摘要:

J. Chem. SOC., Faraday Trans. I , 1988, 84(3), 751-764 Photoluminescence Properties of MgO Powders with Coordinatively Unsaturated Surface Ions Masakazu Anpo,* Yoshiaki Yamada and Yutaka Kubokawa Department of Applied Chemistry, College of Engineering, University of Osaka Prefecture, Sakai, Osaka 591 , Japan Salvatore Coluccia and Adriano Zecchina Istituto di Chimica Fisica, Universita di Torino, Corso Massimo d’Azeglio 48, 10125 Torino, Italy Michel Che Laboratoire de Rkactivitk de Surface et Structure, Universite‘ P. et M. Curie, U. A . l106-CNRSl 4 Place Jussieu, Tour 54, 7.5252 Paris Cedex 0.5, France The photoluminescence spectrum of MgO degassed at high temperature has been reinvestigated using a standard JRC-MgO sample because the contribution of low-coordination surface sites (MgZ+,-Ot;) to the observed photoluminescence of the degassed MgO samples has been recently questioned.The JRC-MgO-I sample exhibits two different types of photoluminescence, i.e. one short-lived with a lifetime of ca. lop4 s, the other long-lived with a lifetime of 1-104 s. The effect of the degassing temperature of the sample and of added quencher molecules indicates that the short-lived photoluminescence observed under U.V. excitation is a radiative decay process from the charge-transfer-excited complex (Mgt,-O;,)* with a lower coordination number of four. However, the luminescence observed after U.V. excitation, i.e. a long-lived emission, is a radiative recombination process of photo-produced electrons and holes via defects such as F+centres.Thus, both charge transfer and defect mechanisms account for the photoluminescence of the MgO degassed at high temperatures, although the long-lived emission is not directly measured in the present work owing to its much smaller contribution. Tench and Pottl have found that degassed high-surface-area powders such as MgO photoluminesce when excited by U.V. light with an energy (ca. 4.6 eV) much lower than the band-gap of the bulk solids (ca. 8.7 eV), although the pure solids in single-crystal form show no absorption in the near-ultraviolet, and that such photoluminescence spectra are completely quenched by the admission of air or oxygen. They have also shown that the values for the surface band-gap, Esg, for different surface planes of MgO calculated from the expression of Levine and Mark by considering the surface Madelung constant are in good agreement with the excitation energies of the photoluminescence spectra of the powders. On the other hand, Zecchina and Stone2 have found that in the U.V.reflectance spectra of well degassed powdered MgO, CaO, SrO and BaO the abnormal absorption appears at much lower frequencies than those of bulk crystals. More extensive studieP have shown that the additional absorption and the observed short-lifetime (1 0-6-1 0-3 s) photoluminescence spectra are associated with the charge- transfer transition on the numerous incompletely coordinated oxide ions located on the surfaces of the well degassed powders hv (Mg2t,-O:L)+ (Mg+,,-O,,)*. 75 1 hv’752 Photoluminescence of MgO Powders Previously the observed luminescence had been explained in terms of the existence of intrinsic defects, such as small impurities, or extrinsic lattice defects such as the F+ centre, an electron trapped at a surface anion v a ~ a n c y .~ ' ~ Recently, it has been found that the cis-trans isomerization of but-2-ene is photocatalysed on degassed MgO powders. The correlation between the intensity of photoluminescence and the rate of photocatalysed isomerization indicates that coordinatively unsaturated surface ions play a significant role in the photocatalytic activity of the degassed MgO powder catalyst^.^^^ However, almost at the same time Shvets et aL9 observed photoluminescence with thermoevacuated MgO powder and claimed that the observed photoluminescence spectra seemed to be in better agreement with the presence of surface F+ centres rather than coordinatively unsaturated surface ions.9 Although their data are very limited and not sufficient to upset the charge-transfer mechanism proposed by Tench and C01uccia,~*~ and Stone and Zecchina,, it is necessary to reconsider the mechanism for the photoluminescence of MgO powders degassed at higher temperatures.Fortunately, we have found that the standard JRC-MgO-I catalyst exhibits two different types of photoluminescence, i.e. emissions with short and with long lifetimes, which are associated with the charge-transfer processes and with the presence of surface F+ centres, respectively. Therefore, in this paper we discuss the features and mechanisms of the observed photoluminescence of MgO powders outgassed in vacuum at various temperatures.Experimental Samples of MgO microcrystals (JRC-MgO-I) were supplied from the Catalysis Society of Japan as a standard catalyst.'' MgO samples were commercially produced from sea water (MgO purity 99.02 YO ; major impurities Ca, Si, and Fe; B.E.T. surface area ca. 40 m2 g-l; bulk specific density 0.42 g ~ m - ~ ) and were degassed for 2 h at the desired temperature. The rate of increase of the degassing temperature was ca. 1 K min-l and ultimate pressures of ca. Pa) were attainable. The photoluminescence spectra were recorded at 293-298 K using a Shimadzu RF-50 1 spectrofluorophotometer (equipped with 500 W Xe lamp as excitation source) with a resolution of 0.3 nm, equipped with colour filters to eliminate scattered light." E.s.r.measurements were carried out at 77 K using a JES-ME-1 @-band) spectrometer. Details of the experimental procedures have been described previously. 7 9 l2 The gases, H, and CO (Takachiho Kogyo Co., 99.9%) were used without further purification and passed through a liquid-nitrogen trap before use. Commercial oxygen was purified by low- temperature distillation. Deionized double-distilled H,O was degassed by alternate freezing and thawing in vacuo. Torr ( Results Effect of the Degassing Temperatures on the Photoluminescence Properties The degassed MgO sample exhibits a photoluminescence spectrum at ca. 340-450 nm when it is excited with U.V. light with 240-280 nm wavelength. Fig. 1 shows the photoluminescence spectrum of the MgO sample degassed at 873 K when it is excited at 280 and 240 nm, respectively.The photoluminescence spectrum of MgO is dependent on the excitation energy, i.e. photoluminescence recorded under excitation at 240 nm is much higher in intensity and much shorter in wavelength than that obtained at 280 nm excitation. As suggested by Coluccia13 and Anpo et aZ.,14 these results indicate that at least two different emitting sites might be involved. As will be described later, the emission observed when MgO is excited by 280 nm radiation seems to be associated with the surface OH groups;** l5 also, with this MgO sample, the photoluminescence yieldM. Anpo et al. 753 wavelength/ nm Fig. 1. Photoluminescence spectra at 298 K of the MgO sample degassed at 873 K for 2 h: (a) excitation 240& 10 nm, recording range 500 mV, (b) excitation 280+ 10 nm, recording range 50 mV.200 250 300 300 350 400 450 500 Fig. 2. Photoluminescence spectrum and its corresponding excitation spectrum at 298 K of MgO sample degassed at 873 K for 2 h (slit width 7.5 nm, recording range 500 mV). wavelength/nm obtained with 240 nm excitation is always much higher than that with 280 nm excitation. Therefore, in the following, we consider mainly the photoluminescence obtained with 240 nm excitation, since to discuss the mechanism of this emission is the purpose of this work. Fig. 2 shows the photoluminescence spectrum of the MgO sample degassed at 873 K754 Photoluminescence of MgO Powders 200 250 300 350 400 450 500 wavelength/nm Fig. 3. Photoluminescence spectrum and its excitation spectrum at 293 K of an MgO sample degassed at 1173 K for 2 h (slit width 7.5 mn, recording range 500 mV). 300 350 400 450 500 wavelengt h/nm Fig.4. Photoluminescence spectra at 293 K of MgO samples degassed at various temperatures for 2 h (excitation 240& 10 nm): 1, degassed at 473 K; 2, degassed at 673 K; 3, degassed at 773; 4, degassed at 1 173 K; 5, degassed at 1273 K; recording range 500 mV for samples 1 , 4 and 5, and 200 mV for 2 and 3.M. Anpo et al. 755 273 473 673 873 1073 1273 degassing temperaturelK Fig. 5. Effects of degassing temperature on the photoluminescence spectrum of MgO: change in intensity (-) and A,,, of the emission spectrum (---) (experimental conditions as for fig. 4. *, MgO was degassed for 3 h rather than 2 h.and its corresponding excitation spectrum at 293 K. Fig. 3 shows the photoluminescence spectrum of MgO degassed at 1173 K and its corresponding excitation spectrum at 299 K. It is clear from these figures that A, of the photoluminescence spectra are different, but that Amax of the excitation spectrum scarcely changes. Fig. 4 shows the photoluminescence spectrum of MgO samples degassed at various temperatures. Fig. 5 shows the relative photoluminescence yields and the Amax values of the photo- luminescence versus the degassing temperatures. It is clearly seen from fig. 4 that on increasing the degassing temperature of the MgO samples the observed photo- luminescence spectrum changes drastically in intensity as well as in wavelength of A, of photoluminescence. As the degassing temperature of the samples increases the intensity of the photoluminescence increases (passing through a maximum at 1173 K) and then decreases, while A, of the photoluminescence spectrum shifts towards shorter wavelength and then levels off at 335 nm for degassing temperatures above 1073 K.During the degassing treatment of samples at various temperatures described above analysis of the desorbed gases was undertaken. The major desorbed gas was found to be H,O and the minor desorbed gas was CO,. H,O desorption was observed above a degassing temperature of 473 K, mainly at 673-873 K. After degassing the sample at 873 K and performing successive degassing treatments above this temperature only a small amount of desorption gas was observed. These results suggest that the removal of surface OH groups from the sample as H,O is closely associated with the appearance of photoluminescence, i.e.photoluminescence behaviour is an intrinsic property of MgO samples. In connection with these results, the effect of the adsorption and desorption of H,O on the photoluminescence spectrum of MgO samples has been investigated. The photo-756 Photoluminescence of MgO Powders 273 473 673 073 1073 1273 degassing tempera t we/ K Fig. 6. Effects of evacuation temperature on the recovery of the intensity of the photoluminescence spectrum of a hydrated MgO sample prepared by adsorption of H,O at 298 K (spectra were recorded at 298 K, water vapour was adsorbed at room temperature until a pressure of a few Torr was reached, after which the evacuation was carried out first at room temperature.The 330 nm emission (0) was excited at 250 nm, and the 420 nm emission (*) was excited at 270 nm; evacuation was carried out for 30 min for each temperature). luminescence described above was completely quenched by the adsorption of H,O at 293 K. However, this photoluminescence was slowly recovered by evacuation of the hydrated MgO sample at various temperatures. Fig. 6 shows the recovery of the photoluminescence intensity with increasing evacuation temperature of the hydrated MgO sample. It is clear that the intensity of the photoluminescence with A, x 330 nm increases slowly with evacuation temperatures up to 1173 K and then decreases with further increase of the evacuation temperature, in good agreement with the feature shown in fig.4 and 5. The photoluminescence spectra observed after degassing above 873 K were identical to the original spectrum, i.e. before the adsorption of H,O. On the other hand, the intensity of the emission with A,,, x 420 nm as a shoulder of the observed emission increases with increasing evacuation temperature of the hydrated MgO sample, passing through a maximum at 573 K, and then decreases with increasing evacuation temperature. This decrease in the intensity of the 420 nm emission band correlates well with an increase in the intensity of the 330 nm emission band. The new luminescence around 420 nm had A,,, x 280 nm in its excitation spectrum. According to Coluccia et aZ.16 and Duley,15 this new luminescence at ca.400-470 nm observed with the hydrated MgO sample could be linked to the presence of surface OH- ions in specific low-coordination sites on the sample, i.e. the emission is either from the lowest quartet state of the hydroxyl radical1' or from a low-lying triplet state of the hydroxide ion.18 Recently, the authors have investigated hydroxyl groups on MgO powders by means of i.r. spectroscopy and microgravimetry, and have shown that there are two different types of OH- groups on the hydrated MgO powders after the evacuation of the sample at temperatures above 473 K.19 These OH- groups show absorption at 3450-3650 cm-l and at 3740 cm-l, and have been attributed to the OH- ions on extended planes and those on corners and edges on MgO powders, respectively. These were removed by evacuation in the same temperature region used for the data of fig.5. Thus, these results, together with those shown in fig. 4, clearly indicate that the photoluminescence at ca. 330-420nm appears at the stage of surface dehydroxylation. In other words, theM. Anpo et al. 757 intrinsic surface sites associated with the photoluminescence are formed by the removal of the surface OH- groups as in the following reaction: Garrone et aL20 have reported the results of U.V. reflectance spectroscopy which show that the bands related to the sites in a specific low-coordination state appear at the final stages of surface dehydroxylation. This is in good agreement with the results mentioned above. Coluccia et aL21 have found that highly dispersed MgO, prepared by thermal decomposition of high-purity hydroxide or basic carbonate in vacuo, and outgassed at 1200 K for 1 h (100-200 m2 g-l), exhibits photoluminescence with A,,, = 390 nm wavelength when excited at ca.230 nm. They have also found that the MgO sample exhibits another emission at ca. 470 nm when it is excited at 274 nm, which is easily quenched by added H,. They assigned the former excitation to the charge-transfer- excitation processes on lower-coordination surface sites, including Mgtg, 0;; and the latter on sites including Mgi;, Oi;, respectively. The JRC-MgO-I sample seems to be less stable on degassing treatment at higher temperature, 22 because after outgassing at higher temperatures the photoluminescence merely decreased without appearance of any new emission to indicate the formation of further low-coordination surface sites.Quenching of the Photoluminescence with added O,, CO and H, As shown in fig. 7, the addition of 0, at 293-298 K onto the MgO sample which was degassed at 1273 K leads to efficient quenching of the photoluminescence intensity without any change of the shape. For example, the addition of oxygen at only 0.033 Torr quenches the photoluminescence by ca. 70%, the addition of ca. 1 Torr of oxygen leading a complete quenching of the emission. As shown in fig. 6, the evacuation of oxygen at 298 K for 20 min, after complete quenching of the emission, leads to a recovery of most of the emission, but not to complete recovery. It was found that the recovery of the emission seems to depend on the exposure time of the sample to the U.V.excitation beam in the presence of oxygen. Fig. 8 shows the effect of the addition of CO molecules at 293 K on the photoluminescence of an MgO sample degassed at 773 K. With increasing pressure of CO the photoluminescence decreases in intensity without any change in the shape of the emission. Evacuation of the sample at 293 K for 30 min, after complete quenching of the photoluminescence with CO at ca. 5.1 Torr, leads to recovery of most of the pho t oluminescence. Almost identical quenching of the photoluminescence with added 0, or CO molecules was observed with other MgO samples degassed at different temperatures.22 As described previous1y,2*11 two different mechanisms may be involved in the quenching: (1) collisional (or weak-interaction) quenching, whereby gaseous molecules interact with the emitting sites in their metastable excited state; (2) quenching due to the formation of an adsorbed complex between the adsorbed molecules and the excited active emitting sites with different pathways.These two mechanisms might operate in the quenching processes to give a non-radiative deactivation pathway, i.e. to enhance the intersystem crossing to the ground state. For MgO samples, the former mechanism seems to be predominant since the reversible nature of the quenching, i.e. recovery of the photoluminescence after the evacuation of the MgO sample at room temperature for 20-30 min, suggests that the added molecules interact weakly with the active surface sites on the MgO sample, except for some small part being irreversible.As described above, the photoluminescence of the MgO degassed at high temperature is easily quenched by added quencher molecules such as 0,. Thus the photophysicalrelative intensity of photoluminescence : relative intensity of photoluminescenceM. Anpo et al. 759 10 i I 8 e 8 '06 4 2 1 0 0.02 0.04 0.06 0.08 0.1 0.1; pressure of O,/Torr Fig. 9. Plots of (Do/@ values uerms the pressure of 0, in the photoluminescence of MgO degassed at various temperatures. Spectra recorded at 298 K, excitation at 240 nm, degassing temperature : 1, 873 K; 2, 1173 K; 3, 1273 K. (---). Data obtained with V,O, supported on porous Vycor glass at 298 K (lifetime = 230 ps was determined by N, laser).24 processes on the MgO surfaces in the presence of quencher molecules can be described as follows: hv 7 photoluminescence (kp).(Mgz-0;;) -+ (MgL,-O;,)* -+ radiationless decay (k,) L deactivation by quencher (kJ. Thus, the following Stern-Volmer equation is obtained for the yield of photo- luminescence from MgO using the steady-state treatment23 (11) Qo/@ = 1 + tk,C where <Do and <D are yield of the photoluminescence of MgO samples in the absence and presence of quencher molecules, respectively, t , k, and C are lifetime of the excited emitting sites on MgO surfaces, quenching rate constant and concentration of quencher on the surfaces, respectively. In fact, as shown in fig. 9, a,-,/@ in the presence of 0, molecules is a linear function of 0, pressure in the low-pressure region, although there were deviations from linearity at higher pressures (fig.9 shows only in the lower-pressure region).,, In other words, in the low-pressure region, for both 0, and CO, dynamic quenching mainly operates, in which the quenching efficiency is dependent on the amount of quencher adsorbed on the surface and in turn the equilibrium pressure of the added quencher molecules. From the Stern-Volmer plots of the photoluminescence quenching by added O,, the values of tk, are 57, 71 and 85 Torr-' for MgO degassed at 1273, 1173 and 873 K, respectively. t and k, cannot be separately determined. However, by assuming that the quenching rate constant of 0, for the excited emitting sites on MgO is approximately equal to that for the charge-transfer-excited triplet state of V,O, catalyst supported on porous Vycor glass (because both sites locate on the surface of oxide catalysts and the quenching rate is mainly controlled by collisional efficiency with 0,), t can be determined for MgO.By using the lifetime of the charge-transfer-excited triplet state of supported760 Photoluminescence of MgO Powders U.V. light off 0 0.4 0.8 12 1.6 0 20 40 60 80 100 pressure of H2/Torr U.V. irradiation time/min Fig. 10. Effects of the addition of H, (a) and U.V. irradiation (6) in the presence and absence of H, on the photoluminescence spectrum of MgO degassed at 1273 K (spectra recorded at 298 K, excitation 240 nm). V,O,, which is known to be 3.04 x T0rr-l s-l at 293 K,24 it is possible to estimate the lifetime of the photoluminescence of the MgO sample.The lifetimes at 298 K for the MgO samples degassed at 1273, 1173 and 873 K were determined to be 2.2 x lo-', 2.4 x and 3.6 x low4 s, respectively. These values seem to be in good agreement with those estimated by Tench and P0tt.l The addition of H, at 295 K to the MgO sample degassed at 1273 K had only a small effect on the photoluminescence measured with 240 nm excitation. As shown in fig. 10, U.V. irradiation of the MgO sample with a 75 W mercury lamp in the presence of H, at only 0.3 Torr was found to decrease the photoluminescence intensity with U.V. irradiation time. When U.V. irradiation ceased the photoluminescence intensity no longer decreased. U.V. irradiation for 350 min in the presence of H, led to a reduction of the photoluminescence intensity by ca.35 %. Coluccia et aL4 and Tench4 found that the photoluminescence with A,,, = 470 nm (excited by 270 nm radiation), due to (Mg::-Oi;), was easily quenched by the addition of hydrogen onto the MgO sample at 300 K. On the other hand, the photoluminescence with A, = 390 nm, due to (Mgt:-Oi;) (excited by 230 nm radiation), was not affected by the addition of hydrogen at 300 K, but was quenched by the U.V. irradiation of the MgO sample in the presence of hydrogen. According to their assignment, the results obtained with the standard JRC-MgO-I catalysts therefore suggest that the unsaturated surface site of (Mgi:-Oi;) is mainly associated with the observed photoluminescence and that the lowest-coordination surface sites, i.e. (Mg::-O&), are not present at a significant concentration in this MgO sample, as mentioned above.The fact that the lowest- coordination surface sites are not formed on the MgO sample seems to be in good agreement with the results obtained by Ito and To~ninaga,~~ who found that the desorption peak from hydrogen adsorbed on the lowest-coordination sites was not observed with the present MgO sample in t.p.d. experiments.M. Anpo et al. 76 I 9, =2.017 1 1.974 t Fig. 11. E.s.r. spectrum of photo-formed oxygen radical anion species and of the F+ centre in MgO sample degassed at 1173 K. U.V. irradiation was carried out at 77 K in the presence of 0,. Initial 0, pressure was 0.7 Torr. Spectrum was recorded at 77 K. Mn2+ ions in MgO powder were used for g value and sweep calibrations. E.S.R. Studies on the MgO Samples It is well known that F+ centres are formed in the course of u.v.- or y-irradiation of MgO samples in the presence of H, (or in vacuo).26 With the present MgO, F+ centres were observed by e.s.r.at 77 K, merely by thermal evacuation at the higher temperatures. Although the details are not clear, these different features may be attributable to the lower purity in the present MgO samples. The addition of oxygen onto the MgO sample with F+ centres which was degassed at 673 K led to no change in the e.s.r. signal. Shvets et a1.' reported that the addition of oxygen onto the MgO catalyst with F+ centres (formed by y-irradiation) led to the formation of 0; radical anions due to the reaction of F+ centres with 0,, although most of them are stable in 0, a t room temperature.Thus, some different surface properties were observed in these MgO samples. Fig. 11 shows the e.s.r. spectrum measured at 77 K after U.V. irradiation of the MgO sample at 77 K in the presence of 0,. A new signal due to oxygen is observed, while no change is seen in the signal due to F+ centres. According to Lunsford and Wong2' the new signal due to oxygen could be assigned to the formation of 0; radical anions, which arise from the reaction of 0, with 0- hole centres (these are produced when the 0;; ion gives up an electron). The 0; radical anions are stable at room temperature on the MgO surfaces, in contrast to those formed on transition-metal oxides supported on SiO, or porous Vycor glass.28129 These results suggest that the formation of stable 0; is closely associated with the quenching behaviour of the photoluminescence by the addition of oxygen.However, it seems that this is not the case with the MgO photoluminescence quenching because Tench3' found that treatment of an MgO sample with a low pressure of hydrogen destroyed the ability to produce such 0- ions without affecting the luminescence properties. He also found that the total oxygen uptake by the MgO sample during U.V. irradiation was ten times higher than the amount of 0; produced. From these results, he suggested that no paramagnetic species of oxygen are involved in the quenching mechanism, and that oxygen may be on the surface in the form of ions such as 0;-. Further studies are necessary to settle these problems.762 Photoluminescence of MgO Powders .-...i I ,\ 0 400 600 wavelengt h/nm Fig. 12. Luminescence spectrum of the u.v.-irradiated MgO samples degassed at various temperatures: (a) 600 "C, (b) 800 "C, (c) 1000 "C. U.V. irradiation was carried out at room temperature in UQCUO for 20 min, and 1-104 s later luminescence was recorded. Discussion Fig. 12 shows the luminescence spectra obtained by Yanagisawa with the same MgO sample (JRC-MgO-I catalysts) degassed at 873, 1073 and 1273 K, re~pectively.~~ These emission spectra were recorded 1-104 s after stopping the U.V. irradiation of the MgO sample at room temperature in vacuo for 20 min with a low-pressure mercury lamp, but not recorded under U.V. excitation, like those mentioned above. In other words, the emission have considerably longer lifetimes, in the range of 1-104 s.In contrast with the features mentioned above, the emission exhibits the following characteristics. A,,, of the emission was almost constant at ca. 400 nm when changing the outgassing temperatures of the MgO sample. The intensity of the emission was highest with the MgO sample degassed at 673 K, decreasing with increasing degassing temperature up to 1273 K. Such features are completely different from those in the photoluminescence spectra obtained under U.V. excitation shown in fig. 4 and 5. Yanagisawa also found that the addition of D, onto the MgO sample led to enhancement of the emission and its extent increases with the pressure of D,. Yanagisawa and H ~ z i m u r a ~ ~ have shown that such a long-lived luminescence is caused by a radiative recombination process of the photo-formed electrons and holes on the defect, such as F+ centres.it is well known that the bulk emission from alkaline-earth- metal oxides such as MgO is caused by radiative recombination of electrons and holes via defects such as F+ and Fo centres. The long-lived emission, which is observable even after stopping the U.V. excitation, clearly indicates that the emission arising from the radiative recombination of photo-produced electrons and holes on the defect sites is also operating with the present JRC-MgO-I samples. The photoluminescence yield obtained under U.V. excitation, i.e. short-lived luminescence, was found to be much higher than As describedM. Anpo et al. 763 that of the emission observed after stopping the U.V.irradiation, i.e. long-lived emission, by ca. two orders of magnitude.33 This result suggests that the former process is much more important than the latter for the deactivation of the excited states of MgO powders. Thus, all these results clearly show that there are two different radiative processes for the degassed JRC-MgO sample. The first radiative process is very sensitive to the added 0, and CO molecules but not to added H, molecules, and is only observable under U.V. excitation, i.e. its lifetime is very short and in the range of lop4 s. The second radiative process is rather insensitive to the added 0, and CO molecules (but added hydrogen enhances the intensity) and is observable even after stopping U.V. irradiation, i.e. its lifetime is long and in the range 1-104 s.Shvets et al.’ observed the photoluminescence with the thermoevacuated powder MgO and claimed that the data are shown to be in better agreement with the attribution of the luminescence at 415 nm to surface F+ centres, rather than to coordinatively unsaturated surface ions, although their data are not better than those reported by Tench and coworker^.^'^ They reported that the adsorption of H, or 0, at room temperature and subsequent outgassing for 15-20 min did not change the luminescence spectrum. Their results are completely different from those obtained by Tench and c o ~ o r k e r s ~ * ~ and those obtained in the present work.? They explained that the disagreement with the results obtained by Tench and Coluccia could be caused by impurities in the H, used by them.However, such an explanation is not only unsound but is also untrue, as indicated in the present work. Although it is not possible to discuss the results reported by Shvets et al.’ because of a lack of experimental detail, it seems that they have observed only the long-lived luminescence, but not the short-lived photoluminescence. As a result, the data obtained by Shvets et al.’ might show better agreement with the defect luminescence mechanism. It seems likely that the yields of the short- and long-lived photoluminescence would change from sample to sample and could depend on the pretreatment of the samples. It is well known that the lower the coordination around a surface oxide ion, the lower will be the frequency of the absorption of the charge-transfer transition associated with it.This is because the Madelung potential progressively decreases as the extent of unsaturation increases. According to Tench and coworker^^.^ the photoluminescences observed at ca. 390 and 470 nm are due to the radiative decay processes of the absorbed photon energy on the coordinatively unsaturated surface site of (Mg,2;-0:;) and that of (Mgi:--Ot;), respectively. With the present work, the photoluminescence at around 390-330 nm (excited at around 240 nm), which could be attributable to a radiative decay process associating with the surface unsaturated sites of (Mg:g-O:;), is observed, and the other photoluminescence, which is attributed to the presence of surface OH groups, is also observed at around 430nm (excited at around 280nm).However, photo- luminescence which is attributable to the radiative decay process associating with the much lower coordination site, (Mgi:-Oi,) was not observed, even with the MgO sample degassed at 1273 K. As described above, the MgO sample used in the present study is known not to be stable for degassing at higher temperatures. Therefore, the reason why the photoluminescence associating with the ( M g ~ ~ - O ~ J is not observed with this MgO might be connected with this instability of the sample. Further studies are in progress using other MgO samples involving a standard JRC-MgO-11. M.A. thanks Prof. H. Hattori of Hokkaido University and Prof. T. Hattori of Nagoya University for supplying standard JRC-MgO samples, and Prof.Y. The spectra obtained by Shvets et d’seem to be in better agreement with those obtained by Tench and coworker^^-^ than with the present work. This may be because spectra from the sample used in the present work are more sensitive to excitation energy and pretreatment of the sample.764 Photoluminescence of MgO Powders Yanagisawa for his permission to use his data before publication. Thanks are due to The Ministry of Education of Japan (Grant-in-aid for Scientific Research, grant nos. 59470007, 59550558, and 61223022). M.A. thanks UniversitC P. et M. Curie for a position as an invited Professor at Paris, and Universita di Torino for an invitation to visit Torino. References I A. J. Tench and G. T. Pott, Chem. Phys. Lett., 1974, 26, 590. 2 A. Zecchina, M.G. Lofthouse and F. S. Stone, J. Chem. SOC., Faraday Trans. I , 1975,71, 1476; F. S. Stone and A. Zecchina, Proc. 6th Int. Congr. Catal. (The Chemical Society, London, 1976), A8; A. Zecchina and F. S . Stone, J. Chem. SOC., Faraday Trans. I , 1976, 72, 2364; 1978, 74, 2278. 3 S. Coluccia, J. F. Hemidy and A. J. Tench, J. Chem. SOC., Faraday Trans. I , 1978, 74, 2763. 4 S. Coluccia, M. Deane and A. J. Tench, J. Chem. SOC., Faraday Trans. I , 1978, 74, 2913; S. Coluccia and A. J. Tench, Proc. 7th Int. Congr. Catal. (Kodansha, Tokyo, 1981), B, 1154. 5 J. Cunningham, in Comprehensive Chemical Kinetics, ed. C. H. Bamford and R. G. Compton (Elsevier, Amsterdam, 1984), chap. 3; J. Nuran, J. Cunningham, A. M. Deane, E. A. Colbourn and W. C. Mackrodt, in Adsorption and Catalysis on Oxide Surfaces, ed.M. Che and G. C. Bond (Elsevier, Amsterdam, 1985), p. 83; J. Cunningham and C. P. Healy, J. Chem. SOC., Faraday Trans. 1, 1987,83, 2973. 6 M. Che, in Adrorption and Catalysis on Oxide Surfaces, ed. M. Che and G. C. Bond (Elsevier, Amsterdam, 1985), p. 11; M. Che and A. J. Tench, A h . Catal., 1982, 31, 77. 7 M. Anpo, Y. Yamada and Y. Kubokawa, J. Chem. SOC., Chem. Commun., 1986, 714. 8 M. Anpo and Y. Yamada, in Advances in Basic Solid Materials, ed. K. Tanabe (Elsevier, Sequoia, 1987), and unpublished data. 9 V. A. Shvets, A. V. Kuznetsov, V. A. Fenin and V. B. Kazansky, J. Chem. SOC., Faraday Trans. I , 1985, 81, 2913. 10 A standard JRC-MgO-I sample was supplied from Catalysis Society of Japan. Various data about this MgO sample are available from a Data Book of 9th Reference Meeting of Catalysis Society of Japan, (Toyama, Tokyo, 1985).11 M. Anpo, C. Yun and Y. Kubokawa, J. Chem. SOC., Faraday Trans. I , 1980, 76, 1014. 12 M. Anpo, Y. Yamada and Y. Kubokawa, in a Data Book of 9th Reference Meeting of Catalysis Society 13 S . Coluccia, in Adsorption and Catalysis on Oxide Surfaces, ed. M. Che and G. C. Bond (Elsevier, 14 M. Anpo, M. Kondo, Y. Kubokawa, C. Louis, M. Che and S. Coluccia, Chem. Expr., 1987,2,61, the 15 W. W. Duley, J. Chem. SOC., Faraday Trans. 1, 1984, 80, 1173. 16 S. Coluccia, M. Deane and A. J. Tench, Proc. 6th Int. Congr. Catal. (The Chemical Society, London, 17 H. J. Maria and S. P. McGlynn, J. Chem. Phys., 1970, 52, 3402. 18 P. B. Merkel and W. H. Hamill, J. Chem. Phys., 1971, 55, 2174. 19 S. Coluccia, L. Marchese, S. Lavagnino and M. Anpo, Spectrochim. Acta, 1987, in press. 20 E. Garrone, A. Zecchina and F. S. Stone, Philos. Mag., Sect. B, 1980, 42, 683. 21 S. Coluccia, A. J. Tench and R. L. Segall, J. Chem. SOC., Faraday Trans. I , 1979, 75, 1769. 22 M. Anpo and Y. Yamada, unpublished data. 23 N. J. Turro, in Modern Molecular Photochemistry (Benjamin/Cummings, Menlo Park, 1978). 24 M. Anpo, I. Tanahashi and Y. Kubokawa, J. Phys. Chem., 1982, 86, 1; M. Anpo, T. Suzuki, 25 T. Ito and N. Tominaga, in a Data Book of 9th Reference Meeting of Catalysis Society of Japan 26 R. L. Nelson, A. J. Tench and B. J. Harmsworth, Trans. Faraday SOC., 1967, 63, 1427; R. L. Nelson 27 J. H. Lunsford, Catal. Rev., 1973, 8, 135; N. B. Wong and J. H. Lunsford, J. Phys. Chem., 1966, 44, 28 M. Anpo, N. Aikawa, Y. Kubokawa, M. Che, C. Louis and E. Giamello, J. Phys. Chem., 1985, 89, 29 M. Anpo, T. Fujii, S. Suzuki and Y. Kubokawa, J. Phys. Chem., 1984, 88, 2572. 30 A. J. Tench, Proc. 6th Int. Congr. Catal. (The Chemical Society, London, 1976), p. 182. 31 Y. Yanagisawa, in a Data Book of 9th Reference Meeting of Catalysis Society of Japan (Toyama, 32 Y. Yanagisawa and R. Huzimura, J. Phys. SOC. Jpn, 1984, 53, 66. 33 M. Anpo and Y. Yanagisawa, unpublished data. of Japan, (Toyama, Tokyo, 1985), p. 25. Amsterdam, 1985), p. 59. revised manuscript was submitted to J, Phys. Chem. 1976), A9. I. Tanahashi, M. Kondo and Y. Kubokawa, Shokubai (Catalysis), 1986, 28, 68. (Toyama, Tokyo, 1985), p. 21. and J. W. Hale, Discuss. Faraday SOC., 1971, 52, 77. 1487. 5689; M. Anpo, unpublished data. Tokyo, 1985), p. 29. Paper 71293; 17th February, 1987

