2+hmpa and Ag2+pin (where pin = pinacol).The positive AS& found for Ag(bipy)2+pin and Ag(bipy)2+ PriOH may also arise in the same way, with the high positive AS, arising from the close approach of the substrate to Ag2+ in the initial state allowed by the removal of one bipyridine from Ag(bipy)i+. It is notable that VOihmpa and V(OH):+hmpa have negative values of AS,*,, although the values of AH&, are close to those found for other aqua-cations.This could arise in two ways: the hydrolysed forms of the cation may prevent hmpa approaching as close to the V atom as it is able to do in the intermediate complexes involving the other cations, resulting in a much smaller positive ASc to compensate for the large negative AS:, or the system may be more rigid in the transition state, resulting in a lower entropy. AS,*, for the two complexes in table 5 involving the anionic ligand hmpa- differ considerably, despite close agreement for their values of AH,*,. However, for Ce*+hmpa-, AS,*, lies at the bottom end of the range AS,*, = 23- 44 J K-l mol-1 found in table 6 for the anions of other hydroxycarboxylic acids and it seems reasonable that this complex conforms to the explanations given above for these anions with Ce;;.On the other hand, the negative AS& for Ag(bipy)2+hmpa- matches those values found for the complexes of Vv with the undissociated acid. It is difficult to visualise some specially rigid requirement for the transition state with this complex, and it is more likely that ASc and AS,* are both very small with the bipyridine ligand restricting the close approach of the chelating hmpa- to the Ag2+ atom, i.e. more like an ion pair, allowing the contribution of the negative AS, to the initial state to dominate in AS,*,. This then raises the question of whether the alternative pathway involving Ag(bipy)i+hmpa is operating, with both bipyridine ligands keeping Ag2+ and hmpa apart. However, this possibility is rejected on the grounds that the redox potential of Ag(bipy)2+ is highe~-l-~ than that for Ag(bipy);+ and the definitive observation of Ag2+hmpa as a pathway indicates that the oxidation prefers Ag2+ in a high redox state.M. P. HEYWARD AND C. F. WELLS 1069 M. P. Heyward and C. F. Wells, J. Chem. SOC., Dalton Trans., 1981, 431. M . P. Heyward and C. F. Wells, J. Chem. Soc., Dalton Trans., 1981, 1863. M . P. Heyward and C. F. Wells, J. Chem. SOC., Dalton Trans., 1982, 2185. M . P. Heyward and C. F. Wells, J. Chem. SOC., Faraday Trans. 1, 1984,80, 2155. C. F. Wells and C. Barnes, Trans. Faraday Soc., 1971, 67, 3297. Z. Amjad, A. McAuley and U. Gomwalk, J. Chem. SOC., Dalton Trans., 1977, 82. A. F. M. Nazer and C. F. Wells, J. Chem. SOC., Dalton Trans., 1980, 2143. a C. Baiocchi, G. Bovio and E. Mentasti, Znt. J. Chem. Kine?., 1982, 14, 1017. C. F. Wells, Tetrahedron, 1966, 22, 2685. lo A. F. M. Nazer and C. F. Wells, J. Chem. SOC., Dalton Trans., 1980, 2143. l1 I. M. Heilbron and H. N. Bunbury, Dictionary of Organic Compounds (Eyre & Spottiswoode, London, l2 C. F. Wells and D. Fox, J. Chem. SOC., Dalton Trans., 1977, 1502. l 3 J. K. Brown, D. Fox, M. P. Heyward and C. F. Wells, J. Chem. SOC., Dalton Trans., 1979, 735. l4 C. F. Wells and D. Fox, J. Chem. Soc., Dalton Trans., 1977, 1498. l5 C. F. Wells and M. Husain, Trans. Faraday Soc., 1970,66, 679. l6 C. F. Wells and M. Husain, Trans. Faraday SOC., 1970, 66, 2855. l7 C. F. Wells and M. Husain, Trans. Faraday Soc., 1971, 67, 1086. l9 Z. Amjad and A. McAuley, J. Chem. Soc., Dalton Trans., 1974, 2521. 4th edn, 1965), vol. 3, p. 1741. A. Prakash, R. N. Mehrotra and R. C. Kapoor, J. Chem. SOC., Dalton Trans., 1979, 205. (PAPER 4/ 1369)

