摘要:

J . Chem. SOC., Furaduy Trans. I, 1987, 83, 1253-1259 Free Energies, Enthalpies and Entropies of the Monothiocyanate Complex Formation of Trivalent Metals in Dimethyl Sulphoxide Solutions Danuta Puchalska,* Waclaw Grzybkowski and Danuta Wojcik Department of Physical Chemistry, Technical University of Gdarisk, 80-952 Gdarisk, Poland Formation of the InSCN(DMSO);+ complex in DMSO solution has been studied spectrophotometrically and calorimetrically and the complete thermodynamic characteristics, AGO, AH" and AS", of the complex-forming reaction have been derived. These results and analogous thermodynamic parameters obtained previously for AlSCN(DMSO)E+, GaSCN(DMS0),2+ and FeSCN(DMSO);+ complex formation are discussed in terms of factors controlling the stability of the complexes. It has been found that the entropy effect is responsible for the stability of the monothiocyanate complexes of A P , GaI", 1n1I1, 1n1I1 and FelI1 in DMSO solutions.The purpose of this paper is to determine the factors which control the stability of monothiocyanate complexes of trivalent metal cations in DMSO solution. It has been shown in previous papers that aluminium(II1) and gallium(1rr) form monothiocyanate complexes of the MSCN(DMSO)i+ type analogous with the FeSCN(DMSO):+ complex.'? Our preliminary results indicate that the entropy change is the main 'driving' force for their formation. In the present paper we report the results obtained for the indium(1rr) system. The free-energy changes have been determined spectrophotometric- ally using an indicator method recently developed in our laboratory, while the enthalpies have been determined by direct calorimetric measurements.Experimental Hydrated indium(Ir1) perchlorate was obtained by dissolving metallic indium in hot perchloric acid followed by repeated crystallizations from water. The solid product was then recrystallized several times from anhydrous DMSO. The stock solution of In(C10,), in DMSO was analysed for indium gravimetrically using 8-hydroxyquinoline. The remaining experimental procedure as well as the measurement techniques concerning the spectrophotometric and calorimetric works were as previously described.'. Spectrophotometric and Calorimetric Study of Indium(I1I) Monothiocyanate Complex in DMSO Solutions The spectrophotometric procedure used in the present study is identical with that applied previously for investigation of the corresponding aluminium and gallium complexes.The treatment consists of studying the electronic absorption spectrum of iron(rr1) for the equimolar mixtures of iron(111) and indium(rr1) perchlorates in the presence of a small addition of complexing agent, i.e. KSCN. The spectral effects induced by variation of composition of the equimolar mixtures are studied and evaluated in terms of the formation of the monothiocyanate complexes. Making use of the equimolar mixtures we accomplish constancy of ionic media, eliminating the problems arising from the variation of the activity coefficients. 12531254 Monothiocyanate Complex Formation 1 400 450 500 X/nm Fig. 1. Absorption spectra of iron(m) in DMSO solutions at 25 "C.(1) Fe(ClO,),-KSCN (2.66 and 1.77 mmol dm-,, respectively). (2-7) Fe(C10,),-In(C10,),-KSCN. Concentrations of Fe(C10,), and In(ClO,), are, respectively, (mmol dm-,): (2) 1.98 and 0.67, (3) 1.59 and 1.07, (4) 1.31 and 1.35, (5) 0.99 and 1.69, (6) 0.65 and 2.02, (7) 0.22 and 2.44mmol dm-3, while the KSCN concentration is constant (1.77 mmol dm-3). It has been well established that iron(m) and indium(rI1) perchlorates dissolved in DMSO exist exclusively as the complex electrolytes of the M(DMSO)i+ - 3C10, t ~ p e . ~ ~ Owing to the high excess of the complex-forming cations we can assume that only the two following equilibria are established in the Fe(ClO,),-In(C10,),-KSCN solution : Fe(DMS0),3+ + SCN- FeSCN(DMSO)E+ + DMSO (1) and In(DMSO)i+ + SCN S InSCN(DMS0),2+ + DMSO.(2) The equilibrium concentrations of both monothiocyanate complexes are determined by the difference of their stabilities and composition of the equimolar mixtures. The intensity of the absorption band with the maximum at 450 nm ascribed to the FeSCN(DMSO)t+ complex is the measured solution property depending on the distribution of the SCN- anion between the FeSCN(DMSO)E+ and InSCN(DMSO)E+ complexes. The value of the mean molar absorption coefficient of iron(m) is directly related to the mole fraction of iron(w) present in solution in the form of the FeSCN(DMSO);+ complex. It has been shown previously that substitution of Al(ClO,), or Ga(ClO,), for an equivalent amount of Fe(ClO,), in the corresponding equimolar mixtures results in a marked increase of the mean molar absorption coefficient of iron(1u) at the wavelength typical for the FeSCN(DMS0),2+ complex.The effect indicates an increase of the moleD. Puchalska, W. Grzybkowski and D. Wdjcik 1255 200c - I E - I I 150C rn E -a IUJ --. 100f I I I I I I I I I Fig. 2. Dependence of the mean molar absorption coefficient of iron(II1) at 450nm on the composition of equimolar mixtures. The total concentrations of both metal perchlorates in equimolar mixtures and KSCN concentrations are, respectively : (a) In(C10,),-Fe(C10,),, 2.66 and I .77; (b) Ga(C10,),-Fe(C10,),, 3.1 1 and 0.63; (c) A1(C10,),-Fe(C10,),, 3.39 and 0.163 mmol dm-3. fraction of iron(rI1) existing as the monothiocyanate complex. This could only result from the fact that aluminium(m) and gallium(rr1) form monothiocyanate complexes less stable than those of iron(m).The opposite effect is observed for the Fe(C10,),-In(C10,), system. Fig. 1 shows the spectra of iron(m) for one of the equimolar Fe(C10,),-In(C10,),-KSCN series along with the spectrum predicted for the Fe(ClO,),-KSCN solution of the same formal ionic strength. As is seen, the substitution of In(C10,), for Fe(ClO,), (curves 2-7) brings about a distinct decrease in the mean molar absorption coefficient of iron(m), clearly indicating a decrease of the mole fraction of iron(II1) present as the FeSCN(DMSO)i+ complex. It follows that the complex of iron(Ir1) is less stable than the corresponding thiocyanate complex of indium. To illustrate the discussed effects more clearly we present in fig.2 the variation of the mean molar absorption coefficient of iron(m) at 450 nm with the composition for one of the Fe(ClO,),-In(ClO,), equimolar mixtures along with corresponding plots for the systems studied previously. Analogous results were obtained for seven series of the Fe(ClO,),-In(C10,),-KSCN equimolar mixtures differing in the total concentration of the metal perchlorates. In order to determine the stability constant of the InSCN(DMSO)t+ complex the spectrophotometric data obtained for the seven equimolar series have been evaluated by means of iterative and successive approximation procedures, both described in detail in previous papers.l9 The derived data for all equimolar series are listed in table 1. The1256 Monoth iocyanate Complex Format ion Table 1.The average values of ‘medium’ stability constants of InSCN(DMSO)f+ complex in equimolar mixtures of Fe(CIO,), and In(CIO,), of total average concentrations C , along with average KInlFe and (Qin/Qfe) values derived from iterative and successive approximation methods, respectively seriesa successive iterative method approximation method C M G S C N /mol dm-3 /mol dm-, log Qin KInlFe log Qin (Q:n/QTe> 0.00052 0.0001 0 3.80 4.25 4.53 5.99 0.00 1 18 0.00026 4.08 5.35 4.38 5.65 0.00208 0.00053 3.61 4.45 4.23 5.18 0.00560 0.001 16 3.10 4.80 4.04 5.85 0.00732 0.001 80 2.99 4.60 3.97 5.81 0.01 121 0.00210 2.40 3.75 3.85 5.88 0.03458 0.00738 1 2.22 4.60 3.49 5.99 __ - ~- ~ ~ __ a Each series consists of at least five equimolar mixtures.only information required in the iterative procedure is the equilibrium concentration of the FeSCN(DMSO)E+ complex, which can be easily calculated from spectrophotometric data on the basis of known molar absorption coefficients of FeSCN(DMSO):+ and Fe(DMSO)i+ complexes. This procedure was applied recently for the Fe(ClO,),- Ga(ClO,), system2 and yields two final values for each of the equimolar series. One of them is the quotient of the equilibrium concentration for the respective complex-forming reaction2 called the ‘medium’ stability constant and defined as [InSCN(DMSO)i+] = [In(DMSO)i+] [SCN -1‘ (3) The second value is the ratio of the ‘medium’ stability constants of the corresponding complexes of indium(m) and iron(Ir1) defined as Taking into account the assumption underlying the applied method, that the respective activity coefficients are independent of the nature of the metal cations and constant within the equimolar series, the relative stability constant KIn/Fe is equivalent to the ratio of the ‘thermodynamic’ stability constants.Thus the value should be constant for all investigated equimolar mixtures, while the ‘medium’ stability constant, Qin, should display dependence upon the total metal concentration as a result of the variation of activity coefficients with the ionic strength. Inspection of the data derived from iterative procedure shows that the ‘medium’ stability constant depends upon the ionic strength, while the ratio of stability constants is constant within an experimental error and amounts to 4.54 0.49.The extrapolation of the ‘medium ’ stability constant to infinite dilution by plotting log Q:* us. I $ / ( 1 + BBIB) results in the thermodynamic stability constant, log pin = 4.62 _+ 0.20. This linear extrapolation reflects to some extent the variation of the activity coefficient arising from the limiting Debye-Huckel equation. The ion-size parameter BB = 2.76 was estimated from conductometric The value of log combined with the value of KIn/Fe yields the value of 3.96 0.15 for log BE;“. This value compares well with the value 4.07 determined previously from independent spectrophotometric results for the Fe(ClO,),-KSCN system.l The successive approximation method applied previously for the Fe(ClO,),-D. Puchalska, W. Grzybkowski and D. Wojcik 1257 Al(ClO,),-KSCN system requires knowledge of the ' medium ' stability constant of the FeSCN(DMSO),2+ complex for each of the series of the equimolar mixtures. The corresponding dependence of log QFe upon ionic strength has been determined previous1y.l On the basis of 'medium' stability constants, known concentrations of the FeSCN(DMSO),2+ complex and material balance the equilibrium concentration of ' free ' ligand concentration has been calculated.In successive approximations the concentra- tions of both complexes were included. In general, the third approximation led to the convergent values of InSCN(DMS0),2+ concentration. The average values of log Q:n for each series of equimolar mixtures are given in table 1 along with the corresponding ratios o f ' medium' stabilities of both monothiocyanate complexes.As is seen, the ratios remain constant for all series, irrespective of total concentrations of both metal perchlorates, providing convincing confirmation for the constancy of the activity coefficients in equimolar mixtures. A similar extrapolation of log Q:. to infinite dilution leads to the value of 4.83k0.07 for logp:", i.e. the thermodynamic stability constant of the InSCN(DMS0),2+ complex compares reasonably well with the value of 4.62 0.20 derived by the iterative procedure. In order to complete the thermodynamic characteristic of InSCN(DMSO),2+ complex the heats of complex formation have been measured calorimetrically. The calorimetric procedure used in the present study is identical with that applied previous1y.lv2 The calorimetric measurements were carried out for two series of In(ClO,), solutions of constant In(ClO,), concentration in each series, while concentrations of KSCN were varied.In each series the concentration of indium(rr1) perchlorate was considerably higher than that of KSCN, preventing the formation of 'higher' indium(rr1) complexes. The obtained calorimetric results are given in table 2. Following from the data the enthalpy of formation of InSCN(DMSO),2+ is practically the same for each method of InSCN(DMSO),2+ calculation and is independent of In(C10,), as well as KSCN concentrations for all investigated solutions. The average enthalpies of formation of the InSCN(DMSO),2+ complex were estimated to be 0.92k0.02 kJ mol-l from the iterative method and 0.69+0.01 kJ mol-1 from the successive approximation treatment.The enthalpy value combined with the value of the standard free energy of formation arising from the stability constant yields the value for the entropy change of InSCN(DMS0),2+ complex formation, which amounts to 92f4 and 94.6 1.4 J K-l mol-1 from iterative and successive approximation methods, respectively. Monothiocyanate Complexes of AlIII, GalI1, In"' and FelI1 in DMSO Solutions The thermodynamic functions of the complex formation for AlSCN(DMS0),2+, GaSCN(DMS0),2+ and InSCN(DMSO),2+ complexes derived by both iterative and successive approximation methods are given in table 3. This table alsocontains the thermo- dynamic characteristics of the FeSCN(DMSO),2+ complex determined independently on the basis of the data obtained for the Fe(ClO,),-KSCN system.As was mentioned above the iterative procedure yields the stability constants of the FeSCN(DMS0),2+ complex along with the corresponding value for the second component of equimolar mixtures. The resulting values of log /3Fe for all investigated systems are also given in table 3. Inspection of table 3 shows that the values of the thermodynamic functions derived by two different methods are nearly the same within experimental error. However, in our opinion the iterative method seems to give more reliable results, as the stability constants of the FeSCN(DMSO),2+ complex derived for all the equimolar mixtures are in good agreement with each other irrespective of the nature of the system investigated.Moreover, the values compare well with the experimental value obtained for the Fe(ClO,),-KSCN system, providing additional support for the reliability of the equi- molar mixtures method. It is of some interest to note that the equimolar mixtures1258 Monothiocyanate Complex Formation Table 2. Enthalpy of formation of InSCN(DMSO)X+ complex as a function of thiocyanate anion concentration in DMSO solutions at 25 "C CIrI(C10& CKSCN AHa AHb /mol dm-3 /mol dmP3 /kJ mol-l /kJ mol-l 0.0226 0.0025 0.88 0.70 0.0226 0.0063 0.94 0.72 0.0226 0.01 12 1.01 0.73 0.0544 0.0083 0.9 1 0.68 0.0544 0.0094 0.92 0.68 0.0544 0.0144 0.88 0.64 a Calculated using 'iterative' method. Calculated using ' successive approximation ' method. Table 3. Standard thermodynamic parameters for the monothiocyanate complexes of trivalent metal cations in DMSO solution at 25 "C log 81 AGo/kJ rno1-I AW/kJ mol-l A F / J K-l mol-l 1% Dl AG"/kJ mo1-I AW/kJ mo1-I A F / J K-l rno1-I successive approximation method 3.87 f 0.06 6.28 & 0.3 95.1 f 0.2 2.66 & 0.15 15.42 f 0.11 103 k 6 - 15.2+ 1.7 -22.1 k0.3 iterative method 2.79 k0.20 3.51 k0.12 - 15.9k 1 .1 - 20.0 & 0.7 14.72 & 0.4 1 1 . 1 k0.2 103 _+ 6 104&3 log B p 4.09 f 0.35 4.02 & 0.25 4.83 f 0.07 0.69 & 0.01 94.6 f 1.5 - 27.6 f 0.4 4.62 f 0.20 0.92 & 0.02 92+4 -26.4+ 1 . 1 3.96f 0.15 Fe3+ + SCN- f FeSCN2+ 1% a1 AG"/kJ mol-l AW/kJ mol-1 A F / J K-l mol-l 4.07 f 0.09 - 23.2 +_ 0.5 5.04 k 0.04 95f2 method is particularly useful in studying the differences between the stability constants, and might result in satisfactory absolute data for both components of the system, as is evident from our results and results recently published.6 Examination of the data listed in table 3 shows that the most characteristic feature of trivalent metal complexes is the positive values of both enthalpy and entropy changes.It follows that the stability of the monothiocyanate complexes is entropy-controlled, as the values of the TAS" term markedly exceed the corresponding AH" values. It is obvious that the entropy gain is related to the release of solvent molecules during complex formation. However, the AS" values are much higher than those reported by Ahrland et al.' for the monothiocyanate complexes of divalent metal cations, varying from 25 J K-l mol-1 for cadmium(r1) to 50 J K-l mot1 for mercury(1r).This striking difference may be explained in terms of the influence of the metal cation on the structure of the bulk solvent. Substitution of the SCN- anion for a solvent molecule in the first coordination sphere of a metal ion results in a substantial decrease of charge density. It is evident that the effect is much more pronounced when the M(DMSO)i+ solvo-cationD. Puchalska, W. Grzybkowski and D. Wbjcik 1259 is transformed into the MSCN(DMSO)i+ complex. Thus the entropy gain on desolvation is increased on account of the loss of order in the bulk solvent. As a result, the total entropy change is practically independent of the nature of the metal ion. Moreover, the values are very close to that determined for the inner-sphere FeSCN(DMSO)E+ complex.1 The conclusion follows that the monothiocyanate complexes of aluminium(m), gal- lium(m) and indium(m) have the same stoichiometry and structure irrespective of their stabilities. Further inspection of table 3 shows that the values of enthalpy of complex formation exhibit a distinct variation with the nature of the central metal ion and decrease rapidly with the radius of the cation. The variation is closely related to the electronic structure and degree of softness of the metal cations. The enthalpy becomes less positive as the metal-to-ligand bond becomes more covalent. The latter effect is responsible for the overall increase of the stability of the complexes within the aluminium(Ir1) to indium(rI1) series. This investigation was supported by grant MR-I- 1 1, Poland. References 1 D. Puchalska and D. Wojcik, J. Chem. SOC., Faraday Trans. I , 1985,81, 1037. 2 D. Puchalska, W. Grzybkowski and D. Wojcik, J. Chem. SOC., Faraday Trans. I , 1986, 82, 1381. 3 M. Pilarczyk and W. LibuS, Bull. Acad. Pol. Sic., Ser. Sci. Chim., 1974, 8, 717. 4 W. LibuS, B. Chachulski, W. Grzybkowski, M. Pilarczyk and D. Puchalska, J. Solution Chem., 1981, 5 D. Puchalska, W. LibuS and A. D. Pethybridge, Electrochim. Acta, 1984, 29, 1453. 6 W. Grzybkowski and M. Pilarczyk, J. Chem. SOC., Faraday Trans. I , 1986,82, 1745. 7 S. Ahrland, S. I. Ishiguro, A. Marton and I. Persson, Acta Chem. Scand., 1985. A39, 227. 10, 1 . Paper 611 112; Received 3rd June, 1986

ISSN:0300-9599    

DOI:10.1039/F19878301253

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I, 1987,83, 1261-1268 Rotating-disc Electrodes ECE and DISPl Processes Richard G. Compton,* Richard G. Harland, Patrick R. Unwin and Andrew M. Waller Physical Chemistry Laboratory, Oxford University, South Parks Road, Oxford OX1 3QZ Theory is presented for the shape of current-voltage curves measured at a rotating-disc electrode for the cases where the substrate reacts via either an ECE or a DISPl process in which the kinetic step is sufficiently fast to allow the use of the reaction-layer concept in the theoretical treatment. When the electrode potential for the electrode reaction can be found under conditions where there are no complications of coupled kinetics, analysis of the shift in half-wave potential with disc rotation speed allows the deduction of the rate constant for the kinetics of the homogeneous chemical process involved in each mechanism.If an independent measurement of this rate constant is available (e.g. from spectroelectrochemistry) then discrimination between ECE and DISPl pathways is possible. Experiments involving the reduction of fluorescein in aqueous carbonate/bicarbonate buffers were found to be in good agreement with theory. Furthermore, comparison of the kinetic data with transient electrochemical e.s.r. data revealed the reaction mechanism to be DISPl, in agreement with previous observations. Many two-electron reductions and oxidations can be understood in terms of the following general kinetic scheme' in which the product, C, of the second step is more readily reduced (oxidised) than the starting material A: A + e - r B GI3 (1) k B + C (11) C + e - e D G / D B + C e A + D .It is clear that there are three limiting cases arising out of steps (1)-(IV). These have been identified as ECE, DISPl and DISP2 by Saveant and coworkers.'s2 In the ECE mechanism step (IV) is kinetically insignificant and both electron transfers occur heterogeneously at the electrode surface. In the DISP mechanism, step (111) is considered as negligible, and the second electron transfer takes place homogeneously via dispro- portionation [step (IV)]. Within the DISP mechanism two cases occur, depending on whether the rate-determining step is the first-order reaction in step (11) (a DISPl process) or the second-order reaction in step (IV) (a DISP2 process).Whilst DISP2 is readily identified by most conventional electrochemical techniques,2 the discrimination between DISPl and ECE using cyclic voltammetry, linear sweep voltammetry or potential step chronoamperometry requires accuracies that are almost unattainable in practice. We have recently shown, through the measurement of transport-limited currents at channel electrodes (either in isolation3 or in conjunction with in situ electrochemical e . ~ . r . ~ ) that ECE and DISPl pathways can be conveniently distinguished. Other workers have 42 1261 FAR 11262 Rota t ing-disc EIec t rodes successfully used micro electrode^^ and double-potential step chronoamperometry2 for this distinction. However, for many electrochemists the rotating-disc electrode (RDE)6 would represent both the most convenient and the most accessible tool for the investigation of the ECE-DISP problem.The necessary theory describing the variation of the transport-limited current with rotation speed has been derived,'* * but shows that while the behaviour under the DISP2 mechanism is such that a distinction from the other two processes may be made, the behaviour of ECE and DISPl processes is so similar that the two mechanisms would be indistinguishable with the accuracies normally employed in rotating-disc voltammetry. The purpose of this paper is to explore the possibility of tackling the ECE-DISP1 problem through the variation of the half-wave potential (rather than the limiting current) with rotation speed. Accordingly the current-voltage behaviour at an RDE for the ECE and DISPl mechanisms is calculated under the reaction-layer approximation.It is shown that when the potential e,B can be found under conditions where the effects of the coupled homogenous kinetics may be considered negligible (e.g. at very fast rotation speeds) a value may be deduced for the homogeneous rate constant k, in step (11). If an independent measurement of this value is available for comparison then discrimination between ECE and DISPl is possible. This is shown to be the case for experiments involving the reduction of the molecule fluorescein to leuco-fluorescein in aqueous carbonate/bicarbonate buffers in the pH range 9.5-10. Experimental results are shown to be in good agreement with theory and it is established that the reduction proceeds via a DISPl pathway, in agreement with previous observation^.^* Theory In this section we consider the variation of the half-wave potential with the rotation speed for the two mechanisms (ECE and DISP1) of interest.Case (i): The ECE Mechanism A general ECE process may be written as electrode : +e- -e A"B - solution : B:C electrode : C 2- D. Now if B is very unstable it will not occupy the whole of the diffusion layer (thickness adif,), but will be confined to a very thin 'reaction layer' (thickness 6,) immediately adjacent to the disc surface. Hence, providing k % 1, transport within this layer may be regarded as being by diffusion (normal to the disc surface in direction z ) alone, and so, where, for steady-state conditions, The surface concentrations of the species participating in the above scheme may be related via the following equations:R.G . Compton et al. 1263 (4) [A],+ 03 = “z=o + [BIZ-0 + P I P 0 where F RT 0 = - ( E - P ) . [C],,, is absent from eqn (4) since it is assumed that E&, is smaller in magnitude than Now the observed current i at the rotating disc (area z r 2 ) is given by where [Alz-m - [Alz-o 6diff also where W is the disc rotation speed (Hz) and v is the kinematic viscosity (cm2 s-l) of the solution. The boundary conditions at the electrode surface stipulate that and so from eqn (2), (9, (8) and (1 1) we see that Likewise from eqn (4), (5) and (12) we find Then eqn (7), (12) and (13) allow us to write We see that, as 0-+ -a, (1 5 ) llim -- - [ A ] ~ - + ~ 2(1 -Bk/Sdiff)/6diff nr2 FD which shows that in the limit of infinitely fast kinetics (6, x 0) a two-electron transfer occurs as would be expected. We thus find that and thus that the equation describing the current-voltage curve for an ECE process is log --1 = (ilr ) 2.3RT 42-21264 Ro tat ing -disc EIec trodes from which we may deduce first that a conventional diffusion-corrected Tafel plot should have a slope of 59 mV per decade and secondly that the half-wave potential E+ varies as F Case (ii): The DISP1 Mechanism The general DISPl mechanism may be written as electrode : +e- -e A e B - solution : B 5 C solution : B+C=A+D.We assume that the second homogeneous step lies heavily in favour of A and D and so each time the first solution reaction occurs two molecules of B are converted into A and D.Thus we need to consider the following steady-state equations (within the reaction layer approximation) : and so the distribution of [B] is given by eqn (2), but now, dk = (D/2k):. (19) Eqn (2) and (19) may be integrated and the following boundary conditions applied: x=O to attain readily wl = [A],,, (dk[exp(0)+i]+- 2 az 2-0 It follows from the above that and so, combining eqn (23) and (24) we see that the DISPl current-voltage curve is described byR . G. Compion et al. 1265 We see that a diffusion-corrected Tafel plot has a slope of 59 mV per decade, as with an ECE process, but that the variation of the half-wave potential with rotation speed is 2.3RT dW 'Og[(!? 0.643V"k]+lOp 2' E+ = P-- F Experimental Rotating-disc measurements were carried out using an Oxford Electrodes rotating-disc assembly and motor controller.Mercury-plated copper electrodes were prepared as described previouslyg and the mercury surface was renewed before the measurement of each current-voltage curve. The diameter of the rotating disc was measured as 3.565 mm. Aqueous solutions were made up using doubly distilled deionised water (resistivity > lo7 cm). All solutions were carefully purged of oxygen before experi- ments were carried out. Reagent-grade chemicals were used throughout. Aqueous solutions in the pH range 9.5-10.0 were obtained using sodium carbonate/bicarbonate buffers to which KCl was added to serve as background electrolyte and to maintain a constant ionic strength of 1 mol dm-3. pH measurements were made using a Radiometer PHM 64 meter.All measurements were made under thermostatted conditions at 25 0.5 "C. Potentials are quoted with respect to the saturated calomel electrode. Results and Discussion We consider first an experiment in which the molecule fluorescein (F) was reduced via the intermediate semi-fluorescein (S) to leuco-fluorescein (L) at a mercury-coated copper rotating-disc electrode in aqueous solutions at pH 9.5-9.7: electrode : F S' +e- -e - solution : S'+ H+ f SH*+ solution : S' + SH'+ e F + L R R1266 Rotating -disc Electrodes dW Fig. 1. The experimentally determined limiting current us. rotation speed data ( x ) compared with the theoretically predicted behaviour for ECE ( * - . ) and DISPl (---) processes calculated from ref. (8) using a value of k deduced from eqn (27) for pH 9.68.(a) n = 1, (b) n = 2. The process has been written as a DISPl reaction in the light of our previous work 3 9 4 at channel electrodes and for reasons which will be elucidated below. Fig. 1 shows the dependence of limiting current on the square root of the electrode rotation speed: a smooth transition from two- to one-electron behaviour is observed as the rotation speed is increased. The solid lines represent pure one- and two-electron behaviour calculated using the diffusion coefficient (3.1 x 10+ cm2 s-l) measured at high pH (0.1 mol dm-3 NaOH), where the reaction suffers no kinetic complications, and a simple one- electron reversible reduction of F to s' is observed. The transition shown in fig. 1 is typical of DISP (and ECE) processes.The data were obtained at pH 9.68. We have previously4 used in situ electrochemical e.s.r. transient data to show that the decomposition of S', under the conditions of these experiments, is a pseudo-first-order process with a rate constant given by log k = 9.922-pH. (27) This result may be used7* * to predict the expected limiting current-rotation-speed behaviour for DISPl and ECE mechanisms, respectively, at the pH in question. The results are displayed in fig. 1 and comparison with the experimental results shows that discrimination between the two mechanisms is not feasible given the experimental accuracies involved. This conclusion, namely that measurements of transport-limited currents at the rotating disc cannot discriminate between ECE and DISPl even if an independent measure of the homogeneous rate constant for the chemical step is available, is in agreement with the observation of other workers [e.g.ref. (2)]. We therefore turn to the consideration of the variation of half-wave potential with rotation speed. Fig. 2 shows data obtained from the fluorescein system at pH 9.68 and 9.51. The half-wave potential is shown to vary with rotation speed. In particular, the lines drawn have a slopeR. G . Compton et al. 1267 -1~llOC I I I I I I I I 1 I I -0.4 - 0.2 0 0.2 0.4 log,, W I H z ) Fig. 2. The variation of the half-wave potential with rotation speed for current-voltage curves run at pH 9.68 (0) and 9.51 ( x ) . Table 1. Values for the rate constant k (s-l) rotating-disc measurements electrochemical ECE DISPl PH e.s.r.transients mechanism mechanism 9.68 1.75 0.5 1 2.03 9.51 2.58 0.75 3.01 of 2.3RT/2F (= 29.5 mV at 298 K) as predicted from eqn (18) and (26) for both ECE and DISP 1 mechanisms. Reasonable agreement between theory and experiment is observed in so far as the slopes of the lines are concerned. Furthermore, on the basis of eqn (27) and eqn (1 8) and (26) it would be expected that the two (parallel) lines should be displaced by 5 mV from one another; excellent agreement is seen between theory and experiment. At this point, since satisfactory straight-line agreement is seen for both ECE and DISP 1 mechanisms, discrimination between these reaction schemes is not possible. However, it is clear that if a value of If can be measured the rate constant (k) for the protonation of S' may be evaluated and compared with the values obtained from electrochemical e.s.r.transients which are generalised in eqn (27). Values of I.? were obtained by going to high rotation speed (ca. 16 Hz) so that effectively only one electron was transferred. Under these conditions it was found that If = - 1.150~0.001 V for both of the solution pH employed. Using this value it is possible to deduce values of k. These are listed in table 1, where they are compared with the numbers obtained1268 Rotating-disc Electrodes independently. It is clear that agreement is only found if the rotating-disc data are analysed assuming a DISPl mechanism, and thus that this is the likely mechanism for the reduction of fluorescein. This is in agreement with our previous deductions made using channel electrodes.3> In conclusion, we have vindicated the theory presented above for the shape of RDE current-voltage curves for ECE and DISPl processes and have shown that provided the first electron-transfer step in the ECE reaction is reversible, and that the half-wave potential may be measured under conditions where the rotation speed is sufficiently fast that the coupled kinetics can be ‘outrun’, a value may be deduced for the rate constant of the homogeneous chemical process in each mechanism.If an independent measure- ment of this rate constant is available, discrimination between ECE and DISPl pathways is possible. Although in this work we have used spectroelectrochemical data for this independent value, it should be emphasised that it could equally well have been determined from conventional electrochemical methods (e.g. at planar electrodes), so the method is independent of the availability of suitable spectroelectrochemical information. Thus it will be seen that the ECE-DISPl distinction can be made by entirely electrochemical measurements (with no unconventional equipment), provided the assumptions of the method described here are justifiable. We thank the S.E.R.C. for a studentship for A.M.W. and St John’s College, Oxford for support for R.G.H. References 1 C. Amatore and J. M. Saveant, J. Electroanal. Chem., 1977, 85, 27. 2 C. Amatore, M. Gareil and J. M. Savkant, J . Electroanal. Chem., 1983, 147, 1 . 3 R. G. Compton and P. R. Unwin, J. Electroanal. Chem., 1986, 206, 1 . 4 R. G. Compton, P. J. Daly, P. R. Unwin and A. M. Waller, J . Electroanal. Chem., 1985, 191, 15. 5 M. Fleischmann, F. Lasserre and J. Robinson, J. Electroanal. Chem., 1984, 177, 115. 6 Yu. V. Pleskov and V. Yu. Filinovskii, The Rotating Disc Electrode (Plenum Press, New York, 1976). 7 L. S. Marcoux, R. N. Adams and S. W. Feldberg, J. Phys. Chem., 1969, 73, 261 1 . 8 S. Karp, J . Phys. Chem., 1968, 72, 1082. 9 R. G. Compton, D. J. Page and P. J. Daly, Anal. Chem., 1983, 55, 1191. Paper 6/ 1 1 13 ; Received 3rd June, 1986