ISSN:0300-9599    

DOI:10.1039/F19888400751

出版商:RSC    

年代:1988

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I , 1988, 84(3), 765-772 Dehydrogenation of Alcohol on Hydride-forming Rare-earth Intermetallic Compounds (RFe, and R,Co,) Hayao Imamura,* Shingo Kasahara, Toshihiko Takada and Susumu Tsuchiya Department of Industrial Chemistry, Faculty of Engineering, Yamaguchi University, 2557 Tokiwadai, Ube 755, Japan The vapour-phase dehydrogenation of methanol, ethanol and propan-2-01 has been studied over hydride-forming rare-earth intermetallic compounds (ErFe,, DyFe,, Nd,Co, and Sm,Co,). Under mild conditions methanol, ethanol and propan-2-01 were selectively dehydrogenated to carbon monoxide, acetaldehyde and acetone, respectively, forming metal hydrides simultaneously. The dehydrogenation in the presence of these hydride- forming compounds is recognized as a useful dehydrogenative method, in which R,Co, and RFe, are amenable to conversion into stable metal hydrides.The dissociative chemisorption of alcohol on the activated alloy surfaces with subsequent migration of the liberated H to the underlying alloy phase is expected. The dehydrogenation of methanol and ethanol is by a rate-limited dissociation reaction. The behaviour of the reaction has also been characterized by isotope techniques. Some rare-earth intermetallic compounds containing 3d transition metals have been found to absorb and desorb rapidly relatively large quantities of hydrogen.'.' LaNi, is considered the paradigm among hydrogen absorbers. Since hydrogen is absorbed dissociatively, the gas must exist, at least fleetingly, as monatomic hydrogen on the surface.This suggests that the surfaces of the rare-earth intermetallics are quite active, which indicates their possible use as hydrogenation and dehydrogenation catalyst^.^ Several reactions are possible by making use of their characteristics : they contain reactive H atoms able to hydrogenate efficiently unsaturated compounds as a means of extracting hydrogen,* and vice versa, hydrogen produced by dehydrogenation can be absorbed intact to form metal hydride~.~ The direction of the reactions is determined thermodynamically. The catalytic properties of these compounds have been studied for hydrogenation, but less so for dehydrogenation. Accordingly, in this work we studied the dehydrogenation reaction over the hydride-forming rare-earth-3d intermetallic compounds. The dehydrogenation reaction is in most cases thermodynamically unfavourable under moderate conditions, but the dehydrogenation equilibrium can often be made favourable by the adroit use of hydride-forming materials.We have proposed the system of the hydride-forming rare-earth intermetallic compounds as a useful dehydrogenation method ;5* it is the hydride formation which provides the driving force for the reaction. We have previously found that the dehydrogenation of propan-2-01 over R,Co, (where R is a rare earth) results in enhanced activity at 3 17 K owing to simultaneous formation of metal hydrides. This type of dehydrogenation also shows interesting features which would appear useful for organic synthesis. ' In the reaction the alloy itself effectively displays abilities of both (i) undergoing catalytic dehydrogenation on the surface and (ii) accepting large amounts of hydrogen in the bulk.In addition, there is necessarily reasonable contact between the active sites and the hydrogen-acceptor sites in order that hydrogen spillover is properly operative.' In the case of dehydrogenation, mixtures of active catalysts and hydride-formers 765766 Alcohol Dehydrogenation on Rare-earth Intermetallic Compounds (Pt/Al,O,-Ti8 or Pt/Al,O,-CaNi,Q) have been examined with similar intentions ; the hydride-forming compound alone shows no activity and separate catalysts are necessarily required. The rare-earth-3d intermetallic compounds are suitable to use for the reaction. However, for this purpose the alloys must be highly activated by successive treatments with solutions of a dihalogenoethane (1,2-dibromoethane or 1,2-di- iodoethane) and sodium hydroxide.Surface studies of R,Co, have shown that this treatment produces a surface that is enriched in Co compared with the bulk.6 It seems that Co precipitated out on the alloy surface acts as a catalyst for dehydrogenation and that dissociated hydrogen species migrate via these active sites to the underlying intermetallic phase by hydrogen spillover. In this study we investigated the dehydrogenation of three alcohols (methanol, ethanol and propan-2-01) in the presence of hydride-forming rare-earth intermetallic compounds (ErFe,, DyFe,, Nd,Co, and Sm,Co,). On the basis of the migration of dissociative hydrogen from the alcohol to the intermetallic phase, we extensively studied the characteristics of this type of dehydrogenation by the combined use of isotopic labelling studies.Experimental Materials The hydride-forming materials used for the dehydrogenation were ErFe,, DyFe,, Nd,Co, and Sm,Co,. These compounds were supplied from the Santoku Metal Industries Co. Ltd and the Shin-Etsu Chemical Co. Ltd. After annealing for homogenization, X-ray diffraction patterns confirmed the existence of the desired compounds and the absence of extraneous phases. The intermetallic compound was then broken down into a powder of particle size 100-200 ,um for subsequent activation. Methanol, ethanol and propan-2-01 were of a special grade and were degassed several times before use. Procedures Activation of the alloy was conducted in a 200 cm3 glass flask, provided with a burette for the addition of 1,2-dibromoethane (DBE) or 1,2-di-iodoethane (DIE), a stirring device and a reflux condenser.6~'o DBE or DIE solutions were added dropwise to a suspension of the alloy powders in tetrahydrofuran (THF) under anhydrous conditions, and the mixture was stirred at refluxing temperatures.The resulting alloy powders were then washed by decantation with 50 cm3 portions of THF, followed by treatment with 20% aqueous sodium hydroxide solutions. The alloy was further washed in distilled water until the pH of the wash water was 7. The washed samples were dried in vacuo at room temperature and were stored under dry argon. The B.E.T. surface areas of the alloy thus treated increased compared with those of the original alloys (ErFe,, 2.4; DyFe,, 1.8; Nd,Co,, 2.0 and Sm,Co,, 2.2 m2 g-').Dehydrogenation Reaction Alcohol dehydrogenation was studied using a gas circulation reactor. The activated alloy (0.2-1.1 g) was transferred to the reactor without exposure to air, outgassed and then brought into contact with alcohol vapour. To follow the reaction the reacting gas was periodically collected and was analysed by a gas chromatograph [tricresyl phosphate (TCP)/Uniport B (60/80), active alumina (60/80) and Porapak Q (50/80)].H . Imamura, S. Kasahara, T. Takada and S. Tsuchiya 767 Results and Discussion Alcohol Dehydrogenation The dehydrogenation of an alcohol is generally an endothermic process, and hence equilibrium is unfavourable unless the temperature is raised.However, the position of equilibrium is progressively shifted to the product side even under milder conditions when the overall reaction includes the heat of formation of metal hydrides (MH,) through transfer hydrogenation as follows : 2 2 RR’CHOH + - M -+ RR’CzO + - MH,. X x Accordingly the dehydrogenation of methanol, ethanol and propan-2-01 over various hydride-forming compounds (ErFe,, DyFe,, Nd,Co7 and Sm,Co,) was carried out. The results of the dehydrogenation in the presence of DyFe, are presented in fig. 1 (a)-(c). That gaseous hydrogen was not detected .throughout the dehydrogenation was a common phenomenon in all the reactions studied; therefore the gas pressure was held constant during dehydrogenation. Most of the hydrogen formed from dissociatively chemisorbed alcohol was successively absorbed in these hydride-forming alloys.This confirms that the overall stoichiometry is given by reaction (1). When the reaction was carried out at higher temperatures one could observe hydrogen in the gas phase owing to the dissociation pressures of the metal hydrides formed. DyFe, was recognized to be amenable to conversion into hydrides stable to ca. 450 K, depending on the dissolved hydrogen concentrations. Irrespective of changes in the alloys, propan-2-01 and ethanol were dehydrogenated to form acetone and acetaldehyde with ca. 100 % selectivity, respectively. This is probably due to a reduction in the number of side-reactions compared with normal dehydrogenation at elevated temperatures.Methanol was also dehydrogenated at 363-443 K to carbon monoxide very selectively. There was no indication of other dehydrogenated intermediates or products such as methane, formaldehyde, methyl formate or carbon dioxide, observed in conventional dehydrogenation. l1 The alloys were readily poisoned by the reaction products, i.e. acetone and acetaldehyde. Poisoning of RFe, with carbon monoxide was less pronounced. For the dehydrogenation of propan-2-01 the initial rate law obtained by adding different pressures of acetone (Pa) gave linear plots of (l/rate) against Pa. Retardation of the reaction was observed for these products rather than a decrease in the hydrogen capacity of the alloy. The presence of carbonyl compounds more strongly adsorbed than alcohol will inhibit the reaction.12 As can be seen in fig.l(a)-(c), reactivity of alcohol over RFe, increased with increasing molecular weight ; methanol < ethanol < propan-2-01. It is generally agreed that a secondary alcohol is more reactive than a primary one towards catalytic dehydrogenation by metals.13 Both the R,Co, and RFe, systems exhibited essentially the same behaviour, although their dehydrogenation activities were significantly different (table 1). R2C07 exhibited an activity over an order of magnitude higher than RFe, for each alcohol. We suppose that 3dmetals that precipitate out on the alloy surfaces as a result of the activation treatment are important factors in determining dehydrogenation behaviour.6 The order of reactivity agreed with that of normal catalytic dehydrogenation over transition metals.l* The hydride stability and dehydrogenation properties varied with the alloys used.The hydride stability varies in the order R,C0715 < RFe,,16 and hence the RFe, system is suitable for this type of reaction at elevated temperatures. However, the surface activity is also important, as shown by the effect of the addition of promoter metals.6768 Alcohol Dehydrogenation on Rare-earth Intermetallic Compounds 0 l 1 1 0 50 100 150 time/min n 0.003 d: E \ 0.002 s z e 0.001 9 2 n 3 - 8 3 s 3=" z e s1 - 0.003 - 0.002 - 0.001 0 0 50 100 time/min 20 I (C) I n - 0.015 6 15 - 1 E 3 s - 0.010 5 E - 0.005 9 3 -0 0 50 1 00 time/min Fig. 1. (a) Dehydrogenation of methanol (40 Torr) over DyFe, (1.1 g) at 423 K: 0, carbon monoxide; ., gaseous H,; 0, absorbed H,.(b) Dehydrogenation of ethanol (21 Torr) over DyFe, (0.8 g) at 413 K: 0, acetaldehyde; ., gaseous H,; a, absorbed H,. (c) Dehydrogenation of propan-2-01 (22 Torr) over DyFe, (0.8 g) at 413 K: 0, acetone; ., gaseous H,; a, absorbed H,.H. Irnarnura, S. Kasahara, T. Takada and S. Tsuchiya Table 1. Results for the dehydrogenation of alcohols by RFe, and R2C07 769 activity/mmol min-l g-' alloy T/K methanol ethanol propan-2-01 ~ ErFe, 423 3 . 6 ~ - 4.7 x 10-4 413 1.2 x 10-5 7.7 x 10-5 3.4 x 10-4 DyFe, - - 1.1 x 10-3 - - 6.9 x 10-3 4.1 x 10-4 356 9.0 x 10-6 2.0 x 10-5 3.5 x 10-3 433 6.3 x 10-5 - - 423 1 . 8 ~ Nd,Co7 317 343 363 3.0 x 1.2 x 2.3 x lop2 - - - - Sm,Co, 333 Mechanistic Interpretations Since hydrogen dissociated from the alcohol is ultimately absorbed in this type of dehydrogenation, it is clear that a sequence of events of considerable complexity is involved in the overall reaction process.First alcohol must be dissociatively adsorbed on the alloy surfaces, followed by participation of the adsorbed alcohol in the hydrogen- transfer step, in which at least the following two processes occur competitively : (i) spillover of atomic hydrogen to the alloy bulk and (ii) H-atom recombination to release to the gas phase. It seems certain that hydrogen spillover precedes the recombination reaction in most cases, particularly for dehydrogenation at lower temperatures. On the other hand, as shown in the reactions of propan-2-01 and ethanol over Nd,Co, at higher temperatures [fig.2 (a) and (b)], the pathway of the recombination of spilt-over hydrogen becomes dominant rather than hydrogen spillover, in which the dissociation pressure of hydrogen for Nd,Co7-H15 at 406 K cannot be overlooked. A comparison of the modes of reaction of ethanol and propan-2-01 dehydrogenation at 406 K is informative (fig. 2). Even under the same reaction conditions, the dehydrogenation of propan-2-01 resulted in a significant increase in pressure, which then progressively decreased owing to hydrogen absorption by the alloy, while for ethanol gaseous hydrogen steadily increased to the dissociation pressure of Nd,Co,-H at 406 K. This reflects differences in the relative ease of the dehydrogenation step of the alcohol used, even though the adsorption of each product significantly affects the later course of the reaction in a different manner.On raising the temperature to 406 K it is apparent that the rates of propan-2-01 dehydrogenation exceed those of hydride formation by Nd,Co,, but this is not the case for ethanol dehydrogenation. Therefore, for the dehydrogenation of ethanol and, moreover, methanol, which is less reactive, the hydride formation process of R,Co7 cannot be a rate-limiting process. Activation energies for the dehydrogenation of alcohol in the presence of several hydride-forming alloys are summarized in table 2, together with the result for SmCo,. This compound scarcely forms the hydride thermodynamically, apart from at lower temperatures. l7 However, dehydrogenation over SmCo, is thermodynamically possible, even in the absence of hydride formation, under the reaction conditions studied (433-483 K ; 20-40 Torr).We confirmed that the dehydrogenation undoubtedly proceeded to release hydrogen in the gas phase. Hence the hydride formation process need not be considered for SmCo,. That the apparent activation energy for methanol and ethanol dehydrogenation over Nd,Co, or Sm,Co,was close to that for SmCo, is consistent with the above speculation. In addition, the values obtained for ethanol and FAR I 26770 Alcohol Dehydrogenation on Rare-earth Intermetallic Compounds 0.4 0.3 ti 0.2 E 0.1 0 0’ 0 50 100 150 200 time/min 0.36 1 2.0 n 0” 0, E s 1.5 0.5 0 0.32 0’ 0 . 0 4 1 ,f 0.15 n 0” u, z s 1 0.10 0.05 0 0 50 100 150 200 Fig. 2. (a) Dehydrogenation of propan-2-01 (27 Torr) over Nd,Co, (0.2 g) at 406 K: 0, propan- 2-01;-0-,acetone; ., gaseous H,, e, absorbed H,.(b) Dehydrogenation of ethanol (27 Torr) over Nd,Co, (0.2 g) at 406 K : A, ethanol; e, acetaldehyde; ., gaseous H,, 0, absorbed H,. time/min methanol dehydrogenation compare well with the activation energies for oxide- supported cobalt or iron.’* The vapour-phase dehydrogenation of propan-2-01 to acetone with R,Co, as hydrogen acceptors was possible even at moderate temperature. AG* at 298 K for the reaction of these alloys with propan-2-01 to form the metal hydride and acetone is negative, indicating that hydride formation provides the driving force for the reaction. This is also reinforced by the observation that the dehydrogenation activity of the alloy markedly decreased with increasing dissolved hydrogen in the alloy.6H.Imamura, S. Kasahara, T. Takada and S. Tsuchiya Table 2. Activation energies for the dehydrogena- tion of alcohols alloy methanol ethanol 88 66 60 Nd,Co, SmCo, 80 56 - DyFe, Sm,Co, 78 59 Table 3. Dehydrogenation of deuteriated methanol over ErFe, activity at 433 K relative methanol E/kJ mol-l /mmol mix1 g-l rate ~~ CH,OH 87 6.3 x 10-5 3 .O CH,OD 88 5.4 x 10-5 2.6 CD,OD 97 2.1 x 10-5 1 77 1 Isotopic Labelling Studies of Methanol Dehydrogenation In methanol dehydrogenation R,Co, and RFe, were completely selective for carbon monoxide formation. The decomposition of methanol to carbon monoxide and hydrogen has been studied on various metal catalysts, in which formaldehyde is proposed as an intermediate ;ll b~ further, the kinetics are probably controlled by the formation process of this intermediate.When formaldehyde was used instead of methanol, the reaction over Nd,Co, and ErFe, yielded mainly carbon monoxide and metal hydrides, the rate being appreciable at temperatures as low as 383 K. The dehydrogenation of formaldehyde was faster by over an order of magnitude than that of methanol. In this case we also observed negligible gaseous hydrogen during the reaction: the hydride formation process cannot be a slow step in methanol dehydrogenation, as in the case of ethanol. However, the formation of a small amount of methyl formate was detected. Therefore, formaldehyde formed at least fleetingly on the alloy surfaces would be subject to the rapid and sequential abstraction of hydrogen in the later course of the reaction before desorption to the gas phase or conversion into methyl formate.This is why formaldehyde and other products were not detected during the reaction. On the other hand, dehydrogenation over ErFe, was conducted with deuteriated methanol under the same conditions, and the results are summarized in table 3. The initial rates of CD,OD conversion were reduced to about a third of those of CH,OH conversion. The apparent activation energies for the reaction also varied from 87 kJ mol-1 for CH,OH to 97 kJ mo1-l for CD,OD. However, when CH,OD was used, only a small kinetic isotope effect was observed in the temperature range studied, and the activation energy was almost equal to that for CH,OH dissociation. The observed kinetic isotope effect is qualitatively consistent with the zero-point energy difference for the C-H us.C-D bond.'' These are compared with the expectation, based on the mechanism of methanol dissociation over nickelll and platinum'l catalysts, in which methanol is thought to undergo successive dissociative dehydrogenations through formaldehyde. Methanol is dehydrogenated via the state of dissociative adsorption, as opposed to the non-dissociative state over copper.11c* 2o Various workers have cited experimental evidence to show that the dissociation of an 0-H bond in adsorbed 26-2772 Alcohol Dehydrogenation on Rare-earth Intermetallic Compounds methanol takes place.,' Blyholder et a1.22 have provided infrared evidence for the existence of adsorbed methoxy groups on Fe/SiO,.Results for the dehydrogenation of CH,OH, CD,OH and CD,OD indicate that hydroxyl hydrogen is preferentially dissociated, while the dissociation of the C-H bond in the adsorbed methoxy intermediate CH,O is kinetically limited. This was further supported experimentally by the ErFe,-catalysed CD,OD-H, reaction at the same temperature as in the case of methanol dehydrogenation. The reaction was conducted by admitting [2H,]methanol at 40 Torr and hydrogen at 40 Torr.? The exchange reaction progressed readily at 423 K and even below methanol dissociation temperatures. Moreover, this was accompanied by rapid hydrogen absorption, indicating that dissociative chemisorption of gaseous H, and the migration of spilt-over hydrogen were not so much affected in the presence of methanol.The exchanged product at low conversion of methanol was predominantly CD,OH, as confirmed by n.m.r. spectroscopy. Reversible formation of a methoxy intermediate and combination of adsorbed H(D) is conceivable for methanol exchange. The incorporation of hydrogen (H) into the CD, group was negligible, and hence multi- exchanged methanol ([2H,]-[2H,]) was not detected. The reaction between CH,OH and D, exhibited similar results. The observed distributions and locations of the isotope in the products are also consistent with this conclusion. References 1 F. A. Kuijpers, Philips Res. Rep. Suppl., 1973 no. 2; G. G. Libowitz, in Critical Materials Problems in Energy Production, ed. C. Stein (Academic Press, New York, 1976), chap.28; K. H. J. Buschow, D. C. P. Bouten and A. R. Miedema, Rep. Progr. Phys., 1982, 45, 937; D. G. hey and D. 0. Northwood, J. Muter. Sci., 1983, 18, 321. 2 See recent numerous papers presented at the International Symposium of the Properties and Applications of Metal Hydrides, pubd in J. Less-Common Met., 1984, 103/104; 1987, 129/130/131. 3 W. E. Wallace, Chemtech, 1982, 752; F. P. Netzer and E. Berter, in Handbook on the Physics and Chemistry of Rare Earths, ed. K. A. Gschneidner and L. Eyring (North-Holland, Amsterdam, 1983), vol. 5, chap. 3; S. T. Oyama and G. L. Haller, Catalysis (Specialist Periodical Report, The Chemical Society, London, 1982), vol. 5, p. 333. 4 K. Soga, H. Imamura and S . Ikeda, Chem. Lett., 1976, 1387; J. Phys. Chem., 1977,81,1762; J.Catal., 1979, 56, 119. 5 H. Imamura, K. Yamada, K. Nukui and S . Tsuchiya, J. Chem. SOC., Chem. Commun., 1986, 367. 6 H. Imamura, K. Nukui, K. Yamada and S. Tsuchiya, J. Chem. SOC., Faraday Trans. 1, 1987, 83, 7 L. A. Paquette, Y. Miyahara and C. W. Doecke, J. Am. Chem. SOC., 1986, 108, 1716. 8 S. J. Fanelli, A. J. Maeland, A. M. Rosan and R. K. Crissey, J. Chem. SOC., Chem. Commun., 1985, 8. 9 H. Imai, T. Tagawa and M. Kuraishi, Muter. Res. Bull., 1985, 20, 511. 743. 10 H. Imamura, Y. Kato, K. Yamada and S . Tsuchiya, Appl. Catal., 1986, 27, 243. 11 (a) A. Lawson and S . J. Thomson, J. Chem. SOC., 1964, 1861; (b) I. Yasumori, T. Nakamura and E. Miyazaki, Bull. Chem. SOC. Jpn, 1967, 40, 1372; (c) E. Miyazaki and I. Yasumori, Bull. Chem. SOC. Jpn, 1967, 40, 2012; (d) D. W. Mckee, Trans. Faraday SOC., 1968, 64, 2200. 12 G. C. Bond, Catalysis by Metals (Academic Press, London, 1962), p. 407. 13 A. C. Neish, Can. J. Res., 1945, 23, 49; W. G. Palmer and F. H. Constable, Proc. R. SOC. London, 14 A. A. Baladin and P. Teteni, Probl. Kinet. Katal., 1960, 339. 15 A. Goudy, W. E. Wallace, R. S. Craig and T. Takeshita, A h . Chem. Ser., 1978, 167, 312. 16 H. A. Kierstead, J. Less-Common Met., 1980, 70, 199. 17 J. L. Anderson, T. C. Wallace, A. L. Bowman, C. L. Radosevich and M. L. Courtney, Los Alamos Sci. Lab. Rep., 1973, LA-5320-MS; R. A. Guidotti, G. B. Atkinson and M. M. Wong, J. Less-Common Met., 1977, 52, 13. Ser. A , 1925, 107, 255. 18 P. Fuderer-Luetic and I. Brihta, Croat. Chim. Actu, 1959, 31, 75. 19 A. Ozaki, Isotope Studies of Heterogeneous Catalysis (Academic Press, New York, 1977). 20 I. Yasumori and E. Miyazaki, Nippon Kaguku Zasshi, 1971, 92, 659. 21 (a) B. A. Sexton, SurJ Sci., 1981, 102, 271; (b) G. W. Rubloff Jr and J. E. Demuth, J. Vac. Sci. Technol., 1977,14,419; (c) J. E. Demuth and H. Ibach, Chem. Phys. Lett., 1979,60,359; ( d ) I. Kojima, H. Sugihara, E. Miyazaki and I. Yasumori, J. Chem. SOC., Faraday Trans. 1, 1981, 77, 1315. 22 G. Blyholder and L. D. Neff, J. Catal., 1963, 2, 138; G. Blyholder and W. V. Wyatt, J. Phys. Chem., 1966, 70, 1745. Paper 7/341; 23rd February, 1987 t 1 Torr = 101 325/760 Pa.

ISSN:0300-9599    

DOI:10.1039/F19888400765

出版商:RSC    

年代:1988

数据来源: RSC

摘要:

