摘要:

J. Chem. SOC., Faraday Trans. 1, 1986, 82, 2481-2495 Adsorption of Organic Substances at the Mercury/Ethylene Glycol Interface Part 2.-Aromatic Compounds Jondo I. Japaridze, Shukri S. Japaridze and Nely A. Abuladze Institute of Inorganic Chemistry and Electrochemistry, Georgian Academy of Sciences, ul. Djikiya 7, 380086 Tbilisi, U.S.S.R. Achille De Battisti Chemical Institute, University of Ferrara, Via Borsari 46, 44100 Ferrara, Italy Sergio Trasatti" Department of Physical Chemistry and Electrochemistry, University of Milan, Via Venezian 21, 20133 Milan, Italy The differential capacitance of a polarized mercury electrode in ethylene glycol solutions containing a series of aromatic compounds has been measured using a stationary drop. Adsorption has been investigated for benzene, toluene, phenol, aniline, dimethylaniline, dimethyltoluidine, di- phenylamine, tribenzylamine and tribenzylammonium sulphate.The results are compared with those for adsorption of the same substances from aqueous solutions. The role of the non-aqueous solvent in determining the mode of adsorption is discussed. ~p~ ~ ~~~ ~~ In Part 1 of this work,l we reported the adsorption behaviour of aliphatic compounds on a mercury electrode from solutions in ethylene glycol (EG). In this paper the study has been extended to aromatic compounds. n-Electrons are known to play an important role in the adsorption of aromatic substances at an electrode, especially when the electrode surface is positively chargedS2 This is the reason why aromatic compounds adsorbed on Hg are not desorbed as the electrode is brought to positive rational potential^.^ Studies of adsorption of aromatic compounds on Hg from non-aqueous solvents are sparse. Most of them deal with solution in dimethylf~rrnamide,~~~ the others with n-propyl alcohol,'* * acetonitrileg and methanol.1° No study has been carried out in ethylene glycol to date.In this work the features of the adsorption of aromatic substances have been investigated by modifying the substituents of the aromatic nucleus in different series of compounds : benzene, toluene, phenol, aniline, dime t h y laniline, dime t hyl toluidine, diphenylamine, tribenzylamine and tribenzylammonium sulphate. The aim of the work has been to elucidate if and to what extent the non-aqueous solvent changes the experimental picture which has emerged from studies in aqueous solutions.2g 3* l1 Experimental Adsorption was studied using a stationary electrode.The experimental capacitance was measured as a function of potential by means of an a.c. bridge working at 400 Hz.12 The temperature of the solution was kept at 20 "C. Potentials were measured and are reported against a 0.2 mol dmP3 calomel electrode 248 12482 Adsorption of the Mercury/ Ethylene Glycol Interface in EG. The potential of this electrode differs by 6 mV from that of an aqueous saturated calomel electrode. All chemicals were purified according to prescriptions given in the literature. Results and Discussion Fig. 1 shows the capacitance-potential curves for adsorption of benzene on Hg from 0.5 mol dm-3 NaClO, solutions in EG.As the benzene concentration increases the capacitance is progressively depressed around - 0.45 V, which testifies to the increasing adsorption of the aromatic compound. Capacitance maxima can be observed both in the positive and in the negative branches of the curves. The negative maximum is better defined and the potential reaches the limiting value of ca. - 0.8 V at 8 = 1. The positive maximum is less marked. At potentials more negative than - 1.4 V, benzene is totally desorbed from the electrode surface. Fig. 2 refers to toluene adsorption. The picture is similar to the previous one. At the potential of maximum adsorption (Em), the capacitance of the surface of Hg saturated with benzene is C, = 4.1 pF cm-2, to be compared with 5.0 pF cmP2 in aqueous solutions (supporting electrolyte: 1.0 mol dm-3 KCl).13 In the case of toluene, C, has been found to be 3.5 pF cm-2.The potential Em is ca. 0.1 V more negative than the potential of zero charge in the base solution, E,,, = -0.35 V.I4 Capacitance curves were integrated twice from a potential more negative than - 1.3 V (cf. fig. l), where all the curves for the organic substance merge with that of the base solution. As fig. 3 shows, toluene is less adsorbed than benzene at positive rational potentials owing to the weakening in the electron interaction. This agrees with the observations made in aqueous s ~ l u t i o n s . ~ ~ ~ l6 The adsorption isotherm at the Em = -0.45 V for benzene and toluene has been derived by means of the equation:l1 8 = (CO - C)/(Co - Cl) (1) using for Co the value of 18.7 pF cm-2 l4 and for C, the values given above.Fig. 4 shows the resulting adsorption isotherms. Toluene is superficially more active than benzene at Em probably because of the lowering of the solubility, since n-electrons are not operative at this potential. Fig. 5 shows a test of the Frumkin isotherm17 in terms of the equation: The slope of the straight line gives the value of the interaction parameter a, while the intercept gives the standard free energy of adsorption at zero coverage: logp = -AG1,/2.3 RT. (3) In the case of benzene a z 1.8 and AGid = -4.4 kJ mol-l. For toluene, a = 1.6 and AGZd = -6.1 kJ mol-l. It is interesting that although toluene is more adsorbed the surface condensation is weaker. This is not observed for adsorption of the same substances on Hg from water,6$ l6 but it has been reported for a Bi electrode.ls Therefore, the electrode-solvent interaction energy is probably important in determining the adsorption parameters.19 The surface pressure at -0.45 V was plotted us.logc. Although the differentiation was somewhat difficult owing to the scarcity of points, the limiting slope, according to the Gibbs equation, gave r = 5.4 x mol cmP2 in the particular case of benzene. With the value of 8 obtained by means of eqn (l), the saturation surface excess Tm has been found to be 5.9 x lo-', mol cm-2. The same procedure gave 5.6 x mol cm-2 for toluene. In aqueous solutions the values of r, are 4.4 x 1 0-lo and 5.0 x 1 0-lo mol cmP2, respectively.l5? l6J .I . Japaridze et al. 2483 40 30 N I E L 20 10 7 I I I I I I 1 I 0 0.4 0.8 1.2 -E/V Fig. 1. Capacitance-potential curves for adsorption of benzene on Hg from 0.5 mol dmP3 NaC10, solutions in ethylene glycol. Concentration of benzene: (1) 0, (2) 0.27, (3) 0.53, (4) 0.63, ( 5 ) 0.73, (6) 0.88, (7) 1.0 mol dmP3. 40 32 N I ' 24 G L Y. 16 a I I I I I I 1 I 0 0.4 0.8 1.2 -EIV Fig. 2. Capacitance-potential curves for adsorption of toluene on Hg from 0.5 mol dm-3 NaC10, solutions in ethylene glycol. Concentration of toluene: (1) 0, (2) 0.07, (3) 0.14, (4) 0.23, ( 5 ) 0.31, (6) 0.4, (7) 0.49, (8) 0.6, (9) 0.7 mol dm-3.2484 Adsorption of the MercurylEthylene Glycol Interface 0.3 0.9 1.5 -EIV Fig. 3. Calculated electrocapillary curves for adsorption of benzene (2) and toluene (3) on Hg from solutions in ethylene glycol. (1) Blank solution 0.5 mol dm-3 NaClO,, ( 2 ) 1 mol dm-3 benzene, (3) 0.79 mol dmP3 toluene.c/mol d ~ n - ~ Fig. 4. Adsorption isotherms of benzene (A) and toluene (A) on Hg from solutions in ethylene glycol at E = -0.45 V. A check on the reliability of the isotherm parameters was carried out by fitting the experimental data of surface pressure and surface excess to the state equation corresponding to the Frumkin isotherm :20 @ = -RTT,[ln(l -0)+a02]. (4) In the particular case of benzene, the values obtained by a non-linear regression analysis have been 1.7 for a, in excellent agreement with the value derived from fig. 5, and 6.5 x 10-l0 mol cmP2 for Tm, which differs by ca. 10% from the value reported above.In view of the few concentrations explored, the outcome is good. The value of the saturation coverage, Tm, has been found to be higher at the Hg/EG interface than at the Hg/water interface6$ l6 for both benzene and toluene. This suggestsJ . I. Japaridze et al. 248 5 0.5 8 1 Fig. 5. Test of the Frumkin isotherm for adsorption of benzene (A) and toluene (A) on Hg from solutions in ethylene glycol at E = -0.45 V (from fig. 4). that the orientation of the molecules is somewhat more vertical in the non-aqueous solution. A higher value of Tm has also been observed in the adsorption of these aromatic compounds on a Bi electrode from aqueous The reason for this may again be the complex adsorbate-solvent electrode interactions both in the adsorption layer and towards the liquid phase.lS The marked decrease in the free energy of adsorption, which is observable whenever non-aqueous solvents are used, is to be attributed to the higher adhesion of the solventz1 only for a minor part, the major contribution being presumably due to the higher solubility of the organic compounds in the non-aqueous so1vent.22T 23 While the above data show that the addition of a methyl group to the benzene ring does not modify the qualitative features of the capacitance curves appreciably, fig.6 and 7 emphasize the remarkable effect of the introduction of the OH and NH, groups. In fig. 6 it can be seen that the maximum capacitance at negative rational potentials does not show up in the case of aniline up to a concentration of 0.17 mol dmP3. On the contrary, the maximum at positive potentials is pronounced.Its height increases, while its potential shifts towards more negative values as the concentration of aniline is increased. The desorption of aniline is already complete at ca. - 1.0 V. Fig. 7 shows that in the case of phenol adsorption, capacitance plateaux rather than maxima are observed at positive rational potentials. The capacitance at the plateaux is seen to decrease with increasing phenol concentration. A small hump develops at negative rational potentials as the phenol concentration exceeds ca. 0.3 mol dmP3. The desorption of phenol is complete at E = - 1.5 V. Fig. 8 clearly shows that phenol is not desorbed at positive potentials (even at E = 0.2 V), as the electrocapillary curves obtained by double integration of the capacitance data suggest.Comparison with data for aqueous solutions shows some dramatic differences. Whereas the capacitance drops to C, = 7.9 pF cm-2 in water around the potential of zero charge,24 it never decreases below ca. 13 pF cm-, in EG in the same potential region. This suggests that the orientation of phenol at the Hg/EG interface is probably always substantially flat even at charges removed from the positive region. Fig. 9 shows the capacitance curves for the adsorption of N,N-dimethyl-o-toluidine (DMOT). The behaviour qualitatively resembles that of the other aromatic compounds,2486 Adsorption of the MercurylEthylene Glycol Interface 50 t 40 20 0 0.4 0.8 1.2 -E/V Fig. 6. Capacitance-potential curves for adsorption of aniline on Hg from 0.5 mol dmV3 NaC10, solutions in ethylene glycol.Concentration of aniline ( x lo2): (1) 0, (2) 0.24, (3) 0.74, (4) 1.1, (5) 2.1, (6) 3.5, (7) 5.8, (8) 10, (9) 14, (10) 17 mol dmP3. particularly benzene and The value of C , is ca. 3 pF cm-* and is reached at a concentration of 0.15 mol ~ l m - ~ . The adsorption-desorption peak at negative rational potentials (curves 4-8) is shifted to more negative electrode potentials and its height increases as the concentration is raised. In the positive branch of the curves, capacitance plateaux are observed in the potential range -0.2 to 0.1 V. The presence of plateaux at anodic rational potentials is characteristic of aromatic compounds and testifies to the reorientation of the adsorbate from a tilted orientation on the negatively charged surface to the flat position at positive charges.In the tilted orientation, the adsorbate causes the expansion of the electrical double layer with depression of the capacitance. In the flat orientation, the depression of the capacitance with respect to the base solution is diminished remarkably. In the case of DMOT, at strongly positive potentials (E > 0.1 V) peaks appear which are not visible in fig. 9. These peaks cannot be attributed to the desorption of the adsorbate unambiguously since their height does not increase and the peak potential does not shift monotonically towards more anodic potentials as the concentration in solution is raised. The bulk concentration needed to reach the maximum surface excess is significantly higher for benzene and toluene than for DMOT (1 .O, 0.49 and 0.15 mol dm-3, respect- ively).In the case of toluene and benzene the negative desorption peak is more rounded and its maximum height at the above concentration of adsorbing substance is ca. 20pFcmP2, to be compared with ca. 3 0 ~ F G c m - ~ for DMOT under similar circumstances.J. I. Japaridze et al. 2487 50 40 N E % 30 G 20 I I I I 0 0.4 0.8 1.2 -EJV Fig. 7. Capacitance-potential curves for Hg in 0.5 mol dmP3 NaClO, solutions in ethylene glycol containing phenol: (1) 0, (2) 0.003, (3) 0.04, (4) 0.08, ( 5 ) 0.1 1, (6) 0.13, (7) 0.23, (8) 0.55, (9) 0.72, (10) 1 mol dm-3. 0.3 0.9 1.5 -EJV Fig. 8. Calculated electrocapillary curves for adsorption of phenol on Hg (cf. fig. 7). (1) 0, (2) 0.13, (3) 0.55, (4) 1.0, (5) 1.18 mol dm-3.2488 Adsorption of the Mercury/ Ethylene Glycol Interface 50 40 30 ri 6 2 a 20 10 I I I I 0 0.4 0.8 1.2 -EIV Fig.9. Capacitance-potential curves of Hg in 0.5 mol dmP3 NaC10, solutions in ethylene glycol containing dimethyl-o-roluidine: (1) 0, (2) 0.01, (3) 0.025, (4) 0.05, (5) 0.075, (6) 0.1, (7) 0.125, (8) 0.15 rnol dmP3. Fig. 10 shows the capacitance-potential curves in the case of dimethylaniline (DMAN) adsorption. As in the case of other substances, DMAN is adsorbed around the potential of zero charge in the potential range -0.4 to -0.6 V. The adsorbate is completely desorbed at ca. - 1.3 V. Comparison of DMOT (fig. 9) and DMAN (fig. 10) reveals that the latter compound is less adsorbable. The adsorption-desorption peaks for DMAN are rather broad and not particularly developed (fig.10, curves 2-6). DMAN starts to affect the shape of the capacitance curves only at concentrations around 0.3 mol dm-3, whereas DMOT already saturates the surface at 0.15 mol dm-3. Peaks develop at positive potentials (0-0.1 V, not reported in fig. 10) in the case of DMAN. They shift towards more negative rather than to more positive potentials as the concentration is increased. A plateau is visible in the range -0.1 to -0.2 V. The positive peaks are lower than the curve for the base solution. The data in fig. 10 suggest that DMAN is adsorbed in a tilted orientation in the potential range -0.4 to -0.6 V, while it takes a flat position at more positive potentials in the range -0.25 to 0 V. Complete desorption takes place in the potential range 0-0.1 V.A second peak appears at strongly positive potentials (not shown in fig. 10) whose interpretation requires further experimental study . Fig. 1 1 shows the capacitance curves of 0.05 mol dmP3 NaClO, solutions in EG with different amounts of diphenylamine (DPA). This substance adsorbs around E = - 0.5 V with depression of the capacity. The potential of maximum adsorption Em, and the peakJ . I . Japaridze et al. 2489 Fig. 35 F) I $ 25 a 14 z. 15 1 t I I I I I I I I 0.2 0.6 1 .o 1.4 1.8 -EIV containing dimethylaniline: (1) 0, (2) 0.3, (3) 0.4, (4) 0.5, ( 5 ) 0.6, (6) 0.7 mol dm-3. 10. Capacitance-potential curves of Hg in 0.5 mol dmP3 NaClO, solutions in ethylene glycol 30 N 'E 20 0 LL =L G 10 0.4 0.8 1.2 1.6 -EIV Fig.11. Capacitance-potential curves of Hg in 0.05 mol dm-3 NaC10, solutions in ethylene glycol containing diphenylamine: (1) 0, (2) (3) 5 x lop4, (4) 5 x lop3, ( 5 ) 0.01, (6) 0.05, (7) 0.1, (8) 0.2 mol dmp3. potential both shift to more negative values as the concentration is increased. The shift of Em is probably due to a diffuse layer effect with &,, the potential at the OHP, becoming more negative as the concentration of DPA is increased. The capacitance decreases also at positive potentials and this can be attributed to the increased adsorption of the substance due to the strong interaction of the z-electrons of the aromatic rings with the positively charged surface, as observed with other aromatic compounds.2490 Adsorption of the Mercury1 Ethylene Glycol Interface 30 N ' 20 G E L L 3- 10 5 6 I I I 0 0.5 1 .o -EIV Fig.12. Capacitance-potential curves of Hg in 0.05 mol dmP3 HC10, solutions in ethylene glycol containing diphenylamine: (1) 0, (2) (3) 5 x (4) lop3, ( 5 ) 0.01, (6) 0.1, (7) 0.2 mol dm-3. Fig. 12 illustrates the effect of pH on the adsorption of DPA. If HC10, is substituted for NaClO, fig. 12 shows that the negative branches differ dramatically, while similar behaviour is obtained in the positive branches. In particular, the adsorption-desorption peaks are absent in fig. 12. This is a consequence of the prevalence in solution of the diphenylammonium cation in equilibrium with the neutral form. Hints of adsorption- desorption peaks appear as the concentration is sufficiently high (curves 5-7, fig.12). However, the proton discharge overlaps with these peaks, which are thus not well developed. The surface activity of DPA in EG is higher than in aqueous solutions.13 The drop in capacitance around Em (-0.7 V in water, -0.5 V in EG) is higher in aqueous solutions (ca. 15 pF cm+) than in the non-aqueous solvent (ca. 6 pF cm-2). A concentration higher in EG than in water by three orders of magnitude is necessary to achieve the maximum drop in capacity. Fig. 13 shows the capacitance-potential curves for adsorption of tribenzylamine (TBA) from 0.25 mol dm-3 NaClO, solutions in EG. Similarly to the situation in aqueous 27 TBA adsorbs preferentially around the potential of zero charge. The maximum decrease in capacity is observed at - 0.4 V. Surface saturation is achieved at the concentration of ca.10+mol drnp3. The corresponding value of C, is ca. 8 pF cm-2. No adsorption-desorption peaks are observed at negative rational potential. Desorption is practically complete at more negative potentials than -0.8 V. At positive potentials, capacitance plateaux can be identified which testify to the adsorption by z-electron interaction. Fig. 13 also shows that the adsorption of TBA is restricted to a potential range of 0.45-0.50 V around the potential of zero charge. Fig. 14-16 illustrate the effect of pH and of the nature of the anion of the supporting electrolyte on the adsorption behaviour of TBA. Marked adsorption of the protonated form, TBA+, is always observed at negative rational potentials. In addition, a remarkable retardation of proton discharge is obtained.J.I. Japaridze et al. 249 1 40 30 N 5 LL 3. ;s 20 10 L5 I 1 I I 0 0.4 0.8 1.2 -EIV Fig. 13. Capacitance-potential curves of Hg in 0.25 mol dmP3 NaClO, solutions in ethylene glycol containing tribenzylamine: (1) 0, (2) 5 x lop4, (3) 7.5 x lo-,, (4) 8.8 x loP4, ( 5 ) mol ~ I r n - ~ . I I I I I 0.4 0.8 1.2 1.6 -EIV Fig. 14. Capacitance-potential curves of Hg in 0.1 mol dmP3 HC1 solutions in ethylene glycol containing tribenzylamine: (1) 0, (2) 0.003, (3) 0.005, (4) 0.01, ( 5 ) 0.05, ( 6 ) 0.075 mol dm-3.2492 Adsorption of the MercurylEthylene Glycol Interface 5,6 - 23 I 1 I I 0 0.4 0.8 1.2 -EIV Fig. 15. Capacitance-potential curves of Hg in 0.1 mol dm-3 HClO, solutions in ethylene glycol containing tribenzylamine: (1) 0, (2) 7.5 x (7) 5 x (8) 7.5 x (9) mol dmP3.(3) lop3, (4) 2 x ( 5 ) 3 x lop3, (6) 4 x Fig. 14 shows that TBA+ adsorbs at higher concentrations than lop3 mol dmP3. Sharp drops in capacitance are observed, corresponding to TBA+ adsorption. The potential where adsorption commences shifts to more positive values while the potential range where adsorption is observed widens as the concentration is increased. The potential at the sharp drop in capacitance can be accurately determined within a few mV. Before the capacitance drop, a feature resembling a capacitance peak is observed in the capacitance curve. The dramatic change in the shape of the capacitance-potential curve is apparently due to the variation in the adsorption features of TBA+. Similarly to other aromatic compounds, TBA+ adsorbs over all the potential range in two different orientations. The capacitance drop marks the transition from one to the other position which takes place in a range of a few mV.The orientation is flat up to the potential of the discontinuity, and it changes to vertical past the critical potential. The latter position is responsible for the low value of the capacitance, ca. 4.5 pF cm-2 in a wide potential region between -0.95 V and - 1.4 V (fig. 14, curve 2). Fig. 14-16 show that the adsorption of TBA in acid solutions retards the proton discharge reaction. The cationic form is in fact remarkably active on a negatively charged Hg surface. TBA+ is adsorbed in a wider potential range, ca. -0.55 to - 1.6 V, as the concentration is increased up to 0.05 mol dm-3.In the saturated solution (0.075 mol dm-3), no hydrogen evolution is observed even down to - 2.0 V. Therefore, TBA+ exhibits remarkable inhibiting properties which are responsible for the shift of up to 1 V of the proton discharge potential with respect to the base EG solution. Fig. 15 shows the features of the adsorption of TBA+ in the presence of HC10, asJ . I. Japaridze et al. 2493 0 0.4 0.8 1.2 -EIV Fig. 16. Capacitance-potential curves of Hg in 0.1 mol dm-3 H,SO, solutions in ethylene glycol containing tribenzylamine: (1) 0, (2) 5 x ( 5 ) 7.5 x lop3, (6) lo-,, (7) 2.5 x (8) 7.5 x (9) 8 x mol dmp3. (3) 7.5 x (4) 2.5 x the base electrolyte. Qualitatively, the picture is very similar to fig. 14. However, C1- is known to be specifically adsorbed on Hg from EG, whereas C10; is While the value of C, is the same, 4.5 pF cm-2, the potential region where it is observed is narrower.Therefore, C1- induces the adsorption of TBA+ because the specifically adsorbed anions create a favourable electric field. Fig. 16 shows the capacitance-potential curves for adsorption of TBA in the presence of 0.1 mol dm-3 H2S0,. Interesting aspects are in this case the effect of a superficially inactive multicharged anion and the effect of the nature of the acid on the formation of the protonated form. The picture in fig. 16 is substantially similar to the other two cases. It can however be observed that H,S04 appears to give the widest potential range where TBA+ is adsorbed in a vertical position. Conclusions The adsorption of aromatic compounds on a polarized Hg electrode from ethylene glycol shows a general pattern similar to the adsorption from aqueous solutions with some definite different details.Adsorption at positive charges is governed by the interaction of the n-electrons of the aromatic ring with the electrode surface. At negative charges adsorption issues from the organic compound being expelled from the solution bulk. These two different energetic2494 Adsorplion of the Mercury1 Ethylene Glycol Interface situations result in two different molecular orientations : flat at positive charges and tilted or vertical at negative charges. While the same picture is observed in the adsorption from aqueous solutions, important differences are also noticed.In some cases Tm, the saturation coverage, differs from that in aqueous solution, with a consequent difference in the value of C,. This indicates that the packing of the monolayer at the electrode surface may depend on the nature of the solvent, probably being the resultant of complex adsorbate-solvent interactions in the bulk and at the interface. In one case (phenol) these interactions are probably responsible for the non-occurrence of the adsorption in the vertical orientation. Lateral interactions are also different in non-aqueous solvents from those in water. While short-range adsorbateadsorbate interactions in the monolayer are not expected to vary at constant orientation in different solvents, the apparent lateral interaction parameter does vary because of variations in the adsorbate-solvent and soivent-solvent intera~ti0ns.l~ Therefore, a systematic study of the apparent interaction parameter should give information on the energetic state of the adsorbate at the interface in different solvents.The surface activity of aromatic compounds in EC is as a rule appreciably lower than in aqueous solutions. This is mainly due to a higher solubility of organic compounds in non-aqueous solvents. To a lesser extent the higher adhesion of the organic solvent to the electrode surface may also play a role. This is a delicate point. The higher adhesion of EG is inferred from its adsorbability from aqueous solutions. However, this does not necessarily mean that AG of the reaction EG(bu1k) -+ EG(surface) is the same in EG as in water.At any rate, no definite conclusions can be reached on this point unless the AG;,, could be corrected for the AGO of solution of the same compounds in EG and in water. This is the only possibility to separate surface from bulk contributions. References 1 J. I. Japaridze, N. A. Abuladze, Sh. S. Japaridze, A. De Battisti and S. Trasatti, Electrochim. Acta, 2 M. A. Gerovich, Dokl. Akad. Nauk SSSR, 1954,%, 542; 1955, 105, 1278. 3 A. N. Frurnkin and B. B. Damaskin, Vest. MGU, Khim., 1967, 5, 27. 4 M. K. Kaisheva, R. I. Kaganovich and B. B. Damaskin, Elektrokhimiya, 1973, 9, 94. 5 M. K. Kaisheva, B. B. Damaskin and R. I. Kaganovich, Elektrokhimiya, 1975,11,431; 1974,10, 1725; 6 R. I. Kaganovich, B. B. Damaskin and M. K. Kaisheva, Elektrokhimiya, 1970, 6, 1359. 7 E.N. Protskaya, V. M. Gerovich, B. B. Damaskin, V. Ya. Rosolovskii and D. G. Lemesheva, Elektro- 8 E. N. Protskaya, B. B. Damaskin, V. A. Safonov, V. M. Gerovich, V. Ya. Rosolovskii and D. G. 9 H. Wegert and H. Baumgartel, Ber. Bunsenges. Phys. Chem., 1980,84, 274. in press. 1972, 8, 1642. khimiya, 1980, 16, 526. Lemesheva, Elektrokhimiya, 198 1, 17, 704. 10 P. Nikitas, A. Anastopoulos and D. Jannakoudakis, J. Electroanal. Chem., 1983, 145, 407. 11 A. N. Frumkin and B. B. Damaskin, Modern Aspects of Electrochemistry, ed. J. O’M. Bockris and 12 J. I. Japaridze, G. A. Tedoradze, and Sh. S. Japaridze, Elektrokhimiya, 1969, 5, 955. 13 J. Dojlido and B. Behr, Rocz. Chem., 1963, 37, 1043. 14 J. I. Japaridze and Sh. S. Japaridze, Adsorbzya i Dvoinoi Sloi v Elektrokhimii (Nauka, Moscow, 1972), 15 R. I. Kaganovich, I. M. Gerovich and 0. Yu. Gusakova, Elektrokhimiya, 1967, 3, 946. 16 K. G. Baikerikar and R. S. Hansen, J . Colloid Interface Sci., 1977, 61, 239. 17 A. N. Frumkin, 2. Phys. Chem., 1925, 116, 466. 18 A. R. Alumaa and U. V. Palm, Elektrokhimiya, 1972, 8, 471. 19 S. Trasatti, J . Electroanal. Chem., 1981, 123, 121. 20 A. Frumkin, 2. Phys., 1926, 35, 792. 2 1 R. Payne, Advances in Electrochemistry and Electrochemical Engineering, ed. P. Delahay and 22 A. R. Alumaa and U. V. Palm, Elektrokhimiya, 1978, 14, 1369. B. E. Conway (Butterworths, London, 1964), vol. 3, p. 149. p. 68. C.W. Tobias (Wiley-Interscience, New York, 1970), vol. 7, p. 1.J . I . Japaridze et al. 2495 23 M. D. Levi, A. V. Shlepakov, B. B. Damaskin and I. A. Bagotskaya, J . Electroanal. Chem., 1982, 24 B. B. Damaskin, V. M. Gerovich, I. P. Gladkikh and R. I. Kaganovich, Zh. Fiz. Khim., 1964,38,2495. 25 A. E. Kakhadze, N. A. Abuladze, V. A. Shikhashvili and Sh. S. Japaridze, Zzv. Akad. Nauk GSSR, 26 M. A. Loshkarev and A. A. Kriukova, Zh. Fiz. Khim., 1957,31,458. 27 A. A. Kriukova and M. A. Loshkarev, Dokl. Akad. Nauk SSSR, 1951, 81, 1097. 132, 1 . Cer. Khim., 1976, 2, 151. Paper 511697; Received 30th September, 1985

