摘要:

,J. Chem. SOC., Faraday Trans. I, 1988, 84(1), 87-96 Adsorption of Carbon Dioxide, Ammonia and Pyridine on Sodium-modified Silicalite Yasuyuki Matsumura and Keiji Hashimoto Osaka Municipal Technical Research Institute, Joto-ku, Osaka 536, Japan Satohiro Yoshida" Department of Hydrocarbon Chemistry, Kyoto University, Sakyo-ku, Kyoto 606, Japan The surface properties of sodium-modified silicalite have been investigated by ESCA, i.r. spectroscopy, adsorption of carbon dioxide and ammonia, and temperature-programmed desorption. The results of carbon dioxide adsorption show that number of basic sites in sodium-modified silicalite is very small, although the existence of sodium ions enhances the adsorption. It has been observed that sodium-modified silicalite adsorbs ammonia and pyridine more stably compared with non-modified silicalite, but the t.p.d. profiles of adsorbed ammonia and i.r.spectra of adsorbed pyridine show that strong acid sites do not exist in the sodium-modified silicalite. These molecules are adsorbed on the silanol groups on the external surface of the silicalite and also on the sodium ions in the silicalite. Silicalite, which is an extremely aluminium-deficient ZSM-5 type zeolite,' has little ion- exchange capacity.' However, when the silicalite is synthesized from a mixture including sodium ions (alkali-metal cations are usually used as a raw material), sodium ions remain in the product as Intentional addition of sodium to silica or silica-alumina is known to increase the number of basic sites ;5* therefore, the acid-base properties of silicalites containing sodium ions are expected to differ from those of sodium-free silicalite, which is not strongly acidic7 or, probably, basic.Previously we found that sodium-modified silicalites catalyse selective dehydro- genation of methanol to formaldehyde and that the sodium ions in the catalyst play an important role in the dehydrogenation.* The contribution to the catalysis of transition- metal contaminants in the silicalite is small.' Alcohol dehydrogenation occurs over some acid-base catalysts.1°-12 It is expected that the surface properties of silicalites, such as their acid-base properties, are changed by sodium modification. These investigations are very important for clarification of the mechanism of catalysis; therefore, in the present work we have investigated the effect of sodium ions on the surface properties of silicalite.Experiment a1 Materials Silicalite was synthesized using the patented r n e t h ~ d . ~ The main starting materials were tetraethyl orthosilicate (purified by distillation), sodium nitrate (GR grade), and tetra- n-propylammonium hydroxide (GR grade). The organic components were removed by heating in air at 770 K for 8 h. The Si/Al ratio of this sample was found to be ca. 4000 by chelatometric titration. l3 The sodium contents of these samples, denoted Na ( 1 . I)-SL, are given in table I . Sodium in silicalite can be leached out by refluxing in water ;' the samples denoted as Na (0.4)-SL and Na (0.01)-SL in table 1 were prepared from Na (1. I)-SL in this manner, and dried at ca.400 K. The sodium content was analysed by 8788 Adsorption on Sodium-modijied Silicalite Table 1. Properties of silicalite samples Na content desorbed watera sample (wtY0) /mmol g-' Na( 1 .1)-SL 1.1 0.07 Na(0.4)-SL 0.4 0.04 Na(O.01)-SL 0.01 0.03 aAmount of water desorbed from the samplp- pretreated at 770 K by heating at 970 K. an atomic absorption method. When the samples pretrear at 770 K were heated at 970 K, desorption of water was complete. The desorbed ailiounts are also presented in table 1. The X-ray diffraction patterns were in agreement with those reported by Olson et a1.' Ammonia and carbon dioxide were high-purity gases supplied by Takachiho Kagaku Kogyo and were used without further purification. Pyridine was dried and distilled over 4A molecular sieves and further degassed by the conventional freeze-pumpthaw technique.Electron Spectroscopy for Chemical Analysis Surface analysis by ESCA was carried out using a Shimadzu ESCA 750 spectrometer. The samples were first heated outside the spectrometer in vacuo at 770 or 970 K and then contacted with air. They were mounted with adhesive tape and set into the spectrometer. After measurement, argon-ion etching of the sample was carried out (beam voltage, 2 kV; Ar gas pressure, 5 x Torr;t emission current, 20 mA). The spectra were measured again after etching, and this procedure was repeated. Adsorption and Temperature-programmed Desorption Adsorption of carbon dioxide or ammonia on the silicalite samples (0.100 g) was measured with a constant-volume method using a calibrated Pirani vacuum gauge.After adsorption equilibrium was attained the samples were evacuated for 1 h at room temperature. The adsorption was then measured again in order to evaluate the amount of reversibly adsorbed molecules. The number of irreversibly adsorbed molecules was calculated by subtraction of the amount of the reversible adsorption from that of the total adsorption. Temperature-programmed desorption (t.p.d.) of ammonia was measured with a conventional apparatus equipped with a thermal conductivity detector. The samples (0.100 g) were exposed to ammonia (ca. 5 Torr) for 0.5 h at room temperature, and evacuated for 1 h. T.p.d. measurements were carried out from room temperature to 770 K with a heating rate of 10 K min-l and with helium as carrier gas (flow rate, 0.020 dm3 min-l).Just before each measurement described above, the samples were heated in vacuo at 770 K for 1 h or at 970 K for 0.5 h. Infrared Spectroscopy 1.r. spectra were recorded with a Nicolet 5DX F.t.i.r. spectrometer. The samples (0.01 g) were pressed into self-supporting wafers and placed into an in situ i.r. cell t Torr = 101 325/760 Pa.Y. Matsumura, K . Hashimoto and S . Yoshida 89 Table 2. ESCA data for silicalite samples" Na(1s) NaKrZLb Si(2p) O(1s) sample l e v l e v l e v /eV Na/Si" Na( 1.1)-SL 1072.7 989.3 103.5 532.9 0.024 Na(0.4)-SL 1072.7 989.3 103.5 532.9 0.029 Na(O.0 1 )-SL 103.5 533.0 - - - Accuracy is estimated as f0.2 eV. Auger peak. ' Molar ratio. 0 1 I I I 0 5 10 15 etching time/m in Fig.1. Change in the Na/Si molar ratio of the sample surface with argon-ion sputtering. The samples were pretreated at 770 K. 0, Na(l.1)-SL; W, Na(0.4)-SL. allowing heating under vacuum. The samples were heated in vacuo at 770 K for I h or at 970 K for 0.5 h before the measurement at room temperature. Before i.r. spectra of adsorbed pyridine were recorded at room temperature, pyridine vapour (ca. 5 Torr) was contacted with the wafers at 420 K for 0.5 h and evacuated at a desired temperature for 1 h. Results ESCA Measurements In order to identify the chemical states of the surface constituents in the silicalite samples (especially sodium) an ESCA investigation was carried out. Typical ESCA results for the samples pretreated at 770 K are presented in table 2. The elements observed were sodium, oxygen, silicon and carbon. Other elements, such as nitrogen, were not observed, although nitrogen exists in the raw material of the samples as a nitrate.The binding energies were calibrated by the C( 1s) line at 284.6 eV; the C( 1s) line is attributed to a hydrocarbon in the adhesive tape holding the samples. The C( 1s) line is a singlet and the C(1s) line of COi- (at ca. 289 eV14) was not observed even in the spectrum for Na( 1-1)-SL without pretreatment. The Na/Si molar ratio was calculated from the area of the Na( 1s) and Si(2p) lines using the atomic sensitivity factors.14 When the samples were pretreated at 970 K, no significant change in these results was observed. 'The value of the Auger parameter plus the photon energy, a + hv ( = the kinetic energy of the Auger line plus the binding energy of the photoelectron line) is a good parameter for- identifying the chemical components such as sodium because the value is independent of the static charge correction of the energy va1~es.l~ The value, a+ hv, for sodium in Na( 1.1)-SL or Na(0.4)-SL is 2062.0 eV; the value is almost the same as that of sodium in sodium chloride,14 suggesting that sodium in silicalite occurs as the monovalent ion.Fig. 1 shows the change in the Na/Si molar ratio of the sample surface with argon-90 Adsorption on Sodium-modijied Silicalite i 3700 I I I 4000 3800 3600 3200 3000 wavenumbedcm Fig. 2.1.r. absorption band of silanol groups in silicalite samples pretreated at 770 K (background- subtracted spectra).(a) Na( 1.1)-SL, (6) Na(0.4)-SL, (c) Na(O.01)-SL. ion sputtering. The sputter rate is estimated to be ca. 10 nm min-l.14 As shown in the figure, the Na/Si ratio of Na( 1.1)-SL in the subsurface layers is discernibly larger than that of the outer layer, but in the case of Na(0.4)-SL the ratio is largest at the external surface of the sample and it decreases appreciably in the subsurface layers. The mean values of the Na/Si ratio are 0.030 for Na( I. 1)-SL and 0.01 1 for Na(0.4)-SL, calculated from the composition (from chemical analyses). The value for Na( 1. I)-SL agrees with the ESCA data quite well; consequently, sodium ions are distributed in Na(l.1)-SL almost uniformly. On the other hand, sodium ions on Na(0.4)-SL are significantly enriched in the surface layer.The high sodium concentration of the external surface of Na(0.4)-SL is probably due to condensation of sodium-containing water coming from inside by evaporation on the outer surface during the reflux for sodium leaching from Na( I . l)-SL. Infrared Spectrum of Silanol Groups in Silicalite Water desorption from the silicalite samples (table 1) strongly suggests that silanol groups exist in the samples pretreated at 770 K. In order to obtain information on silanol groups, i.r. measurements were carried out and the results are shown in fig. 2 (background-subtracted spectra ; we used the spectra of the samples pretreated at 970 K as background spectra because no silanol band was observed in the spectra). In the range 3400-3800 cm-l, the silicalite samples pretreated at 770 K showed a weak absorption band, which can be assigned to the silanol stretching vibration,15-17 but the samples pretreated at 970 K did not show any silanol band.The absorption band of the silanol groups was present at 3700 f 20 cm-' for Na( 1 .I)-SL, 3640 f 20 cm-' for Na(0.4)-SL and 3720 20 cm-l for Na(O.01)-SL.Y. Matsumura, K. Hashimoto and S. Yoshida 91 Table 3. Adsorption of carbon dioxide and ammonia on silicalite samples at room temperature NH3 - co2 pretreatment irreversible irreversible temperature uptakea adsorptionb uptakea adsorption sample /K /mmol g-' /mmol g-' /mmol g-' /mmol g-' Na( 1.1)-SL 770 0.04 (0.03) 0.0 1 0.37 (0.17) 0.20 Na( 1 .I)-SL 970 0.05 (0.04) 0.0 1 0.27 (0.10) 0.17 Na(0.4)-SL 770 0.02 (0.02) 0 0.22 (0.10) 0.12 Na(0.4)-SL 970 0.02 (0.02) 0 0.16 (0.09) 0.07 Na(O.01)-SL 770 0.01 (0.01) 0 0.12 (0.08) 0.04 Na( 0.0 1 )-SL 970 0.01 (0.01) 0 0.08 (0.05) 0.03 a .4mount of adsorption at 1 Torr.In parentheses, the result of the second adsorption is described. Calculated amount of irreversibly adsorbed molecules. Adsorption of Carbon Dioxide and Ammonia, and T.P.D. of Ammonia Adsorption of acidic and basic gases is often used for determination of acid-base sites.18-20 We adopted carbon dioxide and ammonia as the adsorption gases because these molecules are small enough to diffuse easily into the silicalite framework at room temperature. The results of the adsorption of carbon dioxide and ammonia on the silicalite samples are presented in table 3 ; the number of irreversibly adsorbed molecules saturates at pressures below 1 Torr.The amount of carbon dioxide irreversibly adsorbed at room temperature is related to the number of basic sites in a ample.^^^^^ Since irreversible adsorption of carbon dioxide was not observed on Na(O.01)-SL, there can be no basic sites on Na(O.01)-SL. On the other hand, Na( 1.1)-SL adsorbs carbon dioxide much more than Na(O.01)-SL does, but the amount of irreversibly adsorbed carbon dioxide is very small (0.01 mmol g-') compared with the amount of sodium ions in Na(l.1)-SL (ca. 0.5 mmol g-'). The same tendency was observed for the adsorption of carbon dioxide on Na(0.4)-SL containing ca. 0.2 mmol g-l of sodium ions. Thus, the number of basic sites in sodium- modified silicalite is very small. Quite a large amount of ammonia was irreversibly adsorbed on the samples.The amount increased with sodium content of the sample, but in the case of Na(O.01)-SL, the amount was ca. 10 times as much as the sodium content of the sample (ca. 0.004 mmol g-l). When the samples were pretreated at 970 K, the amount adsorbed was decreased compared with that for the samples pretreated at 770 K. The differences in the amount adsorbed with the pretreatment temperature were constant at ammonia pressures above 1 Torr. Ammonia t.p.d. data provide information on the binding strength of ammonia to adsorption site^.^,^^,'' As shown in fig. 3, the t.p.d. profiles for Na(O.01)-SL have a peak at 350 K and a small shoulder at 42M70 K, and a peak at 470 K and a shoulder at 370-420 K for sodium-rich silicalite samples, Na( 1 .1)-SL and Na(0.4)-SL, respectively. When Na( 1.1)-SL and Na(0.4)-SL were pretreated at 970 K, the shoulder at 370-420 K was appreciably smaller than that for these samples pretreated at 770 K. In fig. 3, the t.p.d. profile for ZSM-5-type zeolite measured by Topsare et al. is also shown for comparison.17 There are three desorption peaks at ca. 370 K (a-state), 460 K (P-state) and 690 K (y-state) in the profile. T o p s ~ e et al. concluded that the y-state (activation energy of ammonia desorption = 162.3 kJ mol-') is related to strong acid sites and that the /3-state (96.7 kJ mol-l) is related to adsorption on sodium-ion sites in ZSM-5; they also related the a-state (84.6 kJ mol-l) to sites located on the external surface or to some92 Adsorption on Sodium- m odiJied Silicalite 770K ( b ) 970 K 770K 300 500 T/K 700 900 Fig.3. T.p.d. profiles of ammonia from silicalite. Conditions: sample weight, 0.100 g; heating rate, 10 Kmin-'; He flow rate, 0.020 dm3min-l. The profile for ZSM-5 is taken from ref. (17). (a) Na( 1.1)-SL, (b) Na(0.4)-SL, (c) Na(O.01)-SL. type of extraneous material, or to interaction of ammonia molecules with surface oxide or hydroxyl groups by non-specific hydrogen bonding. l7 Since their experimental conditions were similar to those of our experiment17 and the crystal structure of ZSM- 5 is identical to that of silicalite,2 their assignment is helpful for us to assign the adsorption state of ammonia molecules on the silicalite. Anderson et al. reported that silicalite shows low a ~ i d i t y .~ The acidity of silicalite is due to the existence of a small number of aluminium ions in the silicalite as impurities because the number of acid sites in ZSM-5 is proportional to the number of aluminium ions22 and silicalite is an aluminium-deficient ZSM-5.l Our silicalite samples also contain 0.01 wt% aluminium. However, the absence of the y-state in our samples shows that the number of strong acid sites is very small if they exist at all. Infrared Spectrum of Adsorbed Pyridine The spectra of pyridine adsorbed on Na(l.1)-SL are shown in fig. 4 (background- subtracted spectra); in the range of 1400 to 1700 cm-' these spectra did not change significantly with the pretreatment temperature of the sample. As shown in fig. 4, we can see two bands at 1597 and 1443 cm-l, even after evacuation at 420 or 470 K for 1 h.Those bands are generally attributed to physisorbed pyridine ; however, pyridine physisorbed on silica or silica-alumina is removed by evacuation at 420 K.23 This result shows that pyridine on silicalite is stabilized more than that on silica or silica-alumina, although the adsorption sites are not acidic in nature. When Na( l.l)-SL was pretreated at 770K, the silanol band at 3700cm-I was observed prior to the adsorption, but itY. Matsumura, K. Hashimoto and S. Yoshida 93 1597 I I 1700 1600 1500 1400 wavenumber/cm -' Fig. 4.1.r. spectra of pyridine adsorbed on Na( 1.1)-SL (background-subtracted spectra) : (a) after exposure to pyridine followed by evacuation at 420 K, (b) evaculation at 470 K.almost disappeared after the adsorption. After evacuation at 520 K for 1 h, no absorption band due to pyridine was observed. In the case of Na(O.01)-SL pretreated at 770 or 970 K, the absorption bands due to physisorbed pyridine were observed after evacuation at room temperature, but they were not observed after evacuation at 420 K. Discussion Interaction between Sodium Ions and Silanol Groups in Silicalite Since silicalite is deficient in aluminium, the counter-ions of almost all the sodium ions in silicalite are not A10, anions in the framework (the usual anions as counter-ions for metal ions in zeolites). Therefore, in order to maintain electrostatic neutrality of silicalite, counter-ions such as SiO- would be needed or sodium ions would exist as sodium oxides such as sodium monoxide.It is known that sodium monoxide is an unstable compound and sodium peroxide reacts readily with carbon dioxide in air and is converted to sodium carbonate.24 The ESCA results for sodium-modified silicali te suggest that Cot- species do not exist, even in Na(l.1)-SL allowed to stand for a long time in contact with air. Moreover, carbon dioxide is scarcely adsorbed on sodium- modified silicalite even if the sample is pretreated above 670 K, at which temperature sodium carbonate begins to lose carbon Thus, the possibility of the existence of sodium oxides in silicalite is rather low and sodium ions in silicalite probably exist as SiO-Na+. The formation of SiO-Na+ is proposed to occur during the crystallization pro~ess.~ In other words, while the condensation of silanol groups gives Si-0-Si oxygen bridges, existence of sodium ions in the silicalite framework hinders the formation of oxygen bridges and gives SiO-Na+ and SiOH.Consequently, the existence of SiO-Na+ must accompany the existence of silanol in the silicalite framework. In fact, the silicalite samples pretreated at 770 K contain silanol groups. The amount of the silanol groups is 0.14 mmol g-l for Na(l.1)-SL, 0.08 for Na(0.4)-SL and 0.06 for94 Adsorption on Sodium-modiJied Silicalite Na(O.01)-SL estimated from the amount of water desorbed by heating (table 1). This finding suggests that the number of silanol groups in the samples increases with the sodium content of silicalite, although most silanol groups are removed by heating. Woolery et al.reported that internal silanol groups present in silicalite as well as terminal silanol, and internal silanol groups give an i.r. absorption band at 3740 cm-'; the band of the terminal silanol may overlap the band of the internal silano1.25 Thus, both internal and terminal silanol groups on Na(O.01)-SL are probably responsible for the i.r. band at ca. 3720 cm-l. On the other hand, the i.r. band of silanol in Na(0.4)-SL is lower by ca. 80cm-' than that of Na(O.01)-SL. Misono et al. reported that the i.r. absorption band of silanol is shifted to lower wavenumbers by coordination with an alkali-metal cation,26 and it is probable that the band shift of silanol in Na(0.4)-SL is due to coordination of sodium ions to silanol groups. Since sodium ions in Na(0.4)-SL are mainly present on external surface [this is the decisive difference of Na (0.4)-SL from Na(l.1)-SL], it is assumed that the coordination of sodium ions takes place on the external surface of silicalite.However, the silanol band of Na (1 .1)-SL is at ca. 3700 cm-I and Woolery et al. also reported that sodium-containing silicalite does not exhibit such an i.r. band shift as Na(0.4)-SL.25 This suggests that on the internal surface the interaction between sodium ions and silanol groups is quite weak; sodium ions on the internal surface may be stabilized mainly by framework atoms surrounding the sodium ions. Thus, the i.r. band of the internal silanol groups in sodium-modified silicalite does not shift obviously, and the absorption band of the silanol on the external surface may overlap the low-wavenumber region of the band at ca.3700 cm-l in the case of Na( 1.1)- SL in which sodium ions are distributed on its surface almost uniformly. Adsorption of Ammonia The results of ammonia adsorption and the t.p.d. measurements show that Na(l.1)-SL and Na(0.4)-SL adsorb ammonia molecules strongly compared with Na(O.0 1)-SL. These results strongly suggest that sodium ions play an important role in the adsorption. The t.p.d. profiles for Na(l.1)-SL and Na(0.4)-SL have a desorption peak at 470 K, and the desorption peaks do not change with the pretreatment temperature of the samples (see fig. 3). The results suggest that silanol groups are not responsible for the peak because the silanol band was not observed in the i.r.spectra of the samples pretreated at 970 K. Since the desorption peak area increases with the amount of sodium ions in the silicalite samples, the desorption peak is probably due to interaction of ammonia with sodium ions. Moreover, the desorption temperature is almost the same as that of the 8-adsorption state in the t.p.d. profile for ZSM-5 reported by Topsse et al. ; p-state is assigned to ammonia adsorbed on sodium ions.17 The sodium concentration on the external surface of Na(0.4)-SL is discernibly larger than that of Na (1 .I)-SL, but that of the subsurface layers is smaller. The desorption peak area of the p-state for Na(0.4)-SL is significantly smaller than that for Na (1.1)-SL, and the difference is related to the difference in the mean concentration of sodium ions in both samples. Thus, the desorption peak is due to adsorption of ammonia on sodium ions mainly located on the internal surface of silicalite.Ammonia molecules are known to be adsorbed on silanol groups via hydrogen bonding.27 Since the results of i.r. measurements (fig. 2) show the existence of silanol groups in the silicalite samples pretreated at 770 K, the silicalite samples pretreated at 770 K probably contain silanol sites which can adsorb ammonia molecules. Actually, the total amount of adsorbed ammonia on the samples pretreated at 970 K is smaller than that for the samples pretreated at 770 K (see table 3), but the difference is ca. 30% of the total amount of adsorbed ammonia. Thus, a small number of the ammonia molecules adsorbed on silicalite interact with the silanol groups. Ammonia molecules adsorbed on the silanol groups in silica are removed byY .Matsumura, K. Hashirnoto and S. Yoshida 95 evacuation at room te~perature.~’ However, the amount of ammonia irreversibly adsorbed on sodium-modified silicalite changes significantly with the pretreatment temperature of the samples (fig. 3 and table 3); this suggests that some of the silanol groups in sodium-modified silicalite interact with ammonia molecules quite strongly and adsorb ammonia irreversibly at room temperature. The t.p.d. profiles for Na( 1 .1)-SL and Na(0.4)-SL show that the area of the shoulder at 370-420 K changes with the pretreatment temperature of the samples; this shoulder is quite small when the samples are pretreated at 970 K.The activation energy for desorption of these ammonia molecules is estimated to be cu. 90 kJ mol-’ on the basis of the results of Topsse et u1.l’ The activation energy suggests that ammonia is chemisorbed on the silanol. Thus the shoulder is attributed to ammonia desorbed from the weakly acidic silanol sites. The number of these silanol sites (referred to as NOH hereafter) can be estimated from the difference in the amounts of irreversibly adsorbed ammonia between the samples pretreated at 770 and 970 K (i.e. the shaded parts in fig. 3). The amount is 0.03 mmol g-’ for Na( 1.1)-SL, 0.05 mmol g-’ for Na(0.4)-SL and 0.01 mmol g-l for Na(O.01)-SL based on the results presented in table 3 [the value of NOH for Na(O.01)-SL may be overestimated because the amount of ammonia desorbed from Na(O.01)-SL observed by the t.p.d.measurements scarcely changes with the pretreat- ment temperature of the sample]. Since NOH for Na(O.01)-SL is appreciably smaller than that for sodium-rich silicali tes, sodium ions in silicalite enhance the interaction between silanol groups and ammonia. However, NOH for Na(0.4)-SL is larger than that for Na( 1.1)-SL containing a large amount of sodium ions compared with Na(0.4)-SL. On the basis of quantum- chemical calculations, it is confirmed that coordination of cations to silanol causes the law electron density of the hydrogen atoms in the silano1.28’29 Taking into account this result, the silanol groups interacting with sodium ions are weakly acidic. Probably, these silanol sites exist on the external surface of sodium-modified silicalite, as described in the previous section, and Na(0.4)-SL contains many more of these sites than Na( 1.1)-SL. Thus, we believe that the silanol sites on the external surface of sodium-modified silicalite are weakly acidic and that these sites are responsible for the shoulder at 370-420 K in the t.p.d. profiles. Adsorption of Pyridine In the case of pyridine adsorption, Na(l.1)-SL adsorbs the molecules more strongly than Na(O.01)-SL, and the pyridine adsorbed on Na(l.1)-SL is similar to a physisorbed species even after evacuation at 470 K. These results strongly suggest that sodium ions on silicalite participate in the adsorption of pyridine and that the adsorption is not due to acid-base interaction. Since the absorption band of the silanol groups on Na( l.l)-SL is reduced by adsorption of pyridine, some of the pyridine molecules probably form hydrogen bonds with the silanol groups.However, the number of pyridine molecules adsorbed on the silanol groups is quite small, because the intensity of the i.r. absorption bands for adsorbed pyridine scarcely changed with the pretreatment temperature of the samples. Adsorption of Carbon Dioxide Adsorption of carbon dioxide on silicalite is also enhanced by the existence of sodium ions. Since the amount adsorbed increases with the sodium content of the samples and not with the concentration of external sodium ions on the samples, sodium ions on the internal surface of silicalite are assumed to be mainly responsible for the adsorption. Probably, silanol groups are not concerned in the adsorption of carbon dioxide because 4 FAR I96 Adsorption on Sodium-modijed Silicalite the amount of adsorbed carbon dioxide hardly changed with the pretreatment temperature of the samples.Conclusions The number of acid-base sites in the silicalite containing a small amount of sodium ions is very small. In the case of sodium-modified silicalite, which was expected to have basic properties, as well as sodium-containing silica, the number of basic sites is very small. On the other hand, sodium-modified silicalite adsorbs ammonia and pyridine stably compared with non-modified silicalites. The main adsorption sites for these molecules are sodium ions in silicalite. The silanol groups on silicalite are also adsorption sites, but they form only a small part of the total adsorption sites.The silanol groups remained in the silicalite pretreated at 770 K, while most silanol groups were removed by pretreatment at 970 K. The silanol groups on the external surface of sodium-modified silicalite are probably weakly acidic, and no appreciable number of strong acid sites exists on the silicalite. The weak acidity of the silanol is supposed to be due to the interaction between the silanol groups and sodium ions on external surface. References 1 C. A. Fyfe, G. C. Gobbi, J. Klinowski, J. M. Thomas and S. Ramdas, Nature (London), 1982, 296, 2 D. H. Olson, W. 0. Haag and R. M. Lago, J. Catal., 1980, 61, 390. 3 US. Patent, 1977, 4061724. 4 S. G. Fegan and B. M. Lowe, J. Chem. SOC., Chem. Commun., 1984, 437.5 S. Malinowski and S. Szczepanska, J. Catal., 1963, 2, 203, 310. 6 T. Tagawa, T. Hattori and Y. Murakami, J. Catal., 1982, 75, 56, 7 J. R. Anderson, K. Foger, T. Mole, R. A. Rajadhyaksha and J. V. Sanders, J. Catal., 1979, 58, 114. 8 Y. Matsumura, K. Hashimoto and S. Yoshida, J. Chem. SOC., Chem. Commun., 1984, 1447. 9 Y. Matsumura, K. Hashimoto and S. Yoshida, J. Catal., 1986, 100, 392. 530. 10 H. Niiyama and E. Echigoya, Bull. Chem. SOC. Jpn, 1971, 44, 1739. I 1 C. L. Kibby and W. K. Hall, J. Catal., 1973, 9, 144. 12 T. Yashima, H. Suzuki and N. Hara, J. Catal., 1974, 33, 486. 13 Japan Industrial Standard, 1965, R3 10 1. 14 C. D. Wagner, W. M. Riggs, L. E. Davis and J. F. Moulder, in Handbook of X-Ray Photoelectron 15 A. Ison and R. J. Gorte, J. Catal., 1984, 89, 150. 16 J. C. Vedrine, A. Auroux, V. Bolis, P. Dejaifve, C. Naccache, P. Wierzchowski, E. G. Derouane, J. B. Nagy, J.-P. Gilson, J. H. C. van Hooff, J. P. van den Berg and J. Wolthuizen, J. Catal., 1979, 59, 248. Spectroscopy, ed. G. E. Muilenberg (Perkin-Elmer Co., 1978). 17 N.-Y. Topsse, K. Pedersen and E. G. Derouane, J. Catal., 1981, 70, 41. 18 K. Tanabe, in Solid Acids and Bases (Kodansha-Academic Press, Tokyo-New York, 1970). 19 G.-W. Wang, H. Hattori and K. Tanabe, Bull. Chem. Soc. Jpn, 1983,56, 2407. 20 W. Ueda, T. Yokoyama, Y. Moro-oka and T. Ikawa, Chem. Lett., 1985, 1059. 21 C. V. Hidalgo, H. Itoh, T. Hattori, M. Niwa and Y. Murakami, J. Catal., 1984, 85, 362. 22 P. A. Jacobs, J. Phys. Chem., 1982,86, 3050. 23 E. P. Parry, J. Catal., 1963, 2, 371. 24 The Merch Index (Merch and Co., 1960). 25 G. L. Woolery, L. B. Alemany, R. M. Dessau and A. W. Chester, Zeolites, 1986, 6, 14. 26 M. Misono, T. Takizawa and Y. Yoneda, J. Catal., 1978, 52, 397. 27 L. H. Little, in Infrared Spectra of Adsorbed Species (Academic Press, New York, 1966). 28 W. Grabowski, M. Misono and Y. Yoneda, J. Catal., 1980, 61, 103. 29 H. Kawakami, S. Yoshida and T. Yonezawa, J. Chem. SOC., Faraday Trans. 2, 1984, 80, 205. Paper 612368; Received 9th December, 1986