J. Chem SOC., Faraday Trans. 1, 1988, 84(3), 773-784 Kinetic and Equilibrium Studies at the Solid-Liquid Interface The Adsorption of Sodium Hexadecyl Sulphate to Polystyrene Latex David Painter, Denver G. Hall and Evan Wyn-Jones* Department of Chemistry and Applied Chemistry, Univer,rity of Salford, Salford M.5 4WT Binding isotherms and kinetic measurements associated with the adsorption of the surfactant sodium hexadecyl sulphate onto polystyrene latex particles at two latex concentrations are reported. The surfactant monomer concentrations were estimated in situ for all dispersions from e.m.f. measurements of a cell containing a surfactant ion-selective electrode. Both isotherms are of the classical B.E.T. type. At low surfactant concentration the initial portions of the isotherms (region I) can be fitted by the Langmuir equation and do not depend on the concentration of latex particles.At higher concentrations (region 11) the isotherms diverge. Bound sur- factant-counter-ion interactions could be responsible for the difference between the binding isotherms. The kinetic data were measured with the pressure jump relaxation technique and revealed a single relaxation process in region I1 only. The observed relaxation times decrease with increasing surfactant concentration. Application of linear phenomenological theory to the combined equilibrium and kinetic data shows clearly that the adsorption rate coefficient increases with increasing amount of bound surfactant and that the desorption rate is proportional to the amount bound in region 11.The rate coefficients measured for the two dispersions agree reasonably well. These findings are consistent with the formation of hemimicelle type aggregates. Application of the Aniansson-Wall kinetic treatment associated with monomer-aggregate exchange to region I1 gives a linear plot for each solution, but the agreement between the desorption rate coefficients is worse than that obtained using the phenomenological approach. The preparation of spherical polystyrene latices in very monodisperse form is relatively easy and these systems have found several applications as model colloids in many different disciplines. In many cases latices are stabilised by the adsorption of surface active agents and the removal of these can lead to loss of stability, through coagulation or flocculation.Fundamental to the interpretation of stability results is a knowledge of the properties of the adsorbed surfactants, especially their interaction with the surface. To date, all studies on latex-surfactant interactions have involved equilibrium-type measurements and in a few cases binding isotherms representing the amount of surfactant bound to the latex as a function of total or monomer surfactant concentration have been The mechanism of these interactions can only be properly understood if the equilibrium measurements are complemented by kinetic studies. The exchange process between surface active agents and colloidal particles is very fast and can only be studied using chemical relaxation methods. As a result of preliminary studies by Blo0r4 in these laboratories, kinetic studies on the adsorption/desorption of surfactants at solid interfaces can be investigated using the pressure jump relaxation technique.Furthermore, the development of surfactant selective electrodes, which in principle give a direct measure of monomer surfactant activity, allows accurate binding 773774 Adsorption of Sodium Hexadecyl Sulphate to Polystyrene isotherms to be determined fairly quickly. In this paper we report equilibrium data on the system sodium hexadecyl sulphate-polystyrene latices in the presence of mol dm-3 sodium bromide which have been complemented by the first reported measure- ments of relaxation data using the pressure jump apparatus. Experimental Materials The preparation ,and characterisation of the polystyrene latex used was carried out as described in the The particles have a narrow size distribution with a mean particle diameter of 200 nm, determined by electron microscopy.Electrophoretic measurements indicate tha: the particles have a small negative net surface charge of 1.5 pC cm-2 presumably dbe to sulphate groups from the initiator. No surfactant additives were used in the preparation. For a 1 % w/v latex the estimated surface area is 300 m2 dm-3. The sodium hexadecyl sulphate was a Lancaster Synthesis product which was carefully purified by repeated recrystallization from ethanol. Electrodes The surfactant-selective membrane electrodes used in this work are a modified version of those described in previous publications. 8* The polyvinyl chloride membrane contains positively charged end groups and a commercially available polymeric plasticizer. The principle of these electrodes show that the monomer surfactant activities in various formulations can be obtained from e.m.f.measurements on the following cell. I test solution electrode I sodium bromide surfactant containing I reversible to electrode mol dm-3 I sodium bromide The surfactant and bromide ions carry the same charge, hence measurements using the above cell give the ratio (surfactant ion activity)/(bromide activity). In favourable conditions, the activity coefficients of the monomer and surfactant co-ions are approximately equal, which in turn means that the ratio (surfactant monomer concentration)/(bromide concentration) is measured.During this work the sodium bromide concentration was kept constant at mol dm-3. Pressure Jump Measurements The kinetic measurements were undertaken using the Dialog pressure jump apparatus with conductivity detection incorporating a rapid data capture and analysis system.loa9 Relaxation data can be measured over the time range 10-10-4 s. The measurements were carried out at 35 "C. At very low concentration of surfactant, the amplitude of the relaxation signal was rather weak and in these circumstances several transients were collected for each solution and the stored data were averaged. Equilibrium Studies The monomer surfactant concentrations in the presence of the polystyrene latex were evaluated as follows. First the e.m.f. of the surfactant electrode relative to the bromide reference electrode was measured as a function of increasing surfactant concentration until the c.m.c.was reached. At concentrations below the c.m.c. all the surfactant is in its monomeric form and in this range the e.m.f. data yield good Nernstian responses for the electrode as shown in fig. l ( a ) and (b). The experiment was then repeated byD . Painter, D. G. Hall and E. Wyn-Jones 150 100 > 50 E > 'u. E a; 0 - 50 775 I I I I log [ SHS] log [ SHSI Fig. 1. (a) e.m.f. data us. total concentration of SHS (C) in the presence and absence of 0.5 % (w/ v) PSL at 35 "C in the presence of mol dmW3 NaBr. 0, SHS with 0.5 YO (w/v) PSL; pure SHS. (b) E.m.f. data us. total concentration of SHS (C) in the presence and absence of 1 YO (w/ v) PSL at 35 "C in the presence of mol dm-3 NaBr.0, SHS with 1 % (w/v) PSL; pure SHS. measuring the relative e.m.f. of the surfactant electrode in the presence of a constant amount of polystyrene latex, again ensuring that measurements were taken below the c.m.c. of the surfactant in the presence of the latex. These measurements on the poly- styrene latex/surfactant formulations were carried out in such a way that the reversi- bility of the binding process was confirmed. Once the e.m.f. measurements of the surfactant/latex formulation was completed the e.m.f. of the electrode was then rechecked against monomer surfactant to ensure consistency. In this work two concentrations of the latex were used, viz. 0.5 and 1 O/O w/v. From these data it is possible to evaluate the monomer concentration m, at each total concentration of surfactant C for which the measurements have been taken.In this work we express the amount bound as the number of moles of surfactant bound per unit area of latex surface expressed by r in mol m-2. At any known concentration of overall surfactant C and latex concentration this can be evaluated from the latex density, mean diameter of the latex, C-m, derived from the e.m.f. data and the equation C = m, + T A where A is total surface area of the latex. Binding isotherms were then constructed in the form of surfactant adsorbed per unit area of latex r plotted against monomer surfactant concentration rn, and are shown in fig. 2 for the two mixtures which were investigated. Note that the values of r measured are amounts specifically bound; strictly they cannot be identified with Gibbs surface excesses because the latter include a contribution from the negative adsorption of surfactant monomer.In both isotherms two distinct binding regions, denoted I and 11, can be identified. In both these binding regions r increases with m, as the total concentration of the surfactant is increased. We now focus our attention on region I of the binding process, where a comparison of the enlarged ciirves is also shown in fig. 2. It is clear that for both latex concentrations776 Adsorption of Sodium Hexadecyl Sulphate to Polystyrene enlargement of region I Fig. 2. us m, for m,/1O4mol dm-3 SHS and PSL latex at 35 "C in the presence of mol with 0.5 % (w/v) PSL; +, SHS with 1 % (w/v) PSL.dm-3 NaBr. 0, SHS the values of r are identical within experimental error. This shows that the amount of adsorbed surfactant is proportional to the surface area. It is also apparent from fig. 3(a) and (b) that in region I the binding data can be fitted well by the expression: +- (1) which is a rearranged form of the well known Langmuir isotherm where KO is the Henry's law equilibrium constant for binding at low surface coverage and rsat corresponds to so called monolayer coverage. However, it does not follow from this that the system conforms to the assumptions made in deriving the Langmuir isotherm, namely that the adsorbed molecules are randomly distributed among fixed sites and do not interact with one another. For example, the adsorption of sodium dodecyl sulphate at the air-water interface can be well fitted by the Langmuir isotherm.lla Yet in this case the notion that there are fixed sites and that adsorbed molecules do not interact are clearly untenable.Indeed, any system for which there is a net repulsive lateral interaction between adsorbed molecules which gets bigger as the molecules get closer together can be expected to conform qqalitatively to the Langmuir isotherm. In the present case the value of l/rsat is ca. 240 A2. In fig. 4 this value is compared with the area that would be obtained by the surfactant monolayer lying flat on the surface in its fully extended form, this shows that at least a configuration is clearly possible. However, we have no other information to support the hypothesis that this is indeed the mode of adsorption in region I.On the other hand, if one refers to treatmentP associated with the 1 1 - 1 - r - KO rsat m1 rsatD. Painter, D. G. Hall and E. Wyn-Jones 777 I I I 1 I I I I 2 L 6 8 10 0 1 .o 2 .o 3.0 4.0 I 1 ( 1 /M)/ 1 0-5 mol-’ dm3 Fig. 3. (a) l / r us. l/m, for region I of 0.5 YO (w/v) PSL and SHS. (b) l / r us. 1/M, for region I of 1 % PSL and SHS. I I I I I I I I I I I I I Fig. 4. Areas of all trans sodium hexadecyl sulphaote molecule = 125 A2. Area of surface occupied = 240 A2. adsorption of polymers onto latex surfaces then the form of the monolayer binding in these cases is normally explained in terms of loops, trains and ends of the polymer chain, in which the ends and loops protrude out into the aqueous solution with the trains lying on the surface.It is probable that the explanation in the present case is somewhere between these two extremes. We now focus our attention on region I1 of the binding isotherms. Although both isotherms are qualitatively similar to the classical B.E.T. type observed in gas adsorption, quantitative agreement with the B.E.T. equation is not good. The data could be fitted at higher concentrations only by sacrificing the fit at low concentrations and vice versa. The difference between the isotherms in the two latex concentrations in region I1 is now considered. Similar differences have been observed in studies of the adsorption of778 Adsorption of Sodium Hexadecyl Sulphate to Polystyrene surfactants onto neutral polymers. In these latter measurements the differences in the isotherms have been explained in terms of interactions between bound surfactants and counter-ions by carrying out measurements at different salt concentrations. l2 Essentially the differences in the isotherms occur as a result of different amounts of counter-ion being bound by the adsorbed surfactant as the overall concentration of surfactant is changed.13 Two features of this explanation are that the differences between isotherms should be least when amounts bound are small and should diminish with increasing concentration of added salt.The fact that the present isotherms coincide at low concentrations is entirely in accord with this viewpoint. In this region the presence of lo-* mol dm-3 NaBr has a sufficient buffering effect on the counter-ion concentration almost to eliminate any differences in binding behaviour that might otherwise be observed.The behaviour at higher concentrations can be related to an effect similar to the role of counter-ions in changing the c.m.c. of a normal micellar solution as a result of the addition of salt by postulating in the present case that the surfactant c.m.c. plays a similar role to the saturation vapour pressure in the B.E.T. equation. The coincidence of the binding isotherms at low concentration is also entirely consistent with this viewpoint. This implies that the additional adsorption in excess of monolayer coverage takes place as in the B.E.T. isotherm. It is more likely that the major driving force for adsorption in this region would arise from the self-aggregation of the surfactant molecules on the surface which results in the formation of localized surfactant aggregates often called hemi-micelles or sub-micelles.This would especially apply for a long-chain surfactant such as the hexadecyl sulphate anion. The formation of such surfactant aggregates could result from each surfactant on the monolayer behaving as a nucleation site for aggregation. This could result from the interaction between the tails and loops of a surfactant anion bound to the surface interacting with additional bound surfactant. The equilibrium data in this region cannot be described by a simple expression involving r and rn,. The alternative method of obtaining a binding isotherm is to centrifuge the dispersion and measure the surfactant concentration in the supernatant.Experiments of this kind in effect monitor the Donnan equilibrium between a concentrated dispersion and the particle-free solution in osmotic equilibrium with it. The above data, together with a theory described earlier,13 indicate that isotherms obtained in this way which measure what is in effect a Gibbs surface excess may differ considerably from those obtained by the in situ method used in the present study which measures the amount specifically bound. We are currently investigating this possibility which has considerable implications for the interpretation of adsorption data. Kinetic Studies The conductivity changes measured in pressure jump experiments are due to changes in the concentrations of ions in the bulk solution which arise from the perturbation of the equilibrium involving the adsorption-desorption of surfactant monomers at the solid/ liquid interface.In all experiments where a measurable effect is observed the results could be interpreted in terms of a single relaxation process. The plots of reciprocal relaxation times as a function of total surfactant concentration are given in fig. 5 and are significantly different for the two latex concentrations. If we now refer to the binding data in fig. 2 then it becomes apparent that the observed relaxation process is confined to region I1 of the isotherm. In these circumstances the equilibrium being perturbed is essentially between monomer surfactant in bulk solution and aggregated surfactant adsorbed on the latex surface.To interpret equilibrium and kinetic data of this kind we make use of the linear phenomenological treatment developed previously, in which the expression for theD. Painter, D. G. Hall and E. Wyn-Jones 779 6.0 - - 5.0- 2 4.0- - 3.0- v) N 1 n \ d J * - O I 1.0 0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 C/ 1 O4 mol dm-3 Fig. 5. Reciprocal relaxation times as a function of total SHS concentration for SHS and PSL at 35 "C. 0, 0.5% (w/v) PSL with SHS; 0, 1 YO (w/v) PSL with SHS. relaxation time z for the process describing the equilibrium between free and bound surfactant is given by - 1 = -I(:) z c where I is the equilibrium forward or backward rate, A is the reaction affinity, c is the extent of the reaction and the subscript c denotes that the total surfactant concentration is fixed.We have shown in the appendix how the thermodynamic term (aA/ac) can be obtained from the binding isotherm and that l / z is given by The advantages of this approach are that it is more general than traditional methods and that it enables us to use the equilibrium binding data as they stand. Since (arn,/ar), can be found from the equilibrium isotherms, eqn ( 2 ) enables RTI, the equilibrium adsorption or desorption rate, to be found from the combined equilibrium and kinetic results at any surfactant concentration. It is apparent from the adsorption binding that in region 11, [ 1 + 1 /A(3rn1/CT),] decreases with increasing m,. Because 1 /z increases with increasing m, it follows that RTl/m, also increases. However, the adsorption model which results in the B.E.T.equation leads to the prediction that l / z should decrease with increasing m, it follows that RTl/m, also increases. However, the adsorption model which results from that of the B.E.T. model despite the qualitative similarity of the adsorption isotherms. To proceed further we consider the respective rates of adsorption, R,, and desorption, R,. In terms of conventional kinetic considerations one would expect the rate associated with the adsorption of surfactant to the latex-surfactant complex to be equal to the function denoted f(r, A ) which is the product at a rate coefficient k, and a term involving r and A multiplied by the monomer concentration, m,. Thus R, =f(r, A)m,. (3 a)780 Adsorption of Sodium Hexadecyl Sulphate to Polystyrene In a similar way the desorption rate at the same concentration would be expected to be a function of the product of the desorption rate coefficient k, and a term involving r and A denoted g(T, A).Thus Rd = g(r9 A). (3 4 At equilibrium these rates are equal; hence the expression m, = g(L A ) / f ( L A ) (4) describes the equilibrium relationship between m, and r for a given A. The kinetic data show that f and g both increase with increasing r. The equilibrium data show that the ratio g / f increases with increasing r. Instinctively one might expect g to be proportional to A at constant r. Also if the desorption rate in region I of the isotherm is small then g might be expected to approach zero as -+ rsat. The plots of g against (r - I-',,,) A for the two isotherms are given in fig.6. Within experimental error, both are approximately linear and pass through the origin. Translated into kinetic terminology this means that the desorption rate is a first-order process which is proportional to the amount of bound surfactant present and can be expressed by the equation to give the following rate constants : kd = 3 x lo2 s-' for the 0.5% w/v latex kd = 2.3 x lo2 s-l for the 1 YO w/v latex. At first sight it is perhaps surprising that these k, values differ. However, if we take into account the experimental error in the l/z values the two rate constants can be regarded as sufficiently close to each other. If counter-ion effects are responsible for the observed differences in the equilibrium binding isotherms it seems more likely that they will increase the adsorption rate by reducing electrostatic interactions rather than by reducing the desorption rate.Note, however, that counter-ion effects have not been allowed for in the derivation of eqn (2). Consequently the thermodynamic term in this equation is not strictly correct. Since this inevitably affects the estimates of k, the agreement between the two values is perhaps as good as can be expected. Although counter-ion effects can be handled theoretically in much the same way as for solutions of ionic surfactantsf5 there are insufficient data available at present to justify applying this more sophisticated approach. Consider now the behaviour off in the forward rate eqn (3a). Since g, f and g / f increase with increasing r and g is proportional to (r-rat) it follows that if fa(T - rsaJn then 0 < n < 1.This suggests that the adsorption process in region I1 is co- operative insofar as surfactant tends to bind to surfactant that is already bound, but that this tendency is reduced with increasing amounts adsorbed because of increasing electrostatic repulsion and, possibly, other steric constraints. For conventional micellisation n x 1 and m, is approximately constant. Overall, the phenomenological analysis of the equilibrium and relaxation data is consistent with the view that small hemi-micellar aggregates are formed in region I1 of the isotherm. Repulsion between these aggregates is greater than in solutions of ionic surfactants at comparable surfactant concentrations because of their closer proximity on the particle surfaces. This results in a smaller value of aggregation numbers than for micelles. If this view is correct it is possible that the analysis of aggregation kinetics predicted by the equation for the fast relaxation time derived by Aniansson and Wall'' will22.0 D.Painter, D. G. Hall and E. Wyn-Jones 0 - 0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 78 1 (r -ra.) x 104 Fig. 6. Backward rate, R, us. r - rsat. for SHS and PSL (0.5 and 1 Yo w/v). 0, SHS with 0.5 Yo (w/v) PSL; 0, SHS with 1 YO (w/v) PSL. describe our relaxation data. For the case of interest the appropriate equation for monomer-aggregate exchange is k- k- (I- - rsat) A l / z = -+- where n is the average aggregation number, 0 is the aggregate distribution width and k- is the rate constant for monomer desorption.Although eqn ( 6 ) was derived for solutions of aggregates with fairly large aggregation numbers it has been applied successfully to solutions in which aggregation numbers are as low as 4-6.'7~18 The plots of 1 /z as a function of (r - T,,,)A/rn, for the two latex concentrations studied are shown in fig. 7. The values of k-/n are given by k-/n = 7.8 x lo2 s-l for the 0.5 % w/v latex k-/n = 2.7 x lo2 s-l for the 1 % w/v latex. These k-/n values differ by a factor which is substantially greater than for the k, values obtained from fig. 6. One could conclude therefore that eqn (6) is successful only insofar as the plots in fig. 7 are linear within experimental error and the intercepts are positive. However, it should be noted that the kinetic treatment of Aniansson and Wall allows neither for counter-ion effects nor for interactions between aggregates. The explicit theoretical treatment of this latter effect in particular poses problems to which there are no obvious approaches which can be expected to succeed. Furthermore, if we compare the relaxation equation of Aniansson and Wall [eqn (6)] with the one for the782 Adsorption of Sodium Hexadecyl Sulphate to Polystyrene 8.0 0 I I 1 1 I 0 0.5 1.0 1.5 2.0 r-rsnt.ml xi04 Fig. 7. l/t, us. (r - rSat)/ml for SHS and PSL. @, SHS with 0.5 YO (w/v) PSL; 0, SHS with 1 % (w/v) PSL. corresponding phenomenological treatment [eqn (2)] the term n / 0 2 in the Aniansson and Wall equation is taken to be constant over the concentration range investigated, whereas, in the phenomenological approach, in eqn (2) the corresponding thermo- dynamic term is evaluated from the experimental data.It is likely that n/02 varies with surfactant concentration during the growth of region 11, and if this factor could be taken into account then it is likely that the above rate constants for the two latex particles would be as close as those found from the phenomenological approach. Finally, we comment on the failure to observe a relaxation process in region I of binding. One possible reason for this is that any relaxation process in region I is too rapid for detection using the pressure jump method. For this to be the case, the steep initial portion of the binding isotherm in region I must arise from a rapid adsorption rate rather than a slow desorption rate.A more likely explanation is that the background conductivity of the bulk solution swamps any conductivity changes due to changes in m,. This can be expected to be the case at low surfactant concentrations even if the relative changes in m, are quite large. At higher concentrations close to the inflexion point on the isotherms the relative changes in m, must be quite small. For a given change in pressure these relative changes are given by eqn (A12) in the Appendix. It is apparent from this equation that the denominator on the right-hand side is greatest near the inflexion point. It is also possible that the volume changes accompanying adsorption are smaller in region I than in region I1 and the surfactant hydrocarbon chains may be more exposed to water.Either factor or a combination of both could explain the absence of an observable relaxation process in region I. Conclusions Equilibrium data have shown that the adsorption of hexadecyl sulphate anion to polystyrene latex is a rather complex mechanism involving different modes of binding.D. Painter, D. G . Hall and E. Wyn-Jones 783 By combining this with relaxation data using the versatile phenomenological treatment, the complex kinetic behaviour in which forward and backward rates balance to give the observed thermodynamic behaviour, has been disentangled leading to new kinetic data and valuable information concerning the properties of these systems. We thank the S.E.R.C. for a CASE award. Appendix A Linear Phenomenological TreatmentI4 of Surfactant Binding to Latices According to the linear phenomenological treatment of a single relaxation process the reciprocal relaxation time may be written as - 1 = -($) z c where the phenomenological coefficient 1 is the equilibrium forward or backward rate of the reaction divided by RT, A is the reaction affinity, c is the extent of reaction and c denotes that the derivative is to be evaluated in a closed system.For the equilibrium involving an exchange process between free and bound surfactant, e.g. free surfactant =$ bound surfactant where a and d denote free and bound surfactant respectively, and p denotes the chemical potential. If we take as our reacting system that amount containing m, mol of monomer then we may identify am, with --ac and eqn (A 1) may be rewritten as Since the total surfactant system it follows that concentration = (m,+TA) and remains constant in a closed dm,+AdT = 0 (A 4) so that eqn (3) may be rewritten as z We now suppose that where u, and vd, respectively, are the partial molar volumes of free and bound surfac t an t .For variations at equilibrium dpa = dpd and we have 0 = (ua-ud) dp+RT d hm,- !!!c dT (4784 Adsorption of Sodium Hexadecyl Sulphate to Polystyrene it follows from this expression and eqn (A 6b) that at constant T and p RT am, = (%) = &F)e where the derivative on the right-hand side describes the equilibrium dependence of m, on r and can be found from the equilibrium-binding isotherm. On substitution for the derivatives in eqn (A5) we have To deal with amplitudes we note that in the absence of added electrolyte the relative change in conductivity during a pressure jump experiment is of the same order as the relative change in m, brought about by the applied change in pressure.This change in m, is readily obtained from eqn (A 8). Bearing in mind that the perturbation takes place at constant C eqn (A 4) applies. On substituting for dT in eqn (A 8) and collecting terns we find that we now substitute for (8pJW) as given by eqn (A 9) and rearrange to obtain where (vd - va) is the volume change accompanying the transfer of the mole of surfactant from solution to the adsorbed state. For small perturbations of the kind usually encountered in relaxation experiments eqn (A 12) shows how the relative changes in m, depend on the perturbation in pressure.It must be stressed that the above derivation makes no allowance for interactions between bound surfactant and counter-ions. References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 P. Connor and R. H. Ottewill, J. Colloid Interface Sci., 1971, 37, 642. I. Piirma and S. R. Chen, J. Colloid Interface Sci., 1980, 74, 90. B. R. Vijagendran, J. Appl. Polym. Sci., 1979, 23, 733. D. M. Bloor and E. Wyn-Jones, J. Chem. Soc., Faraday Trans. 2, 1982, 78, 657. J. Lyklema, K. Fursurawa and W. Norde, Kolloid 2. 2. Polym., 1972, 250, 908. J. W. Goodwin, J. Hearn, C. C. Ho and R. H. Ottewill, Br. Polym. J., 1973, 5, 347. B. H. Bijsterbosch, Colloid Polym. Sci., 1978, 256, 343. S. G. Cutler, D. G. Hall and P. Meares, J. Electroanal. Chem., 1977, 85, 145. S. G. Cutler, D. G.. Hall and P. Meares, J. Chem. Soc., Faraday Trans. 1, 1978, 74, 1758. (a) W. Knoche and G. Weise, Chem. Instrum., 1973,5(2), 91. (b) M. Kriznan and H. Strehlow, Chem. Instrum., 1973, y2), 99. (a) E. H. Lucassen-Reynders, Anionic Surfactant Physical Chemistry of Surface Action (Marcel Dekker Inc., New York, 1981), ed. E. H. Lucassen-Reynders, ch 1, p. 28; (b) M. A. Colen Stuart, Adv. Colloid Interface Sci., 1986, 24, 143. M. L. Fishman and F. R. Eirich, J. Phys. Chem., 1971, 75, 3135. D. G. Hall, J. Chem. Soc., Faraday Trans. 1, 1985, 81, 885. D. G. Hall, J. Gormally and E. Wyn-Jones, J. Chem. SOC., Faraday Trans. 2, 1983, 79, 645. D. G. Hall, J. Chem. Soc., Faraday Trans. I , 1981, 77, 1973; J. Chem. SOC., Faraday Trans. 2, in press. E. A. G. Aniansson, S. N. Wall, M. Algrem, H. Hoffmann, H. Kielman, I. Ulbrecht, R. Zana, J. Lang and C. Tondre, J. Phys. Chem., 1976, 80, 905. D. Callson, J. Gettins, J. Gormally, R. Greenwood, N. Natarajan and E. Wyn-Jones, J. Chem. SOC., Faraday Trans. 2, 1981, 77, 143. P. Jones, E. Wyn-Jones and G. J. T. Tiddy, J. Chem. Soc., Faraday Trans. 1 , 1987, 83, 2735. Paper 7/346; Received 23rd February, 1987

ISSN:0300-9599    

DOI:10.1039/F19888400773

出版商:RSC    

年代:1988

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I , 1988, 84(3), 785-799 Domain Complexions in Capillary Condensation Part 1 .-The Ascending Boundary Curve Vicente Mayagoitiat Ecole Nationale Supe'rieure d'lnge'nieurs de Ge'nie Chimique, Chemin de la Loge, 31078 Toulouse Cedex, France Fernando Rojas and Isaac Kornhauser Departamento de Quimica, Universidad Autdnoma Metropolitana-Iztapalapa, Apartado Postal 55-534, Mkxico 13 D.F., Me'xico A statement of the general principles of capillary condensation in porous networks and the ascertainment of its particularities for a given structure are difficult, since either independent or dependent vapour-liquid transitions arise at each point of the network and also because porous materials occurring in nature and in industrial processes possess extremely variable morphologies.However, the following stages enable one to achieve these objectives readily : (i) development of general expressions for the probability that the various elements fill with capillary condensate, according to their type (sites or bonds) and size, (ii) classification of all possible morphologies of porous structures into a few unambiguous types and (iii) for each of these types, simplification of the general expressions to obtain particular equations allowing a straightforward derivation of domain complexions and ascending boundary curves. It appears that, even if in one structural type, the less frequently encountered domains behave as though independent, for the other types, corresponding to most materials, an interdependence must be taken into account.As an extreme case of domain interactivity (also pertaining to structures represented fully, once a certain degree of filling is reached, a phenomenon arises in which the whole configuration of capillary condensate becomes unstable, the entire network then being filled. The theory of interactions during capillary condensation in a porous network develops readily and directly from Everett's proposition of an assisted filling of a cavity from a window.' Morioka and Kobayashi2 were the first to study quantitatively these effects by Monte Carlo methods. The basic ideas concerning such a mechanism, as well as the preliminary probabilistic treatment, were outlined in a previous paper.3 The porous network is visualized as being formed of an ensemble of two kinds of alternated elements : sites (cavities or antrae) and bonds (capillaries or windows) possessing their own size-distribution functions, F,(R) and F,(R), which are normalized and expressed on the basis of the number of elements: S(R) = lFs(R)dR; B(R) = F,(R)dR.SI (1) These are, respectively, the probabilities of finding a site or a bond having a size R or smaller. As each bond is smaller than or at most equal to any of the two sites delimiting it (which constitutes the 'construction principle '), the size distributions must satisfy the following requirement :* b S(R) 6 B(R) for every R. (2) 7 Permanent address: Departamento de Quimica, Universidad Aut6noma Metropolitana-Iztapalapa, Apartado Postal 55-534, Mexico 13 D.F., Mexico. 785786 Domain Complexions in Capillary Condensation For the same reason, the events of finding a size R, for a site and a size RB for one of its bonds are generally not independent: where F(R,, RB) is the probability density associated with the fulfilment of these joint events.g5 is expressed as4 S@S) dS eXP(-J -) (4) It is then possible to conceive networks free of topological size-correlations between the elements (g5 = 1, i.e. element sizes distributed fully at random) only when overlap is zero, since in this case the construction principle cannot be violated. The connectivity of the network, C, is the mean number of bonds meeting in a site. The critical mean radius of curvature,? R,, of an element for capillary condensation to occur at a certain relative pressure Pv/Po (Pv being the equilibrium pressure in the vapour phase and Po the saturated vapour pressure) is, in accordance with the Kelvin equations S(RB) B - W,) - S(R,) * #(RS,RB) = where 0 and vL are, respectively, the surface tension and the molar volume of the adsorbate, R, is the gas constant and T i s the absolute temperature.The sites, idealized as hollow spheres, eventually locate critical menisci of the same geometry having a mean radius of curvature of the same size, so that such entities of size R d R, should in principle be filled with capillary condensate. However, the bonds which are idealized as cylindrical in nature owing to their character as passages, possess critical menisci for condensation associated in principle with cylindrical geometry and then having a mean radius of curvature that is twice their own radius; thus bonds having a size R < Re/2 are liquid-filled. For additional details concerning the problems of characterizing an element by means of a unique quantity, i.e.its size, consult ref. (3), p. 2934. In particular the factor, 2, between the critical radius for condensation and that for evaporation, both in bonds, could be allowed to vary in a more general fashion which is, however, out of the scope of this work. The network effects arising during capillary condensation can be visualized as follows. The bonds have the opportunity to fill by two mechanisms: they may fill on their own from a cylindrical meniscus, behaving as cylindrical capillaries open at both ends (bond sizes < R,/2), or in the case that at least one delimiting site becomes filled, the preseme of a hemispherical meniscus at the filled-site extremity of the bond permits an assisted condensation corresponding to the behaviour of cylinders open at only one end (bond sizes < Rc).The sites suffer a delaying effect shown in fig. 1 (a). Even if R d R,, condensation is impossible in such sitest still having several empty bonds, since a lack of continuity of the critical meniscus makes its straightforward advancement in the entity impossible. Thus, for condensation to take place into a site it is necessary to fulfil at the same time the condition that R < R, and of having at least C - 1 bonds already filled, the complete invasion by liquid then being guaranteed [fig. 1 (b)]. In this figure it is intended to show, that, despite a departure from ideal spherical and cylindrical t The radius is taken as positive if its centre of curvature lies on the vapour-phase side.5 Zero contact angle is assumed, i.e.perfect wetting. f. Indeed, the expressions ‘cylinder open at one end’ and ‘cylinder open at both ends’, which are widely used in the literature, allow corresponding terms to be applied to describe a site as a ‘sphere open at one pole’ and a ‘sphere open at several poles’.V. Mayagoitia, F. Rojas and I. Kornhauser 787 ( b ) Fig. 1. (a) Delay of capillary condensation in a site surrounded by several empty bonds, even when R, d R,; (b) course of capillary condensation in the same site keeping the condition R, d R,, when there are at least C- 1 bonds already filled.Shaded areas represent the solid matrix. The vapour phase is located on the concave side, while the liquid phase is on the convex side of the menisci appearing in the site. geometries for sites and bonds, the phenomena in question actually occur independently of the real shapes, and in the same way, size is defined on the basis of a critical meniscus and not on real shape. The interest in treating these interactions comes from both fundamental aspects (the theory of phase transitions, percolation etc.) and practical applications (texture determination, natural and industrial processes of adsorption in mesoporous solids etc.). First, a set of general relations between the degree of filling of sites and bonds of a given size and the critical radius of curvature is presented.In view of their convenient application to particular structures, a classification of porous morphologies (intended only for the process corresponding to the ascending boundary curve) is proposed. Once the possible types of networks have been clearly identified, we proceed in each case to obtain specific equations allowing the construction of domain complexions, which are needed to predict the ascending boundary curves. We acknowledge the importance of multimolecular adsorption, which affects the geometry of critical menisci, modifies the relationship between R, and Pv/P,, and in most cases makes a significant contribution to the uptake of adsorbate. All these effects should be taken into account, but we prefer to present only the primary aspects of menisci interactions during capillary condensation, treated systematically for the different types of networks,788 Domain Complexions in Capillary Condensation General Equations Before developing the probabilistic treatment of capillary condensation, it is necessary to introduce several definitions concerning the extent of filling as well as a brief remark on topological size correlations &(R) and 8,(R) are, respectively, the degrees of filling by capillary condensate of sites and bonds of size R, on a number-of-elements basis.The corresponding overall degrees of filling are: O,, the degree of filling of the whole pore system on a volume basis is the quantity directly attainable from experiment. In order to relate it to the former variables we take into account V(R), the volume of an element of size R, and that in any network for every site there exist C/2 bonds: The probability density of finding a size RB for a bond linked to a site of size Rs is flRBIRS) = F(RS, RB)/FS(RS).(8) The probability of finding, for a bond linked to a site of size R,, a size between RBI and RB, corresponds to the integral of this conditional density between these limits. From eqn (3) and (8) we have eqn (4) giving the expression for 4. Treatment of Sites All sites bigger than R, are empty of condensate, i.e. OS(R > R,) = 0, while the evaluation of B,(R < H,) is complex. The probability for a bond linked to a site of size Rs to be filled with capillary condensate is Thus the probability of having a site with all C bonds filled is F, whilst that corresponding to C- 1 filled bonds and the remaining one empty (the identity of the latter being irrelevant, there arise C possible configurations) is CF-' (1 -I).Con- sequently, the probability for a site of size equal to or smaller than R, to be filled (i.e. to have at least C- 1 filled bonds) is the sum of these quantities :V. Mayagoitia, I;. Rojas and I. Kornhauser According to eqn (6) and (1 l), the overall degree of filling of sites is 8, = 1;' [p + Cp-' (1 - 41 F,(R,) dR,. 789 (12) Treatment of Bonds All bonds of size equal to or smaller than R,/2 fill on their own, whilst those bigger than R, are empty, so that B,(R < R,/2) = 1 and 8,(R > R,) = 0. Bonds of intermediate sizes can be assisted to fill. Consider a site linked to a bond of size R,.Disregarding the state of such a special bond, the probability for the other C - 1 bonds of this site to be filled (the bond we are interested in being distinguishable, there is only one possible configuration) is F-'. The probability of filling for this site (i.e. of having a size R, < R, and these C - 1 bonds filled) is J = ~~~-'g(R,,R,)F,(R,)dRs. (13) The probability that at least one of the two sites lying at the extremities of the bond in question is filled is 1 -(1 -a2. Then, the degree of filling of bonds according to the size family to which they belong is R < R,/2 R > R, (14) 8B(R) I' 1 -(I --q2 for R,/2 < R < R,. 1 0 the overall degree of filling of these elements being, from eqn (6) and (14): or Classification of Porous Structures The complexity of the general equations previously developed, as well as the enormous variety of morphologies displayed by porous materials restricts both qualitative and quantitative descriptions of capillary condensation for a given network.Fortunately, a classification of porous materials (which we visualize exclusively in relation to the process of our present interest) allows us to make considerable simplifications while retaining the strictness and consistency of the general treatment. Before introducing the classification that seems to us to be the most convenient, we discuss the parameters which make this possible and present several examples of the advantages it permits. A characteristic which arises naturally when trying to classify porous structures is the overlap, or fraction of area below the size distributions which is common to both sites and bonds, its possible values lying between zero and one: when overlap is zero S(RB) is zero for the entire span of sizes corresponding to the bond-size distribution, whilst B(R,) = 1 for every site radius.Under these circumstances eqn (4) leads to q5 = 1, which means an absence of size interactions between the elements. Expression (10) becomes 8B(&) FB(RB) dRB = 8, for zero overlap (17)790 Domain Complexions in Capillary Condensation Table 1. Types of porous materials type overlap W , ) I zero + O I1 zero ca. 0.5 I11 zero - + I IV medium V complete and eqn (13) is transformed into As the overlap increases the order in the structure also increases, and when this parameter reaches its limiting value of one, q5 = 6R,, RB/FB(RB), where 6R,, RB is a Kronecker delta, i.e.6 = 1 for RB = R,, otherwise 6 = 0 (each site possesses bonds exclusively of its own size4). As is shown later, the description of condensation for this limiting case turns out to be straightforward. However, zero overlap still includes types of multiple behaviour which are qualitatively different, their corresponding structures being distinguished by means of some other function. FOT this we propose the ‘overall probability of coalescence’, S(Rc) : which allows one to assess, directly from the shapes of the size distributions, the degree of cooperativity during condensation :3 ’-+ 0 weak or negligible interactions, inde- pendent filling of entities x 0.5 moderate interactions + 1 strong interactions, hindered filling of sites and assisted filling of bonds are frequent.S(Rc) is simply the mean of S(R,), the fraction of site sizes favourable for condensation, over the critical mean radii of curvature, R,/2 (i.e. as cylinders) of bonds, then constituting a measure of the extent sites. 1- The greatest simplification arises t Eqn (19) may also be given as of impending coalescence of bond menisci in when S(Rc) = 0, since then entities behave a complementary function being rlV. Mayagoitia, I;. Rojas and I. Kornhauser 79 1 R 2R (e 1 Fig. 2. Several possible relative locations of site and bond-size distributions for porous structures : (a) type I, (b) type 11, (c) type 111, (4 type IV and (e) type V.independently. Setting S(R,) = 1 reduces the difficulty of the treatment, since from the beginning of filling all the sites are smaller than R,. The above survey leads us to suggest the classifications shown in table 1, which contains the precise definition of the five types, while fig. 2 shows some of the possible situations of the size distributions, for: (i) zero overlap, which means that the smallest site is bigger than the biggest bond, embracing the following cases: (a) the smallest site (ss) is more than twice as large as the biggest bond (bb), type I, fig. 2(a), (b) the normal situation of zero overlap, type 11, fig. 2(b), and (c) the biggest site (bs) is less than twice as big as the smallest bond (sb), type 111, fig. 2(c), as well as, (ii) the normal situation of overlap, or medium overlap, type IV, fig.2(4, and (iii) complete overlap, type V, fig. 2(e), all our concepts and treatments being equally valid for any shape of the size distributions. It is difficult to provide adequate examples of real porous solids for each of these types, principally for two reasons : (i) an appropriate determination of the dual-size distribution is still to come, so that the assignment of a given material to one or another type is possible only by inference from indirect facts, and (ii) types I1 and IV are the most currently encountered but at the same time the most difficult to illustrate, owing792 Domain Complexions in Capillary Condensation precisely to their intermediate character; on the other hand, types I, I11 and V correspond to extreme situations, their peculiarity making it possible to conceive well defined prototypes for them, even if they are rarities.Particular Equations and Typical Types of Behaviour General Treatment for Zero Overlap The introduction of eqn (17) into eqn (1 1) gives (0; + COg-l(l- e B ) for R G R, 10 R > R, U R ) while eqn (12) and (24) lead to e, = s(R,) [e; + ceg-l(i - OB)] the corresponding expressions for bonds being obtained from eqn (14) and (18): R B R,/2 (26) I' &dR) 1 -[1- S(R,)Og-1]2 for R,/2 < R G R,. 1 0 R > R, and from eqn (16) and (26) 8, = B(R,/2) + [B(R,) - B(R,/2)] { 1 - [ 1 - S(R,) eg-']'}). e B = B(R,/2) + [ 1 - B(R,/2)] { 1 - [ 1 - S(R,) eg-']'). (26') (27) As R,/2 reaches the size of the biggest bond, R, is still smaller than any site radius, so that S(R,) = 0 in eqn (26) and e B = 1 in eqn (24).All the entities fill independently. The equations for this case are as follows: Since for zero overlap B(R,) = 1 if S(R,) > 0, eqn (26') simplifies to give Type I R < R, 10 R > R, B,(R) " for and 8, = B(R,/2). (3 1) The corresponding domain complexions at several stages of filling are represented in fig. 3. This mechanism, the simplest possible, has been the only one visualized by most authors. Nevertheless, type I, to which this situation pertains, is scarcely found. These structures could result as a consequence of topochemical reactions in well consolidated solids: first, the gaseous products are trapped and constrained to form large holes (the sites).Only after subsequent failure of the tightness of the solid do these gases create narrow passages (the bonds) to escape. The product of the thermal dehydration of gibbsite at 205 O C 7 ? * is a rare example of type I. Its nitrogen hysteresis loop is extremely wide, the ascending curve becoming steep only at a relative pressureV. Mayagoitia, F. Rojas and I. Kornhauser 793 0 (el Fig. 3. Evolution of the site- and bond-domain complexion diagrams for a type I structure, according to increasing R,: (a) 18, (6) 22, (c) 27, (4 33 and (e) 35 nm. Size distributions for sites and bonds are taken as Gaussians with means Rs = 34 nm and RB = 10 nm and standard deviations gS and C T ~ equal to 0.6 nm. The connectivity, C, is equal to 6. The shaded areas represent elements filled with condensate: /// filled sites, \\\ filled bonds.> 0.95, while the descending curve is horizontal down to 0.8. The volume of the bonds seems to be negligible. Other solids displaying wide hysteresis loops of the B-type in de Boer’s classificationg could be considered as belonging to our type I in the sense that the filling of domains is independent, but these slit-like formations (as the nickel silicates extensively studied by Broekhoff and van BeeklO) fill exclusively by growth of the adsorbed layer, capillary condensation then being absent. Type II Further simplication of expressions (24)-(27) is not possible for this ordinary situation of zero overlap. The course of condensation is outlined in fig. 4. Initially, bonds start filling independently while sites remain empty.At intermediate stages phase transitions in both entities occur simultaneously and cooperatively. The final pattern depends on the relative situation of the size distributions : (i) if these are not too close [& = 1 before S(R,) = 11, the last sites fill exclusively according to their sizes, or (ii) if these lie not too194 Domain Complexions in Capillary Condensation t -R 0 (e ) Fig. 4. Evolution of the site- and bond-domain complexion diagrams for type I1 structures, according to increasing R,: (a) 29, (6) 32, (c) 33 and (d) 34 nm. (a)-@ Rs = 2RB = 31.5 nm, a, = oB = 2 nm, C = 6. For (e) R, = 35.5 nm, R, = 30 nm, RB = 17.5 nm, a, = aB = 1.75 nm and C = 6. far apart [S(R,) = 1 before 8, = 11, filling of the remaining sites depends on filling of bonds but not on their own sizes.Type I1 structures are naturally less unusual than those of type I. Their characteristic hysteresis loop, still broad, corresponds to type E in the de Boer's classification, e.g. the loops of nitrogen on two virgin synthetic silica-alumina catalysts; Socony tcc beads and Aerocat microspheres, studied by Ries;" the loops of xenon and krypton on porous glass, analysed in detail by Brown,12 Blakeney-Edwards13 and Everett,14 and the loop of nitrogen on partially sintered thoria presented by Wall and Br~wn.'~ As the sites of these solids are of intermediate size, the adsorbed layer contributes markedly to the sloping character of the ascending boundary curve. The descending boundary curve exhibits a plateau before falling abruptly at a low relative pressure, indicating unequivocally (in spite of network effects15 and the mechanical failure of the liquidlo$ 17) bonds of small dimensions.V. Mayagoitia, F.Rojas and I. Kornhauser 795 0 (el Fig. 5. Evolution of the site- and bond-domain complexion diagrams for according to increasing R,: (a) 39, (b) 39.5, (c) 40, (d) 40.5 and (e) 41 nm. 20 nm, us = uB = 1.25 nm, C = 6. Type IZZ From the onset of condensation, all the sites are smaller than R,. become e, = e,(R) = o; + ce;-i(i - eB). a type I11 structure, Rs = 27.5 nm, RB = Eqn (24) and (25) (32) Condensation at each site takes place whatever the size of the element and is subjected only to the state of its bonds, so that sites fill homogeneously. For bonds, conditions (26) and (27) undergo a slight simplification, as for all purposes S(R,) = 1 : R d R,/2 (33) (34) 8,(R) P 1 - (1 - Og-1)2 for R,/2 < R \< R, 'lo R > R, 8, = B(R,/2) + [ 1 - B(R,/2)] [l - (1 - e;-')']].This process is illustrated in fig. 5 . Structures of type I11 are uncommon, as it is required that the distributions come very close and be narrow enough as to have R,, \< q,. (35)796 Domain Complexions in Capillary Condensation -R Fig. 6. Evolution of the site- and bond-domain complexion diagrams for a type IV structure, according to increasing &: (a) 48, (6) 49, (c) 50, (4 51 and (e) 52 nm. RS = 27 nm, RB = 25 nm, a, = oB = 1 nm, C = 6. An example of a family of structures with which this type is associated is the nearly homogeneous packing of monodisperse solid globules. RB could be taken as the radius of the inscribed circle in foramina, while R, corresponds to the radius of the inscribed sphere in cavities.Even if these materials are carefully synthesized in order to achieve a larger degree of regularity, some variations are expected around the mean values, EB and R,. For example, in simple cubic packing - - d2- 1 RBI& = - and if Gaussian distributions with dispersions S, and S, are assumed in such a way that SB/RB = S , / R , = d, the limits of the dual distribution are given in practice by (37) From eqn (35H37) we obtain d < 0.031 ; thus if the radius of globules is 100 nm, the permitted variations for a nearly regular packing to belong to type 111 are f2.5 nm for RB and f4.5 nm for R,.t - R,, = RB-~SB = RB(l -2d);Rb, = $+26, = Rs(l +2d).t The critical meniscus for independent condensation in bonds could present an anticlastic character, leading to an ‘effective’ bond-size distribution shifted to the right. However, as the definition of the sizes of bonds is based on hemispherical meniscus neither the formal bond distribution (and its topological consequences) nor the mechanism of condensation suffer any change in the case of type I11 structures.V. Mayagoitia, I;. Rojas and I. Kornhauser 797 (el Fig. 7. Evolution of the site- and bond-domain complexion diagrams for a type V structure, according to increasing R,: (a) 38, (b) 50, (c) 55, (d) 60 and (e) 72 nm. R, = R, = 27.5 nm, 0, = gB = 2.5 nm, C = 6. Hysteresis loops of these materials are narrow, and both the ascending and descending boundary curves are very steep.Descending boundary curves present an initial horizontal part owing to pore blocking during evaporation. l8 Type IV In this case of medium overlap the general equations (11), (12), (14) and (16) cannot be simplified without committing serious errors, as the effect of q5 is intense: a site of size R, is delimited by bonds obeying the modified distribution function [FB(RB) q5(Rs, &)I, and conversely a bond of size R, links sites whose size is distributed according to [Fs(R,)#(R,,RB)]. The effect of q5 is then to promote the reunion of elements of similar size as well as to disfavour the meeting of entities of very different sizes. Regions formed of large sites and bonds combined and regions of small elements emerge.Cooperative interactions between such regions are difficult, so that the ascending curve is sloping rather than steep. Thus, the sloping character of ascending boundary curves is due not only to a broad distribution of element sizes or to the contribution of reversible adsorption, as de Boer’ pointed out, but also to topological aspects. Characteristic domain complexions are shown in fig. 6. Coarse materials belong to this type. We may mention disordered globular solids :798 Domain Complexions in Capillary Condensation Table 2. Simple mechanisms type of structure 8s type of B(RC/2) behaviour I W C ) 8, independent strongly V 8, 0, pseudo- 111 e, + ce;-l(i - 8,) 1 -eB 1 - interdependent (1 - eg-')2 independent commercial gels such as Aerogel S silica," Harshaw activated aluminall and a Rhone-Poulenc SCS- 100 activated alumina.l9 Under some conditions sintering seems to promote overlap: several products corresponding originally to type I1 become of type IV following a strong treatment, such as the silica gels dosed with ions and heated at 150 "C under atmospheric pressure to dry, presented by Mougey et a1.,20 Aerocat silica-alumina microspheres, steam-treated," and Davison Diakel cracking catalyst. l1 TYPE v As overlap becomes complete, the segregration introduced by q5 reaches a maximum, and the structure becomes an ensemble of ' homotatic ' regions (each possessing equal- sized interdependent elements) separated perhaps by abrupt boundaries and then behaving independently. Within each of these homotatic domains condensation in bonds automatically involves the filling of sites.The relevant equations are : OS = 8, = S(R,/2) = B(R,/2). (39) This simple process is depicted in fig. 7. We are unable to give an example of a real material corresponding to this extreme case of overlap. However, suppose that during the formation of a globular structure (e.g. polymerization in emulsion and subsequent drying) a large size segregation of globules minimizes the energy of the system. If enough regularity of packing is achieved, this will result a structure which behaves according to eqn (38) and (39). Discussion Sites and bonds inevitably alternate. To assign a size distribution to each type of such entities is in the nature of things. An analysis of the relative positions these distributions may adopt suggests that there exist well defined types of structures on which the resulting course of capillary condensation is qualitatively different.Thus it is possible to propose a classification of porous materials (at least with regard to capillary condensation in mesopores) based in structure and mechanism. The problem of determining the texture of materials from experimental ascending curves is not yet solved, and must be carefully rethought from the ideas arising from this work. To perform the general analysis of condensation for a given solid, is so difficult that it would be preferable, from a practical point of view, to decide to which type, among those leading to simple mechanisms, it resembles better, and then to follow the analysis according to the particular equations corresponding to such a type.For convenience we reproduce in table 2 some of the results obtained for these simple types.V. Mayagoitia, F. Rojas and I. Kornhauser 799 Conclusions Each type of porous structure undergoes capillary condensation according to a different mechanism, which produces particular domain complexions. Structures rarely found in nature and industry correspond to easily described relationships. Conversely, for types fully represented by real materials, the treatment is very intricate. This work was supported by the Ministry of Public Education of Mexico (Secretaria de Educacion Publica) under contract no. C-87-01-0116 (DGICSA reg. no. 862550). References I D. H. Everett, in The Solid-Gas Interface, ed. E. A. Flood (Marcel Dekker, New York, 1967, vol. 2, 2 Y. Morioka and J. Kobayashi, J. Chem. SOC. Jpn, 1979, 2, 157. 3 V. Mayagoitia, F. Rojas and I. Kornhauser, J. Chem. Soc., Faraday Trans. I , 1985, 81, 2931. 4 V. Mayagoitia and I. Kornhauser, Principles and Applications of Pore Structural Characterization, ed. 5 D. H. Everett, in The Solid-Gas Interface, ed. E. A. Flood (Marcel Dekker, New York, 1967), vol. 2, 6 W. Thompson, Philos. Mag., 1871, 42, 448. 7 J. H. de Boer, A. van den Heuvel and B. G. Linsen, J. Catal., 1964, 3, 268. 8 B. G. Linsen and A. van den Heuvel, in The Solih-Gas Interface, ed. E. A. Flood (Marcel Dekker, New 9 J. H. de Boer, Colston Papers, vol. X : Structure and Properties of Porous Materials, ed. D. H. Everett p. 1083. J. M. Haynes and P. Rossi-Doria (J. W. Arrowsmith, Bristol, 1984), p. 15. p. 1077. York, 1967), vol. 2, p. 1041. and F. S. Stone (Butterworths, London, 1958), p. 68. 10 J. C. P. Broekhoff and W. P. van Beek, J. Chem. Soc., Faraday Trans. I , 1979, 75, 42. 11 H. E. Ries, Adv. Catal., 1952, 4, 87. 12 A. J. Brown, Ph.D. Thesis (University of Bristol, 1963). 13 N. J. Blakeney-Edwards, Ph.D. Thesis (University of Bristol, 1963). 14 D. H. Everett, in The Solid-Gas Interface, ed. E. A. Flood (Marcel Dekker, New York, 1967), vol. 2, 15 G. C. Wall and J. C. Brown, J. Colloid Interface Sci., 1981, 82, 141. 16 M. M. Dubinin, Pure Appl. Chem., 1965, 10, 309. 17 C. G. V. Burgess and D. H. Everett, J. Colloid Interface Sci., 1970, 33, 61 1 . 18 F. Rojas, Ph.D. Thesis (University of Bristol, 1982). 19 I. Kornhauser, M S c . Thesis (Universidad Autonoma Metropolitana-Iztapalapa, Mexico, 1983). 20 C. Mougey, J. FranCois-Rossetti and B. Imelik, Colston Papers, vol. X : Structure and Properties of p. 1108. Porous Materials, ed. D. H. Everett and F. S. Stone (Butterworths, London, 1958), p. 266. Paper 7 / 4 6 ; Received 17th March, 1987