ISSN:0300-9599    

DOI:10.1039/F19868202481

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I , 1986, 82, 2497-2503 Chemical Relaxation in Mixed Micellar Solutions containing Surface-active Drugs and Hexadecyltrimethylammonium Bromide Micelles John Gormally* Department of Chemistry, University of Salford, Salford M5 4 WT Santosh Sharma Department of Pharmacy, University of Nottingham, Nottingham NG7 2RD A study of chemical relaxation in mixed micellar solutions containing surface-active drugs and hexadecyltrimethylammonium bromide micelles is reported. The incorporation of drug molecules into the micelles takes place at a rate which appears to be diffusion controlled and the rate coefficient for the release of drug increases with the proportion of drug within the micelle. The ionic nature of the drugs appears to have little effect on the kinetic behaviour.A number of cationic drugs are surface active and form small micellar aggregates in aqueous so1ution.l The kinetics of this aggregation process has been studied by the technique of ultrasonic relaxation2 and it appears that micellisation in the drug solutions is very similar to that found in solutions of long-chain surfactant molecules. Solutions containing a cationic drug and large cationic surfactant micelles may therefore be expected to contain mixed micellar aggregates. Here we report on a study of the kinetics of the interaction between two cationic drugs and hexadecyltrimethylammonium bromide (HDTABr) micelles. The interpretation of the results is very similar to that given recently for the solubilisation of pentanol by cationic micelles." Experimental Materials Propranolol hydrochloride (PH) was obtained from ICI and penthianate methobromide (PMB) from Winthrop Laboratories.Hexadecyltrimethylammonium bromide was obtained from BDH. These substances were used as supplied. Ultrasonic Measurements Sound velocities in solutions were measured using the resonance technique4 at a frequency of 2 MHz. The values obtained are relative to the literature value for water at 25 "C (1496.58 f0.04 m s-').~ Measurements of sound velocity could be made to an accuracy of kO.1 m s- l. Ultrasonic absorption measurements were made with the resonance technique for frequencies in the range 1-20 MHz and with a pulse apparatus 24972498 Relaxation in HDTABr Micelles (Matec) from 15 to 95 MHz. For all measurements the solutions were thermostatted at 25 f 0.05 "C.Equilibrium Composition of Solutions The concentration of HDTABr was 0.1 mol dm-3 in all solutions which contained this component. As the critical micelle concentration of HDTABr is below low3 mol dm-:, we could safely assume that this surfactant was almost totally in the micellar form. In order to understand the behaviour of a cationic drug in these solutions it is necessary to determine how the drug is partitioned between the aqueous and micellar phases. The degree of partitioning was estimated from sound-velocity measurements described below. In fig. 1 we give plots of sound velocity us. concentration for solutions containing HDTABr alone, a drug alone and the drug in the presence of 0.1 mol dm-3 HDTABr.It can be seen that the sound velocity varies little with concentration for HDTABr 1520 - 'm 151 6 E d 1512 --. Y x .- $ 1 5 0 4 1500 0.05 0.1 0.15 concentratiodmol dm-3 Fig. 1. Plots of sound velocity us. concentration for: 0, penthianate methobromide; 0, penthianate methobromide in 0.1 mol dmP3 HDTABr; 0, HDTABr. A departure from linearity can be seen for the highest concentration of PMB in HDTABr. Although small, this effect is outside the limits of experimental error and indicates that the partition constant, K,,, is not strictly constant over a large concentration range. A similar pair of lines was obtained for the PH-HDTABr system. solutions. As these changes in sound velocity are largely due to changes in the solution compressibility,6 this indicates that the compressibility of HDTABr micelles approximates to that of HDTABr solution at the c.m.c.This behaviour is peculiar to HDTABr and was one of the reasons for selecting this surfactant. Relatively large velocity changes were found in solutions of the drug alone. In the concentration range studied the drug was entirely in the non-aggregated form and the observed increase in sound velocity with drug concentration can be attributed to a reduction in solution compressibility due to hydration of the ionic drug. This behaviour is typical of ionic species and it has been used as a means of estimating hydration number~.~-~ The presence of HDTABr in theJ . Gorrnally and S . Sharma 2499 Do 4 concentration Fig. 2. (a) The variation of sound velocity in solutions containing the drug alone.This line intercepts the ordinate at C,, the sound velocity in water. (b) The variation of sound velocity in solutions containing drug in the presence of 0.1 mol dmP3 HDTABr. This line intercepts the ordinate at C,, the sound velocity in 0.1 mol dm-3 HDTABr. (c) Is (b) displaced downwards by the amount C,-C,. (c) Is taken to represent the effect upon the sound velocity of drug which is in the aqueous phase alone, the effect of micellar drug having been ahwed for by the displacement mentioned above. When the total drug concentration is D,, the concentration in the aqueous phase is D,. drug solutions can be seen to reduce the increases in sound velocity which result from increases in drug concentration. This can only be due to the incorporation of some drug within the HDTABr micelles to give rise to a mixed micelle which has a compressibility similar to that of the HDTABr micelle alone.If we assume that these mixed micelles have a compressibility which is the same as that of HDTABr micelles, then it is a simple matter to estimate how much of the drug is within the micellar phase and how much is in the aqueous phase by the procedure indicated in fig. 2. It is unlikely that the incorporation of drug molecules within HDTABr micelles will have no effect upon their compressibility and we must consider this method to be a means of estimating the partitioning of the drug, rather than a means of accurate determination. However, the interpretation of the kinetic data and the conclusions which arise from this interpretation are dependent upon the accuracy of the partitioning data.As these conclusions appear to be generally consistent and quantitatively reasonable, we feel that the partitioning data must represent an adequate approximation to the truth. Results and Discussion Equilibrium Measurements It is clear from the linear nature of the plots shown in fig. 2 that there is a linear relationship between the concentration of drug in the aqueous phase, D,, and the concentration of drug in the micellar phase, Dm. We can therefore define a partition constant, Kp, by:2500 Relaxation in HDTABr Micelles For the PH/HDTABr system, Kp has a value of 2.3 and the corresponding value for the PMB-HDTABr system is 4.1 when the HDTABr concentration is 0.1 mol drnp3. Chemical equilibrium between the aqueous and micellar phases requires that Pa = P m (2) where pa and ,urn are the chemical potentials of the drug in the aqueous and micellar phases, respectively.The constancy of Kp with drug concentration is consistent with and p,=,@+RTlnD, pm = pg+RTln D, (3) where p p and ,ug depend only upon temperature and pressure. These expressions will be used in treating the kinetic data. Kinetic Measurements The ultrasonic relaxation technique has been used previously to study aggregation in solutions of propranolol hydrochloride and penthianate methobromide. Both drugs aggregate in a micellar manner with c.m.c. values of 0.106 mol dm-3 (PH) and 0.22 mol dm-3 (PMB).2. lo Chemical relaxation was not observed for drug concentrations below these values.In the present study, the concentration of drug in the aqueous phase was restricted to values below the drug c.m.c. so as to eliminate effects due to the self-aggregation of the drug. HDTABr was always present at a concentration well above its c.m.c., but the relaxation which occurs in this solution is too slow to be observed with the ultrasonic techniquell and no relaxational behaviour was found in solutions of HDTABr alone. The relaxation behaviour which was observed in solutions containing a drug and HDTABr must therefore be due to interaction between monomer drug and the mixed drug-HDTABr micelles. For all solutions the relaxation behaviour could be described in terms of a single relaxation process and relaxation times, z, and amplitudes, pmax, could be derived.In general, the behaviour was very similar to that found in solutions containing pentanol and hexadecylpyridinium chloride, which we reported recently . The relaxation data have been analysed according to the methods which have been applied and discussed el~ewhere,~~ 12, l3 Central to this approach is the relationship in which A V is the reaction volume change, IC, is the adiabatic compressibility of the solution, V is the volume of solution considered (in practice V = 1 dm3)I3 and 1 is a phenomenological coefficient, which is directly proportional to the equilibrium rate of reaction. To calculate A V we can eliminate 1 from eqn (4) using the relationship12 in which A is the reaction affinity and < denotes the extent of reaction. For the case considered, A = u , - u ~ ( 6 ) and eqn (5) becomes3 where C denotes the concentration of micellar surfactant and M is the concentration of micelles.Eqn (7) applies to the fast exchange process involving the transfer of drugJ . Gormally and S . Sharma 250 1 0 10 20 30 40 (D, D,/Dt)/ lo3 mol dm-3 Fig. 3. Plots of pmax us. D, Dm/D, [eqn (S)] for: 0 , PMB in 0.1 mol dm-3 HDTABr; 0, PH in 0.1 mol dmP3 HDTABr. These lines must pass through the origin. molecules between the aqueous and micellar phases. This process occurs at constant micelle concentration, M. From eqn (3), (7) and (4) we obtain Z A P D,D, rUmax= ~ ~ 2rc, RT D, in which D, is the total drug concentration, D, = D,+D,. Eqn (8) is plotted in fig. 3, from which it can be seen that the anticipated linear dependence is what is observed.From the gradients of these plots we can evaluate the volume changes associated with the transfer of a drug molecule from the aqueous to micellar environments. For the PH-HDTABr system a value of 2.8 cm3 mol-l is obtained and the corresponding value for the PMB-HDTABr system is 4.16 cm3 mol-l. The values obtained for the micellar aggregation of these drugs are 2.3 cm3 mol-l (PH) and 2.4 cm3 mol-1 (PMB),2 and A V for the solubilisation of pentanol in hexadecylpyridinium chloride micelles was estimated to be 4.7 cm3 m o E 3 It can be seen that the values obtained in the present study are within the expected order of magnitude. For the drug transfer reaction we can write down the equilibrium reaction rate in the form ~ = k-D, (backward rate) R TI V (9) = k+DaM (forward rate).(10) These equations define the rate coefficients k+ and k-. If k- is constant then eqn (4) and (9) imply that the quantity prnax/drn is also constant. The plots in fig. 4 show clearly that this quantity is not constant but increases with Dm in what appears to be a linear manner. This behaviour is very similar to that reported in connection with the micellar solubilisation of pentan01.~ We therefore write an expression for k- in the form: k- = k,( 1 +all,) (1 1) in which k, and a are independent of D,. The increase of k- with D, implies that the intrinsic rate with which drug molecules are released from the mixed micelles increases as the proportion of drug in the micelles increases. This is not surprising since the environment of the micellar drug molecules is expected to change substantially as D,2502 Relaxation in HDTABr Micelles f 4 Fig.4. Plots of,u,,,/(tD,) us. D, for: 0, PMB in 0.1 mol dmP3 HDTABr; 0, PH in 0.1 mol dm-3 HDTABr. Note that the highest value of the mole ratio of drug within the micelles is 0.375, corresponding to D, = 0.06 mol drnp3. increases. Taking the limit D, + 0 we can estimate the values of I;, from the intercepts on the ordinate in fig. 4. From eqn (4) it can be seen that k , is given by The values of k, so obtained are 2.7 x lo6 and 3.7 x lo6 s-l for the PH/HDTABr and PMB/HDTABr systems, respectively. From eqn (9) and (10) it can be seen that if k- increases with D, then the product k+M must also increase with D,, since D, is proportional to D,.There is no reason to expect that k+ should increase and the fact that the relaxation behaviour is observed in the ultrasonic time scale suggests that this quantity is close to being diffusion controlled. The remaining possibility is that M increases as the proportion of drug in the micelles increases. This must certainly occur to some extent, as one would expect that the addition of a cationic species to a cationic micelle would tend to destabilise the micelle. If we assume that k+ are diffusion controlled and that their values can be equated to the values found for the micellisation of these drugs then we can estimate the mean aggregation number for the HDTABr micelle in 0.1 mol dm-3 solution. The mean aggregation number will be given approximately by 0.1 M n=-.(13) The point of this exercise is to check that the numbers which emerge are not wholly unreasonable. It should be stressed that this procedure is not put forward as a method of determining mean aggregation numbers. In the limit D , -+ 0 we have 0.1 n k,D, = k+D, -. The values for k+ are 0.88 x lo9 and 0.82 x lo9 dm3 mol-1 s-l for PH and PMB, respectively.2 Using the values for k, and DJD, given earlier, we obtain for the mean aggregation number of the HDTABr micelle: n = 75 (from the PH-HDTABr data) and n = 92 (from the PMB-HDTABr data). These values refer to HDTABr solutions in theJ . Gormally and S. Sharma 2503 absence of drug and are in good agreement with values between 79 and 96 to be found in the literature.l49 l5 It may be noted that at the concentration used here, the HDTABr micelles are expected to be essentially spherical, the transition to rod shapes occurring at concentrations in excess of 0.2 mol dm-3.15 Conclusions The general features of the interaction between the two cationic drugs studied and HDTABr micelles are very similar and also similar to the interaction between pentanol and cationic mi~elles.~ In the three cases, the rate of association with the micelle approximates to being diffusion controlled and the intrinsic rate of release from the micelle increases as the proportion of HDTABr within the micelle decreases.That the solutions containing cationic: drugs should behave in a similar manner to those containing pentanol is surprising and suggests that the charge carried by the drug molecules does not significantly modify the kinetic processes.The mixed micellar solutions to which the above observations apply had a composition such that the mole fraction of drug within the micelles never exceeded a value of ca. 0.375. There was evidence that at higher drug concentrations the velocity curves of fig. 1 become non-linear. In this situztion, the method of treating the data would become difficult to apply with any confidence. The conclusion that k- increases with the concentration of micellar drug must clearly break down when the mole ratio of drug within the micelles becomes closer to unity. The small departure from linearity seen in fig. 1 , and also found on the corresponding curves for the propranolol-HDTABr system, is consistent with a decline in the rate of increase of k- with proportion of micellar drug.This is what one would expect to happen with increasing drug concentration. The conclusion that k- increases linearly with D, must therefore be considered to be an approximation which is valid only when the proportion of drug within the micelles is relatively low. References 1 D. Attwood, Aggregation Processes in Solution, ed. E. Wyn-Jones and J. Gormally (Elsevier, Amsterdam, 1983). 2 D. Causon, J. Gettins, J. Gormally, R. Greenwood, N. Natarajan and E. Wyn-Jones, J. Chem. SOC., Faraday Trans. 2, 1981, 77, 143. 3 J. Gormally, B. Sztuba, E. Wyn-Jones and D. G. Hall, J. Chem. Sac., Faraday Trans. 2, 1985, 81, 395. 4 F. Eggers, Acustica, 1968, 19, 323. 5 A. J. Barlow and E. Yazgan, Br. J. Appl. Phys., 1966, 17, 807. 6 D. M. Bloor, J. Gormally and E. Wyn-Jones, J . Chem. SOC., Faraday Trans. 1, 1984, 80, 1915. 7 A. Passynski, Acta Physicochim. U.R.S.S., 1938, 11, 606. 8 Y. Miyahara, Bull. Chem. SOC. Jpn, 1953, 26, 390. 9 K. Shigehara, Bull. Chem. SOC. Jpn, 1965, 38, 1700. 10 D. Attwood, J. Phys. Chem., 1976, 80, 1984. 11 E. A. G. Aniansson, S. N. Wall, M. Almgren, H. Hoffmann, I. Kielmann, W. Ulbricht, R. Zana, 12 D. G. Hall, J. Gormally and E. Wyn-Jones, J. Chem. SOC., Faraday Trans. 2, 1983, 79, 645. 13 J. Gormally, N. Natarajan, E. Wyn-Jones, D. Attwood, J. Gibson and D. G. Hall, J . Chem. SOC., 14 H. V. Tartar, J. Colloid Sci., 1959, 14, 115. 15 P. Ekwall, L. Mandell and P. Solyom, J. Colloid Interface Sci., 1971, 35, 519. J. Lang and C . Tondre, J. Phys. Chem., 1976,80, 905. Faraday Trans. 2, 1984, 80, 243. Paper 511698; Received 30th September, 1985