ISSN:0300-9599    

DOI:10.1039/F19858101057

出版商:RSC    

年代:1985

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I, 1985, 81, 1071-1086 Multicomponent Ion Exchange in Zeolites Part 1 .-Equilibrium Properties of the Sodium/Calcium/Magnesium- Zeolite A System BY KEVIN R. FRANKLIN AND RODNEY P. TOWNSEND* Department of Chemistry, The City University, Northampton Square, London EClV OHB Received 7th August, 1984 The results of a detailed study of the ternary exchange equilibrium system Na/Ca/Mg in zeolite A are discussed. A complete ternary isotherm (25 "C) is presented, together with isotherms for the three conjugate binary systems (Na/Ca-A, Ca/Mg-A and Na/Mg-A). The Na/Ca-A isotherm is compared with previously published data, and reasons for the discrepancies in the corresponding standard free energies are discussed in terms of both experimental uncertainty and interference through partial hydronium exchange.The exchange isotherm for the Ca/Mg-A system shows enormous preference for calcium over magnesium; it was only found possible to replace 1.8 % of the calcium by magnesium at 25 "C. Despite many attempts, it was also found impossible to replace sodium completely by magnesium in zeolite A: a maximum level of exchange of 85 % was achieved. This observation contradicts previous reports in the literature. The selectivity trends observed in the ternary Na/Ca/Mg-A system are next described quantitatively in terms of ternary and pseudo-binary separation factors, and some features of the ternary system which could not have been predicted from the binary data are highlighted. It is emphasised that separation-factor data should not be used to infer details regarding the sitings of the ions within the zeolite.The potential use of zeolite A as a builder in has led to considerable interest being shown in the ion-exchange properties of this zeolite. Of especial interest has been the exchange selectivity of the sodium form for the so-called 'hardness ions' calcium and magnesium. Equilibrium studies reported in the l i t e r a t ~ r e ~ - ~ have concentrated mainly on sodium-calcium exchange. Considerable discrepancies regarding the reported select- ivity of zeolite A for calcium are found in the literat~re.~-* Barri and Rees7 have investigated the sodium-magnesium exchange also, and have made a partial study of the Na/Ca/Mg ternary system. This latter study indicated that zeolite A exhibits very marked preferences for both ions, especially calcium, but that these preferences change markedly as the relative loadings of these two ions in the zeolite were changed.The aims of this present study were to investigate in much more detail the selec- tivities shown in the Na/Ca/Mg ternary systems and to try to explain some of the discrepancies found in the literature. EXPERIMENTAL MATERIALS All chemical reagents used in this study were purchased from B.D.H. and were of AnalaR grade. Zeolite A was kindly supplied in the sodium form by Laporte Industries (Widnes). 10711072 ION EXCHANGE IN ZEOLITES Table 1. Analysis of zeolite Aa Na-A Ca-A 22.26 24.92 32.73 32.36 27.69 27.46 16.56 0.336 15.00 99.24 100.10 1 .oo 1 .oo 0.98 1.02 - a Unit-cell compositions: Na-A: Na,,.9 (A102)12.0 (Si02)12.03 27.8 H,O Ca-A: Na,.,, Ca,., (A10z)12.0(Si02),,.,,.31.4H,0. Before use the zeolite was left in contact with sodium chloride solutions (1 mol dmP3 concentration) to ensure (as far as possible) that it was initially in the homoionic sodium form.After subsequent washing and drying at 80 "C the material was equilibrated with water vapour by placing it in a desiccator over a saturated sodium chloride solution at room temperature for four weeks. A sample of calcium A was next prepared by exchanging the sodium form exhaustively (at room temperature) with 0.5 mol dm-3 calcium chloride solutions. After exchange the Ca-A sample was also equilibrated with water vapour in a desiccator. The results of the chemical analyses for these two exchanged forms of zeolite A are given in table 1.ANALYSES OF ZEOLITES Water, silicon and aluminium were determined by standard gravimetric methodsg Sodium, calcium and magnesium contents were determined after the zeolite sample had been dissolved in nitric acid. Sodium was then analysed by flame photometry, and calcium and magnesium by atomic absorption spectrophotometry using nitrous oxide + acetylene and air + acetylene flames, respectively. Aluminium interference was countered in both cases by adding a quantity of aluminium to the standard solutions which was equal to that present in the samples. In addition, improved reproducibility of the magnesium analyses was achieved by the addition of a large background level (400 ppm) of calcium ion. ANALYSES OF ISOTHERM SOLUTIONS Sodium was again determined by flame photometry.For high concentrations of either calcium or magnesium in solution, these ions were determined separately by titrations with EDTA at pH 10.lo When both calcium and magnesium were present, a total divalent ion content was obtained by this method. Calcium was next determined separately by an indirect titration using EGTA.ll Magnesium could then be inferred by difference. For solutions containing small quantities of both calcium and magnesium, atomic absorption spectrophotometry was employed in the same manner as for the samples of the zeolite. EQUILIBRIUM MEASUREMENTS The equilibrium isotherms were constructed by mixing 0.2 g aliquots of zeolite Na-A (composition in table 1) with solutions of a total normality* of 0.1 equiv dmP3, which contained the appropriate quantities of sodium, calcium and/or magnesium chloride. The suspensions were continuously shaken in sealed plastic bottles for 10 days and kept at a temperature of 25 "C.Na-A was used as the starting material for all exchanges except for the construction of the Ca/Mg binary isotherm, where the starting material used was Ca-A (composition in table 1). * Throughout this paper, one 'equivalent' refers to one mole of unit positive or negative charges.K. R. FRANKLIN AND R. P. TOWNSEND 1073 When equilibration was complete the two phases were separated by centrifugation, The zeolite phase was washed briefly with water and then both phases were analysed for all three ions. Reversibility was investigated using the 'wet method 'I2 reported previously. Equilibrium concentrations of each ion in each phase were next calculated from the experimental data in terms of equivalent fractions on the assumptions that and E,, + E,, + EMg = 1 Ena + Eca + EMg = 1 where Ei and Ei are the equivalent fractions of ion i in the solution and crystal (zeolite) phases, respectively. The data are therefore adjusted for any hydronium exchange which may also have occurred.13 RESULTS AND DISCUSSION Analyses showed that 2.3 % of the sodium initially in the zeolite was not exchangeable with either calcium or magnesium.All isotherm points were therefore corrected to eliminate this non-exchangeable sodium. A similar amount of non-exchangeable sodium in zeolite A was also reported recently by Barri and R e e ~ .~ BINARY EQUILIBRIA The Na/Ca isotherm (fig. 1) shows a very strong preference for calcium over sodium, especially at low calcium loadings (e.g. when ECa = 0.097, E,, has a value of only ca. 1.8 x lo-*). The isotherm is seen to be reversible and (in general) agrees well with those previously published. 3-8 The Na/Mg isotherm (fig. 2) is highly sigmoidal, the zeolite showing a strong preference for magnesium at low loadings and a converse strong preference for sodium at high loadings of magnesium. Regarding the shape of the isotherm, good agreement Fig. 1. Ion-exchange isotherm for Na/Ca exchange in zeolite A. Measured at 0.1 normal and 298 K: 0, forward points, x , reverse points.1074 E& 0 . 4 - 0.2- ION EXCHANGE IN ZEOLITES i -- 0 I 0.2 0 .4 0.6 0.8 E,, Fig. 2. Ion-exchange isotherm for Na/Mg exchange in zeolite A. Measured at 0.1 normal and 298 K. Symbols as in fig. 1 . is observed with the data of both Barri and Rees7 and Wolf and F ~ r t i g . ~ In the present study, however, the maximum level of magnesium exchange was found by repeated checks and rigorous analyses to be only 85.2%, whereas for the two previous studies5? it was assumed the curve terminated at 100% at 25 "C. Neither of these isotherms, however, shows any experimental points above 85% , 5 7 7 and therefore there seems little justification now for extrapolating these data to 100%. However, Barri and Rees7 found a higher exchange level at 65 "C. The Ca/Mg isotherm (fig. 3) shows a remarkably low maximum exchange level for magnesium.It was found to be impossible to replace more than 1.8% of the calcium ions by magnesium at 25 "C. This indicates a preference for calcium over magnesium which is so great when starting with pure Ca-A as to almost totally exclude magnesium from the zeolite. It is of interest that Barri and Rees7 found a maximum magnesium exchange level of 40% for the same system at 65 "C, which shows that this exchange system is very strongly temperature-dependent. Standard free energies for the binary exchanges were calculated using standard procedures :14 AG* = -(RT/zA zB) In K, (1) where zA and zB are the valencies of ion A (ingoing) and ion B (outgoing), respectively, and K, is the thermodynamic equilibrium constant. K, was determined using the Gaines and Thomas approach,15 viz.where (3)K. R. FRANKLIN AND R. P. TOWNSEND 1075 I I 1 I 0.2 0 . 4 - 0.6 0.8 E M g 0 Fig. 3. Ion-exchange isotherm for Ca/Mg exchange in zeolite A. Measured at 0.1 normal and 298 K. Symbols as in fig. 1 . Table 2. Standard free-energy data for the systems Na/Ca-A and Na/Mg-A best-fit - system EA(max) AGg,/kJ equiv-l polynomial equationsa Na/Ca-A 1 Na/Mg-A 0.852 +0.50 In Kg = 6.681-15.578E+13.998(q)2-11.851 (3)3 - 2.83 In KG = 6.883-1 5.696 EA + 23.399 E i - 14.195 E l a The superscript N refers to normalised va1~es.l~ KG is the Gaines and Thomas corrected selectivity quotient, and mA and mB are the concentrations (mol dm-3) of ions A and B in solution. r is a total correction for non-ideality in the solution phase, and was calculated using Glueckauf s mode1.I6 Recently, Reed7 questioned the applicability of the Glueckauf model for the calculation of solution-phase activity coefficients in mixed NaCl + CaCl, solutions, since the results disagreed with some experimental results obtained by Moore and Ross.'~ However, other more recent measurement^^^ indicate that the data of Moore and Ross were in error.The integral on the right-hand side of eqn (2) may be determined analytically by integrating a best-fit polynomial of the experimental plot of In KG against EA. These polynomial coefficients and the resulting standard free energies for both the Na/Ca-A and the Na/Mg-A exchanges are given in table 2. The standard free energy for the latter system was determined using the usual normalisation procedure for partial exchange.,O No thermodynamic treatment of the Ca/Mg-A system was possible due to the very low maximum level of exchange obtained for magnesium.Despite the good agreement in the literature regarding the general shape of the1076 ION EXCHANGE IN ZEOLITES Table 3. Standard free-energy data collected from the literature investigators zeolite source AGg,/kJ equiv-1 ~~ ~ Barrer et aL3 Linde Co. -0.59 Ames4 Linde Co. - 2.27a Wolf and Furtig5 ? - 1.12b'.c Sherry and Walton6 Linde Co. - 3.08 Barri and Rees' Laporte Industries - 2.69 Wiers et a1.8 Huber Corp. -2.15 present work Laporte Industries -2.83 ~~ a Recalculated by Wiers et AGg3. Calculated from selectivity-coefficient data. -2- 0 0.2 0 . 