ISSN:0300-9599    

DOI:10.1039/F19888400785

出版商:RSC    

年代:1988

数据来源: RSC

摘要:

J. Chern. SOC., Faraday Trans. 1, 1988, 84(3), 801-813 Domain Complexions in Capillary Condensation Part 2.-Descending Boundary Curve and Scanning Vicente Mayagoitiat and Bernard Gilot Ecole Nationale Supkrieure d’lngknieurs de Gknie Chimique, Chemin de la Loge, 31078 Toulouse Cedex, France Fernando Rojas and Isaac Kornhauser Departamento de Quimica, Universidad Autdnoma Metropolitana-Iztapalapa, Apartado Postal 55-534, Mkxico 13 D.F., Mkxico The study of capillary condensation and evaporation in all possible types of porous structures leads to the following conclusions : blocking phenomena of the ‘ network’ kind occurring in evaporation present several forms according to the degree of overlap between the site and bond size distributions. These phenomena are as follows : (i) intense, with a definite percolation threshold, for random structures pertaining to zero overlap ; (ii) moderate, in the case of entities topologically correlated in size as the degree of overlap is medium; (iii) non-existent, for homotactic domains related to full overlap.Cooperative transitions during condensation are even more sensitive to porous morphology since for each type of structure there is a different set of equations to describe them. Each of these types also has its characteristic shape of boundary loop, and even particular forms of scanning curves. Starting from experimental data for a given solid, this property could permit one in principle to decide, according to the shape of its boundary and scanning curves, to which type of structure this solid belongs, and then to select the appropriate set of equations allowing the correct exploitation of these data.Everett’ proposed a classification of adsorption hysteresis based on the pressure range over which the main boundary loop extends, giving many examples of each type of loop and presenting an excellent survey of all kinds of interactions leading to such behaviour. As we wish to consider vapour-liquid transitions exclusively, this classification is too general for us. de Boer,2 assigning one characteristic (steep or sloping) to each boundary curve and taking into account the region of relative pressure at which steps can occur, distinguished five types of hysteresis loops. This classification proves to be incomplete since, as this author3 himself pointed out, there are numerous examples of loops having one or both of their curves combining both a steep and sloping appearance at different regions of relative pressure, and also because these five types, all with at least one steep curve, do not include the frequently encountered case of both boundary curves being sloping. de Boer based his classification on the shape of the hysteresis loop, and afterwards he tried to explain these forms in terms of the structure of the porous material (the shape of capillaries and their relative positions and sizes) which determines the precise routes followed by the processes of condensation and evaporation.However, it seems to us that the appropriate strategy to establish a convenient classification of hysteresis should follow the opposite direction : such a classification should be based on the recognition of t Permanent address : Departamento de Quimica, Universidad Autonoma Metropolitana-Iztapalapa, Apartado Postal 55-534, Mtxico 13 D.F., Mtxico.27 80 1 FAR I802 Domain Complexions in Capillary Condensation several unambiguous types of porous morphologies, each leading to an individual mechanism of hysteresis associated with a characteristic form of the boundary loop. In Part 1,4 we proposed five well defined types of porous structures (as a function of the relative positions of their site- and bond-size distributions) which produce different particular types of behaviour along the ascending boundary curve of capillary condensation. In this paper we assess whether such a classification remains useful for other processes associated with hysteresis, such as the descending boundary curve of capillary evaporation and the primary scanning curves.In particular, we will examine whether it is possible: (i) to simplify the general treatments of these processes, at least for some types, while maintaining strictness and consistency, and (ii) to recognize a characteristic shape of the hysteresis loop for each of these types. As we are mainly interested in the basic mechanism of capillary condensation and evaporation, we do not take into account for the moment the influence of multimolecular adsorption, but the further incorporation of this effect does not seem insurmountable. Most of the considerations as well as the terminology of Part l 4 will be extensively used in this paper.Descending Boundary Curve General Equations The mechanism of capillary evaporation from a previously saturated condition has been given by Everett5 and Barker.' As the possibility of bubble nucleation in mesopores is vanishingly small,' the liquid held in a site or bond, can evaporate only if a liquid-vapour meniscus, coming from neighbouring entities, is able to invade such an element . Treatment of Sites The probability that a site of size R, be full of condensate is O,(R,). If a precise size RB is imposed on one of its bonds while Rs can adopt any value, the probability for such site to be filled is (taking into account the necessary topological correlations of size between both elements) rco since the lower limit for the size of the site is RB.Let us now consider a site of size R, whose invasion is to be analysed; for the moment only one meniscus will be considered. For this meniscus to reach effectively the site of interest, the first-order neighbouring site from which it arrives must be empty and the bond linking these sites must be able to be permeated, i.e. it must be equal or larger in size than the critical radius, R,. The probability for this invasion, related to only one meniscus, is expressed (again taking into account the topological correlations which render self-consistent any porous network) as: Rs K = JRc (1 - JIB 8 ~ ~ s ) $ ( ~ s F X R ~ ) d ~ s ) $(RB, ~ s ) FB(RB) ~ R B . (2) Thus the probability for a site of size R, 2 R, still to be filled is equal to that of preventing the entry of a liquid-vapour meniscus from any of its C bonds, while if R, < R, the element is clearly inaccessible to vapour: The overall degree of filling of sites along the descending boundary curve is consequently 8, = S(R,) + /Ic (1 - K)C F,(R,) dR,.(4)V. Mayagoitia et al. 803 Treatment of Bonds For any one of the delimiting sites of a given bond of size R, to be responsible for invasion, such a site must be previously invaded from any of its remaining C - 1 bonds. The probability of this previous invasion is so that the probability of invasion of such a bond when R, 2 R, from one of its delimiting sites is A bond of size R, 2 R, remains filled if none of its two delimiting sites allows invasion, while if R, < R, evaporation of its condensate is impossible even if the bond in question is surrounded by menisci: R b R, R < R,‘ for B,(R) { (’ - L)2 1 Finally, the overall probability for a bond to be full of condensate is ( 7 ) Particular Equations and Typical Types of Behaviour What follows is an exploration to see if the general treatment can be simplified for each of the types of porous structures proposed in Part 1.* Zero Overlap Since for zero overlap $ = 1, eqn (2) simplifies appreciably : K = ff ( 1 - 0,) FB(RB) dRB = (1 - 6,) [ 1 - B(Rc)] while eqn (6) leads to L = 1; (1 - 6g-l) F,(R,) dR, = 1 - 6g-l.(9) Introducing eqn (9) into eqn (3) as well as into eqn (4), and considering that for zero overlap R, > R, from the beginning of evaporation [which requires that S(R,) = 01, we obtain which can be also presented in the form B(R,) = (6t’c - OS)/( 1 - 6,).(12) This expression is due to Iczkowski.* As in this case of zero overlap the sizes of the elements are distributed fully at random, the degree of filling of sites is the same for every site size. For bonds, eqn ( 7 ) and (10) lead to (13) 27-2804 Domain Complexions in Capillary Condensation Thus, all bonds larger than R, have the same probability to be invaded by vapour, regardless of their size. Substituting eqn (10) into eqn (8) gives an equation equivalent to eqn (12), but for (14) bonds For the descending boundary curve, no qualitative distinctions appear in the treatment for the structural types I, I1 and I11 we proposed in Part 1.4 These have their common treatment, which we have just outlined.B(R,) = (e, - e y - i ) ) / ( i - ey-1)). Type IV When overlap is medium the general equations (1)-(8) cannot be simplified without the possibility of committing serious errors, so that these expressions remain as the particular equations for type IV. The topological size-correlations attenuate blocking effects with respect to totally random structures, as large elements are linked by other large entities, so that invasion by vapour to these elements is easier than if they were distributed totally at random. As overlap increases, the width of the hysteresis loop decreases. When complete overlap is reached, the network is formed by an ensemble of ‘ homotactic ’4 regions, each being constituted of equal-sized and totally interdependent elements, while each region is completely independent of the others.In this case the simplification of expressions is drastic : R >, R, Os(R) = 8,(R) for (15) 1; R<R, and es = e, = S(RJ = B(R,). (16) Thus, blocking phenomena of the ‘network” type are absent in type V structures, the hindrance to evaporation being due only to the ‘pore” hysteresis, and more precisely to a ‘ delayed meniscus ’lo mechanism. Evaporation The importance of blocking phenomena decreases as overlap of the site- and bond-size distributions increases. For zero overlap these phenomena are intense, the width of the hysteresis loop decreasing from type I to type 111. No qualitative differences appear in the treatment of evaporation for these types. For medium overlap (type IV) the interactions during evaporation are moderate, while the treatment remains very complicated, as size correlations are important.For complete overlap (type V), as blocking phenomena of the network type are totally absent, the descending boundary curve reveals in a direct way the pore size distribution of the material. Consequently, it could be equaIly misleading to ignore blocking phenomena for broad loops than to incorporate uncritically their treatment for narrow loops. Primary Ascending Scanning Curves Such curves begin at some reversal point on the boundary descending curve whose conditions will be denoted by an asterisk, thus R,* is the mean radius of curvature of a point lying on the descending boundary curve, at the onset of the process under consideration.V . Mayagoitia et al. 805 General Equations The treatment follows closely that presented in Part 1,* eqn (10)-(16), where the definitions of I and of J can be found.Treatment of Sites The degree of filling of sites according to their size-group is 1 R < R,* { O,*W R, < R. All sites smaller than R,* are always full of condensate, as the previous evaporation process has not been sufficiently thorough to empty them. Some sites bigger than R,* were already filled at the reversal point (O,*), but in particular those previously unfilled (1 - 9:) and having a size between R,* and the actual value of R, can become filled during the ascending process (the probability of such event being expressed in terms of 1) by a cooperative mechanism. The overall degree of filling for sites is &(R) O,*(R)+[l -O,*(R)][F+CF-' (1 -81 for R,* < R < R, (17) 8, = S(R,*) + 11; (O,*(R) + [ 1 - O,*(R)] [F + C P 1 (1 - Z ) ] } x F,(R)dR+ JIcO,*(R)F,(R)dR. (18) Treatment of Bonds All bonds which are filled at the reversal point remain in the same state.If R,/2 > R,* all the bonds of sizes between these two values become filled on their own. Empty bonds of size between Rc/2 and R,* can fill only by a cooperative mechanism. 9 should be either R,* or Rc/2 whichever is greater: R < & R > R, for & < R < R, (19) The overall degree of filling for bonds becomes J W J R C Particular Equations and Typical Types of Behaviour Each of the five types of porous structures exhibits different qualitative behaviour resulting in particular expressions, as one must include the contribution of an ascending process.General Treatment for Zero Overlap This situation, embracing types 1-111, is associated with the condition q5 = 1, which renders O,* = O,*(R,) for all R, and [see eqn (17) and (18) in Part lI4806 Domain Complexions in Capillary Condensat ion As the size distributions remain apart, then always R, > R,* so that only two kinds of site sizes are considered. The overall degree of filling for sites is an expression that may be compared with eqn (25) in Part l . 4 for (22) es = e: + (1 - e:) s(R,) [e", + c eg-l(i - e B ) ] For bonds R < B R > R , + [ 1 - 8;S(R)] { 1 - [ 1 - S(Rc)B,C-1]2) 92 < R < R, (23) and an equation which, by virtue of eqn (13) and since B(Rc) = 1 for S(R,) > 0, transforms obtained considering also eqn (14) and Eqn (25') can be compared with eqn (27) in Part 1.* Type I When sites are becoming filled, R,/2 is bigger than the biggest bond, so that all bonds are filled.Sites fill independently : R < R, &(R){ for (27) 0: R ' Rc. The overall degree of filling is For bonds or This equation could be obtained directly from eqn (25') when considering that cooperative behaviour is totally absent in a type I structure. Type II As this type corresponds to the general situation of zero overlap, there is no further simplification of expressions (21)-(25').V. Mayagoitia et al. 807 Type III The course of condensation is clearly divided in three regions. First, elements are unable to refill until R, becomes such as to make S(R,) > 0.This regime originates an initial plateau in the primary ascending scanning curve: which can be compared with eqn (26). Secondly, sites and bonds fill exclusively by a cooperative mechanism. Eqn (21)-(25’) apply without any further simplification, except that RE > RJ2, so that 9 = RE. In the third region, when R,/2 > R,*, independent filling of bonds can resume again [note that all the sites are in a supersaturated condition, i.e. S(R,) = 11 : ( 3 5 ) 8, = o,(R) = e,* + (1 - e,*) [e; + ce;-l(i - 6B)] R d R,/2 O;S(R) + [ 1 - 8E(R)] [ 1 - (1 - 0g-1)2] for R > R,/2 8B(R) { (36) and Note the similarity of eqn ( 3 5 ) and (37) to eqn (32) and (34) of Part 1.4 Type IV Owing to the complexity of type IV structures it is necessary to apply the general equations (1 7)-(20), which remain in this case without any possible simplification.Type v To refill elements already emptied at R,* it is necessary that R,/2 > RE and advances progressively : and 8, = 8, = S(9) = B(9). (39) These equations are exactly the same as eqn (38) and (39) of Part 1,4 since pseudo- independent behaviour always occurs in ‘ homotactic ’ domains, no matter the process in question. Primary Ascending Scanning Processes Each type of structure reveals its particularities in scanning. For type I it is necessary that R,/2 reaches RE. Then bonds refill on their own before any site can be refilled once more ; afterwards empty sites start refilling independently. For type I1 it is not absolutely necessary that R,/2 > RE for the curve to bend upwards: after the reversal point, as when R, is such that some empty sites are smaller than it, cooperative filling may take place again.For type 111, however, the main mechanism by which entities are refilled808 Domain Complexions in Capillary Condensation with condensate is through cooperative interactions. Type IV shows a very general process of refilling, while type V is characterized by a plateau extending over the entire region between the boundary curves of hysteresis. Primary Descending Scanning Curves Such curves start from reversal points lying on the ascending boundary curve, the properties of which points are again denoted by an asterisk. The treatment will be presented briefly since it can be derived readily from previous arguments.General Equations Treatment of Sites 0 R > R,* es(R) { O,*(R) (1 -P)' for R,* 2 R 2 R, KYR) R < R, es = rc@(R)Fs(R)dR+ r'O$(R)(l -P)'F,(R)dR RC 0 R > R,* P2 R,* 2 R 2 R, 8;s(R, R , > R > @ 1 R < @ for Treatment of Bonds e m J* corresponds to J in eqn (13) of Part 1,4 except that the upper limit, R,, of the integral is replaced by R,*. K* is equal to Kin eqn (2), except that e:(R) appears as an additional factor inside the integral, since sites of the e,* kind own bonds of the 0: kind exclusively. In this case B' is R,*/2 or R,, whichever is smaller. Finally (43) 8, = B(9')+ rct9:(R)FB(R)dR+ 9' Particular Equations and Typical Types of Behaviour General Treatment for Zero Overlap Using eqn (18) of Part l 4 and eqn (9), J* and K* may be greatly simplified.For sites : R > R,* (44) for 1 -B(R,)-(1 -8;) '-'(@) )r R<R,* 1 - B(R,*/2) and taking into account that J o it is found that 1 - B(RE/2) 1 -B(R,)-(i -e;) Instead of eqn (12) we obtainV. Mayagoitia et al. 809 Fig. 1. Domain complexions for descending boundary curves : (a) type I, R, = 1 1 nm, (b) type 11, R, = 17.5 nm, (c) type 111, R, = 21 nm, (d) type IV, R , = 26 nm, (e) type V, R, = 57.5 nm. The shaded areas represent elements filled with condensate: (///) filled sites; (\\\) filled bonds. 0 R > R,* S(R,*)2 8i(C- 1) e m 1 R < 9’. R,* 2 R 3 R, R, > R > 9’ for For bonds O,(R) Eqn (26) leads to the expression 8, = B(9’) + ( 8’-B(”’2)) [B(R,) - B(9’)] + [B(R,*) - B(R,)] [S(R,*) 8:-ll2. (49) 1 - B(R32) Again a unique equation applies to types I, I1 and 111, with reference to this descending process. Domain complexions, however, will be influenced by the particular state reached at the reversal point for each of the types.810 Domain Complexions in Capillary Condensation (el Fig.2. Domain complexions for primary ascending scanning curves : (a) type I, R, = 33 nm, R,* = 1 1 nm, (b) type 11, R, = 30.75 nm, R,* = 17 nm, (c) type 111, R, = 28.25 nm, R,* = 20.5 nm, ( d ) type IV, A, = 27 nm, R,* = 26 nm, (e) type V, R, = 115 nm, R,* = 55 nm. The shaded areas represent elements filled with condensate: (///) filled sites; (\\\) filled bonds. The broken line represents the boundary between full and empty elements at the point of reversal. Type IV General expressions (40)-(43) cannot be simplified further.To remove condensate in previously filled elements, it is required that R, decreases to R32 so thatV. Mayagoitia et al. 81 1 (el Fig. 3. Domain complexions for primary descending scanning curves: (a) type I, R , = 11.4 nm, R,* = 33 nm, (b) type 11, R, = 17.5 nm, R,* = 33 nm, ( c ) type 111, R, = 20.75 nm, R,* = 40.5 nm, ( d ) type IV, R, = 26 nm, R,* = 51 nm, (e) type V, R, = 55 nm, R,* = 115 nm. The shaded areas represent elements filled with condensate: (///) filled sites; (\\\) filled bonds. The broken line represents the boundary between full and empty elements at the point of reversal. Primary Descending Scanning Processes Vapour-filled elements at the reversal point are either (i) fully spread across the network, which permits a very efficient invasion of the whole structure by vapour (as for zero overlap) or (ii) progressively segregated from filled entities as overlap increases, so that the descending scanning curve does not fall abruptly before its meeting with the boundary loop.In contrast to ascending curves, there it is impossible to have characteristic expressions for each of the types I, I1 and 111 during a descending process, and this may suggest that more information about the nature of the porous network would be available from ascending rather than descending curves.812 Domain Complexions in Capillary Condensation Domain Complexions Everett'' has proposed domain complexions to represent the state (empty or full of capillary condensate) of independent domains ordered according to their characteristic condensation and evaporation properties, x,, and x,, .Our domain complexions are for interdependent elements. As we consider two kinds of them: sites and bonds, a complexion is twofold. The areas in such diagrams correspond to fractions of elements in number, each distribution being normalized. They can be constructed from the calculations that we have extensively developed in previous sections, and represent &(R) and 8B(R), each in its distribution, for every value of R, the element size. These diagrams are important because they represent the information necessary to predict the hysteresis curves from the statistical characteristics of a porous network, and they may be very useful in the characterization of porous materials from experimental curves. If the volumetric uptake of capillary condensate for a certain state needs to be obtained, the volumetric degree of filling, 8,, may be calculated as assuming volumes Vs(R) and VB(R) for sites and bonds of size R, respectively.In fig. 1 to 3 several examples of domain complexions are presented, corresponding to each of the five types of porous structures; each figure depicts one of the particular processes discussed in this paper. For these examples, Gaussian site and bond size distributions were considered, the parameters of which are exactly the same as those reported in Part l 4 (cf. fig. 3-7). The connectivity, C, is always taken equal to 6. These figures clearly show the relative role played by porous entities, according to their kind and size, with respect to cooperative effects such as the assisted filling during ascending processes or the blocking behaviour during evaporation phenomena.Finally, we believe that from these diagrams one might extract topological information on the repartition of phases throughout a porous structure. Conclysions The five types of porous networks introduced in Part l 4 to describe different condensation regimes still remain useful in dealing with their associated processes involving evaporation. Previous authors studying capillary condensation and evapor- ation have proposed very simple models which are limiting cases of the present treatment. This point may be analysed more carefully elsewhere. In fact, porous networks are much more complex to deal with, since interdependence of pore domains seems to be the rule rather than the exception.However, the indiscriminate use of crude models for cooperativity may have extremely misleading results ; for example, blocking phenomena could be thought practically absent in structures having a high degree of correlation between the sizes of neighbouring elements. This work was supported by the Ministry of Public Education of Mexico (Secretaria de Educacidn Publica) under contract no. C-87-01-0116 (DGICSA reg. no. 862550).V. Mayagoitia et al. 813 References 1 D. E. Everett, in The SolikGas Interface, ed. E. A. Flood (Marcel Dekker, New York, 1967), vol. 2, 2 J. H. de Boer, in Colston Papers, vol. X : Structure and Properties of Porous Materials, ed. D. H. Everett 3 J. H. de Boer, in Everett and Stone, p. 90. 4 V. Mayagoitia, F. Rojas and I. Kornhauser, J. Chem. Soc., Faraday Trans. 1, 1988, 84, 785. 5 D. H. Everett, in Colston Papers, vol. X: Structure and Properties of Porous Materials, ed. D. H . Everett 6 J. A. Barker, in Colston Papers, vol. X: Structure and Properties of Porous Materials, ed. D. H. Everett 7 D. H. Everett, in The Solid-Gas Interphase, ed. E. A. Flood (Marcel Dekker, New York, 1967), vol. 2, 8 R. P. Iczkowski, Ind. Eng. Chem. (Fundam.), 1968, 7, 572. 9 P. H. Doe and J. M. Haynes, in Characterization of Porous Solids, ed. S . I . Gregg, K. S. W. Sing and H. F. Stoeckli (SOC. Chem. Ind., London, 1979), p. 253. p. 1059. and F. S. Stone (Butterworths, London, 1958), p. 68. and F. S. Stone (Butterworths, London, 1958), p. 117. and F. S. Stone (Butterworths, London, 1958), p. 125. p. 1087. 10 A. G. Foster, Trans. Faraday Soc., 1932, 28, 645; L. H. Cohan, J. Am. Chem. Soc., 1944, 66, 98. 11 D. H. Everett, Trans. Faraday Soc., 1954, 50, 1077. Paper 71946; Received 17th March, 1987