ISSN:0300-9599    

DOI:10.1039/F19868202497

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

J. Chem. Soc., Faraday Trans. I, 1986, 82, 2505-2514 Thermal Desorption and Infrared Studies of Primary Aliphatic Amines adsorbed on Haematite (a-Fe,O,) Ute Marx, Rolf Sokoll* and Hartmut Hobert Sek t ion Chern ie , Fr iedr ich- Sch ille r - Un ive rsit at, 6900 Jena , German Democratic Republic The adsorption of n-octadecylamine on a-Fe,O, at the solid/liquid interface, and of n-butylamine at the solid/liquid and solid/vapour interfaces has been studied by infrared spectroscopy. To obtain further information about the nature of desorbing products, temperature-programmed desorption experi- ments were made with n-butylamine-a-Fe,O, adsorbates. No difference could be detected by i.r. spectroscopy between the nature of adsorbates formed under the various mentioned conditions. Adsorption of n- octadecylamine and n-butylamine on a-FeTOs mainly involves coordinative interactions between amine and Lewis-acidic surface sites (Fe3+ cations).Furthermore, hydrogen bonds are formed between surface hydroxy groups and adsorbed amine molecules. n-Butylamine adsorbed on a-Fe,O, gave four different desorption peaks (I-IV) which are formed by n-butylamine (I: 423 K), butyronitrile (11: 530 K), CO, (111: 630 K) and H,O (IV: 713 K). Desorption of CO, and H,O is caused by the oxidation of amine molecules strongly adsorbed on two different types of coordination sites. -~ ~~~ ~ _ _ _ ~~~ ~~ ~ Until now no publications exist which deal with infrared spectroscopic investigation of the adsorption of amines on the surface of iron oxides. This is surprising because of the high practical importance of NH-containing substances, e.g.as corrosion inhibitors or as additives for lubricating oils. Not only are data necessary concerning the adsorption states, but also knowledge about thermally induced reactions between amines and iron oxides. Therefore we used infrared spectroscopy and temperature-programmed desorption (t.p.d.) to investigate the interactions between primary aliphatic amines and haematite (a-Fe,O,). To find out whether alkyl chain length or adsorption conditionsinfluence the results, n-octadecylamine (ODA) and n-butylamine (BA) were adsorbed from solutions of carbon tetrachloride and cyclohexane (ODA, BA) and from the vapour phase (BA). Experimental Materials Samples of a-Fe203 were obtained by vacuum decomposition (523 IS, 1 x Pa) of pressed discs of goethite (a-FeOOH)' in the infrared cell. The B.E.T.surface area of the resulting haematite sample was ca. 160 m2 g-I. X-Ray powder diffraction analysis confirmed that samples prepared under these conditions are polycrystalline a-Fe,O,. ODA, BA and both solvents were purified by distillation and were degassed immediately before their use in adsorption experiments. Apparatus and Procedures Infrared spectra were recorded by a Specord IR 75 spectrometer coupled with a computer KRS 4200 in the range 4000-1200 cm-l. The infrared cell used for adsorption from 83 2505 FAR 12506 Aliphatic Amines adsorbed on Haematite solution and the procedure for measuring the spectra were the same as in ref. (2). Investigations of the solid/vapour interface were carried out in a conventional infrared cell with KBr windows and an optical path length of 130 mm.In order to obtain identical surface states the oxide wafers were initially activated under vacuum (p = 1 x lo-, Pa) at 573 K. At higher activation temperatures a deoxidation occurs which gives a surface-reduced phase containing Fe2+ ions, possibly magnetite, responsible for a progressive loss of transmission.3* Spectra were recorded of the oxide sample activated under vacuum, in contact with the amine, after evacuation at beam temperature and finally after heating at increasing temperatures. For t.p.d. experiments, pretreatment of the samples was carried out as follows. 20 mg of goethite was placed in a quartz furnace and evacuated at 673 K and 6 x lo-* Pa for 2 h.The sample was then exposed for 0.5 h to various pressures of BA at room temperature to realize different degrees of surface coverage, and re-evacuated for 3 h before the thermal desorption run with a heating rate of 20 K min-l. A mass spectrometer CH8 (Varian Mat C.m.b.H.) connected to the t.p.d. equipment was used for monitoring the desorption spectra and for identification of the evolved desorption products. In some cases the infrared cell for adsorption from the vapour phase was coupled with the mass spectrometer to investigate special intermediate stages of the desorption process by infrared spectroscopy. Results Infrared Absorption Spectra In fig. 1 the infrared spectra are depicted of a typical experiment in which ODA was adsorbed from CC1, solution onto the surface of a-Fe,O,.The freshly prepared and activated oxide shows five bands at 371 1, 3648, 3618, 3469 cm-l [fig. 1 (A)(a)] and 1535 cm-l [fig. 1 (B)(a)]. According to Rochester and Topham, the bands between 3800 and 3600 cm-l can be ascribed to different types of isolated hydroxy groups which are unperturbed by lateral hydrogen bonding interactions with adjacent hydroxy groups. The broader maximum at 3469 cm-l can be attributed to surface hydroxy groups which are involved in strong hydrogen bonding interactions with adjacent groups. The frequency and intensity of the weak absorption band at 1535 cm-l are strongly dependent on the preparative and pretreatment conditions4 and this band seems to be due to a combination of lattice vibrations of the oxide.Immersion of the oxide disc in CCl, results in a perturbation of the isolated hydroxy groups (3607 cm-l and a shoulder at 3670 cm-l), whereas the band of interacting groups remains unaffected [fig. 1 (A)(b)]. After subsequent immersion of the disc in a solution of ODA in CCl, [0.01 mol drn-,, fig. 1 (A)(c)] a decrease of the band at 3607 cm-l and a shift of the band at 3469-3429 cm-I can be noticed. Furthermore a broad maximum is formed between 3600 and 3000 cm-l on which four sharper bands are placed at 3383, 33 12,3222 and 3 126 cm-l. Only three of these bands remain at 3308,3222 and 3 126 cm-l after subsequent replacement of the amine solution by the pure solvent in such a way that the oxide wafer is always immersed in a liquid phase [fig.1 (A)(d)]. Therefore these infrared bands are caused by adsorbed ODA, whereas the disappearing ones are due to the non-adsorbed excess of amine in solution. Infrared bands at 2917 and 2847 cm-l can be ascribed to CH stretching vibrations of ODA in the adsorbed state, compared with those at 2922 and 2850 cm-I for ODA in solution. Between 1800 and 1200 cm-l the only visible bands are those of the CH bending vibrations of adsorbed ODA at 1456, 1364 and 1303 cm-l, whereas the NH, bending vibration 6(NH,) is overlapped by the strong absorption of CCl, at ca. 1550 cm-l [fig. 1 (B)]. Therefore other experiments were carried out using cyclohexane as solvent, the results of which are represented in fig. 2. When an a-Fe,O, wafer was immersed in aU. Marx, R.Sokoll and H . Hobert 2507 ~ 28"O 3800 3500 wavenumber/cm-' Fig. 1. Infrared spectra of a-Fe203: (a) in vacuum; immersed in (b) CCI,, (c) ODA-CC1, at a concentration of 0.02 mol dm-3, (d) CCl, after adsorption (1 h). c ..-I + c x .- c 1700 1500 wavenum ber/cm -' Fig. 2. Difference spectra [(a) = (2) - (I), (b) = (2) - (3), ( c ) = (3) - (l)] of a-Fe20, immersed in (1) cyclohexane, (2) ODA+yclohexane at a concentration of 0.02 mol dmP3, (3) cyclohexane after adsorption (1 h). 83-22508 Aliphatic Amines adsorbed on Haematite I 1 R .. - . . . . . . . . . . . . . . , ! . . . . , : . . , . . . . . \- I t I 1 3 000 3500 3000 2800 1800 ls00 1200 Fig. 3. Infrared spectra of a-Fe,O,: (a) in vacuum; ( b k ( d ) after immersion in ODA4yclohexane at a concentration of 0.02 mol dm-, and subsequent drainage followed by evacuation at (b) beam temperature, (c) 443 K and (d) 493 K.wavenum ber/ cm- ' c .3 CI u E X ." c1 ~ 1500 12 3 800 3500 3000 2800 1800 wavenum ber/ cm -' 10 Fig. 4. Infrared spectra of a-Fe,O,: (a) in vacuum; (b)-(d) after adsorption of BA from the vapour phase and subsequent evacuation at (b) beam temperature, (c) 443 K and ( d ) 493 K. solution of ODA in cyclohexane, two bands at 1612 and 1577 cm-l appear in the difference spectrum. The former band is due to d(NH,) of free amine in solution and the latter is caused by adsorbed ODA. This illustrates the difference of the spectra of a-Fe,O, immersed in pure cyclohexane after removal of the ODA solution and before adsorption [fig. 2(c)], where only the band at 1577 cm-1 is seen.Between 2800 and 1800 cm-1 no bands of adsorbed ODA or of ODA in solution appear, therefore these parts of the infrared spectra are not shown here.U. Marx, R. Sokoll and H . Hobert 2509 Table 1. Wavenumbers of bands of ODA and BA adsorbed on a-Fe203 wavenum ber /cm-l ODA-Fe203 BA-Fe20, assignment 3308 3222 3126 2952 (sh) 2917 2847 1577 1456 1364 1527 1407 3315 3226 3129 2950 2923 2866 1580 1457 1371 1537 1414 a Above 473 K. 1800 1500 1200 wavenum ber/ cm -' Fig. 5. Difference spectra [(a) = (2) - (l), (b) = (3) - (l), (c) = (4) - (l)] of a-Fe203 : (1) in vacuum; (2)-(4) after adsorption of increasing amounts of BA from the vapour phase and subsequent evacuation at beam temperature. To compare the nature of the adsorbed species which can be observed in situ a t the solid/liquid interface with those existing on the surface of a-Fe,O, after complete separation of the solid and liquid phases, fig. 3 shows the infrared spectra of a-Fe,O, evacuated for 0.5 h at various temperatures after adsorption of a solution of ODA in cyclohexane and subsequent repeated washing with pure solvent.When the ODA-a-Fe,O, adsorbate was evacuated at beam temperature [curve (b)] the spectrum exhibits completely2510 Aliphatic Amines adsorbed on Haematite 300 600 500 600 700 800 TfK Fig. 6. Thermal desorpt1,n spectra of BA adsorbed at room temperature on a-Fe,O, : ( a ) complete saturation of the surface, (b) partially covered surface. 300 400 500 600 700 800 TI K Fig. 7. Thermal desorption spectrum of BA adsorbed at room temperature on a-Fe,O,: (a) total amount; (b)-(e) desorption products represented by the most intense peak in their mass spectrum : (b) BA, (c) butyronitrile, ( d ) H,O, ( e ) CO,.the same infrared bands as in the case of direct contact between the solid and the ODA solution. Increasing the temperature up to 473 K [curve (c)] causes a decrease of the intensity of all bands of adsorbed ODA and that of the NH, bending vibration shows a continuous shift to lower frequencies from 1577 cm-l at beam temperature to 1571 cm-1 at 473 K. The bands of hydroxy groups are shifted to higher wavenumbers without reaching the positions of those of freshly activated a-Fe,O,. After heating the adsorbate above 473 K a fundamental change in the spectrum occurs [curve (41. Simultaneous with the strong loss of transmission, all infrared bands of the surface hydroxy groups andU.Marx, R. Sokoll and H. Hobert 251 1 those at 3308, 3222, 3126 and 1577 cm-l completely disappear and the bands at 2917, 2847, 1456 and 1364 cm-l are greatly reduced in their intensity. On the other hand, two new bands become visible at 1527 and 1407 cm-l. By heating at temperatures higher than 523 K the sample completely loses its transmission. Adsorption of n-butylamine on a-Fe,O, at the solid/liquid and solid/vapour interfaces results in the same general phenomena as in the case of ODA adsorption from solution. This can be seen from fig. 4, in which the infrared spectra are shown of the adsorption of BA at the solid/vapour interface at beam temperature and of the BA-a-Fe,O, adsorbate after heating at increasing temperatures, respectively.Table I summarizes the results of the amine adsorption on haematite obtained by infrared spectroscopy, and an assignment is given of all observed bands. All the spectra discussed above were recorded after complete saturation of the surface of a-Fe,O, by the amines. In fig. 5 infrared spectra are depicted which show the influence of increasing the degree of surface coverage on the bending vibration of butylamine adsorbed on a-Fe,O, from the vapour phase. At very low coverages a small band of d(NH,) can be seen at 1555 cm-l which is greatly increased in intensity and shifted to higher wavenumbers when more amine is admitted. This result refers to the existence of different adsorption centres on the surface of haematite.Thermal Desorption Spectra Fig. 6 shows the thermal desorption spectra of BA adsorbed on a-Fe,O, at room temperature for two different initial coverages. The following characteristics can be noted: (1) In the case of a saturated surface [curve (a)] four maxima are present at (I) 423, (11) 530, (111) 630 and (IV) 713 K. (2) With decreasing coverage (down to the lowest investigated value) maximum (I) disappears and maximum (11) is reduced in intensity, whereas maxima (111) and (IV) remain unaffected. This result strengthens the concept of the existence of different adsorption centres on the surface of a-Fe,O,, which was also found by infrared spectroscopy. Mass spectrometric studies showed (fig. 7) that the four maxima are formed by the desorption of n-butylamine (I), butyronitrile (II), water and carbon dioxide (111, IV).Discussion The represented results show that no fundamental differences can be detected regarding the adsorption behaviour of the two investigated amines on a-Fe,O, at the solid/liquid and solid/vapour interfaces. Therefore a common discussion of the three cases is possible. Adsorption at Beam Temperature Many workers have investigated the adsorption of various organic molecules on a-Fe,O, by infrared spectroscopy. From these results it can beconcluded that mainly coordinatively unsaturated Fe3+ cations and surface hydroxy groups act as adsorption centres [e.g. ref. ( 5 ) ] . Therefore our results will be discussed now regarding the following two forms of adsorbate complexes : /H /* \H ‘H Fe3+.- .N-R Fe-OH- - -N-R. type A type B ODA molecules adsorbed on a-Fe,O, give rise to NH stretching vibrations which are shifted to lower wavenumbers in comparison with their positions in the case of the free amine in CCl, solution (vas: 3383 + 3308 cm-l, v,: 3317 -+ 3222 cm-l). The bending2512 Aliphatic Amines udsorbed on Haematite vibration [d(NH,)] is also shifted to lower wavenumbers (1 6 12 + 1577 cm-l), which indicates that ODA acts as an electron-pair donor bonded to the haematite surface via its nitrogen atom.,$ A comparison with previous results of the adsorption of ODA on SiO, [ref. (7)] and y-Al,O, [ref. (S)] shows that only an interaction as strong as in a type A adsorbate complex is able to cause such large shifts of the NH bands (ODA-SiO,: v,, = 3360 cm-l, v, = 3295 cm-l, 6 = 1585 cm-l, formation of hydrogen bonds between ODA and surface hydroxy groups; ODA-AI,O,: v,, = 3306 cm-', v, = 3220 cm-l, 6 = 1577 cm-l, formation of coordination bonds between ODA and A13- cations).Furthermore, on the surface of a-Fe,O, at least two different types of Lewis-coordinated amine species exist which are formed with increasing surface coverage. This is indicated by the abovementioned desorption behaviour of amine-a-Fe,O, adsorbates and by the increasing frequency of 6(NH,) at stepwise admission of the amine to the solid. Such different coordinatively unsaturated cations were also found on the surface of Al,O, by adsorption of ~yridine.~ On the other hand, after adsorption of ODA on haematite the infrared spectra show a broad absorption maximum between 3500 and 3000 cm-l and a shift of the hydroxy band from 3469 to 3429 cm-l.These phenomena can be attributed to the formation of hydrogen bonds between ODA molecules and surface hydroxy groups (type B), and possibly to an interaction of the alkyl chains of coordinatively bonded ODA molecules with adjacent hydroxy groups, respectively. Desorption Behaviour of the Amine-a-Fe,O, Adsorbates In the case of BA-a-Fe,O, adsorbates the desorption process can be divided into three fundamental ranges: (a) Between beam temperature and 473 K only BA desorbs. The infrared spectra show that all bands of adsorbed BA are reduced in their intensity. (b) Between 473 and 573 K the desorption of butyronitrile occurs. From the infrared spectra it can be seen that in this temperature range the bands of adsorbed BA and those of the surface hydroxy groups completely disappear, whereas two new bands become visible at 1537 and 1414 cm-l.(c) Above 573 K only CO, and H,O desorb. Infrared spectra are not available because of the strong loss of transmission at these temperatures. Desorption of unchanged BA in range ( a ) can be attributed to BA molecules which were hydrogen bonded to surface hydroxy groups of haematite and possibly coordinatively bonded to weak Lewis-acidic centres. In range ( b ) only coordinatively bonded BA molecules are dehydrogenated which leads to the desorption of butyronitrile. This dehydrogenation must happen at the moment of desorption because no nitrile adsorbed on the a-Fe,O, surface can be detected in the infrared spectra.Protons which are evolved from this process react with the surface hydroxy groups, forming water. That is the reason for the disappearance of the hydroxy bands at temperatures below the initial activation temperature of the haematite sample [fig. 3 (41. The simultaneously appearing infrared bands at 1537 and 1414 cm-l can be attributed to the stretching vibrations of carboxylate species which are following way : R C I &-(/ "o-6- . . 'Fe3' - - - antisymmetric and symmetric bound to Fe3+ cations in the This interpretation agrees with literature data concerning the adsorption of carboxylic acids on a-Fe,O,.*~ 5 7 lo* l1 The process which leads to the formation of such carboxylate species is assumed to be the well known reaction between nitrile and H,O, both formed by the dehydrogenation of BA molecules coordinated to strong Lewis-acidic surface centres : CH,-CH,-CH,-CN + 2H,O + CH,-CH,-CH,-COOH + NH,.U.Marx, R. Sokoll and H. Hobert 2513 T/K Fig. 8. Thermal desorption spectra of (a) BA and (b) propionic acid adsorbed at room temperature on a-Fe,O, (partially covered surfaces). The following facts support this assumption : (a) Nitrile desorption and carboxylate formation occur simultaneously. (b) Water, which was formed by dehydrogenation of BA, is neither adsorbed on the a-Fe,O, surface nor desorbed at this temperature, for which reason it must participate in a following reaction. ( c ) Ammonia was detected in the vapour phase by infrared spectroscopy and mass spectrometry in the temperature range of nitrile desorption.Furthermore, there seem to exist two different surface centres of haematite on which these processes take place. On the first ones formation of the carboxylate species competes with the desorption of nitrile. This is manifested by the fact that the amount of carboxylate species increases when the heating rate is decreased. On the other hand the desorption experiments showed that in the case of very small amounts of adsorbed BA, only the desorption maxima at 630 K and 713 K appear, which are formed by CO, and H,O. This refers to the existence of other surface centres on which adsorbed BA is converted into carboxylate species without simultaneous desorption of butyronitrile. But in this process nitrile also can be the intermediate product.The existence of two different coordination sites of different acid strength on the a-Fe,O, surface was also indicated by the infrared spectra of chemisorbed pyridine.12 Finally in range (c) only CO, and H,O desorb, indicating that a complete oxidation of all species which remained on the surface up to these temperatures occurs. The resemblance of the desorption behaviour in this temperature range to those of e.g. propionic acid adsorbed on a-Fe,O, (fig. 8) refers again to the above discussed formation of carboxylate species. Desorption experiments with ODA-a-Fe,O, adsorbates were possible between room temperature and 573 K. Therefore only ODA and n-heptadecylnitrile were detected as desorption products. All these results agree well with other literature data concerning the adsorption of various organic molecules on the surface of haematite.Also, in the case of oxygen- containing (e.g. methanol and formaldehydell. 13) as well as other adsorbates (e.g. butadiene and butene,l* ethylene15 and benzene16) the formation of oxidized products is observed. In the absence of gaseous oxygen this process must be accompanied by reduction of the oxide.2514 Aliphatic Amines adsorbed on Haematite References 1 C. H. Rochester and S. A. Topham, J . Chem. Soc., Faraday Trans. I , 1979, 75, 591. 2 R. Sokoll and H. Hobert, J . Chem. SOC., Faraday Trans. I , 1986, 82, 1527. 3 C. H. Rochester and S. A. Topham, J . Chem. Soc., Faraday Trans. I , 1979, 75, 1073. 4 V. Lorenzelli, G. Busca and N. Sheppard, J . Catal., 1980, 66, 28. 5 C. H. Rochester and S . A. Topham, J . Chem. Soc., Faraday Trans. I , 1979, 75, 1259. 6 E. L. Zhukova and I . I. Shmanko, Opt. Spektrosk., 1972, 32, 514. 7 R. Sokoll and H. Hobert, Z . Phys. Chem. (Leipzig), in press. 8 R. Sokoll and H. Hobert, 2. Phys. Chem. (Leipzig), in press. 9 C. Morterra, A. Chiorino, G. Ghiotti and E . Garrone, J. Chem. SOC., Faruday Trans. 1, 1979, 75, 271. 10 A. D. Buckland, C. H. Rochester and S. A. Topham, J . Chem. Soc., Faraday Trans. I , 1980, 76, 302. 11 G. Busca and V . Lorenzelli, J . Catal., 1980, 66, 155. 12 V. Lorenzelli and G. Busca, J . Catal., 1981, 72, 389. 13 J. Novakova, P. Jiru and V . Zavadil, J . Catal., 1971, 21, 143. 14 M. C. Kung, W. H. Cheng and H. H. Kung, J . Phys. Chem., 1979,83, 1737. 15 G. Busca, T. Zerlia, V. Lorenzelli and A. Girelli, J. Catal., 1984, 88, 125. 16 G. Busca, T. Zerlia, V. Lorenzelli and A. Girelli, J . Catal., 1984, 88, 1 3 1. Paper 511704; Received 1st October, 1985

ISSN:0300-9599    

DOI:10.1039/F19868202505

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

J. Chem. Soc., Faraday Trans. 1, 1986, 82, 2515-2519 Metal Particles supported by Porous Glass Roderick N. Edmonds, Martin R. Harrison and Peter P. Edwards* University Chemical Laboratory, Lensfield Road, Cambridge CB2 I E W We have investigated the use of porous ' Vycor ' glass as a supporting matrix for silver and alkali-metal particles. By different methods of preparation it was possible to produce silver particles either within the pores of the glass or on the external surface. The optical spectra of silver particles are reported and discussed. Alkali-metal particles gave strong electron spin resonance spectra, but in the case of silver weak electron spin resonance signals were detected from the surface particles only. ~ ~~ ~ _ _ _ _ _ _ ~ ~ ~ The use of porous 'Vycor' glass as a support for metal particles for catalytic applications has featured in several publications.l* This glass is perforated with channels between 4 and 7 nm in diameter and thus contains larger cavities than zeolites, whose use as metal particle supports has recently been in~estigated.~3 Porous supports such as these provide a means of stabilising very small particles which must normally be prepared in either a low-temperature matrix5 or an ionic crystal,6, or otherwise protected to prevent coagulation.In this paper we examine a variety of methods of preparing Vycor-supported silver particles, which are investigated by electron microscopy, X-ray diffraction, optical spectroscopy and electron spin resonance (e.s.r.) spectroscopy. We also report the first preparation and e.s.r.investigation of alkali-metal colloids supported by Vycor glass. Experimental Porous Vycor quartz glass (Corning glass no. 7930) was obtained in the form of plates, granules and powder. Organic impurities were removed by heating to 500 "C in air for 3 h, after which the glass was colourless and displayed no e.s.r. signals in the free-spin region at - 196 "C. Optical spectra were measured on plates 2 mm thick; granules and powder were used in the other experiments. To impregnate with silver, the cleaned glass was heated to 150 "C to remove water from the pores, immersed in a silver nitrate solution in either ammonia or triply distilled water, washed to clean the external surface and dried at room temperature. The silver was then reduced by one of the five methods described below; a typical loading was 0.05 % silver by weight from a 0.01 mol dm-3 silver nitrate solution.(i) Hydrogen reduction: samples prepared using ammonia were reduced in a stream of hydrogen gas for 12 h at room temperature to give a yellow colour. Room-temperature hydrogen reduction also produces silver particles in zeolites4 and in silver oxide solution.8 At 50 "C, hydrogen reduction resulted in blackening of the Vycor. (ii) Carbon monoxide reduction: samples prepared using ammonia were reduced in a stream of carbon monoxide gas for 6 h at room temperature to give a brown colour. (iii) Formaldehyde redu~tion:~ samples prepared using ammonia went grey or (at high silver concentrations) black on treatment with formaldehyde for 10 min.(iv) Borohydride reduction: samples prepared without ammonia were treated with a 0.00 1 mol dmP3 solution of sodium borohydride for 5 min, resulting in a yellow coloration or blackening at high silver concentrations. This method has been extensively used for preparing colloids in aqueous solution.lo (v) Irradiation : samples prepared without ammonia were exposed to 6 Mrad of 6oCo y-radiation. This resulted in a yellow colour which turned to black on subsequent annealing. 25152516 Metal Particles supported by Porous Glass 300 400 500 600 700 wavelength/nm Fig. 1. Optical absorption spectra of silver-impregnated Vycor plates reduced with (a) hydrogen and (6) carbon monoxide. The ultraviolet cut-off is due to the glass. To impregnate Vycor with alkali metals, granular samples of cleaned glass were exposed to sodium, potassium, rubidium or caesium vapour in quartz reaction vessels at high temperature.3 The glass was first heated to 400 "C in the reaction vessel to remove all absorbed water.Alkali metals were purified by melting under vacuum and forcing into glass capillary tubes. Oxidation of the metals was confined to the ends of the tubes which were discarded before use. A suitable amount of metal (typically 0.1 g for 1 g of Vycor) was sealed into the reaction vessel under vacuum and vaporised in a furnace. The Vycor granules immediately assumed a blue-black colour on contact with the metal vapour . Optical spectra were recorded by inserting coloured Vycor plates into a Pye Unicam 8800 u.v.-visible spectrometer.Our electron microscope was a Jeol JEM-200CX instrument operating in transmission mode. E.s.r. measurements were made on a Varian E-109E spectrometer with a flow of nitrogen gas to control the cavity temperature. Results and Discussion Optical Spectra Vycor plates reduced with hydrogen or borohydride appeared yellow; those reduced with carbon monoxide were red-brown in colour. Fig. 1 shows the optical spectra after hydrogen and carbon monoxide reductions; the absorption maxima occur at 400 and 425 nm, respectively. Borohydride reduction gave a similar absorption signal to that from hydrogen reduction; similar spectra have also been recorded from silver colloids in aqueous suspension,1° in crystals6 and in glasses,ll and are typical of silver particles less than ca.10 nm in size. There has been much discussion on the size dependence of the optical absorption signal,12> l3 which indicates that the broader signal at longer wavelength resulting from carbon monoxide reduction arises from larger metal particles. By splitting open the coloured plates we found that reduction by carbon monoxide and borohydride had coloured only the surface of the glass, while the colour from hydrogen reduction penetrated deep into the interior. The explanation is probably that reduction first occurs at the openings of the pores and the metal particles thus formed obstruct all but the small hydrogen molecule from further entry into the material.J. Chem. SOC., Faraday Trans. 1 , Vol. 82, part 8 Plute 1 Plate 1. Electron micrographs of Vycor-supported silver particles; (a) reduced using carbon monoxide at room temperature, (b) reduced using hydrogen at 50 "C. R.N. Edmonds, M. R. Harrison and P. P. Edwards (Facing p . 25 17)R. N . Edrnonds, M. R. Harrison and P. P. Edwards 2517 100 G - Fig. 2. E.s.r. spectrum of surface silver particles produced by carbon monoxide reduction. The low-field side of the resonance is overlapped by a cupric impurity signal. Electron Microscopy Plate 1 shows electron micrographs of two samples prepared by impregnation with a 0.01 mol dm-3 silver nitrate solution and reduction with carbon monoxide at room temperature or hydrogen at 50 "C. The carbon monoxide reduction produced large surface particles with an average diameter of ca. 50 nm. Hydrogen reduction gave particles up to 20 nm in size; of these the smaller ones (< ca.10 nm) are located mainly in the pores, while the others are probably on the external surface. Formaldehyde reduction gave large surface particles, while borohydride produced a similar size distribution to hydrogen, although, as remarked above, these particles are probably confined to the glass surface. Under high-resolution conditions, lattice images were visible in both the large and small particles. The crystalline nature of the surface particles was confirmed by X-ray diffraction using a concentrated sample, but at normal silver loadings the X-ray pattern was too weak to observe. Electron Spin Resonance No e.s.r. signals were detected from particles in Vycor pores. y-Irradiated samples (both before and after annealing) showed only F-centre and free radical signals together with weak doublets due to atomic ~i1ver.l~ Yellow hydrogen or borohydride reduced samples gave no e.s.r.signals, but a weak signal appeared in the black samples which had been hydrogen reduced with heating. This signal also appeared in carbon monoxide and formaldehyde reduced samples (fig. 2) and is assigned to surface particles; it has g = 2.02f0.01 and a peak-to-peak linewidth of 350 G. On heating from 100 K to room temperature it diminished in intensity without broadening. In contrast, the bulk silver resonance15 at g = 1.983 is too broad to observe except at the very lowest temperatures.16 There has been much discussion of the 'quantum size effects' in small metal particlesf7 and particularly of the e.s.r.spectrum in particles so small that the mean electronic- energy-level spacing exceeds the microwave energy. Several authors have reported e.s.r. studies of small silver particles with g-values above free spin, similar to our own.l87 l9 In very heavily loaded samples no e.s.r. signal was detected at 77 K, suggesting that the particles were large enough to assume bulk-metal characteristics. Our results indicate that the smallest silver clusters do not give e.s.r. signals and in particular that those trapped in Vycor pores do not do so. This has been noted in Vycor before,l* although signals from particles less than 10 nm in size have been observed in low-temperature matrices5. and on other supports.20 The alkali-metal impregnated samples gave much stronger e.s.r.signals than those loaded with silver. The g-values and room-temperature linewidths for one sample each of sodium, potassium, rubidium and caesium are given in table 1. The linewidths were independent of temperature, except in the case of potassium where the remark- ably narrow room-temperature signal broadened to 9.5 G at 77 K. The linewidths of2518 metal Metal Particles supported by Porous Glass Table 1. E.s.r. parameters for alkali-metal species sodium potassium rubidium caesium ~ g-value particles in Vycor particles in linewidth/G g-value zeolite Y a bulk metalb 60 1.996 f. 0.001 1.9997 2.0015 5.5 1.999 0.001 1.9978 1.9997 80 1.998 & 0.001 1.989 1.9984 - 400 2.0 10 & 0.003 2.013 a From ref. (3). From ref. (22) and (23). individual samples varied with metal concentration, but no distinct trend was identified.The sodium resonance lineshape was slightly asymmetric; the others were symmetric at room temperature and at 77 K, indicating that the metal particles were much smaller than the microwave skin depth21 of ca. 500 nm. Comparison with the results for silver suggests a particle size much smaller than this, while the glossy black appearance of all alkali-metal samples indicates metal particles on the glass surface. The g-values in table 1 are very similar to those of metal particles in zeolite Y:< and in some cases differ significantly from the bulk value^,^^^ 23 which are given for comparison. Our temperature-independent linewidths are inconsistent with bulk metal measurements16 and, except for sodium, the room-temperature linewidths are much smaller than in the bulk metals, indicating less-efficient spin relaxation in the small particles. Explanations for the g-values and linewidths have been proposed.3 Conclusions Our experiments have demonstrated the potential use of porous Vycor glass as a support for small metal particles.The main advantage of Vycor over conventional silica supports lies in its porosity, which facilitates particle-size control as well as vastly increasing the number of particle sites. Our results show, however, that the method of reduction must be carefully chosen in order to impregnate the pores with metal and it appears that hydrogen is the most suitable agent in the case of silver. We have reported the first e.s.r.measurements of silver and alkali-metal particles supported by Vycor and have obtained signals fundamentally different from those of the bulk metals. Our optical measurements have demonstrated the use of porous glass as a convenient stabilising support for metal particles which, without such support, would normally be prone to coagulation. We thank J. M. Thomas for permission to use the Jeol instrument, We are grateful to D. G. Duff, D. A. Jefferson and T. Rayment for assistance with the measurements and Jan Schreurs for supplying Vycor glass. We thank the S.E.R.C. for financial support, and Jesus College, Cambridge for the award of a Research Fellowship to R. N. E. References 1 R. B. Clarkson and A. C. Cinllo, Jr, J. Vac. Sci. Technol., 1972, 9, 1073.2 R. B. Clarkson and A. C . Cirillo, Jr, J . Catal., 1974, 33, 392. 3 M. R. Harrison, P. P. Edwards, J. Klinowski and J. M. Thomas, J . Solid State Chem., 1984, 54, 330. 4 D. Hermerschmidt and R. Haul, Ber. Bunsenges. Phys. Chem., 1980, 84, 902. 5 A. Chatelain, J-L. Millet and R. Monot, J. Appl. Phys., 1976, 47, 3670. 6 S. C. Jain and N. D. Arora, J . Phys. Chem. Solids, 1974, 35, 123 1. 7 M. A. Smithard, Solid State Commun., 1974, 14, 41 1. 8 R. Monot, C. Narbel and J-P. Borel, Nuovo Cimento B, 1974, 19, 253. 9 M. Jarjoui, P. C . Gravelle and S. J. Teichner, C. R. Acad. Sci. Paris, Ser. C, 1975, 280, 705.R. N . Edmonds, M . R. Harrison and P. P. Edwards 2519 10 J. A. Creighton, C. G. Blatchford and M. G. Albrecht, J. Chem. Soc., Furaduy Trans. 2, 1979, 75, 790. 11 M. A. Smithard and R. Dupree, Phys. Stat. Sol. A , 1972, 11, 695. 12 A. E. Hughes and S. C. Jain, Adv. Phys., 1979, 28, 717. 13 U. Kreibig and C. V. Fragstein, 2. Phys., 1969, 224, 307. 14 R. Monot and J-L. Millet, J. Phys. Lett. (Paris), 1976, 37, L45. 15 S. Schultz, M. R. Shanabarger and P. M. Platzman, Phys. Rec. Lett., 1967, 19. 749. 16 P. Monod and F. Beuneu, Phys. Rev. B, 1979, 19, 91 1. 17 A. Kawabata, J . Phys. Soc. Jpn, 1970, 29, 902. 18 J-P. Borel, Phys. Lett. A , 1976, 57, 253. 19 S. C. Jain, N. D. Arora and T. Rs. Reddy, Phys. Lett. A , 1975, 54, 53. 20 A. Abou-Kais, M. Jarjoui, J. C. Vedrine and P. C. Gravelle, J . Cutul., 1977, 47, 399. 21 F. J. Dyson, Phys. Rev., 1955, 98, 349. 22 F. Beuneu and P. Monod, Phys. Reo. B, 1978, 19, 2422. 23 S. Schultz and M. R. Shanabarger, Phys. Rev. Lett., 1966, 16, 178. Paper 511728; Receiced 7th October, 1985