4 - 0.6 0.8 1 ECa Fig. 4. Plots of In K , against E,, for the Na/Ca-A exchange, as taken from the literature.BRW, Barrer Rees and Ward,3; A, A m e ~ ; ~ WF, Wolf and F ~ r t i g ; ~ SW, Sherry and Walton;6 BR, Barri and R e e ~ ; ~ WGC, Wiers, Grouse and Cilley;* FT, present study. Na/Ca-A isotherm, inspection of table 3 shows a wide variation in the derived standard free energies. Fig. 4 shows the plots of In KG agains &., used to calculate these free energies. Obviously the largest differences occur at low values of Eta. Rees17 has attributed the large differences between the isotherms of Barri and Rees7 and Barrer, Rees and Ward3 to experimental error (in the earlier work) in the determination of Eca at low values of Eta. It is indeed in this region that KG is very sensitive to small errors in the measurement of Ec,, and this must therefore largely explain the discrepancies in the plots in this region.The smaller differences between the plots at higher values of EC, may not be due simply to experimental error however. One possibility is that the discrepancies mayK. R. FRANKLIN AND R. P. TOWNSEND 1077 be due to some fundamental differences in selectivity resulting from different modes of preparation of the zeolite A samples. The occurrence of some non-exchangeable sodium in the zeolite tends to suggest this may occur (see beginning of results section). However, Wiers et ~ 1 . ~ noticed that the total ion content recovered from the crystal phase (i.e. Na + Ca) was low by between 5 and 13 % after equilibration. This was also noticed in the present study for both the Na/Ca and Na/Mg exchanges, but the strong trend of increasing recovery as ENa decreased which was observed by Wiers et a1.' was not as obvious in this work.This loss is likely to be due to hydronium exchange.8 Such a conclusion is supported by the work of Drummond et U Z . , ~ ~ who noticed that rapid hydronium exchange occurred when zeolite A was placed in water, and also by other present studies reported elsewhere,13 where similar rapid exchange was observed even in solutions of 0.1 mol dmP3 sodium chloride, although the rate of pH increase was seen to slow down as the concentration of salt in the solutions was raised.13 Further investigations showed that, as might be expected if an exchange process were involved, if increasing quantities of zeolite were suspended in the same volume of a solution of given salt concentration, then the percentage of the zeolite that became hydronium exchanged decreased.Since the level of hydronium exchange will in- evitably affect the preference of the zeolite for calcium over sodium, the Na/Ca-A exchange isotherms measured using different experimental conditions (viz. solution volume and zeolite weight) by different workers (fig. 4) may well be slightly different. Drummond et aL21 suggest that the hydronium exchange is accompanied by a structural breakdown of the zeolite and then a subsequent release of aluminium and silicon species. However, analyses carried out in this work did not result in any aluminium or silicon being detected in solution, so these studies do not support this suggestion.21 TERNARY EQUILIBRIA The ternary isotherm was constructed from 123 pairs of experimental points (fig.5). Fig. 5(a) shows that the range of possible ternary compositions for the solution phase was covered adequately. Inspection of fig. 5(b) shows, in conformity with the concomitant binary study, that complete magnesium exchange could not be achieved. Ten of the pairs of experimental points have been labelled in order to provide some indication of the overall selectivity behaviour of the system. A better method of representing the isotherm is to distort the triangular coordinates corresponding to the solution phase so that each measured solution-phase composition lies directly on top of its corresponding equilibrium crystal-phase point.22* 23 For systems where the selectivity of the zeolite is small and uniform, a purely mathematical method for doing this has been used.22 In this present case however, modifications to the procedure, involving the 'by eye' fitting of data, were found to be necessary.The ternary diagram produced in this way is seen in fig. 6. The distorted solution-phase grid lines are shown in 10% increments superimposed on top of the regular (undistorted) crystal- phase grid. The line corresponding to ENa = 0 now marks the boundary for the maximum attainable level of exchange in the zeolite. It is seen clearly (fig. 6) that the maximum attainable value of EMg increases smoothly as the calcium loading is lowered. The selectivities displayed by zeolite A for each ion separately are seen better in fig.7(a)-(c). Here the appropriate sets of grid lines (which together comprise the complete ternary diagram in fig. 6) are shown separately, enabling one to assess by eye the general preferences exhibited by the zeolite for each ion in turn. Thus, for example, in fig. 7 (a) the points on which the solution-phase grid lines fall on the Na/Ca and Na/Mg crystal-phase edge extrema give the binary Na/Ca-A and Na/Mg-A isotherms, respectively, and the manner in which the grid lines move over the surface1078 ION EXCHANGE IN ZEOLITES Fig. 5. Experimental points for the Na/Ca/Mg-A exchange at 0.1 normal and 298 K: (a) solution phase and (b) crystal phase. Correspondingly labelled points represent examples of tie-lines between the two diagrams.K. R. FRANKLIN AND R.P. TOWNSEND Na 1079 Fig. 6. Ternary isotherm for the Na/Ca/Mg-A exchange at 0.1 normal and 298 K, depicting the distorted 10% solution-phase grid lines on the regular crystal-phase grid. [For a detailed description of the procedure used to obtain such plots see ref. (22).] in between indicates the effect(s) of introducing a third ion. It is clear from these figures that many of the features are as might be expected from the binary isotherms (e.g. the strong preference for both calcium and magnesium at high loadings of sodium, and the increasing preference for both calcium and sodium as the loading of magnesium in the zeolite is increased). However, one important feature not expected from the binary data is seen in fig. 7(6). Taking for example the Ec, = 0.1 grid line, calcium is found to be less preferred by the zeolite when both sodium and magnesium are present than is the case when either sodium or magnesium is present alone.Ternary data may be quantified conveniently by defining a separation factor, which is analogous to separation factors used commonly in binary exchange studies.14 For binary exchanges, one such function is defined as Then if ga > 1 the zeolite prefers ion A, and the converse. Using the same approach, the corresponding ternary separation factor is1080 ION EXCHANGE IN ZEOLITES Na Na Fig. 7. For legend see opposite.K. R. FRANKLIN AND R. P. TOWNSEND Na 2 Ca O.' Mg 1081 Fig. 7. Solution- and crystal-phase grid lines depicting selectivity for each ion separately: (a) sodium, (b) calcium and (c) magnesium.The condition for ion A being preferred to a combination of ion B and C is then that Fig. 8(a), (b) and (c) show these separation factors for sodium, calcium and magnesium, respectively. Calcium is seen to be preferred to a combination of sodium and magnesium at all loadings [fig. 8(b)], while magnesium is preferred only at high sodium loadings [fig. 8(c)]. Sodium becomes preferred to a combination of calcium and magnesium at high loadings of either calcium or magnesium [fig. 8(a)]. Appropriate combinations of these ternary separation factors lead also to pseudo-binary factors, which describe the selectivity of the zeolite for one ion over another in the presence of the third: €&cq ' 1. Plots of In && are shown in fig. 9. These plots are useful, even though the k functions are essentially empirical and therefore care should be taken in their interpretation.For example, in the region of high sodium loadings it is seen [fig. 9(a) and (b)] that both calcium and magnesium are preferred over sodium, and fig. 9(c) shows that in this same region the zeolite shows a slight preference for magnesium over calcium. However, these results should not be interpreted as implying that there is necessarily any direct competition between calcium and magnesium for ion sites within the zeolite. On the contrary, the evidence is that calcium and magnesium ions are found at different sites in zeolite A at high sodium 10adings.~ Rather, these data should be taken 36 FAR 11082 Na 6 . 2 5 Ca Mg Fig. 8. For legend see opposite.Ca K.R. FRANKLIN AND R. P. Na TOWNSEND 1083 Fig. 8. Logarithmic contour plots of the ternary separation factors as a function of crystal-phase composition: (a) In cpMga, (b) In N2,aMga and (c) In Nzgaa. merely to indicate that the zeolite shows (in this composition region) a stronger preference for magnesium over sodium than for calcium over sodium. This preference is not evident from the binary isotherms (fig. 1 and 2) but it is seen not to be in conflict with trends in the binary ka data (fig. lo), where a preference for magnesium over calcium in Na-A is implied with ENa > 0.9. Despite the recent development of a suitable thermodynamic model for the rigorous treatment of ternary systems in the calculation of standard free energies of exchange has not been possible for the Na/Ca/Mg-A system because of the occurrence of partial exchange (fig.2, 3 and 6). Normalisation procedures,2o derived for binary systems, cannot be applied to ternary exchange isother~ns.~~~ 25 However, the inability to calculate values of AGe does not preclude necessarily the accurate prediction of selectivities over a range of solution composition. The means by which accurate prediction may be undertaken, and a validation of the method using the Na/Ca/Mg-A data reported here, will be discussed in Part 2 of this series. K. R. F. gratefully acknowledges an S.E.R.C. CASE award with Unilever Research. 36-21084 ION EXCHANGE IN ZEOLITES Na Fig. 9. For legend see opposite.Fig. 9. K. R. FRANKLIN AND R. P. TOWNSEND Na A87 4.80 ' A Logarithmic contour plots of the pseudo-binary separation factor as a crystal-phase compositions : (a) In %&, (b) In gi& and ( c ) lng$'&.- 6- 6 7 . ) 0.4 0.6 0.8 0 0.2 Ej 1085 \ 1 Mg function of Fig. 10. Logarithmic plots of the binary separation factors a as a function of crystal-phase compositions: 0, Na/Ca-A (Ei = &a); A, Na/Mg-A (Ei = EMg).1086 ION EXCHANGE IN ZEOLITES M. J. Schwuger and H. G. Smolka, Colloid Polym. Sci., 1978, 256, 1014. D. J. Sherman, AICHE Symp. Ser., 1978,74,98. R. M . Barrer, L. V. C. Rees and D. J. Ward, Proc. R. SOC. London, Ser. A, 1963, 237, 180. L. L. Ames, Am. Miner., 1964, 49, 1099. F. Wolf and H. Furtig, Kolloid Z.Z. Polymer, 1965, 206, 48. H. S. Sherry and H. F. Walton, J. Phys. Chem., 1967, 71, 1457. 7 S. A. I. Barri and L. V. C. Rees, J. Chromatogr., 1980, 201, 21. B. H. Wiers, R. J. Grosse and W. A. Cilley, Environ. Sci. Technol., 1982, 16, 617. R. M. Barrer and R. P. Townsend, J. Chem. SOC., Faraday Trans. I , 1976, 22, 661. lo A. I. Vogel, Textbook of Quantitative Analysis (Longmans, London, 1978), p. 320. l1 A. I. Vogel, Textbook of Quantitative Analysis (Longmans, London, 1978), p. 327. l2 P. Fletcher and R. P. Townsend, J. Chem. SOC., Faraday Trans. I , 1981, 77, 497. l3 R. P. Townsend, K. R. Franklin and J. F. O’Connor, Adsorption Sci. Technol., 1984, in press. l 4 A. Dyer, H. Enamy and R. P. Townsend, Sep. Sci. Technol., 1981, 16, 173. l5 G. L. Gaines and H. C. Thomas, J. Chem. Phys., 1953, 21, 714. l6 E. Glueckauf, Nature (London), 1949, 163, 414. l7 L. V. C. Rees, in Properties and Applications of Zeolites, ed. R. P. Townsend (Chem. SOC. Spec. Publ. no. 33, the Chemical Society, London, 1980), p. 218. E. W. Moore and J. W. Ross, J. Appl. Physiol., 1965, 20, 1332. J. Ananthaswamy and G. Atkinson, J. Solution Chem., 1982, 11, 509. 2o R. M. Barrer, J. Klinowski and H. S. Sherry, J. Chem. SOC., Faraday Trans. 2, 1973 67, 1669. 21 D. Drummond, A. De Jonge and L. V. C. Rees, J. Phys. Chem., 1983,87, 1967. 22 P. Fletcher, K. R. Franklin and R. P. Townsend, Philos. Trans. R. SOC. London, Ser. A , 1984, 312, 23 R. K. Bajpai, A. K. Gupta and M. Gopala Rao, J. Phys. Chem., 1973, 77, 1288. 24 P. Fletcher and R. P. Townsend, J. Chem. SOC., Faraday Trans. 2, 1981, 77, 955. 25 P. Fletcher and R. P. Townsend, J. Chem. SOC., Faraday Trans. 2, 1981, 77, 965. 141. P. Fletcher and R. P. Townsend, J. Chem. SOC., Faraday Trans. 2, 1982, 77, 2077. (PAPER 4/ 1395)