ISSN:0300-9599    

DOI:10.1039/F19888400801

出版商:RSC    

年代:1988

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I , 1988, 84(3), 815-826 The Kinetics of the Oxidation of Hydrogen Peroxide by Bis(2,2'-bipyridine)manganese( 111) Ions in Aqueous Perchlorate Media Malcolm P. Heyward and Cecil F. Wells" Department of Chemistry, University of Birmingham, Edgbaston, P.O. Box 363, Birmingham B1.5 2TT The kinetics of the oxidation of hydrogen peroxide by bis(2,2'-bi- pyridine)manganese(m) ions have been investigated in aqueous perchlorate media. The kinetics are found to vary with the initial ratio [H,O,]/[Mn(bipy)&] : if k, is the observed pseudo-first-order rate constant, linear plots are found for k;l against [H,O,]-' at low [H,O,] and linear plots of k , against [H,O,] with an intercept on the k, axis are found at high [H,O,]. From a detailed consideration of the variation of the kinetics with [H,O,] and [H'], the first-order decomposition of three complexes, [Mn(bipy),HO,li,+., [Mn(bi~~>,(H,o,),l~,. and [Mn(bipy),H2O2HO21~,.are found to contribute to the oxidation. Following the investigation of the kinetics of the oxidation of hydrogen peroxide by complexes of 2,2'-bipyridine with nickel(1II)l and with silver(Ir)2, to compare with the kinetics of the oxidation of hydrogen peroxide by transition metal aqua-cations4 such as Mng.,475 Ag:.,'v5 Ceir,,436 C0g.,~1~9~ and Fei!.,* in acidic conditions, we now report our results for the kinetic and stoichiometric study of the oxidation of hydrogen peroxide by bis( 2,2'- bip yridine)manganese( 111) ions. Experiment a1 Materials Solutions of bis(2,2'-bipyridine)manganese( 111) ions were preparedg by the anodic oxidation of tris(2,2'-bipyridine)manganese(11) ions under nitrogen at a platinum electrode using a current density of 0.022 Acrn-' at 36V.Solutions of sodium perchlorate were prepared by neutralising 12 mol dm-3 perchloric acid with AnalaR sodium carbonate followed by boiling to expel CO, and filtration to remove trace impurities. 86 YO stabiliser-free hydrogen peroxide was diluted as required and standardised by titration with 0.05 mol dm-3 CeIV using ferroin as the indicator. AnalaR perchloric acid was used and water was distilled once in an all-glass still. Procedure Concentrations of MnIII were determined by sampling into a solution containing excess Fe" followed by titration of the FeI' remaining against a standard CeIV solution. Rates of oxidation of H,02 by the complex were followed using a Durrum-Gibson stopped-flow spectrophotometer : reaction traces were photographed on the storage screen of a Telequipment DM64 oscilloscope.Thermostatting was achieved by circulating water from a thermostat for temperatures above the ambient temperature, and for temperatures below the latter circulation from a cryostat was used. 815816 Ox id at ion of H 20 by Bis(2,T-bipyr idine)manganese( III) 2Or '"I 7 x / 1 1 I 1 I 1 1 I (1/[H,0,1)/dm3 mol-' 0 40 80 120 Fig. 1. Plots of k;' us. [H,O,]-' for the Mn(bipy)i+ + H,O, reaction at [H,O,] < ca. 0.10 mol dm-3, 25.0 "C and an ionic strength = 1.00 mol dm-3 with varying [HClO,]/mol dm-3: a, 0.20; A, 0.40; 0, 0.60; 0, 0.80; x , 1.00.Results and Discussion Stoichiometry The bis(2,2'- bipyridine)manganese( 111) complex reacted with a small excess of hydrogen peroxide at 25 "C and an ionic strength of 1.00 mol dm-3, adjusted by the addition of sodium perchlorate. Samples from these mixtures and from similar ones containing no MnIII were pipetted into solutions containing an excess of FeII which was titrated against a standard solution of CeIV. From the concentrations of H202 determined in this way before and after reaction with MnIII the consumption ratio [Mn1'1]/[H20,] was obtained. The mean value for this ratio for varying [HClO,] (0.2-1 -0 mol dm-3) = 2.05 0.03, showing that the reaction proceeds as in the normal manner for a non-chain oxidation,1-6 viz. 2Mn'" + H202 + 2Mn" + 0, + 2H+.Kinetics of the Oxidation at Low [H,O,] Reaction traces were determined at 400 nm and 25 "C at an ionic strength of 1.00 mol dmP3, adjusted by the addition of sodium perchlorate with [H202] + [Mn'"] and with [H202] c 0.1 mol dm-3 and [Mn(bipy)i+] x 2 x loP4 mol dm-3 in the acidity range [HClO,] = 0.2-1.0 mol dm-3. In all cases, a plot of log(absorbance) against time was linear and values for the pseudo-first-order rate constant k, calculated from the slopes of these plots are collected in table 1. In some cases, as indicated in table 1, an excess addition of 4.00 x mol dm-3 of 2,2'-bipyridine was made for 0.20 mol dm-3 HClO, and the values of k, in table 1 show that this addition had no influence on the rate constant. Plots of k, against [H202] at constant acidity are curves, but, as fig.1 shows,C . F. Wells and M. P. Heyward 817 Table 1. k,/10-' s-' for Mn(bipy)i++H,O, with varying [H,0J/10-2 rnol dm-3 and [H,O,] < mol dm-3 at an ionic strength = 1 .OO mol dmP3 0.10 mol dm-3 for an initial [MnTT1] z 2 x 0.20 0.20 0.20 0.20 0.20 0.20 0.20 0.20 0.20 0.20 0.40 0.40 0.40 0.40 0.40 0.40 0.40 0.60 0.60 0.60 0.60 0.60 0.60 0.60 0.80 0.80 0.80 0.80 0.80 0.80 0.80 0.80 1 .oo 1 .oo 1 .oo 1 .oo 1 .oo 1 .oo 1 .oo 1 .oo 1 .oo 1 .oo ~- ~-~ ~~ 25.0 "C 15.9 "C _ _ _ _ _ _ ~ [H'I /mol dm-3 [H,O,] 0.73 0.78 1.10 1.22 1.27 2.44 2.54 3.80 4.88 5.07 0.76 1.10 1.27 1.77 2.54 3.80 5.07 0.76 1.10 1.27 1.77 2.54 5.07 0.82 1.36 I .90 2.69 4.08 5.38 - - - 0.76 1.30 I .90 2.72 3.80 4.08 5.20 __ - - 1.14" 1.19 1.41 1.52" 1.61 1.96" 1.98 2.15 2.35" 2.35 0.90 1.01 1.15 1.18 1.26 1.47 1 S O 0.75 0.79 0.89 0.97 0.99 1.10 0.61 0.74 0.75 0.8 1 0.86 0.93 - - - 0.53 0.60 0.68 0.63 0.63 0.67 0.64 - - - 0.28 0.43 1.13 6.55 - - - - - - 0.29 0.43 0.58 1.13 2.83 - - 0.34 0.43 1.13 2.83 2.89 4.34 5.66 0.28 0.43 0.58 2.83 5.66 - - - 0.28 0.29 0.43 0.58 1.45 2.83 2.89 4.34 5.66 - ko 0.53 0.75 1.12 1.54 - - - - - - 0.46 0.57 0.62 0.86 1.03 - - 0.43 0.50 0.67 0.79 0.83 0.90 0.92 0.36 0.43 0.49 0.62 0.70 - - - 0.30 0.33 0.38 0.42 0.52 0.56 0.56 0.59 0.61 - 31.5 "C 40.2 "C 0.26 0.32 0.40 0.65 1.29 2.8 1 5.16 0.26 0.40 0.53 1.05 2.63 - 0.26 0.32 0.65 I .29 2.8 I 5.62 0.26 0.40 1.05 2.63 5.26 - - 0.26 0.36 0.40 0.60 1.05 1.21 2.63 3.95 4.82 5.62 ko 0.96 1.16 1.25 1.66 2.30 3.05 2.85 - _- - 0.86 1.01 1.20 1.41 1.64 - - 0.73 0.78 0.99 1.16 1.29 1.86 0.65 0.73 0.87 0.97 1.09 - - - - 0.5 1 0.60 0.56 0.68 0.77 0.79 0.76 0.76 0.83 0.9 1 a Contains an extra 4.0 x lop3 rnol dmP3 bipyridine.0.39 9.64 1.29 2.57 5.14 - - - __ - 0.4 1 0.82 1.29 2.57 3.26 6.12 0.43 0.76 0.87 1.52 1.73 4.33 0.39 0.64 0.82 1.13 1.63 3.26 5.14 6.12 0.34 0.41 0.58 0.82 1.13 1.63 3.26 6.12 - - - - 1.90 2.60 3.40 3.80 4.70 - - ~- - - 1.45 1.81 1.84 2.20 2.30 2.95 1.23 1.40 1.40 1.58 1.63 1.75 0.92 1.10 1.08 1.14 1.22 1.53 1.62 1.61 0.78 0.79 0.87 0.93 0.96 1.12 1.52 1.61 - - - - plots of k;' against [H,O,]-l are straight lines, excluding the points at high [H,O,] where, as will be shown later, an alternative mechanism starts to have an influence on the kinetics. Similar linear plots for log(absorbance) against time were observed at 15.9, 3 1.5 and 40.2 "C and the values of k, for a range of acidities for each temperature are also given818 Oxidation of H20, by Bis(2,2'-bipyridine)manganese( III) in table 1.Again, plots of kil against [H,O,]-l for constant acidity and temperature are linear, those at 15.9 "C being shown in fig. 2. In all these cases, as illustrated in fig. 1 and 2, although the slopes of the plots of k;l against [H,O,]-' are insensitive to changes in [H+], the intercepts show a considerable variation. Kinetics of the Oxidation at High [H202] As indicated above in fig. 1, deviations from the linearity of plots of k;' against [H20,]-' tend to occur at [H20,] = 0.10 mol dm-3. For [H20,] > 0.10 mol dm-3, plots of log(absorbance) against time are always linear and the values of k, with varying acidity for all the temperatures for [H,O,] > 0.1 mol dm-3 are given in table 2.In this region, a plot of k, against [H,02] is found to be linear at constant acidity and constant temperature. This is illustrated in fig. 3 for the data at 25 "C: now the points at low [H20,] (ca. 0.10 mol dm-7 deviate from these straight lines owing to the change over in this region from one mechanism to another. Excluding points in fig. 3 where [H,O,] < 0.2 mol dm-3, straight lines can be drawn with an intercept on the k,-axis. Mechanism of the Oxidation The linear plots of kil against [H20,]-' obtained at [H20,] < 0.1 mol dm-3 suggest the involvement of an intermediate complex between [Mn(bipy),]:,+. and H,O,. We therefore consider pre-equilibria ( 1)-(4) with processes ( 5 ) and (6) as possible rate-determining steps producing HO; radicals which will be consumed in a subsequent rapid reaction (7).K h [Mn(biPY),l:,+.= [Mn(biPY),OHl:,+. + H+ (1) [Mn(biPY),l::. + H,O,= [Mn(biPY),H,O,l:,+. (2) (3) B1 K, W(biPY) 2H2021:,+.= [Mn(bipy )2Ho21::. + H+ [Mn(bipy),H,O,]:,f.: Mn" + H0; + H+ [Mn(bipy),HO,]z,f. 5 Mn'I + HO; ( 5 ) (6) (7) MnIrr + HOizt Mn" + 0, + H+ If an intermediate 1 : 1 complex is formed between Mn(bipy):+ and H202 at low [H,O,], it is likely that complexes involving more than one H,O, attached to each Mn1I1 will be involved at higher [H,O,], and the existence of the intercepts in fig. 3 shows that more than one rate-determining oxidative step is involved when [H,O,] > 0.10 mol dm-3.Possible pre-equilibria involving complexes with more than one H,O, as a ligand are (8)-(11) at [H,O,] > 0.10 mol dm-3. 82 [Mn(biPY),H,O,I:,f. + H,O,* [Mn(biPY),(H,O,),I~,+. (8)C. F. Wells and M . P. Heyward 819 I I I I 4 0 100 200 300 400 ( 1 /[H202 ])/dm3 mol-' Fig. 2. Plots of k;' us. [H,O,]-' for the Mn(bipy)i+ + H202 reaction at [H,02] < ca. 0.10 mol dmP3, 15.9 "C and ionic strength = 1.00 mol dm-3 with varying [HClO,]/mol dm-3: m, 0.20; A, 0.40; 0, 0.60; 0, 0.80; x , 1.00. 0.2 X I I I I 1 0 0.2 0.4 0.6 0.8 1.0 [ HzOz ]/mol dm-3 Fig. 3. Plots of k, us. [H,02] for the Mn(bipy)i+ + H202 reaction at [H20,] > ca. 0.10 mol dmP3, 25.0 "C and an ionic strength = 1.00 mol dmP3 with varying [HClO,]/mol dm-': ., 0.20; A, 0.40; 0, 0.60; 0, 0.80; x , 1.00.820 Oxidation of H,O, by Bis(2,2'-bipyridine)manganese( 111) Table 2.ko/s-l for Mn(bipy),3+ + H202 with varying [H20,]/ lo-' mol dmP3 and [H202] > 0.10 mol dm-3 for an initial [MnI'I] x 2 x lop4 rnol dm-3 at an ionic strength = 1.00 mol dm-3 25.0 "C 15.9 "C 31.5 "C 40.2 "C 0.20 1.35 0.20 2.25 0.20 2.44 0.20 4.50 0.20 5.13 0.20 6.75 0.20 7.70 0.20 8.98 0.20 9.00 0.40 1.35 0.40 2.42 0.40 7.73 0.40 10.3 0.40 - 0.60 1.35 0.60 2.42 0.60 2.55 0.60 5 .OO 0.60 5.10 0.60 7.50 0.60 10.3 0.80 1.26 0.80 2.49 0.80 2.50 0.80 5.16 0.80 7.50 0.80 7.6 1 0.80 10.0 0.80 10.1 1 .oo 1.22 1 .oo 2.58 1 .oo 5.07 1 .oo 7.73 1.00 10.1 1.00 10.3 2.80 3.50 3.80 4.85" 5.5 5.9 6.5 6.7 7.3" 2.10 2.45 3.85 4.80 1.46 2.20 2.10 2.05 2.10 2.65 3.75 1.18 1.35 1.39 1.87 2.15 2.80 3.30 3.50 0.85 1.40 1.76 2.25 2.80 2.75 - 1.31 2.54 2.62 5.08 5.18 5.24 7.62 10.2 - 2.62 5.24 7.86 10.5 - 2.61 5.23 7.76 10.5 - - - 2.61 5.23 7.84 10.5 - - - - 2.54 2.59 5.08 7.62 10.2 - 1.83 1.96 2.00 2.20 2.30 2.25 3.35 3.85 1.26 1.54 1.90 2.30 1.03 1.27 1.27 1.93 - - - - - 0.77 1.02 1.30 1.38 - - - - 0.75 0.65 0.84 1.15 1.64 - 2.57 5.14 7.71 10.3 - - - - - 2.55 5.10 7.65 10.2 - 2.55 5.10 7.65 10.2 - - - 2.68 5.36 8.04 10.7 - - - - 2.57 5.14 7.7 1 10.0 - - 5.10 6.50 10.2 12.6 - - - - - 3.25 5.4 6.9 8.7 3 .OO 4.60 6.5 8.4 - - - - 2.50 3.90 5.7 7.0 - - - - 2.15 3.34 4.45 5.9 - - 2.56 4.92 7.67 9.84 - - - - - 2.56 4.92 7.67 9.84 2.59 5.18 7.76 9.84 10.2 - - - 2.53 5.06 7.59 10.1 - - - - 2.53 5.06 7.59 10.1 - - 7.8 12.4 16.1 19.6 - - - - - 7.1 10.4 14.4 17.0 18.6 6.0 9.4 12.1 14.3 - - - 4.60 7.8 10.7 13.2 - - - - 3.70 5.2 8.2 10.7 - - " Contains an extra 4 x mol dm-3 bipyridine.and these could be followed by rate-determining oxidative steps (1 2)-( 14), [Mn(bipy),(H,O,),]~. 2 MnI' + H202 + H0; + H+ [Mn(bipy),(H,O,)HO,]~,+. 5 Mn" + H,O, + HO; K" [Mn(bipy),(HO,),]zq. -3 Mn" + HO, + HO;. Again, HO; radicals will be consumed by the rapid reaction (7) producing molecular oxygen.C. F. Wells and M . P. Heyward 82 1 Table 3. Slopes = Kh/2k'4 K, and intercepts = (h + K,)/2k'KI of the plots of k;' against [H,O,]-' at [H,O,] < 0.10 rnol dmT3 and at ionic strength = 1.00 mol dmP3 [H'l slope T/"C /mol dmP3 rnol s dm-3 intercept/s 25.0 25.0 25.0 25.0 25.0 15.9 15.9 15.9 15.9 15.9 31.5 31.5 31.5 31.5 31.5 40.2 40.2 40.2 40.2 40.2 0.20 0.40 0.60 0.80 1 .oo 0.20 0.40 0.60 0.80 1 .oo 0.20 0.40 0.60 0.80 1 .oo 0.20 0.40 0.60 0.80 1 .oo 3.84 & 0.16 3.94 k 0.3 1 3.92 f 1.43 3.55 k 1.63 4.03f1.18 3.64 k 0.19 3.91 k0.23 4.26f0.19 3.84 f 0.18 4.45 f 0.2 1.93 f 0.26 1.61 kO.10 1.71 f0.06 1.59 +O.16 2.06 & 0.2 1 1.26 f 0.07 1.32 0.24 1.16 & 0.10 1.29 + 0.20 1.28 & 0.08 3.47 f 0.1 1 6.0 & 0.2 8.1 2 1 11.1 f 1.2 13.6 & 0.8 5.5 k0.3 8.5 f0.5 10.6 0.3 14.1 k0.4 16.1 f0.4 2.66 k 0.59 5.5 f0.2 7.3fO.I 9.5 k0.3 11.6 f 0.4 1.98 f 0.10 3.87 & 0.43 5.6f0.1 7.5 k0.3 9.3 f0.2 Several of the pathways to this final stage are mutually indistinguishable kinetically, notably (1) + (4) and (2) + (3) or (8) + (9) and (3) + (1 l), and the former set will be used in each case. The overall kinetic equation for the decay of the total concentration of MnII' is (1 5 ) - 2{k + k'K, h-l+ (k, + ki K2 h-' + ki K, K; h-2)P2 H202)~,[Mn111]to,,,[H,0,] - 1 + Kh h-l+ { 1 + Kl h-' + (1 + K2 h-l+ K, Ki h-2)~2[H,0,]}~l[H202] where h = [H;J At [H20,] < 0.10 mol dm-3, if it is assumed that Kh h-' + (1 + K , h-l)P,[H,O,] % 1 + (1 + K2 h-' + K2 K;h-2)PlP2[H20,]2 k'Kl h-l % k + (k, + k; K2 h-l + k; K2 K; h-*)P2[H2O2] and eqn (15) can be rearranged to give (h + Kl) +- k, 2k'/?,Kl[H2O,] Zk'K, * 1 - K h - _ Table 3 contains the values of the slopes and intercepts of the plots of k,' against [H20,]-' at [H20,] < 0.1 mol dm-3calculated using the least-squares procedure.This table shows that the slopes of these plots are independent of the acidity at each temperature for [H,O,] < 0.10 mol dm-3, as required by eqn (16): at 25 "C, the mean slope = 3.86k0.17 x lo-'; at 15.9 OC, the mean slope = 4.02f0.29 x at 31.5"C, the mean slope = 1.78kO.18 x lo-,; and at 40.2 O C , the mean slope = 1.26kO.09 x lo-,.Fig. 4822 Oxidation of H202 by Bis(2,2'-bipyridine)manganese(111) Fig. 4. Plots of the intercept ( J ) of the plots of k;' us. [H,O,]-' at [H,O,] < 0.10 mol dm-3 against h = [HClO,] for the Mn(bipy)i+ +H,O, reaction at various temperature/"C: 0, 15.9; x ,25.0; A, 31.5; 0, 40.2. Table 4. Values derived for k', k'K,, Kl and PJK, for [H,O,] < 0.10 at ionic strength = 1.00 mol dm-3 k'/10-' dm3 w 1 0 - 2 (BllK,) T/"C rno1-l s-l k'Kl/10-2 s-' mol dm-3 /dms mol-2 15.9 1.72k0.25 3.73f0.17 21.7f4.2 335 25.0 5.9f1.6 3.94+0.10 6.7f2.1 330 31.5 6.6k3.1 4.57k0.18 6.9k4.4 620 40.2 30.7 _+ 15.5 5.5 0.1 1.8-t 1.2 730 shows that the intercepts of these plots for [H202] < 0.10 mol dm-3 in table 3 vary linearly with h at each temperature, in conformity with eqn (16).From eqn (16), the slope of fig. 4 = 1 / 2 k K , and the intercept = 1 /2k', from which the values of Kl can also be determined : the values for k , k'K, and Kl calculated from a least-squares analysis of the slopes and intercepts of fig. 4 are given in table 4. From eqn (16), the slopes of the plots of k;' vs. [H20,]-' given in table 3 = Kh/2k'KJ?,; the values for obtained by combining the values of these slopes with k K l in table 4 are also given in table 4. From the linear plot of log k' against the reciprocal of the absolute temperature using the least- squares procedure, AH* = 40 +_ 8 kJ mol-l and AS* = 25 f 50 J K-' mol-' for the rate- determining step (6); and for a similar plot of log k K l against the reciprocal of theC.F. Wells and M. P . Heyward X 823 X L - I 0 0.5 1 .o 0 0.5 1 .o [H20,1/mol dm-3 [H,O,]/mol dm-3 Fig. 5. Plots of k,(h+Kl)h against [H,O,] for [H,O,] > 0.10 mol dm-3, at a constant ionic strength = 1.00 mol dmP3 for (a) 31.5 and (b) 40.2 "C for the Mn(bipy)i+ reaction with varying [HClO,]/mol dm-3: W, 0.20; A, 0.40; 0 , 0.60; 0, 0.80; x , 1.00. absolute temperature, the least-squares procedure produces AH* = 3.6 and AS* = - 238 1.2 kJ mo1-I 8 J K-l mol-l for the overall process (3) + (6). At [HzOZ] > 0.10 mol dm-3, if it assumed that (1 + Kl h-l) pl[H,02] 9- 1 + K , h-l+ (1 + K2 h-l+ Kz K ; k 2 ) p 1 p2[H202]z, eqn (17) can be derived by rearranging eqn (15): 0.5k0(h + K l ) h = k'Kl h + ( k , h2 + ki K2 h + ki K , Ki)p2[H20,].(17) Values for the left-hand side of eqn (17) can be calculated from the values for k, in table 2 and the values for Kl in table 4. Plots of k,(h + K,) h against [Hz02] for [H20,] > 0.10 mol dm-3 at constant acidity and temperature are straight lines with an intercept on the ordinate: examples of these for 31.5 and 40.2 "C are shown in fig. 5. The slopes and intercepts of these plots determined using the least-squares method are collected in table 5. According to eqn (1 7), the intercepts = 2k'K, h for the plots in fig. 5, from which the value of k K , can be calculated.These latter values are given in table 6 and they agree well with the values of k'Kl in table 4 obtained from the plots in fig. 4. A least-squares treatment of the linear plots of log k'K, against the reciprocal of the absolute temperature for the data in table 6 gives AH* = 6.2k0.3 kJ mol-1 and AS* = -222f 12 J K-l mol-l, which agree well with AH* and AS* calculated above from the data in table 4 obtained at [H20z] < 0.10 mol dm-3. Plots of the slopes S of the variation of ko(h+ Kl)h with [H,02] against h at constant temperature are curves, indicating that more than one function of h is involved. As these curves pass through the origin, it is reasonable to assume that ki K2 Ki 4 k , h2 + ki K,h, and, accordingly, fig. 6 shows that the plots of Sh-l against h using the data in table 5 are straight lines for 15.9, 25.0 and 31.5 "C, with an intercept on the ordinate, with some scatter on the plot at 40.2 "C.The values for kz& and k;P2K2 obtained using a least-squares calculation are given in table 6. From the application of the least-squares method to the plots of log k2P2 and log k; p2 K2 against the reciprocal of the absolute temperature, AH* = 49.1 k 4.0 kJ mol-' and AS* = 70f26 J K-l mol-1 for k2PZ and processes (8)+(12) and AH* = 18.8k3.2 and AS* = - 133 & 21 J K-l mol-1 for kip, K, and processes (8) + (9) + (1 3).824 Oxidation of H , 0 , by Bis( 2,2'- b ipy Y idine)mang anese( III) Table 5. Slopes and intercepts from the plots of k,(k+K,)h against [H,O,] at [H,O,] > 0.10 at ionic strength = 1 .OO mol dm-3 [H+] slope/ 1 0-2 mol T/"C /mol dmP3 s-' dm-3 intercept/ mo12 dmP6 s-l 25.0 25.0 25.0 25.0 25.0 15.9 15.9 15.9 15.9 15.9 31.5 31.5 31.5 31.5 31.5 40.2 40.2 40.2 40.2 40.2 0.20 0.40 0.60 0.80 1 .oo 0.20 0.40 0.60 0.80 1 .oo 0.20 0.40 0.60 0.80 1-00 0.20 0.40 0.60 0.80 1 .oo 2.86 5 0.23 5.5 * 0.2 9.9f0.1 16.9+ 1.6 22.3 f 1.3 1.88 & 0.33 3.31 fO.O1 5.1 f 0.5 6.2 f 2.9 12.8 & 1.9 5.4 f 2.4 13.2 f 0.7 27.45 1.4 40.7 f 1.2 54f2 7.0 f 0.2 25.5f 1.1 48.0 f 4.3 80f3 103 f 5 1.21 f0.12 3.13f0.14 3.81 k0.63 5.2f 1.0 7.0 f 0.8 1.12 f 0.17 2.16f0.14 3.89f0.31 5.1 f 1.9 5.4f 1.1 1.19+ 1.95 3.00 & 0.48 5.3 f0.7 5.9 & 0.8 7.7 f 1.3 1.73 k0.13 4.38 & 0.77 7.5 f 2.6 7.8f 1.6 10.3 f 3.2 Table 6.Values for k'K1,-k, /?, and ki 8, K2 with [H,O,] > 0.10 mol dm-3 at ionic strength = 1.00 mol dm-3 k$2/10-2 dm3 T/"C k'K,/10-2 s-' mol-' s-' k i p , K,/lO-, s-' 15.9 2.92 & 0.25 1.5+ 1.5 3.85 1.1 25.0 3.37 + 0.3 1 5.8f 1.2 5.3 k0.8 31.5 3.73 f0.47 18.1 f2.4 10.1 f 1.6 40.2 5.2 & 0.7 39.4 f 8.3 13.8 f 5.5 Comparison with other Cation + H202 Systems Although there is uncertainty from the kinetics about which equilibria participate in the production of the species involved in the rate-determining steps, the kinetics do specify that such pre-equilibria between Mn(bipy)&.and H20, must exist. The kinetics also specify which rate-determining processes contribute to the oxidation, depending on the conditions employed. Thus, for initial ratios of reactants [H,O,]/[Mn(bipy)~&] < 500, only one oxidative step is involved, process (6) involving [Mn(bipy),HO,];,+.i.e. when only 1 : 1 complexes are formed between [Mn(bipy),]Ei. and H202, there is only one oxidative step resulting in the conversion MnI'I -+ Mn". However, when the initial ratio [H202]/[Mn(bipy)3,+] exceeds 500,2: 1 complexes of H,O, with [Mn(bipy),]z. begin to be formed and, in addition to reaction (6), reactions (12) and (1 3) begin to contribute to the conversion MnI'I -+ Mn" : although reaction (14) cannot be totally excluded, its contribution to the conversion Mn'II-+Mn" must be very small compared with reactions (6), (12) and (13). It is interesting to compare this mechanism for the oxidation of H202 by bis(2,2'-C. F. Wells and M . P. Heyward 825 Fig. 6. [H,O,I Plots of Sh > 0.10 mol reaction bipyridine)manganese(m) ions with those found for the oxidation of H202 by complexes of 2,2'-bipyridine with other transition-metal cations and with those found for the oxidation of H202 by aqua-cations of transition metals.For the oxidation of H20, by Ni(bipy)i+, where the Ni cation is entirely surrounded by bipyridine molecules, no intermediate complexes were detected ;' and, likewise, for the oxidation by Ag(bipy)i+, which also appears to have no water in its coordination sphere,, no intermediate complexes have been detected.2 The complex of bipyridine with MnIII has only two bipyridine ligands attached to each Mnlll,g and this complex cation has been formulated by Cooper and Calvin'' as [Mn(bipy),(H,O)OH]E,., with two water molecules coordinated to each Mn cation.It is perhaps not surprising then that this cation, possessing additional space to accommodate two H,O molecules in the inner coordination sphere, can interchange both sequentially with H202 molecules. Moreover, just as [Mn(bipy),(H,O)OH]~,. has one water molecule with a proton lost, similarly [Mn(bipy),(H,O)HO,]~~. has a proton removed from the H,O,. In the oxidation of H202 by Co(NH,)i+, the suggested substitution control'' is unlikely owing to the contrast12 of AS* with AS* for substitution by Cl- and Br-, and no intermediate cation + H,O, complexes have been detected for initial ratios [H,O,]/[Co(NH,)~+] < 300-1000.12 With Co(en)(H,O);+, a cornpari~on'~ of a rate constant for the oxidation of H,02 at one temperature with those for substitution by C1- and Br- at another temperature is inadequate grounds for concluding that the oxidation is substitution controlled : with the initial ratio [H,O,]/[Co(en)(H,O>~+] limited to < 63, no intermediate complexes were detected.826 Oxidation of H ,O , by Bis(2,2'- bipy r idine)manganese( III) Such ~xidations~-~ by Co'", Mni:., Ceii.and F e g involve intermediate complexes with H,O,, but only with FeiThave complexes with a ratio of H,O,:cation > 1 : 1 been suggested. In a comparison of AH* and AS* for the oxidation of H,O, by aqua- complexes of transition-metal cations with E" of the cation, Cog. deviates from the order found for the other aqua- cation^.^ Similar deviations have been found with Cog. for the oxidation of other substrates by these cations." It has now been confirmed in a careful redeterminati~n'~ that E" for Co;.is 1.88 V as assumed in the above comparisons, contrary to a suggestion that E" = 1.45 V,16 so that Cot: remains a kinetic deviant from the other aqua-transition-metal cations in these oxidations. An explanation suggested* for this deviation is that CoiE participates in two-electron transfers, possibly via a dimeric species, with the other aqua-cations involved only in single-electron transfers : this was supported by the apparent necessity to postulate the existence of such dimers to account for the observed kinetic orders found in the oxidation of water7. l7 and of Br,18 by Cog. The independence of E" of log[CoI'I] that only one species is present in the high concentration range [CO"'] = 0.018-0.63 mol kg-l (the kinetics use [CoI''] < 1 x loA3 mol dmP3) without specifying which species this is.In common with earlier measurements of Eo for C O ~ . , ~ ' Biedermann et aZ.15 found it necessary to add Ag+ ions as a potential mediator, and whilst, of course, this does not affect the thermodynamic conclusions, neither does it resolve the mechanistic doubt about the species of Cog. responsible for the slow equilibration with the electrode,l57 l9 the complex kinetic orders in the oxidation of water7* l7 and Br,18 and the deviations found from other transition-metal cations reacting with H202 and other substrates. One possibility is that at the high ionic strengths used in the kinetic investigations, species of CoIII may cluster together and pairs may participate in two- or one-electron transfers as appropriate. References 1 C.F. Wells and D. Fox, J. Chem. SOC., Dalton Trans., 1977, 1498. 2 M. P. Heyward and C. F. Wells, J . Chem. SOC., Dalton Trans., 1981, 1863. 3 H. N. Po and K. D. Chen, Inorg. Chim. Acta, 1975, 14, 173. 4 C. F. Wells and D. Fox, J . Inorg. Nucl. Chem., 1976, 38, 107. 5 C. F. Wells and D. Mays, J. Chem. SOC. A , 1968, 665. 6 C. F. Wells, and M. Husain, J. Chem. SOC. A , 1970, 1013. 7 J. H. Baxendale and C. F. Wells, Trans. Faraday SOC., 1957, 53, 800; C. F. Wells and M. Husain, Trans. Faraday SOC., 1971, 67, 760. 8 W. S. Anderson, Acta Chem. Scand., 1948, 2, 1 ; 1950, 4, 914; J. A. Christiansen and V. S. Anderson, Trans. Faraday SOC., 1950, 4, 1538; W. G. Barb, J. H. Baxendale, P. George and K. R. Hargrave, Trans. Faraday SOC., 1951,47,59; J. Koefoed, Acta Chem. Scand., 1955,9,283; P. Jones, R. Kitching, M. L. Tobe and W. F. K. Wynne-Jones, Trans. Faraday SOC., 1959,55,91; M. L. Kremer and G. Stein, Trans. Faraday SOC., 1959, 55, 959. 9 M. P. Heyward and C. F. Wells, Transition Metal Chem., 1987, 12, 179. 10 S. R. Cooper and M. Calvin, J. Am. Chem. SOC., 1977,99, 6623. 1 1 I. Bodek and G. Davies, Inorg. Chem., 1975, 14, 2580. 12 A. F. M. Nazer and C. F. Wells, J. Chem. SOC., Faraday Trans. I , 1985, 81, 801. 13 N. S. Rowan, C. Y. Price, W. Benjamin I11 and C. B. Storm, Inorg. Chem., 1979, 18, 2044. 14 C. F. Wells and D. Mays, J . Inorg. Nucl. Chem., 1971, 33, 3855; R. Varadarajan and C. F. Wells, J. Chem. SOC., Faraday Trans. I , 1973,69,521; 1980,76,2017; C . F. Wells and A. F. M. Nazer, J. Chem. SOC., Faraday Trans. I , 1976, 72, 910; M. P. Heyward and C. F. Wells, J . Chem. SOC., Dalton Trans., 1982, 2185. 15 G. Biedermann, S. Orecchio, V. Romano and R. Zingales, Acta Chem. Scand., Set A , 1986, 40, 161. 16 A. ,L. Rotinjan, L. M. Borisowa and R. W. Boldin, Electrochim. Acta, 1974, 19, 43. 17 B. Sramkova, J. Zika and J. Doleial, J . Electroanal. Chem., 1971, 30, 169. 18 C . F. Wells and D. Mays, J. Chem. SOC. A , 1968, 2740. 19 B. Warnqvist, Inorg. Chem., 1970, 9, 682; G. Davies and B. Warnqvist, J . Chem. SOC., Dalton Trans., 1973, 900; A. A. Noyes and T. J. Deahl, J. Am. Chem. SOC., 1937, 59, 1337. Paper 7/497; Received 19th March, 1987