ISSN:0300-9599    

DOI:10.1039/F19888400087

出版商:RSC    

年代:1988

数据来源: RSC

摘要:

J. Chem. Soc., Faraday Trans. I, 1988, 84(1), 97-109 Mechanism of Electrohydrodimerization of Cyclohex-2-en- 1-one on Mercury from Aqueous Solutions Part 1.-Results obtained in the Absence of Surfactants M. Yolanda Duarte Department of Chemistry, Nordeste University, Corrientes, Argentina Corrado Malanga and Lamberto Nucci Department of Chemistry and Industrial Chemistry, Pisa University, Pisa, Italy M. Luisa Foresti and Roland0 Guidelli" Department of Chemistry, Florence University, Florence, Italy The diffusion-controlled one-electron wave (wave I) due to cyclohex-2-en- 1 - one (R) electroreduction on mercury from aqueous solutions in the range 2.5 < pH < 4 is shown to be consistent with the following mechanism : R+ HA+ RH+ +A- RH+ + e $ RH' ( 1 4 ( 1 4 2RH' r - R2H2 ( 1 c) where HA is a proton donor and the rate-determining coupling step (1 c), denoted by rds, proceeds in the adsorbed state.Double-potential-step chronocoulometric measurements indicate that R adsorption lies below the limits of sensitivity of the method, whereas R2H2 adsorption is appreciable at all potentials positive to - 1.4 V us. SCE. The limiting current of wave I in the range 4 < pH < 6 is controlled by the protonation step (1 b), which is shown to proceed both heterogeneously and homogeneously. The con- tribution of the various proton donors, including adsorbed water, is pointed out. The diffusion-controlled one-electron wave (wave 11), which develops at the expense of wave I at pH > 7, is shown to be consistent with the R + e+ R'- (7 4 following mechanism : R'- +HA e RH' +A- (7 4 2RH' rx R,H2 (7 c) for 7 < pH < 10, and with the mechanism R+ e s R'- (lOa> 2R'- Rip (lob) ( W for pH > 11.For mechanisms (7) and (10) the rate-determining coupling step is homogeneous. The mixtures of isomeric forms of the hydrodimer R2H2 obtained at pH 5 uia the heterogeneous coupling step (1 c) and that obtained at pH 9 uia the homogeneous coupling step (7c) have practically the same composition. Ri- + 2H20 e R2H2 + 20H- 97 4-298 Electrohydrodirnerization of Cyclohex-2-en- 1 -one A systematic investigation of the mechanism of electrohydrodimerization (EHD) of diactivated alkenes on mercury from aqueous media has been carried out in this laboratory.'" As distinct from a number of diactivated alkenes which undergo EHD only in the presence of strong surfactants, chalcone undergoes EHD in aqueous media both in the presence4 and in the absence5 of strong surfactants such as Triton X-100.While chalcone EHD in the presence of Triton X-100 takes place in the non-adsorbed state, that in the absence of strong surfactants takes place in the adsorbed state and yields a different mixture of hydrodimer isomers. To verify whether this behaviour is shared by other a, B-unsaturated carbonyls, we have thoroughly investigated the mechanism of EHD of cyclohex-2-en- 1 -one by direct current (d.c.) polarography and chronocoulometry. This paper reports the results obtained in the absence of surfactants, whereas Part 2 deals with cyclohex-2-en-1-one EHD in the presence of Triton X-100.The electroreduction of cyclohex-2-en- 1 -one (henceforth denoted by R) has been the subject of several studies in aqueous,'. ' hydroalcoholics and non-aqueous lo The studies in aqueous' and hydroalcoholic' media agree in ascribing the diffusion- controlled one-electron wave observed at pH < 4 (wave I) to electronation of the protonated form RH+ of the reactant followed by coupling of the RH' radicals and hydrodimer formation. The one-electron wave (wave 11), which develops progressively at the expense of wave I and the pH is gradually increased above 4, is instead ascribed to direct electronation of the reactant R followed by further steps yielding the h ydrodimer . Experimental All chemicals were analytical reagent grade supplied by either Merck or Fluka, and were used without further purification.All solutions were prepared from triply distilled water treated with active charcoal. Mercury was purified by a wet process followed by three distillations. All potentials were measured against a saturated calomel electrode (SCE). Polarographic measurements were carried out at 25 f 0.25 "C with an Amel model 473 polarograph. In all measurements the drop time, t,, was kept equal to 2 s. Double- potential-step chronocoulometric measurements were carried out with a computerized apparatus described in ref. (1 l), using a pressurized hanging-mercury-drop electrode12 which was renewed under computer control. All measurements were carried out in buffered solutions in which the buffer concentration was no less than ten times the reactant concentration. The Na+ concentration in all solutions was kept constant at 1 mol dm-3 with NaCl. In this way the bulk concentration of the ions which predominate in the diffuse layer at the extremely negative potentials at which EHD takes place is kept constant, and the extent of ion pairing of the intermediate anion radicals is also kept constant.Results Cyclohex-2-en-1-one EHD was investigated over the pH range from 1.5 to 13. The pH was controlled with citric buffers from 1.5 to 3, formic buffers from 2.5 to 4, acetic buffers from 3 to 5, phosphate buffers from 5 to 7, borate buffers from 8 to 10 and with NaOH from 11 to 13. Higher pH values cannot be explored, since the height of the R- wave decreases in time because of reactant decomposition in the bulk.It is possible that at high pH values the ethylenic double bond of R undergoes hydration with formation of a hydroxy derivative, as observed with other a, B-unsaturated carbonyl comp~unds.'~~~* Polarographic Measurements The diffusion-controlled one-electron wave (wave I) observed in acidic solutions decreases in height as the pH is gradually increased above 4, while a more negative waveM. Y. Duarte et al. -1.30- -1.28 -1.26 - 1.24 99 - - - - - - - 1 I - I - 1 . 4 -1.3 w z ? - .I 4N 4 -1.1 # P r’ 1 /. b / / / / / / 1‘ 4 -1.7 w u m -1.6 2 > =+ 4 -1.5 I I I I I 4 6 0 10 12 PH Fig. 1. (a) Ei us. pH plot for mol dmP3 formic buffer (a), mol dm-3 phosphate buffer (m); the dashed line has a -60 mV slope. (b) Er us. pH plot for loP4 mol dmP3 R reduction from mol dm-3 borate buffer; the dashed line has a -40 mV slope.(c) Er us. pH plot for loP4 mol dmP3 R reduction from loP3 to lo-’ mol dm-3 NaOH solutions. mol dmP3 R reduction from mol dm-3 acetic buffer (0) and 5 x100 Elect ro h y dr odimer iza t ion of Cy clo hex- 2- en- 1 -one I f i I 0 2 4 6 8 [H']f/10-3 mol* dm-) Fig. 3. The solid curve (a) is a plot of </(fd - () us. [H+]i for phosphate buffer with [KH,PO,] = C,, = 5 x mol dm-3 R reduction from a rnol dm-3. For curves (b) and (c) see the text. 0 0.1 0.2 0.3 ciA /mola dm-* Fig. 4. </(id-<) us. dHA plot for mol dm-3 R reduction from a pH 5.46 phosphate buffer (HA = KH,PO,). (wave 11) develops and increases at the expense of wave I. Ultimately, at pH > 7 wave I vanishes and only the one-electron wave (wave 11) is left.Wave I At pH < 2.5 wave I splits into an adsorption prewave and in the corresponding 'normal' wave, provided the reactant concentration C: is > 2 x lo-* mol dm-3. The limiting height of the adsorption prewave points to a maximum surface coverage by the adsorbed EHD product of 3 x 10-l' mol as already observed by Denney and Mooney.6 AtM . Y. Duarte et al. 101 EIV us. SCE Fig. 5. (a) Plot of Aa'(E, + - 1.75 V) us. E, as obtained from a solution of 0.01 mol dm-3 NaOH + 0.99 mol dm-3 NaCl both in the absence and in the presence of mol dm-3 R; the polarogram in the presence of the reactant is also shown (a'). (b) Plot of AaM( - 1.75 V --* E,) us. Ef as obtained from a solution of 0.01 mol dmV3 NaOH + 0.99 mol dm-3 NaCl+ mol dm-3 R.(c) Plot of Ac'(-0.92 V + Ef) us. E, as obtained from a solution of 1 mol dm-3 mol dm-3 R, whose polarogram is shown in (c'). HCl+ pH values > 2.5 no adsorption prewave is observed. Under these conditions the plot of the half-wave potential Ei of wave I vs. pH at constant C$ has a slope of ca. -60 mV [see curve (a) in fig. 13. As the reactant concentration is gradually increased at constant pH, wave I shifts towards more positive potentials. The rate aE{/a log Cg of the half- wave potential shift increases with increasing log C:, attaining a maximum limiting value of 30 mV at Cg > mol dm-3 [see curve (a) in fig. 21. At pH values low enough to ensure a diffusion limiting current id for wave I, Ei is independent of buffer concentration. Thus, an increase in buffer concentration by almost two orders of magnitude under otherwise identical experimental conditions leaves Ei practically unaltered. As the limiting current 6 of wave I starts decreasing with respect to its diffusion- limiting value id because of an increase in pH above 4, it becomes dependent on the buffer concentration.Thus 6 decreases not only with a decrease in H30+ concentration at constant concentration C,, of the acid component HA of the buffer (see fig. 3), but also with a decrease in C,, at constant pH (see fig. 4). The more 6 decreases with respect to id the more the plateau of wave I deviates from horizontal behaviour, causing ( to become progressively more dependent on potential. In plotting fig. 3 and 4, the arbitary procedure of measuring il at a potential E = El - 100 mV was adopted.Wave II In well buffered solutions of pH ranging from 8 to 10, when the limiting current of wave I1 has already attained its diffusion-limiting value id, the half-wave potential Ef of this wave shifts towards more negative values by 40 mV per each unitary increment of pH [see curve (b) in fig. 11. Then, as pH is increased above 1 1 , Eil tends to become102 Electrohydrodirnerization of Cyclohex-2-en- 1 -one I I I o o o o o o o o o o o o o o o ooooo 0 0 O O O O O t/ms Fig. 6. Q(t) us. t curve as obtained from a solution of 0.01 mol dm-3 NaOH + 0.99 mol dm-3 NaCl+ mol dm-3 R upon stepping the applied potential from E, = - 1.75 V to Ef = - 1.00 V at t = 0 and then backward at t = z = 50 ms. independent of pH [curve (c) in fig.11. Provided that over the pH range 7-10 the solution is well buffered (buffer concentration 2 10 Cz), El' is independent of the buffer concentration. In well buffered solutions of pH 7-10 and at NaOH concentrations 3 lOCg, El' at constant pH shifts towards more positive values by ca. 20mV per unitary increment of log C;lt [see curves (b) and (c) in fig. 21. Chronocoulometric Measurements The presence of an adsorption prewave in acidic solutions denotes a notable adsorption of the electrode-reaction product, at least over the potential range covered by wave I at these low pH values. Single and double-potential-step chronocoulometric measurements were performed to determine the potential dependence of product adsorption and to ascertain whether the reactant R is also appreciably adsorbed.Curve (a) in fig. 5 shows the changes AaM in capacitive charge involved in a series of potential jumps from a variable initial potential Ei to a fixed final potential Ef = - 1.75 V us. SCE in a pH 12 solution free from reactant, as plotted against Ei. The slope of this plot is almost constant, and corresponds to the differential capacity on mercury at potentials negative to the P.Z.C. in the absence of specific adsorption (ca. 17 pF cm-2). The same single- potential-step chronocoulometric measurements were then performed in the presence of mol dm-3 reactant, which under these conditions yields wave I1 with Ei* = - 1.61 V [see fig. 5(a)]. The resulting plots of the time-dependent charge Q(t) flowing at Ef us. the square root ti of the electrolysis time are linear, in agreement with the fact that Ef lies $long the diffusion-controlled plateau of wave 11.The intercepts of these plots on the ti = 0 axis at different values of the initial potential Ei yield AoM (Ei) at the interphase between mercury and the reactant solution. The resulting AaM us. Ei plot practically coincides with curve (a), indicating no detectable adsorption of R. Double-potential-step chronocoulometric measurements carried out under the same experimental conditions show that the charge Q(t-z) following the backward potential jump at t = z fromM. Y. Duarte et al. 103 Ef = - 1.75 V to Ei decreases abruptly and then remains perfectly constant in time. This indicates that the primary reaction product is converted into the electroinactive hydrodimer R2H2 so rapidly as not to give rise to a detectable reoxidation current to the reactant.To detect product adsorption, double-potential-step chronocoulometric measure- ments were performed in which the mercury drop was kept at a fixed initial potential Ei = - 1.75 V along the plateau of wave I1 for the majority of drop life. Towards the end of drop life the applied potential was then stepped to a more positive value Ef for a time period z = 50 ms during which adsorption equilibrium of the hydrodimeric product at Ef was established (see fig. 6). Note that the diffusion layer produced by R2H, diffusion towards the bulk at Ei = - 1.75 V before the potential jump is much thicker than that produced by any R2H2 diffusion towards the electrode surface following its adsorption at Ef during z.Hence we can safely assume that, as far as the diffusion and adsorption of R2H2 at Ef is concerned, the electrode behaves as though it were immersed in a solution of R2H, of bulk concentration C:. After z, the potential was stepped back from Ef to Ei = - 1.75 V. As shown in fig. 6, the charge Q(t - z) following the backward potential jump first decreases abruptly by an amount equal to the capacitive charge AaM at the interphase between mercury and a mol dm-3 solution of R2H2, and then more slowly on account of R electroreduction at - 1.75 V. The quantity AaM(Ef) corresponding to the various Ef values was obtained by graphical extrapolation to t = z. Such a procedure involves an error no greater than 0.5 pC cm-2.The resulting plot of AaM(Ef) us. Ef is shown in fig. 5(6). This curve has the typical shape of charge us. potential curves of neutral organic surfactants. The intersection point of curves (a) and (h) locates the potential Em,, = -0.6 V of maximum adsorption for the product R2H,, whereas the merging of curves (a) and (b) at E < - 1.4 V denotes the lack of detectable R,H2 adsorption at these negative potentials. The differential capacity of an R2H2 solution in contact with mercury at Emax amounts to ca. 7 pF cm-2. The same value was obtained by performing double-potential-step chronocoulometric measurements anal- ogous to those in fig. 6 with a lop3 rnol dm-3 solution of R in 1 mol dmP3 HC1, upon jumping from an initial potential Ei = -0.92 V along the plateau of the adsorption prewave [see the profile of the whole wave I in fig.5 (c')] to more positive final potentials. The resulting AaM(Ef) us. Ef plot is shown in fig. 5 (c). The potential Emax = -0.6 V has the property that a potential jump from Emax to a final potential negative enough to exclude R2H2 adsorption (say < - 1.4 V) involves a capacitive charge AaM which does not depend upon the concentration of R2H2 around the electrode. By performing a double-potential jump from Ei = Emax to t3, = - 1.75 V in a pH 12 solution of mol dmP3 R, the capacitive change in charge, Q(z + dt) - Q(z - dt), following the backward potential jump Ef -+ Emax is practically e,qual to the cfiarge, ca. 17.2 pC cmp2, obtained from the intercept of the Q(t < z) us.ts plot on the ti = 0 axis, where Q(t < z) is the charge flowing at Ef before the backward potential jump. This confirms unequivocally the absence of detectable reactant adsorption, at least at Cz < mol dmP3. Higher reactant concentrations were not investigated pecause of the rapid decrease in the accuracy of the extrapolation of Q(t < z) us. tz plots with increasing CE. Macroscale Electrolysis Macroscale electrolysis was carried out in a conventional cell described in ref. (5). Cyclohex-2-en-1-one EHD was carried out both in a pH 9 borate buffer (0.16 mol dmp3 Na,B,O, + 0.68 mol dmd3 NaCl) at - 1.6 V, where the coupling reaction takes place homogeneously, and in a pH 5 acetic buffer (1 mol dm-3 CH,CO,H+ 1 mol dm-3 CH,CO,Na) at - 1.4 V, where the coupling reaction takes place mainly in the adsorbed state.During electrolysis in the borate buffer solution the pH value was controlled by104 Electrohydrodimerization of Cyclohex-2-en- 1 -one a glass electrode immersed in the cathodic compartment ; whenever pH increased beyond 9.2-9.3, the original value was restored by adding a few drops of 6 mol dm-3 HC1. A coulometric analysis at controlled potential yielded a value of almost unity for the number of electrons per molecule of the reactant in both buffer solutions. Electrolysis was conducted up to 91 YO conversion in the pH 9 borate buffer and up to 77% conversion in the pH 5 acetic buffer. After electrolysis, the cathode-compartment contents were extracted with ethyl acetate in a separation funnel. The organic phase was washed with water up to neutrality and dried with Na,SO,. The extracts were then stripped of the organic solvent under vacuum ( mmHgt). The oily crude product so obtained was examined by thin layer chromatography (t.1.c.) on silica and by high- performance liquid chromatography (h.1.p.c.) on a Lichrosorb Si60 column using an ethyl acetate-hexane 80/20 (v/v) mixture as eluent.This crude product turned out to consist of three different compounds and to be practically the same no matter if obtained from borate or acetic buffer. The main compound obtained by fractionating the crude product in a Lobar Lichropur Si60 column yielded the same lH-n.m.r. and i.r. spectra and the same chromatographic response as the starting crude product. This behaviour denotes the establishing of a slow equilibrium between this compound and the other isomers of the crude reaction mixture.It is possible that the attainment of this equilibrium is favoured by the silica employed for the chromatographic separations. Lr., 'H-n.m.r. and m.s. analyses were therefore carried out directly on the crude product. 