ISSN:0300-9599    

DOI:10.1039/F19858101071

出版商:RSC    

年代:1985

数据来源: RSC

摘要:

J . Chem. Soc., Faraday Trans. I, 1985, 81, 1087-1094 Thermal Unimolecular Decomposition of j?-Propiolactone (Oxetan-2-one) BY H. MONTY FREY* AND IVY M. PIDGEON Chemistry Department, Reading University, Whiteknights, Reading RG6 2AD Received 9th August, 1984 The thermal decomposition of oxetan-Zone has been investigated in the gas phase in the temperature range 262-322 "C. The reaction, which yields ethylene and carbon dioxide, is homogeneous and obeys first-order kinetics. There is a minor heterogeneous isomerization to yield acrylic acid. The decomposition is almost certainly a unimolecular process, which in the pressure range studied, < 6 Torr, is in the fall-off region. High-pressure rate constants were determined by extrapolation using three procedures that are discussed.The preferred method using plots of k-l against p-o.61 yielded the Arrhenius equation logkls-l = 14.86f0.30- 180.46f3.20 kJ mol-l/RTlnlO. Theoretical (RRKM) calculations are close to the experimental fall-off curves on the basis of the strong-collision assumption or with (AE)d (stepladder model) b 24 kJ mol-l, from which it is hardly distinguishable. Certainly it is necessary to assume very efficient intermolecular energy transfer. Comparison with the result from other studies suggests a concerted decom- position via an activated complex with zwitterionic character rather similar to those involved in cyclobutanone decompositions. It was pointed out recently1 that in comparison with the relative wealth of kinetic data on the thermal unimolecular decompositions of cyclobutanes, cyclobutanones and oxetanes, there was almost no quantitative information available on the gas-phase kinetics of the decomposition of oxetanones (Q-lactones).Since in many respects cyclobutanones behave rather differently from cyclobutanes and oxetanes, which are rather similar to one another, there is some interest in determining into which class of compounds the oxetanones fall. Thus despite real experimental problems, mainly associated with the very low volatility of oxetanones and also their ease of hydrolysis, work was undertaken on P-butyrolactone. The work on P-butyrolactone showed that the thermal decomposition was uni- molecular and that an activated complex with considerable zwitterionic character was probably involved. Comparisons with the only other gas-phase study on an oxetanone, P-propiolactone itself,2 was only of limited usefulness since the work on the substituted lactone suggested that the Arrhenius parameters for the former compound were probably both too high. To try and resolve this problem as a prelude to studies on a series of substituted oxetanones, we undertook a reinvestigation of the pyrolysis of P-propiolactone. EXPERIMENTAL /?-Propiolactone was a commercial sample whose purity was checked by gas chromato- graphy (see later).Other reagents used for calibration or identification were all commercially available. A conventional static high-vacuum system was used which employed greaseless valves 10871088 H. M. FREY AND I. M. PIDGEON (Teflon-glass) throughout. To minimise adsorption problems the vacuum line was maintained at ca.90 "C by heating with Electrothermal tape. The Pyrex reaction vessel was fully immersed in a high-temperature fused-salt thermostat and attached to a Baratron pressure gauge. The Baratron head, which was kept at 100 "C, was of special design so that it had virtually no dead volume and the connection from it to the reaction vessel was by capillary tubing. Signals from the Baratron were fed via a suitable interface to an Apple I1 microcomputer, which enabled the pressure in the reaction vessel to be recorded at a number of times at given time intervals, and these data were stored on disc for subsequent rate-constant calculations. Analysis was by gas chromatography employing instruments equipped with both f.i.d.and thermistor detectors. Signals from these detectors were integrated using a Hewlett Packard 3380 A instrument. The purity of the P-propiolactone was checked using a 3.4 m x 2.2 mm column packed with 15 ?< w/w Carbowax 1500 on 80/ 100 mesh Chromosorb P. The column was operated at 80 "C with 50 p.s.i. (3.5 x lo5 Pa) of N,. Other than a trace of ethylene there was only one peak, indicating a purity of > 99%. The earlier work of James and Wellington, had shown that the only volatile products of the thermal decomposition of the lactone were ethylene and carbon dioxide. To confirm these findings we carried out analyses on the gaseous products of decomposition using several columns and both f.i.d. and thermistor detectors. With a Poropak N column two peaks were obtained with retention times corresponding to C,H, and CO,.An attempt to detect CO using a molecular sieve (4A) column showed that it could not have been present in more than trace quantities. Thus the only decomposition pathway leading to volatile products detectable by our chromatographic analysis would appear to be The p-propiolactone was stored as a liquid in a small sample vessel attached to the vacuum line. Immediately before use, any volatile impurities were removed by pumping for ca. 20 s. Some of the lactone was then transferred (distilled under vacuum) to another vessel containing molecular sieve 4A and dried for 1 h before use in the kinetic studies. Considerable difficulties were experienced in obtaining reproducible rate data as determined from pressure measurements.This was manifested in departures from first-order behaviour at short reaction times and values for P, appreciably less than 2P0. We suspected that the problems were caused by heterogeneous reactions on the surface of the Pyrex reaction vessel and we attempted to eliminate there by 'ageing' using conventional methods; in this case by the pyrolysis of various compounds including hexamethyldisilazane, ally1 bromide, 2, Sdihydrofuran and hexamethyldisiloxane. All were ineffective, as also was the procedure of carrying out numerous pyrolyses of the reactant itself. The most successful ageing procedure proved to be the pyrolysis of 1 -ethylcyclopentene in the reaction vessel at ca. 500 "C. A number of samples of this reagent at pressures a little greater than 12 Torr were introduced into the vessel and left for b 1 h before removing and refilling the vessel. This was continued over a period of several days.After this treatment a number of samples of P-propiolactone were then introduced into the reaction vessel. At the highest temperatures at which kinetic runs were carried out the value of P, now approached 2P0 closely. At lower temperatures the final pressure was 2 1.85P0 but in all cases the kinetic behaviour of the appropriate pressure plot now showed a first-order fit over a reasonable extent of reaction. In an attempt to define the system more precisely and in particular to establish the reason for the failure of P, to equal twice the initial reactant pressure, calibration experiments were carried out with ethylene.By filling the reaction vessel with known pressures of ethylene and then analysing by gas chromatography it was established that the total pressure change which occurred when the propiolactone was pyrolysed corresponded closely to reaction (1). However, at the low pressures used and the rather large errors involved in sharing the contents of the vessel with the chromatographic sampling volume we could not distinguish between the possibilities that the pressure deficit was the result of a polymerization reaction yielding noTHERMAL DECOMPOSITION OF OXETAN-2-ONE 1089 volatile products and an isomerization reaction yielding a volatile product not detected in our analytical system. The only reasonable isomerization product which might have escaped detection was acrylic acid : CHZ- CO I I + CH,=CHCOOH. CHz-0 We were able to demonstrate that with our chromatographic system it would not have been detectable in the quantities required by the pressure deficit.Accordingly, after a run at relatively low temperature and long time the contents of the reaction vessel were frozen into a small volume of water. After allowing this to warm to room temperature the solution was found to be acid. Titration with 0.008 mol dm-3 sodium hydroxide solution (phenolphthalein indicator) gave a quantitative measure of the acid yield close to that required if the pressure deficit was due to the isomerization. A 'blank' had previously been obtained by filling the reaction vessel with ca. 3 Torr of CO, which was then frozen into a small volume of water and titrated as in the run with reactant. Nevertheless the errors involved are large and we cannot rule out a small fraction of the pressure deficit being due to a polymerization reaction. (It is perhaps relevant to note that /I-propiolactone does have a tendency to polymerize even at room temperature and we observed over a period of some weeks that the lactone kept in a storage vessel formed a clear viscous liquid which turned into a white opaque solid.The isomerization to acrylic acid is a rapid acid-catalysed reaction in the liquid phase which may also be of some relevance.) We return to these problems and their implications for data handling later in this paper. The procedure employed in a typical kinetic run depended on the pressure required in the reaction vessel.For pressures < 2 Torr the small storage vessel containing the dried (liquid) /I-propiolactone was opened to the heated vacuum line and the contents of the line subsequently shared with the previously evacuated reaction vessel. For pressures > 2 Torr the sample was heated to 100 "C and the joint between the storage vessel and the heated line warmed using a hot-air blower. Times and pressures were recorded as soon as the lactone was admitted to the reaction vessel. At the end of a run when chemical analysis was required the contents of the reaction vessel were shared with an evacuated, heated gas-sample pipette and subsequently transferred to the heated gas-sample volume of the gas chromatograph. RESULTS AND DISCUSSION After the ageing of the reaction vessel a run was carried out for ca.8 half-lives at 263.96 "C to obtain a value for P,. A plot of In (P, -Pt) against time was found to be linear up to 3 half-lives. With a small change in P, of < 1 % the plot became linear to > 3 half-lives. This suggests that the decomposition to ethylene and carbon dioxide, which is responsible for the pressure change, is a first-order process. Our preliminary experiments have indicated that the discrepancy between 2P0 and P,, is entirely or mainly due to reaction (2), though we cannot rule out some polymerization as a contributory process. If we assume that both of these processes are surface catalysed then the difficulty in obtaining reproducible rates in the untreated reaction vessel is not surprising.Furthermore, the fact that the values of the ratio of P, to Po were all smaller in untreated compared with treated vessels and also smaller at lower temperature is consistent with a surface-catalysed process (we would expect the surface reaction to have a lower energy of activation than the homogeneous decomposition). To handle the pressure against time data we assume that the isomerization reaction is entirely responsible for the pressure deficit and that it occurs by a first-order process. This leads to the following expression for the pressure at any time, Pt; pt = Po exp(-kt)+P,[l -exp(-kt)] (3)1090 H. M. FREY AND I. M. PIDGEON where k is the overall rate constant and is related to k , the rate constant for the homogeneous reaction [reaction (l)] by k = k , Po/(P, - Po).Values of Po for any particular run were obtained by extrapolation of Pt against time to t = 0. A computer program based on the Simplex method was used to obtain the ‘best’ value of P,. Time data extending to at least 3 half-lives were used. If rather than reaction (2) it is assumed that the pressure difference between 2P0 and P, is due entirely to a polymerization reaction and in the further assumption that this process obeys first-order kinetics (k,) then it is easy to show that Pt = Poexp[-(kl+k,)t]+P,[l-exp(-klt)]. (4) On the basis of our pressure against time data we were not able to decide whether eqn (3) or (4) provided a better fit. In fact for most runs the difference between the calculated values of k, from the two assumptions was small. Since acrylic acid was found we have used values derived from eqn (3) in the remainder of this paper.Further confirmation that the model was a realistic one was obtained from a series of runs in which k, was calculated from the gas-chromatographic analysis of the ethylene yields. Although they showed considerably greater scatter than the values of k , calculated from pressure data the agreement between them was good (to within + 5 % ) . Because of experimental and apparatus limitations we were only able to work with initial reactant pressures < 6 Torr. At these pressures we observed, in agreement with the previous work of James and Wellington, that the reaction is in its ‘fall-off’ region. Thus to obtain values of k , , at any particular temperature an extrapolation procedure was required.Before the advent of modern theories of unimolecular reactions it was the practice to obtain the high-pressure limiting rate constants by an extrapolation of a k-l against p-l curve, which was based on the Lindemann-Hinshelwood theory. Unfortunately such plots are strongly curved asp-’ approaches zero. Rabinovitch and Miche13 noted that if the reciprocal of the rate constant was plotted against rather than p-l a much more satisfactory extrapolation was possible, especially if data were available near the high-pressure limit. More recently Oref and Rabinovitch* have discussed the problem, introducing an adjustable parameter a for plots of k-l against pa. In the present work we have used various extrapolation procedures to obtain k , values.Plots of k-l against p-0.5 gave good straight lines at all temperatures. However, plots of k-l against pa, where a was varied to produce the best least-squares line, yielded values of a which differed from -0.5 and indeed were different for each temperature. Thus for example at 185.97 “C we illustrate the plots obtained from the experimental data in fig. 1 (a)-(c), in which k-’ is plotted as a function ~ f p - O . ~ , p-0.56 and P - - O . ~ ~ , respectively. In these plots -0.56 represents the value of a at this temperature and hence is the ‘best’ line and -0.61 is the average value of a at all temperatures. Note that while all the plots look very similar, indeed to the ‘eye’ they look equally good, they do produce significantly different intercepts.If very-low- pressure data are included it becomes clear that the log k against p-0.5 plots are appreciably curved, such that a linear extrapolation leads to an underestimate of k,. Similarly the k-l against P - O . ~ plots yield an overestimate of k,. The plots of k-l against pa yield values of k , which fall between these ‘extremes’. At each of 10 temperatures in the range 262-322 “C, I6 to 20 complete runs were carried out covering the pressure range 0.27-6 Torr. In every run the pressure was monitored for at least 3 half-lives. At the highest temperature pressures were recorded at 2 s intervals. While the Baratron and microcomputer system could easily cope with an appreciably faster rate of data handling, we were limited by the rate at which the reactant could be admitted to the reaction.vesse1 and hence the accuracy with which to and Po could be determined.THERMAL DECOMPOSITION OF OXETAN-2-ONE 1091 28 0 210 20 0 tA --.y^ . c1 W 160 120 80 ~ 0 0.5 1 1.5 20 0.5 1 1.5 2 0 0.5 1 1.5 1 (p /Torr)-O.s (p /TOIT) -0.56 TOIT IT)-^.^^ Fig. 1. Extrapolation procedure to obtain high-pressure rate constants for P-propiolactone at 285.97 "C. Table 1. Extrapolated high-pressure rate constants at various temperaturesa T/"C 262.45 269.87 274.56 279.59 k,,/10-3 s-1 1.9026 3.1377 4.3687 6.2500 T/"C 185.97 294.56 301.1 1 306.65 k,,/10-3 s-' 9.3371 17.937 26.731 41.391 T/"C 313.61 321.45 k1,/10-3 s-1 62.775 101.31 a Temperatures and rates are reported as measured and calculated and as used to obtain the Arrhenius equation.The number of figures quoted is not intended to reflect the accuracy of the data. Temperatures are probably accurate to between 0.15 and 0.20 "C. Since values of k,, depend so strongly on the method of extrapolation it is difficult to estimate their accuracy. In table 1 we report the values of k, at the various temperatures obtained from thep-0.61 plot. These values are a little greater than those obtained from a log k against p - O a 5 plot (we have used this type of extrapolation previously1) but smaller than those obtained from a k-l against p-0.5 plot. An Arrhenius plot of the data in table 1 yields the equation log kloo/s-l = 14.856 f 0.296 - 180.46 f 3.20 kJ mol-l/RT In 10 where the error limits are two standard deviations (95% confidence).To give some idea of the very large effect of different methods of extrapolation on the Arrhenius parameters we note the two extreme sets of values we have obtained. From the log k against P - O . ~ extrapolation the values of log A / s - l and E.JkJ mol-1 were 14.371 and 175.73, respectively, whereas from the k-l against plot the corresponding values were 15.563 and 187.13. Thus our preferred extrapolation procedure gives parameters not far from the mean of these extremes. The highly correlated nature of Arrhenius parameters makes such large differences possible even when the differences between the rate constants from which they are derived are quite small.1092 H. M. FREY AND I. M. PIDGEON 0 1 2 3 4 log (p/Torr) Fig. 2. Experimental and calculated fall-off curves (559 K): (-)experimental; (----) RRKM strong-collision hypothesis ; (.. . . . -) RRKM stepladder model, ( AE)d = 2000 cm-l ; (- . - .- . -) RRKM stepladder model, ( A E ) , = 1000 cm-'. For assignments see Appendix. 0 - 0.2 -0.4 n -0.8 0 1 2 3 4 log (p/Torr) Fig. 3. Experimental and calculated fall-off curves (568 K): (-) experimental; (-----) RRKM strong-collision hypothesis; (..-...) RRKM stepladder model, ( A E ) , = 2000 cm-' ; (-. - . -. - ) RRKM stepladder model, = 1000 cm-l. For assignments see Appendix.THERMAL DECOMPOSITION OF OXETAN-2-ONE 1093 The Arrhenius parameters obtained from the values of k , at 5 Torr are almost insensitive to the method of extrapolation and are log k1(5 Torr)/s-l = 13.658+0.118- 169.36f 1.26 kJ mol-l/RTln 10 where as before, the error limits are two standard deviations.We have carried out RRKM calculations based on a concerted process (for assignments see the Appendix) and the theoretical curves obtained are shown together with the experimental results in fig. 2. Inspection of fig. 2 reveals that the experimental results lie to the low-pressure side of the ‘strong-collision’ RRKM curve. This is an unusual situation, although it should be noted that in the earlier work of James and Wellington2 the experimental curve was even further from the theoretical curve. It is not always appreciated that although RRKM calculations are relatively insensitive to the detailed assignment of the activated complex (once AS$ has been fixed) they are very sensitive to the A factor of the reaction. For the present work we illustrate this effect by carrying out an RRKM calculation based on the Arrhenius parameters obtained from the log k against p - O e 5 plot.The results obtained are shown in fig. 3. There is now a good fit between the experimental results and the theoretical calculations although it is difficult to distinguish within experimental error whether this fit is better with the strong-collision assumption or with a stepladder model and a relatively large step size of 2000 cm-l (24 kJ mol-l). Note that in the study on P-butyrolactone a good fit between the experimental points and the RRKM calculations (strong collisions) was obtained with a virtually identical (low) A factor. Thus on the basis of a fit with theory one would be led to favour a relatively low A factor for the decomposition.Certainly as has been indicated earlier1 the A factor reported by James and Wellington (log A / s - l = 16.2) appeared to be too high based on thermochemical kinetic arguments and our present work supports this. Also, as found for P-butyrolactone, there would appear to be little doubt that the average energy transferred during a collision between an energized lactone molecule and a ‘bath gas’ lactone molecule is relatively large, which adds weight to the hypothesis that such efficient intermolecular energy transfer is related to the high polarity of these molecules. We clearly have contradictory findings and to some extent the choice is open at this point. The extrapolation which leads to a relatively low A factor is consistent with RRKM theory using either the strong-collision hypothesis or a stepladder model with large step sizes.The more favoured extrapolation gives a higher A factor with the RRKM fit much poorer (but it could be improved if a relatively large collision diameter for the reactant was employed, and indeed there are arguments that can be adduced for highly polar molecules being treated in this way). At present we prefer the extrapolation giving the higher A factor since we have preliminary data on more highly substituted lactones (which are thus studied nearer to their high-pressure limiting values) that are close to this ‘higher’ A factor. The relatively low energy of activation for the decomposition allows us to rule out a pathway involving a biradical intermediate or a biradical-like activated complex [for detailed arguments see ref.(l)]. This then makes the decomposition of lactones closely related to cyclobutanones rather than oxetanes. Not only are the A factors similar but also we now see that the methyl-substituent effect is almost identical. Thus in the case of the cyclobutanones methyl substitution in the 3 position leads to rate enhancement5 by a factor of just under 30. For oxetanes the appropriate comparison is with the 2-methyl compound. There is a slight complication in this case since such oxetanes can decompose by two pathways, but even taking the sum of these two1094 H. M. FREY AND I. M. PIDGEON channels the methyl derivative only reacts 50% faster than the unsubstituted parent.6 (This is very close to the effect of methyl substitution in cyclobutane.) For p- propiolactone and P-butyrolactone (3-methyl-p-propiolactone) at ca.250 "C the increase in rate caused by methyl substitution is a factor of 19. These arguments, which are of necessity circular, nevertheless do support the earlier suggestion' that the Arrhenius parameters of James and Wellington are too high. It is also perhaps pertinent to note that the rate constants calculated from our Arrhenius equation are smaller than those calculated from the Wellington study. Work that is under way on other substituted oxetanones should yield further information on these problems. We thank the referees for their helpful remarks. APPENDIX Input information for RRKM calculations. Collision diameter 5.3 A. FIG. 2 A S = 25.95 J K-l mol-l (corresponding to log A/s-l = 14.856). T = 559.12 K. Vibrational frequencies (molecular)/cm-l: 3028, 3001, 3000, 2935, 1882, 1477, 1427, 1319, Vibrational frequencies (complex)/cm-l: 3028, 3001, 3000, 2935, 1882, 1477, 1427, 13 19, 1199, 1184, 1139, 1093, 1046, 1006,924,891, 790, 746, 513,490 and 191. 1199, 1184, 1139, 1046, 790, 746, 513,490, 255, 250, 250 and 191. FIG. 3 A S = 16.53 J K-' mol-l (corresponding to log A/s-l = 14.37). T = 567.71 K. Molecular vibration frequencies as for fig. 2. Vibrational frequencies (complex)/cm-l: 3028, 3001, 3000, 2935, 1882, 1477, 1427, 1319, 1199, 1184, 1139, 1046, 790, 746, 513,490, 376, 375, 375 and 191. H. M. Frey and H. P. Watts, J. Chem. SOC., Faraday Trans. I , 1983, 79, 1659. T. L. James and C. A. Wellington, J. Am. Chem. Soc., 1969,91, 7743. B. S. Rabinovitch and K. W. Michel, J. Am. Chem. Soc., 1959, 81, 5065. I. Oref and B. S. Rabinovitch, J. Phys. Chem., 1968, 72, 4488. H. M. Frey and R. A. Smith, J. Chem. Soc., Perkin Trans. 2, 1977, 752. P. Hammonds and K. A. Holbrook, J. Chem. Soc., Faraday Trans. I , 1982,78, 2195. (PAPER 4/1410)