ISSN:0300-9599    

DOI:10.1039/F19888400815

出版商:RSC    

年代:1988

数据来源: RSC

摘要:

J . Chem. SOC., Faraday Trans. I , 1988, 84(3), 827-840 Absorption, Magnetic Circular Dichroism and Magnetic Circular Polarization of Luminescence Studies of Ru(bpy)z+ and Complexes with a Di(ethoxycarbony1)-substituted Bipyridine Ligand as a Probe of Rigid Environments Elmars Krausz Research School of Chemistry, Australian National University, G.P.O. Box 4, Canberra 2601, Australia The ruthenium tris(bipyridine) (RBY) chromophore and mono di(ethoxy- carbonyl) derivatives have been studied by low temperature absorption, magnetic circular dichroism (MCD) and magnetic circular polarization of luminescence (MCPL) spectroscopy in a range of rigid systems of varying character. These include 4 : 1 ethanol-methanol glass (EM), poly- (methylmethacrylate) (PMMA), poly(viny1 alcohol) (PVA) and Nafion, a transparent ion exchange medium based on poly(tetrafluoroethane). Absorption and MCD spectra of RBY show some variation in spectral width with medium, but the luminescence and MCPL spectra show more significant variations in position, magnitude and profile as a function of environment. The reduction in MCPL and profile changes in luminescence for RBY in Nafion are taken as evidence for a strong (environmentally induced) distortion in the excited state, giving some degree of localization in this environment.This is matched with the markedly increased photo- chemical activity of RBY in the Nafion environment. MCD spectra of the substituted complexes are markedly different to those of the parent RBY, indicative of a dominant low symmetry potential influencing the metal ti core in the excited state.Luminescent states are less (but significantly) affected, indicative of their largely spin-derived magnetic moments. The Ru(bpy)i+ chromophore/luminophore has been the subject of a great number of studies.l It is well known for its photochemical activity and its uses as a sensitizer, and for its central role as a model charge-transfer luminescent system, and these properties have prompted a great deal of discussion in the recent literature. Of particular interest to us most recently has been the change in luminescence properties of RBY and related systems in passing from rigid to fluid environments. Evidence has been given for time- dependent luminescence spectra,2. changes in the excited-state resonance Raman (ERR) spectra4 and associated changes in the luminescence profile together with a sharp drop in the magnetic circular polarization of luminescence (MCPL) magnitude. '* These phenomena can be attributed to an environmentally driven localization process of RBY.In crystalline hosts, there is little doubt that the absorbing and luminescent states of RBY, characterized as a metal-ligand electron transfer process, can only be described as involving the three bpy ligands equivalently (a delocalized process). This is clearly established by excitation polarization studies6 on RBY in a single crystal Zn(bpy), (PF,), host lattice. In fluid environments, the excited state distorts rapidly, in less than 10 ps in water at room temperature as judged from time resolved luminescence work,' so that the excitation seems principally localized on a single ligand on this timescale. Localization is very strongly evidenced by solution ERR measurements being consistent with a bpy- species' and a marked MCPL reduction relative to the rigid solution Measurements of ERR in the solid phase however lead to an entirely different spectrum for RBY but a similar spectrum for a distinctly localized c~mplex.~ 827828 Absorption, MCD and MCPL Studies of Ru(bpy);+ Arguments supporting localization in the fluid phase are also made from studies of excited-state absorption studie~.~ Recent measurements of excited-state absorption spectra," extended into the near i.r., show significant changes in passing from rigid to fluid phases, invalidating previous assignments made in the visible and near i.r.excited- state absorption spectra. In this paper a comparison of absorption and MCD, luminescence and MCPL spectra are given for RBY and [Ru(4,4'-(C02Et),bpy) (bpy),I2+ (R4M) and [Ru(5,5'-(C02Et), bpy) (bpy),I2+ (R5M) substituted complexes, in a number of rigid environments of varying character. MCD measurements are compared to measurements some of which have been previously reported by Thomson et a1.l' for RBY in EM and by Ferguson et a1.l' for R4M and R5M in PVA. MCPL measurements are related to those for RBY, R4M and R5M in EM reported by Ferguson and K r a u ~ z . ~ ~ ~ ~ ~ Experiment a1 The spectroscopic techniques used have been previously described. l5 Luminescence spectra were corrected for the spectral response of the instrument and were excited using 180" excitation geometry and polarization-scrambled radiation at 458 nm from an Ar+ laser.Care was taken to avoid sample heating by the exciting radiation. The excitation dependence of luminescence and MCPL spectra was not investigated, though small effects have been shown to exist for RBY in PVA and EM.14 Absolute magnitudes of A I / I are correct to 10 %, though relative measurements are more accurate. Rigid alcohol glasses were made by rapid cooling of degassed solutions of Ru(bpy),- C1, dissolved (ca. mol dm-3) in 4: 1 ethanol-methanol mixtures. Glasses were then warmed to 120 K and then cooled further. This process removed the residual strain generated by the initial cooling process and little significant depolarization of transmitted light occurred after this treatment.Strain birefringence was monitored by measuring a standard CD solution of resolved Co(en);+ placed after the sample and observing any change in the CD spectrum. The original Nafion 117 plastic supplied by DuPont is extremely birefringent and useless for MCD or MCPL spectra. Plastic films of PVA and PMMA were made by evaporating aqueous and CCl, solutions of the chloride salt and the polymer, respectively. Nafion films were prepared from solutions of 117 Nafion (Du Pont) as previously described,16-la and soaked in dilute aqueous solutions of RBY, R4M and R5M (the latter as the perchlorate salt) until the maximum absorbance in the visible region approached unity. The counter-ion or solubility of the species is of no consequence in preparing films, which constitutes a considerable practical advantage over PVA and PMMA where achievable con- centrations can be quite limited.Nafion films were cast either at room temperature (Nafion A) or 100 "C (Nafion B). The films cast at higher temperature seemed less crystalline, and were much less brittle than those cast at room temperature, as initially described. Species infused into these two different types of Nafion films had slightly different spectra. Both types could (importantly for MCD and MCPL) be made free of significant birefringence or absorption in the near u.v.-visible and near i.r. regions but the B-type films thermally cycled without cracking and were generally much more flexible and easy to handle. This clearly reflects the lower degree of crystallinity arising from their formation at higher temperatures.Photochemical damage to RBY in both Nafion A and B films with high power laser excitation (second harmonic Nd/YAG at 530nm) at any temperature, or by long exposure to sunlight, was noted. The characteristic yellow colour changed to a more reddish hue and red-orange luminescence became almost completely quenched in the limit of extended illumination. RBY in PVA films showed far less photochemical sensitivity. It was determined however that all the samples used in this work did not degrade significantly during the process of measurement,E. Krausz 829 Fig. 1. Absorption (lower) and MCD spectra taken at 5 K and 5 T applied field for Ru(bpy)i+ in 4 : 1 ethanol-methanol (- - - - -), poly(viny1 alcohol) (-), and Nafion A (see text) (---).Absorption spectra have been scaled to the same peak absorbance to allow a comparison of absorption and MD profiles. Molar extinction values have been presented by Thomson et al." Results and Discussion Absorption and MCD Spectra RBY Fig. 1 shows the absorption and MCD of RBY in EM, PVA and Nafion A. A useful absorption spectrum in PMMA could not be obtained because of the limited solubility of the chromophore, which caused crystallization of the solute when the film was formed at high concentrations. The MCD spectrum of a dilute PMMA film was observed and was altogether very similar to those in fig. 1. RBY in Nafion B has very similar absorption and MCD spectra to Nafion A except for slightly greater (by ca.10 O/O) linewidths. The PVA-based spectrum is in reasonable accord with that presented by Ferguson, et a1.12 and the EM based ones in quite good agreement with that reported in EM by Thomson et al." There are five reasonably easily identifiable bands in the visible near U.V. region reported here, each having its own characteristic MCD features. The assignments of Thomson et a1.l' have been shown to be in error by Ferguson et aZ.,'' as they associate the strongest MCD feature with the strongest absorption feature, and this is clearly incorrect. The strong negative A term arises from a shoulder near 21 500 cm-l that can be seen clearly from the spectrum of RBY in EM in particular. This shoulder is also the transition that gives rise to a strong CD signal'' and furthermore is the source of the830 Absorption, MCD and MCPL Studies of Ru(bpy)i+ excitation polarization anomaly in EM in the same region.6' 2o It had been incorrectly postulated21 that this anomaly was intrinsic to the RBY system itself, and indicated a point in the absorption spectrum where the RBY system transformed from localized to delocalized behaviour, but these ideas are not now tenable.Those bands that have an A term can be confidently assigned as E states, and other (non-degenerate levels) assigned as A, or A , in 0:. The main peak, labelled A,, between 22000 and 22 500 cm-l in the systems under study, does not give rise to a distinct A term but perhaps has some negative B term component, giving rise to some of the asymmetry in the A term observed.The peak Eb near 24000 cm-' gives a clear positive A term, and its magnitude is very similar to that associated with the shoulder Ea, corresponding to an aJD, of around 1 bohr magneton.7 The smaller feature near 26000 cm-l seems to also have a positive A term MCD and can thus be usefully associated with band Eb and is labelled pb' The separation between these bands in each case is close to 1800 cm-l, which most likely corresponds to a ring mode in the bpy system. Bands A,, Ea, Eb and p b all shift to higher energy by equal amounts (ca. 500 cm-l) upon going from EM to PVA to Nafion. This shift corresponds to a reduction in optical dielectric constant that reacts to the charge-transfer process in each case, and the shift can be compared to shifts in solvents of varying dielectric constant reported by Kober et aL2, In this paper, the solution shifts are analysed in terms of a dielectric continuum model and are found to favour a localized description, though these ideas have been ~0ntested.l~ The MCD results in this and otherll' l2 papers would counter the localized interpretation.Another band near 28000 cm-l does not have such a systematic behaviour and represents a state with quite a strong B term. Its MCD sign and magnitude and its shift in the different environments do not support its easy assignment as a vibrational satellite of the other transitions. It may be a ligand centred transition and is labelled A, on the basis of its not having a distinguishable A term. It may of course be an E state with a small magnetic moment but a large B term ; our results cannot distinguish between these possibilities without a firm theoretical model.Weak, fairly diffuse features are seen in the triplet region around 19 000 cm-l ; a clear positive A term of similar aJD, is seen in the single-crystal MCD spectra reported.12 Triplet state A term features are seen in solid solutions, especially EM, but MCD features seen in other environments are consistent with a reduction of the signal due to increased linewidths in the (inhomogeneously broadened) solid solution environments. R4M and R5M Fig. 2 shows the absorption and MCD spectra of R4M in Nafion B. As was the case for RBY, almost identical results are seen for Nafion A and the spectrum is notably in good accord with that observed in PVA, though the published PVA spectrum12 does not display the region to 30000 cm-l.The main absorption features are shifted by ca. 300 cm-l to higher energy in Nafion as compared to those in PVA. The PVA spectrum was performed at the limits of solubility of the material, but due to the ion-exchange character of the Nafion polymer, no such difficulty arises. The similarity between the spectra confirms the validity of the PVA-MCD result, which could perhaps be held in question on this account (reduction in MCD due to microcrystal formation). A plastic film (PVA for example) containing small crystallites of the solute formed upon evaporation of the solvent may give a good semblance of the absorption spectrum but completely distort or destroy the MCD spectrum.The remarkable aspect of the MCD in fig. 2 is that although the linewidths of the features are no greater than those in RBY (they are in fact less), no significant A term MCD signals are seen, and if present, are at least an order of magnitude less than the 1 bohr magneton magnitude figure indicated t 1 bohr magneton = ca. 9.3 x J T-l.E. Krausz 83 1 wavenumberlcm-’ Fig. 2. Absorption (-) (left-hand scale) and MCD spectra (---) (right-hand scale taken at 5 K and 5 T applied field for (R~(bpy),[4,4’-(CO,Et),bpy]}~+ RAM in Nafion B (see text). Fig. 1-3 have absorption and MCD spectra on the same relative scales. above for RBY-MCD spectra. The absorption feature near 28 000 cm-l (A,) has, however, a strong positive B term of very similar BID magnitude to that seen almost in the same position in RBY.The marked absence of any A terms in the main absorption region is taken as an indication of the quenching of any electronic degeneracy in R4M, and this in turn allows the possibility of localization of charge transfer excitation23 in this material, which is in agreement with previous conclusions.’2 The absorption and MCD spectrum of R5M in Nafion B is shown in fig. 3. It was not possible to get useful spectra in EM or PVA owing to the very limited solubility of this material (in these media) with any common counter-ion. The absorption spectrum is in good accord with solution spectra. Owing to the increased electrophilic character of the 5,5’-substituted bpy ligand in RSM, the lowest energy excitation in this system is expected to be quite strongly localized on the substituted bipyridine, and the lowest- energy band could be assigned to such a transition.The MCD is entirely consistent with an assignment of a metal to (single) ligand charge-transfer band to the 18 500 cm-’ band, as it indeed shows a very weak MCD signal with no evidence for the characteristic A term of the lowest-energy spin-allowed excitation in RBY (E,). Some weak MCD features are shown 22000-25 000 cm-l region. This region seems to be assignable easily, on the basis of its characteristic profile and position to higher energy, to a charge-transfer excitation to the (two) unsubstituted bpy ligands. The energy of such excitations would be expected to move to higher energy by the influence of the $5’-substituted ligand, and this is observed.We note, however, that excitation of the bpy groups does not now give rise to the A term structure seen in the parent RBY. Such structure would not in fact be expected due to the lower symmetry in R5M, removing the degeneracy of E,,, levels necessary for such an MCD feature. Removal of E state degeneracies would usually give rise to complex B terms, and this in fact seems to be the case. The strongest MCD feature is, as for R4M, a positive B term, again close to 28000 cm-’. A BID term of similar magnitude has been seen in Fe(bpy)i+ near 25300 cm-l but in the tris-4,4’- and 53’- complexes near 27000 and 28000 crn-l, respectively. l1832 Absorption, MCD and MCPL Studies of Ru(bpy)i+ I A 0 I T 4 0 8 2 -I 30000 25000 20000 I5000 Fig.3. Absorption (-) (left-hand scale) and MCD spectra (---) (right-hand scale) taken at 5 K and 5 T applied field for (Ru(bpy),[5,5’-(C02Et),bpyl)2’ R5M in Nafion B (see text). wavenum ber/cm-’ Luminescence and MCPL RBY The luminescence of RBY in PVA and EM has been studied in detail in a previous publication15 and has been shown to have parallel behaviour in the two environments, although signals in EM seem characteristically, but unaccountably, weaker. The PVA- based spectra are basically understandable as being due to luminescence from two E states separated by ca. 50 cm-l in thermal equilibrium (above 10 K), with the lower E having a larger magnetic moment than the upper by a factor of 2-4. This analysis relied on temperature-dependent lifetime and quantum efficiency data presented for RBY in PMMA by Harrigan and CrosbyZ4 over ten years ago.The temperature-dependent MCPL for RBY in PMMA was studied in this work and was found to be not significantly different to that in PVA. This validates one component of the previous analysis,15 in that spectra of RBY in PVA and PMMA seem essentially the same. Fig. 4 presents temperature dependent luminescence and MCPL data taken for RBY in both Nafion A and B. One immediately notes that the luminescence profile (as well as position) is quite significantly different to that observed in EM, PVA and PMMA. The second peak becomes more intense than the first and a red shift (relative to those in fig. 4) is seen, though the absorption is slightly blue-shifted (relative to PVA). This red shift and profile change is reminiscent of the change in RBY luminescence in passing from rigid to fluid environments where an environmentally induced localization is considered to take place.2 Fig.5 presents luminescence spectra, scaled to constant quanta over the detected range, and appropriately scaled MCPL spectra, at 22 K for the full range of environments studied. It also presents the MCPL ratio AZ/Z, this quantity being related to c / D in the MCD The isoquantal representation of the luminescence spectra most clearly shows the changes in profile, together with corresponding changes in magnitude of the MCPL. Although the luminescence profile varies markedly, the MCPL profile, though shifted, is virtually unchanged, peakingE. Krausz 833 0 0 18000 16000 14000 12000 waven urn be r/ cm - ’ Fig.4. Temperature dependence of the total luminescence intensity I = Il +I, (lower sections) and the MCPL A I = Il -I, (upper sections) of RBY introduced into ca. 50 ,urn-thick membranes of Nafion A (lower set) and Nafion B (upper set). The applied magnetic field was 5 T and the MCPL was linear in field up to this level. Temperatures (1-5) are 10, 20, 40, 80 and 140 K. before the main luminescence peak with a second peak much less intense than the first. The relevant positions and calculated integrals are presented in table 1. R4M and R5M Fig. 6 shows the luminescence and MCPL over the same range of temperature used in fig. 4, for R4M and R5M in Nafion B. Similar experiments were not performed in Nafion A, though they could reasonably be expected to be rather similar.Fig. 7 shows the luminescence, MCPL and A I / I ratio for the two materials at 20 K, and compares them with those observed in EM at the same temperature. Some relevant data calculated from these results are presented in table 2. Although the luminescence maxima for R4M in EM and Nafion B are similar in position, with only a modest increase in width for the Nafion environment, the spectra for R5M are notably different, with the maximum appearing ca. 1000 cm-l lower in energy. To this extent it appears to be appropriate; the luminescence position and shape are remarkably similar to those observed in a partially softened EM glass, at ca. 135 K. There is also a very significant high-energy tail, apparent in the softened glass spectrum, apparent to a less extent in EM, which has, in both cases, vanishing MCPL signals. In EM the 20 K MCPL ratio [fig.9 of ref. (1 3)] is flattest for R5M ; all other luminophores studied,l3~ 14, 26 in varying environments, show a very characteristic increase of the relative MCPL signal A I / I at the high-energy edge of the luminescence (see fig. 5). 2R FAR 1834 Absorption, MCD and MCPL Studies of Ru(bpy)i+ ’ t 1 0 ” 18000 16000 14000 12000 wavenumber/cm-’ Fig. 5. Solid curves represent luminescence (lower) and MCPL (upper) spectra at 22 K in an applied field of 5 T, of Ru(bpy)i+ in a range of environments. The total luminescence spectra have been normalized to constant quanta over the detecting energy range. The top panel also displays ( - - .) the calculate ratio A I / I (right-hand scale).Environments are x , EM; +, PVA; *, PMMA; 0, Nafion A; 0, Nafion B. Table 1. MCPL and luminescence (lum) data at 22f 1.5 K for Ru(bpy)T environment EM PVA PMMA Nafion A Nafion B error 17 300 15 900 1400 17 400 16 000 1400 - 100 - 100 0.027 0.028 (maximum/shift)/cm- 17 300 17 300 16 000 16 000 1300 1300 17 500 17 500 16 100 16 100 1 400 1400 - 200 - 200 - 100 - 100 MCPL magnitudes 0.048 0.035 0.050 0.038 1 17 450 16 150 1 300 17 650 16 250 1350 - 200 - 100 0.021 0.025 17 350 16 000 1350 17 500 16 150 1350 - 150 - 150 0.030 0.034 50 50 70 40 40 60 80 80 0.002 0.0005 a Gap is shift between MCPL and luminescence.E. Krausz 835 18300 16000 14000 12000 wavenumber/cm-' Fig. 6. Temperature dependence of the total luminescence and MCPL spectra (laid out as in fig.4) for { R~(bpy),[4,4'-(CO,Et),bpy]}~+ (lower) and {Ru(bpy),[5,5'-(CO,Et)bpy]}~+ (upper) at concentrations with peak visible absorbances close to 0.5 units, in ca. 50 pmol dm-3 Nafion B membranes (see text). Temperatures (1-5) are 10, 20, 40, 80 and 160 K. Table 2. MCPL and luminescence data at 20 K for [Ru(bpy),LI2+ environment EM Nafion B material : R4M R5M R4M R5M error (maximum/shift)/cm-l - lum 15 800 15 150 15 600 14 350 100 MCPL 16 150 15 200 16 100 14 850 100 gap - 350 - 350 - 500 - 500 100 MCPL magnitudes (WI) - - 0.037 0.035 QJQ, 0.0 15" 0.0 18" 0.035 0.041 " Data from ref. (13). 28-2836 Absorption, MCD and MCPL Studies of Ru(bpy)i+ 2 4 0 c\I d I 0 2 h a % .- - 2 $ 1 v 4 C I8 I I I 1 I I t.. 00 16000 14000 I2000 w avenum ber/cm-' 0.10 4 .d " 0.05 0 Fig. 7. Solid curves represent the total luminescence I (lower) and MCPL AI (upper) of 0, R4M, and A, R5M, at 20 K and 5 T applied magnetic field in Nafion. The dashed lines are the total luminescence spectra for each compound in 4: 1 ethanol-methanol. The dotted curves are the calculated ratios A I / I (right-hand scale) for the Nafion B environment. Total luminescence spectra are again normalized to constant quanta over the range used. Analysis and Conclusions As previously noted, the question of delocalization of RBY in crystalline environments seems beyond doubt. The observation of temperature dependent MCPL of RBY doped in Zn(bpy),(BF,), single crystals taken in the 10-50 K range2' can only be simply understood if the emitting state(s) maintains an electronic degeneracy to the level of ca.kT and has a substantial magnetic moment. The RBY system, being an even electron ti configuration in the ground state and ti-n* in the (lowest) MLCT states, has no (symmetry determined) degenerate states in any symmetry less than D,. For RBY in glasses, where anomalous excitation polarization ratios ( > 1/7) are seen, l9 similar temperature-dependent MCPL spectra are observed and this immediately implies that the low-symmetry potentials present in these inhomogeneous environments only split the degeneracy of the emitting state by less than kT (ca. 10 cm-l). The emitting state(s) then must gain their magnetic moment largely from spin angular momentum, making it relatively inert to crystalline potential, and giving the increased radiative lifetimes seen below 50 K.The spectroscopy at temperatures lower that 10 K becomes much more complex, owing to non-thermalization, a strong red shift and marked field-induced changes in the profile and linear polarization of the luminescence. 27-29 We do not consider these phenomena here, although again some type of dynamic localization process has been suggested.E. Krausz 837 The MCPL for RBY in the (hexagonal) Zn(bpy), (BF,), crystal host is larger, relative to the luminescence, than for solid solutions at a given temperature and applied magnetic field.14 This confirms that the molecular and crystal c axis coincide in this host. Zn(bpy),(PF,), also crystallizes as hexagonal needles, but crystallographic determina- tions and spectroscopic data have shown' an entirely different structure where the (approximate) molecular axes lie almost perpendicular to the c axis.The MCPL in the latter host is reported to be vanishingly weak,25*28 and this is quite consistent with the fact that g, for an even electron system is identically zero by Thus an Estate would not split and give temperature-dependent MCPL for this orientation of the applied field. A more detailed assignment and analysis of the MCD of RBY is not possible as we do not currently have a model of the electronic structure that can account for an (spin- allowed) E, - A - Eb pattern with approximate intensity ratio 1 : 2 : 1 and the lowest state (E,) carrying CD. Current models fail to account for the observed intensities and polarizations.l2, 30-33 The excited-state magnetic moments in RBY are entirely consistent with those arising from a hole in a t: core of the metal, and these are not expected to be very strongly quenched by low-symmetry potentials." The widths of the absorption features and their varying linewidths preclude a satisfactory moment analysis,''* 25 but the MCD of RBY in single crystal and solution environments are entirely analagous, though it is known that some distortions occur in solution, from excitation polarization measurements mentioned above. However, the obvious and substantial reduction in MCD in R4M and R5M is compelling evidence for a strong, low-symmetry potential in the (prompt and environmentally unrelaxed) excited state of these complexes.This is in turn consistent with the idea of strong localization of the excitation onto the single substituted ligand in each case. The MCPL spectra, in every case, were found to be substantially temperature dependent. The difficulty in quantifying this is that the MCPL and luminescence do not have the same profiles and several luminescent states are known to be involved. Furthermore the critical AI/Z ratio does not have the same shape at different temperatures, as well as falling with increasing temperatures. Generally, the discrepancy between MCPL and luminescence profiles becomes greater at lower temperatures. An average of the ratio AZ/I at a particular temperature can be calculated, but this is rather sensitive to those weaker regions that have a strong ratio and becomes particularly prone to (zero) errors in the equipment. A more convenient measurement is the ratio of the integrals of the (corrected) MCPL and luminescence defined as Q,/Q, = J AI dv/J I dv.This, of course, becomes identical to the average of ( A I / I ) when the profiles of I and AI are the same. Tables 1 and 2 show that the average ratio and Q,/Q, do not, surprisingly, differ greatly near 20 K. Fig. 8 shows the temperature dependence of Q,/Q, for the systems studied in detail, and they are quite similar to each other and those in fig. 11 of ref. (1 3 ) and fig. 1 1 of ref. (14), except for an overall reduction of magnitude compared to RBY in PVA. The MCPL ratio from a simple, isolated. E level must give rise to a linear 1/T plot, passing through the origin, analogously to a C / D term in MCD.25 Most surprising in this work is the very substantial MCPL (but not MCD) of R4M and R5M.In the previous work in EM, RBY Q,/Q, was larger than for R4M which again was substantially larger than for R5M. In the current work in Nafion B, Q,/Q, is practically the same for the three materials, over the entire temperature range. This is irrespective of the fact that the AZ/Z (at 20 K and at other temperatures) is quite different in shape (fig. 4 and 7 ) . A rationalization of these phenomena can be made by noting the Q,/Q, values for838 Absorption, MCD and MCPL Studies of Ru(bpy)i+ 6 h - M E . 9 4 0 2 2 0 10’ KIT Fig. 8. Experimental values of QJQ, (see text) plotted against reciprocal absolute temperature for RBY in *, Nafion A; + , Nafion B; A, R4M and 0, R5M in Nafion B.RBY in the various environments in table 1. The highest value (0.05) occurs for PVA followed by PMMA, Nafion B, then Nafion A at 0.23. Neglecting the EM value as anomalously low, which may be due to local strain effects, the Q,/Q, value can then be related to the necessity for the complex to ion pair with the charge compensators (counter-ions), which then creates a distortion in the environment and a strong tendency to localize the excitation. In PMMA the dominantly hydrocarbon nature of the polymer would induce strong ion pairing. In Nafion the fluorocarbon backbone could also be expected to induce strong binding to the sulphonate anions. Thus in Nafion, from the MCPL data, as well as the profile and Stokes shift of the luminescence, RBY would have to be considered substantially localized in the luminescent state.The even larger Stokes shift for RBY in Nafion B compared to Nafion A (fig. 4) and the very large shift for R5M in Nafion B indicate the possibility of an (environmental) dipole relaxation process taking place, even at the lowest temperatures. The temperature-dependent luminescence spectra in fig. 6 show that a substantial red shift and reduction in quantum efficiency is seen near 160 K for RBY, R4M and particularly for R5M. to consist of two zones, the hydrophobic fluorocarbon backbone and clusters of sulphonate groups, and water molecules ca. 5 nm in diameter. Mossbauer experiments on ion salts infused in Nafion by observation of the disappearance of the resonant nuclear absorption, that the hydrophilic environment is still ‘liquid’ near 200 K.Water in these 5 nm regions did not freeze at 100 K below its normal freezing point. Our measurements indicate some softening of the environment and dipole relaxation at an even lower temperature, near 160 K. In conclusion, the reduction of the MCPL magnitude and the distinctive luminescence profile in Nafion compared to other plastics indicates a substantial distortion in the luminescent state, and most likely, localization of charge-transfer excitation. MCD data provide evidence for a strong distortion in the (prompt) excited states of R4M and R5M The Nafion environment has beenE. Krausz 839 in Nafion, but not for RBY in the same system which has a very similar MCD to that found in single crystal and glass and similar spectra in other plastics.The magnetic moments of the luminescent states in these systems seem not to be quenched as strongly as those states seen in absorption. This is consistent with a basic (spin-only) magnetic moment and the long radiative lifetimes seen in luminescence spectra. The MCPL magnitude drops by a factor of ca. 2 upon localization in RBY, consistent with previous studies on softening in EM glasses.2*3 It now seems reasonable to search for time-dependent luminescence spectra in Nafion membranes at quite low temperatures. Such studies may allow control over the photochemical activity and degradation of the RBY system by varying the degree and dynamics of charge localization and the inhibition of a reverse electron transfer from the photochemical product by a dynamic change in the activator complex.I thank Dr Fritz Herren for the samples of R4M and R5M and Dr Albert Mau for the improvements in the preparation procedures of Nafion membranes. References 1 Reviews of recent work are given in T. J. Meyer, in Progress in Inorganic Chemistry, ed. S . Lippard (Wiley, New York, 1984), vol. 30, p. 389; K. Kalyanasundaram, Coord. Chem. Rev., 1982, 46, 159; J. Ferguson, F. Herren, E. R. Krausz and J. Vrbancich, Coord. Chem. Rev., 1985, 64, 21. 2 J. Ferguson and E. Krausz, Chem. Phys. Lett., 1986, 127, 551. 3 J. Ferguson, E. R. Krausz and M. Maeder, J. Phys. Chem., 1985, 89, 1852. 4 E. Krausz, Chem. Phys. Lett., 1985, 116, 501. 5 J. Ferguson and E.Krausz, Inorg. Chem., 1986, 25, 3333. 6 J. Ferguson and E. Krausz, Znorg. Chem., 1987, 26, 1383. 7 L. A. Phillips, W. T. Brown, S. P. Webb, S. W. Yeh and J. H. Clark, ZCCC Hawaii Conference Abstract, December 1984. 8 R. F. Dallinger and W. H. Woodruff, J. Am. Chem. SOC., 1979, 101, 4391; P. G. Bradley, N. Kress, B. A. Hornberger, R. F. Dallinger and W. H. Woodruff, J. Am. Chem. SOC., 1981, 103, 7441. 9 G. A. Heath, L. J. Yellowlees and P. S. Braterman, Chem. Phys. Lett., 1982,92, 656; P. S. Braterman, A. Harriman, G. A. Heath and L. J. Yellowlees, J. Chem. SOC. Dalton Trans., 1983, 1801. 10 A. Hauser and E. Krausz, Chem. Phys. Lett., 1987, 138, 355. 11 A. J. Thompson, V. Skarda, M. J. Cook and D. J. Robbins, J. Chem. SOC. Dalton Trans., 1985, 12 J. Ferguson, E. Krausz and J. Vrbancich, Chem. Phys. Lett., 1986, 131, 463. 13 J. Ferguson and E. Krausz, J. Phys. Chem., 1987, 91, 3161. 14 J. Ferguson and E. Krausz, J. Lumin., 1986, 36, 129. 15 E. R. Krausz and A. Ludi, Znorg. Chem., 1985, 24, 939. 16 N. Oyama and F. C. Anson, J. Electrochem. SOC., 1980, 127, 247. J. A. Bruce and M. S. Wrighton, 17 E. Krausz and A. W. H. Mau, Znorg. Chem., 1986, 25, 1484. 18 E. Krausz, Chem. Phys. Lett., 1985, 120, 113. 19 I. Fujita and H. Kobayashi, Znorg. Chem., 1973, 12, 2758. 20 C. M. Carlin and M. K. DeArmond, Chem. Phys. Lett., 1982, 89, 297; C. M. Carlin and M. K. 21 P. S. Braterman, G. A. Heath and L. J. Yellowlees, J. Chem. SOC. Dalton Trans., 1985, 1081. 22 E. M. Kober, B. P. Sullivan and T. J. Meyer, Znorg. Chem., 1984, 23, 2098. 23 J. Ferguson, A. W-H. Mau and W. Sasse, Chem. Phys. Lett., 1979, 68, 21. 24 R. W. Harrigan and G. A. Crosby, J. Chem. Phys., 1973, 59, 3468. 25 S. B. Piepho and P. N. Schatz, Group Theory in Spectroscopy, with Applications to Magnetic Circular 26 J. Ferguson and E. Krausz, unpublished work. 27 J. Ferguson and E. Krausz, Chem. Phys. Lett., 1982, 93, 21. 28 R. W. Harrigan, G. D. Hager and G. A. Crosby, Chem. Phys. Lett., 1973, 21, 487. D. C. Baker and 29 J. Ferguson and E. Krausz, Chem. Phys., 1987, 112, 271. 30 J. Ferguson and F. Herren, Chem. Phys., 1983, 76, 45. 31 E. M. Kober and T. J. Meyer, Znorg. Chem., 1982, 21, 3967. 1781. J . Am. Chem. SOC., 1982, 104, 72; DuPont Co. (Wilmington Delaware, U.S.A.). DeArmond. J . Am. Chem. SOC., 1985, 107, 53. Dichroism (Wiley-Interscience, New York, 1983). G. A. Crosby, Chem. Phys., 1974, 4,428.840 Absorption, MCD and MCPL Studies of Ru(bpy):' 32 E. M. Kober and T. J. Meyer, Inorg. Chem., 1984, 23, 3877. 33 A. Ceulemans and L. G. Vanquickenbourne, J. Am. Chem. Soc., 1981, 103,2238. 34 Perfluorinated Ionomer Membranes. (ACS Symposium series No. 180, Washington, 1982), ed. 35 Ref. (34), p. 171. A. Eisenberg and H. L. Yeager. Paper 7/509 ; Received 20th March, 1987