ISSN:0300-9599    

DOI:10.1039/F19868202515

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

J . Chem. SOC., Faraday Trans. I , 1986,82, 2521-2530 Reactivity of Solvated Electrons in Tetrahydrofuran Abdul A. H. Kadhum and G. Arthur Salmon" The University of Leeds, Cookridge Radiation Research Centre, Cookridge Hospital, Leeds LS16 6QB The rates of reaction of solvated electrons (e;) with 23 solutes have been measured in tetrahydrofuran at 22 & 1 "C. The values range from (1.5 _+ 0.4) x lo9 dm3 mol-1 s-l for phenol to (1.56k0.01) x lo1' dm3 mol-1 s-l for benzonitrile. It is concluded that the diffusion-controlled rate constant for reactions of e; in this solvent is ca. 1.7 x 10" dm3 mol-l s-l, which indicates that the sum of the reaction radii of e; and the solute is ca. 1.67 nm. This value is close to the value of 1.38 nm, estimated for the maximum reaction radius of e- in the gas phase.The effect of AG* for the reaction on the rate constant is investigated and evidence is presented for an inversion in the rate in the highly exergonic region, although a few solutes deviate from the general trend. The temperature dependence of the rate constants for five solutes has been measured and the values of the activation energy lie in the range 11.7-15.4 kJ mol-I. It is concluded that the rate constants for the electron-scavenging process qualitatively conform to the predictions of quantum-mechanical theories for electron-transfer reactions. ~ ~ ~______ ~~~~ Extensive studies have been made of the reactivity of solvated electrons (e;) in water,'? alcohols3? * and hydrocarbons [see for example, ref. (5)]. In water and alcohols where the electron is strongly solvated there are wide variations in reactivity up to the diffusion-controlled limit of ca.1O1O dm3 mol-1 s-l, but the reactivity of the less-reactive solutes seems to be governed primarily by the entropy of activation, since the energies of activation vary little from those for diffusion in the solvent. Marcus6,' has pointed out that the reactions of solvated electrons with solutes are a special case of electron-transfer reactions and has given a theoretical treatment of the dependence of the rate constant on the overall free energy of the reaction (AGe). In common with Marcus's treatment of electron-transfer reactions, this predicts a parabolic dependence of log k on AG* with k decreasing in the highly exergonic region. However,* for reactions of eLq no evidence has been found for this inversion of the rate, but for excess electrons in hydrocarbons there is evidenceg that the rate is governed by energy consideration?.In hydrocarbons the mobility of the excess electrons be) depends strongly on the structure of the solvent, with values ranging from 430 cm2 V-l s-l for liquid methanelo to 0.08 cm2 V-l s-l for n-hexanell and 0.013 cm2 V--l s-l for trans-decalin.12 For solvents with p e < 1 cm2 V-l s-l the rates of electron attachment to solutes have been shown to be diffusion controlled with rate constants varying linearly with p e [see for example ref. (5)]. On the other hand, in solvents with large values of pe, the rates of attachment are less dependent on pe and seem to be governed mainly by energetic considerations, which results in the rate being dependent on &, the energy of the electrons in the conduction band of the 1iquid.ll Although some values have been reported for the rate constants for the reactions of solvated electrons in a number of solvents of polarity intermediate between the hydrocarbons and the hydroxylic solvents, no systematic study appears to have been made in a single solvent.In this paper we report such a study in tetrahydrofuran (THF), which is a weakly polar aliphatic ether with relative permittivity of 7.4 and in which the solvated electron (e;) has A, = 2100 nm.13 252 12522 Solvated Electrons in Tetrahydrofuran Experimental The techniques of pulse radiolysis employed in this laboratory have been described elsewhere.14-16 In these experiments the irradiation cell for the measurements at room temperature was of 10 x 10 x 7 mm with optical path length of 1 cm and that for measurements over a range of temperatures was 5 x 15 x 30 mm with optical path length of 1.5 cm.These cells were joined to 100 cm3 bulbs, which were closed by a Young's greaseless stopcock and acted as reservoirs for the solutions under study. Experiments at temperatures other than room temperature used the apparatus described e1sewhere.l' The majority of the absorption measurements on the decay of e; were made at 990 nm using an E, G and G SHS-100 silicon photodiode with rise time of ca. 8 ns, but studies on galvinoxyl solutions were made at 1500 nm using a Barnes Engineering A-1 00 In-As photodiode (risetime = 80 ns) since the galvinoxyl radical anion was shown to absorb significantly at 990 nm.At both wavelengths the intensity of the analysing Xe arc lamp was enhanced for a period of ca. 5 ms by discharging a bank of condensers through the lamp via an electronic control circuit and the electron pulse was timed to occur during the early part of the light flash. Irradiation was generally with 25 ns pulses, but 50 ns pulses were used in a few experiments. Signals were recorded by a Tektronix R7912 transient digitiser and were fed to a DEC 11/23 computer for storage and processing. Tetrahydrofuran (Hopkin and Williams, analytical grade) was purified by first distilling to remove the stabiliser and then distilling under vacuum onto Na/K alloy, over which it was stirred with an all-glass magnetic stirrer until it developed the characteristic blue colour of Na-.The solvent was stored over the alloy until required. All the solutes used were purified by standard methods except for galvinoxyl (Aldrich) and sulphuric acid (B.D.H. AristaR), which were used as received. Solutions of involatile solutes were prepared by first making up, in a dry glove-box, a concentrated solution in purified THF. A small aliquot of this solution was added to the pulse radiolysis sample bulb, which was then closed by the Young's stopcock, removed from the glove-box and connected to the vacuum line. The appropriate quantity of purified THF was then distilled under vacuum off the Na/K alloy into the sample bulb and the solution so-obtained was further deaerated by three freeze-pumpthaw cycles.The volume of THF added was determined by weighing the sample bulb. Solutions of methyl and ethyl bromides were prepared as described previously. l8 Solutions of oxygen and nitrous oxide were made up by attaching the sample cell to the vacuum line and after distilling in the appropriate quantity of solvent the bulb was immersed in a water bath to reach thermal equilibrium. The required gas was admitted to the cell and after shaking for 5 min to ensure saturation, the pressure was measured using a Bourdon spiral gauge. The concentration of the gas was assessed using Charles' law and the absorption coefficients of the gases were measured using a technique similar to that of Markham and Kobe.lg The solubilities of oxygen and nitrous oxide in THF at 20 "C were found to be 3.4 x and 4.0 x lop2 mol dm-3, respectively, under partial pressures of the gases of 760 mmHg.? Results Reactivity at Room Temperature The rates of reaction of e; with a number of solutes e;+S -+ S- or R'+P- (1) were measured at room temperature (22k 1 "C) by observing the effect of solute concentration on the rate of decay of e;.The concentrations of the solutes used were chosen so that the time-scale of the decays was appropriate to the detector. In every case t 1 mmHg M 1.336 x lo2 Pa.A . A . H. Kadhum and G. A . Salmon 2523 the decays were observed to be pseudo-first-order with rate constants that varied linearly with the solute concentration. The slopes of these concentration dependences, i.e.the bimolecular rate constants for reaction (I), and the standard error in the slope were obtained using a least-squares linear-regression treatment of the data and are listed in table 1, together with the doses and range of concentrations used. Dependence on Electron Affinity of Solute It was desired to test the dependence of the rate of reaction (1) on the overall free-energy change, AGe, for the process, which is given by AGe = - EA(S) + AGp(S-) - AG,(e-) (2) where EA(S) is the gas-phase electron affinity of S, AG,(S-) is the free energy of polarisation of S- in the liquid and AGs(e-) is the free energy of solvation of e-. Values of EA(S) were mainly taken from the compilation of Beitz and Miller,20 which are based on half-wave reduction potentials in non-aqueous solvents.Where values of EA(S) were not available, they were estimated from Ei values using the same method as Beitz and Miller,20 which uses eqn (3) below and is based on the correlation used by Chen and (3) Went wor th :21 where P- is identical with AGp(S-), Ei is the polarographic half-wave reduction potential of S us. the saturated calomel electrode (SCE) and C is a constant that depends on the reference electrode. The half-wave potential for phenol is not available, but its electron affinity has been estimated by Younkin et a1.22 Values of AGp(S-) were estimated from EA(S) = Ei + P- + c the Born equation: - e2 (4) where E , is the relative permittivity of the solvent, E~ the permittivity of free space, e the electronic charge and r the effective radius of the molecular ion, S.This latter quantity was estimated from the molar volume of the neutral molecule in the condensed state, which is given by 1 M N d 4ny3 = - - where M is the molar mass, d the density of the solute and N the Avogadro number. It is necessary to correct the values of k , for the influence of diffusion and this was done 1 1 = +- 1 using the equation - - kobs kdiff kact in which the rate constant for the diffusion-controlled process, kdiff, was chosen to be 2.0 x loll dm3 mol-1 s-l, i.e. 25% higher than kobs for benzonitrile. The observed values, k,, and the corrected values, kaCt, together with EA(S), I?+ and AG,(S-) for a range of solutes are listed in table 2. The choice of kdiff is somewhat arbitrary since the kobs values for those solutes with E+ more positive than for benzonitrile do not remain constant over an extended range of Ei.Whereas the value of kact for benzonitrile is very sensitive to the choice of kdiff, this is much less so for the other solutes. However, with kdif, = 2.0 x 10l1 dm3 mol- s l, kact for benzonitrile is compatible with that for the other solutes (see below). If AG,(S-) in THF is equal to P- for the same solute in the solvent used to measure E+ then eqn (2) and (3) can be combined to give A G e = - Ei - AG,(e-) - C. (7) Since E , is 37.5 for acetonitrile and 36.7 for N,N'-dimethylformamide, which are the2524 Solvated Electrons in Tetrahydrofuran Table 1. Rate constants for the reaction of e; with some solutes solute dose per pulse k /lolo dm3 mol-1 s-' PI / mol dmP3 /GY 2,4,6-trimethylpyridine phenol 4-aminobenzonitrile p yridine pyrimidine benzonitrile 9,lO-diphenylanthracene nitrobenzene m-dini tro benzene tetranitromethane p-benzoquinone tetracyanoethylene galvinoxyl p-tetrachlorobenzoquinone 2,2-diphenyl- 1 -picrylhydrazyl methyl bromide ethyl bromide tetrachloromethane triphen ylmeth ylchloride benzyl chloride sulphuric acid oxygen nitrous oxide 1.2-6.7 0.44.8 0.6-4.8 0.2-0.8 O .M . 5 2.2- 1 3.6 0.3-2.2 0.3-2.3 0.2-6.3 o . w . 2 0.1-3.0 0.1-1 .o 0.4-4.3 0.1-0.6 0.2-1.5 0.5-3.2 0.4-3.5 0.3-2.5 0.4-3.8 0.1-0.8 0.1-0.7 0.1-1 .o 0.4-3.5 54 17 13 22 13 21 12 62 33 30 29 25 18 8 14 16 27 14 5 13 39 34 13 0.15 f 0.04 0.18 f 0.02 1.96 +O. 15 5.2 k 0.3 6.75 & 0.23 15.6f 1.0 7.6 0.3 5.41 & 0.13 4.51 f 0.04 2.36+ 0.14 2.0 f.0.4 2.12 & 0.17 2.85 f. 0.17 9.45 f 0.24 6.24 f 0.04 6.05 f 0.07 2.45 f. 0.40 9.2 k 0.3 3.17 k 0.05 10.03 + 0.04 1.41 f 0.03 1.88 & 0.20 3.33k0.14 solvents generally employed to measure Ei values, reference to eqn (4) and the values of AG,(S-) in table 2 indicates that AG,(S-) will not differ by more than 0.25 eV from P-. As AG,(e-) is a constant this suggests that a plot of k,,,t us. Ei will provide a fairly accurate assessment of the dependence of kact on AGe for a range of solutes. Such a plot is shown in fig. 1. As indicated above, the overall shape of this plot is relatively unaffected by the choice of kdiff, but if kdif, is chosen to be less than cu. 2.0 x lo1' dm3 mol-1 s-l, kact for benzonitrile is too large in relation to those of the other solutes.In fig. 1 three compounds, namely pyrimidine ( 9 , p-tetrachlorobenzoquinone (1 8) and 2,2'-diphenylpicrylhydrazyl ( 19), are somewhat displaced from the general trends of the graph. Possible reasons for this could be that: (i) the reported E+ values are in error and (ii) reaction (1) yields an excited state of the solute radical anion for solutes (18) and The E: values for these compounds were checked? by cyclic voltammetry with dimethylformamide as the solvent and a Ag/AgNO, reference electrode. The values obtained were referred to the saturated calomel electrode (SCE) on the basis that the potential of the above reference electrode is +0.30 V against the SCE.28 The values obtained for p-tetrachlorobenzoquinone and 2,2'-diphenyl- 1 -picrylhydrazyl were - 0.29 and -0.19 V, respectively, us.SCE. The discrepancy of -0.32 V in the value for p-tetrachlorobenzoquinone does not alter the main conclusion that the value of kact for this solute is higher than expected from the general trends in fig. 1. If the second explanation accounts for the higher values of kact for solutes 18 and 19 (19). t We thank Dr N. Taylor (Department of Physical Chemistry, University of Leeds) for performing these measurements.Table 2. Rate constants and free energy parameters for the reaction of e; with various solutes solute 1 2,4,6-trimethylpyridine 2 phenol 3 4-aminobenzonitrile 4 pyridine 5 pyrimidine 6 a-methylstyrene 7 benzonitrile 8 biphenyl 9 trans-stilbene 10 pyrene 1 1 9,lO-diphenylanthracene 12 nitrobenzene 13 m-dinitrobenzene 1 4 te trani trome t hane 15 p-benzoquinone 16 tetracyanoethylene 17 galvinoxyl 18 p-tetrachlorobenzoquinone 19 2,2-diphenyl- 1 -picrylhydrazyl 0.15 Ifr 0.04 0.18 k 0.02 1.96 Ifr0.15 5.2 f 0.3 6.75 k 0.23 9.84 & 0.38d 15.6 kO.1 11.0 k 3.W 1 1.6 & 0.03d 1 1.5 & 0.3d 7.6 & 0.3 5.41 k0.13 4.5 1 k 0.04 2.36 k0.14 2.0 Ifr 0.41 2.12 *o.10 2.85 Ifr 0.17 9.45 Ifr 0.24 6.24 -t 0.04 0.15 0.18 2.2 7.5 11 23 31 36 35 14 (1 68) 7.9 6.1 2.7 2.3 2.4 3.4 9.8 21 2.8 2.9gb 2.80" 2.65 2.33 2.54e 2.42 2.58 2.18e 2.09 1.83 1.12 0.85" 0.52 - -0.17 - 0.09' - 0.03 0.18" -0.48 -0.66 -0.48 -0.33 -0.01 -0.22 -0.1 0.04 0.47 0.63 0.88 1.20 1.45 1.63 1.86 2.49 2.58 2.65 2.9 1.66 1.89 1.96 1.97 1.67 1.81 1 S O 1.48 1.56 - 1.81 1.78 1.72 1.95 1.85 1.18 1.23 1.63 1.96 1.45 1.71 1.54 1.95 2.19 3.01 3.23 3.35 3.81 4.34 - - b b 3 a Measured against the saturated calomel electrode (SCE).Estimated using eqn (3). Ref. (23). Ref. (24). Ref. (25). f Ref. (26). Ref. (27).2526 12 A- 1 1 E E v1 4 0 m P \ V - r3. 10 9 Soluated Electrons in Tetrahydrofuran I I 1 I 0 7 2 0 I 0 3 13 18 19 e \: 3.0 2.0 1.0 0.0 Ef vs.SCE/V Fig. 1. Dependence of kact on E4 of solute. Solutes labelled as in table 2. 3.0 3.5 4.0 lo3 KIT Fig. 2. Effect of temperature on k, for five solutes: (1) benzonitrile (0, -), (2) biphenyl (0, ---), (3) pyrimidine, (4) kdiff = 8000 RT/3q, ( 5 ) p-benzoquinone, (6) phenol.A . A . H. Kadhum and G. A . Salmon 2527 Table 3. Arrhenius parameters for reaction of several solutes withe; solute log (A/dm3 mol-l s-l) EJkJ mol-1 benzonitrile 13.3f 1.4 12.5k2.5 biphenyl 13.2 f0.4 12.4+ 1.4 pyrimidine 12.7k0.3 11.7k1.4 p-benzoquinone 12.6 k0.6 15.4 2.8 phenol 11.9k1.0 15.1 k2.5 k(S to kes-Einstein) 11.4 7.5 ~~ be expected that emission from the excited states of the radical anions of then it might 1 these solutes would be observable on the same time-scales as the removal of e;.Attempts to observe such emissions proved negative. Temperature Dependence The effect of temperature on the reactivity of e; towards electron scavengers was studied for five solutes; benzonitrile, biphenyl, pyrimidine, p-benzoquinone and phenol. These compounds were selected since they lie in the various regions of the plot of logk,,, us. El (fig. 1). Pulse radiolysis experiments for the various solutes were performed for a range of temperatures between 295 and 229 K, using a range of solute concentrations, and values of k, were determined as indicated above.Arrhenius plots for the solutes studied are shown in fig. 2, along with the corresponding plot for the diffusion-controlled rate constant, kdiff, based on the Stokes-Einstein relationship : kdif, = 8000 RT/3q dm3 mo1-l s-' (8) where 'I is the coefficient of viscosity of THF at the temperature Tin units N s m-2. The Arrhenius parameters derived from the plots in fig. 2 are listed in table 3. Discussion For several solutes with Ei between -2.09 and -2.42 us. SCE k , exceeds 1.1 x lo1' dm3 mol- s-l and for benzonitrile the value is 1.56 x 10" dm3 mol-l s-l. As discussed above, kdiff cannot be less than 1.6 x 10,' dm3 mol-' s-l and is probably ca.2.0 x lo1, dm3 mol-1 s-l. The reaction radius R can then be evaluated from kdiff = 4 0 0 0 ~ NRp, kT/e (9) where p e is the mobility of the electron, k the Boltzmann constant and e the electronic charge (all in S.I. base units), kdif, is in units dm3 mol-1 s-l, and N is Avogadro's constant. m2 V-l s-l 29 gives R = 1.67 nm. Warman5 has drawn attention to the similarity between the reaction radius for the reaction of excess electrons with solutes in a number of hydrocarbons in which the electron has a low mobility and (&), the thermally averaged value of 7t, where k = L/2n and II is the de Broglie wavelength. At room temperature (A) = 1.38 nm5 and, therefore, in THF R is slightly larger than (k). This may be due to the effective radius of the scavenger also being included in R 3 0 The earlier assumption31 that the reaction radius for the reaction between e; and biphenyl in THF and similar solvents is between 0.3 and 0.5 nm seems to be without foundation. Logan32 has shown that for reactions which are near to being diffusion controlled the observed energy of activation, E,, is given by Using the above value of kdif, and p, = 6.2 x E, M B+RT (10)2528 Solva ted Electrons in Tetra h y dro fur an where B is the activation energy for self-diffusion in the solvent. Eqn (10) predicts that E, in THF should be ca.7.5 kJ mol-l, i.e. the value deduced on the basis of the Stokes-Einstein relationship. The fact that E, for biphenyl and benzonitrile are higher than this value suggests that the energy of activation for migration of e; in THF, Ep, is ca.10 kJ mol-1 and that the migration of e; is not governed solely by the viscosity of the solvent. Ep has not been measured for THF, but it has been that for several ethers it is significantly larger than the energy of activation for viscosity of the solvent owing to thermal excitation of the electron to the quasi-free state. The most striking feature of the data in fig. 1 is that, despite some deviations from the general trend, they show clearly a maximum in the rate constant for solutes with Ei z - 2.4 V us. SCE, followed by a decrease as E; becomes more positive. This reduction in rate in the highly exergonic region has not been previously observed for reactions of solvated electrons nor for the more general case of intermolecular electron-transfer in 34 although a similar effect has been reported for an intramolecular electron- transfer reaction.35 However, as reference to fig.1 shows, this reduction in rate is not expected to be observed if the rate constant for the diffusion-controlled reaction is less than ca. lo1' dm3 mol-1 s--l. It is clear that the general form of the dependence of kact on AG* does not conform in detail to the Marcus which predicts a parabolic dependence of logk on the free energy of reaction. This is obviously so on the branch of the curve corresponding to large negative values of the free energy of reaction, i.e. values of E; > - 2.0 V us. SCE. This departure from Marcus theory can also be seen in the magnitude of the energy of activation for the reaction of p-benzoquinone.Marcus theorys* predicts that the energy of activation for the reaction is given by E,=w+- 1+ 4 "i AG:-lvY where w is the work involved in bringing the reactants up to the mean separation distance in the activated complex, 1 is the sum of the reorganisation energies of the solvated electron and the other reactant, and AGe' is the standard free energy of reaction modified for the loss of translational free energy due to the disappearance of the electron. Where one of the reactants is neutral, w can be approximated to zero and d is then given by the value of AGe' corresponding to the maximum rate. Estimates of AGe' require a knowledge of AGs(e-) in eqn (2). This quantity is not known with certainty in THF, but it is reasonable to assume that the optical absorption of e; corresponds to the transition from the bound state to the quasi-free state of the electron and if we further assume that V,, the energy of the bottom of the conduction level, is ca.-0.2 eV then AG,(e-) z -0.8 eV. Thus the maximum in fig. 1 would correspond to AGe z -0.7 eV, which on correction for loss of translational motion of e; gives AGe' z -0.95 eV, i.e. i, z 0.95 eV. For p-benzoquinone AGe' z - 3.3 eV, from which eqn (1 I ) yields E, z 140 kJmol-l, which is to be compared with the experimental value of 15.4 kJ mol-l. This difference in E, corresponds to a difference in reaction rate of ca. loz2. The form of the dependence of kact on AGe indicated in fig. 1 is of the general form predicted by quantum-mechanical theories of electron transfer reactions [see for example ref.(34) and (36) and the references quoted therein]. The asymmetry in this dependence is in keeping with high-frequency intramolecular vibrational modes of the system being important in determining the electron-transfer rate, with significant displacement of the potential-energy surfaces of these modes in going from reactants to products. The lack of an entirely consistent trend in the dependence of logk,,, on AGe is to be expected since the molecular parameters determining this dependence certainly vary from solute to solute. However, as can be seen in fig. 1, these individual variations are not such as to obscure the general trend.A . A . H . Kadhum and G. A . Salmon 2529 It is possible that factors other than quantum effects contribute to the asymmetry in the exergonic region34 and in this respect it is useful to compare the present data with the data of Miller et al.35 on intramolecular electron transfer within bifunctional steroid molecules where non-quantum contributions to the asymmetry are likely to be less important.In both cases the decline in the rate on the exergonic branch of the curve is approximately 30 fold, which in the intramolecular electron-transfer case occurs over a change in AG of ca. 1.2 eV, whereas in the present data it occurs over ca. 2.0 eV. It is notable that the solute showing the largest deviation from the general trends of fig. 1 is p-tetrachlorobenzoquinone. For simple chlorinated aromatic compounds, such as chlorobenzene, electron addition results in dissociation of the carbon-halogen bond37 and it would, therefore, be reasonable to suppose that for p-tetrachlorobenzoquinone, although dissociation does not occur, considerable stretching of the C-C1 bond is likely to occur on addition of e-. Large displacements between the potential-energy surfaces of the reactants and products result36 in much increased asymmetry in the plot of log kact us.A G e and hence in a small effect of AG* on logk,,, in the exergonic region. An alternative explanation forp-tetrachlorobenzoquinone (1 8) and 2,2’-diphenylpicryl- hydrazyl (19) having larger values of kact than would be anticipated from the general trend of fig. 1 is that for these solutes reaction (1) involves the formation of an excited state of the electron adduct, thus reducing the free energy change on the reaction by the excitation energy of the electron adduct. These excitation energies are 0.93 and < 2.4 eV for 2,2’-diphenylpi~rylhydrazyl~~ and p-tetrachlorobenzoquinone,20 respectively, which would account for the effect.However, as indicated above no emission was observed in either case, thus rendering the hypothesis less plausible. A.A.H.K. acknowledges the award of a studentship by the Ministry of Higher Education of the Republic of Iraq. We also acknowledge support from S.E.R.C. for the provision of computing facilities and we thank Dr G. V. Buxton for valuable discussion and a referee for useful comments. References 1 E. J. Hart, The Hydrated Electron (John Wiley and Sons, London, 1970).2 M. Anbar, M. Bambenek and A. B. Ross, Selected SpeciJic Rates of Reaction of Transients from Water in Aqueous Solution. 1. Hydrated Electron. NSRDS-NBS 43 (U.S. Department of Commerce and the Natl Bur. Stand., Washington, 1973). 3 G. L. Bolton and G. R. Freeman, J. Am. Chem. Soc., 1976,98,6825. 4 A. M. Afanassiev, K. Okazaki and G. R. Freeman. J. Phys. Chem., 1979, 83, 1244. 5 J. M. Warman, in The Study of Fast Processes and Transient Species by Electron Pulse Radiolysis, ed. 6 R. A. Marcus, Adv. Chem. Ser., 1965, 50, 138. 7 R. A. Marcus, J. Chem. Phys., 1965, 10, 3477. 8 See ref. (l), p. 186. 9 A. 0. Allen, T. E. Gangwer and R. A. Holroyd, J. Phys. Chem., 1975, 79, 25. 10 A. 0. Allen, Nat. Bur. Stand. Rep. no. NSRDS-NBS 58 (Natl Bur. Stand., Washington, 1976).1 1 A. 0. Allen, T. E. Gangwer and R. A. Holroyd, J. Phys. Chem., 1975, 79, 25. 12 J. M. Warman, P. P. Infelta, M. P. de Haas and A. Hummel, Can. J. Chem., 1977, 55, 2249. 13 F. Y. Jou and L. M. Dorfman, J. Chem. Phys., 1973, 58, 4715. 14 F. S. Dainton, E. A. Robinson and G. A. Salmon, J . Phys. Chem., 1972, 76, 3897. 15 T. J. Kemp, J. P. Roberts, G. A. Salmon and G. F. Thompson, J. Phys. Chem., 1968, 72, 1464. 16 D. H. Ellison, G. A. Salmon and F. Wilkinson, Proc. R. SOC. London, Ser. A , 1972, 328, 23. 17 G. V. Buxton, F. C. R. Cattell and F. S. Dainton, Trans. Faraday SOC., 1971, 67, 687. 18 A. A. H. Kadhum and G. A. Salmon, Radiat. Phys. Chem., 1984. 23, 67. 19 A. E. Markham and K. A. Kobe, J. Am. Chem. Soc., 1941, 63,449. 20 J. V. Beitz and J. R. Miller, J. Chem. Phys., 1979, 71, 4579. 21 E. C. M. Chen and W. E. Wentworth, J. Chem. Phys., 1975,63, 3183. 22 J. M. Younkin, L. J. Smith and R. N. Compton, Theoret. Chim. Acta, 1976, 41, 157. 23 B. J. Tabner and J. R. Yandle, Reaction of Molecules at Electrodes, ed. N . S. Hush (Wiley-Interscience, J. H. Baxendale and F. Busi (D. Reidel, Dordrecht, 1982), p. 495. London, 1971), p. 289.2530 Solvated Electrons in Tetrahydrofuran 24 J. R. Langan and G. A. Salmon, J . Chem. Soc., Faraday Trans. I , 1983,79, 589. 25 M . Swarc, Carbanions, Living Polymer and Electron Transfer Process (Wiley-Interscience, New York, 26 B. Bockrath and L. M. Dorfman, J . Phys. Chem., 1973, 77, 1002. 27 N. T. Ioffe, A. J. Prokof’ev, S. P. Solodovnikv, A. A. Vodod’kim, G. A. Nikiforov and V. V. Ershov, 28 C . K. Mann and K. K. Barnes, Electrochemical Reactions in Non-aqueous Systems (Marcel Dekker, 29 C. Kilner and G. A. Salmon, to be published. 30 F. J. Davis, R. N. Compton and D. R. Nelson, J . Chem. Phys., 1973, 59, 2324. 31 J. A. Delair, M. 0. Delcourt and J. Belloni, J . Phys. Chem., 1980, 84, 1 186. 32 S. R. Logan, J . Chem. Sue., Faraday Trans. I , 1977, 73, 592. 33 J. P. Dodolet and G. R. Freeman, Can. J . Chem., 1975, 53, 1263. 34 P. Siders and R. A. Marcus, J. Am. Chem. Soc., 1981, 103, 748. 35 J. R. Miller, L. T. Calcaterra and G. L. Closs, J. Am. Chem. Soc., 1984, 106, 3047. 36 J. Ulstrup and J. Jortner, J . Chem. Phys., 1975, 63, 4358. 37 D. Razem and I. Dvornik, J. Phys. Chem., 1980,84, 3577. 38 G. A. Salmon, unpublished results. 1968). Izv. Akad. Nauk SSSR, Ser. Khim., 1971, 2844. New York, 1970)’ p. 27. Paper 51 1743 ; Received 8th October, 1985