1.r. spectra obtained from a liquid film of the crude product showed a band at 3460 cm-', ascribable to OH bond stretching, and a band at 1730 cm-l, ascribable to C=C and/or C=O double-bond stretching. 1.r. spectra of a solution of the crude product in CDC1, showed four distinct bands at 3650, 3600, 3460 and 3150 cm-l. A comparison with i.r. spectra of analogous compounds in CDC1,15~'6 permitted us to ascribe this behaviour to a mixture of the following hydrodimers.0 This conclusion was supported by m.s. analysis carried out by direct introduction of the raw product, which showed a spectrum identical to that reported for diketone 111.'' Combined gas chromatography-mass-spectral analysis of the raw product, carried out using a 2 m x 0.29 cm column filled with 8 YO CW20M + 2 YO KOH on 80-100 mesh Chromosorb W DMCS (CW20W) and helium as carrier gas, yielded four chro- matographic peaks having retention times of 7.12, 7.90, 11.80 and 18.02 min. The compound with 7.12 min retention time was ascribed structure I (M+ = 194, M+- H,O = 176), that with 7.90 min retention time structure I1 (M' = 194, M+-H20 = 176, M+-2H20 = 158), whereas the compounds with higher retention times were identified as degradation products of structure 111.'H-N.m.r. analysis of the crude product yielded signals at 6.0-5.4 and 3.4-2.6 ppm, which are ascribable to olefinic and alcoholic t 1 mmHg = 13.5951 x 980.665 x lo-' Pa.M. Y. Duarte et al. 105 EIVvs. SCE -1.50 -1.55 -1.60 1 I I \ h \ \ \ \ \ ~ ~~ -1.25 -1.30 -1.35 E/V us. SCE Fig. 7. (a) Plot of log [(I - i/~;)~/(i/i~)] us. E for buffer. (b) Plot of log [( 1 - i/id)/(i/i:)i] us. E for mol dmP3 R reduction from a pH 4.6 acetic mol dm-3 R reduction from a pH 8.8 borate buffer. protons, respectively, as well as signals at 2.6-0.8 ppm, which are ascribable to the cyclohexane and cyclohexene rings of structures I, I1 and 111. Upon addition of D,O to the reaction mixture, the signals due to the alcoholic protons disappeared. On the other hand DCl addition caused a slow conversion of product I into product IV: Formation of the latter compound was deduced from the disappearance of the signals of the H, and Hb.protons and the appearance of a structured doublet (5.5-5.3 ppm) and of a large multiplet (4.0 ppm) ascribable to the H, and H, protons, respectively. 'H-N.m.r.spectra of the crude product, as recorded in pyridine, showed several similarities in chemical shifts with the spectra of analogous compounds under the same experimental ~0nditions.l~ In particular, the alcoholic protons of structures I and I1 resonate at 5.0 ppm, whereas the Ha and Hb protons resonate at 6.2-5.6 ppm. On the basis of the above spectroscopic evidence, the 'H-n.m.r. spectra of the crude product allowed us to conclude that such a product consisted of compounds I and I1 in equilibrium as well as of compound I11 in a 60/40 approximate ratio.106 Electrohydrodimerization of Cyclohex-2-en- I -one The experimental observation that heterogeneous coupling at pH 5 and homogeneous coupling at pH 9 yield the same mixture of hydrodimers seems to contrast with chalcone EHD,5 which leads to different isomeric forms of the hydrodimer depending on whether the coupling reaction takes place in the adsorbed or non-adsorbed state.However, chalcone EHD in the absence of strong surfactants such as Triton X-100 takes place at an electrode surface which is almost fully covered by reactant and product. Under these conditions the neutral radicals which undergo coupling in the adsorbed state are likely to have a well defined orientation imposed by the high surface coverage.On the other hand, cyclohex-2-en- 1 -one is adsorbed very weakly. Hence the adsorbed neutral radicals RH' which undergo heterogeneous coupling are almost certainly free to assume quite different orientations relative to the electrode surface, and hence they behave towards coupling as though they were in the non-adsorbed state. Discussion The behaviour of wave I over the pH range 2.5-4, where no adsorption prewave is observed and the limiting current is diffusion-controlled, is consistent with the following mechanism : 1 R + HA RH+ +A- (1 a ) EtJ RH+ + e + RH' (1 b) where the rate-determining coupling step (1 c), denoted by rds, takes place in the adsorbed state. In eqn (1) HA is a proton donor and A- its conjugated base.The diffusional problem for mechanism ( I ) under the assumption that the reactant is weakly adsorbed (Henry-isotherm behaviour) is entirely analogous to that examined in ref. (9, the only difference being represented by the fact that here the rate-determining heterogeneous coupling step is preceded by the protonation step ( l a ) in quasi- equilibrium. The presence of this further step is readily accounted for by applying the equilibrium condition to both steps (1 a) and (1 b) under the assumption that the protonation equilibrium (1 a ) is almost completely shifted towards the unprotonated form R in the bulk solution: (2) CRH+/CRH = exp HE- E,)] = C,[H+]/(K,, CRH). H e r e f r F/RT, the C denote volume concentrations at the electrode surface and Kd = CR[Hf]/CR,+ is the dissociation constant of RH'.Using eqn (2) in place of eqn (A 4) of ref. (9, the final current us. potential characteristic of eqn (A 7) of ref. ( 5 ) is modified as follows: E = E, +f-' In ([Hf]/Kd) + (2f)-' In (k6/?kH Cg/D) + (2f)-' In [( 1 - i/Ld)2/(f/fd)]- (3) Here Cc is the bulk reactant concentration, PltH is the adsorption coefficient of the RH' intermediate, 6 is the diffusion-layer thickness and D is the diffusion coefficient, assumed to be the same for all diffusing species. According to eqn (3), the half-wave potential of wave I is given by (4) E$ = E,, + (2f)-l In [kdp2,,/(2D g)] - 2.3 f-' pH + (2f)-' In CE. The predictions of this equation are verified. Thus (t3EI/t3pH),; is = -60 mV as shown in fig.1, whereas (aE[/a log C*,),, is =-30 mV as &own in fig. 2. A further confirmation comes from the plot of log [ ( l -i/id)2/(i/id)] us. E in fig. 7(a), which is linear and has a slope of (30 mV)-', in agreement with eqn (3). This experimental behaviour does not agree with any of the most familiar homogeneous EHD mechanismsM. Y. Duarte et al. 107 (radical-radical, radical-substrate and ion-substrate coupling mechanisms)." Eqn (3) and (4) also hold in the case in which protonation follows charge transfer rather than preceding it. However, this mechanism is inconsistent with the decrease of the limiting current of wave I below its diffusion limiting value cd as observed at pH > 4. The heterogeneous nature of the coupling step is confirmed by the effect of an addition of lo-* mol dmP3 Triton X-100 to a pH 4 solution of mol dm-3 R.Such an addition causes a moderate negative shift of wave I (ca. -20 mV) and a change in wave shape. Thus the difference (I$ - 4) between the potentials at which i = </4 and 3i,/4 passes from 41 to 47 mV. In Part 2 it will be shown that the presence of Triton X-100 causes wave I to satisfy the requirements for a homogeneous radical-radical coupling step (e.g. &?!/a log Cc becomes ca. 20 mV). Triton X-100 concentrations as low as rnol dmP3 are indeed sufficient to remove both the reactant and the intermediate products completely from the adsorbed state, thus deactivating the heterogeneous pathway to radical-radical coupling in favour of the homogeneous one. The slight shift of El following Triton X- 100 addition denotes that the rates of the two parallel homogeneous and heterogeneous pathways are only slightly different. This is not surprising since the surface concentration of the protonated radical RH' should not differ much from that of the reactant R, which lies below the limit of sensitivity of the double-potential-step chronocoulometric technique (ca.0.3 ,uC cm-2, corresponding to a surface coverage 0 x 0.005 if the maximum surface concentration is taken as equal to that deduced from the limiting height of the adsorption prewave). The slope of the El us. log C;F. plot being less than 30 mV at Cg < lo-* mol dmP3 [see curve (a) in fig. 21 is explained by noting that the homogeneous coupling step is characterized by 8E!/a log Cg = 20 mV, whereas the heterogeneous one is characterized by log CE =' 30 mV.For Cc 6 lo-* rnol dmP3 the homogeneous pathway may therefore yield a more positive 4 value, and hence a higher current at any given potential, than the corresponding heterogeneous pathway. Under these conditions the homogeneous pathway will prevail over the parallel heterogeneous one, yielding &!?!/a log C;F. = 20 mV. As Cg becomes sufficiently high, however, the opposite will be trbe, whereas at intermediate Cg values the slope of the Ef us. log Cg plot will gradually change from 20 to 30 mV. 'At pH > 4, when the limiting current 6 of wave I becomes lower than its diffusion- limiting value, the rate of the overall process is simultaneously controlled by the protonation step (1 a), by the coupling step (1 c) and by reactant diffusion towards the electrode. Naturally, the contribution of the coupling step to the control of the electrode process decreases as we proceed along the rising portion of the wave and ultimately vanishes along the plateau of wave I.Contrary to what was reported by Ivcher et al.,7 we found that the dependence of 6 upon pH and buffer concentration does not satisfy the requirements for a rate-determining homogeneous protonation step. To explain such a dependence we must assume that protonation takes place both homogeneously and heterogeneously. An approximate expression for a limiting current controlled by a mixed homogeneous-heterogeneous protonation is as follows : 6 /(id - i,) = A + 0.8 1 2( [ H+] t d k,/Kd)i ( 5 ) with The first and second terms on the right-hand side of this equation express the contributions from heterogeneous and homogeneous protonation, t d is the drop time, k, and k , are the heterogeneous and homogeneous protonation rate constants respectively, whereas PR and Tm are the adsorption coefficient and the maximum surface concentration of R.Eqn ( 5 ) is obtained from eqn (474) of ref. (19), which represents a slight improvement pver Mairanovskii's formula,2o upon disregarding PR Tm with respect to ( 12Dta/7n)1. The latter simplifying assumption is thoroughly justified in view of the weak adsorptivity of R.108 Electrohydrodimerization of Cyclohex-2-en- 1 -one In view of the principles of general acid-base catalysis, both rate constants k, and k, consist of contributions from all proton donors present in the solution.The mechanistic study of EHD reactions of activated alkenes in aqueous media has revealed that adsorbed water molecules are much stronger proton donors than non-adsorbed We will therefore assume that by far the major contribution to k, stems from adsorbed water molecules, whereas that to k, stems from both the H30+ ion and the acid form HA (6) of the buffer: The linear plot of </(i,-il) vs. &HA at pH 5.46 in fig. 4 agrees with eqn (5) and (6), provided that k,, HA CHA 9 kv, H+ [H+] at this relatively low hydroxonium ion con- centration. The positive intercept A = 0.45 of this plot points to a heterogeneous protona!ion pathway. The slope of the plot yields a (kv. HA/&)' value ofca. 2.6 x lo3 dm3 mol-1 s-2. With these A and (kv,HA/K,)i values we can draw a plot of ~+0.812([H+]- c,, 'dk,, HA/&)' at constant CHA against [H']', expressing the asymptotic behaviour of </(id-iJ at low [H+] values when the homogeneous protonation by HA prevails over that by H,O+.This plot is expressed by the dashed straight line (b) in fig. 3, which fits nicely the lower portion of the experimental curve (a). Deviations between curves (b) and (a) at the higher [H+]i values are then used to estimate the kv,H+/kv,HA ratio on the basis of eqn ( 5 ) and (6). The dashed curve (c) was calculated from eqn ( 5 ) and (6) for a kv,H+/ k,,,,value of 1.5 x lo3. The moderate potential dependence of the limiting plateau of wave I for tl c id is such that the unavoidably artibrary procedure for measuring < may affect the above rate-constant values by factors as high as 3.However, the mechanistic conclusions drawn from fig. 3 and 4 are not appreciably affected by the choice of the procedure adopted to measure I;. The behaviour of wave I1 over the pH range from 8 to 10 is consistent with a mechanism which differs from the mechanism of eqn (1) by the fact that the protonation step in quasi-equilibrium follows the electron-transfer step : kv = kv, H+ [H+I+ kv, H A CHA- Re- + HA e RH' +A- (7 b) 2RH' 5 R2H2. (7 4 Moreover, the rate-determining coupling step (7 c) is now homogeneous. The current us. potential characteristic for this mechanism is reported in ref. (4), where the Koutecky-HanuS-Mairanovskii equation21* 22 for a reversible electron transfer followed by a coupling step was modified to include a preceding protonation in quasi-equilibrium. The modified equation is written as E = E[* +f-1 ln [25 (1 - f/id)/(i/id)'] 4' = Eb + (3f)-' In {[k'62[H+]2C:/(3D K:)} (8) (9) with where Ki = C,-[H+]/C,,.. The experimental value of -40 mV for aE!'/apH in fig.1 and that (20 mV) for aE['/a log C;it in fig. 2 (b), agree with the predictions of eqn (9), whereas they contrast with the predictions for homogeneous radical-substrate and ion-substrate coupling mechanisms1* and ,for any kind of heterogeneous coupling me~hanism.~ The plot of log [(I -i/&)/(t/t,)s] vs. E shown in fig. 7(b) is h e a r and has a (60 mV)-' slope, in agreement with eqn (8). The fact that the radical-radical coupling step is homogeneous along wave I1 whereas it is mainly heterogeneous along wave I is not surprising if we consider that RH' adsorption is likely to decrease towards increasingly negative potentials, in a way analogous to that shown by the resulting coupling product R2H2 (see fig. 5).Moreover, RH' adsorption is already small alongM. Y. Duarte et al. 109 wave I, where the rates of the two parallel homogeneous and heterogeneous coupling steps are almost comparable. At,pH values > 1 1 wave I1 is still characterized by a linear plot of log[(l -i/fd)/ (i/fd)'] us. E of slope (60 mV)-' in agreement with eqn (8), and by a value of aE!'/a log Cz of -20 mV [see fig. 2(c)]. This behaviour points again to a rate-determining homogeneous radical-radical coupling step. The tendency of to become pH independent as observed at 1 1 < pH < 13 can be explained on the basis of the following mechanism : R+e-GR'- (W 2R'- + Ri- rds Ri- + 2H,O e R2H2 + 20H-.(W Any protonation in quasi-equilibrium prior to the rate-determining coupling step, no matter the nature of the proton donor, would indeed cause the kinetics to be pH- dependent. Almost certainly the intermediate radicals R'- are not present around the electrode as negatively charged kinetic entities, but rather as ion pairs with the Na+ ion. The mechanism of eqn (10) in alkaline media was also hypothesized by Ivcher et al.' This work was supported by the Consiglio Nazionale delle Ricerche (Progetto finalizzato 'Chimica Fine e Secondaria'). Thanks are due to the Italian Ministry of Foreign Affairs for a fellowship to M. Y. D. during the tenure of which most of the present results were obtained. References 1 R. Guidelli, G. Piccardi and M. R. Moncelli, J. Electroanal. Chem., 1981, 129, 373. 2 M. R. Moncelli, F. Pergola, G. Aloisi and R. Guidelli, J. Electroanal. Chem., 1983, 143, 233. 3 C. Amatore, R. Guidelli, M. R. Moncelli and J. M. Saveant, J. Electroanal. Chem., 1983, 148, 25. 4 M. R. Moncelli, L. Nucci, P. Mariani and R. Guidelli, J. Electroanal. Chem., 1984, 172, 83. 5 M. R. Moncelli, L. Nucci, P. Mariani and R. Guidelli, J. Electroanal. Chem., 1985, 183, 285. 6 E. J. Denney and B. Mooney, J. Chem. SOC. B, 1968, 1410. 7 T. S. Ivcher, E. N. Zil'berman and E. M. Perepletchikova, Zh. Fiz. Khim., 1965, 39, 749. 8 E. Brillas and A. Ortiz, Electrochim. Acta, 1985, 30, 1185; J . Chem. SOC., Faraday Trans. I , 1986, 82, 9 P. Tissot and P. Margaretha, Helv. Chim. Acta, 1977, 60, 1472; Electrochim. Acta, 1978, 23, 104. 495. 10 P. Tissot, J. P. Surbeck, F. 0. GulaCar and P. Margaretha, Helv. Chim. Acta, 1981, 64, 1570. 11 M. L. Foresti, M. R. Moncelli and R. Guidelli, J. Electroanal. Chem., 1980, 109, 1. 12 M. L. Foresti and R. Guidelli, J. Electroanal. Chem., 1986, 197, 159. 13 P. Carsky, P. Zuman and V. Horak, Collect. Czech. Chem. Commun., 1965, 30, 4316. 14 P. Zuman, J . Electroanal. Chem., 1977, 75, 523. 15 J. Wiemann, S. Risse and P. F. Casals, Bull. SOC. Chim. Fr., 1966, 381. 16 E. Touboul, F. Weisbuch and J. Wiemann, Bull. SOC. Chim. Fr., 1967, 4291. 17 J. Dunogues, R. Calas, M. Bolourtchian, C. Biran and N. Duffaut, J. Organomet. Chem., 1973, 57, 18 L. Nadjo and J. M. Saveant, J. Electroanal. Chem., 1973, 44, 327. 19 R. Guidelli, in Electroanalytical Chemistry, ed. A. J. Bard (Marcel Dekker, New York, 1971), 20 S. G. Mairanovskii, E. D. Belokolos, V. P. Gul'tyai and L. I. Lishcheta, Elektrokhimiya, 1966, 2, 693. 21 J. Kouteckjr and V. HanuS, Collect. Czech. Chem. Commun., 1955, 20, 124. 22 S. G. Mairanovskii, Zzv. Akad. Nauk SSSR, Otd. Khim. Nauk, 1961, 12, 2140. 55. pp. 337-340. Paper 6/2499; 31st December, 1986