ISSN:0300-9599    

DOI:10.1039/F19888400827

出版商:RSC    

年代:1988

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I, 1988, 84(3), 841-850 Complexation of Polymer-bound Imino Diacetate-type Chelating Agents with some Transition-metal Ions Effect of Charged Polymer Chains on Chelate Formation Reactions Yoshimi Kurimura* and Kiyomi Takato Department of Chemistry, Ibaraki University, Bunkyo, Mito, Ibaraki 31 0, Japan Acid dissociation constants (K,) of the polymer-bound imino diacetate analogue, 4-vinylbenzylamine-N, N'-diacetate co-styrenesulphonate (P-SS) and 4-vinylbenzylamine-N, N'-diacetate co-N-vinylpyrrolidone (P-VPRo), and the formation constants of (P-SS )-M2+(M = Co, Ni, Cu) chelates have been determined by means of potentiometric titration and spectropho- tometry, respectively. The values of K . , and K,, of P-SS and P-VPRo are smaller than those of the corresponding low molecular weight model compound, benzylamine-N, N'-diacetate (BDA), and the formation con- stants of the (P-SS)-M2+ chelates are larger than those of the corresponding BDA-M2+ chelates by a factor of 1&102, depending on the metal ions.The results obtained in the polymer system are largely explained in terms of a powerful microheterogeneous field effect of the anionic polyelectrolyte chain. Reaction rates involving ionic species are known to be varied by the presence of polyelectrolytes owing to the existence of large electrostatic potentials in the polymer dornains.lp5 In a polyelectrolyte solution, increases in the local distributions of the oppositely charged low molecular weight ionic species in the domain are usually a consequence of electrostatic attraction.l* 2 * polyacids," and other polymer-bound ligands such as polyvinylpyridinel2> l3 and polyvinylimidazolel* have been carried out and the results discussed in terms of macromolecular chain effects.Recently, we prepared polymer-bound benzylamine-N, N'-diacetate, and the tris- (bipyridine)ruthenium(rI)-photosensitized reduction of CoI'I-Schiff base complex has been investigated in this polymer s01ution.l~ It was found that the efficiency of the charge separation, i.e. formation of Co'' Schiff-base complex, in the polymer- bound chelating agent solution is greatly affected by (i) the degree of adsorption of the reagent ions, Ru(bpy):+, and the Co"' complex ions and (ii) the pattern of the local distributions of the reagent ions in the polymer network.A large number of the formation constants of aminopolycarboxylate with divalent transition-metal ions have been determined by means of potentiometric titration and spectrophotometry. However, few studies on the complex formation between polymer- bound aminopolycarboxylate and metal ions have been reported. It is of interest to broaden the investigation of the chelate-formation reactions to those of polymer-bound chelating agent-metal ions systems and to discuss the effect of polymer chains on the complex formation reactions. Studies of the complexation of metal ions with 84 1842 Complexation of Polymer-bound Chelating Agents with Transition-metal Ions Experiment a1 Materials The metal ion solutions, prepared from a reagent grade nitrates of Co", Ni" and Cu", were standardized on an atomic absorption spectrometer (Hitachi model 200-10). N-Vinylpyrrolidone (VPRo) and 4-vin ylpyridine (VP) were distilled under reduced pressure and sodium p-styrenesulphonate (NaSS) was recrystallized from ethanol.Vinylbenzylamine-N, N'-diacetic acid (VBDA) was prepared according to a literature method. l6 Polyvinylpyridine, degree of polymerization 98, was a gift from Professor E. Tsuchida (Waseda University). Partially quaternized poly-4-vinylpyridine (QPVP) was prepared by reaction of the PVP with ethyl bromide in ethanol at 60 "C for 2-6 h. Polymerizations VP-co-NaSS copolymer was prepared by radical polymerization of VP with NaSS in dimethyl sulphoxide at 60 "C for 1 h. VBDA-co-VPRo (P-WRo) was prepared by the procedures reported in the previous paper.15 The molecular weights of the copolymers prepared were estimated to be greater than several thousands since the reaction mixtures were dialysed using a cellulose dialysis tubing (Nakarai Chemical Co., mol.wt cut-off 8000) for several days. Chemical compositions of P-SS and P-VPRo were determined by elemental analysis of C, H, N and Na and, in addition, by the mole ratio method in the case of P-SS (using P-SS-Cu2+ solution). The amount of chelating agent contained in the polymers obtained by elemental analysis agreed with that obtained spectrophotometrically to within f 3 O/O. Acid Dissociation and Chelate Formation Constants All titrations were carried out under an argon atmosphere in order to remove carbon dioxide effect.Hydrogen ion concentrations of the solutions were adjusted using dilute hydrochloric acid and/or aqueous sodium hydroxide solution to desired values. For determination of the chelate formation constants, solutions were made up which contained equimolar amounts of the metal ions and the chelating agent. The points of incipient precipitation were above pH 4.5 for all the metal ions used: no data obtained above pH 4.5 were used in the calculations of the stability constants. The solutions were brought to equilibrium in a bath thermostatted at 25 "C before absorption measurements were made. All the measurements were carried out at Z = 0.1 mol dm-, (NaNO,) and 25k0.1 "C. Absorption measurements were made in a thermostatted quartz cell (10 mmx 10mm) with a Hitachi model 320 spectrophotometer.A Hitachi F-5 pH meter was used to measure pH values. Calibration of the pH meter was carried out using phosphate and acetate buffers, pH 4.60 and 6.81, respectively, at 25 "C. Hydrogen ion concentrations were calculated from the activity of hydrogen ion assuming that the value of the activity coefficient was 0.781, i.e. that of 0.1 mol dm-3 NaN0,.17 Calculations The experimental results of the titrations can be expressed by the modified Henderson-Hasselbach equations :la (1) (2) pH = pK,, - n, log (1 - a)/a pH = pKa2 - n, log (2 - a)/(a - 1) where a is the degree of neutralization and n, and n2 are the constants, and Kal and Ka2Y. Kurimura and K. Takato 843 are the first and second acid dissociation constants, respectively.The values of n, and n,, which vary in value somewhat above 1 to about 2, are a measure of the extent of the effect of the neighbouring groups on dissociation of the first and second acid groups in the chelating agent moieties.ll The values of pK,, and n, were obtained from the intercept and slope of the straight line of pH us. log (1 - a)/a plot, respectively. The values of pK2 and n2 were similarly determined from pH us. log (2-a)/(a- 1) plots. It is reasonable to presume that only 1 : 1 chelates are formed under the conditions employed. In this case the formation constant is expressed by [MA1 [M2+] [A2-] K = (3) where [MA], [M2+] and [A2-] represent the concentrations of metal chelate, metal ion, and deprotonated form of benzylamine-N, N'-diacetate moiety, respectively.The absorbance of the reaction solution around the absorption maxima of the d-d transitions in terms of its components is expressed by A = &,[M2+] +&,[MA] (4) [M2+] = [MIT - [MA] [A], = [A], - [MA1 where [MI,, [A],, and [A], are the total concentrations of M2+, uncoordinated chelating agent, and chelating agent respectively, and E, and E , are the molar absorptivities of the metal ion and metal chelate, respectively, at a given wavelength. For all the metal ions used, the curves for plots of A/[M], us. pH show flat regions in the higher pH ranges, i.e. pH 6.5-7.9 for Co2+, 4.5-6.5 for Ni2+ and 3.5-6.0 for Cu2+, under the conditions employed, suggesting that the predominant species in these pH regions are 1 : 1 chelates. Then, the values of ~~(m01-l dm3) for (P-SS)-M2+ were estimated to be 9.8k0.2 at 510 nm for CoA, 4.1 f0.02 at 630 nm for NiA, and 72f 1 at 720 nm for CuA from the values of the absorbances at the flat regions.The values of e,(mol-l dm3) used for the calculations were 4.17 at 510 nm (Co), 1.50 at 630 nm (Ni), and 8.20 at 720 nm (Cu). The concentrations of MA could be determined photometrically at various pH values using eqn (7) -&1IM1T [MA] = E2 - E l (7) where D is the absorbance at this wavelength. The concentration of A2- can be expressed by [A2-], [H+] [H+I2 * [A2-] = 1L-I- Substituting eqn (5), (6) and (8) into eqn (3), an expression for the formation constant becomes Eqn (9) will then take the form844 Complexation of Polymer-bound Chelating Agents with Transition-metal Ions Table 1.Copolymerizations of VBDA with NaSS or VPRo" mole fractions mole fractions in feed in copolymers found (YO) (calc.) (YO) polymer VBDA NaSS(VPRo) mb nb C H N Na P-ss- 1 0.250 0.750 0.118 0.882 42.28 (41.23) P-ss-2 0.500 0.500 0.493 0.507 48.82 (49.03) P-VPRO 0.091 0.909 0.093 0.907 60.79 (60.95) P-VPRO-2 0.250 0.750 0.246 0.754 56.48 (56.61) 4.54 (4.76) 4.69 (4.90) 7.92 (7.71) 7.37 (7.20) 0.64 (0.64) 2.6 1 (2.69) 10.63 (10.69) 8.50 (8.55) 9.09 (9.20) 8.90 (8.96) 1.66 (1.63) 3.55 (3.45) a Calc. for (C13Hl,04NNa -2H20)o~ll,(C,H7S03Na 2H20)0~,,2 for P-SS-1, (C13H 0 "a - p-VPR,-1, and (C!,H1504 NNa - 3H20)o~246(C6HgON)o~754 for P-VPRo-2. The values of rn and n are the mole fractions of VBDA and NaSS, respectively. H20)o,4g3(C,H7S03Na. H2°)0,507 for p-ss-2, (C,-3H1504NNa * 3H20)0.093 (c6H90Ni" O.:O7 for Eqn (10) can be applied to Co2+, Ni2+ and Cu2+ to determine the values of the chelate formation constants: log K can be obtained from the intercept of the plot of log (l + [H+I/Kal + [H+12/Ka1 &2) us' log (iM1T- Results and Discussion In the present study, two kinds of polymeric chelating agents, P-SS and P-VPRo, were synthesized : the chemical compositions and structures of the polymer-bound chelating agents are presented in table 1 and fig.1, respectively. All the polymers presented in table 1 are soluble in water. Concentrations of the polymer chelating agents are expressed as monomeric units of vinylbenzylamine-N, N'-diacetate unit by mol dm-3. A typical titration curve for one of the chelating agents is shown in fig.2; titration curves for other polymeric chelating agents were similar to that shown. All the titration curves have two distinct inflection points corresponding to a dibasic acid, showing values of the first and second acid dissociation constants which can be calculated separately from the titration curves. Modified Henderson-Hasselbach plots, i.e. pH us. log (I - a)/a (a = degree of neutralization) for the first acid dissociation step and pH us. log (2 -&)/(a - 1) for the second dissociation one for P-SS and P-VPRo, are as shown in fig. 3 and 4, respectively. For all the polymeric chelating agents, plots of pH us. log(1 -a)/a show a slight curvature at higher values of a, but this does not impede determination of intercepts since such curvatures appear in the region of log (1 -a)/a > 0.The estimated values of pKa,, pKa,, n, and n, are summarized in table 2 along with pKa, and pKa, of the low molecular weight analogue. Table 2 shows that the values of pKa, and pKa, of the polymeric chelating agents are greater than those of the corresponding pKa values of the low molecular weight analogue by ca. 1.2-1.5. Futhermore, the values of pKa, for P-SS are larger than those of P-VPRo, reflecting the larger negative charge density in the domain of P-SS than that of P-VPRo. The results seem to indicate that high negative charge density in the polymer network of P-SS supresses the dissociation of protons from the acid moiety on the polymer chain. Thus, an increasing order of pKa becomes BDA < P-VPRo < P-SS. No noticeable differences in the pKa,(pKa,) value of P-SS-1 and P-SS-2 were observed or between those of P-VPRo and P-VPRo-2.Y.Kurimura and K . Takato 845 SO3Na P-VPRO: R = Do Fig. 1. Chemical structure of P-SS and P-VPRo. 0 2 4 6 8 10 NaOH/cm3 Fig. 2. Titration of 2.00 x mol dmP3 P-SS-1 with 0.102 mol dmP3 NaOH at I = 0.1 and 25 "C. The dependence of pKa on the charge distribution on the polymer backbone was also examined by the use of more simple copolymers such as VP-co-NaSS and VP-co-VPRo (VP = N-vinylpyridine). Fig. 5 shows clearly that an increase in the degree of the negative charge density due to the sulphonate groups on the polymer chain causes an increase in the pKa of the pyridine ring, whereas the pK, value decreases with an increasing degree of quarternization, indicating that the proton dissociation is suppressed by the increase in the positive charge density on the polymer.846 Complexation of Polymer-bound Chelating Agents with Transition-metal Ions 5 4 5: a 3 '.0 A I I - 1 0 log ( 1 -ac)/ac 1 Fig. 3. Plots of pH us. log (1 -a)/a. (a) P-SS-1, (b) P-SS-2, (c) P-VPRo-1, ( d ) P-VPRo-2. 1 1 10 E 9 I I - 1 0 1 log (2 --a)/(a - 1 ) Fig. 4. Plots of pH us. log (2-a)/(a- 1). (a) P-SS-1, (b) P-SS-2, (c) P-WRO-1, ( d ) P-VPRo-2.Y. Kurimura and K. Takato 847 Table 2. Values of pKa,, pKa2, n, and n2 for P-SS, P-VPRo and BDA at I = 0.1 and 25 OCa chelating agent PK,, n1 ~Ka2 n2 P-ss- 1 3.32 1.1 10.27 1.1 P-ss-2 3.38 1.1 10.06 1.2 P-VPRo-1 2.94 1.2 9.97 1.3 P-VPRO-2 2.78 1.2 9.52 1.2 BDA 2.24 - 8.90 - a The errors of the pK and n values are within & 7 and & 5 YO, respectively 0 40 8 0 mole fraction of VBDA (0); degree of quarternization (0) Fig.5. Dependence of pKa on the mole fraction of vinylpyridine (VP) in VP-co-NaSS (0) and on the degree of quaternization (0) ; I = 0.1 and 25 "C ; A, pyridine (0 % quarternization). For P-SS and P-VPRo, the values of n, and n, are in the range 1.1-1.3, indicating relatively small neighbouring group effects in the polymer-bound chelating agent solution. The formation constants of P-SS, P-VPRo and BDA with the transition metal ions were determined spectrophotometrically by the procedure described in the experimental section. Absorption spectra of (P-SS)-Co2+ are shown in fig. 6. A typical example of a D/[M], us. pH plot ( D = absorbance) is shown in fig. 7.From the data obtained from this plot, the concentrations of MA at various pH values were determined. Values of log (1 + [H+]/K,, K,,) are plotted as a function of -log ([MIT - [MA])([MA], - [MA]) (fig. 8). As shown in this figure, all the plots are linear with slopes of unity. The values of the formation constants are summarized in table 3. The value of K for the848 Complexation of Polymer-bound Chelating Agents with Transition-metal Ions 0.10 I 4 00 500 6 00 7 00 h/nm Fig. 6. Absorption spectra of P-SS-Co2+ solutions containing 1.00 x 1 .OO x mol dm-3 P-SS-1 and mol dm-3 Co2+ at I = 0.1 and 25 "C; curves in ascending pH values: 3.65, 3.90, 4.61, 4.91, 5.44, 6.53 and 7.24. 2 i 2 3 5 7 PH Fig. 7. Plots of A/[Co2+IT vs.pH for (P-SS-1)-Co2+ and BDA-Co2+ systems at 720 nm. (a) (P-SS- 1)-Co2+, (b) BDA-Co2+. BDA-M2+ complex obtained from the present experiments are in satisfactory agreement with those obtained by And0.l' The results show an enhancement in chelate formation in the polymer system ovek that of corresponding low molecular weight systems: K values for all the polymers are 1-2 orders of magnitude greater than those of the corresponding low molecular weight analogues.Y. Kurimura and K. Takato 849 I I I I I I 2 4 6 Fig. 8. Plots of log (1 + [H+])/(K,, + [H+I2/K,, Ka2) us. log[M],[A],/[MA]. (a,) P-SS- 1/Cu2+, (a,) BDA/Cu2+, (b,) P-SS-l/Ni2+, (b,) BDA/Ni2+, (c,) P-SS- 1/Co2+, (c,) BDA/Co2+. Table 3. Chelate formation constants of Co2+, Ni2+ and Cu2+ with P-SS or BDA at I = 0.1 and 25 O C a log K chelating agent CO2' Ni2+ cu2+ P-ss- 1 8.68k0.12 9.74k0.18 11.50&0.12 P-ss-2 8.96 & 0.09 10.07 & 0.16 11.76 & 0.14 BDA 6.59k0.10 8.07 k 0.16 10.70k 0.14 (6.88) (7.98)b (10.51)b a 1 .OO x 1 O-, mol dmP3 M2+ and 1 .OO x 1 O+ mol dmP3 chelating agent.Data from ref. (19). It is well known that polyanions such as polyvinylsulphonate and polystyrene- sulphonate concentrate the oppositely charged ionic species in their Thus the complexation process for the polymer chelate might proceed by two steps: (a) accumulation of M2+ ions into the P-SS domain (pre-equilibrium process) and (b) complexation of M2+ with the chelating agent moieties (chelation process). Higher electrostatic potentials on the polymer-bound chelating agents carrying their negatively charged moieties closer the polymer chain would attract a large number of divalent metal ions, leading to high local distribution of M2+ ions in the domains.For P-SS, larger formation constants compared with those of the corresponding low molecular weight systems would be attributed to the existence of such a pre-equilibrium process.850 Complexation of Polymer-bound Chelating Agents with Transition-metal Ions Enhancement factors (F) defined by F = KJK,, where Kp and K , are the formation constants of (P-SS)-M2+ and BDA-M2+, respectively, were found to be ca. 120 for Co2+, 50 for Ni2+, and 15 for Cu2+. The results seem to suggest that the enhancement factor decreases as the corresponding K , increases. For metal complexes having larger K , values, the degree of electrostatic accumulation of metal ions into the polymer domains would be smaller than those having small K , values since relatively small amounts of the metal ions remain in the bulk of solution as the larger part of the metal ions are already combined with the chelate groups under the conditions employed.This work was partially supported by a Grant-in-aid for Scientific Research on the Priority Area of Macromolecular Complexes from the Ministry of Education, Science and Culture, Japan. References 1 B. Vogel and H. Morawetz, J. .Am. Chem. SOC., 1968,90, 136. 2 H. Morawetz, Acc. Chem. Res., 1970, 3, 354. 3 T. Ohkubo and N. Ise, J. Am. Chem. SOC., 1973,95,2293. 4 N. Ise, J. Polym. Sci., Polym. Symp., 1978, 62, 205. 5 C. Tondre, J. Chem. SOC., Faraday Trans. 1, 1982, 78, 1795. 6 M. Mandel and J. C. Lyte, J. Polym. Sci., 1978, A2, 2883. 7 T. Pecht, A. Levitzki and J. Anber, J. Am. Chem. SOC., 1967, 89, 1578. 8 M. Hatano, T. Nozawa, S. Ikeda and Y. Yamamoto, Macromol. Chem., 1971, 141, 1. 9 L. Toshi and A. Garnier, Biochem. Biophys. Res. Commun., 1974, 58, 427. 10 A. Garnier and L. Toshi, Biopolymers, 1975, 14, 2247. 11 H. P. Gregor, L. B. Luttinger and E. M. Loebl, J. Phys. Chem., 1955, 59, 559; 990. 12 H. Nishikawa and E.' Tsuchida, J. Polym. Chem., 1972, 79, 2072. 13 Y. Kurimura, K. Wakayama, N. Nishikawa, and E. Tsuchida, Mukromol. Chem., 1979, 180, 339. 14 D. H. Gold and H. P. Gregor, J. Phys. Chem., 1960, 64, 1464. 15 Y. Kurimura, K. Takato, M. Takeda and N. Ohtsuka, J. Phys. Chem., 1985,89, 1023. 16 A. Uehara, E. Kyuno and R. Tsuchiya, Bull. Chem. SOC. Jpn, 1970, 43, 1394. 17 H. Homed and W. Hamer, J. Am. Chem. SOC., 1933, 55, 2194. 18 H. Morawetz, Macromolecules in Solution (Wiley, New York, 1965). 19 T. Ando, Bull. Chem. SOC. Jpn, 1952, 35, 1395. Paper 7/55; Received 27th March, 1987