ISSN:0300-9599    

DOI:10.1039/F19868202521

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I, 1986, 82, 2531-2545 On the Fundamental Concepts underlying Henry-law Adsorption and Adsorbed Gas Transport in Porous Solids John H. Petropoulos* Physical Chemistry Laboratory, Democritos Nuclear Research Center, Aghia Paraskevi, Athens, Greece Vassiliki I. Havredaki Physical Chemistry Laboratory, University of Athens, Navarinou Street, Athens, Greece The basic concepts of the theories of adsorption and adsorbable gas flow (with particular emphasis on the latter) are examined critically in the light of (i) model calculations supplementing previous results of Nicholson and Petropoulos and (ii) new data on the permeability of a series of gases through a mesoporous colloidal graphite compact. The results of (i) indicate that the usual concepts of a constant surface diffusion coefficient and corresponding activation energy (E,) are tenable, at least approximately, for sufficiently strong adsorption and narrow pores; but, even in this case, the parameters in question lack the physical meaning conventionally attributed to them.These results lead to notable new physical insights, particularly in connection with the relation between the values of E, and of the energy of adsorption observed experimentally. The data of (ii) confirm the breakdown of the conventional surface flow theory in the weaker adsorption region, in accord with the predictions of the ab initio calculation approach of Nicholson and Petropoulos. The effect of pore size predicted by the said approach also helps to explain the difference between the flow behaviour reported here and that observed previously on similar graphite compacts of smaller pore size.The theories of gas adsorption and flow in porous adsorbents are based on the concept of the Gibbs excess concentration and a comparable concept of ‘excess flow’ (referred to as ‘surface flow’, or more recently as ‘extra flow’), respectively.132 By means of ah initio calculations of gas flow in model pores (or model porous media) in the Henry law adsorption and Knudsen diffusion regimes, Nicholson and Petropoulo~~-~ examined critically the theory based on the latter concept with the following results. In the stronger adsorption region (higher gas adsorbability, low temperature) the fundamental functional dependence of the ‘excess flow’ on the adsorption field strength and the temperature, predicted by the above conventional surface flow theory (in spite of its shortcomings), is confirmed.The significance of other important concepts of this theory, namely that of the surface diffusion coefficient and the activation energy of surface diffusion, can similarly be examined by complementing the flow results of Nicholson and Petropoulos with the corresponding adsorption calculations. This task is undertaken here. In the low-adsorption region (low gas adsorbability, high temperature) negative excess flows were found. Hence, conventional surface flow theory breaks down completely (since the relevant surface diffusion coefficient is negative). An inversion of the normal tendency of the ‘reduced permeability’ P d ( M / T ) (where P is the gas permeability, M the molecular weight of the gas and T the temperature) to decrease with diminishing gas 253 12532 Gaseous Adsorption and Flow in Porous Adsorbents adsorbability or increasing temperature was also predicted. These theoretical results yielded a natural explanation of earlier (originally misinterpreted) observations7 of such inversion phenomena in ‘ Vycor ’ porous glass at elevated temperatures (which have lately been confirmeds).Similar phenomena have been observed by us9 at ambient temperatures in silica and alumina porous diaphragms. These experimental observations notwithstanding, inversion phenomena of the type described above are still a rarity and had not been detected in our previous extensive measurements of the flow of weakly adsorbed gases in microporous colloidal graphite compacts.1° The results of Nicholson and Petropoulos4-6 suggest, however, that for sufficiently narrow or constricted pores, even the adsorbability of helium may be too high for an inversion phenomenon to appear at ambient temperatures.If so, the effect in question may be expected at larger effective pore sizes. In the present paper, we report gas permeability data for a mesoporous colloidal graphite compact, which strikingly confirm this expectation. Adsorption and Diffusion in the Stronger Adsorption Region The theoretical description of adsorption and flow referred to in the previous section is given by:’$ 2 y lo C = EC,+AC, = ( E + Ak,) C, (1) and P = P g + P , = ~ D g + A k , D , respectively, where C is the concentration (per unit volume of the porous medium) of adsorbed gas in equilibrium with a gas-phase concentration Cg; C, is the corresponding excess Gibbs concentration or concentration of adsorbed gas (per unit surface area) and k , is the Henry adsorption constant; E is the porosity and A the specific surface area (per unit volume of porous medium) accessible to gas; Pg is the (expected) permeability of the same gas in the absence of adsorption (‘calibration gas’) and Ps is the observed ‘excess permeability’ (conventionally attributed to flow of the adsorbed gas molecules); D,, D , are the corresponding gas-phase and surface diffusion coefficients.D, can be evaluated for model pores of simple geometrical shape on the basis of a suitable assumption concerning the nature of the scattering of the gas molecules at the pore walls.Normally, scattering is assumed to be perfectly diffuse and this postulate is maintained in the work of Nicholson and Petropo~los.~-~ On this basis, we where R is the gas constant; rh is the hydraulic radius of the model pore set equal to that of the porous medium (&/A); u1 is the one-dimensional mean gas molecular speed; k , is a geometrical factor depending on the cross-sectional shape and length to radius ratio of the model pore ( k , = 16/3 for a long cylinder); and I C ~ is another geometrical or structure factor incorporating all deviations of real pore structure from that of the model pore. Conventional Surface Flow Approach To evaluate D,, the adsorbed gas is conventionally regarded as a dilute monolayer on the pore wall surfaces and surface flow as a two-dimensional random walk of mean step length A,.Because of its atomic structure, the pore wall surface (even when free of defects) is energetically heterogeneous. The admolecules are preferentially located at ‘ adsorption sites ’ (molar adsorption interaction energy E*), wherein they are laterally confined by potential barriers of height El (per mol).J. H. Petropoulos and V. I. Havredaki 2533 Only admolecules possessing activation energy E, = El (i.e. adsorption interaction energy E*-El) contribute to surface flow. Hence eqn ( 2 ) may be conveniently rewritten as: where (b differs from P d ( M / T ) only by a constant and k$ and Di refer to activated admolec~les.~ A model energetically heterogeneous surface comprising one type of adsorption site (E* = E: = constant) and one type of activated admolecule site (Et = E i = constant) is considered (e and Ei may be regarded as representative averages of the variable E* and El of imperfect real surfaces). In eqn (4), Di is evaluated from3~l1 0: = K , k , ?,, B, = K , k , a, o1 where a, is the distance between neighbouring adsorption sites, k , is a geometrical factor dependent on the nature of the surface atom latticell and K , is a structure factor analogous to K~ (except for the fact that it is also affected by gas adsorbability and temperature1,).The evaluation of k , and kg is based on suitable approximations to the adsorption potential function, which may be represented as U = Uo(x, y)F(z), where x, y and z denote the spatial coordinates parallel and normal to the surface, respectively.13 F(z) describes the shape of the adsorption potential well, which may be represented as a 9:3 p ~ t e n t i a l , ~ .~ ~ i.e. (assuming an isolated surface, cf. line D of fig. 1): F(z) = 2.598 [ (:y - (3y] where zo represents the distance of closest approach of a gas molecule to the surface layer of atoms of the solid. Uo(x,y) is the depth of the adsorption potential energy well (in J mol-l), which varies periodically along the surface between E: and E: - Ei. The precise form of the Uo(x, y ) function is of minor importance for the evaluation of k,, ki. Hence, we adopt here the simplified square-wave representation of Steele and Halsey,15 according to which Uo here takes two values only, namely Uoi = Uol = Et or Uoi = Uoz = E,* - Ef .Eqn (6) does not yield analytical expressions for k, and kg ; but for Uo 9 R T the simple harmonic oscillator approximation leads to uo i (z)' RT exp - 81 uoi kSi = - exp - = 0.5793 - vZi R T (7) where kSi = k,, (adsorption site) or kSi = k,, = kg (activated admolecule); and the second version of eqn (7) is derived by evaluating the simple harmonic oscillation frequency vZi on the basis of eqn (6).16 It is usually considered that k , = k,, . Strictly speaking, however, one should write15 (8) where y , is the fraction of A covered by adsorption sites. On the basis of k , z k,, and eqn (4), ( 5 ) and (7), we find ks = Y s ks, + (1 - r,) kB kt Dt K , k2 a, 8, v,, UO2 - Uol RT D , = U - - k , vz2 )+ exp (-3) RT where K,, is constant for constant EQ/E,*.2534 Gaseous Adsorption and Flow in Porous Adsorbents -0.2 -0.4 -0.6 -0.8 -1.0 O t - - - - - -1.2, D I I I I I I -1.2 r O t I -0.2 c 8:: -0.4 c ; -0.8 -1.0 Fig.1. (a) 9:3 adsorption potential in slit-shaped pores, Ui(za d z d zM), for zM/z, = 2.25 (A), 2.85 (B), 3.50 (C), cc (D), with z, = zo (W) or z, = zmin (W’); (b) model triangular adsorption potential in slit-shaped pores Ui(O 6 z d zM) for zM/uM = (A) and zM/aM = 2 (B). Ab-initio Calculation of @s Approach The ab-initio calculations of the reduced excess flow 4, reported by Nicholson and Petropo~los~-~ refer to model pores with energetically homogeneous (structureless) wall surfaces. Hence, in terms of the treatment of the preceding subsection, all admolecules therein must be considered as activated and the corresponding Henry law adsorption constant should be equivalent to ki here. Furthermore, the calculated ds values are not significantly affected by the introduction of the lateral periodic potential U,(x, J J ) .~ It is therefore evident that the aforesaid Qs calculations, in conjunction with the corresponding ki ones reported below, provide a method of evaluating Di, which circumvents all the assumptions of conventional theory leading to eqn (5). One may similarly deduce D, by evaluating k, for the model heterogeneous surface defined above. It is thus possible to establish if the concept of surface diffusion coefficient is tenable, at least formally. The requisite kSi may be obtained from and k, from eqn (8).In eqn (lo), z is the distance of the gas molecule from the pore wall with z = z, or z = zM for a gas molecule in contact with the pore wall or at the centre of the pore, respectively. The effective radius of the pore is thus r = zM - z,. Nicholson and Petropoulos6 found that the behaviour of 4, for two-dimensional slit, or cylindrical, model pores was very similar. Accordingly, here we confine attention to the former, for which more accurate and extensive data are available. Ui(z) in slit-shaped pores was represented as follows :3-5J. H. Petropoulos and V. I . Havredaki 2535 (a) 9 : 3 Potential [cf. jig. 1 (a)] Here eqn (6) is used with additional terms (due to the interaction of the gas molecule with both walls of the slit), namely where Uoi is the depth of the adsorption potential-energy well when zM -+ co.For finite zM the depth of the adsorption potential well Umi is higher: U(zo < z < zM) = - Umi at z = zmin(Umi 2 Uoi). Normally, in eqn (lo), z, = z,,; but, in ref. ( 5 ) and (6), it was mathematically convenient to set z, = zmin. As shown later, the choice of z, makes no difference to the adsorption behaviour of interest here. (b) Model Triangular Adsorption Potential [cf. jig. 1 (b)] Here, Ui(z) = - UOi(l -z/a,) for 0 < z < aM, if zM > uM; or for 0 < z < zM, if zM < a,; and Ui(Z) = 0 (1 2 4 for aM < z < zM; if zM > aM. In eqn (12), a, is the maximum width (at the top) of the triangular potential well at an isolated surface; Uoi is the depth of the well (independent of zM, i.e.Umi = Uoi); and z, = 0. Physically, this potential is artificial (it corresponds to a constant force emanating from the pore wall and vanishing abruptly at z = aM, if zM < u,; or at z = zM, if zM ,< aM), but it is mathematically simple [it leads to an analytical result for eqn (lo)] and also provides useful insight into the effect of the functional form of Vi(z). For the type of model pores considered here r = Yh. Furthermore, it is convenient to define a dimensionless radius R = r/a (where a = zo or a = aM in the case of the 9 : 3 and triangular adsorption potentials, respectively) and a dimensionless adsorption potential energy 0 = U/RT. Results and Discussion Adsorption Parameters Examples of the results of model calculations of kSi and k, based on the treatment described above are given in fig.2 and 3. The reported k, values refer to a model energetically heterogeneous surface described by eqn (8) with y, = 4 and Uoz/Uol = 0.60 corresponding to Et/E: = 0.40. Lines A, B and C in fig. 2 illustrate the effect of different choices of adsorption potential on the In kSi us. Voi( = Uoi/RT) plot for an isolated surface. In the region Voi M 2-8, these plots can be represented to a reasonable approximation by straight, nearly parallel lines [cf. ref. (16)] given by where a, M 0.82-0.86. Fig. 2 and 3 further show that kSi for narrow pores or k, for energetically heterogeneous pore surfaces behave similarly, except for differences in the value of a,. As the pore radius decreases, a, tends to increase (a, M 0.87 for line H, and a, M 0.92 for line I, in fig.3), although this tendency is markedly reduced if kSi is made a function of umi (a, z 0.85 for line I' in fig. 3). Fig. 2 shows that the plots of Ink, us. uol tend to be appreciably less steep (a, M 0.77 for line D; a, M 0.74 for line E) than the corresponding In k,, us. ool plots (represented by lines A, a, M 0.85, and B, a, M 0.82, respectively). One may similarly fit eqn (13) to the kSi values obtained from eqn (7), or the k , values obtained from eqn (7) and (8), in the range of Uoi(Uol) M 2-8 (cf. fig. 2). kSi = kti exp (a, Uoi/RT) (13)2536 Gaseous Adsorption and Flow in Porous Adsorbents 0 2 4 6 8 10 uoi Fig. 2. Adsorption constants kSi (lines A, B, C and F) and related k , of eqn (8) (lines E, D and G) for isolated flat surfaces (zM + 00) calculated by (i) the conventional treatment of eqn (7) (lines F and G), or (ii) the present treatment of eqn (lo), in conjunction with either eqn (1 1) [z, = zo (C), z, = zmin (lines B and E)] or eqn (12) and (12a) (lines A and D).The resulting a, are perceptibly higher than those recorded for isolated surfaces above (a, z 0.90 for line F; a, z 0.81 for line G), but the adsorption behaviour predicted by eqn (7) and (10) is otherwise very similar at VOi > 2 [cf. ref. (16) and note that the significant discrepancies referred to in ref. (14) probably arise from incorrect evaluation It is worth noting that conformity of k , to eqn (1 3) is equivalent to a constant energy (14) of VZi]. of adsorption AE, = -a, Uol = - a, E,*, since k , = k: exp (- AE,/RT). The results for a, reported above indicate that the values of -AE, predicted by the treatment of eqn (10) are somewhat lower than those given by the conventional treatment, which in turn tend to be somewhat lower than E,*(or Umi in the case of narrow pores) in the E,*/RT range of chief practical interest. Surface Diffusion Parameters The higher values of the reduced excess flow 4, (namely ds 2 3) calculated by Nicholson and Petropoulos6 can be represented to a reasonable approximation by analytical expressions analogous to that for kSi (k,) given above:17 4, = KO RWm exp (aU,,/RT) (15)J.H. Petropoulos and V. I. Havredaki 0 6 4 h u -k Y r: - 2 0 2537 -2L I I t I 1 L I I 0 2 4 6 8 o o i Fig. 3. Adsorption constants kSi for isolated surfaces (lines A and B) or narrow pores (lines H, I and 1’), with 9:3 (za = zmin, lines B, I and 1’) or model triangular (lines A and H) adsorption potential, as a function of Uoi (lines A, B, I and H) or Umi (line 1’).(Note that Umi = DOi for lines A, B and H.) where KO, rn and a are empirical constants. Data in the appropriate range were obtained for R = 0.5-2.5 (pore length to radius ratio = 20) and R = 1.05-2.30 (pore length to radius ratio z 10) for the triangular and 9:3 (2, = zmin) adsorption potentials, respectively, yielding a = 0.78-0.79, rn z 2.1, KO z 0.045 (triangular potential); or a z 0.79-0.82, rn z 2.8-3.0, KO z 0.15 (9: 3 potential). From eqn (3), (4), (1 3) and ( 1 5 ) an approximate analytical expression for Di for the model pore ( K , = 1) can be deduced: which, in the case of the triangular potential, reduces, to a first approximation (rn z 2, a z a,) to the form of eqn ( 5 ) : (164 Note, however, that the constant factor in eqn (16a) is not determined by structural characteristics of the surface as implied by eqn (5).In fact, as suggested by eqn (16) (since a < a,) and shown in fig. 4, there is a noticeable downward drift of Di with Dg z (KO k , a2/k:,) 0,. 84 FAR 12538 Gaseous Adsorption and Flow in Porous Adsorbents 0 A A J I I I I 1 I 3 4 5 6 7 8 9 6 0 2 Fig. 4. Surface diffusion coefficients D, and corresponding Di calculated from the numerical data of ref. (5) and (6) and of the present work; 9:3 adsorption potential (za = zmin): R = 1.05 (a), 1.65 (M), 2.30 (A); model triangular adsorption potential: - R = 1 (O), 1.5 (0). The slope of a D, plot corresponding to E, = E6( = iU,,,) is indicated by line A.increasing Uoz. The more marked downward trend of Dg (with decreasing vo2) in the lower uO2 region is simply due to breakdown of eqn (15) (& < 3), when uo2 and R are not sufficiently high and low, respectively. Fig. 4 further shows that the behaviour of 0: in the case of the 9:3 potential is not materially different. The lack of a reasonably clear-cut analytical approximate result in this case is no doubt attributable to the significant dependence of Ui(z) on R in the range R = 1.05-2.30 (corresponding to zM = 2.25-3.50) used [cf. fig. l(a)], in contrast to the complete lack of such dependence when R 2 1, in the case of the triangular potential [cf.fig. 1 (b)]. One may similarly evaluate D, for the model pore with energetically heterogeneous pore wall surface, to obtain Comparison of eqn (1 7) with eqn (9) shows that the activation energy of surface diffusion E, (which is conventionally identified with Ei) is here given by Since a, > a (note that % must be determined in the uol range corresponding to the Uo2 region used in the determination of a, hence the x, values applicable here are significantly higher than those previously deduced from fig. 2 and 3), E, can exceed E6 appreciably, as indicated by the slopes of the D, plots of fig. 4. In practice, the chief quantity of interest is the ratio -E,/AE, which, according to the conventional treatment of eqn (7) and (9), should not differ materially from Ei/E,*.However, the experimental values of -E,/AE, typically quoted1* 2 , lo are of the order of 0.6 or higher, whereas values of E i / E t obtained by model calculations are markedly lower (e.g. 5 0.3).16, As shown above, the present treatment predicts higher E, and E, = aE;+(a,-a)E,*. (18)J . H . Petropoulos and V. I. Havredaki A 'p 2539 I I I I 1 I I I 1 I 2 4 6 8 10 12 In ( k , / a ) or In (kilo) Fig. 5. Surface diffusion coefficients in relation to the corresponding k,; D,, k , (0, 0, 0 , W, A) were calculated exactly as for fig. 4; for I l k , ki (0, 0, A) a calibration gas of finite adsorbability (UO2 = 0.5) was assumed; 9:3 adsorption potential: R = 1.05 (a), 1.65 (m), 2.30 (A); model triangular adsorption potential: R = I (0, S), R = 1.5 (0, I), R = 2.5 (A); - E,/AE, (deduced from the linear parts of the plots) E 0.49 (0, n), 0.52 (a), 0.50 (a, A), 0.49 (8, U, A).Note that the DL plots have been displaced downwards for clarity. lower - AEs (especially when energetic surface heterogeneity is taken into account) than the conventional one and thus helps partly to explain this discrepancy. The actual predicted values of -E,/AE, for the model pores and the model energetically hetero- geneous surface assumed here can be determined directly from the slope of plots of the calculated In [(D,/a) (M/RT)$] us. In @,/a), since eqn (1 4), ( I 7) and (1 8) yield (19) ES ES AES A E S In [(D,/a) (MIRT);] = In KL -- In ki +- In k , where Kb is a constant. The relevant plots (fig.5 ) show good linearity in the high k , region with slopes which are remarkably insensitive to the value of R or the form of the adsorption potential [the only perceptible difference is found in the case of the narrowest pore with 9: 3 potential where, as shown by line (A) of fig. 1 (a), the adsorption force fields emanating from opposite pore walls overlap extensively]. The resulting values of -E,/AE, exceed the value of Ei/E; assumed here by ca. 20%. Fig. 5 also includes examples from a further series of calculations intended to simulate the fact that, in practice, q5s is of necessity determined in relation to a calibration gas of finite adsorbability (usually He). Values of E,* as high as 2.5 kJ mol-1 have been quoted for He. l5 Accordingly, surface diffusion coefficients DI; were calculated assuming a calibration gas with oo2 = (E,* -Ei)/RT = 0.5 (corresponding to E,* = 2 kJ molkl at T = 300 K or E,* = 2.5 kJ mol-lat T z 375 K).Theresultingplots(fig. 5)attainlinearityat substantially lower k, values, but the slope of the linear region is essentially unchanged.A further factor which may affect the experimental value of -Es/AEs is K , , which should appear in eqn (16) or (17) when applied to porous media. As pointed out previously, K , is not independent of uoi. Its effect can be studied by means of suitable model calculations, such as those of Nicholson and Petropoulos,6 who constructed a series of simple model porous media consisting of parallel (P) or serial (S) arrays of model pores of two different radii (Rl, R2) combined in various proportions (number fractions f i , f i ) .Calculations of bS then showed that its dependence on Uo2 (in the appropriate 84-22540 Gaseous Adsorption and Flow in Porous Adsorhents -2 -’L -5 t ‘& O O n A L I I I 4 6 8 10 In (k,lQ) Fig. 6. Surface diffusion coefficients in relation to the corresponding k , for model porous media consisting of parallel (A, a) or serial (A, 0, 0) arrays of narrow (R, = 1) and wide (R, = 5) model pores with triangular adsorption potential.6 Number fraction of narrow pores,f, = 0.9 (a, A), 0.5 (0, a), 0.1 (0). The linear parts of the serial array pore plots yield -E,/AE, 0.49 (A), 0.56 (01, 0.64 (W. range) could still be described by eqn (15); the value of a was unaffected by P-type structure, but could be considerably lower, in the VO2 range of practical interest, in the case of S-type (constricted pore) structure.Where k, is concerned, the arrangement of the pores is obviously immaterial and the value of a, will not be perturbed significantly, unless very narrow pores are present. Hence, in accordance with eqn (18), S-type pore structure should tend to enhance the value of - E,/AE,. This is confirmed by the relevant plots, shown in fig. 6, which exhibit good linearity in the higher U,,, range. Fig. 6 further shows that the linear region of the plot becomes both more extensive and steeper, as the proportion (fi) of narrow sections (i.e. constrictions) in the serial pore array becomes lower; for fi = 0.1, the value of -Es/AEs exceeds that characteristic of single pores (fig.5 ) by over 25%. It is also important to note that axially non-uniform porosity (expected to be present in the vast majority of porous diaphragms studied in practice as a result of the pelletization process)19 has been shown20 to have the same effect as S-type pore structure. We conclude that the discrepancy between experimental - Es/AEs values and Ei/E$ determined by model calculations can be attributed in part to oversimplifications in the conventional surface flow approach and in part to the effect of the pore structure and macroscopic non-homogeneity of real porous media. Another question of considerable practical importance concerns the value of E,/AE, for a series of gases in a given porous solid. If Ei/E$ = constant within the series, the results presented in fig.5 and 6 indicate that E,/AE, = constant only if the relevant k, range falls wholly within the linear part of the log [D,/(M/RT)&] us. log k, plot pertaining to the effective mean pore size and pore structure of the porous medium concerned. If the lower end of the k, range in question extends into the curved part of the said plot, a tendency for -E,/AE, to decrease in the case of the lighter gases is to be expected. Detailed numerical data for k, and D, (using He as calibration gas) suitable for applying the above ideas are given in tables 11.4 and 11.5 of ref. (2). The resulting plots are shown in fig. 7. The heavier gas data for ‘Carbolac’ diaphragms (including such diverse gases as Ar, CH, and CO) are clustered remarkably closely around a straightJ .H . Petropoulos and V . I . Havredaki 254 1 --3.5 \ ‘\ 2.5 - Fig, 7. Correlation based on eqn (19) between D, (in cm2 s-l) and k, (in nm) measured by Ash et a1.2 for a variety of gases, temperatures and porous diaphragms. Membrane L (‘Carbolac’ carbon): lower open points; Membrane M (‘Carbolac’ carbon): filled points; Membrane N (‘Graphon’ graphitized carbon): upper open points; H 2 : u ; D2:(); N,: 0; Ar: 0; Kr: A; Xe:D; CH,: V; CO: Q. line corresponding to -E,/AE, z 0.56. The points for the lightest gases (H, and D,) fall well below this line and yield -E,/AE, z 0.34 and 0.40, respectively. The difference between the aforementioned E,/AE, values and those quoted in ref. (2) is no doubt largely due to our definition of E, according to the Eying transition-state theory and our method of direct evaluation of EJAE,, as opposed to the Arrhenius definition of E, and the individual evaluation of E,, AE, used in ref.(2). In the case of the ‘Graphon’ diaphragm, data for four gases only are available, but the general pattern of behaviour, at least as far as defined by the noble gases, does not appear to be different. The straight line defined by the points in the high k , region corresponds to -E,/AE, = 0.46 as compared to -E,/AE, z 0.35 for Ar. Hence, in this case, the diminution in -E,/AE, is noticeable at k, values higher than before, in keeping with the larger pore size of the ‘Graphon’ (rh z 5 nm) as compared with the ‘Carbolac’ (rh z 0.5 nm) diaphragm. Thus, the experimental behaviour revealed by fig.7 is in at least qualitative accord with the theoretical results of fig. 5 and 6. On the basis of the conventional treatment, on the other hand, any variation in -E,/AE, must be interpreted in terms of a corresponding variation in Ei/E,*. Model calculations on smooth surfaces, however, show that Ei/E,* should tend to decrease for the bulkier gas molecules,16 which is the reverse of what is observed here. Before closing this section some comment is called for in connection with the considerable shift of the N, points in fig. 7 for the ‘Graphon’ diaphragm relative to those of Ar. This is presumably attributable to the effect of specific gas-solid interactions (quadrupole interactions in the case of N2). It is noteworthy that analogous effects are not visible in the case of the ‘Carbolac’ data.Clearly, this interesting point can be pursued further only in the light of more data and of suitable elaboration of the treat- ment of Nicholson and Petropo~los.~2542 Gaseous Adsorption and Flow in Porous Adsorbents Adsorption and Diffusion in the Weak Adsorption Region In the weak adsorption region, the usefulness of the ‘excess parameters’ k, and D, becomes increasingly questionable from the point of view of physical meaningfulness (although the problem of reliable experimental determination is also not unimportant). The deviations of the plots of fig. 2,3 and 4 from linearity at low ool or O,,, respectively, cause AE, and E, to lose their usual physical meaning. The occurrence of negative excess flow at still lower oO2, which renders D, meaningless, has already been noted in the introductory section.The occurrence of negative Gibbs excess concentrations is also noteworthy.21 Accordingly, we adopt here a formulation of adsorption and flow akin to that of Nicholson and Petropoul~s,~- who reported their calcuated flow results in the form of In 4 us. OO2 plots. As previously pointed out, the former parameter differs only by a constant factor from the experimental ‘reduced permeability’ P d ( M / T ) . Do, is less convenient from this point of view. Accordingly, it is replaced here by the logarithm of a ‘reduced sorption coefficient’ k which, using eqn (l), is given by k = C/EC, = 1 + Ak,/c = 1 + ks/rh and is, therefore, easily deduced here from the k, values previously calculated. At the same time, k differs from the sorption coefficient determined experimentally, S = C/C,, only by a constant factor, namely E ; if E is determined by helium pyknometry, then k = S/SHe.One may similarly normalise P d ( M / T ) with respect to He to obtain The advantage of this formulation is that the finite adsorbability of the ‘ normalizing gas ’ merely shifts the relevant plot parallel to the axes without changing its form. The computation results of Nicholson and Petropoulos for slit-shaped pores (together with an example for a model porous medium),^^ have been plotted in this manner in fig. 8 for the cases of EQ/Et = 0 or 0.4. Experiment a1 A mesoporous diaphragm of cross-sectional area 0.795 cm2, thickness 3.37, cm and porosity 0.362 was constructed by compaction of very fine graphite powder (Acheson DAG 621 dispersion powder of specific surface area 72 m2 g-l and ash content 0.29%) in a cylindrical metal holder.To minimize non-uniformity of packing density, the powder was added to the holder in 15 small portions and compressed each time symmetrically by means of close-fitting piston^.^ The effective hydraulic radius was rh = &/A = 3.6 nm. The gases used were of > 99.9% purity. Permeation measurements were carried out by attaching the specially designed diaphragm holder directly through soft metal gaskets to a permeation apparatus incorporating a mercury manometer and manostatic device on the upstream side and a McLeod gauge on the downstream side.g The diaphragm was first thoroughly outgassed at temperatures up to 470 K.After completion of the permeation runs, the diaphragm holder was quickly transferred to a sorption bulb, which was attached to a volumetric sorption apparatus. Special metal attachments were used to minimize the dead volume of the sorption bulb. Prior to the adsorption measurements, the diaphragm was subjected to a second, thorough, but considerably milder, outgassing treatment. Results and Discussion Adsorption and permeation measurements were carried out for He, H,, N,, Ar, Kr, CH,, C2H,, and CO, in the pressure range 30-120 Torr.? Henry’s law was obeyed throughout this range for all gases, except the last two at the high-pressure end. The t 1 Torr = 101 325/760 Pa.J. H . Petropoulos and V. I. Havredaki 2543 1 0.5 8 S d 0 -c.5 I I I I 1 I 0 2 4 6 In k or In k, 1 0.5 8 s - 0 -0.5 I I I I 1 1 0 2 4 6 In k or In k 2 Fig.8. Reduced permeabilities q!~ calculated for single model pores (or a model porous medium)jT in relation to the corresponding reduced sorption coefficient k (assuming Ei/E: = 0.4; full lines) or k , (Ei/E,* = 0; broken lines): (a) slits with 9: 3 potential (z, = zmin) and R = 1.05 (A), 2.30 (B), 4.80 (C), 9.80 (D), 24.80 (E); (b) slits with model triangular potential and R = 1 (A), 3(B), 10 (C), 25 (D); or a serial array (with R , = 1, R, = 5 andf, = 0.5) of effective dimensionless hydraulic radius R = 3 (B’). corresponding permeabilities were constant (and showed no definite tendency to vary, even outside the Henry law region in the case of C,H, and CO,). The results are plotted in fig.9(6) in the form proposed in the previous section but one, in comparison with those of Savvakis et al.1° on a series of compacts constructed from the same graphite powder, but of considerably smaller pore size [fig. 9(a)]. The present data exhibit an unmistakable minimum in 4 as gas adsorbability is varied, in conformity with the general pattern predicted by the Nicholson-Petropoulos approach (fig. 8). The lack of a similar pattern in the results of ref. (10) [fig. 9(a)] can now be explained on the basis of the effect of pore size. As shown in fig. 8, the minimum in 4 shifts to lower k as R is reduced, with the result that it will be unobservable in practice if it occurs at k < k,, at the temperature of observation. Pore constrictions have a similar effect [cf.lines B and B’ in fig. 8(6)]. The quantitative differences among the different diaphragms in fig. 9(a) are no doubt attributable to their diverse macroscopic structures;10 whereas the deviation of the data of fig. 9 (b) from a single smooth curve are presumably largely due to specific gas-solid interactions of the kind noted earlier in the ‘Graphon’ diaphragm of fig. 7.0,s - 0.4 z a 8 E . v - 0.2 0 2544 0 ?b 0 3 Gaseous Adsorption and Flow in Porous Adsorbents 8 4” 4 0.1 Lk h Lt $ 8 a E . v t 4 - 0 4 4 Fig. 9. Experimental reduced permeabilities in relation to the corresponding reduced sorption coefficients (normalized with respect to He in both cases) for a series of gases in a series of porous colloidal graphite compacts : (a) microporous compacts (&/A = 1 nm) C( b), D (Q), E (0-), F (a), G (o> of ref.(10) at T = 296.4 K (a), 323 K (A), 348 K (A); (b) mesoporous compact (&/A = 3.5 nm) of present work at T = 298.2 K (open points), 348.2 K (filled points). Gases used: N~(o), H, (o), ~r (A), N, (v), CH, (o), ~r (()I, co, (0, C,H, (a). Conclusion The theoretical results obtained here indicate that, on the basis of the Nicholson- Petropoulos approach, the basic concepts of conventional surface flow theory are formally valid to a first approximation in the stronger adsorption region when applied to sufficiently narrow pores. Thus, ‘excess flow’ is indeed governed by a ‘surface diffusion coefficient’ D,, approximately independent of pore size, and an activation energy of surface diffusion E,; although the value of D, does not reflect solely the structural characteristics of the pore wall surface and Es is not identical with the height of the lateral potential barrier Ei, as implied by conventional theory.The effect of pore structure on the value of E, has also been shown to be important, with the result that the total deviations E, < Ei and -AE, < E,* revealed by our model calculations go a long way towards explaining the discrepancies commonly found between - E,/AEs as determined experimentally and Ei/E,* as calculated for realistic smooth model surfaces. The correlation between D,2/(M/T) and k, proposed here in terms of eqn (19) has the virtue of yielding Es/AE, directly and represents a considerable advance over the empirical correlation between P z / ( M / T ) and k, given by Ash et a1.2 By plotting the data of the latter authors according to eqn (19), a clear tendency for -Es/AE, to diminish at low k , has been revealed, in accord with the Nicholson-Petropoulos treatment, but not with the conventional one.The experimental results reported in the present paper, on the other hand, provide further strong support for the Nicholson-Petropoulos approach, particularly with regardJ . H. Petropoulos and V. I . Havredaki 2545 to the predictions of this approach concerning the dependence of flow behaviour in the weak adsorption region on pore size. The present work was carried out at the Democritos Centre and is based in part on a Ph.D. Thesis presented by Ms V. I. Havredaki to the University of Athens. Thanks are due to Prof. Th. Yannakopoulos, Prof. R. M. Barrer and Dr D. Nicholson for useful discussions. References I e.g. R. M. Barrer, Appl. Muter. Res., 1963, 2, 129. 2 R. Ash, R. M. Barrer, J. H. Clint, R. J. Dolphin and C. L. Murray, Philos. Trans. R . Soc. London, 1973, 3 D. Nicholson and J. H. Petropoulos, J. Colloid Interface Sci., 1973, 45, 459. 4 D. Nicholson and J. H. Petropoulos, J. Colloid Interface Sci., 1979, 71, 570. 5 D. Nicholson and J. H. Petropoulos, J. Colloid Interface Sci., 1981, 83, 420. 6 D. Nicholson and J. H. Petropoulos, J. Colloid Interface Sci., 1985, 106, 538. 7 S. T. Hwang and K. Kammermeyer, Can J. Chem. Eng., 1966,44, 82. 8 Y. Shindo, T. Hakuta, H. Yoshitome and H. Inoue, J . Chem. Eng. .I@, 1983, 16, 120. 9 V. Havredaki and J. H. Petropoulos, J . Membrane Sci., 1983, 12, 303. 275, 255. 10 C. Savvakis, K. Tsimillis and J. H. Petropoulos, J . Chem. Soc., Faraday Trans. I , 1982, 78, 3121. 11 T. L. Hill, J. Chem. Phys., 1956, 25, 730. 12 D. Nicholson and J. H. Petropoulos, J . Phys. D, 1968, 1, 1379. 13 W. A. Steele in The Solid-Gas Interface, ed. E. A. Flood (Marcel Dekker, New York, 1967), vol. 1, 14 D. H. Everett and J. C. Powl, J. Chem. Soc., Faraday Trans. I, 1976, 72, 61. 15 W. A. Steele and G. D. Halsey Jr, J . Phys. Chem., 1955, 59, 57. 16 W. A. Steele, Interaction of Gases with Solid Surfaces (Pergamon Press, Oxford, 1974), chap. 2 and 3. 17 J. H. Petropoulos, J . Membr. Sci., in press. 18 e.g. F. Ricca, C. Pisani and E. Garrone, J . Chem. Phys., 1969, 51, 4079. 19 e.g. C. Sawakis and J. H. Petropoulos, J. Phys. Chem., 1982, 86, 5128. 20 J. H. Petropoulos, J . Chem. SOC., Faraday Trans. I , in press. 21 R. M. Barrer and J. H. Petropoulos, Surf: Sci., 1965, 3, 126. chap. 10. Paper 5 / 1754; Received 9th October, 1985