ISSN:0300-9599    

DOI:10.1039/F19858101087

出版商:RSC    

年代:1985

数据来源: RSC

摘要:

J . Chern. Soc., Furaday Trans. I, 1985, 81, 1095-1100 Alk yl-r adical-Chloride-ion Adduc t s Formed in the Radiolysis of Chloroalkanes An Electron Spin Resonance Study BY MARTYN C. R. SYMONS* AND IAN G . SMITH Department of Chemistry, The University, Leicester LE 1 7RH Received 30th August, 1984 Exposure of dilute solutions of chloromethane in [2H3]cyanomethane to W o y-rays at 77 K gave CH,- - -C1- adducts by electron capture. These were characterised by their e.s.r. spectra, which showed a slightly reduced proton hyperfine coupling (-22 G) and a clear quartet splitting from 35Cl and 37Cl [ A , , x 4 G and A , x (-)2 GI. On annealing, normal methyl radicals were formed irreversibly. Also, t-butyl chloride in tetramethylsilane or adamantane matrices gave two types of rotating t-butyl radicals, one having A(’H) = 22.7 G and the other having A(lH) = 21.1 G.The former is certainly normal t-butyl radicals and the latter, converted irreversibly to the former on annealing to ca. 180 K, is assigned to chloride-ion adducts with negligibly small isotropic coupling to chlorine. These features were broad at 77 K but did not exhibit well defined anisotropic splitting from chlorine nuclei. We conclude that, as with alkyl bromides and iodides, halide-ion adducts are formed on electron capture and that these are best viewed as ‘collision complexes’ or charge-transfer complexes, held together by the rigid matrices. They are not properly described as radical anions, and in our view their ready formation and low-temperature stability precludes the possibility that true radical anions are formed in condensed phases.The results show that the extent ofcharge transfer (ca. 4%) is less than that deduced for R’Br- adducts (ca. loo/;) or R’I- adducts (ca. 17%). The trend follows the ionization potentials of the halide ions, as expected. These results are also compared with those for the isoelectronic radicals H3NAC1, H3NABr and H3N21, which are clearly o* radicals rather than being halide-ion complexes of H,N+ radicals. Some years ago Sprague and Williams established that methyl radicals formed by electron capture by bromomethane in cyanomethane at 77 K are weakly complexed to bromide i0ns.l Later, we extended these observations to the iodides and pointed out that the fact that these ‘adducts’ are formed rather than radical anions is of considerable significance, since such anions are frequently postulated as being iinportan t reaction intermediates.Since then these adducts have been the centre of two controversies. One centred around the claim by Wood and Lloyd3 to have prepared P-bromo- and P-iodo-alkyl radicals in adamantane which exhibited very low isotropic hyperfine coupling to bromine and iodine nuclei. In view of our own work, in which we claimed to show that P-bromoalkyl radicals exhibit very large hyperfine coupling to bromine (,4,,, z 350 G),4 we suggested that these species were alkyl-radical-halide-ion adducts rather than being P-bromoalkyl radical^.^ Since this suggestion was firmly rejected,6 we endeavoured to establish our case more In the course of doing so, we established that the perpendicular hyperfine coupling to bromine and iodine is negati~e,~ and that the isotropic hyperfine coupling is close to zero.This small 10951096 ALKYL-RADICAL-CHLORIDE-ION ADDUCTS isotropic coupling is thought to be of major diagnostic value in distinguishing between true o* radical anions and adducts.1° The other controversy has been concerned with the nature of the bonding between carbon and halogen in these adducts. When comparing results for F,C'X- (where X is a halogen) radical anions1' and these adducts, we suggested that one of the factors contributing to the remarkable difference between the two electron-addition species is the fact that 'CF, radicals retain approximately the original degree of orbital hybridisation at carbon whereas 'CH, radicals are planar.l1 However, this point of view has been severely criticised,12 in our view unjustifiably.l0 A major factor emerging from this controversy is that, in addition to this change in shape and consequent rehybridisation, the presence of electronegative groups attached to carbon reduces the disparity in energy between the two orbitals comprising the o bond, thus increasing its strength and reducing the bond-breaking effect of the added electron.Thus, at least in condensed phases, it seems that for alkyl halides, as the carbon-halogen bond lengthens to accommodate the excess electron, the alkyl group flattens, thereby weakening the bond still further. This stretching and flattening is envisaged by us as being a continuous process leading finally to bond-breakage, with no potential minimum corresponding to the o* radical anion.Clarke1, has carried out some interesting ab initio calculations on electron capture by chloromethane which compare the process in the gas phase and in simulated condensed phases; these suggest that, in the latter, the process should indeed be purely dissociative. The key point is that the acceptor orbital for MeCl in the gas phase is very diffuse and in no sense resembles the C-C1 o* orbital, which is thereby 'protected'. This outer orbital is not available in condensed phases. EXPERIMENTAL Chloromethane (B.D.H.), [2H,]cyanomethane (N.M.R.), tetramethylsilane (B.D.H.), t-butyl chloride (B.D.H.) and [2H,,]adamantane (Merck, Sharpe and Dohme) were used as supplied.The purities of substrates and matrices were checked using n.m.r. spectroscopy prior to use. Solutions (ca. 1 % mol fraction) were prepared by volume and degassed using the freeze-thaw method. Samples in adamantane were prepared by recrystallizing the adamantane from the substrate and pressing the solid into a hard pellet. Samples were exposed to ,OCo y-rays at 77 K in a Vickrad cell at a nominal dose rate of 1 Mrad h-' for up to 4 h. E.s.r. spectra were recorded using a Varian E-109 spectrometer calibrated with a Hewlett-Packard 5246L frequency counter and a Bruker B-H 12E field probe standardised with a sample of diphenylpicrylhydrazyl (DPPH). Spectra at 4 K were obtained using an Oxford Instruments helium Dewar and those above 77 K were obtained using a variable-temperature system.RESULTS AND DISCUSSION The e.s.r. spectrum for chloromethane in CD,CN at 80 K after exposure to 6oCo y-rays at 77 K is shown in fig. 1 . The stick diagram is for the H,C- - -C1- adduct. The spectrum at 4 K was similar but less well resolved. Annealing to 155 K resulted in an irreversible loss of the features assigned to the adduct and growth of the normal quartet for 'CH, radicals [A(lH) = 22.8 GI. Central features are due to solvent radicals.14 The e.s.r. spectrum for t-butyl chloride in tetramethylsilane at 1 1 1 K is shown in fig. 2. At 77 K the inner components, assigned to But- - -C1- adducts in fig. 2, were too broad to detect, but those assigned to normal But' radicals remained narrow and showed no marked anisotropy. Spectra in [2H16]adamantane were comparable, againM.C. R. SYMONS AND I. G. SMITH 1097 1 32 60G I%+ H Fig. 1. First-derivative X-band e.s.r. spectrum for a dilute solution of chloromethane in [2H,]cyanomethane after exposure to 6oCo y-rays at 77 K showing outer [M,(lH) = +g] features for H,C- - -C1- adducts. The central features, due to solvent radicals, mask the &; lines. 9 32306 4J I I---- 1 +% I l l +"/2 + % - 3 - I I gain x 10 -But radical 1 But'/CI- adduct -Y2 Fig. 2. First-derivative X-band e.s.r. spectrum for a dilute solution of t-butyl chloride in TMS at 11 1 K after exposure to 6oCo y-rays at 77 K showing features assigned to normal t-butyl radicals and to But- - -C1- adducts.1098 ALKYL-RADICAL-CHLORIDE-ION ADDUCTS Table 1. E.s.r. parameters for R - - -X- adducts 35C1 SlBr 1271 7 , hyperfine coupling/Ga decrease T / K in A(lH) adduct (matrix) I1 I is0 a;(%>b A('H)/G (%> Me'/Cl- Me'/Br- Me'/I-" But' /C1- But'/Br- But'/I- But/C1- But/Br- But/I- H3N-Cle? f H3N-Brey f H3N-Ig 4 -2 57 - 28 I08 - 60 - - - - - - - - - - - - 43.3 13.5 426.4 106 530 200 0 0.3 -4 d 1.0 5.6 ca.6 < 1.0 6.7 7.0 23.4 212.8 310 4.0 -22 11.5 -21 24.0 -20.6 ca. 21 ca. 21 ca. 21 - - - 21.9 21.4 21.1 - - - 19.0 ca. 23 43.0 ca. 16 68.0 - h 4.4 8.7 10.5 - - - 3.5 5.7 7.0 - - - a 1G = Ref. (2). T; g values ca. 2.001. % p-orbital character obtained from data in ref. (15). Ref. (16). f 14N: All = 46.5 (Cl), A, = 9 (Cl), Aiso = 21.5 (Cl); Ref. Ref. (8) and (9). All = 51.5 (Br), A, = 36 (Br), Aiso = 41.2 (Br); Ail = 48.0 (I), A, = - (I), Aiso = - (I).(17). Not resolved. showing features for the adducts as well as the normal radicals. At 4 K the adduct features remained broad and chlorine hyperfine coupling remained poorly defined. (Unfortunately, we have not been able to prepare any methyl-radical adducts in adamantane, presumably because the methyl radical is able to escape from the large solvent cages.) Parameters derived from these spectra, together with those for some bromide and iodide complexes, are given in table 1. We also include data for some H,N'+ derivatives for comparative purposes. We justify our assignments as follows. (i) The quartet splittings seen on the M,(lH) = *# features can only reasonably be assigned to hyperfine coupling for 35Cl and ,'Cl (we would not be able to distinguish between these isotopes for such small splittings).Although parallel and perpendicular features are not well defined, the marked broadening of the M , = chlorine features indicates strong anisotropy, and simulations suggest that our values are accurate to ca. 10%. We assume that Al is negative by analogy with the well established results for the bromine and iodine adducts.'-g Absence of any chlorine coupling for the rotating complexes in adamantaneM. C. R. SYMONS AND I. G. SMITH 1099 and tetramethylsilane supports this assignment. (ii) There is a clear reduction in A(lH) for the adduct species relative to the 'free' radicals. This reduction is smaller than those observed for bromide and iodide adducts, as expected for chloride adducts (see below). (iii) It is normal experience that solutes form electron-gain rather than electron-loss centres in these matrices.Therefore, the only radicals we would expect to form at 77 K from the dilute solutes are 'CH, and H,C- - -C1- adducts and 'CMe, and Me,C*- - -Cl- adducts. All four of these products were identified. STRUCTURE There can be little doubt that these adducts have the structure proposed for similar alkyl-radical-anion adducts, namely the normal, essentially planar, alkyl radical adjacent to, but not bonded to, the anion. The small (ca. 4%) fall in Aiso(lH) is commensurate with the small anisotropic coupling to anion nuclei. For the chloride, our approximate estimate of of4 G leads, in the normal way, l5 to an estimated 3p, population of ca. 4%. In the limit of zero charge transfer there should still be a small dipolar hyperfine coupling to chlorine from the adjacent radical.If we assume an effective point-charge separation of ca. 2.6 A, then using standard equations we calculate a parallel coupling to 35Cl of 0.3 G. This is appreciably less than the experimental value, so it is necessary to invoke some charge transfer. As mentioned above, absence of significant isotropic coupling proves the absence of significant (T bonding. This again supports our theory that the interaction is a charge-transfer process. This involves electron donation from the anion and hence should increase as the ionization potential falls, namely in the order C1- < Br- < I-. That this requirement is fulfilled is established by the results in table 1.+ COMPARISON WITH RESULTS FOR 'NH, RADICALS In an interesting study Patten has shown that 'NH: radicals formed from NH; cations in irradiated ammonium chloride and bromide react with neighbouring anions to give H3NACl and H,N-Br radicals.lG We have extended this study to include H,N21 and a range of amine-halogen radicals, R3N-X.17 The results can only be interpreted in terms of B bonding between nitrogen and halogen, the SOMO being I \ \ I I I \ \ \ \ I I \ \ \ Br Fig. 3. Qualitative energy-level diagram designed to illustrate the effect of rehybridization of a methyl group and the difference between H,C- and H,N- bonding.1100 ALKYL-RADICAL-CHLORIDE-ION ADDUCTS the o* nitrogen-halogen bond. For the chloro-derivative the H,N- unit is nearly planar [p:s(14N) = 191 (see table I), but deviation from planarity increases steadily on going to the bromo- and iodo-derivatives.This is paralleled by a steady increase in calculated spin density on halogen, from ca. 19 % on chlorine to ca. 43 % on bromine and ca. 68% on iodine. These results nicely illustrate the postulate that as the bond stretches, so the H3N- unit (or H,C- unit) flattens. Thus the H3NACl results are relatively close to those expected for the limiting H,N'+- - -C1- structure found for the isoelectronic H,C- - -C1 system. The bonding is retained for the nitrogen derivatives because the orbitals involved in the 0 bonding are much closer in energy than for the carbon derivatives (fig. 3). OTHER SYSTEMS We have made a cursory study of other alkyl chlorides and, in all cases, there was evidence of adduct formation but spectra were no better defined than those illustrated herein.Also, several alkyl chlorides gave, after irradiation, definite adduct features which changed irreversibly to those for normal alkyl radicals on annealing. We conclude that dissociative electron capture occurs, that chloroalkane anions are not formed but that in these rigid matrices the radicals are trapped in the same cavity as the ejected chloride ions and there is a weak charge-transfer interaction detected by e.s.r. spectroscopy. We thank the S.E.R.C. for a grant to I.G.S. E. D. Sprague and F. Williams, J . Chem. Phys., 1971, 54, 5425. S. P. Mishra and M. C. R. Symons, J . Chem. Soc., Perkin Trans. 2, 1973, 391. R. V. Lloyd, D. E. Wood and M. T. Rogers, J . Am. Chem. Soc., 1974, 96, 7130; R. V. Lloyd and D. E. Wood, J . Am. Chem. Soc., I975,97, 5986. A. R. Lyons, G. W, Neilson, S. P. Mishra and M. C. R. Symons, J . Chem. SOC., Faraday Trans. 2, 1975, 363. D. J. Nelson and M. C. R. Symons, Tetrahedron Lett., 1975, 34, 2953. D. E. Wood and R. V. Lloyd, Tetruhedron Lett., 1976, 35, 345. ' I. G. Smith and M. C. R. Symons. J . Chern Soc., Perkin Trans. 2, 1980, 1362. M. C. R. Symons and I. G. Smith, J . Chem Soc., Perkin Trans. 2, 1981, 1180. M. C. R. Symons and I. G. Smith, J . Chem. SOC., Faradav Trans. 1 , 1981, 77, 2701. D. J. Nelson and M. C. R. Symons, Chem. Phys. Lett., 1977,47, 436. T. Clark, J. Chern. Soc., Chem. Commun., 1984, 93. M. C. R. Symons. Chemical and Biochemical Aspects of' Electron Spin Resonance Spectroscopy (Van Nostrand Reinhold, Wokingham, 1978). l6 F. W. Patten, Phys. Rev., 1968. 175, 1216. J. B. Raynor, I . Rowland and M. C. R. Symons. unpublished results. lo M. C. R. Symons, Chem. Phys. Lett., 1980, 72, 559. l 2 J. T. Wang and F. Williams. Chem. Phys. Lett., 1980, 72, 556. l 4 R. J. Egland and M. C . R. Symons, J . Chem. Soc. A, 1970, 1326. (PAPER 4/ 1504)