ISSN:0300-9599    

DOI:10.1039/F19888400097

出版商:RSC    

年代:1988

数据来源: RSC

摘要:

J. Chern. SOC., Faraduy Trans. 1, 1988, 84(1), 11 1-1 16 Empirical Analysis on the Constituent Terms of Transfer Enthalpies Quarternary Ammonium Ions in Acetonitrile-Methanol Mixtures Y asuhiko Kondo,* Ry6ichi Uematsu, Y oshinobu Nakamura and Shigekazu Kusabayashi Department of Applied Chemistry, Faculty of Engineering, Osaka University, Suita, Osaka 565, Japan Through an empirical analysis of transfer enthalpies in acetonitrile-methanol mixtures using ten quaternary ammonium ions, an equation which expresses the behaviour of the transfer enthalpy as a function of solvent composition has been derived: AH,A""'"(RR',N+) = AH,AN+mix[(B~')4N+] + 6, AH,AN +MeoH (RR;N+)x,,,, + A(RRjN+)x,,,, xAN. On the basis of this equation it is suggested that the transfer enthalpy includes at least three types of interaction, a protophobic interaction, an electrostatic interaction and another interaction, presumably a specific interaction such as dipole-dipole or acid-base association.Quaternary ammonium ions, together with alkali-metal and halide ions, have been used widely as a family of univalent ions in various fields of solution chemistry.'-'' In these studies a continuous variation in ionic radii has implicitly been assumed by systematically changing the alkyl chains in the ion, without introducing substantial modifications of solute-solvent interactions among the series. However, unusual patterns of behaviour have been found for the tetramethyl- ammonium ion, and even for the tetraethylammonium ion, through a more sophisticated treatment of the experimental results.6* 11-14 Protic us.dipolar aprotic solvent effects are still significant even when specific anion-protic solvent interactions have no direct rele~ance.~. l5 A dissection of thermodynamic quantities into their constituent terms seems to be a fundamental step in order for empirical energy correlations, such as an extended Brarnsted treatment and entropy-enthalpy compensations, to be made on a more theoretical basis.l'* l7 In this work the empirical but quantitative dissection of transfer enthalpies into their constituent terms will be carried out over a series of quaternary ammonium ions in acetonitrile-methanol mixtures, and discussion will be made as to its physical significance. Experiment a1 Materials Adamantane and norbornane were recrystallized three times from acetone and dried over P,O, for a few days.3-Ethylpentane (a commercial product) was used without further purification. Tetra-n-propylammonium iodide was recrystallized twice from acetone-propan-2-01 mixtures and dried over P,05. Tetramethylammonium bromide and perchlorate were recrystallized twice from acetonitrile-ethanol mixtures and dried over P,O, for a few days. N-Methylquinuclidinium iodide, prepared from quinuclidine 111112 Transfer En t halpies in Ace t on it r ile-Me t hanol Mixtures and methyl iodide in acetonitrile, was recrystallized three times from acetone-ethanol mixtures and dried over P,O,. (Found: H, 6.40; C, 38.05; N, 5.50; I, 50.32. Calcd for C,H,,NI : H, 6.37; C , 37.96 ; N, 5.53 ; I, 50.13 YO .) 1 -Methyl-4-dimethylaminopyridinium iodide, prepared from 4-dimethylaminopyridine and methyl iodide in acetonitrile, was recrystallized three times from ethanol-methanol mixtures and dried over P,O,.(Found : H, 4.87; C, 36.48; N, 10.63; I, 47.81. Calcd for C,H,,N,I: H, 4.96; C, 36.38; N, 10.61; I, 48.05 YO .) 1 -Methyl-4-t-butylpyridinium iodide, prepared from 4-t-butylpyridine and methyl iodide in acetonitrile, was recrystallized six times from butan-2-one-ether mixtures and dried over P,O, for a few days. (Found: H, 5.73; C, 43.16; N, 5.07; I, 45.89. Calcd for C,,H,,NI: H, 5.82; C, 43.34; N, 5.06; I, 45.79%.) l-Methyl-4- cyanopyridinium iodide, prepared from 4-cyanopyridine and methyl iodide in acetonitrile, was recrystallized three times from methanol-ethanol mixtures and dried over P,O, for, a few days.(Found: H, 2.82; C, 34.02; N, 11.48; I, 51.52. Calcd for C,H,N,I : H, 2.87 ; C, 34.17 ; N, 1 1.39 ; I, 5 1.58 YO .) Acetonitrile was successively distilled from calcium hydride, phosphorus pentaoxide and again from calcium hydride. Methanol was distilled after refluxing with magnesium metal and then twice under a nitrogen atmosphere. Heat of Solution Measurements Heats of solution were measured at 25.0k0.05 "C with a Tokyo Riko twin isoperibol calorimeter (TIC-2D) ;l6-" the final concentration ranges used for the measurements were ca. 8.5 x 10-3-3.8 x lo-, mol dme3 for norbornane and adamantane, (3.7- 9.1) x mol dm-3 for 3-ethylpentane and (0.75-1.5) x lo-, mol dm-3 for the salts. Experimental errors were estimated to be ca.2% from duplicate or triplicate runs. Results The enthalpies of solution determined in this work are summarized in table 1. In order for a quantitative analysis on transfer enthalpies to be made without any possible errors due to extrathermodynamic assumptions, relative transfer enthalpies, BRAHtN+miX (RRiN+), were defined by the equation BRAHtN+miX (RR;N+) = AHtN+miX(RRiN+X-) - AHtN+miX[( Bu"),N+X-] (1) in which tetra-n-butylammonium ion was taken as a standard ion. In eqn (1) X- indicates the common anion for the particular pair of salts: results are summarized in table 2.* Tetramethylammonium bromide is practically insoluble in acetonitrile, as is tetramethylammonium perchlorate in methanol, as can be partly inferred from table 1. This prevents the single-ion transfer enthalpy for tetramethylammonium ion, AH,AN'MeoH(Me4N+), to be derived according to the conventional procedure. One advantage of the relative transfer enthalpy defined above is exemplified as follows.First, the relative transfer enthalpies were calculated on the basis of eqn (1) for the tetramethylammonium perchlorate-tetra-n-butylammonium perchlorate pair and plotted in fig. 1 as closed circles; these values will be referred to as 'direct values'. An analogous quantity, except that methanol is chosen as a reference solvent instead of acetonitrile, BRAH~eoH+mix(RRiN+), is related to the original quantity by eqn (2) through a thermodynamic cycle : dRAH:N+miX(RRLN+) = C ~ R A H ~ ~ + ~ ~ O ~ ( R R ~ N + ) + C5RAHFeoH+mix(RR;N+). (2) Secondly, the second term on the right-hand side of eqn (2) was calculated for the tetramethylammonium bromide-tetra-n-butylammonium bromide pair. Adjustment of * Some of the values published previouslyl"ls were redetermined or supplemented in this work.Y.Kondo, R. Uematsu, Y. Nakamura and S. Kusabayashi 113 Table 1. Enthalpies of solution for acetonitrile-methanol mixtures (25.0 "C) (kJ mol-') xWeOH 3-ethylpentane adamantane norbornane Pr",NI Me,NBr - 0 10.6 21.8 12.8 14.3 0.10 10.3 21.3 12.4 10.5 10.4 0.25 9.66 20.95 11.9 11.3 10.9 0.50 8.50 19.8 10.6 14.5 15.8 0.75 7.04 18.2 9.29 18.4 21 .o 0.90 - - - 21.5 25.1 1 .o 4.9 1 16.8 7.68 25.2 29.4 Me,NClO, @Me1 M e , N o M e I - B u ' o M e I - NC c 1 NMeI 15.9 9.56 14.3 8.74 8.62 - 6.19 10.9 5.05 5.46 18.65 7.60 12.2 6.23 6.71 23.4 12.0 16.9 10.3 11.6 29.45 17.4 22.1 15.0 17.35 - 21.8 26.7 18.8 22.0 - 26.7 31.2 23.4 26.95 - m d 5 1 .o 0.5 0 XMeOH xWeOH, mole fraction of methanol.10.0 8.0 4 -0 0 0 0 Fig. 1. Relative transfer enthalpy, dRAH~N'miX(RRjN+) as a function of solvent composition. A, 0 0, and 0, experimental results for Pr",N+, Me," and N-MePy+, respectively; (-) calculated values according to eqn (4) with the parameters summarized in table 2.114 Transfer Enthalpies in Acetonitrile-Methanol Mixtures Table 2. Relative transfer enthalpies, dRAH,AN'"'"(RR~N+) for acetonitrole-methanol mixtures and two parameters (25.0 "C) (kJ mol-') 0 0 0.10 0.2 0.25 0.6 0.50 0.8 0.75 1 .7 0.90 1.9 1 .o 2.2 ~ R A H i A N - M C O H 2.2 0 0 I .O 1.65 3.05 4.6 5.8 6.0 6.0 0 0.30 0 0.5 - 1.1 1.25' 3.2 2.9f 5.4 5.05 6.6 - 7.7 7.7 - 3.2 0.15 - 0 0.6 1.45 2.5 4.7 5.8 7.1 7.1 .4.2 - 0.30 0 0.6 1.6 3 .O 5.4 6.9 8.4 8.4 .4.8 - 0.25 0 0.6 1.55 3.2 5.85 7.8 9.4 9.4 -6.0 0.20 0 0.6 1.5 3.2 5.4 7.1 8.2 8.2 0.24 -4.6 0 0.3 1.1 2.2 3.9 4.8 6.0 6.0 0.18 - - 3.2 0 0.8 1.7 3.6 6.3 8.1 9.6 9.6 .4.8 0.18 - 0 0 0.5 - 1.4 1.7 - 2.6 3.0 - 3.4 5.4 -4.85 7.5 -6.1 9.0 - 5.0 9.0 0.32 -6.0 These values were calculated using the data published in a ref.(16) and (19); ref. (16); ref. (17); 2,3-dihydro- 1,2,5-trimethy1-3-methylthio- 1,2,4-triazolium ion, ref. (1 8). AHtN+mix[(Bun)4N+], ref. (16). f direct value, see text. the calculated values through the addition of the constant, GRAHtN+MeoH(RRjN+) according to eqn (2) leads to an indirect estimate of the relevant quantity, SRAHtN+miX (RRjN+).The adjustment of the constant value was repeated until the indirect values overlapped the direct values over an intermediate solvent composition. The most plausible set of indirect values are indicated in fig. 1 as open circles; the value SRAHtN+MeoH(Me4N+) was found to be 7.7 kJ mol-'. With respect to the transfer of tetra-n-propylammonium ion as well as of the tetraethylammonium ion, the relative transfer enthalpy shows a linear change as a function of solvent composition, i.e. (see fig. 1). However, for the ions in which at least one alkyl group is substituted by a methyl group, a negative deviation from linearity becomes significant. This effect can be accommodated into the scheme by introducing the term A(RRjN+)x,,,, xAN into eqn (3), i.e. byY .Kondo, R. Uematsu, Y . Nakamura and S. Kusabayashi 115 Furthermore, the value A(RRjN+) can effectively be evaluated by use of the transfer quantity in equimolar mixtures, i.e. by eqn (5), without recourse to any rigorous treatment : A(RRjN+) = [GRAH,AN+o.5(RRjN+) - 0.5SRAH,AN+MeoH(RRjN+)]/0.25. ( 5 ) The two parameters, c~~AH,A~+~'O~(RR;N+) and A(RRjN+), together with the standard deviation from the regressions, are summarized in table 2. Some comparisons of the calculated and experimental values are shown in fig. 1. Considering that the experimental error in the transfer enthalpy amounts to ca. 0.4 kJ mol-l, the reproducibility of this simple treatment seems to be satisfactory. Discussion Among the methods developed for deriving single ionic quantities, the most frequently used are reference electrolyte methods, especially the Ph,As+/Ph,B- assumption.20* 21 Although doubt has sometimes been expressed concerning the extension of the method to protic-solvent systems, no practical method is available to test this point.Provided that a central charge is sufficiently prevented by four phenyl or alkyl groups from affecting the surrounding media, a transfer quantity of the reference electrolyte could naively be expected to simulate that of the relevant non-electrolyte. Transfer enthalpies of three hydrocarbons indicate almost the same pattern of behaviour, i.e. AH,AN'MeoH amounts to ca. -5 kJ mol-1 and AH,AN+miX shows linear changes with solvent composition. The single-ion transfer enthalpy for tetra-n-butylammonium ion derived on the basis of the (Bu"),N+/( Bu"),B- assumption behaves similarly, except that the tetra-n-butylammonium ion shows a slightly more exothermic trend by ca.1.5 kJ mo1-l at intermediate solvent compositions (see tables 1 and 2). Thus the credibility of single-ion values derived by the present authors so far is estimated to be (ca. 2 kJ mol-l. Along the series of tetraalkylammonium ions, (Pr"),N+ to Me,N+, the relative transfer enthalpy, BRAH,AN +MeoH(R4N+) gives endothermic values increasing with decreasing ionic radii. The ion-dipole interaction also predicts an increasing endothermicity with decreasing ionic radii for ion transfer from acetonitrile ( p = 3.92 D) to methanol ( p = 1.70 D).22 The relative transfer enthalpy, G,AH,AN+MeoH(R4N+) can be taken as a measure of exothermicity derived presumably from protophobic interactions around the quaternary ammonium ion being replaced by endothermicity due to electrostatic interactions (largely ion -dipole interactions).This view is in accord with the argument that electrostatic interactions are still important for the R,N+ ion in non-aqueous solvents. 23 One characteristic feature observed for tetramethylammonium ion lies in the presence of concavity in the plot as represented by the term A(Me4N+)xMeoHxAN (see fig. 1). This feature is confirmed by the series Et4N+, Et,MeN+, N-methylquinuclidinium and N- methylpyridinium ions : along this series the value dRAHtN-+MeoH(RRjN+) increases, while the value A(RRjN+) decreases. Along the series, a closer approach of solvent molecules to a cationic centre becomes more feasible, either by decreasing the conformational freedom of the alkyl chains, Et,MeN+ to the N-methylquinuclidinium ion, or by decreasing steric requirement, the N-methylquinuclidinium to N-methyl- pyridinium ions.Recent Monte Carlo simulations indicate the presence of a definite, although less marked, hydration shell around the tetramethylammonium ion, with an oxygen atom pointing towards the cationic centre.24 From another point of view, a larger dipole moment is expected to be induced along the rotational axis in the ion, since along the series the polarizability difference becomes more marked between the methyl group and the other groups located at the opposite side of a cationic centre. Phenomenologically, a negative deviation from linearity, i.e.a negative value of116 Transfer En t halp ies in Ace t on it rile- Me t han ol Mixtures A(RRjN+), corresponds to preferential solvation of the ammonium ion by acetonitrile. Thus an increasing concavity might be ascribed to the increasing dipole-dipole association between the dipolar ammonium ion and the dipolar acetonitrile. Of the substituents introduced into the para-position in the pyridine ring, two groups, Me2N- and But-, have electron-donating properties, while one NC- has an electron- accepting property. Neither the term dRAHtN+MeoH(RR;N+) nor the term A(RR;N+) follow this trend. Interestingly, the increasing trend in the term c~RAH.~+~~O~(RR;N+) along the series, But- < Me2N- < H-, NC- may have its origin in the steric bulk of a substituent, since it is supposed to affect the closer approach of the solvent to a dispersed cationic charge over the ring.Again in the pyridine series, two factors, steric control of a closer approach of the solvent to a cationic centre and of the alignment of the solvent required for dipole-dipole association, possibly together with other factors, are thought to exert concomitant effects on the parameters. AHP""'"(RRjN+) = AH~N+"ix[(B~n)~N+] + C ~ R C ( H ~ ~ + ~ ~ O ~ ( R R ~ N + ) X M ~ ~ H -k A(RR;N+)XMeOH xAN (6) Rearrangement of eqn (4) leads to the quantitative relation. This derivation makes it possible to suggest that the transfer enthalpy for quaternary ammonium ions consists of at least three terms, protophobic interactions, ion-dipole interactions and another factor, presumably a specific interaction such as dipole4ipole or acid-base association between the quaternary ammonium ion and the solvent.Furthermore, this type of empirical analysis would be useful for testing the internal consistency of the experimental values. We thank Dr M. H. Abraham, University of Surrey, for a critical reading of the original version of this manuscript. References 1 R. A. Robinson and R. H. Stokes, Electrolyte Solutions [Butterworths, London, 2nd edn (revised), 2 F. Franks, in Hydrogen-bonded Solvent Systems, ed. A. K. Covington and P. Jones (Taylor and 3 Wen-Yang Wen, in The Physical Chemistry of Aqueous Systems, ed. R. L. Kay (Plenum Press, London 4 C. M. Criss, in Physical Chemistry of Organic Solvent Systems, ed.A. K. Covington and T. Dickinson 5 M. H. Abraham, J. Chem. SOC., Faraday Trans. I , 1973,69, 1375. 6 M. H. Abraham, J. Chem. SOC., Faraday Trans. I , 1978, 74, 2101. 7 C. de Visser and G. Somsen, J. Chem. Thermodyn., 1972, 4, 313. 8 C. de Visser and G. Somsen, J. Phys. Chem., 1974, 78, 1719. 9 F. Kawaizumi and R. Zana, J. Phys. Chem., 1974,78, 1099. 19651. Francis, London, 1968), p. 31 and other articles therein. and New York, 1973), p. 155. (Plenum Press, London and New York, 1973), p. 23 and other articles. 10 B. S. Krumgalz, J. Chem. SOC., Faraday Trans. I , 1980, 76, 1887. 11 C. V. Krishnan and H. L. Friedmann, J. Phys. Chem., 1971,75, 3606. 12 R. L. Kay, G. P. Cunningham and D. F. Evans, in Hydrogen-bonded Solvent Systems, ed. A. K. 13 R. Zana, J. E. Desnoyers, G. Perron, R. L. Kay and K. Lee, J. Phys. Chem., 1982, 86, 3996. 14 M. Costagnolo, A. Sacco and A. de Giglio, J. Chem. SOC., Faraday Trans. I , 1984, 80, 2669. 15 R. Fernandez-Prini and G. Atkinson, J. Phys. Chem., 1971, 75, 239. 16 Y. Kondo, M. Itto and S. Kusabayashi, J. Chem. Soc., Faraday Trans. I , 1982, 78, 2793. 17 Y. Kondo, M. Ogasa and S. Kusabayashi, J. Chem. Soc., Perkin Trans. 2, 1984, 2093. 18 Y. Kondo, M. Inoue and S . Kusabayashi, J. Chem. SOC., Perkin Trans. 2, 1983, 1217. 19 Y. Kondo, H. Shiotani and S. Kusabayashi, J. Chem. SOC., Faraday Trans. I , 1984, 80, 2145. 20 A. J. Parker, Chem. Rev., 1969, 69, 1. 21 R. Alexander, A. J. Parker, J. H. Sharp and W. E. Waghorne, J. Am. Chem. SOC., 1972,94, 1148. 22 A. D. Buckingham, Discuss. Faraday SOC., 1957, 24, 151. 23 R. Zana, G. Perron and J. E. Desnoyers, J. Solution Chem., 1979, 8, 729. 24 W. L. Jorgensen and J. Gao, J. Phys. Chem., 1986, 90, 2174. Covington and P. Jones (Taylor and Francis, London, 1968), p. 249. Paper 71008; Received 2nd January, 1987