ISSN:0300-9599    

DOI:10.1039/F19868202531

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I, 1986, 82, 2547-2556 Thermodynamic Study of Organic Compounds in Octan- 1-01 Processes of Transfer from Gas and from Dilute Aqueous Solution? Paolo Berti, Sergio Cabani," Giovanni Conti and Vincenzo Mollica Dipartimento di Chimica e di Chimica Industriale, Universita degli Studi di Pisa, Via Risorgimento 35, 56100 Pisa, Italy Free energies and enthalpies of solvation of water and some hydrocarbons (hexane, cyclohexane), ethers (diethyl ether, tetrahydrofuran) and ketones (propanone, pentan-3-one, cyclopentanone) in octan- 1-01 have been deter- mined at 298.15 K from vapour-pressure measurements of dilute solutions and from limiting heats of solution. These solvation functions in octanol have been used together with the corresponding hydration functions in order to obtain water-octan- 1-01 partition coefficients and their dependence on temperature.A comparison is made with the practical partition coefficients relative to mutually saturated solvents. Many thermodynamic studies have been carried out these last two decades for the transfer process of organic molecules from ideal gas to aqueous so1ution.l The obtained standard thermodynamic functions of hydration, AXE (X = G, H , S , C,), were related to the molecular structure of the solutes through a scheme of group contributions. This procedure allowed some features arising from the water-functional group interactions to be classified according to the nature, number and position these groups occupy in the molecular structure of the so1utes.la Much attention has also been devoted to the transfer of organic and inorganic non-electrolytes from gas to dilute non-aqueous solution,2 but unfortunately a systematic study on the thermodynamics of solvation of organic molecules in non-aqueous solvents is hitherto lacking, although it presents many interesting aspects. In fact, a knowledge of the values of contributions of many organic functional groups to the thermodynamic functions of solvation in many different solvents could improve our possibilities of predicting and understanding of the role of the solvent on the following: chemical equilibria, kinetics of reaction, gas solubilities, partitioning of a solute between immiscible solvents etc.As part of a planned study on the thermodynamics of solvation of non-electrolytes in non-aqueous solvents, we report in this paper some results concerning the free energies, enthalpies and entropies of transfer of some small organic molecules from gas to dilute solution in octan-1-01.Free energies of solvation have been obtained from liquid-vapour equilibria and enthalpies of solvation have been obtained from calorimetric heats of solution and from known values of heats of vaporization. Finally, entropies of solvation have been calculated from the values of the above free energies and enthalpies. These functions have been used, together with the literature values of the corresponding thermodynamic properties of hydration,l a for calculating the thermodynamic functions of transfer of solutes from pure water to pure octanol. Many values of free energy for the transfer process from octanol-saturated water to water-saturated octanol are available from experimentally determined partition coef- ficients, Poct,w.3 These data are of pharmacological and environmental interest for their t Taken in part from a Ph.D.Thesis (P.B.), Universita di Pisa, 1985. 25472548 Study of Organic Compounds in Octan-1-01 correlation with drug activity4 and the solubility in water of liquid and solid non-ionic compo~nds.~ A few values of enthalpy of transfer have also been reported, obtained from studies of the dependence of log(Poct,w) on temperature,6 as well as from direct microcalorimetric measurements carried out by flow techniq~es.~-l~ Free energies, enthalpies and entropies of the transfer process, evaluated by considering separately the solution processes in water and in octanol, allow us to gain much more information on the causes driving the distribution.Moreover, the possibility of investigating both the distribution between the pure solvents and the distribution between reciprocally saturated phases is an important step in understanding the effects of the mutual water-octanol solubility on the partitioning of solutes. For this purpose some calorimetric measurements have also been carried out to study the differences between the enthalpies of solution in pure octanol and in octanol containing known quantities of water. Literature data have been used to examine this effect on the free energy of transfer. Experimental Materials All examined solutes were commercial products of the best grade available.The ketones (propanone, pentan-3-one, cyclopentanone) were dried over anhydrous calcium sulphate. The alkanes (hexane, cyclohexane) and the ethers (ethyl ether, tetrahydrofuran) were dried over calcium hydride. All compounds were then fractionally distilled in an inert atmosphere. The purity of the fractions collected at constant boiling point was checked by g.1.c. and was in all cases better than 99.8%. The water used in the experiments was first deionized and then distilled from alkaline KMnO,. The solvent, octan-1-01, was a Fluka product of purissimum grade ( > 99.5 % ). It was refluxed over metallic sodium and then distilled under nitrogen at reduced pressure. The samples used showed a g.1.c. purity better than 99.9% and a water content < 0.01 % , determined by Karl Fisher titration.Vapour Pressure Measurements The total vapour pressure, pt, of the solutions was measured at 298.15 K with a precision of 2.5 Pa by a static apparatus described in aprevious paper.ll Particular care was taken in deaerating the solutions by repeated freezings and meltings under vacuum. The partial pressure of solutes, P,, was then calculated using the relationship : P, = pt - (1 - X,) Po,,, where X , and Pact, are the solute mole fraction and the vapour pressure of pure octan- 1-01, respectively. Po,, was determined in separate experiments using octanol samples treated as described above. The measurements were made in the temperature range 293-303 K. At 298.15 K we found Poet = 14.5 Pa and its derivative with temperature (dPoct/dT)298 = 1.43 Pa K-l.These data are in good agreement with literature data.I2 The composition of the liquid phase in equilibrium with vapour was determined, according to the various solutes, by U.V. absorbtion or by density measurements of the mixtures. The density values were obtained using an Anton Paar digital vibrating density meter (DMA 602) with a precision better than 3 x lop3 kg m-3. Heats of Solution The heats of solution in octanol for the compounds considered were measured by an isoperibol calorimeter built for this purpose. The calorimetric cell was a glass bottle of ca. 100 cm3 volume, with a very reduced vapour space (< 1 cm3). It was contained in a copper cylindrical vessel within a sophisticated air circulation thermostatic bath, capable of maintaining the temperature within f 5 x K even for long periods of time.Successive additions of solutes were made through an external digital microburette having a total volume of 1 cm3 and a resolution of lop4 cm3. Before entering the cell,P. Berti et al. 2549 the liquids flowed through a heat exchanger constituted by a loop of PTFE tube wound around a massive brass cylinder situated beside the cell. A miniaturized relief valve at the end of the dispensing tube prevented diffusion of the solute in the solvent during intervals between successive runs. The temperature changes in the cell caused by addition of solutes were read by means of a thermistor (5000 R at 293 K) connected to a Wheatstone bridge. Signals from an unbalanced bridge, via an AD converter, were transferred to a AIM 65 Rockwell microcomputer for the collection and the elaboration of calorimetric data.The heat capacity of the system was obtained for each run by repeated electric calibrations. The stirring inside the calorimeter was maintained constant during the experiments. It was possible, however, to change the stirring speed in a wide range, using a stepping motor, in order to obtain rapid mixing with low noise. Repeated experiments allowed us to evaluate a A T resolution within 5 x lop5 K, corresponding to 1 x J on the heats, owing to the heat capacity of our system. The accuracy of the apparatus was tested by neutralization experiments using aqueous HC1 and NaOH solutions. For each compound at least 10 measurements were carried out in the concentration range 0.002-0.04 mole fraction.No appreciable dependence on concen- tration was observed for all the compounds examined and the uncertainty on the limiting heats of solution did not exceed 0.10 kJ mol-l. Results In table 1 are reported the thermodynamic quantities related to the isothermal transfer at 298.15 K of a mole of solute from the ideal-gas state to infinitely dilute solution. The free energy change is associated with the process: S (ideal gas, C, = 1 mol dm-3) + S (ideal solution in octan-1-01, Coct = 1 mol dm-3) where S is a generic solute, Cg and Coct are the molar concentrations of S in the two states. The standard free energy of solvation has been calculated by extrapolating at Xs = 0 the function : At least five different values of AGb,, have been determined for X s ranging from 0.01 to 0.15 for each solute examined.Fig. 1 and 2 show, in this concentration range, a linear relationship between AGict and Xs, which allowed us to obtain extrapolated AG$t values with an uncertainty of ca. 0.17 kJ mo1-l. The limiting value, AGZCt, corresponds to the transfer of a mole of solute from an ideal gaseous state at P = 1 atmt to a hypothetical ideal solution state in octan-1-01 at X s = 1. This quantity was then converted to the standard free energy of solvation, AG:ct, relative to process (1) by: (1) AG;,, = RT In (Ps/Xs). (2) AGEct = AGzct - RT In (RT) - RT In (Poct/M,,,). (3) A value of poet = 821.72 kg mV3 (determined by us) for the density of pure octanol at 298.15 K was used, in accordance with the value of 821.77 kg m-3 reported in ref.(1 3). In eqn (3) Moct represents the molecular weight of octan-1-01. The enthalpies of solvation, A g c t , have been calculated at 298.15 K using the relations hip : (4) where AKol, is the limiting heat of solution in octanol and AWvap is the enthalpy of vaporization. These latter values have been taken, except for water,14a from ref. (14b). A K c t = A%ol* - m a p t 1 atm = 101 325 Pa.2550 Study of Organic Compounds in Octan-1-01 Table 1. Thermodynamic standard functions of solvation, AXEct ( X = G, If, S ) and heats of solution, AKoln in octan- 1-01 at 298.15 K" ~ ___ Solute AGct W C t - TAS0,ct A f C O h water hexane cyclohexane ethyl ether tetrahydrofuran propanone pentan-3-one cyclopentanone - 16.48k0.17 - 13.89f0.08 - 15.56 f 0.04 - 12.35k0.17 - 16.18f0.21 -12.93f0.13 -17.99kO.13 - 20.88 f 0.21 -40.57 k0.08 - 30.79 f 0.08 - 31.55 f 0.06 - 24.86 0.36 -28.32 +O.15 -22.37k0.16 - 32.86 k 0.21 - 34.34 0.24 24.09 k 0.25 16.90 f 0.16 15.99 f 0 . 10 12.51 k0.53 12.14f0.36 9.44 f. 0.29 14.87 k 0.34 13.46 0.45 3.44 f 0.07 0.76 f 0.04 1.50 & 0.02 2.29 k 0.1 1 3.68 0.06 8.47 & 0.12 5.72 f 0.08 8.38 k 0.06 a All data are in kJ mol-l. AGEct refers to the transfer of solutes from 1 mol dm-3 ideal gas state to 1 mol dmP3 hypothetical ideal solution state in octan-1-01. AWoct is the calorimetric enthalpy change for the transfer from ideal gas to infinitely dilute solution. T A g c t is obtained from - T A g c t = AGZct - A K c t . The standard entropies for process (1) have been finally calculated from the free energies and enthalpies of solvation using the well known relationship : A x k t = (Wet -AG&t)/T* Our value of the free energy of solvation of hexane is in good agreement with the value of AG:,, = - 13.56 kJ mol-1 reported by Abraham.2b An acceptable agreement also exists between our AG:ct value for water and those calculated by using the solubilities of water in the octanol phase equilibrated with aqueous saline solutions of known water activity, reported by A ~ e l b l a t .~ In effect, this author considered solutions in octan- 1-01 of higher water content than ours. However, both our data and his may be satisfactorily described by the same quadratic equation (fig. 2), which gives an intercept (AG$, = -4.01 kJ mol-l) very near to the value obtained from our data alone (AG,*,, = -4.02 kJ mol-I).An appreciable difference is instead found between the value AGZct = - 15.56 kJ mol-1 obtained by us for cyclohexane and the value calculated from data reported by Abraham AG:,, = - 13.26 kJ mo1-1.2a In this latter comparison the results are affected by the fact that Abraham considered the free energy of partitioning referred to a system whose phases are water-saturated octanol and octanol-saturated water. The effect of the mutual water-octanol solubility on the partition coefficients is showed by fig. 3, where the AGf-,,,, data, obtained from direct measurements of partition coefficients, are compared with the values of standard ideal free energy of transfer AG;,,,, calculated from the free energies of solvation in the pure phases: It may be seen that differences of ca.1-2 kJ mol-l are common and may be positive or negative according to the nature of the solute. No comparable data are reported in literature for the enthalpy of solvation in octanol for the compounds examined here. As far as the dependence of the enthalpy of solvation on the water content of the octanol solvent is concerned, some experiments were conducted for propanone. In fig. 4 is reported the trend of the limiting heat of solution of propanone when octanol (with increasing amounts of water) is used as a solvent. The difference between the heat of solution in octanol and in water-saturated octanol is 2.8 kJ mol-l. Noticable differences are also exhibited by the phenol derivatives studied by Beezer et aZ.,7 whereas Riebesehl did not find any appreciable difference in the enthalpy of solution of n-alkanols in pure and water-saturated octan01.~ The influence of water content on the thermodynamic functions of solvation of organicP.Berti et al. 2551 Fig. 1. AGb,, [eqn (2)] us. the solute mole fraction in octan-1-01, X,. Solute: (a) cyclohexane, (b) hexane, ( c ) cyclopentanone, (d) pentan-3-one, (e) THF, (f) ethyl ether. compounds in octanol will be considered elsewhere. In this paper we will limit ourselves to a first comparison between the thermodynamics of solvation in octan-1-01 and in water and to deduce their role in the transfer process of solutes between these two pure liquids. Discussion In fig.5 are reported the standard thermodynamic functions of solvation in octan-1-01, together with the values of the thermodynamic functions of hydration, AX:, and the standard thermodynamic functions of transfer of solutes from an ideal 1 mol dmP3 solution in water to a hypothetical ideal 1 mol dm-3 solution in octanol. Values of AX; (j = h, oct, w -+ oct) for methane, obtained from studies of solubility at various temperatures, are also reported.17 A first glance analysis of this figure allows us to recognize that: (a) Except for hydrocarbons in water and methane in octanol, the solution is more stable than the gas phase. (b) The stabilization of the solution is due, for both solvents, to an enthalpic effect, which prevails on the unfavourable entropic term.(c) The order of the compounds on2552 Study of Organic Compounds in Octan-1-01 \ \ I I I Fig. 2. AGA,, us. A', for (a) propanone and (b) water: 0, this work; ., calculated from ref. ( 5 4 . the scale of increasing enthalpies of solvation in octanol differs from the order found on the scale of the corresponding entropies. The same behaviour is observed when the solvent is water. This means that the rule of enthalpy-entropy compensation is not strictly observed.lB ( d ) The order of the compounds in the scales of the functions of solvation in octanol is different from that found in the corresponding scales of hydration. Obviously, this is a consequence of the different interactions the polar and the non-polar groups have with the two solvents considered. ( e ) As far as the hypothetical partitioning between pure octanol and pure water is concerned, the octanol phase is favoured in respect to the aqueous state, except for propanone.The reason lies in the positive entropic effect, which prevails on the positive value of the enthalpy of transfer. Riebesehl found an endothermic enthalpy of transfer from water to octanol even for the n-alkanols studied by flow-microcalorimetry.9 This behaviour is not exhibited by phenol derivatives, which show negative enthalpies, as shown from measurements of heats of solution in water and in octanol carried out by Beezer et al.' and Haberfield et aZ.,lo or from the dependence of experimental partition coefficients on temperature.6 c $ d * l9 From fig. 6 we find that the entropies of solvation in octanol and in water are linearlyP.Berti et al. 2553 -30 - 20 -70 0 A G: - OCt /kJ mol-* Fig. 3. Free energy of transfer from octanol-saturated water to water-saturated octanol, AGF +,,., us. free energy of transfer from pure water to pure octanol, AG;,,,,. Points 1-8 are as follows: 1, methane; 2, ethane; 3, hexane; 4, heptane; 5, octane; 6, cyclohexane; 7, pentan-3-one; 8, tetrafluoromethane. AGf,,,, values were calculated from partition coefficient data [AGf,o,t = --RTln (Poctlw)] taken from ref. (2b), (15) and (16). AG~,,,, values were obtained using eqn (5) from AGZct and AGE data taken from ref. (1 a ) and from this work and ref. (2b), respectively. c -. ---aw- l I I 1 x w 0 .I 0.2 0.3 Fig. 4. Limiting values of the enthalpy of solution, A g o , , , of propanone in an octan-1-01-water mixed solvent us.the water mole fraction, X,. related to the intrinsic volume of solute molecules, estimated as the van der Waals volume, Vw.20v 21 Contrary to what happens in water, the free energies and enthalpies of solvation in octanol are mainly related to the size of solutes. Water is an evident exception, in fact it exhibits a large affinity for octanol, in spite of its small van der Waals volume and its substantially hydrophobic environment. To this purpose, it may be noted that the solubility of water in 2,2,4-trimethylpentane is Xw = 8.76 x 10-4,22 whereas in octan-1-012554 40 Study of Organic Compounds in Octan-1-01 - AH" 20 - I .-. E c, -. A Y 0 - 2 c -20 -40 - A G O TAS" 7.2.1 4';!6 ."~' 3 1 I Fig. 5.Standard thermodynamic properties of solvation, AX; ( X = G , H , S ) , in octan-1-01, ( j = oct), in water, ( j = h), and of transfer from water to octanol, AXO,,,,, (= AXZCt-AX;). 1, ethyl ether; 2, THF ; 3, hexane; 4, cyclohexane; 5, propanone ; 6, pentan-3-one; 7, cyclopentanone; 8, methane. AXE,, data are taken from this work, except for methane, ref. (1 7); AX; data are from ref. (1 a). Hatched areas show AX:ct, shaded areas show AXO,,,,, and unfilled areas show AX;. X, == 0.275.5d The high solubility of water in octanol may be justified by the formation of a tetrahedral complex in which a molecule of water is hydrogen-bonded to four hydroxy groups of octanol m01ecules.~~ The formation of this species is accompanied by a large heat of solvation, which is similar in magnitude to the enthalpy of vaporization of pure water.This enthalpic effect prevails on the unfavourable entropy of solvation, which is, however, noticeably more negative than the entropy of solvation of the organic compounds examined here. Although less evident, we can however recognize the different behaviour of hydro- carbons in octan-1-01 from that of the other organic compounds of similar size but containing heteroatoms, the former showing less stability. Cyclisation increases the stability of the compounds in octanol by ca. 3-4 kJ mol-1 and is due to an enthalpic effect, being the entropy of solvation of the open-chain and their analogous cyclic compounds practically the same. A more detailed analysis in terms of the contributions that the polar and non-polar groups bring to the thermodynamic functions of solvation in octan-1-01, is at the moment premature because of the few experimental data available.This work has been supported by Minister0 della Pubblica Istruzione (Roma).P. Berti et al. 2555 vw Fig. 6. Thermodynamic properties of solvation in octan-1-01, AXct, and in water, A%, ( X = G, H , S ) us. the intrinsic molar volume, V,. AX,",, data are taken from this work and ref. (2b), AX: data from ref. ( l u ) . The V, values were estimated according to ref. (20) and (21). ., water; 0, alkanes (methane, ethane, cyclohexane, hexane, heptane, octane); 0, ketones (propanone, cyclopentanone, pentan-3-one); 0, ethers (THF, ethyl ether); (A), octan-1-01. (a) AHgct, (b) AH:, (4 TAgct, (4 TAG, (4 AGiCt, (f) AGE.References 1 See e.g.: (a) S. Cabani, P. Gianni, V. Mollica and L. Lepori, J . Solution Chem., 1981, 10, 563; (b) E. Wilhelm, R. Battino and R. J. Wilcock, Chem Rev., 1977, 77, 219; (c) M. H. Abraham, J . Chem. Soc., Faraday Trans. I , 1984, 80, 153. 