ISSN:0300-9599    

DOI:10.1039/F19878301261

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I , 1987,83, 1269-1280 Origin of Idealized Static Thermodynamic Forces inducing Solid Particle Motion, Orientation and Related Effects in a Solute Concentration Gradient Ernest A. Boucher School of Chemistry and Molecular Sciences, University of Sussex, Brighton BNI 9QJ The forces acting on a solid particle in the concentration gradient of a solution or mixture are given for a carefully specified thermodynamic model. From this analysis it is possible to predict the initial nature of any ensuing motion of the particle and, in particular, to give the potential propelling force on a particle at rest as 9 = -Z S IdA,(do,/dt) where the integration is over each slice of solid of surface area dA,, with s denoting different types of surface, and dos/dr is the variation in the interfacial tension for each s at position <. An orientation effect is predicted for certain solids with composite surfaces. The propelling and orientation forces can be related to the surface excess adsorption of a solute at a solid surface.Aspects discussed are the implications for using density-gradient columns, the behaviour of particles at liquid interfaces and a variety of qualitative phenomena possible in special circumstances (some of biological relevance) including two-particle effects, the potential for the complementary movement of liquid over or through a stationary solid and conditions whereby a collection of particles might become separated. Some of the effects might apply to macromolecules to which the concepts of surface area and interfacial tension are not relevant.All predictions are based on static considerations using equilibrium thermodynamics : no attempt is made to describe actual motion. The objective of this paper is to explain in thermodynamic terms the reasons for the possible linear and orientational motion of a solid particle in a composition gradient (i.e. a chemical potential gradient) within a solution or mixture. Thus the forces acting on a particle at rest are given for an ideal model. Were the particle to be released it would move, and conversely the liquid would be expected to move when the solid is held. The idealized forces are those expected immediately after adsorption has occurred at the solid surface. The basic premise is that equilibrium thermodynamics when applied to a carefully specified, but artificial model is capable in principle of giving the magnitude and direction of the forces on the particle, and hence an indication of impending motion, although no attempt is made to describe the motion.This paper does not aim to give an alternative account of diffusiophoresis involving irreversible thermodynamics and the laws of motion: it is more limited in intentions, but it complements accounts discussed below. Young et af.l attributed the movement of bubbles in a liquid possessing a temperature gradient to stresses resulting from the thermal variation of the surface tension at the bubble surface. It was found that a bubble could be held stationary within a liquid by a negative temperature gradient and an expression was obtained to describe this phenomenon.Levich and Kuznetzov2 examined the motion of a droplet of one fluid in 12691270 Particles in a Concentration Gradient another whose concentration varied with direction, and Dukhin and Derjaguin3 described the phenomenon known now in several variants as diffusiophoresis. Mansfield* also examined the motion of droplets of one liquid in another, with special reference to the use of density-gradient column. A density-gradient column consists of a cylindrical vessel containing a solution of varying composition such that its density decreases with height (usually linearly, and determined by calibration). A particle or droplet placed in the solution should, by simple considerations of neutral buoyancy, come to rest at a position such that its density is the same as that of the bulk solution, level with the centre of gravity of the particle.If by surface effects the particle would move in a concentration gradient of the kind described above, then it can be argued that an error will be incurred in using the density-gradient column. The particle equilibrium will be affected by these surface effects, and the experimentally determined concentration gradient might be inaccurate since it is based on the use of calibration floats (glass ‘divers’), although their surface area might be relatively much less than that of the particles. Kawai and Keller5 pointed out the possibility of errors, especially for solids with large surface areas, in the determination of the density of polyethylene crystals by this method: they used an argument by C .R. Burch that a gradient in energy (surface implied) produces a force. Ziabicki and Wasiak6 also discussed this source of error giving the difference, pt - p l , between the true and liquid densities as - A do’ Pt--P’ = - g 6) (g) where A is the specific surface area, pl is the liquid/air surface tension, d is the distance up the column and g is the acceleration due to gravity. Blackadder and Lewel17 took the precaution of using polyethylene crystals of relative low area in carbon tetrachloride- p-xylene. Pearson* argued that use of the energy gradient explanation is only valid for a forcefequal to VF (where F is the energy) if, for the energy in question, the mechanical component can be separated from the other contributions to the total energy: the surface energy cannot be so separated from the thermal energy.Pearson went on to discuss density variations in the fluid surrounding the particle, but regarded these as small. The full implications of these possible effects have not been fully assessed; some aspects will emerge from the analysis below. It can be argued that although the solid is at rest in the density gradient column, the components of the liquid or of the absorbate will themselves move. The density-gradient column itself will not be in physicochemical equilibrium, but the timescale for diffusion is long compared to that of the experiment. This is a very reasonable assumption. The arguments which follow depend on the further assumption that the presence of the solid does not speed-up motion near the surface to such an extent that one cannot even examine the forces caused by the supposed static concentration gradient and accompanying equilibrium (matching the bulk concentra- tion) adsorption on the solid.One can suppose that certain solid/liquid combinations will be more appropriate than others for this approach and that droplets might be the least suitable, but there could still be qualitative value in this analysis for liquids. Prieve et aZ.g discussed particle migration in an ionic concentration gradient near a chemically reacting metal surface (as the basis of coating metal by latex). Ruckensteinlo discussed the phenomenon of motion in an electrolyte solution in terms of an interfacial tension gradient.Anderson et al.l1? l 2 gave a detailed treatment of particle motion. They state that : ‘ Although the usual Marangoni motion is precluded by the non-slip condition at a fluid/solid interface, it has been argued theoretically that gradients of non-electro- lytes can propel rigid particles through fluids or generate osmotic flow through capillaries. The origin of such flows is the diffuse nature of the force field between solute molecules and the solid surface, which enters directly into the fluid momentum equations,E. A . Boucher 1271 whereas this interaction is lumped into a boundary condition in the usual analysis of the Marangoni effect. ’ There have been related attempts to solve similar problems for particle motion in a concentration gradient, but because of the complexities of systems involving particle motion, an analysis has been carried out in thermodynamic terms for a strict set of conditions applying to a static situation.There is available little knowledge on rates of diffusion relative to rates of adsorption at interfaces with an adsorbate concentration gradient and it is difficult to judge the limitations on applicability of even the qualitative aspects. It is the essence of an analysis of this kind that one uses a postulated model, regardless of its precise physical relevance and unencumbered by speculation. The relevance must be judged later after differing approaches have been compared. Then it might be possible to see the extent to which phenomena similar to those predicted arise, not only with solids, but with liquid droplets, with colloidal particles (where Brownian motion will compete) and with flexible and rigid polymers of synthetic or biological origin.The Model for Non-electrolytes The model system is shown in fig. 1. The bulk concentration of component i in the elemental slice between z and z + 6z is ci, : the concentration is uniform in the other two directions except near the solid surface. It is supposed that an externally applied restraining force Fr, provisionally regarded as acting in the direction opposite to that of increasing z, is necessary to keep at rest a rigid particle of solid of arbitrarily, but conveniently chosen cylindrial shape. The position of the solid or of an element of the solid is specified by <, and the whole system consists of contiguous slices 6z.All extensive variables are of the type such that, for example, the solid area and volume are A =16A and V s =j6Vs, and the total number of moles of component i is ni = Sdn,. The solid/liquid interfacial tension for the portion within 6z is denoted oz. The complete differential of the internal energy of a slice of the system containing the solid is C (2) In addition to the usual thermodynamic quantities, areas A’ and A” of the cylindrical ends are introduced, and the end surfaces are assigned interfacial tensions o’ and a’’, which are expected to depend on the solution composition. It is assumed that the chemically inert solid is of constant volume. Integration of each term in eqn (2) over the whole system gives (3) Eqn (3) is not applicable to highly curved surfaces, but this is not a significant limitation. We take for comparison a system at the same temperature, pressure, volume and composition except that the solid is absent or is replaced by a fictional solid of the same volume, but having no active surface: d(6U) = Td(6S)-P1d(6V1)-PSd(6VS)+azd(SA)+ 2 pid(6n,).i-1 d U = TdS - P1d V + oz d( 6A) + o’d A ’ + a”d A I’ + 2 pi d( 6ni ) . s i s dU* = TdS*-PldV+Xp,dn,*. (4) a The Gibbs function defined as U+PV- TS is introduced to give the fundamental equation for the characteristic function A*G = (G - G*), under isobaric conditions for the whole system, dA*G = -A*SdT+ oz d(6~)+~’”’+o’’d~’’+2 pi d(6A*ni). ( 5 ) i s1272 Particles in a Concentration Gradient Ci. o ci concentration -c I , p-6A solid/liquid CJ I I U;' U ;' Fig.1. System of a solid particle in a solution or mixture of varying composition in the z direction. The position of the solid is specified by [, whereas the concentration c ~ , ~ , the chemical potential and the solid/liquid interfacial tension depend on the location of the elemental slice between z and z+&. The solid is held at rest by force Sr acting in the opposite direction to the driving force 9 created. A more compact form of eqn ( 5 ) for isothermal as well as isobaric conditions is dA*G = sos, d(dA,) +Z pi d(dA*n,). i s It is seen from eqn ( 5 ) that a surface type, denoted by s in eqn (6), whose interfacial tension os,2 also depends on location, gives the same contribution to dA*G whether it occurs at either solid end or at a solid side, and therefore at any orientation.Not so obviously, eqn (6) contains the form of one definition of the amount adsorbed. For a two-component solution with one type of surface, eqn (6) can be written as Within an elemental slice 62, 6A*ni gives the difference between the number of moles of component i, with the adsorbing solid present less the number when the inactive fictional solid of the same size and shape is present: (8) The quantity 6ng is by definition the surface excess which can be positive or negative. The position of the solid surface may seem implicit. The essence of Gibbs's approach is to recognize that if one uses a dividing surface between the bulk phases, chosen to be parallel to the physical surfaces, then dny, or the areal quantity dry = 6nf/6A, will depend precisely on the location of this Gibbs surface.We need not concern ourselves with all the defining technicalities, but one choice of surface position in the Gibbs reference system can be that where there is no excess of, say, component 1, i.e. dry = 0 (for each slice). The relative adsorption, Xi1), for a slice dz is based on that for a uniform macroscopic system, viz. (9) which is then invariant to the location of the Gibbs surface. Taking a two-component liquid as an example, it is usual to express the experimental bulk solute mole fraction dA*n, = dn, -an,* = dng. ry = rg - (ci/c,) rlE. A . Boucher 1273 change Ax, on adsorption as an adsorption isotherm of the specific surface excess, nAx,/m, of component 2 per unit mass of solid for a liquid of n moles, as a function of x,.It then follows that rp = nAx,/x, A . (10) Restraining and Propelling Forces For the system with the solid held in a fixed position, eqn (6) can be integrated at constant chemical potential and constant interfacial tension within each slice Sz giving r s J i J Eqn (1 I ) can be differentiated and eqn (6) subtracted giving the appropriate form of the Gibbs-type adsorption equation : f r On the timescale of an experimental observation, diffusion of the components of the solution is very slow and remains constant. Then in eqn (6), d(SA*ni) = 0, and the integrated form reduces to f s J The restraining force exerted on the solid to keep it at rest is c'A*G/c'C, whence Eqn (14) implies that a restraining force will be required to prevent motion if the particle area, A , varies with position, (, at constant particle/solution interfacial tension, os.z . We shall subsequently concentrate attention on the first contribution, so that for a rigid particle the restraining force is Fr = C: (c?os, J?() SA. (15) ,s The propelling force, ideally given by 9 = -Fr, is that which would in the absence of the externally applied force (see model represented by fig. 1) propel the particle when it is exactly counteracted by frictional or viscous forces, provided equilibrium can be maintained to comply with the model. For a single type of surface this ideal (hypothetical) propelling force acting on the solid at constant temperature, pressure, chemical distribution and composition, is F = - (20,/2<)6A.(16) h 9 is positive if oz decreases as a slice of solid is moved to the right in fig. 1. The force contribution from each solid band or slice is directly proportional to its area SA, justifying to some extent the recommendation that the density-gradient column should not be used for the accurate determination of the density of a solid with a large surface area. The factor daz/d< gives the variation of the interfacial tension of a band of surface area 6 A with change, d<, in its position in the fluid. It is emphasised that do,/d< is strictly not the gradient, do,/dz: the variation of az with the position of the band SA on the solid surface for a stationary solid. Indeed, the solid could consist of contiguous elemental solid cylinders, each of area 6AS of different surface composition. In this case da/dz at constant 5 has little significance.1274 Particles in a Concentration Gradient Eqn (1 5 ) contains no reasons for o depending on position, i.e.on solution composition, nor is it essential to the basic argument that equilibrium conditions prevail. The difficulty of measuring solid/liquid interfacial tensions is well known. It is implied that o is related to adsorption phenomena and that any interpretation using experimentally determined adsorption isotherms entails equilibrium and, in that sense, idealized behaviour. Features such as the diffusion, adsorption and desorption rates, and fluid flow, will not be discussed here. For a single surface type eqn (1 2) can be written for unit area of an elemental slice as When eqn (17) is used in eqn (16) the propelling force for a rigid particle of uniform surface type can be written as c f We have another thermodynamic expression, namely, the Gibbs-Duhem equation for bulk solution, C and it is consistent with the definition of ril) given by eqn (9) to eliminate p1,* from eqn (1 7).The result is that under isobaric and isothermal conditions C Then ril) can be used instead of rf in an expression analogous to eqn (18), enabling the role of the various components to be assessed. The dependence of c ~ , ~ on z is established in setting up the experiment, but the dependence of I?:') on ( is determined by the adsorption isotherm. The treatment given above presupposes that it is the solid which will move in a stationary liquid when the applied restraining force is removed. It is easily seen that one could have carried out an analysis in much the same way for a liquid of non-uniform bulk concentration over a plane solid in a capillary tube or in a porous solid or compact.One could more readily have regarded the tendency of the liquid to move relative to the solid as a case of preferential wetting of the solid either by the solution of low solute concentration, or by the solution of high solute concentration. A similar phenomenon can then be expected from, say, solvent and solution separated within a capillary tube by a diffuse boundary. With all these cases complications of fluid flow, e.g. due to hydrostatic pressures developed, can be expected. The thermodynamic treatment does not give the plane of shear: the limiting plane as the solid is approached from the liquid side, just able to withstand the moderate shear stress arising from the relative movement of solid and liquid.With particles in the 1 ,um size range the thickness of the stagnant layer will be relatively insignificant, but this will not be so with small or rough particles. The analysis given above should be applicable to liquid droplets, except that the concept of differing surface types loses its significance. Liquid droplets can therefore be expected to move according to forces of the type given by eqn (15) and (18), but the orientation effect of the next section will not occur, and Marangoni effects (see below) will tend to cause mixing at the liquid/liquid interface.Then the analysis is expected to be of limited qualitative relevance.E. A . Boucher 1275 I I1 Fig. 2. Depiction of the torque F required to rotate a cylindrial particle having curved surface types s’ and s” and inert ends through angle o. The elemental slices are greatly exaggerated to show the change in surface environment for rotation by do. Particle Orientation It can be demonstrated that in principle the orientation of a particle in a concentration gradient is possible when the overall surface consists of discrete patches of at least two types, s’ and s”, differing in their value of da’/dy and da”/dy. The supposition is that a preferred orientation can exist if one of the types gives a greater lowering of surface free energy than the other, either at the high-concentration or at the low-concentration side of the particle. Fig.2 represents for a solid cylinder seen end-on with two curved surface types of area 6A’ and JA”, and indifferent or inactive end surfaces. Assuming that state I is the most favourable orientation it is only necessary to show that a torque Y is needed to produce an incremental rotation 60 of the cylinder about its axis. From the geometry of the simple system it can be deduced that the work done on the system by applying the torque is (21) In the model 6A” = 6A’ and, therefore, a torque is needed to produce a thermodynami- cally reversible rotation unless da’/dy = da”/d<. Clearly the finite work done in a change is given by integration, and the torque itself can be obtained in terms comparable to those for the ideal propelling force, e.g.arising from eqn (15), or given by eqn (18). We know of no study where orientation has been proposed. n w = 6Ayaa///ay) sy - 6Ayaai/ar) sy. Neutral Buoyancy Conditions Comparison of eqn (1) with eqn (15) shows a basic similarity of form. In view of the difficulty in measuring ts for immersed solids it is perhaps not surprising that o1 for the liquid surface has been substituted for the polyethylene/liquid value @l is related to c in the density-gradient column: see below), and that da/dz is assumed to be constant. We shall not attempt here to give a detailed and complete analysis of the problem of finding where a suspended object in a density-gradient column will come to rest. Several types of fundamental phenomenon might be involved and experimental measurements and extrathermodynamic arguments might be needed.In many, but by no means all, cases of adsorption from solution a monomolecular model for surface adsorption is1276 Particles in a Concentration Gradient sufficient, and since this is of negligible thickness compared to the size of visible particles, the effect on density measurements is simple to illustrate. Imagine that fig. 1 has been rotated anticlockwise through 90" and that in a gravitational field the solute concentration and hence the density of the solution now decreases as z increases vertically. Suppose there exists a linear dependence of density p on height, and that the variation of a hydrostatic pressure P follows accordingly: p =po-az dP/dz = -g(po-az) (22) where a is a constant and g is the acceleration due to gravity.The cylindrical solid (with axis vertical and freely suspended) experiences a body force mg, and is subjected to a buoyancy force to which in principle there is added a force of the kind given by eqn (1 5) or (18). From statics, mg = Pgpm -m dA(c?a/c?<) (23) 1 where Vs is the solid volume and pm is the density of the bulk fluid level with the centre of the solid. The difference between the true density, ps, of the solid and that measured, It is suggested that this is for most purposes an appropriate equation for the correction of the measured density, or for correcting a calibration curve, with the possible expression of the force in terms of the relative adsorption.Eqn (24) might not represent the only, or even the most important source of systematic error in the technique. With irregularly shaped particles the accuracy of solid density measurement might be limited by the uncertainty in the position on the solid to use to get the corresponding liquid density. Polymer Molecules and other Particles A small but visible solid polymeric particle of the kind that might be used in a density-gradient column can be assigned a surface area and an interfacial tension. Regardless of how difficult it is to measure the interfacial quantities the above treatment should be applicable to its motion, orientation and neutral buoyancy conditions. The same cannot be said of the following types of species which have some particulate characteristics : synthetic flexible, rigid and highly cross-linked polymers, biopolymers (e.g.globular proteins), cells, bacteria and viruses. Latex particles, depending on their type, may behave as droplets, solids or macromolecules. The concept of surface can be replaced by that of sites for specific interactions (ignoring charged species in the first instance) with components of the solution or liquid mixture, i.e. energy changes may accompany competitive binding and there may be configurational changes. One can then redefine terms in a thermodynamic or statistical mechanical treatment,l3> l4 and anticipate analogous expressions and conclusions in most cases. One can expect mobility and orientation effects as well as possible differential motion or separation.Chemotaxis of cells, for example, is well documented, but fundamental phenomena have yet to be precisely explained in physicochemical terms. In the absence of the complete analysis the full range of phenomena remains obscure, as do the lower limits on size and flexibility of species which, for example, move other than by diffusion in a concentration gradient. Liquid/Vapour Interfacial Effects We can see that the formal treatment given above can be qualitatively or even quantitatively applied to the behaviour of solids, floating drops or lenses at the freeE. A. Boucher 1277 surface of a liquid having a variable composition in at least one direction. There will exist a variation in the surface tension, a;, say. Three consequences of this would need to be investigated separately to ascertain their relative importance, e.g.their magnitude and timescale. Solutal Marangoni effects are surface flow phenomena tending to cause mixing and elimination of surface tension gradients. A solid particle and a drop or lens of a second (immiscible) liquid would be subjected to unequal surface tensions in different directions and as a result would generally move (rather like a camphor boat, but with a different origin of the unequal spreading pressure). In addition, it is predicted that the submerged portion of a surface object will be subjected to a driving force of the kind given by eqn (15). This seems to be a novel suggestion, and yet it could possibly play a role in the motion and predator-prey strategies of small surface creatures of the kind which are known to secrete chemicals.Electrolyte Solutions A full treatment of a particle in an aqueous salt solution has not yet been completed, but the following outline seems tenable. The solid which is held as in fig. 1 will now be an electrode, and therefore, subject to careful description [e.g., by specifying electrode type (below) and having a composite electrode with infinitesimal insulating spacers] its electrical potential, Ed, can be measured relative to a suitable electrode. It is not implied that mechanical and electrical connections to the solid particle are generally practicable, but they are necessary for a formal characterization of the system. The electrochemical potential of component i in the bulk solution phase within 6z is denoted pi.The (electrochemical) Helmholtz free energy, 6F, of the slice within 6z depends on the variables according to 6F = SF( T, 6 V1, 6 Vs, 6AS1. 6ni, . . . an',,,, any , . . dn;,, ). (25) The Gibbs function for the slice is d(6G) = - 6 s dT- 6 V' dP' + 6 VS d P + 0, d(6A) +z z pg d(6@) (26) a i where a now denotes the phases present, each containing some of the c+ 1 components (including electrons), denoted i. For the whole system the complete differential of the Gibbs free energy is now dG = - 6 s dT+ 6V dP1+ 6VS d P + a,d(aA)+z I: ~7 d(6ng). (27) I S I S s i I The reference system (*), has the same temperature, pressure or pressure variation, and the same variation in chemical composition as the complete system above except that the ficticious solid now has a surface which does not experience adsorption phenomena or become charged. For this new system the reference Gibbs function is dG* =- 6S*dT+ SV'dPII*+ 6VSdPSI*+C z &! d(dn?)l*.(28) s s a i s The complete differential of the new Gibbs function, A*G = G-G*, is a(A*V') dP1+ 6(A*Vs) dPS + so, d(6A) +C C s[ ,uf d(dA*nq) + zt l+,b d(dng)] (29) I a i1278 Particles in a Concentration Gradient where Fis the Faraday constant, 4 is the potential, and zi is the charge on species i. Clearly dA*F would be chosen if V and P were required as independent variables in terms of the form, JP1d(6A*V'). Usually isothermal conditions will obtain, but it is possible to use an analogous treatment to account for possible particle motion in a temperature gradient, T = T(z), rather than in a concentration gradient, ci, , = ci(z).Recall that Young et a1.l realized bubble motion in a liquid with a temperature gradient, although bubble surfaces are usually more mobile than those of a solid. It is not necessary to presume that conditions are isobaric, although we conveniently invoke this condition to obtain dA*G = o, d(6A) +C, C [elf d(SA*nf) + zf Fij d(dnf)]. (30) s a i 1 A more serious consideration at this stage is the relevance of the 'excess' quantities, of the type A*Y, to systems where electrochemical effects are present. In the non- electrolyte case the strict Gibbs definition of adsorption with respect to a dividing surface is appropriate and it leads to unequivocal thermodynamic expressions for particle propelling and orientation forces.In electrochemical systems the distribution of charges within phases, i.e. within the solution and within the solid, is important and has often led15-17 to the use of a model with finite adsorption thickness as advocated by Guggenheim,1s-20 and recently critically discussed on general grounds uis-h-uis the Gibbs model by Rusanov.21 By definition, rf is given as before, but additionally there are the electroneutrality conditions within a bulk phase a, and within the adsorbed substances C, zini = 0 and Z zirf = 0. (31) i i The usual methods for obtaining Gibbs-Duhem relationships when applied to eqn (29) give 0 = B(A*S) dT-x 6(A* V a ) d P + SA do, I a I I Under isothermal and isobaric conditions eqn (32) gives the equivalent adsorption equation 0 = J6A do, +Z C [dn; dpf, ,+C Z j z f FSnf d&.a i a i For each bulk phase a there will also be a Gibbs-Duhem equation, isothermal and isobaric conditions can be written for each slice 6z concentrations, ci, ,, and chemical potentials, cf. eqn (1 9), of the Gibbs (33) which under in terms of (34) During the timescale of an experiment it is assumed that d(6A*ni) = 0 and d ( 6 n i ) = 0 within a slice, i.e. that diffusion in the solution is relatively slow. The implication is that the last two terms in eqn (29) are zero, whence the restraining force is given by aA*G/a<, and it is exactly the same relationship as was obtained for the non-electrolyte case. The idealized force, 9, acting on the particle, which in the absence of the artificial constraint would give motion, is given by I-E. A .Boucher 1279 Each term on the right-hand side of eqn (35) gives a positive contribution if both 0, and 6 A decrease as l increases, but the detailed reasons for CJ and A depending on l will differ from those for the non-electrolyte case. It might be possible to proceed further by using a completely or perfectly polarized electrode as the particle, characterized by there being no charge crossing the interface. The components of the system are electrons and metal ions on the metal side, and the constituent ions in water on the solution side. The model would then be that of a two-part interface, or interphase, consisting of separate contributions to the surface properties from the metal and solution sides of a dividing surface. We proceed no further along these tentative lines at this stage since eqn ( 3 9 , subject to the qualifications regarding A*Y and the reference system, seems well-founded and does not contain any new electrochemical features.Discussion and Conclusions It has been shown, using a well defined thermodynamic model, that an applied force is required, given by eqn (14) or (15), or by -9 in eqn (39, to keep a solid particle at rest in a gradient of concentration (i.e. of chemical potential) in a liquid solution or mixture. By the same token, the analysis shows the origins of the force which would initially propel the particle, and provide orientation of a composite particle, according to the torque given by eqn (21). The same kind of analysis might be relevant to liquid droplets, except that only one type of surface can be envisaged, and greater movement of the components of the fluids is expected. Other predictions of a more qualitative nature have been made.It has been shown by a more rigorous analysis than that used before that particles in a density gradient column might be ascribed an incorrect density (some aspects of this problem having been recognized for several years). Also, floating objects at liquid surfaces might be subjected to a comparable propelling force, and particles in the bulk phase might become separated during motion, if their values of d a / d l and/or if their surface areas differ. Particle motion, even based on the simple assumption the ideal forces would be maintained and balanced by a modified form of Stokes’ law,22 is not discussed : Anderson and P r i e ~ e ~ ~ gave a detailed account of the hydrodynamic aspects of motion.It is also not possible at this time to quantify the many predictions that have been made. This paper has been developed from considerations of the possible errors in density- gradient column measurements (being made at Bristol by R. J. Manning in 1968), and it benefitted from discussions with the late A. J. B. Cruickshank and J. M. Haynes. It has been greatly extended, and subsequently improved by discussions and correspon- dence, notably with J. L. Anderson, D. G. Hall, J. C. Melrose and D. C . Prieve, whose contributions are also gratefully acknowledged. Glossary U A constant in eqn (22) area of solid/liquid interface; 6A area within element 62; A’ and A” areas of portions of solid area of solid of discrete type s number of components in liquid phase concentration of species i in a solution or mixture the Faraday constant Helmholtz free energy are respectively, the restraining force to keep particle at rest and the propelling force, 9 = -9,.1280 g G, A*G ni, n np, A*ni S T .F u, u* Xi zi S Z a rg rp s Pi, z Pi 5 P a Particles in a Concentration Gradient acceleration due to gravity Gibbs function, U+ PV- TS; see eqn (5) number of moles of i and total number in solution surface excess of component i, eqn (8) denotes surface types, i.e.patches cornprizing overall surface entropy absolute temperature torque needed to disorientate a particle: eqn (21) internal energy in systems with active and inactive solids; see eqn (3) and (4) mole fraction of component i in solution distance into a solution from reference wall electrical charge on species i denotes phases present in a system surface excess of component i per unit area relative adsorption of component i given by eqn (9) relates to an elemental slice of the system or solid at a given constant bulk solution composition chemical potential of component i in elemental slice 6z electrochemical potential of i position of solid particle or solid elemental slice in the solution denotes liquid densities specified in the text interfacial tension, usually solid/liquid or liquid/liquid and depending on the location of solid in liquid, i.e.oz, and on surface type, os or os,z or 0’ and 0’’ electrical potential angle of rotation of a particle by a torque denotes quantities relating to system with hypothetical inactive solid, see eqn (4) References 1 N. 0. Young, J. S. Goldstein and M. J. Block, J. Fluid Mech., 1959, 6, 350. 2 V. G. Levich and A. M. Kuznetzov, Dokl, Akud. Nuuk S S S R , 1962, 146, 145. 3 S. S. Dukhin and B. V. Derjaguin, Dokl. Akud. Nuuk SSSR, 1964, 159, 401. 4 W. W. Mansfield, Trans. Furuduy Soc., 1970, 66, 341. 5 T. Kawai and A. Keller, PhiloLi. Mug., 1963, 8, 1993. 6 A. Ziabicki and A. Wasiak, Kolloidn. Zh., 1967, 215, 158. 7 D. A. Blackadder and P. A. Lewell, Polymer, 1968, 9, 249. 8 J. R. A. Pearson, Polymer, 1968, 9, 283. 9 D. C. Prieve, R. E. Smith, R. A. Sander and H. L. Gerhart, J. Colloid Interface Sci., 1979, 71, 267. 10 E. Ruckenstein, J. Colloid Inferface Sci., 1981, 83, 77. 1 1 J. L. Anderson, M. E. Lowell and D. C. Prieve, J. Fluid Mech., 1982, 117, 107. 12 D. C. Prieve, J . L. Anderson, J . P. Ebel and M. E. Lowell, J. Fluid Mech., 1984, 148, 247. 13 T. L. Hill, Thermodynumics Jor Chemists und Bio1ogi.st.s (Addison-Wesley, Reading, Mass., 1968), 14 T. L. Hill, Thermodynamics ufSmull Systems, Part 1 and Part 2 (Benjamin, New York, 1963 and 1964). 15 R. Parsons, in Modern Aspects of Electrochemistry, ed. J. O’M. Bockris (Butterworths, London, 1954), 16 R. Parsons, Can. J . Chem., 1959, 37, 308. 17 R. Parsons, in Physical Chemistry: Enriching Topics From Colloid and Surfuce Science, ed. H. 18 E. A. Guggenheim and N. K. Adam, Proc. R. Soc. London, Ser., A, 1933, 139, 218. 19 E. A. Guggenheim, Trans. Furuduy SOC., 1940, 36, 397. 20 E. A. Guggenheim, Thermodynamics (North-Holland, Amsterdam, 4th edn, 1959). 21 A. I . Rusanov, in The Modern Theory of Cupillurity, ed. F. C. Goodrich and A. I. Rusanov (Akad. 22 e.g. K. E. van Holde, Physical Biochemistry (Prentice-Hall, New Jersey, 1971), chap. 4. 23 J. L. Anderson and D. C. Prieve, Sep. PuriJ Methods, 1984, 13, 67. chap. 3 and 6. chap. 111. van Olphen and K. J. Mysels (Theorex, La Jolla, California, 1975), chap. 17. Verlag, Berlin, 198 1 ). Puper 6/ 1 12 I ; Receiued 4th June, 1986