ISSN:0300-9599    

DOI:10.1039/F19888400111

出版商:RSC    

年代:1988

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I , 1988, 84(1), 117-132 Nuclear Magnetic Resonance Spectroscopy applied to Minerals Part 6.-Structural Iron in Kaolinites as viewed by Proton Magnetic Resonance William E. E. Stone* Section de Physico-Chimie Mine'rale (MRAC), Place Croix du Sud 1 , B-1348 Louvain-la-Neuve, Belgium Rosa-Maria Torres-Sanchez? Groupe de Physico-Chimie Mine'rale et Catalyse, Place Croix du Sud 1 , B-1348 Louvain-la-Neuve, Belgium This paper examines to what extent the proton n.m.r. signal of hydroxyl groups can be used as an internal probe to investigate the distribution of structural iron in the framework of kaolinite. To this end, nine samples with iron contents varying between 0.06 and 1.86 wt % in Fe,O, have been examined at 90 and 24 MHz. Measurements involved determinations of the various contributions to the second moment of the proton line by solid-echo techniques and also spin-lattice relaxation times (TJ. Although the lineshapes of all samples are fairly constant, the values vary considerably (over three orders of magnitude).These values can be accounted for by a paramagnetic induced relaxation mechanism using (apart for the purest sample) a multi-impurity approach. These experiments clearly show that part of the iron is not evenly distributed within the structure. The results are finally discussed in relation to the problem of structural disorder in kaolinites and are compared with other experimental approaches. Because of its use in many industrial applications, kaolinite has been the subject of numerous studies.l9 In soils and weathered lateritic materials, the association and interaction between iron and kaolinite constitutes an important and active area of re~earch.~ Kaolinites are dioctahedral 1 : 1 layer silicates with practically no exchange capacity.They are formed by a single tetrahedral sheet (with all sites occupied by silicon) combined with a single octahedral sheet (wherc one third of the sites are vacant and the rest occupied by aluminium). These layers, 7 A thick, are stacked on top of each other along the c* axis. The outer plane of the octahedral sheet is entirely composed of hydroxy groups, forming a nearly close-packed surface. An inner hydroxy group is found in the plane shared by the tetrahedral and octahedral sheets. The ideal structural formula is Si,Al,O,,(OH), and most natural samples closely approach this composition.Natural samples, however, are far from being perfect well ordered specimens and exhibit various types of disorder : variation in the degree of isomorphous substitution, random location of the vacant site, variation in the structural stacking sequences, distorsions in the regular disposition of layers (occlusions, voids) and presence of contaminants. To determine the deviation from ideality and degree of structural disorder of these mineral structures, X-ray diffraction studies have been widely used, as exemplified by the detailed study given by Tchoubar et aZ., To classify kaolinites following their degree of crystallinity, several tests based on physical properties of the material have been used.' It is found that they all correlate very strongly with the amount of iron present in the t Present address : C.I.B.I.O.M., Cerrano 669, 1414 Buenos Aires, Argentina. 117118 Structural Iron in Kaolinites Table 1.List of the kaolinite samples studied (wt % Fe,O, and K,O obtained by chemical analysis, H.I. is the Hinckley index) sample origin Fe,O, (%) K,O (YO) H.I. K GB 1 P1 P3 P4 FU8 P2 ONNE, V W D Keokuk (U.S.A.) St Austell (G.B.) Georgia (U . S . A.) Charentes (F) Charentes (F) Charentes (F) Soil Cuba Charentes (F) Soil Nigeria 0.06 0.34 0.51 0.93 1.17 1.35 1.52 1.62 1.86 0.01 1.57 0.55 1.47 1.06 0.88 0.07 0.32 0.55 - 0.51 - 0.44 0.1 0.82 0.28 0.39 0.14 sample. This observation explains why techniques such as e.s.r. and Mossbauer spectroscopies have extensively been used to try and understand the role played by this element in the crystallochemical properties of kaolinite~.'-~ The frequency and intensity of the OH bands are sensitive indicators of minor variations in the crystal structure.Infrared spectroscopy is therefore one of the techniques which has been used to classify kaolin^.^^^^^-^^ The protons of the hydroxy groups of kaolinite can also be studied by n.m.r., as shown by a recent study on the dehydroxylation of kaolinites by Otero-Arean et a1.I2 This paper examines, by n.m.r., the relaxation properties and lineshapes obtained for protons in a collection of kaolinites with different iron contents. It is observed that the response obtained varies according to the sample examined. By using the proton as an internal probe, information concerning the distribution of iron in the lattice can be reached. These results add further evidence to the established relation found between iron content and structural disorder of kaolinites.Experimental The kaolinite samples studied here (see table I) were taken from a collection of samples which had previously been examined by e.s.r., i.r. and analytical electron microscopy. The iron content expressed as weight percent in Fe,O, varies from nearly zero to ca. 2 %. In the analysis of results, it is considered that the majority of iron is in the form of FeIII ions substituted in A1 sites in the octahedral sheet ; when necessary, a deferration process is used to eliminate most of the existing freely accessible iron oxides and the corresponding samples have symbols with a subscript D.These samples were deferrated following the standard Segalen u.v.-oxalate procedure.' The degree of crystallinity of the samples is given by their Hinckley index', (H.I.) (a well ordered structure has a high H.I.). To eliminate physisorbed water, the samples were pumped overnight and then sealed in 5 mm diameter tubes. The samples examined were in the powdered form. The n.m.r. measurements were carried out, at room temperature, on a Bruker CXP-90 spectrometer. Samples were examined at 90 and 24 MHz. Spin-lattice relaxa- tion times were measured by a (1 8O0-t-9O0) pulse sequence ; solid echoes were obtained after (9Oo,-r-9O0,) sequences. The signals were scanned a few hundred times. Proton Relaxation Results As stated in the introduction, the amount of iron present in kaolinite is found to be in direct relation to the amount of structural disorder existing within the crystallites.W.E. E. Stone and R-M. Torres-Sanchez 119 (dl Fig. 1. Absorption curves of samples VW taken in various experimental conditions (see text). 1.0 n 8 v ZN N 1 h c U n f 8 v ZN 0.1 0 0.5 1.0 1.5 2.0 tls Fig. 2. Semi-logarithmic plot of the recovery of the longitudinal magnetization of sample K. @, 90 MHz; 0, 24 MHz. Probably equally important is the way the amount of iron is distributed within the framework. This is certainly true as far as the proton n.m.r. results are concerned. The presence of iron impurities affects the proton magnetic resonance properties in several ways : the resonance line and its intensity may be altered, the time dependence and speed of recovery of the magnetization during a spin-lattice relaxation experiment is changed.120 Structural Iron in Kaolinites 1.0- v s - h 8 \ *r W n T " - 0 .1 0 10 50 100 150 200 2 50 ; O-- t+/psf Fig.3. Semi-logarithmic plot of the recovery of the longitudinal magnetization of various samples with increasing iron contents at 90 MHz. (a) Complete recovery plots. 0, GB1; 0, P1; 0, P3; 0, P2; A, ONNE. (b) Exploded short-time portion. 0, P1; 0, P2. In kaolinites, three different n.m.r. signals have been observed (see fig. 1): (1) a fine central line whose amplitude varies from sample to sample and which is due to physisorbed water (this line disappears when the sample is degassed overnight and all samples will be submitted to this type of pretreatment); (2) a large Gaussian-type absorption line due to the hydroxyl groups; and ( 3 ) a small very broad line which is present for samples with at least 0.5 % Fe,O,.As shown in fig. 1, the three lines relax at different rates: curves (a) and (6) correspond to the samples as such, and (c) and ( d ) to the samples degassed overnight. Curves (a) and (c) correspond to a completely relaxed situation (i.e. observed after a time much longer than TI), whereas (6) and (4 are taken at some intermediate time (< &), It is clearly observed that the main OH line relaxes more slowly than the others. The effect of the presence of paramagnetic iron on the spin-lattice relaxation of certain of our kaolinite samples can be seen in fig.2 and 3. Fig. 2 shows the results obtained for the purest sample (K) at 90 and 24 MHz. The normalized magnetization recovery M,(t) is satisfactorily exponential with respect to time t . Fig. 3(a) shows the magnetization recoveries of five other samples with iron concentrations increasing from sample GB1 to sample ONNE. The curves shown were recorded at 90 MHz. UnlikeW. E. E. Stone and R-M. Torres-Sanchez 121 1 10- VJ \ c 1 0- 10- 10- ! n W 10 100 ~ ~ / 1 0 ~ ~ at ~ r n - ~ Fig. 4. values obtained at 90 MHz for the various samples as function of the total number of Fe atoms per cm3. 0, Main line; A, broad line. sample K, it is observed hyre that the recovery of the magnetization has, for a large part, the form exp [ - ( t / T , ) z ] (also true for samples P4, FU8 and VW).For clarity, the first experimental points of the relaxation curves are not given in fig. 3(a); two typical initial portions of these curves are given (for P1 and P2) i? fig. 3(b), where it can be observed that at short times the curves deviate from the tz law. The reasons for this bending over of the curves at short t are discussed below. The results presented in fig. 2 and 3 correspond to measurements of the signal intensity taken close to the summit of the n.m.r. absorption curve, i.e. the main OH line. The spin-lattice relaxation of the broad component is extracted by measurements taken in the wings of the absorption curve after correcting for the contribution of the main slower relaxing central line. In fig. 4 are given the TI values obtained at 90 MHz for the various samples as a function of N p , the number of FeIII atoms ~ m - ~ , derived from chemical analysis assuming that all the iron is evenly distributed throughout the sample.Analysis At room temperature, in solids such as kaolinite, the electron S spins of the paramagnetic iron and the nuclear I spins of the protons occupy fixed positions in space. Spin-lattice relaxation of I spins occurs because the rapid flipping of S spins (through their own rapid q) modulates the dipolar interaction between S and I. The relaxation122 Structural Iron in Kaolinites rate 1 / perturbation theory to be14 for a nucleus Z located at a distance r from an impurity S can be estimated from 1/T, = C/r6 (1) where and y are the gyromagnetic ratios, 8 the angles between r and the applied magnetic field H,, co, the Larmor frequency of spin Zand z the correlation time of spin S, inducing the relaxation of spin I.Eqn (1) and (2) therefore describe in terms of microscopic parameters the relaxation of a given nucleus. The next problem is then to describe the behaviour of the measured macroscopic magnetization M, which results from the sum over all nuclei present in the sample. These nuclei, however, may be associated with the impurities in different manners. Nuclei located close to a paramagnetic ion may go undetected owing to the large local static field induced by the impurity. To take this effect into account, it is usual to define a small volume around each impurity. This volume, of 'cut-off' radius b,, includes all nuclei having their resonance frequency shifted away from the range of observation.As shown by eqn (l), the direct relaxation effect, due to the impurity on the neighbouring visible nuclei, falls off rapidly with distance. Because of nuclear spin-spin interactions this relaxation information can however be transferred throughout the lattice, to further nuclei, by a spin-diffusion mechanism. l5 The overall relaxation mechanism is therefore a complicated function which will depend on the relative values for C, the spin-diffusion coefficient D and the mean distance between impurities R. When the impurity concentration is small, the magnetization recovery is exponential and expressions for the relaxation rates, in several limiting cases, have been proposed and tested on simple doped material.16 For all kaolinites, except sample K, the recovery of nuclear magnetization is non- exponential.The effect of a relatively high concentration of FeI'I impurities will be to minimize spin-diffusion effects ; the interpretation of the observed relaxation rates can no longer be based on single-impurity models1' but must integrate the presence of an ensemble of paramagnetic impurities. Multi-impurity approaches have been proposed in the case where the paramagnetic ions are randomly distributed on the lattice At every given point in the lattice the magnetization recovery is exponential, but from point to point the configuration of impurities, as seen by the nuclei, varies. This results in a distribution of relaxation rates, the average of which is not necessarily exponential.Finally, an ensemble average over all possible impurity configurations, around a given nuclear site, should be performed. All nuclei lying at a distance smaller than b, are excluded. The analytical expression for the recovery of the magnetization gives an asymptotic long-time dependence of the form exp - ( t / T , ) r with'' when ( 3 ) (4) where C is the powder average value of C [eqn (2) with the angular terms set equal to 3. The effect of the finite cut-off radius b, is to cause deviations to the above law at short times. This bending over of the relaxation curves is indeed observed in our experiments, as illustrated in fig. 3(b). Eqn ( 3 ) will tentatively be applied to all examined kaolinites except sample K.For this particular kaolinite, as shown above, the magnetization recovery is found to be exponential. It is also observed that its spin-lattice relaxation time is directly proportional to the applied magnetic field H,. This behaviour corresponds (provided co,z > 1) to one of the limiting cases examined by Lowe and Tsel' for samples with aW. E. E. Stone and R-M. Torres-Sanchez 123 relatively low paramagnetic concentration and has been named the spin-diffusion ( 5 ) vanishing case. Then where A is a constant with a value of ca. 50 for randomly distributed ceptres. In tohe case of sample K, N , = 1.15 x lo9 Fe atom cm-3 with a distance R = ($nNp)5 = 27.4 A; these values are in the range of those studied by Lowe and Tse.16 In the evaluation of C, which is found in both eqn (3) and (9, it is considered that the only paramagnetic centres responsible for the proton relaxation are Ferrr ions; these centres being high-spin S state ions, &hS(S+ 1) = pg (i.e.square of the FeIII magnetic moment p, = -5.47 x lo-,' erg G-l). The only remaining unknown of C is then z, the correlation time of the paramagnetic spin. It will be assumed in what follows that, between 24 MHz and 90 MHz, the frequency dependence of z is small, so that by comparing the corresponding values of T,, a rough estimate for the order of magnitude of z can be obtained. We shall now consider the information that can be extracted from the measured relaxation T, values for three typical samples. 1 / T, z AA$ (CD); Kaolinite K This kaolinite is the purest and best crystallized sample available; it can be considered as a standard reference material to which eqn (5) can be applied with a relatively high degree of confidence.With this equation, an estimate for z can be obtained by taking the ratio of the experimentally determined T, values measured at 90 and 24 MHz. The assumption that z is frequency independent in this range is partly justified by the observed linear dependence between 1/T, and H,. The spin-diffusion coefficient D is related to the second moment of the proton line.'* It is observed (see below) that the second moment is frequency independent and therefore D falls out in the expression of the q ratio. With q = 1.3 and 0.4 s at 90 and 24 MHz, respectively, a value for z = 1.2 x lo-' s is obtained.This value compares relatively well with other estimates for z obtained for FeIII in other minerals.lg* 2o Considering that chemical analysis gives a correct estimate for N,, the value for D which can then be extracted is ca. cm2 s-l. This very small value for D is not all together unexpected. The lamellar structure of kaolinite makes the proton lattice highly anisotropic and spin diffusion throughout the particles shall be governed by the slowest process, i.e. normal to the structural layers (for this sample, with a random distribution of ions, the amount of Fe'II per unit cell is 1 /30). Such small values for D have been found in solids with highly irregularly distributed nuclei.21 Finally, an estimate for b,, the cut-off radius, can be obtained from the relation16 where HL is the nuclear local field proportional to the second moment of the n.m.r.line H , z (M2)i. A value of b, z 6 A is obtained. b: = G J I H L (6) Kaolinite GBl In our collection of samples, GB1 is next in terms of iron conten$. If the iron is evenly distributed then N, = 6.5 x lo1' Fe atom cm-3 with R = 15.4 A. This kaolinite also contains 0.55% K,O associated most probpbly with a mica contamination phase. The measured T, values are extracted from the ti recovery of the magnetization [see fig. 3 (a)] and are, respectively, 54 and 34.2 ms at 90 and 24 MHz. As for kaolinite K, no broad component centred about the Larmor frequency is observed. With eqn (3) an estimate for z can be obtained from the ratio of the measured T,, z = 1.4 x lo-' s.With this value of z and the experimental value for T,, the number N , of paramagnetic species responsible for the proton relaxation can be extracted from eqn (3). The value found, N , = 2.1 x lo1' Fe atom ~ m - ~ , corresponds to approximately one third of the iron124 Structural Iron in Kaolinites Table 2. Comparison between the amount of Fell1 given by the chemical analysis and that responsible for the measured q N,/lOl9 Fe atoms ~ r n - ~ chemical calculated from sample analysis q (main line) difference K 1.15 GB 1 6.50 PI 9.75 P3 17.77 P4 22.36 FU8 25.80 vwn 29.05 P2 30.96 ONNE, 35.54 1.15 2.14 2.48 4.82 8.69 7.12 7.10 7.42 10.38 0 4.36 7.27 12.95 13.67 18.68 21.95 23.54 25.16 present in the sample. For all other samples, the same trend is observed: from the relaxation of the main central line, the calculated Np is smaller than that given by chemical analysis.Part of the iron is therefore not evenly distributed within the kaolinite sample (see table 2). One possibility is to locate these ‘extra’ Fe’II ions in the mica contamination phase, which would then correspond to a mica having an iron content of 4 5 % . Following our previous study on micas,22 this would give rise to a proton line with a second moment of ca. 5 G2. These mica protons would, however, be difficult to detect given the small percentage of mica present in the sample (& 5 YO mica in which protons are more diluted than in kaolinite). Another possibility is that instead of being randomly distributed, some of the iron ions are more densely packed. Qualitatively, it is clear that given a number of Np ions, the measured increases when some of these ions are grouped together.This point will be taken up later. From the bending over, at short times, of the ti magnetiza!ion recovery, an estimate for b, can be obtained from eqn (4) and is found to be 6.5 A. Kaolinite V WD This is one of the highly iron-charged kaolinites which contained 1.75 YO Fe203 before and 1.52 YO Fe203 after deferration. For the pretreated sgmple, a random distribution of iron gives Np = 29 x lo1’ Fe atom with R = 9.4 A. The chemical analysis shows the presence of 0.44% K,O. Some typical n.m.r. spectra are given in fig. 1. Unlike the two other cases discussed above, an