2 See e.g.: (a) M. H. Abraham, J . Am. Chem. Soc., 1979, 101, 5477; (b) M. H. Abraham, J . Am. Chem. Soc., 1982, 104, 2085; (c) C. L. De Ligny, N. G. van der Veen and J. C. van Houwelingen, Znd. Eng. Chem. Fundam., 1976,15, 336; ( d ) C . V. Krishnan and H. L. Friedman, in Solute-Solvent Interactions, ed. J. F. Coetzee and C . D. Ritchie (Marcel Dekker, New York, 1969), vol 11, chap. 9, pp. 1-103; (e) E. Wilhelm and R. Battino, Chem. Rev., 1973, 73, 1 ; (f) H. L. Clever and R. Battino, in Solution and Solubilities, Techniques of Chemistry, ed.M. R. Dack (John Wiley, New York, 1975), vol. 8, part 1, chap. 7. 3 See e.g.: (a) T. Fujita, J. Iwasa and C. Hansch, J . Am. Chem. Soc., 1964, 86, 5175; (b) C. Hansch and S. M. Anderson, J , Org. Chem., 1967,32,2583; (c) A. Leo, C. Hansch and D. Elkins, Chem. Reu., 1971, 71, 525; ( d ) A. Leo, J . Chem. Soc., Perkin Trans. 2, 1983, 825; (e) I. Moriguchi, Y. Kanada and K. Komatsu, Chem. Pharm. Bull., 1976, 24, 1799.2556 Study of Organic Compounds in Octan-1-01 4 See e.g.: C. Hansch, Acc. Chem. Res., 1969, 2, 232; (b) R. F. Rekker, The Hydrophobic Fragmental Constant (Elsevier, Amsterdam, 1977); (c) N. H. Anderson, M. James and S. S. Davis, Chem. Znd., 1981, 677; ( d ) C. Hansch and A. Leo, Substituent Constants for Correlation Analysis in Chemistry and Biology (John Wiley, New York, 1979); (e) R.N. Smith, C. Hansch and M. M. Ames, J. Pharm. Sci., 1975, 64, 599. 5 See e.g. : (a) C. Hansch, J. E. Quinlan and G. L. Lawrence, J . Org. Chem., 1968,33,347; (b) Y. B. Tewari, M. M. Miller, S. P. Wasik and D. E. Martire, J. Chem. Eng. Data, 1982, 27, 451; (c) D. Mackay, A. Bobra and W. Y. Shiu, Chemosphere, 1980, 9, 701 ; ( d ) A. Apelblat, Ber. Bunsenges. Phys. Chem., 1983,87,2; (e) M. M. Miller, S. Ghodbane, S. P. Wasik, Y. B. Tewari and D. E. Martire, J. Chem. Eng. Data, 1984,29, 184; (f) Y. B. Tewari, D. E. Martire, S. P. Wasik and M. M. Miller, J. Solution Chem., 1982, 11, 435. 6 See e.g.: (a) I. Kojima and S. S. Davis, Int. J. Pharm., 1984, 20, 247; (6) M. James, S . S . Davis and N.H. Anderson, J . Pharm. Pharmacol., 1981, 108P; (c) N. H. Anderson, S. S. Davis, M. James and I. Kojima, J. Pharm. Sci., 1983, 72, 443; ( d ) A. E. Beezer, W. H. Hunter and D. E. Storey, J . Pharm. Pharmacol., 1980, 32, 815. 7 A. E. Beezer, W. H. Hunter and D. E. Storey, J . Pharm. Pharmacol., 1983, 35, 350. 8 A. E. Beezer, W. H. Hunter and D. E. Storey, J . Pharm. Pharmacol., 1981, 33, 65. 9 W. Riebesehl, VIth Convegno Nazionale di Calorimetria ed Analisi Terrnica (AICAT), Napoli, 4-7 December 1984; C 52 Atti Convegno. 10 P. Haberfield, J. Kivuls, M. Haddad and T. kzzo, J . Phys. Chem., 1984, 88, 1913. 11 S. Cabani, G. Conti and L. Lepori, Trans. Faraday SOC., 1971, 67, 1933. 12 T. Boublik, V. Fried and E. Hala, The Vapour Pressures of Pure Substances (Elsevier, Amsterdam, 1973). 13 A. J. Treszezanowicz and G. C . Benson, J . Chem. Thermodyn., 1978, 10, 967. 14 (a) D. Eisenberg and W. Kauzmann, The Structure and Properties of Water (Oxford University Press, Oxford, 1969); (b) J. D. Cox and G. Pilcher, Thermochemistry of Organic and Organometallic Compounds (Academic Press, London, 1970). 15 C. Hansch, A. Vittoria, C. Silipo and P. Y. C. Jow, J . Med. Chem., 1975, 18, 546. 16 C. Hansch, A. Vittoria, C. Silipo and P. Y. C. Jow, J . Med. Chem., 1976, 19, 61 1. 17 R. J. Wilcock, R. Battino, W. F. Danforth and E. Wilhelm, J . Chem. Thermodyn., 1978, 10, 817. 18 R. Lumry and S. Rajender, Biopolymers, 1970, 9, 1125. 19 J. A. Roggers and A. Wong, Znt. J . Pharm., 1980, 6, 339. 20 A. Bondi, J . Phys. Chem., 1964, 68, 441. 21 J. T. Edward, J. Chem. Educ., 1970, 47, 261. 22 W. Riebesehl and E. Tomlinson, J . Phys. Chem., 1984, 88, 4770. 23 A. S. C. Lawrence, M. P. McDonald and J. V. Stevens, Trans. Faraday Soc., 1969, 65, 3231. Paper 511755; Receiued 9th October, 1985

ISSN:0300-9599    

DOI:10.1039/F19868202547

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

I. Chem. SOC., Faraday Trans. 1, 1986,82, 2557-2564 Remarks on Dependence on Temperature ‘ at Constant Volume ’ Patrick G. Wright Department of Chemistry, The University, Dundee DDl 4HN It is sought to extend recent treatments of the dependence of equilibrium constants on temperature ‘at constant volume’, with explicit attention to the behaviour of equilibrium constants based on various different types of specification of concentration. It is shown that for reactions in solution, if ‘at constant volume’ means ‘at constant molar volume of the pure solvent’, then the dependence on temperature is indeed rclated to a certain type of AU in precisely the way that has for a long time been taken to be the case. If, however, ‘at constant volume’ means ‘at constant total volume of an actual reacting system’, then the dependence on temperature may be incompletely specified except in the limit of infinite dilution, when it becomes the same as at constant molar volume of the pure solvent.In contrast to the situation which obtains with the dependence on temperature at constant pressure, the use of amounts of solute per unit volume involves no complications in the dependence at constant molar volume of the pure solvent. 1. Introduction In a recent paper, Blandamer et a1.l have clarified some aspects of the dependence of equilibrium constants on temperature ‘at constant volume ’. Some extension of the argument seems capable of attaining rather greater definiteness in some of the conclusions. To do so, however, it is convenient to use a notation which goes beyond the normal set of superscripts.For example, the symbol AUm as normally used means where the partial derivative relates to a closed system and the limiting process envisaged is an approach to infinite dilution. In the argument which follows, however, a symbol is needed not only for this quantity but also for the quite distinct quantity 2. Notation to be used A sufficient symbolism for the present purpose can be provided by the following. Homogeneous systems alone will be considered : further complications would arise if attention were to be directed to such situations as, say, partition of a solute between two solvents. ( a ) Let { X ) denote the measure-number of a physical quantity X in terms of some appropriate unit. Let { { X } } denote the ratio of a physical quantity X to some constant having the same dimensions as X .This constant may or may not be equal to the unit that is used. ( X } depends not only on X but also on the unit that is chosen to be used. On the other hand, ( ( X } } depends on X and on the constant, but is independent of the choice of unit. For example, if Xis the pressurep, then ifp = 64.17 kPa and the constant is 101.325 kPa, then ( p } is 64.17 in terms of the unit kPa, 6.417 x lo4 in terms of the unit Pa, 0.6333 in terms of the unit atm, 481.3 in terms of the unit mmHg and so forth, but ( { p ) } is 0.6333 whichever unit of pressure is used.2558 Dependence on Temperature ‘ at Constant Volume ’ (b) Let ‘concentration-variable’ be used as a generic term for such quantities as: (i) the amount of a gas or a solute per unit volume, (i’) the mass of a gas or a solute per unit volume, (ii) the formal partial pressure x i p of a gas, (iii) the amount of a solute per unit amount of solvent, (iii’) the mass of a solute per unit amount of solvent, (iii”) the amount of a solute per unit mass of solvent, (iii”’) the mass of a solute per unit mass of solvent, (iv) the mole fraction of a substance in a solution, (iv’) the mass fraction of a substance in a solution. Some such variables are dimensionless, and some are not.(c) Both for dilute gaseous systems and for dilute solutions, some quantities exhibit a logarithmic asymptotic dependence on concentration-variables. For example, for any particular concentration-variable q, there are asymptotic dependences pi N ‘constant’ + RT In ({qi)) (1) 2: ‘constant’+Z vi RT In { { q i } ] i 2: ‘constant’ -C vi R In ((qi)).( 1 ”1 a Let tX denote what will here be called the ‘non-logarithmic part of’ a quantity X , i.e. the quantity which remains after the logarithmic terms, from it. Thus, for example: t ~ i = pi - RT In {(rill and but if any, have been subtracted is equal to (i3H/2t)T, itself. ( d ) For any extensive thermodynamic quantity 2, let and A211 denote l i m t - (3, p (2”) (3) (3’) (where V is the actual volume of an actual closed system), and for a reaction in solution let A2?l denote l i m t (3’’) (where V i is the molar volume of the pure solvent-species A); with ‘lim’ everywhere referring to the limit of infinite dilution. If 2 is U or H , the limiting values could also be specified as the appropriate sorts of lim(aZ/ac), but if 2 is S, A ( = U - TS), G, J(= - A / T ) or Y( = - G / T ) , then the specification has to be in terms of a non-logarithmic part.If Z is U or H, and to some extent for all sorts of 2 in the case of the special stoichiometry xi vi = 0, the limiting values AZ11, AZ)) and A211 are unique. However, if Zis S, A , G, J o r Y, and Ci vi # 0 (or i f x i vi = 0 and ‘mass-based’ concentration-variablesP. G. Wright 2559 are considered), then the limiting values AZll, A D ) and A211 are affected by which particular concentration-variable r , ~ is used, and by the magnitude of the constant q e by which q is divided in obtaining ({q)). To distinguish which particular concentration-variable has been used, it would be possible to distinguish different sorts of (say) AG]] as: AGill, AGi’ll, AGiill, AGiiiJ1, AGiii’ll, .. . where the numerals refer to the quantities listed in section 2(b). For reactions of gases, the ordinary A G e is AGiiIl. A g l , when it exists (i.e. when there is a solvent), is (unless the solvent is incompressible) automatically equal? to AZll. For and for constant T the molar volume F/; of the pure solvent is constant if p is constant; and thus It thus suffices to consider only AZIl’s and AZ))’s; the latter of which are related to partial derivatives (aZ/o?t),, r , ~ $ always exhibits an ‘ asymptotic constancy’, whichever particular concentration-variable is considered. In such terms, it is possible to write at constant actual volume of an actual reacting system.(e) The equilibrium value of the quotient of powers of concentration-variables lim (n ~ { i ) , ~ = K,, (4) i where K,,, especially if Zi vi # 0, depends on which of the concentration-variables is used. For reactions of gases (for the moment ignoring non-ideality), K,, is K, if ri is the amount per unit volume n i / V ; K p if qi is the partial pressure x i p ; and K e if qi is the related quantity x,p/p*. 3. Relations for Reactions of Gases and for Reactions in Solution For reactions of gases, AG]] and so forth depend on T but not on p . The dependence of any Kq on T can be inferred from either of the equivalent relations (which apply whichever sort of q is used), but while AG]] = - RT In {{K,,}}, AY]] = R In ({K,,)} ( 5 ) dAGiill/dT = - A p i l l , d A p i l l / d T = + A H i i l l / p and thus d In ({Kp})/d T = AH11/RT2 no such relation applies to K,. As has often been pointed out (if not quite in the present notation), if E i v i # 0, then: dAGi’l/dT # - AF]], dA Yill/dT # AHi]]/ T 2 and d In ((K,})/dT = AUlI/RT2 # AH11/RT2.(7) It may be noted in passing that for reactions of gases: AH]] = AH) and AU]] = A U ) t Equal to AZ]], and not (as might momentarily have been conjectured) to A Z ) .2560 Dependence on Temperature ' at Constant Volume' and that, since the Kq's just mentioned exhibit only a dependence on T and not also a dependence on p or V, the dependence on T is the same at constant V as it is at constant p . Well known relations2-6 for reactions in solution include, in the present notation: 8 ln((K'I}} AH]] ( aT )P = a 1 n W >> - Avll types (iii) and (iv) ( ap'I IT- RT for concentration-variables of butt (9) (10) for concentration-variables of type (i) (where a; is the coefficient of thermal expansion of the pure solvent, and isothermal compressibility). is its The dependences on T at constant V l are readily inferred, for For concentration-variables of types (iii) and (iv), this becomes and the limiting values of these four partial derivatives are, respectively: AU]], AU), (aU;/aV&, and A V ] .It follows that the above expression for reduces to the simple forms AU) RT2 t This is a point at which further complications can enter if attention is not restricted to homogeneous 3 Z.e. the A I ~ of Blandamer et al.' is equal to AU)).What basically is involved is the relation systems. AH']- TAVll(ap/aT)V~ = AU" equations equivalent to which have frequently appeared in the literature (but without full recognition that what is involved is the dependence of K,, on T at constant molar volume of the pure solvent).P. G. Wright 2561 This derivation relates to concentration-variables of types (iii) and (iv) and it will now For concentration-variables of type (i) : be shown that precisely the same result holds for concentration-variables of type (i). and this expression differs from the preceding one [eqn (1 l)] by: which is equal to zero, since A basically simple relation can be discerned, that: For all of the sorts of Kq most usually used? for reactions in solution, a relation holds which resembles that for the effect of temperature on K, for reactions of gases.This may be contrasted with the less simple but more familiar proposition that: For a K for reactions in solution that is based on amounts of solute per unit mass of solvent, or on mole fractions, but not for one that is based on amounts per unit volume, a relation AH]] = lim (g) i2 ln({K,)) - AH]] ( aT )P-w, T, P holds which resembles that for the effect of temperature on Kp or K* for reactions of gases. The simplicity of the result appears only in terms of A V ) . In terms of AU]] [ E lirn(aU/ac),, the relation which holds is (but even this has the feature of being the same for all of these sorts of KJ. A partial derivative (a In {{Kv}}/i3T)v, with V denoting the actual volume of an actual reacting system, is for many sorts of KV not a properly defined quantity.(For any KV which depends on both T and p , the derivative would be properly defined if p were completely determined by T and V. Normally, however, this is not the case: p would be completely determined by T, V and amounts of substances, but not by T and V alone.) In the special case of the limit of infinite dilution, (a ln((K,>}/aT), does become a properly defined quantity, but it is then equal to (a ln{(K,J)/C)T)v;. Consequently, no further relations arise. The one special feature of V, as distinct from V l , in the dilute limit is that '@V/a<),, v' is identically zero, but (i3P'/ac)T, v; is equal to (aV/2<),, P . As Blandamer et aZ.l point out, relations amounting to (8 ln((K,))/aT), = A U ) / R T 2 t A K based on volume fractions can exhibit special features.This question is treated in a further communication.2562 Dependence on Temperature ' at Constant Volume ' have frequently appeared in the literature without explicit statements that they actually relate to constant Vl; or, if to constant actual volume, to constant actual volume in the limit of infinite dilution. Such relations have, moreover, often appeared without clear indication of what sort or sorts of Kv are being considered. 4. Relations Involving a AC, Blandamer et a1.l raise a related point, concerning the dependence of AU)) on temperature. A clear-cut and simple relation holds for the dependence at constant V i (though not for that at constant actual volume, save in the limit of infinite dilution).The quantity @A w / a T),; This second derivative whose limit has to be taken will necessarily satisfy the equation From the relations: it follows that which implies that = 0. Therefore, eqn (14) implies that By using limit (13), the right-hand side of this equation is equal to (aA U)/a T ) , . The left-hand side is equal to which is equal to AC)?. Therefore, it is now seen that t Here, and elsewhere, all double limits are taken to behave 'non-pathologically' (as they will, unless a gas/liquid critical point, critical point for partial miscibility, or other critical point, is involved).P. G. Wright 2563 As Blandamer et a1.l indicate, this simple relation holds for a dependence at constant molar volume of the pure solvent, and not for a dependence at constant pressure.A corresponding relation for constant pressure is, however, easy to infer. Since and it can be seen that eqn (15) implies that: The right-hand side of this equation could easily differ appreciably from ACV. Typical values of (ap/aT),; are ca. I MPaK-l and, judging by analogy with experimental observations on the dependence of AH]] values and activation energies on pressure, values of (aAU’))/C3p)T of ca. m3 mol-l do not seem impossibly large (at least for some reactions). These values would correspond to (aAU))/alJ, differing from ACI by ca. 10 J mol-l K-l; and if (aAV)/i3p)T were, rather, ca. 10+ m3 mol-l, then the difference would still be ca. 1 J mol-l K-l. 5. Dependence on Pressure at Constant Molar Volume of the Pure Solvent It is evident that and it follows that (and this holds for a Kv based on amounts per unit volume just as much as for one based on mole fractions or on amounts of solute per unit mass of solvent).Otherwise expressed, the Ad of Blandamer et a1.l is equal to - AU))/T(ap/aT),;. This ‘ dependence on pressure ’ reduces virtually to a dependence on temperature. 6. Concluding Remarks It is hoped that the present analysis will provide a conclusive identification of various ‘ isochoric ’ dependences on temperature, frequently contemplated in the past, in the light of specific matters which Blandamer et a1.l have called into question. The equilibrium relations relied upon by such authors as Brummer and Hills’ and WhalleyR are indeed correct, but as Blandamer and coworkers have pointed out in their detailed analysis of this subtle problem, there has been an overriding need for clarification of the precise significance of ‘isochoric’ in these relations.Past ambiguities are perhaps related to the fact that each of the definitive relations (a ln((K,>)/c7T),~ = AU))/RT22564 Dependence on Temperature ‘at Constant Volume ’ and (aAU)/aT),; = AC; involves two senses of‘ isochoric’, and not just one. The explicit partial derivative relates to constancy of the molar volume of the pure solvent, while A@) and AC$ are quantities of the type formed from a partial derivative which, on the contrary, relates to constancy of the total volume of an actual system. It may finally be stressed that every sort of K,, considered here is equal to a limiting value lim (az/at I T , v lim (n r& i and is not the quotient of concentration-variables in some actual system which is not necessarily particularly dilute. There is no necessity for RT2@ In cccn rSi),,>>/~T)v: i to be equal either to AU) = lim (aU/at),, , or to the value of (aU/at),I, , for the actual experimental situation.Similarly, it has often been realised (and unfortunately, sometimes has not been realised) that in cases where there is no necessity for (a In {(K,)}/eT)p = AH11/RT2 to be equal either to AH]] = lim(C7H/a<)T,. conditions of the actual experimental situation. A similar remark applies to or to the value of (i3H/at)T, for the Afl] 3 lim(aV/ag),, and the value of (aV/ay),, for the actual experimental situation. References 1 M. J. Blandamer, J. Burgess, B. Clark and J. M. W. Scott, J . Chem. SOC. Faraday Trans. 1, 1984, 80, 2 E. A. Guggenheim, Trans. Furuduy SOC., 1937,33,607; Thermodynamics (North Holland, Amsterdam, 3 S . D. Hamann, Physico-Chemical Efects of Pressure (Butterworths, London, 1957). 4 G. S. Kell, J. Chem. Eng. Data, 1975, 20, 97. 5 J. B. Rosenholm, T. E. Burchfield and L. G. Hepler, J . Colloid Interface Sci., 1980, 78, 191. 6 L. G. Hepler, Thermochim. Acta, 1981, 50, 69. 7 S. B. Brummer and G. J. Hills, Trans. Faraday Soc., 1961, 57, 1816; 1823. 8 E. Whalley, Adu. Phys. Org. Chem., 1964, 2, 93. 3359. 3rd edn, 1957) p. 319. Paper 5 / 1786; Received 15th October, 1985