ISSN:0300-9599    

DOI:10.1039/F19888400841

出版商:RSC    

年代:1988

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. 1 , 1988, 84(3), 851-863 Thermal Stabilities of Hexacarbonyl and Subcarbonyls of Molybdenum encapsulated in NaY and NaX Zeolites Y asuaki Okamoto,* Akinori Maezawa, Hiroshige Kane, Isao Mitsushima and Toshinobu Imanaka Department of Chemical Engineering, Faculty of Engineering Science, Osaka University, Toyonaka, Osaka 560, Japan Decompositions of Mo(CO), and the resulting subcarbonyl species adsorbed on NaY and NaX zeolites have been investigated utilizing i.r., temperature-programmed decomposition and X.P.S. techniques. It was found that Mo(CO), encapsulated in the NaX zeolite was considerably less stable than that contained in the NaY zeolite, while intermediate subcarbonyl species, Mo(CO),,,,, showed a completely reversed thermal stability. The drastic difference in the stabilizing properties between X and Y zeolites is thought to result from the difference in the basic strength of the zeolite framework oxygen.The structure of Mo(CO),,,, is proDosed on the basis of i.r. spectra. The X.P.S. results suggest that the adsorbed Mo(CO), species is decomposed to Mo-metal aggregates at 473 K. Transition-metal carbonyls supported on organic or inorganic matrices show prominent catalytic properties. Various types of interaction modes between metal carbonyls and (surface) functional groups have been proposed and recently reviewed.” Among inorganic materials having high specific surface areas, zeolites provide potential media for producing well defined catalytically active metal species, because of their well characterized stru~tures.~ Low-valent molybdenum catalysts prepared from Mo(CO), encapsulated in zeolites are examples of such systems.Interesting catalytic features of Mo(CO), supported on decationized Y zeolites (HY) have been reported by Yashima et aL4 for polymerization and hydrogenation of ethylene and metathesis of propylene. Their results suggest great potential for a precise control of catalytic reactions by employing zeolites as supporting matrices. Infrared studies of adsorption and decomposition of Mo(CO), on NaY and HY zeolites have been conducted by Gallezot et aL5 and Howe et al.‘?’ to elucidate interaction modes between the carbonyls and the zeolites and to reveal intermediate subcarbonyl species. With NaY zeolite, it is suggested that Mo(CO), is thermally decomposed to molybdenum metal6* ’ in contrast to Mo(CO), encaged in HY where Mo is shown to be oxidized by the reaction with surface hydroxyl groups in the zeolite.The subcarbonyl species Mo(CO),,,, has been suggested to be thermally stable in HY5 and Nay7 zeolites. On the other hand, Mo(CO),,,, has been proposed as a stable intermediate subcarbonyl species on Al,O,s-lo and Zn0.l’ The difference in the stable subcarbonyl species between these matrices seems to be important to the understanding of the chemistry of anchoring metal carbonyls. Thermal properties of metal carbonyls adsorbed on supporting materials provide fundamental knowledge about the interaction modes between the carbonyls and the surface functional groups and about the optimization of catalyst preparation. In the present study, decompositions of Mo(CO), and the resulting subcarbonyls encapsulated in sodium-type zeolites were investigated by utilizing i.r., temperature- 85 1852 Thermal Stabilities of Molybdenum Carbonyl Compounds in Zeolites programmed decomposition, and X.P.S.as a function of zeolite compositional Si/A1 ratio (Nay and NaX). It was found that decomposition temperatures of the carbonyl species depended strongly on the zeolite composition. Experimental Two kinds of NaY zeolites of different composition (Si/Al = 2.43: SK-40, Nikka Seiko Co. and Si/Al = 2.78 : JRC-Z-Y5.6, Catalysis Society of Japan1,) and an NaX zeolite (Si/Al = 1.23: 13X, Gasukuro Kogyo Co.) were utilized in the present study. Hexacarbonylmolybdenum was supplied by Strem Chemicals and was used as received.Powdered zeolites were pressed into self-supporting wafers (thickness, ca. 10 mg ern-,) for i.r. studies. The samples were pretreated in a vacuum (< 1 x Pa) at 673 K for 1-2 h in an in situ i.r. cell. After having recorded a background spectrum, the zeolite wafers were exposed to Mo(CO), vapour (ca. 8 Pal3) at room temperature (ca. 295 K) for a selected period. The i.r. spectra were recorded at room temperature in the transmittance mode on a Hitachi double beam spectrophotometer (EPI-G). The resolution was 2.5 cm-l at 3000cm-l. In t.p.d. (temperature-programmed decomposition) studies of Mo(CO), encaged in zeolites, zeolite samples were used in powder form. After having been evacuated at 673 K for 1-2 h, the zeolites were exposed to Mo(CO), vapour at room temperature for 12 h.The decomposition of Mo(CO), adsorbed on the zeolites was conducted in a dynamic vacuum and the evolved gases were continuously and repeatedly analysed by a mass filter (ULVAC, MSQ-150A) over a mass range of m / z = 1-51 (1 cycle; 130 s). The decomposition temperature was increased at a rate of 1.8 K min-' from room temperature to ca. 600 K. In the present zeolite systems CO (m/z = 12 and 28) was the only major desorption product, accompanying a negligible amount of H, (m/z = 2). No formation of CH, ( m / z = 15) and CO, (m/z = 44) was detected during the decomposition of Mo(CO),-NaY or -NaX under vacuum. The X.p. (X-ray photoelectron) spectra of the zeolites were measured at room temperature on a Hitachi 507 photoelectron spectrometer using an A1-K, radiation (1486.6 eV; 9 kV, 50 mA).The zeolite samples were mounted on double-sided adhesive tape and were evacuated at room temperature in a pretreatment chamber (< 1 x Pa) prior to the X.P.S. measurements. The binding energies were referenced to the C 1s band at 285.0eV due to adventitious carbon. Mo(CO),-NaY samples that were prepared and activated at elevated temperatures in a conventional vacuum line were transferred to the pretreatment chamber by using a nitrogen-filled glove box. In these samples, the binding energy of the Si 2p level from the NaY zeolite predetermined above was utilized for establishing the charging correction. Results and Discussion The i.r. spectra of Mo(CO), adsorbed on Nay, (JRC-Z-Y5.6) are depicted in fig.1. It is evident that on room-temperature adsorption, the structure of Mo(CO), is retained almost intact with some distortions in the symmetry of the molecule. The lowering of the molecular symmetry is evidenced by the appearance of the normally infrared-inactive v, bands around 2120 cm-l [fig. 1 (b)]. The wavenumbers of the fundamental vibrations of Mo(CO), encapsulated in the NaY and NaX zeolites were calculated from the various combination bands at 2500-1900 cm-l. They are summarized in table 1 and compared with the literature values14-16 for Mo(CO), in various phases. Our results are in excellent agreement with those calculated from the data reported by Yon-Sing and Howe' except for the additional v1 band at 21 17 & 2 cm-l in the present study.The fundamental frequencies of Mo(CO), encaged in the NaY seem to be rather close to those for vapourY. Okamato et al. 853 f I co 0 03 A u 2200 2000 1800 (1111111 2200 2000 1800 1600 - 2200 2000 1800 1600 wavenumber/ cm - ' Fig. 1. Infrared spectra of adsorption and decomposition of Mo(CO), on NaY (JRC-Z-Y5.6) zeolite: (a) background (evacuated at 673 K for 1 h); (b) exposed to Mo(CO), vapour for 1 min, followed by evacuation for 10 min at room temperature; (c) evacuated at 373 K for 20 min; ( d ) evacuated at 423 K for 20 min. or for solutions, implying weak adsorptive interaction of Mo(CO), in the zeolite supercage. Concerning the doubled v, band in fig. I (b), equivalent pair peaks were also observed for Nay, (SK-40). The spectral intensities of these two v1 bands were found to grow simultaneously with increasing adsorption of the carbonyl up to saturation.However, the 21 17 cm-l band seemed to be removed more easily on evacuation of the sample at room temperature than' the band at 2123k2 cm-l, resulting in the formation of subcarbonyl species (e.g. 1808 cm-l). Accordingly, the double peak may suggest the existence of two kinds of Mo(CO), with slightly different configurations as a result of different interaction modes with the zeolite framework. A similar double v, peak has also been noted for Mo(CO), dissolved in CHC1,l5 as quoted in table 2. This was reconfirmed in the present study. On evacuation of the sample at room temperature, three new bands appeared at 1835, 1808 and 1790 cm-l as shown in fig.1 (b) and gradually grew with increasing evacuation time, accompanied by a colour change from white to light yellow. The yellow colour may indicate the formation of Mo(CO), species. These observations show a gradual decomposition of Mo(CO), to produce subcarbonyls. The i.r. spectrum changed drastically on evacuation of the sample at 373 K, showing the formation of a new fairly stable subcarbonyl species [fig. 1 (c)]. Some of this species remained anchored even at 423 K [fig. l(d)] and disappeared completely on further evacuation at 473 K. This subcarbonyl species is characterized by major absorption bands at 19 15, 1796 and 1768 cm-'. These spectral features for the NaY zeolite are quite consistent with the results reported by Howe and coworkers.'* '854 Thermal Stabilities of Molybdenum Carbonyl Compounds in Zeolites Table 1.Fundamental frequencies of Mo(CO), encaged in NaY and NaX zeolites NaY solutions approximate normal" spectral" this mode activity work Howeb NaX vapourC CHC13d CH2C12a solid" V l v2 v3 v4 v5 v7 VS V9 VlO v11 v12 '6 '13 Re 2123 2127-2123 2123 2124 2115 2117 2112 - 21 17 - - - 2110 - - 40 1 R 415 418 R 2016 2014 - 2027 2021 2019 2003 R - - - 344 - 394 389 555 IA - - - 481 - - i.r. 1973 1975 1955 2004 1983 - 1990 593 - - - 593 - - i.r. i.r. 357 357 368 - - 370 i.r. 81 84 - - - 506 - 448 468 R R 110 109 - 81 - 91 100 478 - - 512 - - IA - - - 62 - - - IA - - 392 - - - - 81 - - - a Ref. (14). Calculated from the data of ref. (7). Ref. (16). Ref. (15). R: Raman; i.r. : infrared; IA : inactive.Table 2. CO stretching frequencies of molybdenum carbonyls encaged in NaY and NaX zeolites carbonyl i.r. bands" zeolite species observed/cm-l assignments NaY Mo(C0)6 I I1 I11 IV NaX Mo(C0)6 11' 111' IV' 2123,, 21 17,, 2048,, 1970,s 1810.- 2046, 1953, 1835 2020, 1918, 1789 1955,,, 1925,,, 1913, 18 1793,, 1765,, 2123, 1955, 2047, 1828 1910, 1782 1916,, 1896, 1763,, " Only resolved bands are shown here: vs: very strong; s: strong; w: weak; and sh: shoulder peak. In the present study, more detailed decomposition steps were investigated by increasing stepwise the evacuation temperature up to 370K. Fig. 2 depicts such i.r. spectra for Nay, zeolite. Nay, showed essentially the identical characteristics. However, it was recognized that Nay, decomposed Mo(CO), more readily at room temperature than Nay,, resulting in stronger bands at 1835, 1810 and 1788 cm-' in a similar treatment sequence [fig.1 (b) us. 2(a)]. After a prolonged evacuation at room temperature, the 1789 cm-l band became the most intensified [fig. 2(b)]. At slightly elevated temperatures (313-318 K), several new bands appeared as shown in fig. 2(c) togetherY. Okamato et al. 855 with the intensified band around 1790 cm-'. On further evacuation at 3 17-33 1 K [fig. 2(4 and (e)], the 2123 cm-l band due to Mo(CO), was completely removed and the 2045 and 1953 cm-' signals were reduced in intensity. However, the absorption bands at 2020 and 19 18 cm-l showed maximum intensity at 3 17-320 K [fig. 2 (d)]. At temperatures higher than 333 K [fig. 2(f)] the stable carbonyl species was observed as in the case of Nay, zeolite [fig.1 (c)], which remained intact even after a prolonged evacuation at 383 K [fig. 1 (g)]. These decomposition steps were completely reversible with regard to evacuation and CO introduction. Treatment at 423 K in vacuo reduced appreciably the whole spectral intensity [fig. 2(h)], whereas thermal treatment at 450 K almost eliminated the carbonyl bands. The intensities of the bands at 1913 and 1925 cm-' were reversed at 423 K. A similar observation was made with Nay, [fig. 1 (c) us. 1 (41. These findings imply that a few stabilized subcarbonyl species with slightly different configurations exist as a consequence of different adsorption sites in the supercage (e.g. S,, and SII,). It is apparent from the i.r.studies that Mo(CO), encaged in the NaY zeolite is decomposed to the stable subcarbonyl species through several intermediate sub- carbonyls. On the basis of the spectral changes, subcarbonyl species I-IV are tentatively proposed and summarized in table 2. It seems that Mo(CO), is thermally decomposed stepwise from I and/or I1 through IV. Comparing Nay, and Nay, zeolites, it is revealed that the thermal stabilities of Mo(CO), and intermediate subcarbonyl species (1-111) are greater in Nay, (Si/Al = 2.78) than in Nay, (Si/Al = 2.43), while the stability of species IV is entirely reversed in these NaY zeolites. The oxidation state of molybdenum species in the NaY zeolite was examined utilizing X.P.S. Fig. 3 shows the X.p. spectra of the Mo 3d level for the MO(CO),-NaY, evacuated at elevated temperatures (A) and subsequently exposed to air at room temperature for controlled periods (B).The binding energies of the Mo 3d levels are summarized in table 3 and compared with those for several molybdenum compounds relevant to this system.17*ls The Mo 3d5,2 energies for the Mo(CO),-Nay, activated at various conditions are very close to those for molybdenum metal and molybdenum carbonyl complexes within the accuracy of the X.P.S. measurements (k 0.2 eV). Slightly broadened Mo 3d spectra observed at high evacuation temperatures (> 470 K) are ascribable to the molybdenum species oxidized to some extent during heat treatment perhaps by adsorbed (residual) water. Accordingly, it is concluded that molybdenum is retained essentially in a zero-valent state during the thermal treatments of Mo(CO), encaged in the NaY zeolites with formation of zero-valent subcarbonyl species at 370 K and final decomposition to molydenum metal (> 470 K).With the molybdenum/NaY zeolites, the reactivity of molybdenum species to air was examined by X.P.S. As depicted in fig. 3 (e), the molybdenum species (subcarbonyl IV in table 2) produced by evacuation at 373 K was completely oxidized to an Mo6+ species, when exposed to air at room temperature for 50 s. This was also inferred by the instantaneous colour change from brown to white observed. On the other hand, the X.p. spectra for the Mo species activated at > 473 K were slightly broadened by relatively short oxidation treatments (50 or 130 s) at room temperature, suggesting that only a small part of metallic species is oxidized to higher-valent species.A prolonged exposure to air produced high oxidation states of molybdenum ; Mo4+ and Mo6+ as deduced from the X.P.S. binding energies. These findings indicate that molybdenum metal aggregates are formed rather than atomically dispersed metal particles, even at 470 K. The aggregation of Mo metal is suggested to occur at 673 K from the pore volume measurements of Mo(CO),-Nay.' Fig. 4 depicts the i.r. spectra of Mo(CO), encapsulated in the NaX zeolite following progressive evacuation at ambient and elevated temperatures. In sharp contrast to the NaY zeolites, the NaX zeolite is shown to decompose Mo(CO), drastically even at room temperature, producing stepwise several subcarbonyl species.Upon mere contact of856 Thermal Stabilities of Molybdenum Carbonyl Compounds in Zeolites i I I I I I I I I 2200 2000 1800 1600 m r- 4 1 1 1 1 1 1 1 1 2200 2000 1800 1600 wavenumber/cm-' Fig. 2. Infrared spectra of the decomposition of Mo(CO), adsorbed on NaY (SK-40) zeolite pre- evacuated at 673 K for 1 h: (a) exposed to Mo(CO), vapour for 30 s, followed by a 10-min evacuation at 293 K; (b) additional evacuation for 60 min at 295 K; (c) evacuated at 313-318 K for 10 min; ( d ) at 317-320 K for 10 min; (e) at 322-331 K for 10 min; at 333-341 K for 20 min; ( g ) at 383 K for 6.5 h; (h) at 415-423 K for 10 min, successively.Y. Okamato et al. 857 B A A 1 I I I 1 I I I 240 235 230 225 240 235 230 225 binding energy/eV Fig. 3.X-Ray photoelectron spectra of Mo(CO),-NaY (JRC-Z-Y5.6) zeolite : (A) (a) Mo(CO), adsorbed on NaY at room temperature; (b) evacuated at 373 K for 20 min; (c) evacuated at 473 K for 20 min, and ( d ) evacuated up to 633 K in a temperature-programmed decomposition experiment at a rate of 1.8 K min-': (B) (e) exposure of sample (b) to air for 50 s at room temperature; cf) exposure of sample (c) to air for 50 s; ( g ) subsequent exposure of sample (f) to air for 24 h; (h) exposure of sample ( d ) to air for 130 s. Mo(CO), with the NaX, the i.r. spectrum appeared to be quite different from that for Mo(CO), adsorbed on the NaY [fig. 4(b)]. The sample turned orange in colour. The bands due to subcarbonyls prevailed over the signals attributable to Mo(CO), (v, 2123 and v, 1955 cm-l in table 1).With increasing evacuation time [fig. 4(c)], the bands at 1782 cm-' increased in intensity, while the signal intensities at 1828, 1910 and 2047 cm-l decreased simultaneously, perhaps after a slight initial enhancement. The appearance of a shoulder peak at 1763 cm-' on prolonged evacuation [lo0 min; fig. 4(e)] may explain the somewhat complicated behaviour of the bands around 1910 cm-l, since the 1763 cm-l band accompanies two strong bands at 1916 and 1896 cm-l, as shown in fig. 4cf). No band due to subcarbonyl I (1810 cm-l) in the NaY zeolites was detected with the NaX zeolite. On evacuation of the sample at elevated temperatures, stable subcarbonyl species, characterized by three bands, was formed [fig. 4 0 1 and retained even at 473 K in spite of a slight decrease in the relative intensity of 1896 cm-' band [fig.4 (g)], as observed with the corresponding absorption bands for the NaY zeolites [(fig. l(d) and 2(h)]. The carbonyl bands vanished at 573 K. Introduction of CO (3.3 x lo3 Pa) at room temperature into the subcarbonyl-NaX system activated at 473 K readily restored the original spectrum. However, after the sample had been treated at 573 K under vacuum, no carbonyl absorption bands were restored on exposure of the sample to CO858 Thermal Stabilities of Molybdenum Carbonyl Compounds in Zeolites Table 3. X-Ray photoelectron spectroscopy data for Mo(CO),-NaY zeolite treated at elevated temperatures in uucuo and for reference compounds binding energy/eV" treatment temperature/K Mo 3d5,2 Mo 3dsi2 ref.ca. 290 373 473 633 MOO, MoV MOO, Mo metal [(?f-C5H5)M0(C0)312 Mo(~~-C,H,)(CO),CI (CH3)3SnMo(C0)3(775-C5H5) Cl(CH,),SnMo(CO),($-C,H5) 228.5 228.6 233.0b 228.2 228.4b 229.9, 232.6c 228.4 228.4d 233.1 23 1.9 229.9 228.4 227.8 229.2 228.9 228.4 231.8 this work 231.6 235.6b 231.2 231.1b 23 1 .O 23 1 .Od 17" 1 8f a Referenced to the Si 2p level at 102.5 eV. Exposed to air at room temperature for 50 s. For 24 h. For 135 s. Referenced to C 1s = 285.0 eV. f The binding energies are shifted by + 1.4 eV to adjust the Mo 3d binding energy for MOO, to that of ref. (17). (3.5 x lo3 Pa) at room temperature. These findings indicate that the decomposition of Mo(CO), to the subcarbonyl species is reversible in the NaX zeolite, while the subcarbonyl species are irreversibly decomposed to molybdenum metal.Similar observations were made for the NaY zeolites. The i.r. bands observed during the decomposition of Mo(CO), in NaX are tentatively attributed to the species 11'-IV' in table 2 on the basis of the spectral features on the thermal activations in fig. 4. Mo(CO), decomposes stepwise to the subcarbonyls 11', 111', IV', and finally to Mo metal. The subcarbonyl species (IV') encaged in the NaX zeolite is considerably more stable than the corresponding species IV in the NaY zeolites, whereas Mo(CO), and the intermediate subcarbonyls show completely reversed thermal stabilities in these matrices. T.p.d. experiments were conducted in a dynamic vacuum to confirm the afore- mentioned i.r. results. Fig. 5 depicts the t.p.d. spectra for Mo(CO), encapsulated in the NaY and NaX zeolites.It is clearly demonstrated that Mo(CO), decomposes in two main steps in these zeolites. This decomposition feature substantiates the existence of thermally stable subcarbonyl species on the zeolites. In the vapour phase, Mo(CO), is known to decompose thermally to MO metal at 423 K.19 Accordingly, it is evident that in the zeolite matrices the decomposition of Mo(CO), is initiated at considerably lower temperature, whereas the final decomposition of the subcarbonyl species is markedly suppressed. The carbonyl decomposition temperatures, TL (lower) and TH (higher), for the zeolites employed herein are summarized in table 3. TL decreases in the order of Nay, > Nay, > NaX, whereas TH increases as Nay, < Nay, < NaX.Comparing these thermal stabilities of hexa- and sub-carbonyl species with those observed in the aforementioned i.r. results, it is concluded that the TL peak corresponds to the formation of the subcarbonyl species IV and IV' by the successive decomposition of Mo(CO), and intermediate subcarbonyls, whereas the TH peaks arise from the thermal decomposition of species IV and IV' to Mo metal.Y. Okamato et al. 859 I I I I 1 1 1 1 I I I I I I I I I wavenumber/cm -' 2200 2000 1800 1600 2200 2000 1800 1600 Fig. 4. Infrared spectra of adsorption and decomposition of Mo(CO), on NaX zeolite: (a) background (evacuated at 673 K for 1 h); (b) exposed to Mo(CO), vapour for 1 min, followed by an evacuation for 10 min at room temperature; (c) evacuated for 40 min; (d) for 70 min; ( e ) for 100 min at room temperature; cf> for 20 min at 373 K; (g) for 20 min at 473 K.860 Thermal Stabilities of Molybdenum Carbonyl Compounds in Zeolites 300 400 500 decomposition temperaturelK Fig.5. Temperature-programmed decomposition spectra of Mo(CO), adsorbed on (a) NaY (JRC-Z-Y5.6); (b) NaY (SK-40); (c) NaX zeolites. Solid line, CO (m/z = 28); broken line, H, (m/z = 2). The stoichiometries, CO/Mo ratio, of subcarbonyl species IV and IV' can be estimated from the t.p.d. spectra. The amount of evolved CO is considered to be proportional to the spectral area. The peak area ratios of the TL and TH peaks were 1.4, 1.3 and 1.0 for Nay,, Nay, and NaX, respectively. These are close to unity. Furthermore, it is considered that the total peak area corresponds to the stoichiometry of CO/Mo = 6, since no formation of CH, and CO, were detected in the present systems.Therefore, it is concluded that the major subcarbonyl species stable at ca. 370 K are MO(CO),,,, rather than MO(CO),,,, or Mo(CO),,,,. The existence of stable MO(CO),,,, species has also been proposed for some oxide supports, Al,03'-lo and ZnO," while Howe et aL7 have claimed the formation of stable Mo(CO),,,, on NaY zeolite. The cause of the difference in the stoichiometry between the present and Howe's studies is not clear at present. However, the i.r. spectra for species IV and IV', which have been regarded as evidence of MO(C0)4,d,,7 do not seem to be in conflict with the Mo(CO),,,, species as discussed below. As shown in the i.r.studies, Mo(CO), is decomposed stepwise to species IV or IV'. Taking into account the stoichiometries of the stable subcarbonyls, the subcarbonyl species differentiated by the i.r. bands are tentatively assigned in table 3. The proposed decomposition steps are given in scheme 1. Scheme 1Y. Okamato et al. 86 1 In the NaX zeolite, the i.r. peak assigned here to Mo, (CO),, was not observed. This may be due to a higher reactivity of the NaX than the NaY zeolites toward the rather unstable species. The existence of stable intermediate species, such as Mo(CO),, was not suggested by the i.r. studies during the decomposition of Mo(CO),,,, to Mo metal. This may also be evidenced by sharpened t.p.d. peaks at TH relative to the peaks at TL. Further studies are required for a decisive conclusion.The i.r. spectrum due to species IV in fig. l(c) or 2Cf)-(h) consists mainly of three bands. Mo(CO), compounds with C,, symmetries should show only two bands, as observed in many organometallic compounds. However, when the structure of Mo(CO), is distorted to a lower symmetry, more than two bands are anticipated. Very recently, Kirlin et aZ.,O have reported the i.r. spectra of the surface species prepared from the reaction of H,Re,(CO),, with A1,0, or MgO. Their spectral features are very close to those for species IV. They have assigned the surface species to Re(CO),(OM)- (HOM), (M = Mg or Al) on the basis of i.r., electron tunnelling, and U.V. spectra. Therefore, it seems reasonable to assign the i.r. spectra of species IV to an Mo(CO), moiety adsorbed on an oxygen triad with a significantly distorted octahedral symmetry.The structure is proposed in fig. 6. The Mo(CO), moiety is considered to be greatly stabilized by forming three dative bonds with zeolite oxygens. From considerations analogous to those of Kirlin et al.,,' it is proposed that two M o t 0 bonds are weaker than the remaining one in NaY and the reversed structure in NaX. The coordinations of oxygen lone-pair electrons to the Mo(CO), moiety are readily broken to restore Mo-CO bonds upon an exposure to CO. Stabilization of surface M(CO), species by dative bonds from surface hydroxy groups and/or lattice oxygens are also proposed for Mo(CO), on A1,0,' and Re(CO), on MgO and Al,O,. 2o The other small shoulder peaks may suggest the presence of more than one Mo(CO),,,, species with slightly different structures (e.g.on site I1 or site 111) and/or possible small perturbations of v(C0) by Na+ cations (Lewis acid-carbonyl interactions). Finally, the electronic structure of the zeolites was evaluated on the basis of the X.p. spectra to understand the dominant factors in elucidating the drastic differences between the zeolites in the thermal stabilities of Mo(CO), and the subcarbonyls. The binding energies of the A1 2p, Si 2p and 0 1s levels for the zeolites examined here are summarized in table 4. The electronic structure of a zeolite system will be discussed in more detail elsewhere.,' The binding energy or charge density of aluminium atoms seems to be invariant over the present range of zeolite composition.However, the Si 2p and 0 1s binding energies decrease with increasing the A1 content, suggesting a decrease in the positive charge on the silicon atoms and an increase in the negative charge on the oxygen atoms. It is unlikely that silicon and aluminium atoms act as Lewis acid sites accessible to adsorbed molecules, since they are well shielded by four oxygen anions. Accordingly, it is natural to propose that the lattice oxygens in the zeolite framework act as interaction sites with the Mo-carbonyls. The X.P.S. results in table 3 clearly demonstrate that the base strength of lattice oxygen increases with increasing A1 content. As mentioned above, TL decreases as the A1 content increases, while TH increases. These results lead us to conclude that the decompositions of Mo(CO), (x = 6-4) are promoted by the basic sites in the zeolite, whereas Mo(CO),,,, is stabilized by the identical nature of the sites ; stronger basic sites induce a greater stabilization of Mo(CO),,,,. The contribution of basic sites to the carbonyl-zeolite interactions is reflected on the observed band positions in table 2.The bands of the carbonyls in the zeolites are considerably red-shifted, compared with those of corresponding organometallic compounds as noted by Howe et al.,,' This is interpretable in terms of the dative interactions with zeolite framework oxygen, resulting in the increased electron density on the molybdenum atom and subsequent red-shifts of the v(C0) bands. In addition, the carbonyl species in the NaX zeolite always exhibit lower wavenumbers than the corresponding species in the NaY zeolites, as shown in table 2.This is amarentlv due862 Thermal Stabilities of Molybdenum Carbonyl Compounds in Zeolites 0-0-0 Fig. 6. Proposed model for Mo(CO),,,,. on NaY zeolites. Table 4. X-Ray photoelectron spectroscopy results on the zeolites and decomposition temperatures obtained by t.p.d. of Mo(CO), adsorbed on the zeolites zeolite decomposition" T/K b.e./eVb composition ; Si/Al TL TH Si 2p A1 2p 0 IS Nay, JRC-Z-Y 5.6 2.78 378 454 102.5 74.2 532.0 Nay, SK-40 2.43 366 459 102.5 74.2 531.8 NaX 13X 1.23 333 498 101.9 74.1 531.1 a Rate of temperature increase: 1.8 K min-'. TL lower decomposition temperature; TH higher one. Referenced to C 1s = 285.0 eV. to a stronger electron donation of the NaX zeolite compared with the NaY zeolites.The stronger electron donation induces stronger Mo-C bonds, resulting in the higher thermal stability of Mo(CO),,,,. Conclusions The decompositions of Mo(CO), encapsulated in NaY and NaX zeolites were investigated to examine the effect of zeolite composition using i.r., temperature- programmed decomposition, and X.P.S. techniques. The salient findings and conclusions in this study are as follows: (1) Mo(CO), decomposes stepwise in a vacuum through intermediate carbonyl species to a thermally stable subcarbonyl species, Mo(CO),,,, and finally to molybdenum metal aggregates ; (ii) the thermal stabilities of Mo(CO), and of the intermediate subcarbonyl species are considerably lower in the NaX zeolite than in the NaY zeolites, while the stability of Mo(CO),,,, is completely reversed in these zeolites; (iii) the marked differences in the thermal stabilities of the molybdenum carbonyl species between the NaX and NaY zeolites are explained in terms of the difference in the basic strength of framework oxygen in the zeolite; (iv) the structure of Mo(CO),,,, is proposed on the basis of the i.r.and X.p. spectra. References 1 D. C. Bailey and S. H. Langer, Chem. Rev., 1981,81, 109. 2 J. Phillips and J. A. Dumesic, Appl. Catul., 1984, 9, 1. 3 J. M. Thomas, in Chemistry and Physics of Solid Surfaces VI, ed. R. Vanselow and R. F. Howe (Springer-Verlag, Berlin, 1986), p. 107. 4 (a) T. Yashima, T. Komatsu and S. Namba, Chern. Express, 1986, 1, 701 and references therein; (6) T. Yashima, T. Komatsu and S. Namba, in Proc. Climax Fourth Znt. Con$ Chemistry and Uses ofY. Okamato et al. 863 Molybdenum, ed. H. F. Barry and P. C. H. Mitchell (Climax Molybdenum Comp. Ann. Arbor, Michigan, 1982), p. 274. 5 P. Gallezot, G. Coudurier, M. Primet and B. Imelik, in Molecular Sieves ZZ, ed. J. R. Katzer (ACS Symp. Ser., 1977), vol. 40, p. 144. 6 S. Abdo and R. F. Howe, J. Phys. Chem., 1983, 87, 1713. 7 Y. Yon-Sing and R. F. Howe, J. Chem. SOC., Faraday Trans. I , 1986, 82, 2887. 8 A. Brenner and R. L. Burwell, J. Am. Chem. SOC., 1975, 97, 2565. 9 A. Brenner and R. L. Burwell, J . Catal., 1978, 52, 353. 10 A. Kazusaka and R. F. Howe, J. Mol. Catal., 1980, 9, 183. 11 K . Tanaka, Y. Zhai and K. Aomura, in Proc. Climax Znt. Conf. Chemistry and Uses of Molybdenum, ed. H. F. Barry and P. C. H. Mitchell (Climax Molybdenum-Comp. Ann Arbor, Michigan, 1982), p. 278. 12 T. Hattori, H. Matsumoto and Y. Murakami, 4th Znt. Symp. Preparation of Heterogeneous Catalysts 13 R. R. Monchamp and F. A. Cotton, J. Chem. SOC., 1960, 1438. 14 R. L. Amster, R. B. Hannan and M. C. Tobin, Spectrochim. Acta, 1963, 19, 1489. 15 L. H. Jones, J. Chem. Phys., 1962, 36, 2375. 16 L. H. Jones, Spectrochim. Acta, 1963, 19, 329. 17 Y. Okamoto, K. Oh-Hiraki, T. Imanaka and S. Teranishi, J. Catal., 1981, 71, 99. 18 S. 0. Grim and L. J. Matienzo, Znorg. Chem., 1975, 14, 1014. 19 E. M. Fednova and J. V. K. Krykova, Russ. J. Znorg. Chem., 1966, 11, 141. 20 P. S. Kirlin, F. A. DeThomas, J. W. Bailey, H. S. Gold, C. Dybowski and B. C. Gates, J. Phys. Chem., 21 Y. Okamoto, A. Maezawa, M. Ogawa and T. Imanaka, unpublished work. (Louvain-la-neuve, 1986). 1986, 90, 4882. Paper 71532; Received 24th March, 1987

ISSN:0300-9599    

DOI:10.1039/F19888400851

出版商:RSC    

年代:1988

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. 1, 1988, 84(3), 865-869 Hydrogen Bonding Part 3.-Enthalpies of Transfer from 1 , 1 ,l -Trichloroethane to Tetrachloromethane of Phenols, N-Methylpyrrolidinone (NMP) and Phenol-NMP Complexes Michael H. Abraham,* Philip P. Duce and David V. Prior Department of Chemistry, University of Surrey, Guildford GU2 5XH Ronald A. Schulz Department of Chemical and Process Engineering, University of Surrey, Guildford GU2 5XH Jeffrey J. Morris and Peter J. Taylor ICI Pharmaceutical Division, Mereside, Alderley Park, Macclesjield SKI0 4TG Enthalpies of solution of seven phenols and of NMP have been determined in 1, I, 1-trichloroethane and in tetrachloromethane at 298 K. Combination with our previously determined enthalpies of complexation, AW, of the phenols with NMP leads to enthalpies of transfer of the hydrogen-bond complexes ArOH NMP from 1,1,1-trichloroethane to tetrachloro- methane.The substituent-dependent behaviour of values of AW in the two solvents arises exclusively from the effect of the solvents on the reactant phenols and not on the complexes themselves. This finding confirms the suggestion that the AHo values in 1,1,1 -trichloroethane contain a contri- bution from association of the phenols with the solvent itself, probably of the dipoledipole type. ~~ In a previous part of this series' we studied the hydrogen-bond complexation of some substituted phenols with N-methylpyrrolidinone (NMP) in the solvents 1,l , 1-trichloroethane (TCE) and tetrachloromethane : ArOH + NMP $ ArOH -.. NMP. For complexation in the latter solvent, a plot of AH" against AGO was linear with a positive slope, exactly as is observed for numerous series of phenols against various bases in non-polar solvents such as benzene, cyclohexane or tetrachloromethane.' 9 However, in the case of TCE the corresponding plot of AH" against AGO has a smaller slope that is negative : a most unusual feature. Details of the complexation constants for some phenols that we have studied are given in table 1, and regression lines for the A W us. AGO plots are, with values in kcal mol-':? AHo(in CCl,) = -2.83 + I .03 AG"(in CC1,) (2) AHo@ TCE) = -7.36-0.62 AG"(in TCE) (3) In eqn (2) and (3) n is the number of data points, r is the correlation coefficient and s the standard deviation in kcal mol-' and F is the F-statistic (variance ratio).We suggest that involvement of the dipolar solvent TCE (p = 1.78 D)$ in the complexation reaction, eqn (l), led to this peculiarity. If TCE formed a loose association with the dipolar phenols, probably of the dipole4ipole type, release of TCE during the complexation reaction would result in A W for eqn (1) being more positive than (1) ( n = 7, r = 0.94, s = 0.23, F = 52). (n = 7, r = 0.74, s = 0.32, F = 63). t 1 cal = 4.184 J. $ 1 D z 3.3356 x C m. 29 865 FAR 1866 Hydrogen Bonding in Phenols and N- Methylpyrrolidinone AH: (ArOH) Table 1. Gibbs energies and enthalpies for complexation of phenols with NMP in TCE and tetrachloromethane in kcal mol-' at 298 K" AH: (NMP) A HZ(comp1ex) AH' (CCI,) phenol TCE AGO AW - cc1, AGO AW 3-methylphenol - 2.58 - 5.63 -2.88' - 5.96' phenol - 2.92 -5.31 -3.18 - 5.69 4-fluorophenol - 3.33 - 5.48 - 3.44 - 6.36 3-trifluoromethylphenol - 3.49 - 5.64 - 3.90 - 7.07 4-trifluoromethylphenol - 3.60 - 5.30 - 4.02 - 7.05 pentafluorophenol - 3.90 -4.57 -4.21 -7.12 3,5-di-(trifluoromethyl)phenol - 4.02 - 4.70 -4.86 - 7.70 a All values from ref.(1) except where shown. This work. expected, owing to the breakage of the phenol-TCE bonding. The more dipolar (and in general the more reactive) the phenol the greater would be the solvent involvement and the more positive (or less negative) would be the observed AH" value in TCE. Thus in the scheme given by reactions (4H6): ArOH - TCE + ArOH + TCE (4) ArOH + NMP --+ ArOH NMP ( 5 ) NMP + ArOH - TCE + ArOH NMP + TCE (6) since A q + A G = AX, where AX is the observed enthalpy change in TCE (AW&), a positive A% value could result in an 'intrinsic' negative enthalpy change, A%, being converted into an observed enthalpy change that is more positive the more reactive the phenol (see fig.1). A corollary of this argument, that we did not pursue before,l is that the difference in AH" values in TCE and tetrachloromethane, corresponding to the different slopes in eqn (2) and (3), should be due to a solvent effect on the phenols and not to a solvent effect on the hydrogen-bond complexes (ArOH -.a NMP). Any solvent effect on NMP itself will cancel out along a series of phenols. A thermodynamic cycle may be constructed [cf. ref. (4)-(6)] in terms of enthalpies of transfer, A%, of reactants and complex that will enable this prediction to be tested.We define A% as the enthalpy of a species in tetrachloromethane less the enthalpy in TCE. Then A% for phenols and NMP can beM. H. Abraham et al. 867 AH' AGol Fig. 1. The possible influence of phenol-solvent association, AK (a), on intrinsic enthalpies of complexation, A% (c), in (A) TCE and (B) tetrachloromethane, as shown by plots against AGO, to give the resultant observed enthalpies of complexation (b). Aq(comp1ex) = Aq(pheno1) + Aq(NMP) + - AH&E. (9) In these two reactions, the enthalpies of complexation in TCE and in tetrachloromethane are denoted as AW&E and AWcc14. Experimental The solvents, phenols and NMP were as described previously.' Enthalpies of solution were obtained using an LKB-8700 solution calorimeter as before.7* The instrument was tested by determining the molar enthalpy of solution of tris(hydroxymethy1)amino- methane in 0.1 mol dm-3 hydrochloric acid, which was found to be 71 14+_ 7 cal mol-', the average of nine experiments.This agrees quite well with our own recent determination,8 7104f 8 cal mol-l, and with the originally reported' value of 7107f4 cal mol-l. Enthalpies of solution of solutes were carried out in duplicate, with final concentrations below the solute self-association limit's of 0.01 mol dm-3. All these enthalpies were corrected for the enthalpy of breaking empty ampoules in the two solvents concerned. 29-2868 Hydrogen Bonding in Phenols and N- Methylpyrrolidinone Table 2. Enthalpies of solution, and of transfer from TCE to tetrachloromethane, in kcal mol-1 at 298 K phenol A% AfC in TCE in CC1, phenol phenol- complex NMP 3-methylphenol phenol 4-fluorophenol 3- triflurometh ylphenol 4- triflurome thylphenol pent aflurop heno 1 3,5-di-(trifluromethyI)phenol NMP 2.59 4.71 5.10 1.12 3.62 3.70 1.29 - 0.48" 3.93 1.34 1.82 1.49 5.78 1.07 1.55 1.77 6.65 1.55 2.03 1.15 3.79 2.67 3.15 1.72 5.61 1.99 2.47 0.72 7.64 3.94 4.42 1.87 4.84 3.55 4.03 1.03 0.00 A G&, a Ref.(1). I I -5 -4 0 -6 - 8 L-10 Fig. 2. Plots against AGE,,, of: A, A% in TCE, assumed to be given by A q (phenol); 0, AH, (AH&,) and + , in TCE assumed to be given by [AH",,, -AfC (phenol)]. Results and Discussion In table 2 are given the enthalpies of solution and desired enthalpies of transfer we have obtained for the seven phenols studied.The total reactant transfer values are given as A K (phenol) plus A% (NMP), with the latter value taken as 0.48 kcal mol-'. Combination with the A.H" values for complexation (table l)t yields AK for the corresponding ArOH...NMP complexes given in the final column of table 2. The A K (phenol) values show a very marked trend with AGEc4 (or AGgCE), with A% becoming larger (more positive) as AGO becomes more negative. This is exactly as expected if sOme loose phenol-TCE association has to be broken on transfer from TCE to tetrachloromethane. The effect of change of solvent on the hydrogen-bond complex itself is almost independent of the phenol substituent, at 1.3 1 & 0.41 kcal mol-l. Since the random error in the enthalpies of complexation are ca.0.2 kcal mol-1 in each case, and that in the AX determination is ca. 0.1 kcal mol-', the combined error in values of tetrachloromethane, but we prefer to use our own values for consistency. t Virtanen and Jarva'O have determined enthalpies of complexation of some phenols with NMP inM. H. Abraham et al. 869 Aq(comp1ex) will be ca. 0.4 kcal mol-l. Hence within experimental error, AK (complex) is constant. We can show graphically exactly how the predicted enthalpy vs. Gibbs energy plots in fig. 1 arise in the present case. If the phenols do not associate appreciably with tetrachloromethane, we can take Aq(pheno1) as an estimate of AX and hence the quantity [AHO,. - A%(phenol)] = [A% - Ae(phenol)] as an estimate of the intrinsic enthalpy, A%.Plots? of A%, AH&. (A%) and AK for complexation in TCE, all against AGO, are shown in fig. 2. The predicted pattern of fig. 1 is followed exactly. A plot of against AGO is not given in fig. 2 in order to simplify the diagram, but lies almost parallel, and quite close, to the calculated intrinsic enthalpy change, A%. Since the solvent effect on AGO values (table 1) is almost independent of substituent, the large enthalpic effect on the reactants is probably counteracted by a substituent- dependent entropic effect. Two important consequences follow. First, the effects of phenol-solvent interactions on the Gibbs energy of complexation will be small or non- existent; hence AGO values in TCE are quite normal by comparison to those in tetrachloromethane. Secondly, because solute-solvent interactions can lead to significant effects on enthalpies and entropies of complexation but not on Gibbs energies of complexation, the latter parameter will be the most useful one to use in any construction of a scale of solute hydrogen-bond acidity.We are grateful to the S.E.R.C. for a CASE award (to D.V.P.). References 1 M. H. Abraham, P. P. Duce, J. J. Morris and P. J. Taylor, J. Chem. SOC., Faraday Trans. 1, 1987, 83, 2 A. S. N. Murthy and C. N. R. Rao, Appl. Spectrosc. Rev., 1968, 2, 69. 3 M. D. Joesten and L. J. Schaad, Hydrogen Bonding (Marcel Dekker, New York, 1974). 4 S. D. Christian and E. E. Tucker, J. Phys. Chem., 1970, 74, 214. 5 W. Libus, M. Mecik and W. Salek, J. Solution Chem., 1977, 6, 865. 6 J. N. Spencer, R. S. Hainer and C. D. Penturelli, J. Phys. Chem., 1975, 79, 2488. 7 M. H. Abraham, P. P. D u e , R. A. Schultz, J. J. Morris, P. J. Taylor and D. G. Barratt, J. Chem. SOC., 8 M. H. Abraham and A. Nasehzadeh, J. Chem. SOC., Faraday Trans. 1, 1981, 77, 321. 9 R. J. Irving and I. Wadso, Acta Chem. Scand., 1984, 18, 195. 2867. Faraday Trans I , 1986, 82, 3501. 10 P. 0. I. Virtanen and M. Jarva, Acta Univ. Ouluensis Chem., 1973, no. 3 . Paper 7/1030; Received 1 Ith June, 1987 t Note that we take AGE,, as AGO in fig. 2, but it would make almost no difference if we had used AG&- instead. Also in ref. (1) it wai suggested that plots of AX, and A% against AGO were not straight lines for the most activated phenols (see fig. 1). However, within experimental error, we are unable to distinguish this feature .

ISSN:0300-9599    

DOI:10.1039/F19888400865

出版商:RSC    

年代:1988

数据来源: RSC