additional broad signal, with a shorter spin- lattice relaxation time, is observed.It should be mentioned that because the contribution of this broad line to the total proton line is small (ca. 10%) the uncertainty on the experimental relaxation points for this component is high. We will first discuss the results obtainled for the main proton line. At both 90 and 24 MHz, the magnetization recovers as ts and the corresponding & values are 4.8 and 2.3 ms. As for kaolinite GB1, with these two values an estimate of 2.4 x lo-’ s for z is derived. The number of iron ion? responsible for the observed relaxation is Np = 7.1 x lo1’ Fe atom ~ r n - ~ with R = 15 A. At this stage, it is of course interesting to see whether the iron remaining unaccounted for (21.9 x 1019 Fe atom ~ m - ~ ) could be responsible for the relaxation of the broader signal. In doing so, it is then implicitly assumed that these randomly distributed iron species and the associated protons are present in a separate non-kaolinite lattice.As discussed above, it seems unlikely that the mica present in the sample constitutes this second phase which, moreover, is less abundant here than inW. E. E. Stone and R-M. Torres-Sanchez 125 t Fig. 5. Normalized total line area for the various examined samples. GB 1. Furthermore, spin-lattice relaxation times measured on mica containing a corresponding iron concentration are at least an order of magnitude larger. In any case, regardless of the details, when the same procedure used for the main proton line is applied to the broad components found in the various samples, the K values, calculated using eqn (3) and the remaining iron, are systematically too high (up to an order of magnitude).One possible explanation for the observed q of the broad component is that instead of being evenly distributed, the remaining FeIII ions are concentrated in certain regions without, however, being clustered on adjacent sites. In this configuration, each FeIII ion will retain its exclusion barrier b, yhich, according to the results obtained from the main line, should be of the order of 6 A. The number of protons which can still be detected in these areas will be relatively small and will correspond to those located in a small volume delimited by spheres having a radius b,. These protons, however, will now all be practically equivalent from a spin-lattice relaxation point of view. The existing distribution of relaxation rates should be narrow, and consequently the observed resultant is exponential with respect to time.As explained above, for this broad component the uncertainty on each experimental point taken during the relaxation process is such that the exact time behaviour for the recovery is difficult to ascertain. The results presented in fig. 4 for the broad components are those extracted from a single t dependence. For an exponential dependence, the theoretical expression for the relaxation rate is then given by the expressionla From the experimental K at 90 and 24 MHz, the parameter C can be estimated ; taking Np as the number of FeIII ions remaining after subtraction of those contributing to the relaxation of theomain line, a value for b, can be obtained from eqn (7).An average value for b, of ca. 6 A is found. The origin of the broad component could, therefore, be associated with accidents in the even distribution of the FeIII ions. This heterogeneity126 Structural Iron in Kaolinites defines with respect to the main kaolinitic lattice certain regions which, depending on their importance and localization, could reflect (or cause) perturbations in the regular structure of the mineral. It should be stressed that the line of thought developed above is indicative only of a certain general trend. With natural samples variations or exceptions to the general rule are to be expected. In the case of sample GB1 for instance, although the total iron content is too high to explain the observed q, unlike the other samples (sample K excepted) no broad component is observed.In fig. 4 it is seen that the broad component & of sample P3 is out of range of those found for the other samples. Such variations from idealized behaviour also show up when the normalized areas of the n.m.r. lines are examined. This is done in fig. 5, which gives as function of the degree of substitution of the A1 atoms by Fe, the variation of the total peak areas (normalized with respect to experimental conditions) referred to the area of the almost iron-free sample K. It is observed that the samples follow two general trends except GB 1. The slope of the curves is related to the number of protons excluded from observation. The distribution of samples on different curves could be due to the presence of paramagnetic clusters, i.e.Ferrr ions localized on neighbouring sites such that R < b,. The number of protons removed from observation will increase with the number or/and size of the clusters. The probability of having adjacent FeIrr ions increases with the iron content; however, as discussed above, the distribution of this element is not entirely ruled by random laws and therefore iron clusters may be present whatever the total iron content. Clusters have a negligible influence on the observed relaxation rates of the main lines; however, their effect on the intensity of the detected signals constitutes an additional parameter, revealing the heterogeneous character of certain of these samples. Lineshape and Second-moment Analysis In solids, n.m.r.lineshapes can be characterized in terms of their moments. The second moment M2 in particular provides information about both the geometrical arrangement of the nuclear spins on lattice sites and the type of spin interaction. The second moment of a line due to I spins will contain a contribution from like spins (Mi') and a contribution from unlike spins (Mis) such that Mi = MF + Mis. Given a model for the distribution of spins and their respective interactions, the measured Mi can be compared with a calculated value. Such an approach may contain a series of pitfalls as amply exemplified by our study conducted on natural crystalline micas.22 Contrary to micas where the OH groups could be substituted in a non-random fashion by fluorine, in kaolinite only one I-spin species is present, i.e.'H, I = i. This should in principle make things simpler. Whatever the case, it is of interest to be able to measure selectively both terms of the total Mi. Here Mis should contain not only the H-A1 contribution but also eventually an H-Fe contribution. When using a pulse spectrometer, the Mi of lines in solids is usually difficult to estimate accurately because of the dead-time following the observation pulse during which part of the signal is lost. To circumvent this problem and almost achieve zero time resolution, a 'solid-echo' technique can be used. It is also possible by modifying the pulse sequence necessary to obtain the solid echo, to identify the various contributions of Mi. The theoretical derivation of these effects can be found e l s e ~ h e r e .~ ~ , ~ ~ Estimation of a Total Second Moment Value Mi Fig. 6 shows, for the two purest samples, a semilogarithmic plot of the proton normalized echo amplitude as function of the square of the time taken from the echo maximum. The echo signal is obtained by a (90°-~-900,,0) pulse sequence. For short z values, the echo shape does not deviate from the free induction decay (obtainedW. E. E. Stone and R-M. Torres-Sanchez 127 0 1 5 10 t 2 / 1 0-'0 s2 1 .o n 0, 4 4 \ c v n 0 1 5 10 t 2 / 10-'0 s2 Fig. 6. Semi-logarithmic plot of the normalized echo amplitude as function of t2 for samples K (a) and GBl (b). @, 90 MHz; 0, 24 MHz. following a single 90" pulse). For these two samples, K and GBl , the points fit quite well a Gaussian function over nearly 80% of the decay so that Mi can be determined to a very good approximation from the initial slope of the curves, given in fig.6, by 1 3 2 ! 4! E( t)/E(O) = exp ( - Mi t 2 / 2 ) = 1 --Mi t2 + - (Mi)' t4 + . . . . Identical results are obtained at both 90 and 24 MHz. For samples with a higher iron content (see fig. 7), the same behaviour is observed except at short t , where a second component with a much higher value for Mi appears. In all cases, the contribution of this second component to the total observed signal does not seem to be very important (apparently 10-15°h, see fig. 7), although an estimate cannot be given because of the difficulty of recovering very broad lines with finite values for z (its second moment is ca.50 G'). As seen in table 3, it is quite remarkable 5 FAR I128 Structural Iron in Kaolinites 0 1 5 10 t 2 / 10-’0 s2 Fig. 7. Semi-logarithmic plot of the normalized echo amplitude as function of t2 for sample FU8 at 90 MHz. Table 3. Second-moment determinations sample )J,2R”(T) echo decay at z = 0 total second moment Mt/G2 hetero- contribution MiS/G2 K GBI P1 P4 FU8 vw, 5.9 6.0 5.5 6.0 5.5 5.4 6.1 6.0 4.3 4.6 6.1 7.0 1.5 1.6 1.6 1.5 1.5 I .5 that, except for FU8, all samples irrespective of frequency have a value for Mi between 5.5 and 6.1 G2, in the range of the calculated Mi for kaolinite (5.5 G2) by Otero-Arean et a1.12 The appearance of a second component in these echo experiments also coincides with what was observed in the If, for short z values, the shape of the echo is not strongly distorted, the amplitude of the echo’s maximum E(z) is a rapidly decreasing function of z.In the case of two spins I and S, the general expression for the echo intensity following the pulse sequence (90°-z-900,00) is E(z, t ) = &,[I- Mi’(t - ~ ) ~ / 2 ! - Mis (t2 + z2)/2! + M4(t - ~ ) ~ / 4 ! +error terms in t2z2] where t is the time measured from the second pulse. The echo is formed at t = z. From eqn (9) it can be seen that a value for the total second moment of the resonant line can be obtained by plotting the second derivative of R = E(T, t)/Eo, - (d2R/dt2),_ = R”(z), as function of T~ and extrapolating to zero. This is illustrated for sample FU8 in fig. 8. As shown in table 3, similar values for Mi to those derived from fig. 6 and 7 are obtained by this method.experiments, (9)W. E. E. Stone and R-M. Torres-Sanchez 10 a 129 0 1 5 10 ?/ 1 0-1° sz Fig. 8. Second derivative of the normalized echo signal at t = z as function of the square of the pulse separation z for sample FU8 at 90 MHz. Estimation of the Homo- and Hetero-nuclear Contributions to M', As shown by eqn (lo), the echo formed at t = z has its amplitude attenuated, for small enough values of z, by the heteronuclear IS interaction : E(z)/E, z 1 -f(2M',')z2. (10) MiS can therefore be extracted by examining how the amplitude of the echo varies with ?. Two examples are shown in fig. 9. The normalized intensity shows a Gaussian dependence with z2 over at least 85% of the decay. A more rapid decay at short times is observed as before for all samples except K and GB1.The values obtained for Mis from the long-time behaviour are given in table 3 and are all ca. 1.5 G2. This corresponds to the H-A1 contribution as estimated by Otero-Arean et a1.12 The MiS value is measured here directly, whereas in the case of the latter authors, it was deduced by the difference between the measured total second moment and the homonuclear contribution Miz determined by them via a multi-pulse Carr-Purcell sequence. With a simple solid echo sequence of the form (90°-z-1800), M',' can also be derived by a relation similar to eqn (10). However, for finite values of z, the formation of a well defined echo requires that Mis > Miz, which is not the case here, as demonstrated experimentally for all our samples on which the latter pulse sequence was tried.From table 3, it appears that, on the whole, iron does not perturb the local kaolinite order as seen by the protons : the second moment is (except for FU8) invariant whatever the iron content. The fact that the echo decays are essentially Gaussian is also indicative that the local fields acting on the proton spins are uniformly distributed. A similar conclusion can be drawn from some recent 27Al and 29Si n.m.r. measurements performed on kaolinites of different degrees of crystallinity and which show no variation in the ~pectra.'~ In our study, however, for samples having 0.5 % Fe,O, or more, this homogeneous distribution of spins is locally perturbed. In these regions, protons have very broad lines which can only be associated with their interaction with FeI'I..The estimated second moments for these broad components are much higher than those measured in the case of micas with comparable iron contents. At these iron concentrations, the iron distribution is therefore no longer uniform in the crystallites. In the regions where part of the iron is concentrated, the local paramagnetic character is certainly reinforced ; the interactions will now depend on the orientation of the protons with respect to the local paramagnetic 5-2130 Structural Iron in Kaolinites a 0 1 5 10 15 20 I I I I 7 2 / 10-10 s2 0 1 5 10 15 20 T2/1O-'O S2 Fig. 9. Attenuation of the amplitude of the echo signal as function of T ~ . a, 90 MHz; 0 , 2 4 MHz. Results for sample K (a) and P4 (b).system of axes. In this case, even if the g factor is isotropic, a substantial increase of the proton second moment is to be expected26 compared to what can be estimated for a random diluted impurity situation2' (in our case, an additional 2-4 G2). N.M.R. Results and the Problem of Structural Disorder in Kaolinites X-Ray diffraction studies4 infer that the main type of disorder in kaolinite is the displacement from one layer to the other (or even within the same layer) of the vacant octahedral site. 1.r.' e.s.r. and Mossbauer6-11 spectroscopies have been used to try to establish more ' local ' disorder indices using certain characteristics of the relevant observed spectra. An exact description of why these determined parameters are in fact modified by the lack of perfect order existing within the kaolinitic framework is still to be worked out.The n.m.r. results-presented here provide us with a complementary picture describing how the proton and iron populations are grossly distributed relative to each other. Because of the heterogeneity of situations probably existing within a badly crystallized sample and because all techniques only supply a response corresponding to an average taken over a broad distribution of possibilities, the quest for a minute description (in terms of vacancy displacements, dislocations, deformed sites etc.) isW, E. E. Stone and R-M. Torres-Sanchez 131 probably illusory. General trends can be established, each technique supplying a characteristic feature of the whole.The samples examined here have been studied previously by e . ~ . r . ~ and it is interesting to see whether any convergent conclusion can be drawn from the results of both techniques. Using e.s.r., several authors have established that in the region of g = 4, the observed signals (an isotropic I line plus a triplet E signal) are due to FeIII ions substituting A1 in an octahedral position. Mestdagh et aL7 clearly demonstrated an inverse relationship betwen the intensity of the isotropic g = 4.2 resonance (the I signal) and the degree of disorder of the sample. The claim is therefore that an interdependence exists between the proportion of defects and the concentration of FeII’ ions in the I sites. Broadly speaking, the n.m.r. results have shown that, with respect to iron, hydroxyls fall into two populations: (I) those located in an environment where iron is randomly distributed on the lattice sites and (2) those found in regions where iron is in a more regrouped situation.If one assumes that these latter regions and the iron associated with them are somehow connected with heterogeneities which destroy the regular pattern of the crystal then the iron responsible for the e.s.r. I signals7 and that determined by n.m.r. should be of the same type. The two sets of results are, however, quite different, which is perhaps is not surprising in view of the various approximations made on both sides. With respect to the H.I. index, the general trend is the same except at high iron content. Discrepancies in this region could be due to the fact that it was assumed here that all of the iron ions are located on octahedral sites, whereas some of them could be present in minute non-extractable iron contaminants.These cannot be detected in the proton n.m.r. spectra. On the other hand, a recent study on some 20 samples by Brindley et a1.’ concludes: (1) that the I signal is appreciable even for the most well ordered kaolinites (which are low in total iron content); (2) that insofar as actual I and E areas are concerned both correlate inversely with the degree of order of the samples and (3) for well ordered samples (having an H.I. 2 1) there is no correlation between the total iron content and the degree of order: ‘the mere introduction of iron ions into the octahedral layer does not necessarily lead to a structural defect in the kaolinite framework ’.Considering the results of the above n.m.r. study, this last sentence can be rephrased: ‘as long as the iron ions are randomly distributed in the framework, the probability of having a distorted kaolinite is low ’. The problem of structural disorder in kaolinites in relation to the presence of iron is therefore still open. Instead of taking the view that iron somehow causes an increase of lattice disorder it could equally be said that the increase in iron content which is observed to parallel the disorder found in kaolinites does not cause the loss of crystallinity but accompanies it during the formation of the mineral. This work was supported by the following organizations : M.R.A.C., F.N.R.S., S.P.P.S., F.D.S.-U.C.L.of Belgium. References 1 Bull. Mineral., 1982, 105, 41 3-58 1. 2 W. B. Jepson, in Iron in Soils and Clay Minerals, NATO Advanced Study Institute (Bad Windsheim, West Germany) (to be published). 3 A. J. Herbillon, in ref. (2). 4 C. Tchoubar, A. Plancon, J. Ben Brahim, C. Chard and C. Sow, Buff. Mineral., 1982, 105, 477. 5 J. M. Cases, 0. Lietard, J. Yvon and J. F. Delon, Bull. Mineral., 1982, 105, 439. 6 P. L. Hall, Clay Minerals, 1980, 15, 321. 7 M. M. Mestdagh, A. J. Herbillon, L. Rodrique and P. G. Rouxhet, Buff. Mineral., 1982, 105, 457. 8 S. A. Fysh, J. D. Cashion and P. E. Clark, Clays Clay Minerals, 1983, 31, 285. 9 G. W. Brindley, Kao Chich-Chun, J. L. Harrison, M. Lipsicas and R. Raythatha, Clays Cfay Minerals, 1986, 34, 239.132 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 Structural Iron in Kaolinites M. Cruz-Cumplido, C. Sow and J. J. Fripiat, Bull. Mineral., 1982, 105, 493. R. Prost, Agronomia, 1984, 4, 403. C. Otero-Arean, M. Letellier, B. C. Gerstein and J. J. Fripiat, Proc. Znt. Clay Conf., ed. H. Van Olphen and F. Veniale (Elsevier, Amsterdam, 1982), p. 73. D. Hinckley, Clays Clay Minerals, 1963, 11, 239. A. Abragam, Principles of Nuclear Magnetism (Clarendon Press, Oxford, 1961). N. Bloembergen, Physica, 1949, 15, 386. [. J. Lowe and D. Tse, Phys. Rev., 1968, 166, 279; D. Tse and I. J. Lowe, Phys. Rev., 1968, 166, 292. D. Tse and S. R. Hartmann, Phys. Rev. Lett., 1968, 21, 511. M. R. McHenry, B. G. Silbernagel and J. H. Wernik, Phys. Rev. B, 1972, 5, 2958. D. Olivier, P. Lauginie and J. J. Fripiat, Chem. Phys. Lett., 1976, 40, 131. B. Stubner, H. Knozinger, J. Conard and J. J. Fripiat, J. Phys. Chem., 1978, 82, 1811, J. Haupt and W. Muller-Warmuth, Z. Naturforsch., 1967, 22, 643. J. Sanz and W. E. E. Stone, Am. Mineral., 1979,64, 119; J. Sanz and W. E. E. Stone, J. Phys. C, 1983, 16, 1271; J. Sanz and W. E. E. Stone, Clay Minerals, 1983, 18, 187. P. Mansfield, Prog. NMR Spectrosc., 1971, 8, 43. M. Engelsberg and R. E. Norberg, Phys. Rev. B, 1972, 5, 3395. S. Komarneni, C. A. Fyfe and G. J. Kennedy, Clay Minerals, 1985, 20, 327. J. A. Ibers, C. H. Holm and C. R. Adams, Phys. Rev., 1961, 121, 1620. T. G. Stoebe, T. 0. Ogurtani and R. A. Huggins, Phys. Rev., 1965, 138, 239. Paper 71009; Received 2nd January, 1987