ISSN:0300-9599    

DOI:10.1039/F19868202557

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I, 1986, 82, 2565-2568 Special Features of Equilibrium Constants that are based on Volume Fractions Patrick G. Wright Department of Chemistry, The Uniuersity, Dundee DDI 4HN Attention is directed to special features of the effect of temperature and pressure on equilibrium constants that are based on certain kinds of volume fractions. These features are akin to the well known complications that arise with equilibrium constants that are based on amounts per unit volume. ~- ~~ Equilibrium Constants based on Volume Fractions Dependence on Temperature Relations for chemical equilibrium are sometimes1$ expressed in terms of quotients of volume fractions; and it is the purpose of this note to point out that such kinds of equilibrium constant can have a dependence on temperature that involves complications similar in origin to the complications which arise3-' with equilibrium constants based on amounts per unit volume.Let K4 denote an equilibrium constant equal to where di is the volume fraction of species i in a solution in a solvent A, and the limiting process envisaged is one of approach to infinite dilution. Then there will be four cases of the dependence of equilibrium constants K4 on temperature, one for each of four distinct meaningss of 'volume fraction'. (a) A first type of volume fraction q5i is the quantity xi q/ xi q all i where xi is the mole fraction of i in the solution, and In this case, di exhibits an asymptotic behaviour is the molar volume of pure i. di xi T/V"A (3) (PA being the molar volume of the pure solvent) and it follows that: (Kz = lim [n ~ ; i ] ~ ~ ) .i By logarithmic differentiation, it follows that : which implies that: A HI] r+) = -+x viaF-ai c vi RT2 i ( 5 ) 25652566 where (using the notation of the previous paperg) AH]] is an ordinary AHm equal to Equilibrium Constants based on Volume Fractions lim (awac)?., p and a: is the coefficient of thermal expansion of pure i. The term -a: C vi is one whose presence was long ago shown3 to be required for equilibrium constants based on amounts per unit volume. The somewhat similar term i via: i is a special feature arising with the present sort of K4, whose dependence on temperature at constant pressure thus involves the thermal expansion of solutes as well as that of the solvent. It is also possible to consider an ‘isochoric’ dependence of Kd, as expressed1* by a partial derivative at constant molar volume of the pure solvent.From eqn (4), it is seen that: (with no third term, since a partial derivative at constant ITA is involved). Now: A U)) v: R P where AU)) is a sort of ‘ AUm ’ equal to lim ( W a O T , v * Further : and therefore it follows that: In contrast to what happens9 with all of the various sorts of K not based on volume fractions, the right-hand side does not reduce to AU))/RT2. (b) A second type of volume fraction bi is the quantity where is the partial molar volume of species i in the solution. In this case, there is an asymptotic behaviour (bi N xi tp/voA (3’) and, by following through algebra exactly analogous to that just given, it is seen that for this sort of K4: andP.G. Wright (c) A third type of volume fraction is the quantity 2567 where qi is a constant (commonly corresponding to some fixed number of sites in a lattice model). In this case, there is an asymptotic behaviour bi 2 xi (constant) and so K4 is equal to the product of Kx and a constant; and therefore K4 has precisely the same dependence on temperature as Kx. ( d ) A fourth type of volume fraction is the quantity xi C(T,P~)/'X xi K(T',PJ i where all the molar volumes relate to a particular fixed temperature and pressure and not to the temperature and pressure of the conditions of an actual experiment. This case8 amounts to a sub-case of the preceding one; and so again corresponds to a K4 which has precisely the same dependence on temperature as Kx.(Cf. a remark by Isaacs. 11) Dependence on Pressure at Constant Temperature (a) From eqn (4) above, it follows by logarithmic differentiation that: Since it follows that for this sort of K4: (a In KJap), = - A ml/RT The isothermal compressibility of solutes is involved, as well as that of the solvent. (b) Similar reasoning shows that for the second sort of K4: (c) and ( d ) For the same reasons as with the dependence on temperature, the third and fourth sorts of K4 exhibit precisely the same dependence on pressure as Kx. References 1 Y. Marcus, Introduction to Liquid State Chemistry (Wiley, New York, 1977), pp. 189-190. 2 Yu. V. Kazakevich and Yu. A. Eltekov, Zh. Fiz. Khim., 1980, 54, 154 (Russ. J . Phys. Chem. Engl. 3 E. A. Guggenheim, Trans. Faraday Soc., 1937,33,607; Thermod-vnamics (North Holland, Amsterdam, 4 S. D. Hamann, Physico-Chemical Efects of Pressure (Butterworths, London, 1957). 5 G. S. Kell, J. Chem. Eng. Data, 1975, 20, 97. 6 J. B. Rosenholm, T. E. Burchfield and L. G. Hepler, J . Colloid Interface Sci., 1980, 78, 191. 7 L. G. Hepler, Thermochim. Acta, 1981, 50, 69. 8 J. H. Hildebrand and R. L. Scott, Solubility of Non-electrolytes (Reinhold, New York, 3rd edn, 1950), Transl., 1980, 54, 82). 3rd edn, 1957), p. 3 19. p. 133; Regular Solutions (Prentice Hall, New Jersey, 1962), p. 11.2568 Equilibrium Constants based on Volume Fractions 9 P. G. Wright, J . Chem. Soc., Faraday Trans. 1, 1986, 82, 2557. 10 M. J. Blandamer, J. Burgess, B. Clark and J. M. W. Scott, J . Chem. SOC., Faradav Trans. 1, 1984, 80, 1 1 N. S. Isaacs, Liquid-Phase High Pressure Chemistry (Wiley, New York, 1981), p. 187. 3359. Paper 511787; Received 15th October, 1985

ISSN:0300-9599    

DOI:10.1039/F19868202565

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I, 1986, 82, 2569-2575 Infrared Study of Pyridine Adsorption on Rutile and Silica-coated Rutile Immersed in Heptane Colin H. Rochester* and David G. Smith Department of Chemistry, The University, Dundee DDI 4HN The surface of rutile immersed in solutions of pyridine in heptane contains Lewis-acidic Ti4+ sites which are involved in coordinative interactions with adsorbed pyridine molecules. Different types of surface hydroxy group behave differently, one type forming hydrogen bonds with pyridine molecules. Treatment of rutile with silicon tetrachloride and water vapour resulted in a coated surface which behaved more like the surface of silica than of rutile in the presence of adsorbed pyridine. The use of infrared spectroscopy for the study of adsorption behaviour in situ at the solid/liquid interface is well established.'* Spectra have been reported of rutile immersed in heptane and mixtures of heptane with aromatic hydrocarbon^.^ Results for rutile immersed in solutions of triethylamine in heptane suggested that the amine was adsorbed in three ways which involved hydrogen-bonding interactions with surface hydroxy groups, coordinative interactions with Lewis-acidic surface sites and dissociative ad- sorption to give diethylamine and adsorbed ethoxide and ethanoate anion^.^ The present work was aimed at confirming the reactivity of Lewis-acidic sites at the rutile/heptane interface by investigating the adsorption of pyridine, which is a good probe molecule in this Triethylamine was adsorbed on rutile which had been preheated at 400 K4 Rutile was here pretreated by evacuation at room temperature or 673 K in order to gain insight into the effects of surface hydroxy groups8 on the availability of adsorption sites.In view of the importance of coated rutile as a pigment the adsorption of pyridine on rutile which was modified by treatment with silicon tetrachloride followed by hydrolysis has been briefly investigated. Experimental Rutile (code CL/D482/1, Tioxide International Ltd) with surface area 30 m2 g-l was freed from surface chloride ions as beforeg the final treatment involving heating in air at 383 K (24 h). Heptane and pyridine were purified by triple distillation followed by a series of freeze-pumpthaw cycles under vacuum and were stored under nitrogen.Rutile was compressed by 4 ton pressure on a 2.5 cm diameter sample weighing ca. 120 mg to obtain a self-supporting disc. This was then mounted in an infrared celllo which was glassblown to a vacuum apparatus designed to allow storage and manipulation of liquids and solutions. Discs were initially heated in oxygen (1 3 kN m-2) at 673 K for 2 h to free the surface from hydrocarbon impurities, then equilibrated with saturated water vapour at room temperature and finally evacuated either at 673 K or at the ambient temperature (ca. 300 K) in the optical beam of the infrared spectrometer. The method of subsequently recording spectra of the discs wetted by solutions of an adsorbate in heptane has been previously described.1° A Perkin-Elmer 683 spectro- photometer was used in conjunction with a model 3500 infrared data station.One rutile disc was modified by treatment with silicon tetrachloride and water vapour. After treatment in oxygen and exposure to water vapour the disc was heated in vacuum (400 K, 2 h), exposed to silicon tetrachloride vapour (5.3 kN m-2) at 523 K (2 h), cooled 85 2569 FAK 12570 I. R . Study of RutilelHeptane 3800 3400 1600 wavenumber/ cm -' Fig. , Spectra of rutile (a) after evacuation at 673 K (22 h), (6) immersed in heptane, (c) in a dilute solution of pyridine in heptane and ( d ) in a more concentrated pyridine solution. ( e ) Spectrum of pyridine in solution in heptane. to ambient temperature and exposed to water vapour (s.v.P.) for 1 h, evacuated at 400 K (1 h), exposed to silicon tetrachloride (5.3 kN mP2) at 400 K and the temperature raised to 523 K for 2 h, cooled to ambient temperature and evacuated, exposed to water vapour (s.v.P., 1 h) and finally evacuated at 673 K for 15 h.Results Heat treatment of rutile at 673 K largely dehydroxylates the surface leaving a low residual population of hydroxy groups responsible for an infrared band at 3700 cm-l [fig. 1 (a)]. A shoulder at 3655cm-l coupled with a weak band at 3410cm-l also showed that dehydroxylation of the exposed { 1 lo} surface planes of rutile was not quite complete.8 A previous study established that the band at 3655 cm-l shifted to 3605 cm-l and the band at 3410 cm-1 did not move when rutile which had been heated at 400 K was immersed in h e ~ t a n e . ~ The present shift of the band at 3700cm-l and shoulder at 3655 cm-l to give a new band envelope centred at 3610 cm-' [fig.1 (b)] for rutile immersed in heptane indicated that hydroxy groups responsible for the maximum at 3700 cm-l were perturbed by interactions with heptane molecules adjacent to the rutile surface. The addition of pyridine led to the appearance of sharp maxima at 1600 and 1580 cm-l [fig. 1 (c)] due to vibrations of pyridine molecules. The latter could be partially ascribed to pyridine in solution but the former must primarily be ascribed to adsorbed pyridine. The absence of any substantial change in the intensity of the bands due to surface hydroxy groups suggests that hydrogen-bonding interactions between hydroxy groups and pyridine molecules were not significant. By analogy with the assignment of a similarC. H.Rochester and D. G. Smith 257 1 I 0 3400 1600 1400 1600 1400 wavenum berlcm-' Fig. 2. Spectra of rutile (a) after evacuation at 673 K (22 h), (b) in heptane, (ck(e) in solutions containing increasing concentrations of pyridine in heptane and (f) after drainage of the liquid phase and heat treatment (673 K, 14 h) in vacuum. ( g ) Spectrum of pyridine-heptane solution. band at 1605 cm-l in spectra of pyridine adsorbed on rutile at the solid/vapour interface6 the band at 1600 cm-1 is attributed to pyridine which was adsorbed through coordinative interactions with exposed Lewis-acidic surface sites. Infrared bands in the spectral region 1400-1 500 cm-l due to adsorbed pyridine57 were obscured by intense bands due to vibrations of liquid heptane.Addition of a higher concentration of pyridine to the liquid phase considerably enhanced the band at 1580 cm-l due to pyridine in solution. The band at 1600 cm-l was also enhanced in intensity and shifted slightly to lower wavenumber. The latter effect could be ascribable to a contribution to the spectrum at ca. 1590 cm-l due to a vibration of pyridine molecules linked to the surface via hydrogen bonds from surface hydroxy groups. The band due to hydroxy groups perturbed by contact with heptane molecules disappeared from the spectrum when the pyridine concentration was increased [fig. 1 (d)]. Weak bands appearing between 3600 and 3800 cm-l were due to pyridine in solution. The final stage of dehydroxylation of rutile at temperatures around 673 K is very sensitive to vacuum conditions in the infrared cell.A repeat experiment led to the complete disappearance of the bands at 3655 and 3410 cm-l due to hydroxy groups on (1 lo} surface planes8 and left only a residual infrared band at 3700 cm-l [fig. 2(a)]. The band shifted to 3630 cm-l (AvOH = 70 cm-l) when the disc was immersed in heptane [fig. 2(b)]. Addition of pyridine resulted in the disappearance of the band at 3630 cm-1 and the appearance of the maxima at 1600 and 1580 cm-l due to vibrations of pyridine 85-22572 I. R. Study of RutilelHeptane 3400 3800 3400 wavenumberlcm-' 1600 Fig. 3. Spectra of rutile (a) after evacuation at ambient temperature, (b) in heptane, (c)-(i) in solutions containing increasing concentrations of pyridine in heptane and ( j ) after drainage and evacuation at ca.300 K. (k) Spectrum of solution of pyridine in heptane corresponding to spectrum (e). [fig. 2(c)-(e)]. The band at 1600 cm-l did not shift towards lower wavenumbers with increasing pyridine concentration possibly because of the absence of hydroxy groups giving the band at 3655 cm-l (for rutile in vacuum). Because a particularly thin film of liquid in contact with the disc was attained in the repeat experiment the intensity of a band at 1462 cm-l [fig. 2(b)] due to vibrations of liquid heptane was sufficiently low to allow observation of bands at 1483, 1470, 1457 and 1440cm-l due to vibrations of pyridine in solution and in the adsorbed state [fig. 2(c)-(e)]. Drainage and evacuation of the disc at ambient temperature left residual infrared bands at 1607, 1585, 1470 and 1440 cm-l which are characteristic of pyridine liganded to Lewis-acidic sites in the rutile surface.6 The strength of adsorption was apparent from the failure of heat treatment at 673 K in vacuum to desorb pyridine completely from the surface [fig.2(f)]. The maximum at 3700 cm-l due to surface hydroxy groups did not reappear in spectra after the disc was drained of liquid and the cell was evacuated. Fig. 3 shows the spectroscopic results for the addition of pyridine in heptane to rutile which had been evacuated at ambient temperature after saturation with water vapour. The band at 3655 cm-l due to surface hydroxy groups on rutile in vacuums shifted to 3605 cm-l when rutile was immersed in heptane [fig. 3(b)].Successive additions of increasing concentrations of pyridine led to decreases in the band intensity at 3605 cm-l and a general increase in absorption intensity in the range 320&3500 cm-l [fig. 3 (cE(h)]. Drainage and evacuation led to the reappearance of the maximum at 3655 cm-l [fig. 3 (j)]C. H . Rochester and D. G. Smith 2573 3 800 3400 wavenumber/cm-’ Fig. 4. Spectra of rutile after modification by treatment with silicon tetrachloride and water vapour (a) in vacuum, (b) in heptane, (c)-(g) in solutions containing increasing concentrations of pyridine in heptane and (h) after drainage and evacuation at ca. 300 K. (5’) Spectrum of pyridine in solution in heptane. suggesting that the hydroxy groups were involved in hydrogen-bonding interactions with adsorbed pyridine at the solid/liquid interface.The narrowness of the regenerated band at 3655 cm-l is consistent with data in fig. 1 and 2 which show that hydroxy groups responsible for a band at 3700 cm-l [present only as a shoulder in fig. 3(a)] were not regenerated as unperturbed groups after removal of the liquid phase and evacuation. Spectra of the hydroxylated rutile in pyridine solutions exhibited three infrared bands in the 1500-1700 cm-l spectral range (fig. 3). The bands at 1600 and 1580 cm-l also observed for the largely dehydroxylated rutile (fig. I) were accompanied by an additional band at ca. 1595 cm-l due to a vibration of adsorbed pyridine molecules involved in hydrogen-bonding interactions with surface hydroxy groups. Pyridine linked to the surface by hydrogen bonds was desorbed after drainage and evacuation [fig.3(j)]. However, a sharp residual band at 1605 cm-l with a weak band at 1575 cm-1 indicated that pyridine liganded to Lewis-acidic surface sites6 was resistant to desorption. Hydroxy groups responsible for an infrared band at 3410 cm-l in spectra of rutiles were unaffected by the immersion of rutile in heptane or by the adsorption of pyridine at the rutile/heptane interface. The dominant feature of the infrared spectrum of rutile treated with silicon tetrachloride and water vapour was a maximum at 3740 cm-l due to the OH stretching vibrations of silanol groups [fig. 4(a)]. The maximum shifted to 3695 cm-l (ATOH = 45 cm-l) when the modified rutile disc was immersed in heptane [fig. 4(b)].Subsequent adsorption of pyridine resulted in decreases in the intensity of the band at 3695 cm-l and the2574 I. R. Study of RutilelHeptane appearance of two bands at 1595 and 1580 cm-l due to vibrations of pyridine molecules [fig. 4(ck(g)]. The gamer is ascribed to pyridine involved in hydrogen-bonding interactions with silanol groups. The latter was partly due to pyridine in solution and partly due to the adsorbed pyridine molecules. The presence of only a weak shoulder at 1600 cm-l [fig. 4(g)] suggests that Lewis-acidic sites on rutile were largely unavailable as adsorption sites for pyridine after the surface modification treatment. This was confirmed by the weakness of the residual band at 1605 cm-l [fig. 4(h)] after drainage of the liquid phase and evacuation.The restoration of the maximum at 3740 cm-l was consistent with the desorption of pyridine involving the breaking of hydrogen bonds between silanol groups and pyridine molecules. Weak residual bands were also present in spectra at 1640 and 1540 cm-l, suggesting that the surface modification had generated a low population of Brsnsted-acidic hydroxy groups.5$ A weak band at 1640 cm-l was detectable in spectra before removal of the liquid phase, showing that the Brarnsted sites were sufficiently acidic to protonate pyridine at the rutilelheptane interface. The band at 1540 cm-l was obscured by an intense absorption maximum due to liquid heptane. Discussion The AvOH shift of 70 cm-1 for hydroxy groups giving the infrared band at 3700 cm-l which occurred when rutile was immersed in heptane compares with the corresponding shift of 50cm-l for hydroxy groups giving the band at 3655 cm-l.The relative magnitudes of AToH for surface hydroxy groups perturbed by adsorbate molecules have been related to the Brsnsted acidities of surface hydroxy groups in the sense that a bigger AvOH value implies a lower pK, or higher acidity.ll The conclusion that hydroxy groups giving the infrared band at 3700 cm-l are more acidic than groups giving the band at 3655 cm-l is compatible with the suggestion that the former exist at apex, step or edge sites,8 since it is reasonable to propose that proton loss to leave a residual surface 02- may be slightly more favoured at these sites than at row A sites8 on the (110) surface planes of rutile.Despite this neither type of hydroxy group exhibited detectable Brarnsted acidity in the presence of adsorbed pyridine or triethylamine4 at the rutile/heptane interface, and therefore the groups must be only weakly acidic. The disappearance of the infrared band at 3700 cm-l after the adsorption of pyridine conforms with previous results for other adsorbates4! 9 9 l2* l3 which showed that hydroxy groups responsible for this band were particularly reactive. Also as before,4, 9 7 l2, l3 hydroxy groups giving the band at 3655 cm-l were less reactive. The formation of hydrogen bonds between row A hydroxy groups in the { 1 10) exposed surfaces of rutile (3655 cm-l band8) and adsorbed pyridine molecules resembled the similar result for the adsorption of ethyl ethan~ate.~ The complete lack of reactivity of hydroxy groups giving an infrared band at 3410 cm-l is consistent with the proposal3? 4 9 lo that these groups are located at sub-surface sites below exposed ( 1 lo} surfaces of the rutile particles.The present results prove conclusively, in accordance with a previous s~ggestion,~ that Lewis-acidic centres on the surface of rutile are active as sites for the adsorption of Lewis bases at the rutile/heptane interface. These sites are believed to be incompletely coordinated Ti4+ cations in the exposed {loo} and { lOl} surface l4 Surface hydroxylation of rutile apparently had little effect on the Lewis acidity. This observation supports the model of rutile hydroxylation in which the generation of hydroxy groups by the chemisorption of water primarily occurs on the { 1 lo} surface planes8 and therefore does not interfere with the behaviour of Lewis-acidic sites on other planes.The high strength of adsorption of pyridine on rutile was previously observed during a study of the adsorption of pyridine vapour on r ~ t i l e . ~ The modified rutile surface contained silanol groups giving an infrared band at 3740 cm-l close to the similar band for isolated silanol groups on silica. The shift of 45 cm-l in the band position when modified rutile was immersed in heptane was similarC. H. Rochester and D. G . Smith 2575 to the corresponding shift of 44 cm-l for pure si1ica.l Furthermore, the adsorption of pyridine on modified rutile predominantly via hydrogen-bonding interactions with surface silanol groups resembles the behaviour of silanol groups on silica particles in contact with pyridine vapour5 or with solutions of pyridine in carbon tetrach10ride.l~ The results show that not only the (1 10) surface planes but also the (100) and { 101) planes were significantly modified by the treatment with silicon tetrachloride and water.It is concluded that the surface behaviour of the modified rutile resembles much more the behaviour of silica than of rutile, although the slight residual Lewis acidity shows that coverage of the underlying rutile surface was not completely effective. The generation of Brsnsted acidic sites on a silica-modified rutile surface was previously reported following a study of the adsorption of pyridine vapour.’ This study confirms that proton transfer also occurs at the rutile/heptane interface to a limited extent.It is unlikely that the acidity results from retention of hydrogen chloride, which is a product of the modification reaction since rutile treated with hydrogen chloride and subsequently subjected to evacuation at 673 K failed to exhibit Br~nsted acidity in the presence of adsorbed pyridine.16 We thank the S.E.R.C. for a CASE studentship and Tioxide International for collab- oration and financial assistance. References 1 C. H. Rochester, Adv. Colloid Interface Sci., 1980, 12, 43. 2 C . H. Rochester, J. Oil Colour Chem. Assoc., 1985, 67, 285. 3 J. Graham, C. H. Rochester and R. Rudham, J. Chem. SOC., Faraday Trans. Z, 1981, 77, 2735. 4 J. Graham, R. Rudham and C. H. Rochester, J. Chem. SOC., Faraday Trans. 1, 1983,79, 2991. 5 E. P. Parry, J. Catal., 1963, 2, 371. 6 G. D. Parfitt, J. Ramsbotham and C. H. Rochester, Trans. Faraday SOC., 1971,67, 1500. 7 G. D. Parfitt, J. Ramsbotham and C. H. Rochester, Trans. Faraday SOC., 1972,68, 17. 8 D. M. Griffiths and C. H. Rochester, J. Chem. SOC., Faraday Trans. I , 1983, 79, 1510. 9 J. Graham, C. H. Rochester and R. Rudham, J. Chem. SOC., Faraday Trans. 1, 1981, 77, 1973. 10 A. Buckland, J. Graham, R. Rudham and C . H. Rochester, J. Chem. SOC., Faraday Trans. 1, 198 1,77, 2845. 1 1 M. L. Hair and W. Hertl, J. Phys. Chem., 1970, 74, 91. 12 D. M. Griffiths and C. H. Rochester, J. Chem. SOC., Faraday Trans. 1, 1978,74,403. 13 D. M. Griffiths and C . H. Rochester, J. Chem. SOC., Faraday Trans. 1, 1977,73, 1913. 14 P. Jones and J. A. Hockey, Trans. Faraday Soc., 1971, 67, 2679. 15 D. M. Griffiths, K. Marshall and C. H. Rochester, J. Chem. SOC., Faraday Trans. I , 1974, 70, 400. 16 G. D. Parfitt, J. Ramsbotham and C. H. Rochester, Trans. Faraday SOC., 1971,67, 3100. Paper 5/1799; Received 16th October, 1985

ISSN:0300-9599    

DOI:10.1039/F19868202569

出版商:RSC    

年代:1986

数据来源: RSC