ISSN:0300-9599    

DOI:10.1039/F19878301269

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

J . Chem. SOC., Faraday Trans. I , 1987,83, 1281-1291 Ionic Solvation in Water-Cosolvent Mixtures Part 14.-Free Energies of Transfer of Single Ions from Water into Water-Ethylene Carbonate and Water-Propylene Carbonate Mixtures Grahame S. Groves, Kamal H. Halawani and Cecil F. Wells" Department of Chemistry, University of Birmingham, Edgbaston, P.O. Box 363, Birmingham B15 2TT The spectrophotometric solvent-sorting method for determining the free energies of transfer for individual ions, AG:(i), from water into a mixture of water and cosolvent has now been applied to cosolvents which increase the dielectric constant and enhance the solvent structure when added in small concentrations to water. The spread of AGY(i) values found, uiz. AG:(i) for i = a cation is negative and AG;(i) for i = an anion is positive, is similar to those found in mixtures where the cosolvent decreases the dielectric constant and enhances the solvent structure or increases the dielectric constant and decreases the solvent structure. Values for AG:(i)e, the free energy of transfer of the charge alone, are found for some large organic anions.The spectrophotometric solvent-sorting methodl for the determination of the free energy of transfer of the proton from water into a water-cosolvent mixture has been applied to mixtures which have a lower dielectric constant than water-using cosolvents such as alc~hols,l-~ acetone,l? dioxan,l- dimethyl sulphoxide,lY ethane-l,2-diol,l~ glycerol1? and ethan~nitrile.~ Of these cosolvents, the alcohols and acetone strongly enhance the structure in liquid water when added in low mole fractions, with dioxan, dimethyl sulphoxide and ethane-l,2-diol having a lesser enhancing effect on the structure, and glycerol having practically no effect.*V7 In all these cases, the free energy of transfer of the proton from water into the water-rich mixture, AG:(H+), is negative, AG,"(X-) determined by combining AG:(H+) with values for AGi(HX) are normally positive and AGF(Mn+), determined by combining AG:(X-) with values for AG:(MX,), are normally negative.l-' This is in contrast to expectations from purely electrostatic considerations in a dielectric continuum having a dielectric constant less than that of water of positive values for AGF(i) for all i ions irrespective of the sign of the charge.lP7 The expectations for the transfer of ions from water into water-cosolvent mixtures with dielectric constants higher than that of water on electrostatic considerations of ions immersed in dielectric continua are that AG:(i) should be negative irrespective of the sign of the charge on i.This has been tested8 for one set of aqueous mixtures using additions of urea to increase the dielectric constant, and the same broad dispersion of values for AG:(i) has been found, generally comparable with the distribution found in aqueous mixtures with a dielectric constant less than that of water: AGF(i) is negative for i = cations and positive for i = anions.8 Moreover, the physical properties of water-urea mixtures suggest, in contrast to the other water-cosolvent mixtures above, that water-urea mixtures have less structure than pure water.8 It is of interest, therefore, to determine AG:(H+), and therefrom AG:(i), in other aqueous mixtures where the effect of adding the cosolvent is to raise the dielectric constant.Such mixtures, for which extensive ranges of E" and solubility measurements exist from which values for AGF(HX) and AG:(MX,) can be calculated, are not available. However, a limited range of values is available in water-propylene carbonate and in water-ethylene carbonate mixtures : the addition of propylene carbonate to water first 12811282 Ionic Solvation in Water-Cosolvent Mixtures raises the dielectric constant slightly, with further additions producing a slight decrease, causing an overall level effect on the dielectric constant before the mixtures separate into two phase^.^ Water-ethylene carbonate mixtures have higher dielectric constants than pure water.lo Experimental Propylene carbonate was of a grade suitable for gas-liquid chromatography and ethylene carbonate was Fluka puriss. grade: all other materials were as used previo~sly.~-~ The concentrations of unprotonated 4-nitroaniline were determined spectrophotometrically at 383 nm and at 25.0 "C, as described ear1ier.l Solutions of 4-nitroaniline in water- propylene carbonate mixtures containing sodium chloride and hydrochloric acid were prepared by transferring pure propylene carbonate (density at 25 "C = I. 198 g ~ m - ~ ) l l volumetrically and making up to a standard volume, to avoid the correction to molar concentrations necessitated by the contraction on mixing propylene carbonate with water.Mixtures containing ethylene carbonate were made up to a standard volume using accurately weighed samples of the solid, the amount of solid needed being calculated from the known densities of water-ethylene carbonate mixtures. lo Results and Discussion Determination of AG(H+) This methodl of deriving values for AG:(H+) considers two sequential processes. First, a singly charged sphere of radius 3r(H,O) [where r(H,O) is the radius of the H,O molecule], H:O(H,O),, is transferred from water (dielectric constant D,) into the mixture (dielectric constant Ds), for which the free energy on electrostatic considerations on the molar (c) scale14 is given by AG(Born) = pE(HZq) [TPn(H,O)+n(RO)]-p,"(H&) [TP(H,O)] = ~ 6r(H20) (Dil -Dil) Ne2 = 167.6 (D;l- Dil) kJ mol-1 (I) at a constant temperature T and constant pressure P for a mixture n(H,O)+n(RO), where N is Avogadro's number and e is the electronic charge.This transfer is restricted to low mole fractions of cosolvent, i.e. essentially water-rich conditions, so that the contributions of other terms to AG,"(H+) for the transfer of a unipositive ion of this size out of water and its initial reception in the mixture, e.g. arising from the cavity collapse in water and cavity formation in the mixture, from dielectric saturation29l2 and from inverse electrostatic terms in rn with n > 1,13 will be very small, especially as AG(Born) forms only a small part of the total AG:(H+). Secondly, solvent molecules around the proton in the mixture relax, producing a free-energy change AG2 = P:(H:olv) [TPn(H,O) + n ( W 1 -PE(HZq) [TPn(H,O) + n(RO)I where the standard state for the proton in the relaxed state and immediately after the transfer from water are both defined in the mixture n(H20) +n(RO)., The latter solvent sorting in the immediate vicinity of the sphere HzO(H,O), in the mixture will be restricted to the process (2) at low mole fractions of cosolvent, where RO = ethylene or propylene carbonate and b 2 5 to allow for solvent exchange immediately outside the sphere H:O(H2O),.For the initial transfer of one mole of H2q between the two solvents in process (I), the free-energy change for the solvent sorting in process (2) is given on the molar scale1 by (H2O),H,+Iv + ROsolv =$ [(H,O),-lROIH~Ol" + H2Osolv + AG2 = - [ROHIRT In (K, [H,O] F,) (3)G.S. Groves, K . H . Halawani and C. F. Wells 1283 + + where ROH = [(H20) o-lRO]H,+,,v, K, = [ROH]/[RO] [Hzq] with H& = (H,O)&,+,,, and F, = y(R0H) y”(H,O)/y(H:J y”(R0). yi is the activity coefficient for solute species i (H,+, and ROH) in the mixture related to the standard state of yi = 1.0 and [i] = 1 .OO m ~ l d m - ~ with yi -+ 1 .O as [i] -+ 0 in that particular mixture,l* and the activity coefficients y” for the bulk components (H20 and RO) of the mixture are defined for each mixture by y” -+ 1.0 as the concentration of the solute species [i] -+ 0.4 These standard states in the mixture for the bulk components are related to the standard states where activity = 1.0 in each of the pure components via + + =fd(H20)Y”’ (H20) y”(H,o) and ARO) =fd(RO)y ”’(RO)y”(RO).Here f(H,O) and f(R0) are the activity coefficients on the mole fraction scale of H 2 0 and RO in the complete mixture of solute and bulk components with respect to activity = 1 .O in the respective bulk component, H20 or RO; y”’(H,O) and y”’(R0) are conversion factors for changing from the molar to the mole fraction scale; andfd(H20) andf,(RO) are degenerate activity coefficients4 defined on the mole fraction scale (x) by &(H2O) ITPn(H2O) + n(RO)l = pi(H20) [ TP(H20)1fRT1nfd(H20) &RO) [TPn(H,O)+ n(RO)] = p:(RO) [ TP(RO)]+RTlnfd(RO) y(i) for a solute species i defined in the mixture can also be related through a degenerate activity coefficient yd(i)* to the situation in pure water via Y’(i) = yd(i) y(i) where y’(i) is the activity coefficient for i in the complete mixture related to the standard state in pure water y’(i) = 1.0 for [i] = 1.0 and [RO] = 0 with y’ -+ 1.0 as [i] -+ 0 as [RO] = 0, and yd(i) is given by ,4(i)[TPn(H20) +n(RO)] = &(i)[TP(H,O)] +RT lnyd(i).The total free energy of transfer on the mole fraction scale is now given by AGY(H+) = AG(Born) + AG2 + RTln (dsMw/dwMs) (4) where subscript w and s indicate pure water and the mixture, respectively, d is density and M is molecular weight with the molecular weight of the mixture M, = 100/[(wt% RO/M,,)+(wt% H20/18.016)]. Values for Ds and ds are available in the literature, but values for K,, F, and [ROH] are determined experimentally. The determination of these latter quantities is done using the spectrophotometric estimation of unprotonated 4-nitroaniline (B) when this is present in low concentrations in the mixture n(H,O)+n(RO) in the presence of added HCl at a constant ionic strength = 1 .OO mol dm-3 maintained by the addition of sodium ch1oride.l Equilibria ( 5 ) and (6) are now involved in addition to (2): + For a constant added concentration of 4-nitroaniline, co = 1.45 x mol dm-3, at constant temperature, an ionic strength = 1.00 mol dm-3 and a total concentration of RO = [RO],,,., eqn (7) can be deduced from a consideration of equilibria (2), ( 5 ) and ( W1284 Ionic Solvation in Water-Cosolvent Mixtures (7) ccR K2F2 cowR cR + wRco -=--- (cR-c) K I F i [Roltot.(co-cR) KIFl[Roltot. where wR is the molar concentration of water in the mixture, Kl and K , are the thermodynamic equilibrium constants for ( 5 ) and (6) with reference to the standard states already specified for each composition of bulk components n(H20) + n(RO), cR is the concentration of unprotonated 4-nitroaniline determined spectrophoto- metrically1 with c, as above for a particular [HCI] and c is the concentration of unprotonated 4-nitroaniline under the same conditions but with RO absent. Fl = y(B)y(H,+,)/y(BH+)y”(H,O) and F2 = y(B)y(ROH)/y(BH+)y”(RO).For other cosolvents mixed with water linear plots have been for CC,/(C, - c) us. c,/ (c, - c,) for constant [cosolvent] and constant temperature with varying [HCl] at c, = 1.45 x mol dm-3. This linearity1-8 confirms that the assumption on which eqn ( 7 ) is based,172T4 viz.KIFlwkl = K;F;w-l, where K;, F; and w are K,, F, and [H20], respectively, in the absence of added cosolvent, holds over the whole range of compositions for which AG:(H+) has been dete~nined.~ In particular, this confirmation disposes of the ~uggestionl~ that more stringent restrictions should be imposed on the composition ranges used because of this assumption: it should be noted that the suggested assumption,14 w = wR, is not invol~ed.~ Moreover, the intercepts of these linear plots coincide with those calculated using K;F;w-l determined in the absence of cosolvent . 1-8 Similar linear plots with appropriate intercepts calculated from an independent determination of K;F;w-l without added RO1 have been obtained when this method is applied at 25.0 “C to a range of concentrations of ethylene carbonate up to 50 wt% and to 5.99 and 12.00 wt % propylene carbonate: the measurements with propylene carbonate are restricted by the separation of the mixtures into two phases which occurs just above the higher concentration in our conditions of 1 mol dm-3 NaCl used to determine the extinction coefficient of 4-nitroaniline.Values for K C C 1 obtained from the (slope)-’ = [RO],,,~K,/c,F, of these experimental plots1 are given in tables 1 and 2. Values for K2F2 determined from the ratio slope/intercept of these plots (using the value of the intercept calculated from K;F;w-l determined independently in the absence of R0)l are also given in tables 1 and 2. As [ROH] = [RO],,,~/([K,F,c,/(co-c,)l+ 11, values for K , = [ROH]/([H+],,,. - [ROH]) ([RO],,,.- [ROH]))can becalculated for each [H+Itot. and these values are also given in tables 1 and 2. For the low concentrations of propylene carbonate to which this investigation is restricted, table 1 shows that the values of K , for each [RO],,,. agree well with the value for K C c 1 for the same [RO],,,.. Similar good agreement between K , and K c C 1 is obtained at the lower concentrations of ethylene carbonate, but as the latter gets higher the values of K , become erratic, as found for all other cosolvents at such high concentrations where [cosolvent - H+] M [H+Itot. making ([H+],,,. - [cosolvent * H+]) = 0 in the calculation of K,. It is calculated that + + + + + + + as found for all other coso1vents.1-8 This experimental conclusion for all cosolvents invalidates the ~uggestionl~ that the involvement of F, = 1 .O should limit the calculation to very low concentrations of cosolvent.Values for AG:(H+) can now be derived using eqn (4) together with eqn (1) and (3). For the mixtures of water with propylene carbonate, the values of D, and d, wereG. S. Groves, K. H. Halawani and C. F. Wells Table 1. Values for K, (dm3 mol-l), KcC1 (dm3 mol-l) and K,F2 for water-propylene carbonate at an ionic strength of 1.00 mol dm3 and at 25 "C total [propylene carbonate] added acidity wt% : 5.99 12.0 /mol dm-3 mole fraction: 0.01 1 0.023 0.10 0.17 0.21 0.42 0.83 slope) 0.296 0.294 0.307 0.325 0.361 50.9 0.313 * 0.00 1 0.3 16 0.3 11 0.323 0.322 0.314 48.4 0.328 * 0.001 1285 Table 2.Values for K,(dm3 mol-l), K2Fz and K,E;-' (dm3 mol-l) for water-thylene carbonate at an ionic strength of 1 .OO mol dm-3 and at 25 "C total [ethylene carbonate] added acidity wt% : 5.00 10.0 20.0 30.0 40.0 42.5 45.0 /mol dm-3 mole fraction: 0.0107 0.0222 0.0487 0.081 0.120 0.131 0.143 0.10 0.16 0.20 0.40 0.80 1 .oo KZFZ KcK1 (from slope) 0.477 0.443 0.437 0.422 0.393 - 39.8 0.398 f 0.002 0.54 0.5 1 0.49 1 0.454 0.425 - 35.8 0.442 f 0.002 0.84 0.7 1 0.69 0.64 0.58 - 27.7 0.57 kO.01 1.04 1.08 1.03 0.89 0.82 - 22.4 0.71 kO.01 - 1.42 1.50 1.33 1.24 1.26 19.2 0.83 * 0.01 - 1.19 1.43 1.60 1.63 1.68 18.0 0.88 f 0.01 - 1.17 1.31 1.71 1.94 1.86 17.6 0.90 k0.01 interpolated from the published datas for 5 , 10 and 15.22 wt% propylene carbonate. Values for AGz were calculated for the concentrations of propylene carbonate at which K, has been determined using eqn ( 3 ) with the values of K,F;l calculated from the slopes only of the plots of CC,/(C, - C) us.c,/(c, - c,) putting F, = 1 .O. [ROH] was calculated from eqn (8) and (9): + + [ROH] = 0.5(A - (A2 -4[RO],,,.)~) A = ([RO],,,, + 1 +K;') and [H,O] = wR from WJR = ( 1 OOOd, - [ RO]t,t. MR,)M,'. From the smooth curve obtained for the plot of AGz us. solvent composition passing through the origin (fig. l), the values of AG2 at 5, 10 and 15.22 wt % were interpolated. The values for AGf(H+) are given in table 3. The same procedure was applied for the1286 Ionic Solvation in Water-Cosolvent Mixtures x2 Fig. 1. Variation of AG2 for eqn (3) for water-ethylene carbonate mixtures (0) and for water-propylene carbonate mixtures (0) at 25 "C with mole fraction of the cosolvent.Table 3. Values of the free energy of transfer of individual ions, AG,"(i)/kJ mol-l, from water into water-propylene carbonate at 25 "C [propylene carbonate] mole wt% fraction H+ Ag+ C1- Br- I- 5.00 0.0092 -0.82 - 0.75 - - 10.0 0.0192 -1.56 - 1.37 - - 15.0 0.0302 -1.83 - - - - 15.22 0.307 -1.84 -1.49 1.55 0.41 0.98 Table 4. Values for the free energy of transfer for individual ions, AGy(i)/(kJ mol-l), and AG,"(i),/(kJ mol-l), from water into water-ethylene carbonate mixtures at 25 "C [ethylene carbonate] AG,"(i) AG,"(i), wt% mole fraction H+ C1- CH,CO, CH2<c0i C02H co, CH2<c0; 5.00 10.00 20.00 30.00 40.00 42.50 45.00 49.75 50.00 50.00 0.0107 -1.18 1.18 - - 0.0222 -2.33 2.25 - - 0.0487 -4.56 4.22 - - -6.3 6.1 - 0.08 1 -7.4 7.4 - - 0.120 0.131 -7.8 7.8 - - 0.143 -7.8 8.0 - - 0.168 -8.1 8.4 - - 0.170 -8.1 8.5" 12.0 10.8 0.170 -8.1 8.4b - - - ~ ~ - ~ _ _ _________ a Using E" values from ref.(1 7). Using E" values from ref. (1 6).G. S. Groves, K . H . Halawani and C . F. Wells 1287 calculation of values of AG,"(H+) for a wider range of concentrations of cosolvent in water-ethylene carbonate mixtures using the values for K,C1 in table 2 with f', = 1.0 and the values for D, and d, interpolated from the published data.lo The plot of AG, us. solvent composition is shown in fig. 1 and the resultant values for AGF(H+) are given in table 4. Free Energies of Transfer of Simple Cations and Anions Values for AG,"(HCl) in water-propylene carbonate on the mole fraction scale have been given by Roy et al.ll determined using the cell Pt, H21 H20 + cosolvent, HCl (AgCl, Ag (11) for 5, 10 and 20 wt% propylene carbonate.Unfortunately, it has been our experience that 20 wt % of propylene carbonate is so close to the point where the mixture separates into two layers for there to be doubt about which side 20% lies. As a result we have decided to restrict our activities to the other two compositions. Values for AG,"(Cl-) calculated by subtracting the values of AG,"(H+) in table 3 from these values for AG,"(HCl) are also given in table 3. The above values for AGF(HC1) were extrapolated to 15.22 wt % to give the value for AG,"(Cl-) in table 3 at this composition. Values for the solubility products for the silver halides have been determined in water and in 15.22 wt % propylene ~arb0nate.l~ The free energies of transfer from water into the mixture can be calculated from these pK, values using AG,"(Ag halide), = 5.71(pK,-pKw) kJ mol-l (12) and these values on the molar scale are converted to the mole fraction scale using AG,"(Ag halide) = AGF(Ag halide), + 1 1.41 log (13) Using AGF(AgC1) and the value for AG,"(Cl-) at 15.22 wt % propylene carbonate, a value for AG,"(Ag+) can be obtained.This latter value can then be used with the values of AG,"(AgBr) and AGF(Ag1) calculated using eqn (12) and (13) to produce values for AGF(Br-) and AG,"(I-) for 15.22 wt% propylene carbonate. All these values for AG,"(i) on the mole fraction scale are contained in table 3. The value for AG,"(HCl) from water-ethylene carbonate can be calculated using two sets of data for cell (1 1) :I6, l7 Halle and BateP give values of E& only for 50 wt % ethylene carbonate, whereas El-Harakany et al.17 record values of E& for a range of solvent compositions.Values for Ek for the solvent mixtures for which AG,"(H+) is known were obtained by interpolation of a plot of E& us. solvent composition using the data of ref. (17). AG,"(HCl), on the molality scale is then given by (14) AG,"(HCl), = 96.5(E: - E,O) kJ mol-l using Eow = 0.2224 V at 25 OC:18 this procedure was also applied to the value of Fm of Halle and Bates at 50 wt% ethylene carbonate. These values for AG,"(HCl), were then converted to the mole fraction scale using AG,"(HCl) = AG,"(HCl), + 1 1.41 log ( ~ 1::6) kJ mol-l and the values for AG,"(H+) in table 4 are subtracted from these data to produce the values for AG,"(Cl-) on the mole fraction scale given in table 4.1288 Ionic Solvation in Water-Cosolvent Mixtures Free Energy of Transfer of Large Anions It has been shownlg for an acid dissociation equilibrium ka Asolv * K O I V + H,+lv with an equilibrium constant Ka that the change in the proton affinity Pa of B- on transferring the equilibrium from water into a mixture water-cosolvent is given by where the superscripts w and s indicate water and the mixture, respectively.For large organic acids A and anions B-, only a very small difference in size and structure exists between A and B- such that AG:(A) = AG:(B),eutral, i.e. the free energy of transfer of uncharged B.Then, the free energy of transfer of the charge on B is given by AG,"(B-), = AG,"(B-) - AG:(B)neutral = AG:(B-) - AG,"(A) = - AP,. (1 8) Such values for AG:(B-)e have been determined for large organic anions in mixtures of water with methano1,l99 2o a ~ e t o n e , ~ ~ 2o d i ~ x a n , ~ * 2o t-butyl and dimethyl sulphoxide.20 pKa values are available for the dissociation of acetic acid21 and malonic acid2, in 50 wt% ethylene carbonate. pKa values for acetic acid have been determined on the molality scale21 from which APa can be calculated on the mole fraction scale using (19) APa = 5.71 (pc-pKa)+AG,"(H+)-5.71 log ('z:") ~ kJ mol-1 with the values for AG:(H+) on the mole fraction scale in table 4. The value for AG:(CH,CO;), in 50 wt% ethylene carbonate is given in table 4.pKa values on the molality scale are available for the dissociation of the first proton, pK,,, and for the dissociation of the second proton, pK2,, from malonic acid2, in 50 wt % ethylene carbonate. Eqn (19) has been applied to these data for the first dissociation using the value of pKla determined independently in water23 and the value of AG:(H+) in table 4 to produce the value of AG: ( CH,, /Ez:H)e given in table 4. Eqn (22) applies for the dissociation of both protons in malonic acid : to produce a value for AG;[CH,(CO;),],, where pK,,, = pKla + pK,,. Values for p K a and p e a in 50 wt % ethylene carbonate give22 PIC;,, = 10.497; excellent agreement exists for p e a = 5.696 at 25 "C between both sets of data,23y24 and with p q a = 2.847 at 25 0C,23 pm,, = 8.543.AG,"[CH,(CO;),], obtained from eqn (20) and AGY(H+) in table 4 is also given in table 4. Comparison of AC(i) Values with those using other Cosolvents Fig. 2 shows that, for both cosolvents, AG:(i) for cations is negative and AG,"(i) for anions is positive, as found in general for all other co~olvents.~-~ This finding supports that using urea as the second component of the mixture,8 that this distribution of values of AGf(i) between cations and anions is the same, irrespective of whether the dielectric constant is greater or lower than that of pure water. Moreover, as mentioned earlier, there is a gradation of the structure-enhancing ability among the cosolvents until, with urea, aG. S. Groves, K. H. Halawani and C.F. Wells 1289 I 0.20 - l o t Fig. 2. Variation of the free energy of transfer of individual ions between water and water-ethylene carbonate mixtures (EC) and between water and water + propylene carbonate mixtures (PC) with mole fraction of the cosolvent at 25 "C. " 0.2 0.4 0.6 0.8 1.0 x2 Fig. 3. Variation of the relative partial molar volumes of water (7, - q) (0) and of ethylene carbonate (V, - c) (0) in water-ethylene carbonate mixtures at 25 "C with mole fraction of ethylene carbonate. structure breaker is obtained,a and this property appears to have little effect on the general distribution of values of AG,O(i) between cations and anions, even though it does affect AG,"(i) for ions containing hydrocarbon-type ligands attached to a central atom.8 With water-propylene carbonate, there is a negative excess volume of mixing which has been ascribed principally to a decrease in the relative partial molar volume of the propylene carbonate, F2 - c, as the decrease in the relative partial molar volume of water, - q, remains essentially zero over this composition range 0 < x, < 0.0307.91290 Ionic Solvation in Water-Cosolvent Mixtures This is comparedg with the sharp decrease in p2 - c found in the same composition range with the alcohols as cosolvents leading to a minimum at low x2,25 which is ascribed to an increase in structure within the 'flickering clusters' of structured waterz6 caused by the external strain imposed by the hydrocarbon groups packed between the clusters of structured water ; correspondingly, the excess viscosity of water-propylene carbonate mixtures increases in the composition range 0 < x, < 0.0307.9 Two sets of data exist for the physical properties of water-ethylene carbonate mixtures.10y27 In both cases, a plot of reciprocal density us.composition is slightly curved. D'Aprano'O has shown that the excess molar volume of the mixture has a minimum at x, x 0.16, which approximates to the highest concentration of ethylene carbonate used here. However, this does not tell us to which component of the mixture the contraction can be ascribed, as done for water-propylene ~arbonate.~ Although the plot of d;l us. composition is a fairly shallow curve, 7, - G and -Vl - c can be calculated for the whole composition range. The result of this treatment is shown in fig.3. This shows that TI-- c approximates to zero for mole fractions x, < 0.17 of ethylene carbonate, but that most of the contraction originates in the negative 7, - in the same composition region; although no minimum is observed in the variation of 7,- with composition, the plot in fig. 3 suggests that a plateau may exist in the region x, M 0.05-0.2 (20-50 wt % ethylene carbonate). It is foundlo? 27 that the viscosity rises gradually as the proportion of ethylene carbonate in the mixture increases. Moreover, D' Aprano et al. find from conductivity measurements10 that a plot of log K(where K is the ion association equilibrium constant) for potassium chlorate in water-ethylene carbonate departs sufficiently from the linearity expected for a dielectric continuum to provide a minimum between 25 and 50 wt% ethylene carbonate.A reasonable conclusion from this overall examination of the physical properties of mixtures of water with either propylene carbonate or ethylene carbonate is that both cosolvents at a low concentration have a structure-forming effect in water, although perhaps small, contrary to an assumption by other authors.2s We can conclude, therefore, for the values of AG;(i), that a similar general distribution, i.e. positive for anions and negative for cations, occurs for mixtures with cosolvents which produce the following combination of effects relative to pure water: structure enhancement and decrease in dielectric constant ;lW7 structure breakage and increase in dielectric constant ;8 and now structure enhancement and increase in dielectric constant. Although the values for AG;(i), for the large organic anions in table 4 seem large compared with the values for AG:(Cl-), they show a close comparison with values for AG:(i), for i = carboxylate ions with other cos~lvents.~~ 5 7 l9, 2o AG;(CH,CO;), and AG: ( CH,, /E:rH), in table 4 compare well with AG;(CH,CO ;), x 1 1-1 7 kJ mol-1 for x, x 0.17 for added ethanol, t-butyl alcohol, dimethyl sulphoxide, acetone or dioxan.,O AG;(Cl-) M 8-12 kJ mol-1 at x, x 0.17 using these latter co~olvents~-~ compares well with AG;(Cl-) in table 4.Water-methanol which has a lower AG;(CH,CO;), (ca. 4 kJ mol-' at x, = 0.1720) also has a lower AG:(Cl-) (ca. 3.5 kJ mol-l at x z = 0.17'9 ,). The large positive increment in AG;(i), in going from B- to B2- in table 4 compares with the large increase in AG,"(i) in going from a singly charged anion, i, to a multiply charged anion, found for water-ethanol, and water-ureas mixtures.References 1 C. F. Wells, Aust. J . Chem., 1983,36, 1739; Trans. Faraday SOC., 1965,61,2194; 1966,62, 2815; 1967, 2 C. F. Wells, J . Chem. SOC., Faraday Trans. 1, 1973, 69, 984; 1976, 72, 601; 1984,80, 2445. 3 C. F. Wells, J. Chem. SOC., Faraday Trans. I, 1974, 70, 694; Thermochim. Acta, 1982, 53, 67. 4 I. M. Sidahmed and C. F. Wells, J . Chem. SOC., Faraday Trans. 1, 1986, 82, 2577. 5 C. F. Wells, J . Chem. SOC., Faraday Trans. 1, 1978, 74, 1569. 63, 147.G. S. Groves, K . H . Halawani and C . F. Wells 1291 6 C. F. Wells, J. Chem. Soc., Faraday Trans. I , 1981, 77, 1515.7 G. S. Groves and C. F. Wells, J. Chem. Soc., Faraday Trans. I , 1985, 81, 1985. 8 G. S. Groves and C. F. Wells, J. Chem. Soc., Faraday Trans. I , 1985, 81, 3091. 9 J. Courtot-Coupez and M. L’Her, C.R. Acad. Sci. Paris, Ser. C, 1972, 275, 103. 10 A. D’Aprano, Gazz. Chim. Ital., 1974,104,91; A. D’Aprano, M. Goffredi, and R. Triolo, Electrochim. Acta, 1976, 21, 139. 1 1 R. N. Roy, W. Vernon, J. J. Gibbons and A. L. M. Bothwell, J. Chem. Soc. A, 1971, 3589. 12 D. R. Rosseinsky, Chem. Rev., 1965, 65, 467. 13 A. D. Buckingham, Discuss. Faraday Soc., 1957, 24, 151; J. S. Muirhead-Gould and K. J. Laidler, Chemical Physics of Ionic Solutions, ed. B. E. Conway and R. G. Barradas (Wiley, New York, 1966), chap. 6. 14 Y. Marcus, Zon Solvation (Wiley, Chichester, 1985), pp. 160-161 and 203-205. 15 M. L’Her, D. Morin-Bozke and J. Courtet-Coupez, J. Electroanal. Chem., 1975, 61, 99. 16 J. C. Hall6 and R. G. Bates, J. Chem. Thermodyn., 1975, 7, 999. 17 A. A. El-Harakany, H. Sadek and A. M . Abdou, Electrochim. Acta, 1981, 16, 1677. 18 H. S. Harned and B. B. Owen, The Physical Chemistry of Electrolytic Solutions (Reinhold, New York, 19 C. F. Wells, J. Chem. Soc., Faraday Trans. I , 1978,74, 636. 20 C. F. Wells, Adv. Chem. Ser., 1979, 177, 53. 21 J. C. Hall6 and R. G. Bates, J. Solution Chem., 1975, 4, 1033. 22 J. C. Hall6 and R. G. Bates, J. Chem. Soc., Faraday Trans. I , 1976, 72, 2866. 23 S. N. Das and D. J. G. Ives, Proc. Chem. Soc., 1961, 373. 24 W. J. Hamer, J. 0. Burton and S. F. Acree, J. Natl. Bur. Stand. USA, 1940, 24, 269. 25 K. Nakanishi, Bull. Chem. Soc. Jpn, 1960,33, 793; F. Franks and D. J. G. Ives, Q. Rev. Chem. Soc., 1966, 20, 1. 26 H. S. Frank and M. W. Evans, J. Chem. Phys., 1945, 13, 507; H. S. Frank and W-Y. Wen, Discuss. Faraday SOC., 1957, 24, 133; G. NCmethy and H. A. Sheraga, J. Chem. Phys., 1962, 36, 3382; 3401; G. Nimethy, Angew. C h e w Int. Ed. En@., 1967,6, 195; W. Laiden and G. Nkmethy, J. Phys. Chem., 1970, 74, 3501. 3rd edn, 1958), p. 456. 27 G. P. Cunningham, G. A. Vidulich and R. L. Kay, J. Chem. Eng. Data, 1967, 12, 336. 28 M. J. Blandamer, J. Burgess, S. J. Hamshere and F. M. Mekail, Trans. Metal Chem., 1979, 4, 77. Paper 6/1137; Received 6th June, 1986