ISSN:0300-9599    

DOI:10.1039/F19888400117

出版商:RSC    

年代:1988

数据来源: RSC

摘要:

J . Chem. SOC., Faraday Trans. I , 1988, 84(1), 133-149 Electrokinetics of Polyelectrolyte Solutions in Capillary Tubes Hans Vink Institute of Physical Chemistry, University of Uppsala, Box 532, S-751 21 Uppsala 1, Sweden The electrokinetic behaviour of polyelectrolyte solutions in capillary tubes has been investigated. The experiments were carried out with polystyrene sulphonate and carboxymethyl cellulose polyelectrolytes in glass capillaries with diameters in the range 0.3-1.5 mm. The electrokinetic effects were very large compared to similar effects in simple salt solutions and could not be explained in terms of the classical theory of electrokinetic phenomena. To interpret the electrokinetic behaviour of polyelectrolyte solutions a more general theory based on non-equilibrium thermodynamics was developed, which was found to be in essential agreement with the experiments.According to the theory, the main cause of the large electrokinetic effect is the high wall-friction coefficient of the polyion, which is a manifestation of large frictional interactions in the polyion matrix. Electrokinetic phenomena involving the transport of water and electric charge in porous media or capillary tubes constitute a well known part of classical electrochemistry. The classical treatment has, in general, been concerned with simple salt solutions, for which an acceptable theory has been developed based on the concept of the Helmholtz double layer and the zeta-potential. However, the effect has an appreciable magnitude only in very fine capillaries or pores, thus impeding the experimental studies of the phenomena.For polyelectrolyte solutions very few studies of electrokinetic phenomena exist.' The present investigation indicates that the effect is much more pronounced in polyelectrolyte solutions and has a different character, requiring a new theoretical approach for the interpretation of the effect. Theory Classical Treatment The theory of electrokinetic flow in capillary tubes was first given in its present form by von Smoluchowski.2 The resulting equation for electro-osmotic flow is valid for large electrokinetic radii (large capillary radius compared to the thickness of the double layer) and has the form E DC R2 J, = - L E - - gradp r 81 where J, is the volume flux, R the capillary radius, q the solution viscosity, p the hydrostatic pressure, E the electric field strength, E~ the permittivity of vacuum, D the dielectric constant and 5 the zeta-potential.For the electro-osmotic counter-pressure we obtain from eqn (1) where P is the reduced electro-osmotic pressure. 133134 Electrokinetics in Capillary Tubes More recently various modifications and refinements of the theory have appeared, 3-8 which extend the range of validity of the original theory and yield detailed information about the velocity field in the capillary. Although eqn (1) and (2) are in good agreement with experiments in simple salt solutions, they are at variance with the electrokinetic behaviour of polyelectrolyte solutions. We therefore consider an alternative approach based on the frictional formalism of non-equilibrium thermodynamics.Theory based on Non-equilibrium Thermodynamics In a previous articleg the theory of electrokinetic phenomena, based on the frictional formalism, and specifically concerned with flow processes in a gel matrix, was developed. With minor modifications this treatment can be generalised to cover flow process in capillary tubes. Starting with the equation for balance of force, we have for the ith ionic species (i = 1,. . . ,n) fiw(Vi - 0,) C, +Lo vi + C fij(vi - vj) C, = zi FE - 8, gradp Cf,w(vw - v i ) ci +fwo o w = - o w gradp (3) (4) wheref-,,f, etc. are the molar friction coefficients between the components specified by the indexes (in N m2 s mol-2) v, and v, are the average velocities, c, and c, are the concentrations (molmT3) and oi and b, the partial molar volumes of the respective components.E is the electric field strength, p is the pressure, zi is the charge number of the ith ionic species and F is the Faraday constant. Friction between the mobile components and the capillary wall is represented by the coefficients fi0 and fw0 (in N s m-' mol-l), which differ dimensionally from the coefficients f,, and fij, because the concentration of the wall component, which is undefined, has been omitted. It has been showng that interionic friction between the different ionic components represents a cooperative process, the effect of which can be lumped into a common factor. This leads to a considerable simplification of the treatment. Thus, introducing the i and for water i fluxes Ji = c,v, and J , = c,v, we obtain Rii Ji + R,, J , = zi( 1 - a) FE - (a, - z, B) C RWi Ji + R,, J , = - ow gradp i where The two parameters a and /? determine the cooperative energy dissipation due to interionic friction.Because they are linear functions of the ionic concentrations ci they may often be neglected in dilute solutions. For convenience, we therefore put a = B = 0 in the following.H. Vink Eqn (6) and (7) may be solved for the fluxes Ji and Jw, and we obtain 1 3 5 where vw + c $hi vi R W W U - 0) S -gradp( ) Jw = EF R ~ ~ ( I - e) s = c a e=c-. R;W i R i i R w w ( 1 4 ) The directly observable fluxes in electrokinetics are the electric charge flux I and the From eqn ( 1 1 ) and ( 1 2 ) we obtain volume flux J,. ( o w + C $i V J 2 j i i Ri, ' w w ( l - @ * ( 1 7 ) A further simplification of these equations is achieved when use is made of the fact ( u; Jv = JwUw+c J i V i = EF -gradp C--+ that f w o , f , o .4 L w c w = L where we introduce the notationf, for the hydrodynamic friction coefficient of the ith component.The relations in eqn ( 1 8 ) may be anticipated, considering that the coefficients to the left represent a surface effect, whereas the coefficient f;: refers to friction between molecularly dispersed components. To substantiate the statement in eqn ( 1 8 ) we evaluate the coefficients involved for some typical cases. From the treatment of viscous flow in the frictional formalismlo we find that when the flow is perfectly uniform, with no sieving effects (all components have the same average velocity), the friction coefficients of the different components are simply related to the Darcy law friction coefficient : f w o = f f i w ; f , o = f v i (19) where f satisfies the Darcy law fu = -gradp where v is the common average velocity of the components.For Poiseuille flow in a tube with radius R we have136 Electrokinetics in Capillary Tubes solutions used in the experiments) we find: For R = 0.15 mm and q- = 0.2 N s m-2 (which is a typical value for the polyelectrolyte f = 7.1 x 10' (N s mV4). (22) For a hydrated Na+ ion the estimated radius'' is 0.33 nm, and the partial molar Thus, according to eqn (19) we obtain for Na+ ions volume has an approximate value of oNa+ = 9 x loF5 m3 mol-l. Lo = 6.4 x lo3 (N s m-l mol-l). (23) The hydrodynamic friction coefficient of an ion can be obtained from the molar conductivity of the ion (A, = F2zf/f,).For the Na+ ion (Ao = 50.1 x lo-' m2 mo1-' S2-l) we have fa = 1.86 x 10l2 (N s m-l mol-I). A comparison of eqn (23) and (24) demonstrates the overwhelming dominance of hydrodynamic friction over the wall friction effect. This explains why a solution in capillary flow in general behaves as a simple fluid, without the occurrence of any detect able sieving effects. Using eqn (18) we obtain the relations $i = Ci/Cw (25) 8," + C $hi 0, = 1 /cw a Rii = f , / c i where Note that eqn (29) is consistent with eqn (19). For the parameter S we obtain (30) 1 1 zicifio %-- c Z( Ci(l -fao/fi) = -- C- Cw i cw t fa where the condition of electroneutrality has been used to obtain the last equality.Inserting these relations into eqn (1 6) and (1 7), we obtain 2, c, v, where i J i is the electrolytic conductivity. The two terms Z . C. is. ci vi2 C+ and c~ i J i i J i in the expression for the volume flux represent specific contributions due to the ionic volumes. These terms are small and may in general be neglected.H. Vink 137 Considering now the solutions of an anionic polyelectrolyte, we assign to the polyion the index 1 and to the counterion the index 2. Expressing the concentration of the polyion in terms of the equivalent concentration C (i.e. the molar concentration of an equivalent amount of charged univalent groups), we have for univalent counterions 21 = -22 = - 1; c, = c2 = c. From eqn (30) we then obtain where we denote the difference of the friction coefficient ratios by O.This parameter obviously determines the direction and magnitude of the electro-osmotic volume flow. Using eqn (33) and neglecting the small ionic volume terms in eqn (31) and (32) the electrokinetic equations take the form (34) FOC I = KE-- gradp f J " = f F°CE-lgradp. f For the reduced electro-osmotic pressure we obtain (35) Pronounced electrokinetic effects are expected when large differences in the wall- friction effects for the two ionic species exist. The experiments indicate that the wall- friction effect for the polyion is always larger than the corresponding effect for the counterion. Since direct friction between the wall and the polyions only occurs at the surface of the capillary, while the coefficient flo refers to the average friction force in the bulk solution, the effect can be large only if the force at the capillary wall is transmitted into the solution.The only conceivable mechanism for such interactions involves friction between different polyions, which can be characterised as a viscous interaction within the polyion matrix. The treatment of this effect requires the introduction of velocity gradients, and thus extends the treatment beyond the domains of conventional non- equilibrium thermodynamics. However, as it is essential for the understanding of the wall-friction effect, a simple model for such interactions is considered in the Appendix. For a proper understanding of electro-osmosis it is important also to consider the flux of the polyion.With the same assumptions as before, we obtain from eqn (1 1) The second term in the first parenthesis represents the electro-osmotic contribution to the electrophoretic transport of the polyion. We have f,.c = f10 c- Vlf2O/f2) c < 1 . f fwo cw +fiO C+f20 c (38) Thus, we find that in electro-osmosis (grad p = 0) the polyion transport always occurs in the direction of its electrophoretic mobility.138 Electrokinetics in Capillary Tubes Experiment a1 Apparatus Details of the measuring cell are shown in fig. 1. It consisted of a U-shaped precision- bore capillary tube, connected to wide Pyrex tubes (12 mm diameter). Into these tubes end-pieces (made of lucite and provided with O-ring seals) could be inserted, which carried the measuring capillaries (with 1.5 mm diameter) and the electrodes.The latter were ion-exchange membrane electrodes (carboxymethylated cellophane with a low degree of substitution12). In general the same solution was used on the two sides of the membrane, which prevented the contamination of the solution in the cell by foreign ions. Although there is an electro-osmotic flow of water through the membranes, this was negligible compared to the flow in the capillary (calculations using the electro-osmotic coefficient of the membrane12 indicate that the flow in the membranes was less than 1 % of the flow in the capillary). Cells with three different capillaries were used and data are listed in table 1. The cell could be used in the horizontal position, to measure the electro-osmotic volume flux, or in the vertical position, to measure the electro-osmotic counter-pressure.In the former case the cell was placed in an air thermostat, whereas in the latter case a water thermostat was used. All measurements were carried out at 25 "C. The positions of the liquid levels in the measuring capillaries were determined with a cathetometer (vertical cell), or a comparator (horizontal cell). The current was supplied by a LKB 2197 constant-power supply. The electric circuit contained a standard resistance of 100 k a in series with the cell, and the current through the cell was determined by measuring the potential over this resistance with a Keithley 136 digital multimeter. The circuit also contained a switch by which the current in the cell could be reversed. Materials The experiments were carried out with polystyrene sulphonate (PSS) and carboxymethyl cellulose (CMC) polyelectrolytes.The PSS samples were obtained by s~lphonating'~ narrow-molecular-weight polystyrene (from Pressure Chem. Corp., Pittsburg). Two samples, PSS 1 and PSS 2 were prepared from polystyrene with molecular weights M, = 293000 and 1.86 x lo', respectively. They were used in the form of acid (HPSS) and sodium salt (NaPSS). CMC was only used in the form of the sodium salt. The NaCMC sample had a degree of substitution of 0.96. The intrinsic viscosity in 0.2 mol . dm-3 NaCl was [71] = 1170 cm3 g-', corresponding to a molecular weight of M, = 9.8 x lo5, according to ref. (14). All polyelectrolyte samples were carefully purified by dialysis. The conductivity water had an electrolytic conductivity in the range (0.4-0.6) x lo-* C2-l m-l.Results and Discussion Electr o-osmosis The electro-osmotic measurements were carried out with the two measuring capillaries in a horizontal plane. To compensate for possible differences in capillary levels the flow was determined as the mean value of two consecutive runs, with the current reversed. The direction of flow was always in the direction of counterion movement, i.e. in the direction of the electric field for anionic polyelectrolytes. The general consistency of the measurements is demonstrated in fig. 2, where the Ohmic behaviour of the measuring system is displayed. The minor differences between measurements in different capillariesH. Vink 139 Fig. 1. Measuring cell with end-piece containing current supply electrode and measuring capillary.A, ion-exchange membrane ; B, platinum electrode and C , measuring capillary. Table 1. Data for the measuring cells capillary capillary cell no. diameter/mm length/cm 1 0.3 10.6 2 0.75 19.0 3 1.5 34.7 are attributed to uncertainties in the capillary dimensions, This consistency of the data rules out the existence of large adsorption and surface conduction effects in the systems investigated. When comparing the electro-osmotic volume flux in different cells the uncertainties in cell dimensions can be eliminated by considering the ratio J,/I. As demonstrated in fig. 3, the ratio J,/I was found to be independent of the electric field and the capillary radius. However, deviations from this behaviour occurred at low polyelectrolyte concentrations. In general, the ratio J , / I increased with decreasing polyelectrolyte concentration (the reversed order of curves 1 and 2 in fig.3 is due to the much higher conductivity of HPSS as compared to NaPSS). The concentration dependence of the electro-osmotic volume flux is displayed in fig. 4 and 5, where the reduced volume flux J,/E is plotted against concentration for the different polyelectrolytes investigated. This quantity decreases with increasing con- centration, rapidly at very low concentrations, more slowly at higher concentrations. An interesting feature is that it has approximately the same magnitude for all polyelectrolytes studied, notwithstanding the very large difference in viscosity of the solutions concerned.This behaviour is not in agreement with the classical theory of electrokinetic140 Electrokinetics in Capillary Tubes 40 - 30 - - I E 4 6 IS --.- N n --.- 20 - "0 500 1000 1500 VIV Fig. 2. Ohmic behaviour of the measuring system. I is the total electric current, I the capillary length and r the capillary radius. + , cell 1 ; a, cell 2 and 0, cell 3. Upper curve for HPSS2 at C = 1.690 equiv. m-3 and lower curve for NaPSS2 at C = 1.620 equiv. mP3 phenomena. In particular, according to eqn (1) the reduced volume flux should be much lower for sample HPSS2 than for sample HPSS 1, because the solution viscosity increases rapidly with molecular weight, whereas [ is expected to be independent of molecular weight. Also, the computed value of ([ x 25 V for HPSS2) is many orders of magnitude higher than the value normally considered in electrokinetics ([ < 100 mV).On the other hand, the frictional model treatment gives a plausible explanation of the behaviour of the electro-osmotic volume flux. According to eqn (35) the reduced volume flux is determined by the ratio aC/' The finding that this ratio is practically independent of the solution viscosity and the capillary radius gives useful information about the structure of the friction coefficients. In eqn (33) we may neglect the small counterion friction term and then obtain From eqn (21) and (29) we obtain aC = f10 C/fl* (39) The approximate constancy of the ratio aC/fimplies that flo C is the dominant term in the expression for f. Consequently the solution viscosity is directly related to the polyion wall-friction coefficient and thus [according to eqn (A 11) in the Appendix] to the friction coefficient fll of the polyion matrix itself.Assuming we obtainH. Vink 141 3 0 2 0 * - t- 0 U 0 0 500 1000 1500 VIV Fig. 3. Dependence of the ratio J,/I on the applied potential difference V and the capillary radius. +, cell 1 ; 0, cell 2 and 0, cell 3. Curves 1 and 3 represent HPSS2 at the concentrations C = 1.690 and C = 0.682 equiv. m-3, respectively. Curve 2 represents NaPSS2 at C = 3.90 equiv. m-3. which yieldsf, 1.4 x 1OI2 N s m-' mol-' for the polyelectrolytes investigated. This is in good agreement with thef, value for Na+ ions in eqn (24) (from transport number measurements12 it follows that the hydrodynamic friction coefficient for the polyion equivalent is of the same order of magnitude as that of the counterion).The decrease of the reduced volume flux with increasing concentration is [according to eqn (42)] due to an increase in the hydrodynamic friction coefficient. This behaviour is in agreement with the concentration dependence of the hydrodynamic friction coefficient of linear polymers, found in sedimentation and diffusion. l5 From these considerations it is obvious that knowledge of solution viscosity is essential for the interpretation of the electrokinetic behaviour of polyelectrolyte solutions. The viscosities of some of the polyelectrolyte solutions were therefore determined. The cells in table 1 were used as capillary viscometers (using end-pieces without membrane electrodes), the measuring procedure being the same as that in ref.(16). The results are shown in fig. 6 and 7. In fig. 6 the concentration dependence of the apparent viscosity, for two different values of the velocity gradient q, is represented in a logarithmic plot. The gradient dependence of viscosity is moderate for low-molecular- weight HPSS and for NaCMC, but very pronounced for the high-molecular-weight HPSS sample, which is further illustrated in fig. 7. In these measurements q refers to the apparent velocity gradient at the capillary wall.142 Electrokinetics in Capillary Tubes Clequiv. m-3 Fig. 4. Concentration dependence of the reduced volume flux. a, HPSS2 in cell 2; +, HPSSl in cell 1. The applied potential difference was loo0 V. t I I I I I I I I 5 Clequiv.m-’ Fig. 5. Concentration dependence of the reduced volume flux. 0, NaPSS2 in cell 2; 0, NaCMC in cell 2. The applied potential difference was 1000 V.H. Vink 143 -1 n p1 E z v) --- W F M - -2 0 1 I I I I I I I I 5 Clequiv. m-3 Fig. 6. Concentration dependence of the apparent viscosity of the polyelectrolyte solutions. and 0, HPSS2 at the velocity gradients q = 0.2 and 2 s-l, respectively; A and A, NaCMC at q = 2 and 20 s-l, respectively; and 0, HPSSl at q = 2 and 20 s-l, respectively. 0.4 0.3 I I I 1 2 3 0 ' 0 4/s-' Fig. 7. Dependence of the apparent viscosity of HPSS2 at C = 3.160 equiv. m-3 on the velocity gradient.144 n E z Q. 4 200 100 40 0 300 - - Electrokinetics in Capillary Tubes t I 1 1 I I I I I 1 50 0' 0 tlmin Fig. 8.Temporal evolution of the electro-osmotic pressure difference for HPSS2 at C = 1.512 equiv. rnp3 in cell 2. Potential difference: 0, 250 V and 0, 500 V. 150 f , I I\ I inn I 50 I I I I I 1 I I L O! 50 t/min Fig. 9. Temporal evolution of the electro-osmotic pressure difference for HPSS2 at C = 2.826 equiv. m-3 in cell 3. Potential difference 0, 250 V and 0, 500 V.H . Vink 145 5 4 3 m I E 4 m . % 2 1 C 5 C/equiv. m-3 Fig. 10. Concentration dependence of the reduced electro-osmotic pressure for sample PSS2. + , 0, and ., HPSS2 in cells 1, 2 and 3, respectively. 0, NaPSS2 in cell 2. Applied potential difference was 200 V for cell 1 and 500 V for cells 2 and 3. Electro-osmotic Counter-pressure Although the theoretical expression for the reduced electro-osmotic pressure is simple and directly related to the parameter CT, the real situation is far more complicated. According to eqn (37) and (38) the volume flux in electro-osmosis is essentially due to the flux of water, the (anionic) polyion flux always being in the opposite direction to the applied field.In electro-osmotic counter-pressure experiments the water flux is practically zero, whereas the polyion flux is increased owing to the action of the pressure gradient. Thus, the velocity gradient in the polyion matrix is changed. Because of the close coupling between the parameter CT and the solution viscosity v, and the pronounced dependence of the latter quantity on the velocity gradient, may appreciably depend on the applied forces. Therefore non-linear effects are expected in the relationship between the applied electric field and the resulting pressure difference.Such effects were also found experimentally. Some typical measurements of the electro-osmotic pressure difference are shown in fig. 8 and 9. A characteristic feature is the existence of a hump in the pressure us. time curves. This is probably due to a time lag in the establishment of steady-state flow conditions in the polyion matrix. The effect was usually small, although pronounced effects were found in measurements with the widest capillary, fig. 9. In the steady-state region often a slow decrease of the pressure difference occurred, which was probably due146 Electrokinetics in Capillary Tubes 0.3 0.2 m E 4 2 v) 0.1 I I 1 I I I I 1 I I 1 I 5 10 01 0 Clequiv.m-3 Fig. 11. Concentration dependence of the reduced electro-osmotic pressure. + , HPSS1 in cell 1 ; 0, NaCMC in cell 2. Applied potential difference 500 V. to concentration polarization. This limited the accuracy in the determination of the electro-osmotic pressure difference. As a rule, the following measurements refer to a time immediately after the establishment of the steady state. It may be seen in fig. 8 and 9 that the pressure difference does not vary linearly with the applied electric field, but increases more slowly when the field is high, probably as a consequence of the velocity gradient effect. The concentration dependence of the reduced electro-osmotic pressure at constant electric field is shown in fig. 10 and 11. For sample PSS2 the curves are linear at first, but level off at higher concentrations.No significant difference between HPSS and NaPSS was detected. For HPSSl the curve levels off at very low concentrations and the electro-osmotic pressure difference is much below that of the high-molecular-weight sample. Similar behaviour was observed with NaCMC. Obviously a close correlation between the electro-osmotic pressure difference and the solution viscosity exists. Such a correlation is inadmissable in the classical theory, because in the stationary state the solution is macroscopically at rest and is hence in a state independent of viscosity. On the other hand, according to eqn (36), in the frictional model the correlation implies a relation between the quantities Q and q, which corroborates the finding in the electro-osmotic flow experiments that the polyion wall- friction term is the main factor determining the solution viscosity in eqn (40).Further support for this finding was obtained from a few tentative experiments carried out with polyelectrolyte solutions in the presence of simple salt. It was found that the solution viscosity and the electro-osmotic pressure difference were markedly depressed by the addition of salt, whereas only a small decrease in the electro-osmotic flow was observed. This implies that the term fi0 C and thus a and q decrease when salt is added, whereas the ratio a/f [or a/q according to eqn (40)] remains essentially unchanged.H. Vink 147 1 4, 5 0 - 1 - 1 - 0.5 0 log R Fig. 12. Double-logarithmic plot of the reduced electro-osmotic pressure 13r 1PSS2 as a function of capillary radius.0, C = 1 equiv. m-3; 0, C = 2 equiv. m-3 and +, C = 3 equiv. m-3. The broken line has slope -2. Finally, it is also of interest to consider the dependence of the electro-osmotic pressure difference on the capillary radius. In fig. 12 the electro-osmotic pressure data obtained from fig. 10 are represented as functions of capillary radius in a double-logarithmic plot. The slopes of the lines have numerical values in the range 1.60-1.67, which is lower than the value imposed by the classical theory according to eqn (2), but in agreement with the frictional model treatment, eqn (A 15). Appendix The treatment of the wall-friction effect is closely related to the treatment of viscous effects in the frictional formalism of non-equilibrium thermodynamics.lo The balance- of-force equations (3) and (4) are valid only for average values of the component velocities. If local velocities are considered the equations have to be supplemented by ' viscous ' friction terms, referring to interactions between molecules of the same species having different velocities. If the flow is laminar, and velocity differences between different layers exist, we may assume in accordance with the general ideas of the frictional model that the friction force between molecules in adjacent layers is proportional to the difference between their velocities divided by their mean spacial separation. This means that in the continuous limit the friction force is proportional to the velocity gradient.Considering the friction force Fi between ions of the ith species, we obtain for the force acting on an annular zone of unit length, between the radii r and (r+dr) in the tube148 Electrokinetics in Capillary Tubes dui dr dFi = - 2nrfii ci - + 2n(r + dr)fii ci =2nf.c aa i(2’ L + r - - T:)dr where u denotes local velocity andLi is the ‘viscous’ friction coefficient. Integrating by parts we obtain In electro-osmosis (grad p = 0) this force is proportional to the total electric force acting on the ith ionic species in the volume considered. Thus Fi = - kz, FEnr2ci (A 3) where k is the constant of proportionality. In general k 4 1, because Fi may be considered as a perturbation of the main force of hydrodynamic friction between the polyions and water. From eqn (A 2) and (A 3 ) we obtain dui - kziFEr - - -- dr X i .Integration yields kz. FE u . = 2 (R2-r2)+uR 4Li where R is the radius of tlie tube and uR is the slip velocity at tile wall. The average velocity in the tube is 1 kzi FER2 v, = n ~ 2 (6 ui2nrdr = +UR. 8f,i To estimate the magnitude of the slip velocity we assume that the force given by eqn (A 3) is transmitted to the wall by surface friction. Thus or kzi FER UR = _____ 2fR where fR is the surface friction coefficient. Inserting into eqn (A 6) we obtain We may now evaluate the wall-friction term in the balance-of-force equation, eqn (3), by equating the total wall-friction force in the bulk solution with the surface friction force. Thus nR2cifiOvi = 2nRci fRuR. (A 10) With the help of eqn (A 8) and (A 9) we obtainH. Vink 149 From this equation we may obtain two limiting scaling laws for the dependence of fi0 on R. If K i / f R ) 4 R we have 8Ai A 0 = 3' This is the case of viscous Poiseuille flow,fai representing the 'viscosity' of the ith component. In this case the slip velocity has an insignificant value, which follows from eqn (A 8) and (A 9): 1 In the other limit, when (fii/fR) 3 R we have 2fE f a 0 = R' In this case the flow resembles a plug flow with pronounced slippage at the wall. We may conclude that in intermediate cases the scaling law has the form Lo a R-*; 1 < n < 2. References 1 M. H. Gottlieb, J. Phys. Chem., 1971, 75 1981. 2 M. von Smoluchowski, Handbuch der Electrizitat und des Magnetismus (Graetz) (Barth, Leipzig, 1921), 3 D. Burgreen and F. R. Nakache, J. Phys. Chem., 1964, 68, 1084. 4 F. A. Morrison Jr and J. F. Osterle, J . Chem. Phys., 1965, 43, 211 1. 5 C. L. Rice and R. Whitehead, J. Phys. Chem., 1965, 69,4017. 6 D. Hildreth, J. Phys. Chem., 1970, 74, 2006. 7 W. Olivares, T. L. Croxton and D. A. McQuarrie, J. Phys. Chem., 1980, 84, 867. 8 V. Vlachy and D. A. McQuarrie, J. Phys. Chem., 1986, 90, 3248. 9 H. Vink, J . Chem. SOC., Faraday Trans. I , 1982, 78, 3115. 10 H. Vink, J. Chem. SOC., Faraday Trans. I , 1983, 79, 2355. 11 R. A. Robinson and R. H. Stokes, Electrolyte Solutions (Butterworths, London, 1955), p. 121. 12 H. Vink, J. Chem. SOC., Faraday Trans. I , 1984, 80, 1297. 13 H. Vink, J . Chem. SOC., Faraday Trans. 1, 1987, 83, 801. 14 W. Brown, D. Henley and J. Ohman, Makromol. Chem., 1963, 62, 164. 15 B. Nystrom and J. Roots, J . Macromol. Sci. Rev. Macromol. Chem., 1980, 19, 35. 16 H. Vink, Makromol. Chem., 1970, 131, 133. vol. 11, p. 366. Paper 71160; Received 29th January, 1987