ISSN:0300-9599    

DOI:10.1039/F19858101095

出版商:RSC    

年代:1985

数据来源: RSC

摘要:

J. Chern. SOC., Furaday Trans. I , 1985, 81, 1 101-1 1 12 Reactivity of OH and 0- with Aqueous Methyl Viologen Studied by Pulse Radiolysis BY SONJA SOLAR, WOLFGANG SOLAR AND NIKOLA GETOFF* Institut fur Theoretische Chemie und Strahlenchemie der Universitat Wien and Ludwig Boltzmann Institut fur Strahlenchemie, Wahringerstrasse 38, A- 1090 Wien, Austria AND JERZY HOLCMAN AND KNUD SEHESTED Accelerator Department, Rise, National Laboratory, DK 4000 Roskilde, Denmark Received 10th May, 1983 The behaviour of aqueous MV2+ towards oxidizing radicals (OH and 0-) has been investigated in the pH range from 6 to 14 by means of pulse radiolysis. A semi-linear optimization method was applied for resolving the complex reaction mechanism. In the pH range from 6 to 8 the rate constant for attack by OH is k = (2.5 f0.2) x lo8 dm3 mol-' s-l.The resulting transient absorbs at Amax = 470 nm ( E ~ , ~ = 1600 f 70 m2 mol-l) and decays with 2k = (1.3 f0.2) x lo8 dm3 mol-l s-l. In strongly alkaline solutions (pH 2 13.8) the 0- radical anion reacts preferentially by hydrogen abstraction from the methyl group, k = (1.4f0.2) x lo9 dm3 mol-l s-l, forming a radical which then decays by reaction with OH- (k = 2.8 x 106 dm3 mol-l s-l) to produce a modified radical cation; this has absorption maxima at 392 and 605 nm (&3g2 = 4300 m2 mol-l, E,,, = 1500 m2 mol-l) and is relatively long lived. The remaining part (< 10%) of 0- attacks MV2+ at the ring carbon atom, k = (l.Ok0.4) x 10" dm3 mol-1 s-l, resulting in an 0- adduct, which has Amax = 470 nm ( E ~ , ~ = 2200+_ 100 m2 mol-l).Methyl viologen, MV2+ ( 1,l -dimethyl-4,4'-bipyridinium ; also called paraquat), is widely used as an electron relay in various devices for the photoinduced generation of hydrogen from water1 and is also a well known herbicide. Therefore, the reaction of the methyl viologen radical cation (MV*+) with oxygen has been studied exten~ively.~-~ Patterson et aL3 observed on pulse radiolysis of an aqueous air-saturated 2 x mol dm-3 MV2+ solution a transient absorption spectrum (25 p s after the end of the pulse) with A, at 470 nm ( E = 1250 m2 mol-l). It was assigned to an OH adduct of MV2+, formed with k(OH+MV2+) = 1.8 x los dm3 mol-l s-l. Recently we reported on multisite H attack on MV2+,536 and gave a preliminary overall rate constant k(OH+MV2+) = 4 x lo8 dm3 mol-l s - ~ .~ The present paper is concerned with the kinetic and spectroscopic characteristics of the transients produced by attack of OH and its dissociated form, 0-, on MV2+ in aqueous solution. These oxidizing intermediates can occur in different systems used for solar-energy utilization. The results may also be of interest for other pyridinium-like compounds and their behaviour towards oxidizing species. EXPERIMENTAL IRRADIATION FACILITIES The pulse-radiolysis experiments were performed at Rise, using the 10 MeV Linac (Haimson Research Corp. HRC-712, pulse length from 100 ns to 1 ps) with complete equipment for optical detection of transients in connection with a Nicolet Explorer I11 digital storage 11011102 REACTION OF OH AND 0- WITH METHYL VIOLOGEN oscilloscope and an on..line PDP 8 computer.' In addition, experiments were carried out on a 3 MeV Van de Graaff accelerator (type K, High Voltage Engineering Co., Burlington, pulse lengths from 200 ns to 4 p s ) at the Max Planck Institut fur Strahlenchemie, Mulheim a.d.Ruhr, F.R.G.It was equipped for fully automatic registration of intermediates in combination with an on-line PDP 11 computer.8 In both cases the applied dose was varied from 3 to 6 J kg-1 (0.3-0.6 krad) per pulse and the measured optical densities were normalized to 10 J kg-l (1 krad). PREPARATION OF SOLUTIONS All solutions were prepared with at least triply distilled water. Methyl viologen dichloride (reagent grade, B.D.H.) and NaOH and Ba(OH), (reagent grade, E. Merck) were used without further purification. The solutions (pH 6-14) were saturated with high-purity N,O in order to convert eiq into OH, k(eiq+N2O) = 9.1 x lo9 dm3 mol-l s - ~ .~ Using lop4 mol dm-3 MV2+ ca. 3% of eiq react with the substrate, whereas ca. 36% of eiq are scavenged by MV2+ when the concentration is 2 x lop3 mol drn-,. This was always taken into consideration. To avoid the influence of carbonate present in NaOH, Ba(OH), was added. The precipitated BaCO, was centrifuged and the supernatant solution was used for adjustment of the pH. It has previously been reported1* that MV2+ is unstable in strongly alkaline solutions, and so this was carefully reinvestigated. Using deoxygenated 2 x lo-, mol dmP3 MV2+ in 0.6 mol dm-3 NaOH (pH 13.8) at room temperature no significant change of the absorption spectrum, compared to that in neutral media, could be observed within 3 h. Preparation of alkaline solutions of MV2+ and performance of the experiments were hence completed in < 1 h.COMPUTER PROCEDURE All computer simulations and optimizations of the kinetic and spectroscopic parameters were performed on the CDC-Cyber computer at the University of Vienna, Austria. RESULTS AND DISCUSSION REACTIVITY OF MV2+ WITH OH Two series of experiments were performed in order to obtain the absorption spectrum resulting from attack of OH on MV2+. First, it was measured by using from lop4 to 2 x lop3 mol dm-3 MV2+ solutions containing 2.8 x lo-, mol dmw3 N20, which competes with MV2+ for eLQ [k(MV2+ +eLq) = 7.5 x 1O1O mol dmp3, ref.(5)].Taking e.g. 5.4 x mol dm-3 MV2+, ca. 87% of eiq (G = 2.35) are converted into OH radicals, while the rest form MV'+ (A,,, at 392 and 605 nm). Some of the H atoms produce MV'+ (G = 0.235) as well as H adducts and H,. Hence, the measured transient absorption spectrum A in fig. 1 is made up from several species. However, using previously reported data5*6 the contributions of both MV*+ and H adducts formed under these conditions were calculated and their sum is presented as spectrum B in fig. 1. By subtraction of spectrum B from A the corrected spectrum C is obtained, which represents only the transient formed by reaction of OH with MV2+ (with G = 3.7). It possesses an absorption maximum at 470 nm, in agreement with Patterson et a1.3 In order to confirm the small residues of the absorption bands at 392 and 600 nm a second series of experiments was carried out in which the solutions were saturated with oxygen (1.4 x rnol dm-3 O,), which acts as a scavenger for H atoms [k = 2.0 x lOlo dm3 mol-l s-l, ref.(l l)] 2nd partly for eiq [k = 2.0 x 1O1O dm3 mot1 s-l, ref.( 12)].Using 5.4 x mol dm-3 MV", immediately after the pulse ca. 50% of eiq form MV*+ and an absorption spectrum with the known absorption bands at 392 and 600 nm appears (spectrum D, insert, fig. 1). However, after ca. 50 ps MV'+ is reformed into MV2+ by electron transfer to 0,: MV'++O, -+ MV2++0; with k = 7.7 x lo* dm3 mol-1 s - ~ , ~ whereas the absorption spectrum of the longer-lived transient at 470 nm resulting from attack of OH on MV2+ attains a maximumSOLAR, SOLAR, GETOFF, HOLCMAN AND SEHESTED 0.08 , I 1103 LOO 500 600 700 A/nm Fig.1. Absorption spectra of transients produced from 5.4 x lop4 mol dm-3 MV2+ in the presence of 2.8 x lop2 mol dm-3 N,O at pH 7 (applied dose: 3 4 J kg-' per 200 ns pulse; OD values normalized to 10 J kg-l). (A) Spectrum measured 50 ps after pulse, (B) calculated contribution of MV'+ formed by partial reaction of e& and H with MV2+ and (C) difference spectrum (C = A - B) representing OH adducts on aromatic ring carbon. Insert: (D) spectrum obtained after end of pulse by irradiation of 5.4 x mol dm-3 MV2+ in the presence of 1.5 x lop3 mol dm-3 0, at pH 7 and (E) measured spectrum 50 ps after end of pulse. (spectrum C , fig. 1). Under these conditions, however, only ca. 66% of the OH radicals (G = 1.9) are involved in the production of this transient and the rest are consumed in radical-radical reactions.Hence, the absorptions at 392 and 600 nm (fig. 1) can be attributed to small amounts of H adducts which absorb at these wavelength^.^^^ From the OD value at 470 nm [taking G(0H) = 3.71 a mean value for the extinction coefficient was calculated, E,,, = 1600 70 m2 mol-l. This value is substantially higher than that previously reported, E,,, = 1250 m2 m01-l.~ By analogy with the multisite reactivity of H atoms with MV2+,57 it was expected that OH radicals should also attack the substrate molecule at different positions. Principally, there are three possibilities: formation of OH adducts on a ring carbon atom (Rl), hydrogen abstraction from the -CH, group (R,) and addition of OH to the N atom (R,): H 3 C - & m & - C H 3 + OH (1) MV2+1104 REACTION OF OH AND 0- WITH METHYL VIOLOGEN Table 1.Reactions and rate constants considered in the pulse radiolysis of the MV2+ systm in neutral and basic aqueous solutions no. reaction 1 2 3 4 5 6 7 8 9 10 1 1 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 Ti: MV2+ +OH R, 2R, -+ products 2R, -+ products R, + OH- -+ R, MV2+ + 0- *-r E: + OH- R, --+ products MV2+ + H MV*+. H + 71 MV-2+. H 2MVW2+ - H -+ products MV" . H+ -+ MV'+ + H + MV2+ + e i q -+ MV'+ OH +OH -+ H,O, OH+H +H,O OH +OH- -+ 0- + H,O OH+O--+HO; OH+HO; --+ H,O+O; OH+O; -,OH-+O, O-+H,O-+OH+OH- eiq + eiq -+ H, + 20H- e;tq + OH -+ OH- elq+H -+ OH-+H, e;tq+H+ -+ H+H,O eiq + HOT -+ 0- + OH- + H,O elq + 0- -+ OH- + OH- e;q+N,O-+N,+OH+OH- H+H+H, H +OH- -+ eiq + H,O 0-+o- +o;- rate constant /dm3 mol-l s-l s-I [ref.( 14) and (1 6)] k , = ? k , = ? k , = ? k, = ? k , = ? k , = ? k , = ? k , = ? ki = ?" k,, = 3.1 x 10' k,, = 2.9 x 10, k;, = 2 x 104" k,, = 7.5 x 1O'O 2k,, = 1.2 x 10" k,, = 2 x 1 O 1 O k,, = 1.4 x 1O'O k,, = 1.8 x 10" k,, = 7.5 x lo9 k,, = 1 x 10'0 k,, = 2 x 106 2k,, = 1.2 x 1O'O k,, = 2.2 x 10'0 k,, = 2.5 x 1O1O k,, = 2.2 x 1 O 1 O k,, = 2 x 1010 k,, = 9.1 x lo9 2k,, = 2.3 x 10" k,, = 1.5 x lo7 2k,, = 2 x 10' 2k,, = 6 x 10, k,, = 3.5 x 109 a k' is given in s-l.Formation of an OH adduct [reaction (l)] is considerd to be the predominant process, as can be concluded from the data concerning the reaction of OH with pyridine and aromatic corn pound^.^^-^^ The R, radical has no absorption in the experimentally accessible range (vide infra).Finally, as previously demonstrated for aliphatic amines,16 pyridones13 and pyridines," attack of OH does not occur at the quaternary nitrogen atom (formation of R3). Therefore, it is assumed that the absorption band at 470 nm represents species R, only. The experimental k(OH + MV2+) value, determined by following the rate of formation of the absorption at 470 nm, was corrected for consumption of OH in radical-radical reactions. A rate constant of k = 2.5 x lo8 dm3 mol-l s-l was obtained.SOLAR, SOLAR, GETOFF, HOLCMAN AND SEHESTED 1105 300 400 500 600 700 X/nm Fig. 2. Absorption spectrum of transients produced by reaction of 0- with 1 x lop3 mol dm-3 MVZ+ in 0.6 mol dmp3 NaOH, saturated with N,O.Range A, 20 ,us and range B, 7 ,us after end of pulse (applied dose: 4-5 J kg-l per 400 ns pulse; OD values normalized to 10 J kg-l). Insert : change of the relative OD values, OD,, as a function of time at A, 392 and B, 470 nm (solution: 5.4 x mol dm-3 MV2+ in 0.6 mol dm-3 NaOH, saturated with N,O). REACTIVITY OF MV2+ WITH 0- Investigations of the MV2+/0- system were performed at pH b 13.8 (0.6 mol dm-3 NaOH) in the presence of 2.8 x lop2 mol dmp3 N,O at 20 "C. Under these conditions the primary species of water radiolysis (OH, H and eiq) are converted into the 0- radical anion within 100 ns of the end of the pulse, since: N,O +eLq -+ N, +OH- + OH[k,, = 9.1 x lo9 dm3 mol-1 s-l, ref.(8)] (28) H+OH- + eZq[k3,, = 1.5 x lo7 dm3 mol-1 s-l, ref.(l l)] (30) (32) Because of the high reactivity of MV2+ with eCq [k = 7.5 x 1Olo dm3 mol-1 s-l, ref.(5)] competition between reaction (14) and reaction (28) (see table 1) takes place, depending on the substrate concentration.Using loL3 mol dm-3 aqueous MV2+ solution, saturated with N,O, ca 23% of the eiq produce the radical cation MV+ (G = 0.8) and the remaining eiq together with the OH radicals and H atoms are converted into 0- (G = 5.4). The measured absorption spectrum under these con- ditions is presented in fig. 2. A comparison of the decay kinetics of the transient indicated the formation of at least two kinds of species, one absorbing in the ranges 30WlOO nm (A, = 392 nm) and 500-760 nm (A,,, = 605 nm) (denoted as A in fig. 2) and a second in the range 420-500 nm (denoted in B in fig.2). The decay of the optical densities for up to 100 ps after the end of the pulse of both types are presented in the insert of fig. 2. The absorption bands A in fig. 2 are practically the same as those of MV+ [see ref.(5) and references therein]. Since the contribution of MV+ in this case is only ca. 13% of the total adsorption, it can be concluded that another transient with similar structure is formed. The small absorption at 470 nm is attributed to an 0- adduct (R4). OH 0- + H+[ppK = 1 1.9, ref.( 1 8)J.1106 REACTION OF OH AND 0- WITH METHYL VIOLOGEN As 0- reacts preferentially with the side group of the methylated abstraction of an H atom,'? 