ISSN:0300-9599    

DOI:10.1039/F19878301281

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

J . Chem. Soc., Faraday Trans. I, 1987, 83, 1293-1305 Reaction of Neopentane with Hydrogen over Pd, Pt, Ir and Rh Zbigniew Karpinski,* Wojciech Juszczyk and Jerzy Pielaszek" Institute of Physical Chemistry of the Polish Academy of Sciences, Kasprzaka 44/52, 01 -224 Warszawa, Poland The skeletal reaction of neopentane with dihydrogen has been studied over powders and (1 11) oriented films of palladium, platinum, iridium and rhodium. Various pretreatments of the metals have been used for changing the surface morphology (0,-H, pretreatment, very high temperature reduc- tion etc.). The effect of 0,-H, pretreatment on the course of the reaction was most pronounced in case of palladium (selectivity change) and platinum (overall activity change). The catalytic behaviour of rhodium and iridium suffers only minor changes after these pretreatments : demethylation always predominates.In contrast to unsintered palladium, heavily annealed Pd samples should have a lower density of surface imperfections. This fact seems to be responsible for the negligible activity of sintered Pd surfaces in hydrogenolysis and, in effect, for the higher selectivity toward isomerization. XRD analysis of variously pretreated metal powders shows that the increase in isomerization selectivity may be attributed to the sintering of palladium. At ca. 200 "C iridium powder was only slightly selective toward isomeri- zation (2-3%). This selectivity is limited to the initial stage of the reaction only and to clean, but not carbided, surfaces of iridium. A carbided Ir surface produces unusually large amounts of ethane among the hydro- genolysis products. The comparison between catalytic behaviour of platinum and other Group VIII metals allows us to rationalize the results in terms of surface morphology and the reaction mechanism.The exceptional catalytic behaviour of platinum cf. other Group VIII metals in alkane rearrangement reactions has stimulated a great number of studies involving various model forms of platinum catalysts, like flat, stepped and kinked surfaces of platinum.l? These investigations identified different active centres on the Pt surface. One may inquire whether all the catalytic information collected for well defined surfaces can be of further use for surface characterization of supported metal catalysts which are more complex in their nature, i.e.whether a reaction such as that of neopentane with hydrogen may be used as a probe reaction. The so-called ' surface-structure sensitivity' found several years ago for various Pt catalysts3 already suggested that such a possibility should exist. More recently, Burch and Garla4 found that the hydrocracking activity of Pt/Al,O, catalysts depended on surface morphology. Our intention was to employ the reaction of neopentane conversion for probing the surface of supported metals, especially when so-called strong metal-support interactions were supposed to exist. However, such interactions may be manifested both in topographical as well as in chemical changes (decoration by support species) of the surface of metal particles and it would be difficult to decide a priori to which extent each of these two factors may contribute to the observed changes in a catalytic behaviour.It seemed reasonable to separate both factors. As has already been said, the effect of surface topography is fairly well known in the case of platinum, but it is less clear for other important Group VIII metals like Pd, Ir and Rh, mainly because these metals have 43 1293 FAR 11294 Reaction of Neopentane with H, over Metals been less-extensively studied in alkane conversion reactions. There is also a considerable disagreement as to the catalytic behaviour of Pd5-12 and Ir536v13-18 in bond-shift isomerization. In this work we have decided to investigate differently modified surfaces of unsupported metals (films and powders).Such studies would furnish, albeit indirectly, some information on the nature of active centres. This information will be utilized later for probing the surface of supported metals. The aim of the following study was to prepare (1 11) oriented films of Pd, Pt, Ir and Rh and to check whether the oxidation-reduction procedure has any considerable effect on the catalytic properties of these metals. Chen and Ruckenstein l9 in their TEM study, showed that palladium crystallites with sizes in the range 30-350 A underwent structural changes during heating at temperatures > 350 "C, resulting in the formation of pits. The process of formation of pits and their development into cavities was temperature- and time-dependent. On the other hand, the same authorsz0 found that the pits which had been formed during heating in 0, disappeared during annealing in H, at 550-600 "C.We believe that such a severe H, treatment could diminish the number of surface irregularities, restoring, hopefully, the original (1 1 1) surface. However, it seems reason- able also to expect that heating the defected Pd specimen (after 0, treatment) in H, at lower temperatures (say at 250-300 "C) for a shorter period of time would be sufficient for surface-oxide reduction, but insufficient for the removal of a considerable number of pits (surface vacancies). We decided also to investigate metal powders. Metal powders should sinter during prolonged heating in H, : either the mean metal crystallite size should increase or, at least, a fraction of small metal particles should diminish in the absence of a support which usually plays a role of a stabilizing medium. Also, the density of surface defects (e.g.vacancies) should decrease during annealing. 1 9 9 2o Experimental The (1 11) oriented films of Pd, Pt, Rh and Ir were prepared by epitaxial growth of an evaporated layer of a metal on a mica sheet in the manner described by Anderson and Avery5 for Pt and by Karpinski97 lo for Pd. The entire side wall of the reactor was covered by the freshly cleaved mica sheet. During evaporation [ca. (1-5) x lop5 Torr, 1 Torr z 133.3 Pa] and subsequent annealing in hydrogen for 2 h (5 Torr) the reactor was maintained at 500 "C. Spectrographically standardized palladium and platinum wires, and 99.9% iridium and rhodium wires used for evaporation were purchased from Johnson Matthey.The investigations of the reaction of neopentane (Fluka, puriss.) with hydrogen were conducted in a static circulation system in the manner described previo~sly.~ The extent of (1 1 1) epitaxy for Rh- and Ir-on-mica films was checked by X-ray diffractometry. Both films withdrawn from the reactor showed a very high degree of (111) preferred orientation with respect to the mica base [apparent absence of (200) and (220) peaks of both metals in the X-ray diffraction spectrum]. Additionally, for a special purpose we prepared one unoriented Pd film by evaporation at - 196 "C and a further instant temperature increase to 268 "C. Platinum powder was prepared from H,PtCl, by reduction with hydrazine hydrate.Rh and Ir sponges (grade 1) were purchased from Johnson Matthey, whereas Pd powder (1-1.5 pm, 99.9%) was an Alfa Ventron product. A tubular (0.25 in diametert) reactor was used. After evacuation down to Torr at room temperature, each powder was pretreated with flowing oxygen (300 "C, 0.25 h), then after a short flush with helium, reduction was carried out in a flow of H, prepurified by passing it through an MnO-SiO, column. The reduction was carried out in two stages: first, 1 h at 300 "C, then 1 h at i 1 in = 2.54 cm.Z . Karpiriski, W. Juszczyk and J. Pielaszek 1295 400 "C. The temperature of the reactor was then adjusted for the reaction: ca. 250-300 "C in the case of Pt and Pd, ca. 150-1 80 "C for Rh and Ir. After removing hydrogen from the catalyst by evacuation, a reaction mixture (usually neopentane: H, ratio of 1 : 10 unless otherwise stated) was introduced to the reactor and the reaction progress was followed by g.1.c.X-Ray diffraction measurements were carried out on Rigaku-Denki diffractometer using Cu, nickel-filtered radiation. Profiles were recorded by a step-by-step technique and then corrected for the Lorentz-polarization factor. The mean particle size diameter was calculated from the Scherrer formula and an instrumental broadening correction was applied using the diffraction profile from well annealed palladium filings as a standard. Results Evaporated Films The reaction of neopentane was carried out in the presence of a 10-fold excess of hydrogen. The alkane partial pressure was ca. 1.7 Torr. Table 1 reports experimental results for Pt and Pd films, table 2 for Rh and Ir.The reaction temperatures, usually constant for each metal, were selected in such a way as to obtain comparable activities for all metals studied. The first column of data for each metal shows the activities and selectivities for a typical (1 1 I) oriented film. The next column describes various film pretreatments and kinetic parameters following these pretreatments. The last column of table I concerns the catalytic behaviour of the originally unsintered Pd/Pyrex film and its changes after subsequent film pretreatments. The calculation of turnover frequencies and initial product distributions, i.e. selectivities, was performed in the manner described in ref. (10). Powders Table 3 shows the results of kinetic studies collected for Pd powder.It should be noted that several kinetic runs were carried out on the same sample of Pd powder. The repeated use of the same sample was always preceded by the 'standard pretreatment': 0,, 300 "C, 0.25 h; H,, 300 "C, I h; H,, 400 "C, 1 h. A few experiments were performed with highly sintered metal powders. Table 4 presents the results for iridium powder, and table 5 those for Rh and Pt powders. Discussion The results obtained for each of the four platinum metals investigated will be discussed in separate subsections. In the final subsection we shall try to present our general opinion on the problem of the bond-shift isomerization over transition metals. Palladium The results presented in table 1 for Pd films and in table 3 for Pd powders show certain similarities, but also several differences. The most visible difference is that the selectivity for isomerization of neopentane can be much higher for a Pd (1 1 1) film (ca.56% ) than for Pd powder (in the range 10-30% ). However, the common feature is that the selectivity is very sensitive to catalyst pretreatment for both films and powders. In table 3 we show reaction rates and initial product distributions for the reaction of neopentane with hydrogen for different partial pressures of both reactants. Our intention was not to study the reaction kinetics in the sense of the reaction orders, but rather to observe the effect of reactant partial pressures on the selectivity. However, as may be 43-2c h, u m Table 1. Turnover frequencies and initial product distributions of neopentane conversion catalysed over evaporated films of platinum and palladium reaction initial product distribution (% ) b temperature TOFa film pretreatment 1°C /molecule atom-' s-l CH4 C,H, C3H, i-C4H,, i-C5H,, fresh (1 11) film O,, 245, 0.25; H,, 245, 0.Y 0,, 245, 0.25; H,, 460, 13 O,, 500, 0.5; H,, 245, 0.17 0,, 245, 0.5; H,, 480, 17 0,, 247, 0.33; evacuation fresh (1 1 1) film 0,, 290, 0.25; H,, 290, 0.6 0,, 290, 0.25; H,, 500, 13.5 0,, 500, 0.5; H,, 300, 0.17 0,, 460, 0.33; H,, 500, 15 0,, 282, 0.33; evacuation unsintered film 0,, 284, 0.25; H,, 500, 17.5 0,, 300, 0.5; H,, 500, 19 0,, 480, 1; H,, 490, 18.5 after decomposition of Q-PdHe 245 243 243 247.5 245.5 245 290 289 289 289 282 282.5 268 285 303 300 303 298 platinum 8.03 x 10-5 4.04 x 10-5 1.62 x 10-3 4.33 x 10-5 8.70 x 10-4 3.09 x 10-4 8.11 x 10-4 4.43 x 10-4 2.97 x 10-3d 1.34 x 10-4 2.69 x 10-4 5.85 x 10-3 2.63 x 10-5 2.28 x 10-4 1.15 x 10-4 7.1 x 10-5 9.93 x 10-4 1.30 x lo-, palladium Pd-on-Pyrex film 3.4 2.7 2.1 5.6 3.3 3.6 10.4 18.7 8.1 23.2 3.6 8.5 25.2 13.4 13.4 7.7 3.6 19.4 0.2 1.5 2.4 traces 1.4 2.6 7.3 2.0 7.7 0.3 0.8 26.7 3.8 5.8 1.9 0.9 15.0 - 0.4 1.5 0.2 2.8 1.1 1.6 6.2 10.1 4.1 12.8 0.4 1 .o 8.5 2.6 10.1 3.2 1.5 21.9 8.4 6.7 7.2 9.1 6.0 7.9 25.1 35.2 18.2 53.4 10.5 20.2 39.7 56.3 38.8 31.9 24.7 42.6 87.6 87.7 90.4 80.1 89.6 85.6 55.6 28.6 67.5 2.9 85.2 69.6 - 23.9 32.0 55.3 69.3 1 .o a Turnover frequency (initial); the roughness factor of all films taken as 1, film area = 400 cm2.in the formation of a given product divided by total consumption.for 0.25 h followed by H, treatment (40 Torr) at 245 "C for 0.5 h. 0.5; H,, 500, 10; H, (500 Torr), 25, 10 (Q-Pd-H formation); evacuation at 25-300 "C (Q-PdH decomposition). Expressed as the percentage of neopentane consumed O,, 245, 0.25; H,, 245, 0.5 means heating a film in oxygen (10 Torr) at 245 "C TOF drops drastically after a few minutes of the kinetic run. After; 0,, 300,Table 2. Turnover frequencies and initial product distributions of neopentane conversion catalysed over evaporated films of rhodium and iridium reaction temperature film pretreatment /"C TOFU /molecule atom-' s-l initial product distribution (% )b CH, C,H, C3H, i-C,H,, i-C,H,, fresh (1 1 1) film 0,, 221, 0.25; H,, 221, 0.Y O,, 219, 0.25; H,, 470, 13 0,, 500, 0.5; H,, 219, 0.17 O,, 219, 0.5; H,, 480, 17.5 0,, 220, 1; evacuation fresh (1 1 1) film 0,, 227, 0.25; H,, 227, 0.5 0,, 227, 0.25; H,, 480, 14 0,, 490, 0.5; H,, 226, 0.2 0,, 224, 0.5; H,, 500, 17.5 0,, 227, 0.5; evacuation 220.5 218.5 218 219 218.5 22 1 225.5 246 227 225 225.5 226 228 rhodium 2.18 x 5.67 x 10-3 3.15 x 10-3 1.20 x 10-3 5.81 x 10-3d 2.23 x 10-3 5.60 x 10-3 1.75 x 10-3 4.75 x 10-3 1.49 x 10-3 3.94 x 10-3 2.17 x lo-, iridium 1.04 x lo-, 75.5 18.5 1.8 77.5 20.0 1.2 80.8 19.2 - 82.9 16.3 0.9 69.9 18.7 2.3 82.1 16.4 0.7 91.1 4.7 - 98.2 1.2 - 89.4 7.7 - 87.2 5.4 - 66.9 16.3 3.4 86.6 9.1 0.5 88.2 5.4 - - 4.2 0.7 2.9 7.4 13.4 6.3 3.8 - - - - - - b i c$ As in table 1.Table 3.Reaction of neopentane on Pd powders initial partial pressure/Torr reaction reaction initial product distributionb (% ) max.sample temperature rate conversion pretreatmenta H, neopentane /"C /mol g-l s-l CH, C,H, C,H8 i-C,H,, i-C,H,, (% ) standard standard standard standard standard standard standard standard standard HT reductionc 59.197 118.779 119.635 109.08 1 81.401 38.992 430 60.453 119.570 66.974 5.943 3.197 12.423 25.172 39.562 41.111 61 6.142 40.353 6.682 286 303 303 320 303 320 30 1 318 286 304 321.5 286 302 321.5 33 1 359 383 302.5 321.5 302.5 320 340 355 374 9.18 x 10-9 1.56 x 10-9 4.27 x 10-9 3.16 x 10-9 1.18 x 10-9 8.80 x 10-9 6.20 x 1.40 x 3.27 x 2.56 x 4.20 x 1.50 x 2.30 x 2.30 x 4.89 x lo-', 7.08 x 2.46 x 10-7 2.29 x 10-9 3.43 x 10-9 2.07 x 10-9 4.33 x 10-9 1.91 x 1.77 x 10-8 4.87 x lop8 21.7 29.6 20.1 26.9 20.0 25.5 18.4 19.0 19 13.3 20.8 15.0 20.0 33.4 11.7 13.8 22.1 14.2 23.0 13.6 17.8 24.9 8.4 14.6 3.9 6.9 11.3 36.3 2.7 2.3 8.4 21.8 1.3 2.1 6.6 17.2 1.1 2.4 4.2 10.7 1 5 1.7 4.3 7.3 17.5 3.5 8.7 8.3 22.5 14.7 28.1 0.3 1.7 3.7 8.8 24.3 3.4 8.4 11.3 35.1 1.1 3.1 3.4 7.9 9.5 37.6 1.8 3.8 6.7 16.2 - 59.5 28.8 57.3 42.7 53.1 47.5 52.9 45.3 55 45.5 38.1 39.5 40.3 23.8 58.3 54.1 41.9 55.8 30.7 47.3 49.8 28.4 36.0 41.5 8.0 17.7 0.2 23.7 3.3 25.3 20.8 20 35.3 16.2 33.3 8.9 29.3 26.7 2.9 18.1 35.0 21 .o 1.7 50.0 21.1 - - - 2.56 25.30 1.28 12.26 0.82 7.77 0.30 2.26 0.05 0.43 2.69 0.58 1.74 3.07 0.24 2.30 8.16 0.95 6.85 0.23 1.20 2.88 0.59 1.52 ~ a Standard pretreatment: 0,, 300, 0.25; H,, 300, 1; H,, 400, 1.As in table 1. 0,, 300, 0.25; H,, 700, 12.Table 4.Reaction of neopentane on Ir powders initial partial pressure/Torr reaction reaction initial product distribution* (% ) max. conversion sample temperature rate pretreatmenta H, neopentane /"C /mol g-l s-l CH, C,H, C,H8 i-C,H,, i-C,H,, (% 1 * 3 210.5 7.33 x 10-8 23.2 13.0 4.5 59.3 11.97 x. 189 2.00 x 10- 18.2 5.7 3.8 69.2 3.0 2.99 9: - standard 97.092 standard 115.944 standard 101.610 standard 85.887 standard 530 HT reductionc 99.462 HT oxidationd 102.242 standard 55.028 carbiding" 55.568 10.304 12.437 24.050 40.550 63 10.236 9.906 5.602 5.490 195 213 196 212.5 193 210 215 190 208.5 226 173.5 193.5 196 173.5 193.5 2.13 x 9.50 x 5.95 x 10- 6.73 x 8.15 x 4.20 x lo-', 1.53 x 10-7 1.70 x 10-7 3.39 x 10-9 7.20 x 10-9 2.96 x 10-9 1.19 x 10-9 2.69 x lop8 3.54 x 10-8 1.26 x 19.1 22.6 21.4 26.4 23.0 28.0 26.5 21.5 19.3 24.0 24.0 23.0 25.2 32.1 33.7 6.3 4.5 9.4 4.4 8.6 5.0 13.2 4.6 8.1 5.0 14.5 5.9 5.5 6.8 24.3 23.7 8.1 4.8 12.9 14.7 5.3 6.5 3.6 5.2 10.2 6.2 35.4 13.1 32.9 2.9 67.2 63.7 62.2 55.6 62.3 50.8 58.9 30.5 59.8 51.4 64.2 66.3 57.2 19.0 30.5 2.9 2.8 0.1 1.6 0.8 2.3 8.0 - - .-.1.94 13.30 y 3.77 c E 14.84 h 3.33 7.56 2.50 0.77 4 0.09 % 6.9 2.00 - 0.33 1.9 3.97 1.2 13.85 - 0.55 traces 5.19 a As in table 3. the reactor at 400 "C. As in table 1. 0,, 320, 0.25; H,, 850, 18. 0,, 500, 0.25; H,, 165, 1. After carrying out the reaction at 400 "C and evacuatingCL w 0 0 Table 5. Reaction of neopentane over Rh and Pt powders initial 2? E % partial pressure/Torr reaction reaction initial product distributionb (% ) max.pretreatmenta H, neopentane /"C /mol g-l s-l CH, C,H, C,H, i-C,H,, i-C,H,, (% 1 sample temperature rate conversion s. x standard 107.890 10.434 193.5 210.5 HT reductionC 107.375 10.094 249.5 267 HT oxidationd 101.519 9.953 196.5 standard 102.75 1 10.192 253 inactive up HT reductione 98.893 9.827 to 500 "C HT oxidationf 102.647 10.420 322.5 344.5 rhodium 1.89 x 10-9 2.41 x 10-9 3.20 x 10-7 1.88 x lo-, 1.64 x lo-, platinum 5.28 x 2.41 x 10-9 8.70 x 10-9 18.5 8.7 6.7 30.0 7.4 9.9 25.9 13.3 2.4 30.1 13.0 1.6 30.4 7.6 26.3 4.1 1.9 2.7 4.1 2.3 3.1 3.7 4.7 6.5 3 - 5 58.6 trace 0.47 2 66.1 - 0.38 52.7 2.78 53.2 2.1 2.99 $ 35.6 - 5. s 19.85 10.2 81.0 8.76 z 0 10.7 79.8 0.66 9.7 75.4 1.99 -. a As in table 3. As in table 1. 0,, 300, 0.25; H,, 900, 12. O,, 500, 0.25; H,, 200, 1.0,, 300, 0.25; H,, 810, 12. f 0,, 500, 1; H,, 300, 1.5. 2Z . Karpinski, W. Juszczyk and J. Pielaszek 1301 h c1 ..d 3 4 W sr d d c1 .I Y .I 0 0 1 I D 40 291" 9 9 0 opoppppo pppp I 61 z e Fig. 1. A part of the X-ray diffraction profile (1 11) from two palladium powders: small circles represent a profile of Pd powder after three consecutive experimental runs, whereas bars represent a profile of Pd powder after twelve runs. judged from the comparison of appropriate figures, the reaction orders, ca. +0.8 versus hydrocarbon and ca. - 1 to - 1.5 versus hydrogen generally found for reaction of alkanes with hydrogen21 describe reasonably most of our data (in some cases, especially when the ratio of hydrogen to neopentane was low, a serious self-poisoning effect was found).More interestingly, the selectivity for isomerization is not very pronounced for Pd powders and depends slightly on partial pressures of the reactants (Sis = 10-25%). It must be emphasized that this moderate selectivity persists only for a very low conversion level, usually < 1.5-2.5%, then a rather abrupt decline in isomerization occurs. This suggests that on palladium powder the hydrogenolysis of isopentane (product of isomerization) proceeds much faster than its formation. Another interesting feature is that the isomerization selectivity increases systematically with the run number, indicating some changes in the catalyst powder. At the same time the overall activity of palladium decreases. This may be due to two obvious reasons: sintering and/or poisoning by carbonaceous residues.Although it is difficult to estimate the extent to which each of the suggested reasons may be responsible for the catalytic effect, we would like to stress here the importance of the sintering effect. First, we believed that the rejuvenation of our Pd powder between kinetic runs by oxygen treatment at 300 "C for 15 min (with following reduction) was complete. This, of course, may not be true. However, a more efficient burning off of carbonaceous deposits at higher temperatures (e.g. at 500 "C) would produce surface defects as evidenced by Ruckenstein and Chen.ls. 2o Besides, we checked that in case of Ir powder, for which sintering is much less likely than for palladium, two consecutive runs gave exactly the same catalyst behaviour : identical activity and selectivity.The mean crystallite size diameter of a fresh (unreduced) Pd powder calculated from XRD line broadening was ca. 314-376 A. The mean particle size of the same powder increased to ca. 518 A after the first three runs and showed no significant change after the subsequent nine runs. This suggests that the main sintering takes place during the first three runs and regenerations between the runs. However, a closer inspection of the X-ray diffraction profiles supports our view that a possible rearrangement of the particle size distribution can take place: the XRD profiles after twelve runs exhibit less-broadened profiles at their base (fig. 1). This could be indicative of a lower content of the small diameter particle size fraction,22 i.e.a further1302 Reaction of Neopentane with H, over Metals sintering of the Pd powder. Thus, the fact that the selectivity for isomerization grows, run by run, up to ca. 30% can be related to the sintering of the Pd powder. Now let us consider the behaviour of Pd films. The (1 11) epitaxially oriented Pd film shows high selectivity towards isomerization (ca. 56% and in our earlier work even 70-80%9310). A very short oxygen treatment at 300 or 500 "C with a following short reduction at 300 "C results in an increase of the overall activity with a rather drastic decline of the isomerization. Such a pretreatment (O,, 300 "C and particularly, at 500 "C; H,, 300 "C) produces, as follows from the Ruckenstein and Chen work,1gt20 a certain number of defects (vacancies) in the surface layer of palladium.An increase, but not a decline, in the overall activity of our Pd films after a brief heating in oxygen at 300 "C supports our belief that the rejuvenation of the Pd powder has been complete. However, the introduction of extra defect sites furnished very active centres for hydrogenolysis. Their relative absence (or rather low abundance) for the highly sintered Pd (1 11) film does not facilitate the hydrogenolysis of product isopentane. Thus the production of smooth Pd surfaces either by preparation of fresh Pd (1 11) films or by annealing rough films at 500 "C in H, increases the selectivity toward isomerization. Table 1 shows how the catalytic behaviour of originally rough Pd films may be changed. First, the unoriented (prepared at lower temperature) Pd-on-Pyrex film becomes progressively more selective after two consecutive overnight reductions in H, at 500 "C.The same specimen loses its high selectivity after introducing new defects. This could be realized by transforming a highly sintered Pd film into p-PdH and then decomposing the hydride phase. Such a procedure should result in a disintegration of metal crystallites via a mechanism involving formation and rearrangement of defect leading ultimately to a material which is very active in hydrogenolysis. After having compared the results collected for Pd films and powders one may ask whether it is possible to pretreat our Pd powders with H, so drastically as to get them as highly selective for isomerization as highly sintered Pd (1 1 1) films are.A few attempts have been made in this respect. We tried to sinter Pd powders at higher temperatures (2 500 "C). In this case it was necessary to employ a fused silica reactor and quartz-wool plugs protecting the catalyst bed. Thus one can expect a danger of a formation of Pd-Si compounds during such severe hydrogen re treatment.,^ Our result after heating a fresh Pd powder at 700 "C for 12 h in H, (see table 3) was that the selectivity for isomerization reached 50%, almost matching the appropriate value for Pd (1 1 1) film. However, a closer analysis of our specimen after the reaction does not allow us to look at this result so optimistically. The original very fine Pd powder (1-1.5 ,urn) produced on sintering two fractions: one, a very large lump of metal and the other a small amount of a black powder mixed with a quartz wool plug.The XRD analysis of both fractions showed that the lump was pure palladium, whereas the powder gave a very complex diffraction pattern, suggesting unequivocally the formation of Pd-Si compounds. Thus we cannot say to what extent the selectivity increase after annealing at 700 "C may be attributed to the metal sintering and to the interaction of Pd with Si. Most probably, both factors suppress hydrogenolysis. As a matter of fact, Moss et aZ.24 reported the interaction of palladium with silicon after heating their Pd/SiO, catalyst in H, at 600 "C. Our XRD profiles (not shown) resemble very much the profiles presented by those authors [fig. 6 (a) in ref. (24)]. The temperature employed in our first attempt (700 "C) was suggested by the fact that Wang et aZ.25 had not detected the presence of Pd-Si compounds by XRD after heating their Pd/SiO, catalyst at 700 "C for 2 h in H,.The reduction of Pd powder at 550 "C again caused very low activity and somewhat higher isomerization selectivity as compared to the initial value of ca. 30%. Again, two fractions of the catalytic material have been found: a metallic lump and some fine powder. However, in this case both fractions appeared to be pure palladium, as far as XRD analysis could show. In order to overcome difficulties associated with formation of the lump of Pd (too low metal surface area to follow the reaction) we tried to mixZ . Karpiriski, W. Juszczyk and J. Pielaszek 1303 Pd powder with a pure silica gel (Davison 62, 80-100 mesh). Such a catalytic material after reduction at 550 "C overnight was very selective towards isomerization (selectivity = 70%).However, again a complex XRD pattern was obtained, suggesting the formation of Pd-Si compounds, although the additional peaks originating from the Pd-Si phase were much less pronounced compared to the dominant f.c.c. peaks from the Pd phase. Finally, the reduction of Pd-SiO, at 500 "C gave the typical palladium behaviour: isomerization selectivity of ca. 15% . Summing up, the results collected for differently pretreated powders and films of palladium show that the catalytic behaviour of this metal is governed by its surface murphology. Surface defects are not plentiful on highly sintered Pd films and therefore such specimens are not active in hydrogenolysis.Iridium As seen from table 2, hydrogenolysis of neopentane, mostly to methane, predominates for Ir films. This result supports some earlier finding^.^^ l2 Both low- and high-temperature oxidation (with or without subsequent reduction) change the selectivity pattern only marginally: methane is always the predominant product. We noticed also only modest changes in turnover frequencies (TOF) after 0,-H, treatments. Prolonged (ca. 15 h) heating in H, restores the initial activity of the Ir film. However, the results for Ir powder (table 4) show some limited selectivity toward isomerization (ca. 2-3%). A severe annealing in H, at 850 "C overnight, thus with a possible formation of Ir-Si compounds, led to an increase in the isomerization selectivity up to 8%. The last figure may be compared with the selectivity value of 9-10% reported by Foger and Anderson13 for the same reaction on low dispersed Ir/SiO, catalysts. However, high-temperature oxidation (500 "C) and low-temperature reduction restore, in principle, normal Ir behaviour, i.e.a very modest level of isomerization selectivity (2-3%). XRD analysis of variously treated Ir powders shows only marginal changes in the crystallite size (all being ca: 500-600 A), indicating that the sintering in the case of Ir powder is negligible. Because there are strong indications in the literaturel59 l6 that carbided rather than clean Ir surfaces isomerize neohexane, we performed a kinetic run on a deliberately 'carbided ' iridium surface.The 'carbiding ' was achieved by carrying out the reaction of neopentane with hydrogen at 400 "C. All the neopentane reacted, giving only methane in the gas phase. After pumping out the reactor at 400 "C down to lo-, Torr, a fresh reaction mixture was introduced onto the catalyst at 173 "C. The experiment (table 4) showed no isomerization at all, but a great change in the distribution of hydrogenolysis products was found. An unusually large amount of ethane (while the conversion level was kept low) suggests ethane was a primary product. At the same time, the amount of propane was low. Thus we do not think that ethane may result from the hydrogenolysis of isopentane, but rather we are inclined to believe that C,, species (CH,,, CH,) which exist on the iridium surface after its carbiding recombine with each other or with C, species originating from the neopentane molecule.Why these C,, species do not recombine with iso-C, adfragments to form isopentane in the way suggested by Gault,,l and Foger and Anderson26 is not clear. Possibly the reaction temperature employed in this study was too low. Ponec and found an enhanced isomerization on carbided Ir, but at higher temperature (2 600 K). This also suggests that isomerization via surface carbenes at low temperatures is not feasible. Thus Ponec's results may satisfactorily explain the literature discrepancies on the iridium behaviour in bond-shift isomerization, but only for higher temperatures of the reaction. Since these discrepancies concern also the low temperatures of the reaction, say at ca.200 "C, they still await explanation.1304 Reaction of Neopentane with H, over Metals Rhodium Both Rh films (table 2) and powders (table 5) were inactive in the isomerization of neopentane. This fact is in a general agreement with common opinion on this 6? l8* 21 Again, a small amount of isopentane was produced on Rh powder after its drastic annealing in H, at 900 "C. As in the case of Pd powder heated at 700 "C, we may expect a strong interaction of rhodium powder with silicon coming from the fused silica reactor or quartz-wool plugs. Heating in oxygen seems to destroy this interaction because the normal catalytic behaviour of Rh is recovered, i.e. only hydrogenolysis takes place after the following pretreatment: O,, 500, 0.25; H,, 200, 1.Platinum Both Pt films (table 1) and Pt powders (table 5 ) are very selective towards isomerization, with the selectivity value of ca. 80-90%, irrespective of the kind of pretreatment. However, the results collected for Pt films show interestingly big changes in the overall activity after 0,-H, pretreatment. Apparently low turnover frequencies of a fresh (1 1 1) Pt film, 8.03 x s-l, rises by factor of ca. 160 after O,, 245, 0.25; H,, 245, 0.5 treatment. The extrapolation of the values of 8.03 x s-l at 245 "C to 300 "C, assuming after Davis et a1.l the activation energy of 55 kcal mol-l, gives a TOF value equal to 0.0134 s-l, which is ca. 10-fold less than the overall activity of (1 11) and (100) single crystals of Pt, but is in good agreement with the activity of various Pt/SiO, catalyst^.^^ However, a similar extrapolation adopted for the TOF value after the 0,-H, treatment raises the overall .activity to 2.67 s-l at 300 "C.The last figure is ca. 20-fold higher than the activity of a virtually clean Pt (1 I 1) single crystal. Therefore, we believe that factors other than trivial cleaning factors are responsible for these catalytic changes. First, let us consider the effect of surface (or subsurface) oxygen left after the possibly inadequate reduction at 245 "C. It is known that the oxygen treatment may change the catalytic behaviour of platinum, usually increasing the overall activity.28* 29 However, this effect is not very pronounced (a few times only) for single crystals and low-dispersed Pt catalysts.As a matter of fact the catalytic activity of our Pt film after oxygen treatment without reduction (only evacuation) reveals some 10-fold increase in respect to a fresh (1 1 1) Pt film, but this cannot explain the previously mentioned rise by factor of ca. 160. We believe, following 2oj 2 9 9 30 that the oxidation-reduction cycle results in a rearrangement of surface atoms, creating a high density of defects (vacancies) in the outer layer of platinum. Therefore, the overall activity should be increased not only because the total number of active centres is much higher (an increase of the roughness factor of films), but mainly because the TOF value for a defect site is much higher than for a plane atom. However, the selectivity for isomerization does not change with the surface structure and this fact is in a good agreement with results reported by Davis et a1.l for various flat, stepped and kinked Pt single-crystal surfaces.Comparison of the Catalytic Behaviour of Platinum Metals The comparison between the catalytic behaviour of platinum and palladium allows us to rationalize these results in terms of the reaction mechanism.l73 l 8 ~ 21 Platinum is less covered with C, fragments during hydrocarbon reactions than any of the other Group VIII metals. These C , entities (e.g. adcarbenes) donate electron charge to the surface thus filling up the surface orbitals. This fact, in turn, prevents bond-shift isomerization, which requires vacant surface orbital^.'^^ This is why Pt, but not other Group VIII metals, is good at bond-shift isomerization, but poor at methane production.In contrast the (1 1 1) face of Pd does not seem to be efficient in carbene formation9~ lo, 31 and therefore behaves like Pt (isomerization occurs). However, when we have Pd surface with defects, carbidingZ. Karpiriski, W. Juszczyk and J . Pielaszek 1305 is very pronounced and this removes the benefit of having defects for bond shift and more methane is produced instead. Rooney and 32 observed very extensive bond-shift isomerization of cage compounds (e.g. protoadamantane -+ adamantane) on all metals where carbiding is not easily possible. In conclusion one can say that the problem of the unique catalytic properties of platinum among other Group VIII metals, which has also been discussed by Rooney and seems to consist in the fact that presumably all Group VIII metals (or at least Pt, Pd, Rh and Ir) are capable of isomerization, but rupture of hydrocarbon adspecies to surface carbenes is very pronounced on Rh, Ir and defected Pd, but not on Pd.Z. K. thanks Prof. J. J. Rooney for helpful discussion. This work was carried out within Research Project 03.10. References 1 S. M. Davis, F. Zaera and G. A. Somorjai, J. Am. Chem. Soc., 1982, 104, 7453. 2 S. M. Davis, W. D. Gillespie and G. A. Somorjai, J. Catal., 1983, 83, 131. 3 M. Boudart, A. W. Aldag, L. D. Ptak and J. E. Benson, J. Catal., 1968, 11, 35. 4 R. Burch and L. C. Garla, J. Catal., 1982, 73, 20. 5 J. R. Anderson and N. R. Avery, J. Catal.. 1966, 5, 446. 6 M. Boudart and L. D. Ptak, J.Catal., 1970, 16, 90. 7 M. Hajek, S. Corolleur, C. Corolleur, G. Maire, A. 0. Cinneide and F. G. Gault, J . Chim. Phys., 1974, 8 Z. Karpinski and T. Koicielski, J. Catal., 1979, 56, 430. 9 Z. Karpinski, Nouv. J. Chim., 1980, 4, 561. 71, 1329. 10 Z. Karpinski, J . Catal., 1982, 77, 118. 1 1 A. Sarkany, L. Guczi and P. Tetenyi, Acta Chim. Acad. Sci. Hung., 1978, 96, 27. 12 M. A. McKervey, J. J. Rooney and N. G. Samman, J . Catal., 1973, 30, 330. 13 F. Weisang and F. G. Gault, J. Chem. Soc., Chem. Commun., 1979, 519. 14 K. Foger and J. R. Anderson, J . Cataf., 1979, 59, 325. 15 Z. Karpinski and J. K. A. Clarke, J. Chem. Soc., Faraday Trans. 1, 1975, 71, 2310. 16 M. W. Vogelzang and V. Ponec, in Advances in Catalytic Chemistry II, Proc. 7th Int. Congr., Salt Lake 17 J. G . van Senden, E. H. van Broekhoven, C. T. J. Wreesman and V. Ponec, J . Catal., 1984,87,468. 18 J. K. A. Clarke and J. J. Rooney, Adc. Catal., 1976, 25, 125. 19 J. J. Chen and E. Ruckenstein, J. Phys. Chem., 1981,85, 1606. 20 E. Ruckenstein and J. J. Chen, J. Colloid Interface Sci., 1982, 86, 1 . 21 F. G. Gault, Adv. Catal., 1981, 30, 1 . 22 J. Pielaszek, J. B. Cohen, R. L. Burwell and J. B. Butt, J. Catal., 1983, 80, 479. 23 J. Pielaszek, J. M. Polukarov, Z. V. Semenova and A. Janko, Elektrokhimia, 1977, 13, 878. 24 R. L. Moss, D. Pope, B. J. Davis and D. H. Edwards, 1. Catal., 1979, 58, 206. 25 S-Y. Wang, S. H. Moon and M. A. Vannice, J. Catal., 1981, 71, 167. 26 K. Foger and J. R. Anderson, J . Catal., 1980, 64, 448. 27 K. Foger and J. R. Anderson, J . Catal., 1978, 54, 318. 28 C..E. Smith, J. P. Biberian and G . A. Somorjai, J. Catal., 1979, 57, 426. 29 I. Onal and J. B. Butt, J . Chem. SOC., Faraday Trans. 1, 1982, 78, 1887. 30 M. J. Ledoux, F. G. Gault, A. Bouchy and G. Roussy, J. Chem. SOC., Faraday Trans. 1,1978,74,2652. 31 A. Janko, W. Palczewska and I. Szymerska, J. Catal., 1980, 61, 264. 32 H. A. Quinn, J. H. Graham, J. J. Rooney and M. A. McKervey, J. Catal., 1972, 26, 333. 33 C . O’Donohoe, J. K. A. Clarke and J. J. Rooney, J. Chem. Soc., Faraday Trans. 1, 1980, 76, 345. City, 1982 (Academic Press, New York), in press. Paper 611 157; Received 9th June, 1986