1 4 9 l9 the following reactions are suggested: H I C - ~ ~ N - ~ H ~ + OH- MV2+ + O - - T - - - - R2 R4 b- Concerning the assignment of the R, transient, there are two indications benzenes by (7) (8) that its main absorption-bands cannot be identicai to those observed at 392 and 605 nm : (a) radicals of this type are not expected to have strong absorptions above 320 nm71 1 4 9 l5 and (b) analysis of the kinetic data for the bands at 390 and 600 nm obtained by varying [MV2+] (series A, fig.3) and [OH-] (series B, fig. 3) yielded a rate constant for the build-up which is dependent on the pH of the solution. This can be explained by a consecutive reaction of R, with OH-, where the measured absorptions at 392 and 605 nm are attributed to the resulting R, transients (see also table 1): In very strongly alkaline solutions reaction (7) is the rate-determining step (fig. 3A) and hence the build-up kinetics is strongly dependent on [MV2+].From the pseudo- first-order rate constants, kA, the overall rate constant for attack of 0- on MV2+ can be determined : k(0- + MV2+) = k, + k, = 1.5 x lo9 dm3 molt1 s-l. The same value is obtained from the pseudo-first-order build-up at 470 nm, which corresponds to absorption of the 0- adduct (R4). The deviation from linearity (fig. 3A) is explained by the fact that with increasing MV2+ concentration the rate of the consecutive two-stage reaction becomes controlled by reactions (6) and (7). In fig. 3B [MV2+] is sufficiently high that reaction (6) is the rate-determining step and therefore a strong dependence on [OH-] is observed. Based on this fact the rate constant k, was determined at 390 and 600 nm to be k,(R,+OH-) = 2.8 x 10, dm3 mol-1 s-l.In this case the deviation from linearity is explained by the increasing rate of reaction (6), whereby its predominance over reaction (7) is reduced. Further support for the assignment of the absorptions at 392 and 605 nm to species R, is provided by the similarity of the absorption spectrum of the radical (R6) previously reported :,O HO-CH2-CH2 A m - C H 2 - C H 2 - OH . R6 Usually, replacement of H or OH by -CH20H in a side chain of an aromatic compound does not have a significant influence on the absorption spectrum.SOLAR, SOLAR, GETOFF, HOLCMAN AND SEHESTED 1107 t " I v) 4 1 5 1 d Y t 0 5 10 0 0.2 0.4 0.6 [ MVZ+] / 1 O4 mol dm-3 [ NaOH] /mol dme3 Fig. 3. (A) Determination of the rate constant for the reaction of 0- with MV2+ at 390 nm (solution: 1 x lop4 to 1 x mol dm-3 MV2+ in 0.6 mol dmp3 NaOH, N,O saturated; applied dose: 2-3 J kg-l per 200 ns pulse).(B) Determination of the rate constant for the reaction of radical R, with OH- at 390 nm (solution: 2 x lop3 rnol dm-3 dm-3 MV2+, 0.024.60 mol dmp3 NaOH, N,O saturated; applied dose: 2 J kg-1 per 200 ns pulse). k , = (k, + k,) [MV2+] and k , = k, [OH-]. Therefore, the reported values of E of radical R, at 392, 470 and 605 nm were used as an approximation for the unknown species R, in the computation procedure. The value of E for MV*+ can also be used in the computation. Taking into account all the experimental results and considerations for the possible attack of 0- on MV2+ the following reaction scheme is suggested: + OH- - Rs k6 w products.(9) k9 The occurrence of reaction (5) is also indicated by the influence of dose rate on the build-up kinetics of species R,. The decay kinetics of species R, (at ;1 = 470 nm) showed dependence on OH- in the pH range from 12 to 14. A change from pseudo-first-order to mixed-order kinetics was observed with decreasing [OH-]. The nature of the products could not be specified. STUDIES IN THE pH RANGE FROM 9 TO 12 In the pH range from 9 to 10.5 a strong decrease in optical density of the 470 nm band (curve A,, fig. 4) and a corresponding enhancement of the initial optical density at 392 nm (curve B, fig. 4), as well as at 605 nm, were established. The measured OD1108 B ? 392 nm measured -I ; Q- o.o' 0 .O' REACTION OF OH AND 0- WITH METHYL VIOLOGEN 392 nm /- .- - 605 nm L70 nm 0.08.0.06- 0.04- 0.02 - 0 - I I I I I I 1 8 9 10 11 12 13 1L PH Fig. 4. Change of the relative OD values as a function of pH at 470 nm (measured and computed curve) and at 392 nm (solution: 5.0 x lop4 mol dm-, MV2+, saturated with N,O). Insert: oscilloscope traces of relative OD values at 392, 605 and 470 nm (solution: 2.65 x lop4 mol dmP3 MV2+, saturated with N,O, pH ca. 9.35). values are independent of the substrate concentration from 5 x lop4 to dm-3 MV2+ (saturated with N,O) in unbuffered as well as in borate-buffered solutions. The build-up rates observed at all three wavelengths are proportional to [MV2+]. The decrease of the OD,,, values cannot be substantiated by 0- reactions, since in the pH range from 9 to 10.5 0- is not involved in the reaction mechanism, even though its rate constant with MV2+ is six times higher than that of OH.For comparison, a computed pH dependence of OD,,, due to the OH/O- equilibrium is presented in fig. 4, curve A,. Although the course of the curves is similar, curve A, is shifted by ca. 1 pH unit. A possible acid-base equilibrium of the MVz+ can be excluded, since the substrate spectrum does not change even in strongly alkaline solution. Although it seems that OH- is involved in the processes, causing the decrease in OD4,,, a satisfactory reaction mechanism for this pH range cannot be put forward. However, one can suggest a reaction between an OH adduct (Rl, A = 470 nm),and an OH- species, leading to a transient product P, with absorption maxima at 392 and 605 nm.The structure of P would be aconjugated system of double bonds like MV*+, but with an E value ca. 5 times smaller: OH adduct (R,) + OH- pK z 9.7 p>-<[.g. H,C-N +Wm) (331 H OHSOLAR, SOLAR, GETOFF, HOLCMAN AND SEHESTED 1109 The transient P may convert by water splitting into an MV'+-like radical. This process is indicated by an additional slow build-up of the 392 nm absorption band on a time- scale of 50-250 ps after the end of the pulse (see insert to fig. 4), where the 470 nm absorption is already slowly decreasing with mixed-order kinetics. This build-up is dependent on neither pH nor [MV2+]. Taking into account pK(32) z 9.7 and that the equilibrium (32) is attained immed- iately after the pulse, the lower limit of k3, (forward) of ca.5 x 1 O 1 O dm3 mol-1 s-l can be estimated. This rate constant is unexpectedly high, but it might be explained by the two positive charges on the R, radical. Using 5 x lo-, mol dm-3 MV2+ in the presence of from lop3 to 10-1 mol dm-3 C1-, OD,,, increases with C1- concentration at the expense of OD,,,. This is consistent with a kinetic salt effect on a reaction between two oppositely charged species. COMPUTER SIMULATIONS In order to check the postulated reaction mechanisms and to resolve the kinetic and spectroscopic parameters of the superimposed transients, a previously described optimization procedure was 22 The calculations are based on the time- dependent change (1-2000 ps after the end of the pulse) in the optical densities (at A = 392,470 and 605 nm) of the irradiated solutions for at least three different MVz+ concentrations [( 1 - 5 ) x lo-, mol dmP3] in the pH ranges from 6 to 8 and from 13 to 14.The optimization method used renders the splitting of the unknown model parameters into linear spectroscopic parameters and non-linear kinetic parameters. Therefore, only initial information about the rate constants in the reaction model, but no spectroscopic data on the transients, is needed in the iterative procedure.21 All the important simultaneously occurring reactions considered in the computation procedure are compiled in table 1 . Since reactions (6)-(9) are insignificant in the pH range from 6 to 8 they are not considered in the first series of computations.Similarly, reactions (1)-(5) were neglected for the experiments performed at pH > 13. The experimentally determined rate constants for formation and decay of the species, produced by the attack of OH or 0- on methyl viologen, were taken as the starting parameters in the iterative procedure. The above reaction mechanisms were confirmed by the simulation procedure. The individual kinetic and spectroscopic parameters obtained for the intermediates produced by attack of OH as well as 0- on MVz+ are summarized in table 2. As an illustration of the ability of the method the resolved superimposed formation kinetics of species R, and R, at 470 nm, formed at pH 3 13, is shown in fig. 5. Note that at all investigated wavelengths the computed (CTA) and measured total absorptions (MTA) showed satisfactory consistency.This indicates that the suggested reaction mechanism (table 1) is in agreement with the experimental results. Characteristic absorptions for the hydrogen-abstraction product, R, (at pH > 13), have not been observed in the investigated wavelength range from 320 to 700 nm. The changes in all important transient concentrations as a function of time are simulated in fig. 6. It is obvious that species R,, produced at pH > 13, are disappearing within 4 ps under the given conditions, converting into the long-lived R, transients. Simulation computations demonstrate that a decrease of MV2+ concentration leads to a longer apparent life-time, but on the other hand to smaller radical concentrations. However, as mentioned above, the main absorption bands of R,-type radicals are expected to be below 320 nm, hence their contribution to the observed absorption is negligible.The computed rate constants for the addition of 0- to the aromatic ring (k,) and for hydrogen-atom abstraction (k,) are in accordance with those previously reported for different methyl-substituted benzene derivatives.'~ l9 The values of k for addition1110 REACTION OF OH AND 0- WITH METHYL VIOLOGEN Table 2. Rate constants for formation and decay and absorption characteristics of transients produced by the reaction of OH or 0- with MV2+ (pH 6-14) absorption characteris tics formation kinetics transient decay A, emax transient pH range (k/dm3 mol-l s-l) (2k/dm3 mol-' s-l) /nm /m2 mol-l Rl 6-8 k, = ( 2 . 5 f 0 .2 ) ~ 10' 2k4 = ( 1 . 3 f 0 . 2 ) ~ 10' 470 1600+70 R2 13-14 k, = (1.4k0.2) x lo9 ki = (2.8 + 0.3) x loea < 320 - R4 13-14 k8 = (1.0k0.4) x 10' ki = ( 1 . 3 f 0 . 3 ) ~ lWa 470 2200f100 2k5 = (2.0f0.4) x 10' R5 13-14 ki = (2.8k0.3) x lo6" (long lived) 392 43OOb 605 1 5OOb a k' is given in s-l. Taken from Farrington et a1.,20 based on the similarity of species R, of MV2+ with those of bipyridyl derivatives. 0 10 20 30 40 50 tlW Fig. 5. Resolved optical densities as a function of time at 470 nm for computed formation of transients R, (0- adduct on ring carbon) and species R, in comparison with the measured (a) and computed (-) total absorption (solution: 5.4 x lop4 mol dm-3 MV2+, 2.8 x lop2 mol dm-3 N20, pH 13.8).SOLAR, SOLAR, GETOFF, HOLCMAN AND SEHESTED 1111 3.0 2.5 m 'E 2.0 a - E 2 1.5 v) .- p! I 1.0. 0.5- I I I I I - 6- \ *. 0- \,".. 2 4 6 8 10 '.... 1 1 1 1 1 1 1 1 1 1 tlPS Fig. 6. Change of transient concentration as a function of time (solution: mol dm-3 MV2+, 2.8 x mol dm-3 N,O, pH 13.8). of 0- to an aromatic ring are usually of the same order of magnitude (k z lo8 dm3 mol-1 s-l), whereas the values of k for addition of OH to aromatic rings vary, depending on the nature of the substrate: k(OH+phenol) = 1.4 x 1O1O dm3 mol-l s-l [ref.( 18)] (strongly activated ring system), k(OH +4.4-bipyridine) = 5.3 x lo9 dm3 mol-l s-l [ref.(l8)] (deactivated ring system) and k(OH+MV2+) = (2.5 0.3) x loa dm3 mo1-1 s-l (strongly deactivated ring system), (see table 2). This is in accordance with the high electrophilicity of OH radicals and the rather less electrophilic behaviour of O-.37 199 2 3 9 24 CONCLUSIONS The investigations reported in this paper were initiated by the fact that MV2+, when acting as an electron mediator in various homogeneous systems for the production of hydrogen from water by solar energy, is decomposed.Since in such devices H and OH (and, under certain conditions, 0-) can appear as intermediates, it was important to investigate their reactivity towards MV2+. As previously shown5* multisite attack of H on MV2+ leads to partial decomposition. OH radicals (with a similar reaction rate as H atoms towards MV2+) and 0-, which is even more reactive, also cause destruction of the substrate. Two of us (N.G. and S.S.) thank Prof. Dr D. Schulte-Frohlinde, Max Planck Institut fur Strahlenchemie, Mulheim a.d.Ruhr, F.R.G., for permission to use the1112 REACTION OF OH AND 0- WITH METHYL VIOLOGEN pulse-radiolysis facility, the Riss National Laboratory (Roskilde, Denmark) for financial support and the Bundesministerium fur Wissenschaft und Forschung, Austria, for financial assistance. N. G. acknowledges a fellowship from the Govern- ment of Denmark. We thank Dr E. Bjergbakke for helpful discussion. The valuable help given by Miss Hanne Corfitzen and Mr Torben Johansen (Riss) as well as by Dipl. Phys. F. Schworer, Mr K-H. Toepfer and Mr F. Reikowski (MPI fur Strahlenchemie) is greatly appreciated. N. Getoff,K. J. Hartig,G. Kitte1,G. A. PeschekandS. Solar, HydrogenasEnergy Carrier; Production, Storage and Transport (in German) (Springer-Verlag, Vienna, 1977).J. A. Farrington, M. Ebert, E. J. Land and K. Fletcher, Biochim. Biophys. Acta, 1973, 314, 372. L. K. Patterson, R. D. Small, Jr and J. C. Scaiano, Radiat. Res., 1977, 72, 218. E. J. Nanni Jr., Ch. T. Angelis, J. Dickson and D. T. Sawyer, J. Am. Chem. Soc., 1981, 103,4268. S . Solar, W. Solar, N. Getoff, J. Holcman and K. Sehested, J. Chem. Soc. Faraday Trans. I , 1982, 78, 2467. S. Solar, W. Solar, N. Getoff, J. Holcman and K. Sehested, J. Chem. Soc., Faraday Trans. I , 1984, 80, 2929. K. Sehested, H. Corfitzen, H. C. Christensen and E. J. Hart, J. Phys. Chem., 1975, 79, 310. N. Getoff and F. Schworer, Radiat. Res., 1970, 41, 1. E. Janata and R. H. Schuler, J. Phys. Chem., 1982,86, 2078. lo J. A. Farrington, A. Ledwith and M. F. Stam, Chem. Commun., 1960, 259. l1 M. Anbar, Farhataziz and A. B. Ross, Selected Spec@ Rates of Reactions of Transients from Water in Aqueous Solution, II. Hydrogen Atom (National Bureau of Standards, Washington, D.C., 1975). l 2 M. Anbar, M. Bambenek and A. B. Ross, Selected Specific Rates of Reactions of Transients from Water in Aqueous Solution, I. Hydrated Electron (National Bureau of Standards, Washington, D.C., 1973). l3 S. Steenken and P. O’Neill, J. Phys. Chem., 1978,82, 372. l4 H. C. Christensen, K. Sehested and E. J. Hart, J. Phys. Chem., 1973, 77, 983. l 5 J. Holcman and K. Sehested, J. Phys. Chem., 1977, 81, 20. l6 N. Getoff and F. Schworer, Radiat. Phys. Chem., 1973, 5, 101. l7 S. Steenken and P. O’Neill, J. Phys. Chem., 1979, 83, 2407. l8 L. M. Dorfman and G. E. Adams, Reactivity of the Hydroxyl Radical in Aqueous Solutions (National l9 P. Neta and R. H. Schuler, Radiat. Res., 1975, 64, 233. 21 S. Solar, W. Solar and N. Getoff, J. Chem. Soc., Faraday Trans. 2, 1983, 79, 123. 22 S. Solar, W. Solar and N. Getoff, Radiat. Phys. Chem., 1983, 21, 129. 23 M. Simic and M. Z. Hoffman, J. Phys. Chem., 1972,76, 1398. 24 N. V. Raghavan and S. Steenken, J . Am. Chem. Soc., 1980, 102, 3495. Bureau of Standards, Washington, D.C., 1973) J. A. Farrington, M. Ebert and E. J. Land, J. Chem. Soc., Faraday Trans. 1, 1978, 74, 665. (PAPER 4/ 1649)

ISSN:0300-9599    

DOI:10.1039/F19858101101

出版商:RSC    

年代:1985

数据来源: RSC