ISSN:0300-9599    

DOI:10.1039/F19878301293

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

J. Chem. Soc., Faraday Trans I, 1987,83, 1307-1314 Micellar Effects upon the Reactions of Complex Ions in Solution Retardation of the Base Hydrolysis of cis-(Azido)(imidazole)bis(ethylenediamine)cobalt(Irr) by Sodium Dodecyl Sulphatel Anadi C. Dash* and Rarnesh C. Nayak Department of Chemistry, Utkal University, Bhubaneswar 751 004, India The kinetics of base hydroloysis of cis-(azido)(imidazole)bis(ethylenedi- amine)cobalt(m) has been studied in the presence of sodium dodecyl sulphate at 40 "C. Unlike the case for univalent anions, the surfactant strongly retards the base hydrolysis of the complex. The rate data have been satisfactorily analysed by the pseudo-phase ion-exchange model : Cbh + Na& Cb& + Nah where Cb+ denotes the imidazole( 1 - ) conjugate base, cis-[(en),Co(im)N,]+, and W and M stand for the aqueous and micellar pseudo-phases, respect- ively.The pseudo-phase ion-exchange equilibium has also been verified by measuring the equilibrium constant using a spectrophotometric method. The second-order base hydrolysis path of Cb+ is, however, absent for the fully micelle-incorporated species, Cb&. Micellar binding of Cb+ also leads to a slowing down of its spontaneous aquation rate to a small extent (by a factor of 2), which may be ascribed to the difference in the polarity of the microenvironment of the reactant in the micellar pseudo-phase and the aqueous phase. It is now well established that surfactants in aqueous media exert significant effects upon the rates and equilibria of a wide variety of reactions by incorporating the reactants into their micellar pseud~-phases.~-~ Most of the reactions which have been studied in the presence of surfactants involve organic substrates.A limited number of studies, however, have demonstrated the rate-influencing effect of sodium dodecyl sulphate (SDS) on the Hg(I1)-induced aquation of (NH,),COC~~+,~ racemization and dissociation of tris( 1,lO- phenanthroline)iron(rI) and tris(2,2'-bipyridyl)iron(11),~ the ligand substitution reactions of some bivalent metal ions7 and the electron-transfer reactions of some metal complexes.*P9 In a preceding paper1 we have shown that both cationic (CTAB) and anionic (SDS) surfactants mediated the rate of substitution of 3N0,-salicylate dianion by water at the cobalt(rr1) centre of the (a~S)<3NO2-salicylato)(tetraethylenepentamine) cobalt(rr1) ion.This paper presents a study of the effect of SDS on the kinetics of base hydrolysis of the cis-(azido)(imidazole)bis(ethylenediamine)cobalt(rrI) ion. It has been reported earlier from our laboratory1° and elsewherell that outer-sphere association of acidopenta-aminecobalt(II1) complexes with anions attenuates their pro- pensity to react with OH-, the effect being much more pronounced for the multivalent anions. In an aqueous medium, SDS at a concentration greater than the critical micelle concentration (c.m.c.) generates an anionic micelle which may be considered to be a macroanion. It was therefore of interest to us to investigate the base hydrolysis reaction of the title complex, a substrate which can exist as a monopositive conjugate base species owing to the NH-deprotonation of the coordinated imidazole, cis-[(en),Co(im)N,]+, in the presence of this anionic detergent.Furthermore, the mechanism of base hydrolysis of the complexes ci~-[(en),Co(imH)X]~+ [imH = imidazole; X = C1-, Br-, N;, NCS-, 13071308 Micellar Effects on Complex Ions in Solution OH(O)C,H,CO;]~~-~~ is a controversial issue in the sense that the spontaneous aquation of the imidazole( 1 - ) conjugate base species, cis-[(en),Co(im)X]+, is kinetically equivalent to the base hydrolysis of its conjugate acid form: +H,O cis-[(en),Co(im)X]+ -+ cis-[(en) ,Co(im)OH2I2+ + X- ~is-[(en),Co(imH)X]~+ + OH- + ~is-[(en),Co(imH)OH]~+ + X- Our aims were therefore (i) to examine the possible interaction of this micelle forming detergent with the cationic substrate, the imidazole( 1 -) conjugate base species in particular, and (ii) to attempt to unravel the mechanism of its base hydrolysis using the micelle as a probe.We expected that the species cis-[(en),Co(im)N,]+, generated in an alkaline medium, could be partitioned into the anionic micellar pseudo-phase of SDS, where its spontaneous aquation could be examined in the absence of its second-order base hydrolysis. This work is therefore expected to provide evidence in support of the spontaneous aquation path of the imidazole(1 -) conjugate base species (see above) postulated by us earlier.12-14 Experimental The cis-(azido)(imidazole)bis(ethylenediamine)cobalt(rIr) dinitrate was prepared and characterized by analytical, i.r., u.v.-visible and lH-n.m.r. spectroscopy as described ear lie^.^ Sodium dodecyl sulphate (Fluka, AG) was purified by repeated crystallisation from ethano1,16 air-dried and stored over fused calcium chloride.The purity of SDS was checked by c.m.c. determination (c.m.c. = 0.0075 mol dm-3; literature value2 0.008 mol drn-,). All other chemicals were of AnalaR grade. The kinetic measurements were made spectrophotometrically using a Varian Cary u.v.-visible recording spectrophotometer model 634s. The cell compartment was thermostatted by circulating water from a constant-temperature bath through the thermostatted cell holder. The reaction was initiated by injecting 0.15 cm3 of a stock solution of the complex into one of the 10 mm quartz cells containing 3 cm3 of the thermally equilibrated reaction mixture of SDS and NaOH of known concentration.Rapid mixing was achieved by quickly shaking the Teflon-stoppered cell several times. The concentration of the complex was (1 .O-1.3) x lo-* mol dm-3. The rate of decrease of absorbance was followed at 335 nm, where the azido complex is the only significantly absorbing species (E,,, = 3543 dm3 mo1-1 cm-l at pH 1). Runs were made under pseudo-first-order conditions and the observed rate constants were calculated from the gradients of plots of In ( A , - A m ) against t . An absorbance change of 0.4'units was monitored for most of the runs. Duplicate runs reproduced the rate constants to 5 % . Results and Discussion Binding of the Substrate to SDS Micelles cis-[(en),Co(imH)NJ2+ is a weak acid, the pKNH of the coordinated imidazole being 10.3 0.1 at 25 "C and I = 0.3 mol A lower value of pKNH at lower ionic strength and higher temperature is expected.l79 l8 The conjugate base species, cis-[(en),Co(im)N,]+, is relatively more absorbing around 320 nm than its acid form, ~is-[(en),Co(imH)N,]~+.Complete NH-deprotonation of the coordinated imidazole at [OH-] = 0.02- 0.06 mol dm-3 is indicated spectrophotometrically, which is in keeping with the pKNH data stated above. The absorption spectrum (280-370nm) of the complex in alkaline medium isA . C. Dash and R . C. Nayak o.8 I 1309 I I I I I 280 300 320 3CO 360 300 X/nm Fig. 1. Effect of SDS on the absorption spectrum of cis-[(en),Co(im)N,]+; [complex], = 1 . 1 1 x mol drn-,, cell path = 10 mm; 0, [NaOH], = 0.02 mol dm-3, A, [NaOH], = 0.02, [SDS], = 0.1 mol drn-,.significantly modified by SDS; the absorption maximum of the conjugate base species, cis-[(en),Co(im>N,]+, at 302 nm shifts to 310 nm in the presence of the surfactant well above its c.m.c. limit (see fig. 1). This is indicative of the interaction of the complex with the anionic micelle. In contrast to this, SDS has little effect upon the absorption spectrum of the acid form of the complex, ~is-[(en),Co(imH)N,]~+, although a relatively stronger interaction between the dipositive complex ion and the anionic micelle is expected. Fig. 2 depicts the variation of ' E , ~ ~ ' at 320 nm (35 "C) with [SDS], at [OH-], = 0.02, 0.04 and 0.06 rnol drn-,. Strikingly enough, E~~~ tends to attain limiting values at each [OH-], which appear to decrease with increasing [OH-],.This leads us to think that both Na+ and the complex ion compete for micellar binding and that the pseudo-phase ion- exchange eq~ilibrium'~ reaction (i) is displaced in the direction favouring more of the complex in the aqueous phase with increasing "a+],: In reaction (i) Cb+ denotes cis-[(en),Co(im)N,]+ and the subscripts W and M stand for the aqueous and micellar pseudo-phases, respectively. The nature of the variation of cobs with [OH-], at a given [SDSIT (see fig. 2) is also inconsistent with significant N- reprotonation of Cb+ in the micellar pseudo-phase forming the less absorbing conjugate acid species, HCb2+M, as the hydrolysis equilibrium Cb& + H,O, e HCbc +OH% will not favour the formation of more of HCbg with increasing [OH-], in the aqueous phase.Furthermore, the absorbance of the reaction mixture containing [OH-], = 0.02, [SDS], = 0.1 and [complex], = 1.79 x lop4 mol dm-, decreased significantly with the addition of NaCl (lo-, E~~~~~~ nm = 5.51, 5.39, 5.08 and 4.92 dm3 mol-l cm-l at1310 Micellar Eflects on Complex Ions in Solution 6.0 --- I L. 0 0 0.1 0 . 2 [ SDS JT/mol dm-3 Fig. 2. Plot of cobs (in dm3 mol-l cm-l) against [SDS], for cis-[(en),Co(im)N,]+ at 320 nm (35 “C); [complex], = 1.13 x [OH-], = 0.02 (O), 0.04 (A) and 0.06 (0) mol dm-3. [NaCl], = 0,0.02,0.1 and 0.15 mol dm-3, respectively). This provides additional support to the competition by the sodium ion for micellar binding, as shown in reaction (i). With the concentration of the micellized surfactant given by [Dn] = [SDS], -C.M.C., “a+], = [NaOH], + [SDS],, [Cb+], = [complex],, and the overall micellar binding parameter given by p = m&, +mSa = [Cb+,]/[Dn] + [Nat]/[Dn] it is easy to express the various concentration terms governing equilibrium (i) as given below : [Cb+,] = m&[Dn] ”a+,] = mS,,[DnI [Cbh] = [Cb’], - [Cb;t;] = [Cb+], - mEb[Dn] “a&] = “a+], - “ah] = “a+], - (D- mEb) [Dn].The equilibrium constant, The concentrations are expressed in terms of the total solution volume, and it is further assumed that the activity coefficient ratio, y,(Na+)/y,(Cb+) and y,(Na+)/y,(Cb+), are each equal to unity. The zero time absorbance data are given by Eobs[Cb+], = e$[Cb$] +c%[Cbit,]. By substituting [Cbk] = [Cb+],-[Cb&] and [Cb+,] = mS,,[Dn] in this expression, the micellar binding parameter for Cb+, m&, can be given asA .C. Dash and R. C. Nayak 131 1 5.0 E 4 . 0 - Table 1. The least-squares best values of the parameters of eqn (3) as calculated using different values of and the c.m.c.a T 0.60 0.867 f 0.046 (0.905 k 0.045) 0.65 0.873 f 0.045 (0.908 f 0.045) 0.70 0.879 f 0.044 (0.912f0.044) 0.75 0.885 & 0.044 (0.916+0.044) 0.80 0.890 f 0.043 (0.920 f 0.044) 0.070 f 0.0 13 (0.048 +O.OlO) 0.076 f 0.014 (0.052 kO.011) 0.082 f 0.01 5 (0.056 f 0.01 1) 0.088 f 0.0 16 (0.060 f0.012) 0.094f0.017 (0.064+0.014) 5.48f0.14 (5.43 f 0.14) 5.48f0.14 (5.43 0.14) 5.47f0.14 (5.42 f 0.14) 5.46 f 0.14 (5.42k0.14) 5.45 f 0.14 (5.41 k0.14) 12.3 f 2.4 (18.9& 1.4) 11.4f 2.2 (17.5 3.8) 10.7 f 2.1 (1 6.2 & 3.5) 10.0 f 1.9 (1 5.2 & 3.3) 9.4f 1.8 (14.2 & 3.1) a c.m.c.= 0.003 (for values in parentheses only) or 0.002 mol dm-3; T = 35.0 "C. 0 10 20 30 X Fig. 3. Least-squares best-line plot of 103/(~ObS-~%), rnol dm-3 cm against X [see eqn (3)] using /3 = 0.75 and c.m.c. = 0.002 mol dm-3 for 0.01 < [SDS],/mol dm-3 < 0.15 and [OH-], = 0.02, 0.04 and 0.06 mol dmP3. where E$, 8% and &,bs denote the molar extinction coefficients of the conjugate base species, Cb+, in the micellar pseudo-phase, aqueous phase and when it is distributed in both the phases, respectively, Eqn (1) can easily be recast as which after substituting for [Cb+],/(m&#h]) [as obtained from eqn (2)] can be rearranged to give: [Cb+]T/(m~b[Dnl) = + x / K e x 1 /(&,bs - E m = 1 /<&% -43 + X{ 1 /(%-I - E 3 &XI (3) where X = ([NaL+IT/{(P- m&)[Dn]} - 1 .o)./3 and c.m.c. were used as adjustable param- eters, the values of which were varied from 0.6 to 0.820 and 0.002 to 0.003 mol dm-3, respectively; the latter was adjudged from the observed increase of absorbance of the complex with [SDS] (see fig. 2). This is also consistent with the earlier observations that electrolytes decrease the c.m.c. value of SDS.3 With E& = (4.33 k0.13) x lo3 dm3 mol-1 cm-l(320 nm), the absorbance data were fitted to eqn (3) by an iterative1312 Micellar Eflects on Complex Ions in Solution Table 2. Rate data for base hydrolysis of cis-[(en),Co(im)N,]+ in an aqueous SDS medium for three values of [NaOH],/mol drn-,: T = 40.0+0.1 "C kv/ 1 O-* s-' 0.02 0.04 0.06 [SDSI /mol dm-3 0.00 0.002 0.005 0.0 10 0.020 0.040 0.060 0.100 0.150 3.17 4.97 7.52 2.97 3.58 6.28 0.96 3.56 6.14 0.80 1.97 3.44 0.46 1.05 2.14 0.35 0.73 1.26 0.28 0.72 1.10 0.24 0.58 0.82 0.20 0.46 0.72 h e a r least-squares program adapted to an IBM 1130 computer; X = {"a+],/ (P[Dn])- 1 .O> was chosen for the first iteration step.The values of the parameters, (&,?jb - &$)-l and Kex, as calculated from the intercepts and gradients of the least-squares best line plots of (&obs-&$)-l against X were found to be insensitive to the values of p and c.m.c. (see table 1). The representative plot of (&obs --&%)-l against X for p = 0.75 and c.m.c. = 0.002 mol dm-3 (see fig. 3) depicts satisfactory fitting of all data at varying [OH-] and [SDS].The mean values of Kex and &Pb (320 nm) turned out to be 15.0 1.6 and (5.42 kO.13) x lo3 dm3 mol-1 cm-l, respectively. The value of Kex is comparable to that for the (CH,),N+/Na+ exchange reaction with SDS micelles {Kex[(CH3),N+/Na+] = 8.8 at 90 0C}.21 Kinetic Model The rate of base hydrolysis of the complex is strongly retarded by SDS (see table 2). Strikingly, this behaviour is similar to the strong retardation of the base hydrolysis of acidopenta-aminecobalt(II1) complexes by multivalent anions reported earlier.lO7 l1 The plots of k , against [SDS], (not shown) at each [OH-], indicate that the pseudo-first-order rate constant decreases rapidly with [SDS], beyond 0.003 mol dm-3 and tends to attain a limiting value.However, in the asymptotic region of the plot, k, increases only slightly with [OH-],, which may be reconciled with the effect of competitive partitioning of Cb+ and Na+ into the micellar pseudo-phase in which the complex undergoes hydrolysis relatively slowly. A reasonable kinetic model is presented in scheme 1, for which the rate law is given by - d[Cb+],/dt = k,[Cb+]T = kZ,[Cb+,] + k,M,[Cb',] where k,, k& refer to the same value of [OH-],; k, = kZs at [SDS], = 0 and k& is the pseudo-first-order rate constant for Cb+ in the micellar phase. Since [Cb+,l/[Cb+,l = Kex"a+,l/"a+,l [see reaction (i)] and [Cb'], = [Chit,] + [Cb+,] the rate law yieldsA . C . Dash and R. C. Nayak 1313 Table 3. Calculated values of the parameters of eqn (7)a c.m.c.kE P /mol dmP3 k$ Ke,/10-4 s-l Kex 110-4 s-1 0.60 0.65 0.70 0.75 0.80 0.85 0.70 0.75 0.80 0.85 0.003 0.003 0.003 0.003 0.003 0.003 0.002 0.002 0.002 0.002 2.67 f 0.64 2.75 f 0.62 2.83 k0.60 2.90 f 0.58 2.96 f 0.57 3.01 k0.56 1.87f0.47 1.99 4 0.46 2.12f0.44 2.22 f 0.43 19.0_+0.5 0.14f 0.03 17.5 _+ 0.5 0.16 f 0.03 16.2 40.5 0.17 f 0.04 15.1 k0.5 0.19f0.04 14.1 k0.5 0.21 f0.04 13.3 f 0.4 0.23 f 0.04 14.1 k0.4 0.13f0.03 13.1 40.4 0.15f0.03 12.3f0.3 0.17f0.03 11.6f0.3 0.1940.04 (k:) mean = (0.17 _+ 0.03) x lop4 s-l a Correlation coefficient = 0.992-0.994. L products Scheme 1. where y = [Na;t,]/[Na&]. The expressions for "ah] and "a&] stated earlier also give Y = (8-mEb> [Dnl/{[Na+l!L'-(P-m~b) [Dnl>* ( 5 ) Eqn (1) on rearrangement yields (m;b)2(Kex - I ) LDn] -mg,(Kex[Cb+], + [Na+]T +flDn] (&x - + Kex p[cb'l'I' = (6) which enables us to calculate mEb.Eqn (5) is then used to calculate y . The calculation of m&, and y under the experimental conditions of the rate measure- ment was made by an iterative procedure. The rate data in the range [SDSIT = 0.01- 0.1s mol dm-3 ([OH-], = 0.02, 0.04 and 0.06 mol dm-3) were fitted to eqn (7), a linearized form of eqn (4): (kowbs - k t y ) / ~ = Kex kty -k& Kex (7) by an iterative least-squares computer program. The value of Kex obtained from the spectrophotometric measurement was chosen as the initial guess and /I and the c.m.c. were used as the adjustable parameters. For each value of /? (in the range 0.6-0.85)20 and the c.m.c. (in the range 0.002-0.003 mol dm-3), the iterative least-squares procedure yielded the best values of Kex and ( k z Kex) as the gradient and intercept of the plot of (k,Wb,-kty)/y against k,.The values of Ke, and k g are to some extent sensitive to the chosen values of p and the c.m.c. (see table 3). The value of Kex obtained from the kinetic data is in satisfactory agreement with that calculated from spectrophotometric measurements. Further, from a consideration of the1314 Micellar Efects on Complex Ions in Solution value of Kex it is evident that the propensity of the species Cb+ for micellar binding is significantly greater than that of Na+. However, it is reasonable to think that the conjugate base species, as required by its charge and the hydrophilic centres, is likely to be present, like Na+, in the water-rich Stern layer of the micelle. The only observable path of hydrolysis of Cb+ in the micellar pseudo-phase is its spontaneous aquation path ( k g ) .This lends indirect support to the existence of the spontaneous aquation path of the conjugate base species of the complex in the aqueous phase. Since the formation of Cb+ is virtually complete the contribution from the second-order base hydrolysis path of the conjugate acid species, cis-[(en),Co(imH)NJ2+, (see introduction) to the overall base hydrolysis rate can be neglected. Based on this the base hydrolysis rate constant (k,wb,) in the absence of SDS and at [OH-], = 0.02, 0.04 and 0.06 mol dme3 given by yielded values of 0.41 x s-l and 1.8 x lo-, dm3 mol-1 s-l for the spontaneous aquation rate constant ( k g ) and the second-order base hydrolysis rate constant (krH) of Cb+, respectively, in the aqueous phase at zero ionic strength from the intercept and gradient of the plot of k,wbs against (yyHyg [OH-]), the activity coefficients being calculated from the Davies equation.22 Using the mean value of kE (see table 3) the ratio kg/k,M, is 2.4.The effect of the micelle on the spontaneous aquation of Cb+ is not large enough to be interpreted in terms of a change in mechanism. This might be related to the difference in the polarity of the microenvironment of the reactant in the aqueous and the micellar pseudo-phases. References 1 Part 1 : A. C. Dash and R. Nayak, Znorg. Chem., 1986,25, 2237. 2 J. H. Fendler and E. J. Fendler, Catalysis in Micellar and Macromolecular Systems (Academic Press, 3 E.H. Cordes and C. Gitler, in Progress in Bioorganic Chemistry, ed. E. T. Kaiser and F. J. Kezdy 4 J. H. Fendler, Acc. Chem. Res., 1976, 9, 153. 5 J. H. Cho and H. Morawetz, J. Am, Chem. SOC., 1972,94, 375. 6 S. Tachiyashiki and H. Yamatera, Bull. Chem. SOC. Jpn, 1984, 57, 1070. 7 P. D. I. Fletcher and B. H. Robinson, J. Chem. SOC., Faraday Trans. I , 1984, 80, 2417, and references 8 K. V. Ponganis, M. A. De Araujo and H. L. Hodges, Inorg. Chem., 1980, 19, 2704. 9 E. Pelizzetti and E. Pramauro, J. Phys. Chem., 1984,88, 990. New York, 1975). (Wiley, New York, 1973), vol. I, p. 1. therein. 10 A. C. Dash and R. K. Nanda, J. Znd. Chem. SOC., 1975, 52, 298. 1 1 M. B. M. Campbell, M. R. Wendt and C. B. Monk, J. Chem. SOC., Dalton Trans., 1972, 1714; M. R. Wendt and C. B. Monk, J. Chem. SOC. A, 1969, 1624. 12 A. C. Dash and S. K. Mohapatra, J. Chem. SOC., Dalton Trans., 1977, 246. 13 A. C. Dash, B. Mohanty and S. K. Mohapatra, Znd. J. Chem., 1984,23A, 312. 14 A. C. Dash and B. Mohanty, Transition Metal Chem. (Wienheim), 1980, 5, 183. 15 R. W. Hay, M. Tajik and P. R. Norman, J. Chem. SOC., Dalton Trans., 1979,636. 16 J. C. Russell and D. G. Whitten, J. Am. Chem. SOC., 1982, 104, 5937. 17 A. C. Dash, M. S. Dash and S. K. Mohapatra, J. Chem. Res., 1979, ( S ) 354; (M) 4531. 18 C. R. Johnson, R. E. Shepherd, B. Marr, S. ODonnell and W. Dressick, J. Am. Chem. SOC., 1980,102, 19 J. H. Fendler, Annu. Rev. Phys. Chem., 1984, 35, 137, and references therein. 20 C. J. Garnett, A. J. Lambie, W. H. Beck and M. Liler, J. Chem. SOC., Faraday Trans. I , 1983,79, 953. 21 C. J. Garnett, A. J. Lambie, W. H. Beck and M. Liler, J. Chem. SOC., Faraday Trans. I , 1983,79,965. 22 C. W. Davies, J. Chem. SOC., 1938, 2093. Paper 6/ 1 186; Rcceived 1 1 th June, 1986 6227.

ISSN:0300-9599    

DOI:10.1039/F19878301307

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. 1, 1987, 83, 1315-1321 Oxidative Coupling of Methane over Samarium Oxides using N,O as the Oxidant Kiyoshi Otsuka" and Tom Nakajima Department of Chemical Engineering, Tokyo Institute of Technology, Ookayama, Meguro-ku, Tokyo 152, Japan Kinetic studies of the partial oxidation of CH, over Sm,O, using N,O and 0, as oxidants have been carried out at 823-950 K and the kinetic features in the reactions with N,O and 0, have been compared. At a low temperature (823 K) and a pressure of the oxidants of 3-13 kPa, the oxidation by N,O produced C,H, with high selectivity, but only deep oxidation took place when 0, was used as the oxidant. However, C,H, was observabIe when the pressure of 0, was lowered below 1 kPa. The reactivity ofOg was two orcters of magnitude greater than that of N,O.0, is a better oxidant than N,O for the production of C, compounds. A reaction mechanism for the oxidation of CH, with N,O is proposed. Kinetic studies of the reaction have shown that N,O decomposition is rate-determining. Kinetic studies of the oxidation with 0, have confirmed the reaction mechanism proposed previously. The partial oxidation of methane, producing methanol and formaldehyde, can be achieved successfully when nitrous oxide is used as the oxygen source over MoO,/Cab- O-Sill, and V,O,/SiO, catalyst3. The partial oxidation of methane by nitrous oxide over ZSM-5-type catalysts gives aromatic compounds in good selecti~ity.~? We have demonstrated that N,O is also a favourable oxidant for the oxidative coupling of methane over La203, Sm,O, and Dy,O,, forming C,H, and C,H, selectively.s However, further studies are needed to determine the features of the reaction in detail itnd to clarify the reaction mechanism.The object of this work was to observe the kinetic features of the oxidation coupling of methane using N,O as the oxidant and to propose a reaction mechanism. The kinetic features of the reactions are compared between the reactions using N,O and 0,. Favourable oxidants, either N,O or O,, for the oxidative coupling of methane will be also discussed on the basis of the yield of C, compounds. Experimental The samarium oxide (Sm,O,) used was reagent grade (purity > 99.9%) in the form of a powder obtained from the Asahi Chemical Industry Co. ; the surface area of the catalyst was 8.4 m2 g-l.The gases used in this study were He (> 99.99%), CH, (> 99.97%), N,O ( > 99% ) and 0, ( > 99 % ). These gases were further purified by passage though a silica-gel column cooled at 77 or 195 K. The experiments were carried out using a fixed-bed reactor with a conventional gas-flow system at a total pressure of 101 kPa. The reactor consisted of a fused quartz tube (10 mm 0.d. x 200 mm). In a typical experiment the reactor contained 7 cm3 of fused quartz sand (for gas preheating) and 0.01-0.25 g of catalyst sandwiched by quartz wool (thickness 1 mm). The catalyst was treated in the reactor at 923 K for 1 h under an oxygen flow of 25 cm3 min-l. After the oxygen treatment a reacting gas mixture was introduced and the catalyst was raised to the desired temperature.All of the gases were analysed by gas chromatography. The column packings were as 13151316 Oxidation of Methane over Samarium Oxide 1001 I 3 1 , -e&+)- 0-0-0- o-o- 1 0 c M 3 k 0 a, Y 4 15 Fig. 1. Effect of temperature on the selectivities and rates of formation of the products: 0, CO,; ., CO; A, C,H,; A, C2H4. Experimental conditions: P(CH4) = 20.3 kPa, P(N20) = 4.5 kPa, and W/F = 0.15 g-cat s ~ r n - ~ . follows: MS13X (1 m) and active carbon (2 m) for the separation of CH,, CO, 0, and N,; Gaskuropack 54 (2 m) for the separation of CO,, N,O, C,H,, C2H6, H,O, C,H6 and C,H,; and TSR-1 10% /Flusin T for the separation of HCHO and CH,OH. A11 kinetic studies were carried out at a conversion of the oxidants (N,O and 0,) of < 5 % .The conversion of methane was proportional to the contact time of the reactants under these experimental conditions. Results and Discussion Partial oxidation of CH,, over Sm,O, using N,O produced C2H6, C,H,, CO, CO,, N, and H,O. C,H, and C,H, were also produced when the conversion of CH, became high.s However, under the experimental conditions in this work, C, compounds were produced in trace amounts. Typically > 90% carbon mass balance and oxygen mass balance were achieved in the analysis. Temperature Effect The rates of formation of the products in the oxidative coupling of CH, by N,O are plotted as functions of temperature in fig. 1. The reaction conditions are indicated in the figure caption. The selectivities of the products are also plotted in the upper part of fig.1 . The selectivities were calculated on the basis of a mole of methane being converted to the products. No oxygen was observed in the gas phase at any of the temperatures examined in fig. 1. A plot of the logarithm of the rate of methane conversion us. l / T showed a good straight line. The apparent activation energy for the conversion of methane calculated from the slope of the line was 155 f 10 kJ mol-l. The formation of C,H, becomes apparent at temperatures > 873 K. The selectivities to C, compounds (C,H, + C,H,) are high ( > 50% ) at any temperature and increased only slightly on raising the temperature (fig. 1). In contrast to this result, the C, selectivities depended strongly on temperature when 0, was used as the ~ x i d a n t .~K. Otsuka and T. Nakajima 1317 P(N,O)/kPa Fig. 2. Effect of pressure of N,O on the selectivities and rates of formation of the products: 0, CO,; 0, CO; A, C,H,. T = 823 K, P(CH,) = 20.3 kPa, and W/F = 0.15 g-cat s ~ m - ~ . P(CH,)/kPa Fig. 3. Effects of pressure of CH, on the selectivities and rates of formation of the products: 0, CO-CO,; A, C,H,; a, N,. T = 823 K, P(N,O) = 4.5 kPa, and W/F= 0.15 g-cat s cm-3. Effect of N,O Pressure The pressure effects of N,O on the rates of formation of the products are shown in fig. 2. The experiments were carried out at 823 K and 20.3 kPa of CH,. The selectivities to the products are also plotted in the upper part of the figure. No C,H, was observed at the experimental conditions in fig. 2. The absence of oxygen in the gas phase was also confirmed here.The results in fig. 2 show that the rate of formation of C,H, does not change with the pressure of N,O. The rates of formation of carbon oxides (CO and CO,), however, increase with a rise in N,O pressure. The C , selectivity at low pressures of N,O is high (> 80%).1318 Oxidation of Methane over Samarium Oxide "0 10 20 P(N,O)/kPa Fig. 4. Rate of decomposition of N20 as a function of the pressure of N20: 0, without CH,; 0, with CH, (20.3 kPa). T = 823 K and W/F = 0.15 g-cat s ~ r n - ~ . Effect of CH, Pressure Fig. 3 shows the rates of formation of the products and the selectivities to C,H, and carbon oxides as functions of the pressure of CH,. There were neither 0, nor C2H4 in the gas phase. The rate of N, formation was also plotted in fig.3. The results in fig. 3 show that the rate of C,H, formation does not depend on the pressure of CH,. However, deep oxidation of CH, (CO and CO, formations) increases with a rise in the pressure of CH,. Thus C, selectivities decrease gradually with CH, pressure. The conversion of N,O (N, formation) is also enhanced on raising the pressure of CH,. Decomposition of N,O The decomposition of N,O on the Sm,O, was examined in the absence of CH,. The temperature effect studies on the rate of decomposition of N,O gave the apparent activation energy of the reaction 167 f 10 kJ mol-1 under the experimental conditions; P(N,O) = 4.5 kPa, the weight of Sm,O, 0.250 g, and T = 800-973 K. The activation energy was the same as that for the conversion of CH, (1 55 10 kJ mol-l) within experi- mental error.Fig. 4 shows the effects of pressure of N,O on the rates of conversion of N,O in the presence (filled circles) and the absence of CH,. In both cases the rates of N,O conversion increase linearly with N,O pressure. The results in fig. 4 indicate that the rate of conversion of N,O in the presence of CH, is always greater than that in the absence of CH, at a fixed pressure of N,O. The enhancing effect of CH, must increase with increasing pressure of CH,. This idea can explain the observation that the rate of conversion of N,O was increased by increasing the pressure of CH, (fig. 3). Oxidation of CH, by 0, Kinetic studies of the partial oxidation of CH, by 0, over Sm,O, have already been reported by one of Kinetic studies were also carried out in this work.Although the experimental conditions and the reaction system were different from those applied in previous ~ t u d i e s , ~ ? ~ the same conclusions were confirmed in this work. These are summarized as follows. (1) The C, selectivities increased sharply as raising temperatures from 823 to 973 K. (2) The rate equation for the conversion of CH, (r,) could be written asK. Otsuka and T, Nakajima 1319 - z E! Y. \ 2 5 0 2 .- +I (r 0 a, Y 4 0 0.5 1 .o 1.5 P(O2)lkPa 0 Fig. 5. Effect of 0, pressure on the selectivity and rates of formation of C,H, at low pressures of oxygen: 0, CO-CO,; A, C,H,. T = 823 K, P(CH,) = 20.3 kPa, and W/P = 6.3 x g-cat s ~ r n - ~ . where P(CH,) and P ( 0 , ) are the pressures of CH, and O,, respectively. k, K, and KO are the constants at a fixed temperature.This rate equation suggests the reaction mechanism that the abstraction of hydrogen atom from CH, by adsorbed dioxygen is the rate-determining step in the oxidation of CH,. The proposed reaction mechanism is shown CH,(g) * CH,(a) (1) O,(g) * O,(a) CH,(a)+O,(a)-+ CHj+HO; CHj + CHj -+ C2H6 CH; + 0, -+ CH,O; -+ -+ CO, CO,. k in eqn (1) is the rate constant for the rate-determining step (iii). Kq.and KO are the equilibrium constants of the adsorptions of CH, and 0, [steps (i) and (ii)], respectively. In the case of the reaction with N,O, however, the effects of P(CH,) and P(N,O) on rc could not be expressed by such a simple equation as eqn (1). This suggests that the reaction mechanism for the oxidation of CH, by N,O is quite different from those described above [reactions (i)-(v)].The oxidation of CH, by 0, under similar reaction conditions to those in fig. 2, i.e. t = 823 K, P(CH,) = 20.3 kPa, P ( 0 , ) = 3-15 kPa and weight of catalyst 0.0 1 1 g, produced only CO and CO,. However, when the reaction was carried out at a pressure of 0, below 1.5 kPa, the formation of C2H6 was clearly observed. The effect of the pressure of 0, at lower pressures is shown in fig. 5. The C, selectivity increases above 50% as the pressure of 0, decreases below 0.2 kPa. The decrease in the rate of formation of C& on raising P ( 0 , ) must be ascribed to preferential oxidation of CHj [eqn (v)] at high P(0,). Note that the reactivity of 0, with CH, is much greater than that of N,O (cf.the rates in fig. 2 and 5). For example, the rate of conversion of CH, at P ( 0 , ) = 1.3 kPa from1320 Oxidation of Methane over Samarium Oxide fig. 5 is 68 pmol min-l g-l and that from fig. 2 at P(N,O) = 3.2 kPa is 2.6 pmol min-l g-l. The maximum yield of C,H, in fig. 5 is 10 pmol min-l g-l at P ( 0 , ) = 0.3 kPa. In the case of the reaction with N,O, the C, yield was 1.2 pmol min-l g-l (fig. 2). These results clearly indicate that 0, is a better oxidant for the production of C, compounds compared to N,O. Reaction Mechanism As described earlier, no oxygen was observed in the gas phase during the oxidation of CH, by N,O. This fact suggests that the oxygen atom released from N,O on the surface would react quickly with CH,. The same apparent activation energies for the conversion of N,O observed in the presence and absence of CH, strongly suggest that the mechanism for the decomposition of N,O does not change in the presence of CH,.Thus the rate of oxidation of CH, with N,O must be determined by the rate of decomposition of N,O. The observation that the rate of conversion of N,O was enhanced by CH, (fig. 3 and 4) can be ascribed to the increase in the number of active sites for the adsorption of N,O. Methane should reduce surface lattice oxygens, increasing the number of vacant oxygen sites on the surface. The vacant sites generated may become the active sites for the decomposition of N,O. The reaction mechanism that we have in mind is as follows: N,O(a) + e st, N,(g) + 0- (vii) 0 - + e -+ 0,- (viii) (ix) CH, + 0- -+ CHj + OH- CHj + CH; -+ C2H6 CHj +02- -+ CH,O- +e -+ --+ CO, CO, CH,+02- -+ -+ CO, CO,, 2e (xi) where 0- is assumed to be the reactive oxygen species for the activation of CH, [eqn (viii)].l? 2* E.s.r.measurements, however, could not confirm the formation of 0- on the Sm,O, sample after adsorption of N,O at 823-973 K and then quenching to 77 K. The unpaired 4f electrons of Sm3+ might interfere with the detection of paramagnetic oxygen ions by e.s.r. spectroscopy. Decomposition of N,O [step (vii)] is the rate-determining step in the overall reaction. The observation that the conversion of N,O depended on the pressure of N,O to first order (fig. 4) suggests a weak adsorption of N,O(a); i.e. the adsorption can be expressed by Henry's isotherm under the experimental conditions in this work. The lattice oxygen 0,- on the surface may further oxidize CH; radicals to CO and CO, [eqn (x)].Eqn (xi) represents the reduction of surface lattice oxygens by CH,. Reduction of Sm,O, by CH, was examined using a closed gas-circulation apparatus. Although the rate was slow, the reduction proceeded at 873 K. The oxygen vacancies and trapped electrons generated in reaction (xi) must exert favourable effects on the number of active sites for the adsorption of N,O and on the rate of decomposition of N,O [eqn (vii)]. As described earlier, the rate of C,H6 formation did not change with the pressure of N,O (fig. 2). This observation implies that the concentration of the 0- ions on the surface does not depend on the pressure of N,O. The sites for the 0- must be saturated with 0- at very low pressures of N,O. Excess oxygen which could not be accommodated on the sites must be converted into lattice 0,- ions according to reaction (viii). This idea explains the decrease in the C, selectivity on raising the pressure of N,O (fig. 2).K . Otsuka and T. Nakajima 1321 References R-S. Liu, M. Iwamoto and J. H. Lunsford, J. Chem. Soc., Chem. Commun., 1982, 78. H-F. Liu, R-S. Liu, K. Y. Liew, R. E. Johnson and J. H. Lunsford, J. Am. Chem. Soc., 1984,106,4117. K. J. Zhen, M. M. Khan, C. H. Mak, K. B. Lewis and G. A. Somorjai, J. Catal., 1985, 94, 501. S. S. Shepelev and K. G. Ione, React. Kinet. Catal. Lett., 1983, 23, 319. J. R. Anderson and P. Tsai, Appl. Catal., 1985, 19, 141. K. Otsuka and T. Nakajima, Inorg. Chim. Acta, 1986, 120, L27. K. Otsuka, K. Jinno and A. Morikawa, J. Catal., 1986, 100, 353. K. Otsuka and K. Jinno, Inorg. Chim. Acta, 1986, 121, 237. K. Aika and J. H. Lunsford, J . Phys. Chem., 1977,81, 1393. Paper 6/1190; Received 12th June, 1986

ISSN:0300-9599    

DOI:10.1039/F19878301315

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

J. Chem. Soc., Faraday Trans. I, 1987,83, 1323-1346 Tin Dioxide Gas Sensors Part 1 .-Aspects of the Surface Chemistry revealed by Electrical Conductance Variations Jerome F. McAleer,? Patrick T. Moseley,* John 0. W. Norris and David E. Williams Materials Development Division, AERE Harwell, Didcot, Oxfordshire OX11 ORA Detailed measurements are reported of the dependence of the conductance of porous, sintered pellets of tin dioxide, on temperature, moisture and oxygen partial pressure. Comparison is made with the behaviour of thin layers prepared by radio-frequency sputtering. The effect of introducing a non-equilibrium mixture of a combustible gas in air, on the conductance and conductance activation energy is described. The theoretical basis for current interpretations of the behaviour of this material is reviewed.The results are discussed using a model in which the conductance of the porous pellets is controlled by different sorts of intergrain contact, represented as necks, necks depleted of conduction electrons, and Schottky barriers. The results are best rationalised by postulating that the conductance is in fact controlled by Schottky barriers separating domains or agglomerates, each comprising a rather large number of crystallites. The effects of temperature, moisture, oxygen partial pressure and combustible gases are discussed in terms of their effect on the Schottky barrier height by way of alterations in the area density, charge and occupancy of surface states (chemisorbed oxygen species). The existence of a surface state level located 2 1.1 eV below the conduction band edge is deduced.Adsorbed water strongly affects both the conductance and the response to combustible gases. Loss of water from the surface over the temperature range 28W50 "C results in a sigmoidal conductance - temperature relationship in moist air. The effect is to lower the resistance at temperatures below this range to no more than one-tenth of the values observed for dried pellets. A surface transformation 02- ++ OH- is inferred. In dried air, with dried pellets, inflexions in the conductance-temperature behaviour at temperatures below 230 "C are tentatively attributed to the effect of02- ++ 0- ++ 0; surface transformations. The onset of aconductance response to the presence of a combustible gas coincides with the onset of a surface-catalysed combustion.At higher temperatures an effect of com- bustible gases is to lower the conductance activation energy for the porous pellets, but not for the sputtered layers. The effect of moisture on the response to CO was to extend the response to lower temperature; on dried pellets in dried air the response disappeared abruptly when the temperature fell below 350 "C. 1. Introduction Since the demonstration some 30 years ago1.2 that adsorption of a gas on the surface of a semiconductor can bring about a significant change in the electrical resistance of the material there has been a sustained and s~ccessfu1~ effort to make use of the change for purposes of gas detection. Although the original measurements were made on zinc ? Present address: Genetics International (UK) Inc., 1 1 Nuffield Way, Abingdon, Oxfordshire.13231324 Tin Dioxide Gas Sensors oxide, most commercial gas detectors available at present make use of SnO, sensing elements because they offer high sensitivity at a lower operating temperature.* A high surface-area-to-bulk ratio is generally achieved in commercial designs either by employing the material in the form of thin (sputtered) films or as highly porous aggregates of loosely sintered particles. Much recent work regarding SnO, has addressed proprietary sensors containing deliberate additive^.^^ Sensors comprising porous pellets of SnO,, and usually containing a precious-metal catalyst, offer a high degree of sensitivity combined with a light, simple construction at low cost.Their principal disadvantages are that they are relatively non-selective, yielding a response to most reducing gases, and that the need to operate sensors at 300-400 "C results in a significant power requirement. Attempts to achieve improvements in either selectivity or reduced power drain can be most effective if based on a full appreciation of the mechanisms by which such devices function. This paper first addresses the different processes which might control con- duction in porous pellets of SnO, and then discusses these with respect to experimental measurements of the dependence of conductance on oxygen partial pressure, moisture and reducing gases. In subsequent papers, we address the effects of catalysts,6 and present determinations of surface state energie~.~ A preliminary report of this work has appeared elsewhere.* 1.1.The Conductivity Mechanism The equilibration of the bulk of a non-stoichiometric oxide, such as SnO,, with the ambient atmosphere can be written? as:5 05 + 2Sn& i0, + VG + 2Sn&, Snk, n' + Sng, in which the loss of oxygen from the bulk lattice causes the generation of oxygen lattice defects and electron donor states, and carriers. Eqn (1) predicts a dependence of conductance on oxygen partial pressure of the form: o = o, Po:'" exp [ - (AG + ED + E,)/RT] where AG denotes the Gibbs energy change in the reaction which creates the defects [eqn (l)], ED denotes the energy to excite an electron from the donor state created in reaction (1) into the conduction band and E, denotes the activation energy for mobility of the electron in the conduction band.The exponent n depends on the stoichiometry of reaction (1) and on assumptions concerning the charge balance in the lattice. For a reaction involving the creation of doubly charged oxygen vacancies, n would take the value 6.5 Frequently, charge-compensating substitutional impurities are present, in which case n takes the value 4. Sintered, porous pellets of SnO, show a substantial conductivity change when only small concentrations of a combustible gas are present in a large excess of oxygen. The mechanism embodied in eqn (1) cannot explain this observation since the oxygen partial pressure would not sensibly be changed in this circumstance. The assumption, therefore, is that the conductance is controlled by surface processes which are not at equilibrium with the bulk [see for example the review in ref.(5)]. Circumstances in which conductance is controlled by surface processes have been elucidated following the study of thin (typically 100 nm) sputtered l a ~ e r s . ~ - l ~ Current ideas concerning the processes which control conduction in porous, sintered pellets arise 7 The Kroger-Vink notation is used in this paper. A prime denotes one negative charge with respect to the lattice, . denotes a positive charge and x denotes neutrality. Subscripts denote the atom which would normally occupy the site; V represents a vacancy and n' denotes a conduction electron.J. F. McAleer et al. film 1325 conduct ion surface barrier donor level band / 0 -/-/.state distance, f Fig. 1. Thin-film tin dioxide gas sensor, showing schematically the principal energy states and the notation. out of different emphases13-17 placed on concepts developed from the study of such thin layers. In this Introduction, we first briefly describe these concepts with the aim of clearly identifying and quantitatively describing the manner in which the different physical effects considered to control the behaviour of thin layers might affect the behaviour of more massive bodies. Porous bodies and thin films are unlikely to exhibit identical properties, but since both forms are used in gas sensing it is useful to develop an understanding of the limiting processes in each case. 1.2. The Behaviour of Thin Layers Windischmann and Mark12 described the mechanism of operation of devices in which a thin (10-100 nm) layer of SnO, is sputtered onto a suitable insulating substrate.The conductance of this layer is a function of the composition of the gaseous ambient. Windischmann and Mark attempted to account for the following facts. (i) Oxygen must be present, i.e. these sensors respond only to non-equilibrium gas mixtures in which both combustible gases (for example CO, hydrocarbons, H2) and oxygen are present. (ii) There is a temperature of maximum response, i.e. the relative change in conductivity upon introduction of a combustible gas increases with increasing temperature then falls to zero at sufficiently high temperature. They also discussed the dependence of conduction on gas partial pressure, observed for CO : where p is generally found to lie in the range 0.3-0.8. In fig.1, the model discussed by Windischmann and Mark is presented in terms appropriate to the discussion which follows. The nature of the surface states is discussed later and in a subsequent paper.' The depletion layer is characterized by two parameters (see Appendix l), namely the depletion layer thickness, d,,, and the surface potential or 0 K P&) 44 FAR 11326 Tin Dioxide Gas Sensors Schottky barrier height, Ad,. Both d,, and Ads depend upon the charge stored in the surface states. The microscopic model adopted for the response of the channel conductance to a non-equilibrium gas mixture is formulated in terms of a dependence of the surface state occupancy, 8, on the partial pressure of a combustible gas.5 The surface states are identified with adsorbed oxygen species.The combustible gas, R, reacts with adsorbed oxygen and establishes a steady-state occupancy of the surface states which is less than the equilibrium occupancy in air: k l k-1 n’+402(g) $0; (3) R(g) + 0 ~ 3 RO(g) + n’ (4) where n’ denotes a conductance electron. A condition for sensitivity to the partial pressure of the combustible gas is then 1 k-1 4 k2 P,; k1 P& >> k, P, ( 5 ) k-, + k, PA, + k2 PR One should now note a particular feature of this model, namely the assumption of an activated adsorption and desorption of oxygen, such that the condition (5) may apply. Various kinetic schemes have been solved by Strassler and Reis;13 power laws of the form 0 z pa (p denoting partial pressure of combustible gas) were derived, with a having a value in the range 0.5-1, depending on the mechanism (number and charge of surface oxygen species) chosen.The derivation of a square-root dependence of conductivity on CO partial pressure requires further assumptions, a point which is not pursued in the original paper of Windischmann and Mark.12 In particular,13 one needs to assume that the layer is fully depleted of conduction electrons (the conductance is surface-trap- limited, Appendix l), that there are two charged surface oxygen species, only one of which is reactive, and that the rate of interconversion of the species is slow compared to the rate of surface combustion reaction. In line with recent e.s.r. evidence,l* we might denote these two surface states as 0; and 0-, and a kinetic scheme which might be postulated is the following: 20H- 11 H20+e- -1 R-e- $02*Os*0,+0- L 02-.(7) RO In this scheme the steady-state condition of the surface is strongly influenced by the temperature dependence of water adsorption and desorption, as observed (see Sections 3.1 and 3.2). 1.3. The Behaviour of Massive Porous Bodies The foregoing treatment of the influence of surface reactions on conduction mechanisms of tin dioxide is directly applicable to the treatment of dense films with a thickness up to ca. 1 pm. In the present study and in many commercially available gas sensors tin dioxide is deployed with a porous structure (offering a high surface area to the atmosphere) and a gross thickness much greater than 1 pm.Considerations of the mechanism of conduction by such porous artefacts need to take account of the structural inhomogeneity which implies lower-resistance paths through the bulk of crystallites, alternating with higher resistance constrictions at the points of contact; the treatment developed for thin films is useful in assessing this latter aspect.J . F. McAleer et al. 1327 Z Fig. 2. Three models for conductance limited by intergrain constrictions. (a) Open neck, (b) Closed neck, ( c ) Schottky barrier. Three general cases may be considered1’ and in each of them the depletion layer represents a substantial fraction of the material comprising the intergrain contact. The different cases imply different dependence of conductance activation energy on the partial pressure of oxygen and of the combustible gas.Fig. 2(a) shows the well sintered case with a fully open ‘neck’ between adjacent grains. Surface states (here drawn as 0;) cause a depletion zone extending to a depth marked by the dashed line on both sides of the neck. In this case the conductivity would be largely that of the undepleted layer in the centre of the neck. The conductance would be determined by activation of electrons from donor states in the bulk (bulk-trap-limited regime), activation energy ED, and would be affected by the gaseous atmosphere through changes in the effective channel width only. Fig. 2(b) shows the case of a ‘closed’ neck where the depletion zones from the two surfaces overlap leaving a higher-resistance ohmic path through the centre.This geometry could arise because less-complete sintering leaves a narrower neck or it could arise from an open neck being constricted by alterations in the composition of the atmosphere causing a change in the depletion layer thickness. 44-21328 Tin Dioxide Gas Sensors In this case, the conductance in the neck would be determined by activation of electrons from surface states into the conduction band (surface-trap-limited regime), activation energy E,, and would be directly affected by the influence of the gaseous atmosphere on the occupancy of the surface states. The situations depicted in fig. 2(a) and (b) imply that a porous body would respond in the same way as a thin-film gas sensor. The expected variations of activation energy with gas concentration depend on assumptions made concerning the distribution in energy of the surface states.1° At low oxygen partial pressure or high combustible gas concentration, the conductance would be bulk-trap-limited [fig.2 (a)], with activation energy ED. At higher oxygen partial pressure or lower combustible gas concentration, the conductance would become surface-trap-limited7 activation energy E, [fig. 2 (b)]. If the surface states were narrowly distributed in energy (discrete surface state assumption) then there would be a change in activation energy from one distinct value (ED) to another (E,) with changing gas composition, whilst if the surface state had an exponential energy distribution (equivalent to a Freundlich adsorption isotherm) then, in the surface- trap-limited regime the activation energy would vary with the gas composition; a linear relationship between activation energy and log (oxygen partial pressure) is calculated.lo Fig. 2(c) shows a third situation applicable to a porous mass, but not to a thin film sensor. Within the two grains shown there is ohmic behaviour, but at the point of contact a Schottky barrier forms arising out of charge trapped in the surface states. In this case, conductivity would be limited by charge transport across the barrier and would be 0 = constant x exp( -eA4,)/kT. The situation occurs as a limiting case as the necks between the grain get shorter and narrower and it gives effects on the conductance which are, in principle, distinctly different from the two other cases.For example, the activation energy for conductance would be eAq5,. It would be directly affected by the charge and fractional coverage of the surface species and hence a function of the composition of the gaseous atmosphere. The origin of the different emphases adopted by different authors is now evident. Clifford15 has advocated that conductance in sintered pellets is controlled entirely by Schottky barriers between grains, whereas Jones et aI.l4 have implicitly assumed a surface-trap-limited conductance. However, it is also evident that simple distinctions like this might actually be rather difficult to make. For example, a real three-dimensional sintered mass is likely to incorporate more than one type of intergranular contact, and there would always be some route around any given high-resistance boundary,lg unless every intergrain contact was of high resistance [i.e.no contacts such as in fig. 2(a)], or unless the grains were packed into larger domains or agglomerates which were exclusively separated from one another by high-resistance contacts. We also note that each of the three cases illustrated in fig. 2 contains a grain boundary across the contact. Grain-boundary trapping states can themselves establish a depletion layer around the boundary. If the grain size were small enough, then the grains might be depleted of conduction electrons, resulting in a conductance limited by excitation from grain boundary traps.20 Finally, in this section, we note two further effects which distinguish the behaviour of porous masses from that of thin films.First, not all of the interior of a porous mass may be easily accessible to a gas. The effect of closed porosity, or of extremely fine porosity, might indeed be to define domains, such that the gas sensitivity of conductance arises from connections between domains. Secondly, the discussion above, confirmed by observation [Section 3.9 below and ref. (21)] relates the gas response to a surface combustion reaction. A consequence of this surface combustion process would be that the concentration of reactive gas would vary throughout the pellet thickness, and this could have a complex effect on the measured conductance, dependent on the geometry of the system, including the geometry of the contacts. Variability, or changes in domain size and diffusion and reaction conditions withinJ .F. McAleer et al. 1329 Table 1. Properties of tin dioxide powders and pellets (a) powder impurities (ppm) source material ~~ impurity Specpurea FEMh SuperliteC BDHd In < 3 120 140 400 Pb < 30 450 750 < 30 A1 80 80 < 20 < 20 Ca 100 300 150 150 Si 900 900 700 900 Fe 450 100 200 150 (b) pellets open porosity (%) 47 45 pore diameter/nme 250k 100 200f20 crystallite size/A(XRD) 500 & 100 a Johnson-Matthey, spectroscopic grade introduced for reference. Keeling and Walker Ltd, prepared by dehydration of meta- stannic acid. Keeling and Walker Ltd, prepared by burning high- purity tin in air. AnalaR grade. Mercury porosimeter. a pellet, could all account for some of the ageing and irreproducibility observed in practical devices, as we tentatively conclude from this work.1.4. The Effects of Water Vapour Semiconductor gas detectors are sensitive to water vapour, and their response to combustible gases is affected by the ambient humidity.l69 22, 23 Indeed, it has been shown that the surface of tin dioxide adsorbs greater quantities of water than of oxygen.24 The origin of the effect of moisture on gas sensitivity has not been clearly established: Heiland22 and have discussed it in terms of the displacement of chemisorbed oxygen by H 2 0 and OH-. In the current work we have explored the effects of moisture in some detail and attempt to provide a clarification of the likely mechanism involved. Another conclusion of this work is that the effects of water vapour could account for ageing and irreproducibility in practical devices.2. Experimental 2.1. Materials High-purity tin dioxides from three different sources were compared. Analyses are given in table 1. The oxides were calcined as a loose powder to 800 "C in air, pressed (0.1 t cmP2) into cylindrical pellets, 0.3 cm long and 1 cm in diameter and fired at 1000 "C for 16 h. The resultant artefacts offered ca. 45% open porosity. There were small differences, also indicated in table 1, in the microstructure of the pellets from different source powders. Pellets of much lower porosity were obtained from Superlite and BDH material by omitting the step of calcination at 800 "C and then pressing to 5 t before firing. The FEM material did not densify under this treatment. t 1 in = 2.54 em.1330 Tin Dioxide Gas Sensors 2.2.Water As sensor response is very sensitive to the presence of adsorbed water, attention was paid to the determination of the water content of the prepared pellets, and to the determination of the temperature range over which this adsorbed water was lost from the pellet upon heating in dried air. Both chemical analysis of an ethanol extract and thermogravimetric analysis showed an overall water content of ca. 0.2%, lost over the temperature range 240-600 "C. Loss of water upon heating was also monitored using a P,05 hygrometer; only a minor loss was observed around 100 "C, terminating at ca. 170 "C, before the major water loss occurred beginning at ca. 250 "C. It is likely that the water is bound as surface Sn-OH groups.Infrared spectra recorded from a fine dispersion of the solid in fluorolube oil showed trace evidence for 0-H absorption in the 3400 cm-l region. Previous studiesz5 have assigned a band at 3640 cm-l to the OH stretching fundamental of surface SnOH groups, and a broad band centred at 3200 cm-1 to the stretching mode of surface hydroxy groups perturbed by hydrogen bonding. Apparently, isolated hydroxy groups are stable at temperatures up to 500 "C. X-Ray photoelectron spectroscopy combined with removal of thin surface layers by ion bombardment showed no unambiguous evidence for any reduced tin (SnII) concentrated near the material surface. 2.3. Electrical Measurements Gold contacts were evaporated onto the two flat surfaces of the pellet, which was then held between two gold electrodes in a jig, inserted into a gas-tight tube housed in a furnace.The curved surface of the pellet was freely available to the gas atmosphere passing through the tube. Furnace components exposed to the gas stream were made either of quartz glass or high-purity alumina: care was taken to avoid the presence of any catalytically active metals which might have caused combustion of the non- equilibrium gas streams. Gas mixtures (1 % of CO, H, or CH, with 78 % N, and 2 1 % 0,) were certified mixtures from commercial sources (Air Products Ltd and British Oxygen Co.). The control of the furnace temperature, measurement of the specimen temperature and impedance, and alternation of the gaseous atmosphere between the non-equilibrium mixtures and air (dried over molecular sieves), were all performed by a microcomputer.The equivalent parallel resistance and capacitance of the specimen was determined at three different frequencies (100 Hz, 1 kHz and 10 kHz) using an impedance meter (Hewlett Packard) based on a phase-sensitive detector. This instrument limits the voltage across the specimen to < 1 V r.m.s. and the current through the specimen to < 40 mA r.m.s. It is capable of determinations over the range 1 mR to 20 MR and 1 pF to 1 nF to a specified accuracy of 0.2% over much of the range, and a resolution of f 1 ma, f 1 pF or +O. I % . In this work, the range was extended to higher resistances by connecting a precalibrated 15 MSZ resistor in parallel with the specimen and computing the specimen resistance from the measured resistance of the parallel combination.The accuracy of this procedure for resistances > 100 MR obviously becomes poor. The capacitance of the wiring was obtained from a measurement with the specimen holder in the furnace empty and was subtracted from the subsequently measured values. Impedance measurements were taken every ca. 15 s and the average values recorded over each interval of 90 s were stored, together with the specimen temperature at that time. Specimen temperature was determined using a Pt/Pt-13 % Rh thermocouple housed in a thin alumina sheath placed directly in contact with the specimen. The composition of the gaseous atmosphere at the furnace exit was measured using a quadrupole mass spectrometer.J . F. McAleer et al.1331 Table 2. Temperatures of physical changes in porous SnO, T/"C physical change ref. 150 160 227 280 350-400 400 450 520 desorption of 0; 0; - 0- transformation dry slope change (point marked E) water loss begins, minimum in air resistivity ; low-temperature limit of peak in apparent gas response maximum in apparent gas response desorption of water from OH- high-temperature limit in apparent gas response; water loss complete; maximum in air resistivity desorption of 0- or 02- 24 18 fig. 3 fig. 3 fig. 9 ( 4 fig. 9 ( 4 fig. 9 ( 4 fig. 3 24 24 2.4. Thin-layer Devices A layer of tin dioxide (ca. 1 pm) was sputtered onto a set of gold interdigitated electrodes on an alumina tile. The tin dioxide target was a compacted powder and sputtering was promoted by a radiofrequency discharge in an argon atmosphere at 10-2mbar.As prepared, the sputtered tin oxide was slightly coloured, but before use the films were fired in air at 600 "C and this treatment rendered them colourless. 3. Results 3.1. Conductivity in Moist Air In laboratory air (relative humidity ca. 50% ) the resistance of the porous tin oxide pellets showed an unusual sigmoid variation with temperature, reversible on temperature cycling (the track ABCDCBA, fig. 3). In table 2, the temperatures on the turning points of this curve are compared with temperatures for the occurrence of other surface phenomena. In the two Arrhenius regimes shown on the diagram the activation energies were 0.28 eV (lower temperature) and 1.1 eV (higher temperature): a ratio of 4: 1. 3.2. Conductivity in Dried Air In a stream of dried air, the resistance of the porous tin oxide showed a pronounced difference between the first and the subsequent temperature cycles (fig.3). Once the temperature rose above 250 "C on the first cycle water was lost from the pellet and the subsequent behaviour therefore would have been the behaviour of the dry material. Thus, the resistivity rise BC (fig. 3) was due to the loss of water. In moist air, the resistivity fall CB on the decreasing temperature cycle was due to the readsorption of moisture. In dry air, on the decreasing temperature cycle, the resistivity followed the track CEF. On subsequent temperature cycles, the resistivity in dry air varied reversibly along the track FECD. However, the behaviour of the dry material still showed two temperature regimes in the Arrhenius plot (CD and EF, fig.3) in which the slopes were still I. 1 and ca. 0.28 eV (a ratio of 4:1 as noted above): i.e., the lower-temperature activation energy was approximately the same for both wet and dry pellets, although the magnitude of the conductivity differed by more than one order of magnitude.1332 Tin Dioxide Gas Sensors c Q W I 1 1 0 1 2 3 4 lo3 KIT 6. Fig. 3. Resistance-temperature relationship for a porous pellet of SnO, (FEM) in air. (i) Track ABCD: first temperature increase of moist pellet in moist or dry air. (ii) Track DCBA: reversible behaviour of pellet in moist air. (iii) Section BC: loss of moisture from moist pellet in dry air. (iv) Track DCEF: behaviour on temperature decrease, first temperature cycle in dry air.(v) Track FECD: reversible behaviour of dry pellet in dry air. TIoC time - 1.5 1 Fig. 4. Variation of capacitance of an undried 'wet' pellet of SnO, heated in a dried air stream. The smooth curve occupying the lower part of the plot over most of the range represents the resistivity of the pellet (left-hand side). The more jagged curve represents the permittivity of the pellet (right-hand side). 3.3. Apparent Permittivity The apparent permittivity in air of an initially undried pellet heated in dried air (fig. 4) increased between 100 and 200 "C, but then fell sharply as the temperature rose through the range where moisture was lost from the surface. The pattern of behaviour as the temperature fell showed a peak at the same temperature as for the temperatureJ .F. McAleer et al. 1333 21 19 17 15 $ 13 n c: ,9 11 ---. c - 9 I 0 1 2 3 4 lo3 KIT Fig. 5. Comparison of pellets prepared from different source materials (dried pellets in dried air). 1, Superlite; 2, BDH; 3, FEM. increasing, but to a lower maximum value and without any relatively sharp transition. The difference is attributable to the moisture removed at the highest temperature. The apparent permittivity was very high, corresponding to an apparent dielectric constant of ca. 1200. 3.4. Effect of Pellet Microstructure There were differences in the resistivi ty-temperature profiles for pellets made from different preparations of tin oxide. Fig. 5 illustrates these differences. First, there were relatively small differences (up to a factor of ca.50 times) in the absolute magnitude of the apparent resistivity of pellets made from different source material and secondly, pellets of one of the materials (FEM) showed an inflexion in the resistivity in the transition between low- and high-temperature regimes. The other materials (Superlite and BDH) showed a smooth transition. Activation energies in the low- and high- temperature regimes were the same for the different preparations. The pellets showed similar levels of minor impurities, and the different results were speculatively attributed (see Section 4) to the effect of small differences in the particle size and microstructure of pellets from different source material (see Section 2), and to segregation of lead and indium impurities. Pellets of much lower porosity showed a different behaviour (fig.6). In laboratory air, the resistance-temperature profile was not reversible on temperature cycling. After following the track ABC on the first increase of temperature, the resistance cycled along the line CDE, of activation energy 0.9 eV, upon temperature cycling. The behaviour (ABC) on the first temperature cycle is attributed to the loss of water from the interior of the pellet. A simple interpretation would attribute the subsequent behaviour to a bulk-trap-limited conductance, with activation energy equal to the donor energy, ED. This pellet showed virtually no gas sensitivity.1334 Tin Dioxide Gas Sensors 1 0 3 ~ / ~ Fig. 6. Behaviour of a low-porosity pellet. ABCD: first temperature increase (BC: loss of water from pellet); DCE: reversible behaviour on subsequent temperature cycling.3.5. Effect of Oxygen Partial Pressure There are two pertinent observations to report under this heading: first, the need to raise the temperature above the transition range in order to obtain equilibration with atmospheric oxygen and secondly the dependence of resistivity and of activation energy on oxygen partial pressure. Fig. 7 illustrates the first point. During a repetitive temperature cycle, the ambient oxygen partial pressure was changed at the lowest temperature in the cycle (room temperature). Then, whilst the temperature increased through the lower temperature range, the resistivity followed exactly the behaviour seen for the initial oxygen partial pressure ; the specimen equilibrated over the transition temperature range, then showed a different behaviour in the lower temperature range on the reverse temperature cycle, namely that relating to the new oxygen partial pressure.It is over the transition range of temperature that the largest gas responses are observed (see below). Fig. 8 illustrates the variation of activation energy and of resistivity, in the higher temperature range, with oxygen partial pressure. The activation energy varied linearly with log (POQ), with slope ca. 0.2 eV per decade. The sets of isothermal In CT us. In po, data taken from fig. 8 (a) all show slopes of ca. -a. By way of example, the data recorded at 500 "C yield a slope of - 0.29 k0.05. 3.6. Effect of the Introduction of 1% H, At ambient temperature the effect of the introduction of a dry gas mixture containing 17; H,/21% 0,/78% N, was to cause an immediate fall of resistance, FA in fig.3, which could not be reversed by switching the gas stream back to dried air. If the pellet was now heated in the dried air stream, the resistance followed the track ABCD; water was detected in the gas stream over the temperature range BC. Thus, at room temperature,J . I;. McAleer et al. 21 19 1 7 . 1 5 ’ h 5 Q 2 13 11 d 9 1335 . - C 7 I I 0 1 2 3 4 lo3 KIT Fig. 7. Effect of changing oxygen partial pressure during temperature cycling, starting from room temperature (FEM material). 1, First cycle, Po, = 3 x atm (15 volume parts per million, v.p.m.), cycle started at point A. 2, Second cycle, Po, = 3 x atm (1.5%), oxygen partial pressure changed at point B, cycle finished at point C.1.1 1.2 1.3 0.2 0.4 0.6 0.8 Fig. 8. Dependence of resistance on oxygen partial pressure in the high-temperature range (CD in fig. 4). (a) Resistance-temperature characteristics at different Po,. (b) Activation energy us. Pot. 103 KIT barrier heightlev1336 Tin Dioxide Gas Sensors 700 600 500400 300 200 100 0 3 2 100 T/"C 0 0 1.14 elapsed time/ 1 O3 min O A 1 I I ' L 1 ' L ' I I ' Fig. 9. (a) Variation of resistivity of a porous pellet of SnO, (FEM), in air, with pulsed exposure to 1 % H,/78 % N,/2 1 % O,, for 15 min in every 30 min, while the temperature is lowered from 700 "C to room temperature. The peaks represent values in air and the troughs values in air-hydrogen. (b) Variation of the response of an SnO, Pt pellet to pulses of 1 % of hydrogen air as the temperature is changed from 100 to 20 "C and back. Peaks in the resistance (lower) plot represent values in air and troughs are values in air-hydrogen.The upper plot shows that permittivity changes are in the opposite sense. exposure of the dried tin dioxide surface to dilute hydrogen had resulted in a reaction forming surface hydroxy groups. Fig. 9(a) shows the effect of alternating the dried 1 % H, mixture with dried air (1 5 min in each gas mixture) as the temperature was slowly ramped from 700 "C down to 20 "C. The resistivity response to the gas mixture containing hydrogen is clear. Furthermore, treatment of the tin dioxide with platinum extends the response to lower temperatures as described in a later paper5 and a substantial response in the apparent permittivity [shown in fig.9(b)], opposite in sign to the resistivity response, is shown. The introduction of hydrogen caused a decrease in resistance and an increase in permittivity. The magnitude of the permittivity response varied with temperature in exactly the same manner as the resistivity response. When the resistivity response to hydrogen is expressedJ . F. McAleer et al. 1337 ' (6) F l o s s o f H , O A I I I I I 1 10 0 200 300 400 500 3' TI"C Fig. 10. Resistivity-temperature profile for a porous pellet of SnO, (FEM) exposed to an alternating gas atmosphere; 15 min in air, 15 min in 1 % C0/78% N,/21 % 0,. (a) T increasing, (h) 1 % CO, ( c ) air, ( d ) T decreasing. (-) Temperature increasing (initially wet pellet, dried air), (- - -) temperature decreasing (dry pellet, dried air).as a fractional change in resistance, 6R/R, a maximum response at ca. 380 "C is shown. However, what is also clear from fig. 9(a) is that this response-temperature profile reflects the variation in conductance of tin dioxide in moist air. The resistance values shown in fig. 9(a) for those parts of the cycle where air was present agree pretty well with those shown in fig. 3. Furthermore, the air-resistance shown in fig. 9(a) was pretty much unaffected by temperature cycling, paralleling the behaviour shown in moist air, but not in dried air. The explanation is of course that moisture was produced by the combustion of the hydrogen on the surface of the pellet.3.7. Effect of the Introduction of 1% CO No moisture is generated by the combustion of carbon monoxide, so the resistance of the tin dioxide pellet at room temperature in the dried gas streams stayed at a very high value. In fig. 10 resistivity-temperature profiles for a pellet subjected to an alternating gas atmosphere (15 min in air, 15 min in 1 % C0/78% N,/21% 0,) are shown. Fig. 10 compares the behaviour when the air was carefully dried and the temperature was decreasing from a high value (i.e. the pellet was dry, too), with the behaviour when the air stream was dried, but the temperature was increasing from room temperature with a pellet newly inserted in the apparatus (i.e. the pellet was initially wet). The activation energy in the higher temperature range is reduced by exposure to the gas mixture containing carbon monoxide.The activation energy in air containing 1 % CO (0.42 eV) is cquivalent to the value for an oxygen partial pressure of 5 x atmt drawn from fig. 8 (b). A similar effect was observed when the pellet was exposed to dry air containing 1% methane. Several points may be noted. First, the response to carbon monoxide in the presence of moisture was larger, peaked at a lower temperature and extended to lower tempera- tures than the response in the absence of moisture. The response in the absence of i 1 atm = 101 325 Pa1338 11 10 z s 1 - 9 - Tin Dioxide Gas Sensors - - h 18 16 14 12 10 8 4 103K/T 1 I (b) Q t I 1 I I " 1.0 1.1 1.2 1.3 1.4 t .5 lo3 KIT Fig. 11. (a) Resistance-temperature profile for a sputtered layer of SnO,-,.(i) EA = 0.41 eV, (ii) EA = 1.1 eV. (b) Arrenhius diagram illustrating the lack of effect of exposure to CO on the activation energy for the sputtered layer (i) EA = 0.5 eV, (ii) air, (iii) air + 1 % CO. moisture disappeared abruptly when the temperature fell below 350 "C. Fig. 10 illustrates that this might have been due either to a lack of effect of carbon monoxide or to the response abruptly becoming too slow to be observable. The behaviour in the higher temperature range in the presence and absence of moisture was the same. The effect of moisture occurred in the lower temperature and transition ranges. Some authors [e.g. ref. (M)] have claimed that no response to carbon monoxide is observed unless moisture is present: fig. 10 shows that this statement is true only at temperatures < ca.350 "C. 3.8. Comparison of the Behaviour of Porous Pellets with that of Thin Layers Fig. 11 (a) shows the conductance of a 1000 nm thick layer on temperature cycling in air. The behaviour is qualitatively similar to that of the porous pellets. Fig. 11 (b) shows the effect of 1% CO in air, in an experiment in which the gas composition was cycledJ . F. McAleer et al. 1339 as described in Section 3.7. In contrast to the behaviour of the porous pellet (fig. lo), the conductance activation energy for the thin layer device was unaffected by the presence of co. 3.9. Comparison of Electrical Response with Combustion Catalysis Combustion temperatures of the gases on a tin dioxide surface were determined by the analysis of the exit gases from a small reactor containing a bed of the powder.The temperature at the onset of combustion was found to agree well with the temperature observed for the onset of an electrical response, as has been noted previously.21 This temperature was higher for methane (500 "C) than for carbon monoxide (300 "C) than for hydrogen (ca. 30 "C). The effect of the presence of precious-metal catalysts on the electrical combustion response will be discussed in detail in a subsequent paper.6 3.10. Nature of the Lower-temperature Conductance caused by Adsorbed Water It is important to distinguish whether the high conductivity induced at lower temperature by adsorbed water was an effect of electron transport or whether it was due to a proton conduction along the surface in an adsorbed water layer.If the conduction were protonic, then the passage of a direct current would result in the evolution of hydrogen at the negative contact onto the pellet. Furthermore, if a 'wet' pellet was so electrolysed at room temperature in a stream of dried air, the resistance of the pellet should rise as the surface water was consumed. Additionally, if the conduction were protonic, then one might expect a low-frequency relaxation in the impedance of the specimen, contributed by electrode reactions at the contacts. However, we did not observe any low-frequency relaxation; indeed the d.c. resistance of the pellets was the same as the a.c. resistance. Passage of a current of 1 mA for 15 h produced no change in the resistance of a wet pellet exposed to dried air and no hydrogen was detected by the quadrupole mass spectrometer in the exit gas (sensitivity ca.lo-' atm). We conclude, therefore, that the effect of adsorbed water was indeed to change the electronic conductivity of the material. 4. Discussion The results presented above afford the opportunity to test the models for conductivity of porous tin dioxide represented in fig. 2 from several points of view and to arrive at some understanding of the gas response of the material. Four groups of data provide fundamental information about the conductivity mechanism : the apparent permittivity, the variation of resistivity with temperature, the variation of resistivity with oxygen partial pressure and the effects of combustible gases.These aspects are addressed in the following paragraphs. 4.1. Apparent Permittivity and its Variation The apparent permittivity was very high (fig. 4 and 9), corresponding to an apparent dielectric constant of ca. 1200. Such high values of permittivity, in the absence of any permanent electric polarization effects such as are found with certain non- centrosymmetric crystal structures, are normally attributable to the presence of thin insulating layers at the grain boundaries of a semiconducting solid. The variation of permittivity with temperature upon exposure to a combustible gas would then be taken to indicate a variation in the thickness of such a layer. The depletion layer caused by oxygen adsorption at the grain surfaces could constitute such a layer. To give an effect on the apparent permittivity the depletion layer must be normal to the direction of flow1340 Tin Dioxide Gas Sensors of the current; the apparent permittivity would then be expected to vary in antiphase with the resistivity, upon gas cycling, as is observed.The depletion layer thickness is sensitive to the surface charge. The effects of moisture and temperature on the permittivity (fig. 4) therefore imply effects on the surface charge, specifically a marked change in the surface charge associated with the loss of water from the surface on heating above 300 "C, and gradual changes in the surface charge of dry material caused by changes in temperature and presumably attributable to changes in the coverage of oxygen surface species. 4.2. Resistivity-Temperature Behaviour The three models schematically represented by fig.2 would be expected to exhibit distinctly different resistivity-temperature behaviour. Open-neck structures should be ohmic and bulk-trap limited, leading to a single activation energy over the whole temperature range. Fig. 3 shows that the experimental behaviour is not this straightforward. The closed-neck structure has an activation energy related to surface states. Thus, following a sequence of increasingly stable species such as 0; < 0- < 02- as tempera- ture rises might lead to two or three straight-line regions with different slopes in an Arrhenius plot. The data for dry pellets of some of the materials (e.g. BDH, fig. 5) show behaviour cpnsistent with this idea. The Schottky barrier model [fig.2(c)] relates the activation energy directly to the barrier height and hence to the square of the surface charge. If increasing temperature resulted in desorption of surface species and a diminution of surface charge then a curved Arrhenius plot, with slope decreasing with increasing temperature, would be expected. A linear plot would be obtained only in the special case of coverage being temperature- independent. Any surface transformations which resulted in a change in the surface charge (e.g. OH- -P 02-, or 0- -P 02-, or possibly 0; -+ 20-) would result in a sigmoid plot in which the slope would change (by a factor of four for a doubling of the surface charge) over the temperature range of the transition. We have shown that the conductance transformation in moist air correlates with the loss of water from the surface.The conductance behaviour in moist air therefore receives a clear explanation, as does the permittivity variation, in terms of Schottky barriers between grains or domains controlling the conductance, and a doubling of surface charge (assumed to be associated with a transformation OH- -+ 02-), over the temperature range where water is lost from the surface. Indeed, we have shown previously' that, at low temperatures, the I-V characteristic is non-linear and have accounted for this non-ohmic behaviour by considering the conduction as an activated transport across a Schottky barrier. A more rigorous analysislg has given the same result as the simplified analysis. The conductance behaviour of dry pellets in dry air is not susceptible to such a clear explanation. Depending on the material, the Arrhenius plots are either continuously curved, or show an inflexion.The simplest rationalization, although not entirely satisfactory, is again that Schottky barriers control the conductance, but that both the charge on the surface species and their coverage varies with temperature, an interpreta- tion which is also consistent with the observed permittivity-temperature behaviour of dry pellets (Section 4.1 above). The temperature reported1* for the transformation 0; t) 20- (160 "C) is significantly lower than that noted here for the conductance changes in dry air (230-35OoC), and it might be that the behaviour of the conductance is influenced by another possible change, namely 0- ++ 02-.This change clearly involves a doubling of the surface charge, whereas the transformation 0; ++ 0- need not necessarily do so. Another key to understanding the complex pattern of behaviour may be provided byJ . F. McAIeer et al. 1341 the variety in the properties of tin dioxide from different origins (fig. 5 ) . If the rather small differences in specifications between these materials are able to influence the distribution of intergranular contacts between the models in fig. 2 then it is not surprising that they should behave in such diverse ways. Yet another aspect, raised in Section 1.3, is that the behaviour of a porous body might be controlled not by the interconnection between individual grains but by those between domains or agglomerates of the mass.The crystallite size in the fabricated pellets was very small (table 1). We have previously calculateds from the low-temperature I-I/ characteristic, an apparent barrier spacing of around 6pm, much larger than the crystallite size. Another estimate can be obtained from the observed dielectric constant of the pellet (cobs z 1200): the ratio of this to the actual dielectric constant of bulk SnO, where x denotes the pellet thickness (2 mm) and N , the number of barriers across the pellet. If d,, z 10 nm (see Appendix 2) then N , z lo3, or a spacing between barriers of ca. 1 pm. These rough calculations point to the concept that the pellets comprise domains of crystallites and that the control of conductance is exercised at the domain boundaries.4.3. Behaviour of the Non-porous Pellets As noted in Section 3.4, on a simple interpretation the conductance activation energy of the non-porous material (0.9eV) could be identified as the donor energy, ED. However, the bulk-trap-limited case defines a theoretical lower limit on the activation energy for porous pellets, so the data for the porous materials require ED < 0.3 eV. Furthermore, 0.9 eV seems a very large value for a bulk donor energy in SnO,: values of up to 0.15 eV have been mentioned in the literature. We can offer an hypothesis, clearly speculative but nonetheless reasonable, which will rationalise these observations. We presume that the surface states which control conductance in porous pellets are also present in the interior of the mass of the ‘non-porous’ pellets, the difference between these two cases being that in the ‘non-porous’ pellets the interior surface cannot rapidly communicate with the exterior gaseous phase.We then simply suppose that in the absence of any equilibrium with the external gas phase, the surface state occupancy does not vary with temperature. The activation energy for the non-porous material is therefore again interpreted as a surface state energy. 4.4. Dependence of Conductance on Oxygen Partial Pressure The behaviour depicted in fig. 8 may be interpreted in two ways. The relevant equations are set out in Appendix 2. First, the data do not accord with predictions deriving from the assumption of a surface-trap-limited conductance with discrete surface states. However, the calculations of Baidyaroy and Marklo show that a variation of activation energy with log (oxygen partial pressure) such as is shown in fig.8(b) can be obtained with the assumption of a surface-trap-limited conductance and an exponential distri- bution of surface state energies. Baidyaroy and Marklo also showed that such a distribution of surface-state energies was equivalent to the assumption of a Freundlich adsorption isotherm: and gave a z TIT, where T, is a parameter describing the width of the site energy distribution (Section 1.3). Cliffordlj has given a z kT/Eo with the parameter E,, z 0.17 eV for SnO,, i.e. a z 0.3 at 800 K, implying Tc z 2000 K. Baidyaroy and Marklo calculated a variation of activation energy with oxygen partial pressure of 0.2 eV @(P) z P (9)1342 Tin Dioxide Gas Sensors per decade for T , = 1000 K and 0.4 eV per decade for T, = 2000 K. The observed slope [ca.0.2 eV per decade, fig. 8(b)] is, therefore, not inconsistent with this interpretation. The data receive an equally consistent interpretation if the conductance is considered to be controlled by Schottky barriers between grains or domains of the mass. The development given in Appendix 2 shows that, if the oxygen-adsorption equilibrium is written m 2 n’ +- 0, 0; then [eqn (A 19)] A4, x constant + [kTm/(2.18) el lnp,, . From fig. 8(b), kTml(2.18)e = 0.09 eV, i.e. (T = 800 K), m = 2.7. This value of m does not seem to be consistent with the descriptions of the surface species that are usually mooted (uia O;, 0- or 02-, with 0,- the species expected at 800 K).However, the description is an approximate one and the experimental activation energies suffer in accuracy because the high-temperature regime, where the data are not complicated by the effects of moisture, covered only a narrow range of temperature in this work. By using different adsorption isotherms and reaction schemes [e.g. schemes such as eqn (7), Section 1.3, with a Freundlich isotherm for the species OSl5], more ‘reasonable’ values can be derived from the charge per adsorbed oxygen atom rn-l in the notation of eqn (1 l ) ] . 4.5. Dependence of Conductance on Combustible Gases, and Surface State Energy The different effect of combustible gases on the conductance activation energy provides a clear distinction between the behaviour of thin sputtered layers and of porous masses of tin dioxide.For the sputtered layers, activation energy was unaffected by the presence of a combustible gas, i.e. unaffected by the surface state occupancy. This observation implies that the surface states present in the higher temperature range have a narrow energy distribution not the exponential distribution discussed in Section 4.4. If the surface states present on the porous mass are assumed to be similar in nature to those on the sputtered layer, the change of activation energy in the presence of a combustible gas then implies that conductance in the porous mass is controlled by Schottky barriers between domains, the conductance activation energy being equal to the barrier height. The alternative interpretation, of conductance limited by narrow necks, requires that the surface-state energy depends on surface state occupancy, an effect which is manifestly not present on the sputtered layers.The various effects discussed so far, therefore, point consistently to a control of conductance in the porous bodies by inter-domain Schottky barriers. Given this interpretation, the surface-state energy, E, in fig. 1 , sets an upper limit on the barrier height, A& Thus, from the conductance activation energy at high temperature, we deduce that, for T > 700 K, E, > 1.1 eV. 4.6. Effect of Moisture on Response to CO Fig. 10 shows clearly that the effect of CO on conductance extends to lower temperature in the presence of moisture than in the absence of moisture.One interpretation is that in a scheme such as eqn (7), Section 1.3, the gas can react both with 0- and with OH- on the surface, e.g. one might write, formally: CO + 20H& -+ CO, + H,O + 2n’. (12)J . F. McAZeer et al. 1343 4.7. Consequence for Mechanisms of Ageing Effects From this work we can identify two effects significant for the ageing and reproducibility of tin oxide devices. (a) The preferred temperature range of operation, around 300 "C, is that where the effects of moisture are most marked. ( b ) The inference that the response is controlled by a relatively small number of inter-domain contacts implies that the response must be sensitive to the stability of these contacts. In addition, we remark that the response is determined by the rate constants for the surface reactions, and that these are sensitive to poisoning. 5.Conclusions The electrical conductivity of porous tin oxide pellets can be interpreted in terms of the properties of narrow regions connecting grains or domains of the structure. For the pellets which we have prepared, the conductance behaved as though it were controlled by Schottky barriers separating domains or agglomerates, each domain including a relatively large number of crystallites. It is the Schottky barrjer element that distinguishes the behaviour of porous tin dioxide sensors from the related thin film devices. The effects of temperature, moisture, oxygen partial pressure and combustible gases on the conductance and relative permittivity can be interpreted in terms of changes in charge and occupancy of electronic surface states affecting the Schottky barrier height.The presence of a surface state associated with an adsorbed oxygen species, located > ca. 1.1 eV below the conduction band edge, is deduced. Moisture is removed from the surface in the temperature range 280-400 "C. Adsorbed water introduced another electronic surface state, formulated for simplicity as OH-, which controls the behaviour at temperatures below this range. The adsorption and desorption of water results in a sigmoid conductance-temperature relationship. Carbon monoxide may react both with the oxygen surface state and with the state associated with OH-. The result is that the conductance response to carbon monoxide extends to lower temperature in the presence of moisture than in the absence of moisture.Previous reports that moisture is necessary for the observation of a CO response are correct only for temperatures below ca. 350 "C. Ageing phenomena reported for tin dioxide sensors may be attributed to changes in water adsorption and changes in the contact at the relatively few barriers which control the conductance, as well as to poisoning of the surface reactions. In any description of the operation of a practical sensor the effects of diffusion and combustion on the profiles of gas composition and conductance within the porous body, should not be overlooked. Financial support from the Electronics and Avionics Requirements Board of the Department of Trade and Industry is gratefully acknowledged. We are also grateful for many useful discussions with Dr B.C . Tofield, for comment by Drs A. Brailsford, R. Bullough and A. M. Stoneham, and for much experimental assistance by Miss G. Tappin and Mr R. J. Powell. Appendix 1 Surface Barrier Layers and Thin-film Devices : Basic Results The band structure of a thin-film tin dioxide gas sensor depleted by surface states associated with the chemisorption of oxygen is shown in fig. 1. If n,(z) denotes the conduction electron concentration at some point along the direction normal to the surface, then the conductance of the layer can be evaluated, with several simplifying assumptions, by introducing the dependence of n,(z) on the local electrostatic potential, V(z) and by using the Poisson equation to relate V(z) to the local charge density (conduction electrons and ionized donor~).~-l~1344 Tin Dioxide Gas Sensors The assumptions that must be made to reduce the problem to a tractable one are equivalent to the following statements : first, conduction electrons and surface-trapped electrons arise only from the ionization of donors and not from the valence band (a ' wide-band-gap ' assumption), and secondly, within the depletion layer the donors are completely ionized so that the charge in the surface states is approximately equal and opposite to the total charge on the ionized donors within the depletion layer.The Fermi level would then lie within the band gap in the depletion layer, sufficiently far from the conduction and valence band edges that Boltzmann statistics could be used to describe the conduction electron density. Under these assumptions, the depletion layer may be characterized by two parameters: (a) the effective thickness, dsc, determined by the charge balance between ionized donors and the surface charge (Q,): d s c N ~ e = QS (A 1) i.e.in which ND denotes the number of ionised donor states per unit volume, Ns,j the number of surface states of the type j with charge qj electronic units, per unit area, and Oj the fractional occupancy of surface states of type j . (b) The surface potential (Schottky barrier height), A#, : A#, = Q~/2moeND Here, the surface charge, Q,, is eZj Ns, qj Oj, and the formulation is appropriate to surface states with discrete energy levels. If a continuous distribution of surface states over an energy range is assumed,lO then the surface charge would be ejE N,(E) P(E) dE, in which P(E) denotes the distribution function.We note that typical values ( E E ~ M 10-l2 F cm-l, ND M 1018-1020 cmP3, A& % 1 V) imply a surface coverage, CNs 8 z 1013 cm-2 and a depletion layer thickness, dsc M 100-1 nm. The major results of previous publication^^-^^ concerning the effect of variations in 8 and N, on the channel conductance in a thin-film device can be illustrated to a sufficient degree of accuracy by dividing the channel into two zones, namely the depletion-layer [thickness given by eqn (A 2)] and the rest of the film [thickness (h - dsc), with h denoting the film thickness], and by assuming that the conductance of the depletion layer is much less than that of the remainder of the film.It may then clearly be seen that there are two classes of behaviour: (a) in the first case the layer is completely depleted (dsc 2 h). The behaviour is that of a surface-trap-limited conductance, i.e. the conductance is determined by the equilibrium between conduction band states and surface states. The activation energy for the conductance would be the surface-state energy, Es. (b) In the second case the layer is not completely depleted (dsc < h) and the behaviour is that of a bulk-trap-limited conductance in a layer of variable width (h - dsc). The activation energy for conductance here would be the bulk donor energy, ED. In the first case, (a) above, the conductance is directly affected by surface gas reactions. In the second case, (b) above, the effect of surface gas reactions on conductance is less direct.As the channel thickness, h, decreases, or as the surface charge (N,8) increases the channel conductance will move from a regime described by case (b) to one described by case (a). The surface charge (N,B) will be a function both of the temperature and of the composition of the gaseous ambient. Reaction schemes such as those given in the main text provide atomistic models for the effects.J . F. McAleer et al. 1345 Appendix 2 Oxygen Partial Pressure Dependence of Conductivity, Coverage and Depletion Layer Parameters In the following, the overall surface equilibrium is formulated as: (A 4) n’+- 0, + 0;. We treat a surface-trap-limited conductance, and a Schottky-barrier-limited conduc- tance, with a Langmuir adsorption isotherm under the assumption 8 < 1.In this case, the overall equilibrium constant for reaction (A 4) is written: rn 2 Formulations using different adsorption isotherms (e.g. Freundlich or Frumkin) differ in the dependence of K on 8. Surface-trap-limi ted In this case, the conductance of a fully depleted thin layer is calculated: There is a mass balance condition: 0 = ep (n’) h . N,h = N , 8 (single-sided channel) and the energy difference between adsorbed and desorbed states is E, (fig. 1): K = exp (E,/kT) (A 8) hence (n’) = wpor12 exp (- E,/kT) (A 9) Ns and N o = ep h2 D p ~ r ’ ~ exp (- E,/kT). NS The conductance activation energy is E,. Barrier-limi ted In this case, transport across a surface barrier is considered: 0 = oo exp (-eAd,/kT) and the objective is to calculate A#, as a function of po,.The source of electrons is the (A 12) undepleted bulk of the solid: (n’) = N , exp (- E,/kT) and the energy difference between adsorbed and desorbed states is ES-eA4,: (A 13) (A 14) K = exp [(E, - eA4,)/kT] 8 = NDpgL2 exp (- E,/kT) exp [(E, -eAq&)/kT]. so However, 8 and A& are related [eqn (A 3)]: A#, = e N: 8 2 / 2 ~ ~ 0 ND . I.e.1346 Tin Dioxide Gas Sensors or, letting @ denote the dimensionless quantity eA4,/kT, 2(Es - ED) e2 Ni ND ln@+2@ = k T + l n ( 2EE0kT)+m lnpOZ' Now, under the conditions considered (Ads = 0.2-1 eV, T = 800 K), In CD is signifi- cantly less than 2@. Empirically one observes that the relationship ln@+2@ = 2.18@+0.56 (A 18) holds over the range considered. Substitution in eqn (A 13) then gives the relationship: (A 19) One may also verify, from eqn (A 15), that with typical values (N, z 1015 cm-2, ND x 10ls ~ m - ~ , EE, z F cm-l), that Aq& increases very sharply with increases of 8. The effect is that 8 does not change much with change ofpo2, although A4, does change. Indeed, if the approximation (valid for large A#s), In (4) < 24, is made, then eqn (A 17) A4, = constant + [kTm/(2.18) el lnpOz. gives which, on substitution into eqn (A 14) gives 8 = ND ( 2 ~ 8 ~ k T / e 2 x ND)i and, from eqn (A 2) d,, = ( ~ E E , k T/e2 N,)i. (A 22) The limiting value of 8 implied by eqn (A 21) is, with the typical values for N,, N , and EE, given above, 8 z lop3, whilst d,, = 10 nm. References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 W. H. Brattain and J. Bardeen, Bell Systems Tech. J . , 1953, 32, 1. G. Heiland, 2. Phys., 1954, 138, 459. T. Seiyama, A. Kato, K. Fukiishi and M. Nagatini, Anal. Chem., 1962, 34, 1502; N. Taguchi, U.S. Patent 3 631 436 (1970); U.K. Patent 1 280 809 (1970); Chemical Sensors Meeting, Fukuoka, 1983. W. Hagen, R. E. Lambrich and J. Lagois, Adv. Solid State Phys., 1983, 23, 259. D. E. Williams and P. McGeehin, in Electrochemistry, ed. D. Pletcher (Specialist Periodical Reports, Royal Society of Chemistry, London, 1984), vol. 9, p. 246. P. T. Moseley, J. F. McAleer, J. 0. W. Norris, B. C. Tofield, G. Tappin and D. E. Williams, to be published. J. F. McAleer, P. T. Moseley and D. E. Williams, to be published. J. F. McAleer, P. T. Moseley, B. C. Tofield and D. E. Williams, Proc. Brit. Ceram. SOC., 1985, 36, 89. P. Mark, Surf. Sci., 1971, 28, 192. S. Baidyaroy and P. Mark, Surf: Sci., 1972, 30, 53. T. A. Goodwin and P. Mark, Prog. Surf. Sci., 1971, 1, 1. H. Windischmann and P. Mark, J . Electrochem. SOC., 1979, 126, 627. S. Strassler and A. Reis, Sensors Actuators, 1983, 4, 465. A. Jones, T,A. Jones, B. Mann and J. G. Firth, Sensors Actuators, 1984, 5, 75. P. K. Clifford, Proc. Ist Int. Conf: Chemical Sensors, Fukuoka, Japan (Kodansha, Elsevier, 1983), p. 135. P. K. Clitford and U. 'I.. Tuma, Sensors Actuators, 1983, 3, 233, 255. N. M. Beekmans, J . Chem. SOC., Faraday Trans. I , 1978, 74, 31. S-C. Chang, J . Vac. Sci. Technol., 1980, 17, 366. A. Brailsford, personal communication. J. Van den Broek, Philips Res. Rep., 1967, 22, 367. N. Yamazoe, Y. Kurakawa and T. Seiyama, Sensors Actuators, 1983,4, 283. G. Heiland, Sensors Actuators, 1982, 2, 343. S. Roy Morrison, Sensors Actuators, 1982, 2, 32P. N. Yamazoe, J. Fuchigami, M. Kishikawa and T. Seiyama, Surf. Sci., 1979, 86, 335. E. W. Thornton and P. G. Harrison, J. Chem. SOC., Faraday Trans. I , 1975,71, 461. Paper 611200; Received 13th June, 1986

ISSN:0300-9599    

DOI:10.1039/F19878301323

出版商:RSC    

年代:1987

数据来源: RSC