ISSN:0300-9599    

DOI:10.1039/F19888400133

出版商:RSC    

年代:1988

数据来源: RSC

摘要:

J. Chern. Soc., Furuduy Trans. I, 1988, 84(1), 151-163 Temperature Dependence of Adiabatic Compressibility of Aqueous Solutions of Alkyltrimethylammonium Bromides Ryszard Zielinski'f and Shoichi Ikeda" Department of Chemistry, Faculty of Science, Nagoya University, Nagoya 464, Japan Hiroyasu Nomura and Shigeo Kato Department of Chemical Engineering, School of Engineering, Nagoya University, Nagoya 464, Japan The adiabatic compressibility of aqueous solutions of octyl-, decyl-, dodecyl- and tetradecyl-trimethylammonium bromides has been determined from measurements of density and ultrasound velocity at different temperatures ranging from 20 to 45 "C at 5 "C intervals. Based on the theoretical treatment in which the adiabatic compressibility is given as a function of concentration, the _apparent adiabatic cpmpressibilities of the surfactant in the monomeric (8,) and micellar (8,) forms are obtajned from the experimental results at each temperature.The values for 8, increase with increased temperature, and its temperature coefficient is constant for methyl, octyl and decyl derivatives, but changes sharply betw_een 30 and 40 "C for the dodecyl and tetradecyl derivatives. The value of 8, increases linearly with increased temperature, and its temperature coefficient inp-eases with increasing alkyl chain length. The linear changes of 8, and 8, with temperature are related to the change of hydrophilic hydration, while the sharp change of 8, for dodecyl and tetradecyl derivatives is attributed to the change of hydrophobic hydration. Several papers have appeared on the temperature dependence of the partial molar volume of surfactants in aqueous ~olutions,~-~ but the temperature dependence of the adiabatic compressibility of surfactant solutions has received considerably less attention.Nomoto and Endo5 have studied the temperature dependence of the adiabatic compressibility of aqueous solutions of polyoxyethylene dodecyl ether containing, on average, six oxyethylene groups. Vikingstad et a1.6 have reported pressure and temperature effects on the partial molar volume and adiabatic compressibility of sodium decanoate micelles, while De Lisi et aZ.' have examined temperature dependence of adiabatic compressibility of nonyl- and decyl-trimethylammonium bromides. Backlund et aL8 have investigated the temperature dependence of ultrasound velocity in aqueous solutions of hexadecyltrimethylammonium bromide.In a previous paperg we developed a simple theoretical treatment, in which the adiabatic compressibility of aqueous solutions of surfactants is given as a function of concentration, in order to obtain the apparent adiabatic compressibilities of the surfactant in both monomeric and micellar forms. We then applied it to the experimental results obtained from aqueous solutions of several surfactants at 25 "C. In the present work we measure the density and ultrasound velocity in aqueous solutions of alkyltrimethylammonium bromides in which alkyl = octyl, decyl, dodecyl or tetradecyl, and we derive the partial molar volume and the apparent adiabatic compressibilities of these surfactants in both monomeric and micellar forms at different temperatures.The temperature is varied from 20 to 45 "C at 5 "C intervals. t Permanent address : Department of General and Analytical Chemistry, Institute of Commodity Sciences, Academy of Economics, 60-967 Poznan, Poland. 151152 Compressibility of Surfactants Methods of Calculation In aqueous solution a surfactant can exist in either the monomeric or the micellar form. If the weight concentrations of monomer and micelle are c, and cm, the total concentration of surfactant, c (g ~ m - ~ ) , is given by c = C,+C,. (1) If the apparent specific volume of monomer and micelle are u", and u", (cm3 g-l) respectively, the density of the aqueous solution of the surfactant, p (g cmd3), is given (2) by P = PO + (1 - 61 P O ) c1+ (1 - u"m P O ) c m where po is the density of the solvent.By differentiating eqn (2) with respect to pressure, P, at constant entropy, S, the adiabatic compressibility of the solution, j? (bar-,), can be derived as a function of the c_oncent_rations. If the apparent adiabatic compressibilities of monomers and micelles are p1 and pm, respectively, and the changes in concentrations of monomer and micelle with pressure, P, take place only through a change in volume of solution with pressure, i.e. (ac,/ap)s = 8; c, (aCm/aP)s = B m c m - then the adiabatic compressibility of the solution, /I, is given by where Do is the adiabatic compressibility of the solvent. According to the Laplace equation and expanding the terms to the first power of concentrations, the ultrasound velocity in the solution, u (m s-l), is expressed by where uo is the ultrasound velocity in the solvent, and u, = l/p, is the specific volume of the solvent.In eqn (2), (3) and (9, the concentrations of monomer and micelle must be given as functions of total Concentration of surfactant. In most cases the pseudo-phase model holds well for the micelle f~rmation,'~-'~ which leads to c, = c, cm = 0 c < c.m.c. (6) c, = c.m.c. c, = c-c.m.c. c 3 c.m.c. (7) where c.m.c. is the critical micelle concentration. We can then expect to have two straight line segments intersecting at the c.m.c., if we plot the density, the adiabatic compressibility or the ultrasound velocity against the total concentration, and if the apparent specific volumes and the apparent compressibilities of the surfactant in the monomeric and micellar forms are independent of concentration.R .Zieliriski, S. Ikeda, H. Nornura and S. Kato 153 Experimental Materials Samples of tetramethylammonium bromide (C'TAB), octyltrimethylammonium bro- mide (C,TAB), decyltrimethylammonium bromide (C,,TAB), dodecyltrimethylam- monium bromide (C,,TAB) and tetradecyltrimethylammonium bromide (Cl,TAB) were the same as those used previo~sly.~ They were used as supplied, without further purification, but after having been dried in vacuo at room temperature for at least 48 h. All solutions were prepared by weighing, using distilled and degassed water. The surfactant concentrations were converted to g cm-3 at each temperature. Measurements Density measurements were carried out, using Ostwald type pyknometers of 20 cm3 capacity.Pyknometers were calibrated at each temperature using distilled and degassed water. The densities of the solutions were calculated from the generally accepted values of the density of water13 and were corrected for the density of air. Uncertainties in the solute concentration and the weighing (mainly the latter) can produce errors in the values of density, ca. 5 x Measurements of ultrasound velocity, u, ( f: 0.15 m s-l) were made using an ultrasonic interferometer working at a frequency of 5.0 MHz. Errors in ultrasound velocity measurements can arise mainly from variations of temperature, T. Since the temperature coefficient of ultrasound velocity, du/dT, in water varies from ca.3.0 m s-l K-' at 20 "C to ca. 1.4 m s-l K-l at 45 "C, the uncertainty of 0.01 "C in the temperature of the solutions during ultrasound velocity measurements produces an error in the values of u of ca. f3.0 x lo-, m s-' at 20 "C andf 1.4 x m s-' at 45 "C. This allows us to calculate the adiabatic compressibilities of the surfactant solutions to an accuracy of better than 0.03 %. g ~ m - ~ . Results The densities of aqueous solutions of alkyltrimethylammonium bromides (C,TAB) at each temperature are plotted as a function of the surfactant concentration. A typical plot for octyltrimethylammonium bromide is shown in fig. 1. Each plot can be divided into two straight line segments with a break point corresponding to the c.m.c. The straight line segments can be described by means of eqn (2).From their slopes the values of the apparent specific volumes of the surfactant in the monomeric form, v"', and that in the micellar form, v",, can be derived at each temperature. We introduce the apparent molar volumes of monomer and micelle by - Vl = M,v", vm = M#, - where M, is the molecular weight of the surfactant. In general, for dilute electrolyte solutions, the apparent molar volume is approximately equal to the partial molar volume. Values of the apparent molar volumes of the surfactants at various temperatures are tabulated in table 1. The change in the apparent molar volume upon micellization, APm, is defined as the difference between the values of the apparent molar volume of the surfactant above and below the c.m.c.: Values of AP, for the surfactants at various temperatures are presented in fig. 2. - - - AV,,, = Vm- Vl. (10)154 Compressibility of Surfactants Fig. 0 50 100 150 concentration/g dm-3 1. Density of aqueous solutions of octyltrimethylammonium bromide as a function of concentration at several temperatures. Symbols: A, 20; V, 25; 0, 30; 0, 35; 0, 40; H, 45 "C. Table 1. Apparent molaL volume of alkyltrimethylammonium bromides in monomeric ( and micellar (V,) forms in water at various temperatures (in cm3 mol-') C,TAB C,TAB C,,TAB C,,TAB C,,TAB T/ "C Pl Vl 'm Vl 'm '1 'm Vl Vm 20 - 223.7 228.1 253.4 260.7 279.9 293.6 305.9 326.2 25 115.2 225.0 228.8 255.0 262.2 283.4 296.2 309.6 329.4 30 115.3 226.3 229.4 256.4 263.4 285.7 297.7 312.0 328.9 35 115.7 227.0 230.4 258.3 264.5 287.1 299.4 315.2 331.8 40 116.1 228.1 231.4 260.5 266.4 288.6 300.2 318.0 332.5 45 116.5 229.3 232.1 261.2 267.2 291.9 301.4 320.4 335.7 Formation of micelles brings about an increase in the apparent molar volume of alkyltrimethylammonium bromides, which indicates that the structure of the micelles is looser than that of the monomers at each temperature.The change in the apparent molar volume due to micelle formation decreases with increased temperature, and this decrease seems to be linear in the range of temperatures investigated. A similar tendency was also observed for nonyl- and decyl-trimethylammonium bromide^,^ sodium dodecyl sulphate,' sodium decanoate6 and sodium sulphonate.2 One can suppose that the decrease in AVm with increased temperature may be brought about by dehydration of the ionic head group.R .Zieliriski, S. Ikeda, H. Nomura and S . Kato 155 24 20 16 d I - 0 E rn $ 12 1 2 --.. 4 8 4 0 15 20 25 30 35 4 0 45 5 0 T/OC Fig. 2. Change in the apparent molar volume upon micellization for alkyltrimethylammonium bromides in aqueous solutions as a function of temperature. Symbols: A, C,TAB; V, C,,TAB; 0, C,,TAB; 0, C,,TAB. In aqueous solution the apparent molar expansion, ii = (avi/aT),, (i = 1 or m) is ex- pected to be a Fseful and sensitive measure of the structural solute-solvent interactions.14 The value of Ei increases with the length of the alkyl chain from 0.212 cm3 mol-' K-' for the octyl to 0.576 cm3 mol-' K-' for the tetradecyl derivative in the monomeric form (i = I), and from 0.193 cm3 mol-1 K-' for the octyl to 0.380 cm3 mol-1 K-' for the tetradecyl derivative in the micellar form (i = m).The higher value for the monomer would suggest stronger hydration for the monomer than for the micelle. Fig. 3 and 4 are representative examples of the changes in the ultrasound velocity in aqueous solutions of octyl- and dodecyl-trimethylammonium bromides with the surfactant concentration at temperatures ranging from 20 to 45 "C. Each plot can be divided into two straight line segments, which correspond to the monomeric and micellar forms of the surfactant in aqueous solution. We can assign the intersection point on each plot to the c.m.c. It should be noted that in the case of octyltrimethylammonium bromide the plots near the c.m.c deviate from straight lines at each temperature, so that the c.m.c.values have to be estimated by linear extrapolation from both sides. A similar deviation can be observed on the plot of ultrasound velocity in aqueous solutions of sodium octyl sulphate at 25 "C published by Bloor et all5 It should be noted that there is a general tendency, especially for higher homologues, that a positive deviation from the straight line or a curvature, concave upward, occurs at the highest micelle concentrations examined. Typical examples of the changes in the adiabatic compressibility of aqueous solutions of octyl- and dodecyl-trimethylammonium bromides with the surfactant concentration at different temperatures are shown in fig. 5 and 6. Each plot can be divided into two straight line segments corresponding to the monomeric and micellar forms of the156 Compressibility of Surfactants 1570 - 1560 - 1550 - 1530 1520 E \ 1510 Y 7 .. 1 5 0 0 v 0 50 100 150 concentration/g dm-3 Fig. 3. Ultrasound velocity in aqueous solutions of octyltrimethylammonium bromide as a function of concentration at several temperatures. Symbols are the same as in fig. 1. ii-r--y 1: 1534 1508 '""k d4---+- 1484 v 1'"g6 1482 I I I I I 0 10 20 30 40 concentration/g dm-3 Fig. 4. Ultrasound velocity in aqueous solutions of dodecyltrimethylammonium bromide as a function of surfactant concentration at several temperatures. Symbols are the same as in fig. 1.R. Zieliriski, S. Ikeda, H. Nomura and S. Kato 157 4.001 I 1 I 0 50 loo 150 concentration/g dm-3 Fig. 5. Adiabatic compressibility of aqueous solutions of octyltrimethylammonium bromide as a function of surfactant concentration at several temperatures.Symbols are the same as in fig. 1. 425 I 0 10 20 30 40 concentration/g dm-3 Fig. 6. Adiabatic compressibility of aqueous solutions of dodecyltrimethylammonium bromide as a function of surfactant concentration at several temperatures. Symbols are the same as in fig. 1.158 Compressibility of Surfactants Table 2. Apparent adiabatic compressibility of alkyltrimethylammonium bromides in monomeric &) and micellar c6,) forms in water at various temperatures (in bar-') C,TAB C,TAB Cl,TAB C,,TAB C,,TAB T/"C B l Bl Bm Bi B m Bi Bm Bl B m 20 -0.33" -0.64 3.36 -0.98 3.69 -1.70 3.88 -2.49 4.04 25 -0.12 -0.17 3.47 -0.55 3.78 - 1.48 4.04 -2.62 4.18 30 0.16 0.24 3.49 -0.19 3.90 - 1 .1 1 4.20 -2.71 4.22 35 0.45 0.59 3.49 0.37 4.03 -0.17 4.31 -0.75 4.39 40 0.72 0.93 3.61 0.83 4.15 0.56 4.40 -0.25 4.51 45 0.83 1.22 3.66 1.18 4.27 0.77 4.63 0.61 4.65 - - - 4.66 - - - - 50 - a Extrapolated value. surfactant. One can draw straight lines by means of eqn (3). It is seen that the plot of the adiabatic compressibility of octyltrimethylammonium bromide as a function of surfactant concentration has a curvature near the c.m.c. at each temperature. Also there is a general tendency, especially for higher homologues, that a negative deviation from the straight line or a curvature, convex upward, occurs at the highest micelle concentration examined. The slopes of the plots for the monomeric forms are negative for all homologues, while the sign of the slope above the c.m.c.depends on the length of alkyl chain and temperature. At a given temperature, the slope of the plot for the micellar form increases with increasing number of carbon atoms. For a given number of carbon atoms in the alkyl chain, the slope increases with increased temperature. Because of deviation from linearity at the highest concentration, all numerical values of the slope are based on the straight line parts of an experimental curve. The apparent _adiaba_tic compressibilities of the surfactants in the monomeric and micellar forms, P1 and am, can be estimated using eqn (3), and their values at different temperatures are listed in table 2. For comparison, some data on tetramethylammonium bromide are also included.The temperature dependence of Bl and Brn for alkyltri- methylammonium bromides are shown in fig. 7 and 8. At 20 "C the values of bl are negative and they increase with increased temperature. The negative values can be attributedjo the effect of water of hydration around the monomer, and the gradual increase in P1 with rising temperature must be caused by th_e partial dehydration of the monomer. It is seen that the temperature dependence of /I1 varies with the length of the hydrocarbon chain of the alkyltrimethylammonium bromide. For the octyl and decyl derivatives it is linear, whereas for the dodecyl and tetradecyl derivatives it is sigmoidal. The values of P1 decrease non-linearly at each temperature as the number of carbon atoms in alkyl chain increases.The value of Pm increases linearly with increased temperature for all the derivatives, and its temperature coefficient is larger for higher homol_ogues. The temperature coefficient of Pm for all the derivatives is + or less of that of PI for the octyl and decyl derivatives. The values of Prn are positive but lower than the adiabatic compressibility of liquid hydrocarbons having the same number of carbon atoms.16* l7 However, while the apparent adiabatic compressibility of the micelles increases with increasing length of the alkyl chain, the adiabatic compressibility of hydrocarbons in the liquid state decreases. The change in apparent adiabatic compressibility of surfactants due to micelle formation can be expressed by ABm = /Tm -&. (1 1)R.Zieliriski, S. Ikeda, H. Nomura and S. Kato 159 1.5 1.0 0.5 - - 0.0 'f' -0.5 - 2 3 6 - -1.0 -1.5 -2.0 - -2.5 - - - b P --- - - 2-4 -3.0 I I I 1 I I I I 15 20 25 30 35 40 45 50 T/"C Fig. 7. Apparent adiabatic compressibility of alkyltrimethylammonium bromides in the monomeric form as a function of temperature. Symbols are the same as in fig. 2. 4.801 4.60 4.40 4.20 I s " 4.00 2 t G 2 P \ 3.8 0 3.60 3.40 3.20 15 20 25 30 35 40 45 50 T/"C Fig. 8. Apparent adiabatic compressibility of alkyltrimethylammonium bromides in the micellar form as a function of temperature. Symbols are the same as in fig. 2. 6 FAR II60 Compressibility of Surfactants 15 20 25 30 35 40 45 50 Fig. 9. Change in the apparent adiabatic compressibility upon micellization for alkyltri- methylammonium bromides in aqueous solution as a function of temperature.Symbols are the same as in fig. 2. T/"C The Yalues of ADm are shown as a function of temperature in fig. 9. A gradual d_ecrease in ADm with rising temperature can be mostly attributed to the increase in PI, gnd, therefore, to the dehydration of the ionjc head groups, but the sharp decrease in ADm is attributable to the sudden increase in P1 caused by partial destruction (melting) of the hydrophobic hydration. The apparent adiabatic compressions of the surfactant in the monomeric and micellar forms, El and Em, can be calculated from the values of the corresponding apparent molar volumes, Fl and Tm, and the apparent adiabatic compressibilities, Bl and Bm : El = TJ1 (1 2) Km = VmPm. (13) (14) - - - The change in the apparent molar compression, AKm, due to micelle formation can be expressed by CI - - AKm = Km - Kl.The values of kl, Km and AKm at various temperatures are tabulated in table 3. At a given temperature El decreases with the number of carbon atoms of alkyl chain, whereas Em increases. Similar observations were reported by Vikingstad et al.l8 for aqueous solutions of the homologous series of sodium alkanoates containing 6-1 3 carbon atoms in hydrocarbon chain, at 25 "C. For a given surfactant both Kl and K,,, increase with increased temperature, but the increase in Kl is larger. A similar tendency was reported by Vikingstad et for sodium decanoate. According to Cabani et al.lg the negative values of El at low temperaturesR. Zieliriski, S. Ikeda, H.Nomura and S. Kato 161 Table 3. Apparent molar compressions and the change in apparent molar compression due to micelle formation of alkyltrimethylammonium bromides in water at various temperatures (1 OP4 cm3 mol-' bar-,) C,TAB C,,TAB C,,TAB C,,TAB T/"C kl km Akm k, km Akm k, km Akm kl Em Akm 20 -14 77 91 -25 96 121 -48 114 162 -76 132 208 25 - 4 79 83 -14 99 113 -42 120 162 -81 137 218 30 5 80 75 - 5 103 108 -32 125 157 -84 140 224 35 13 81 68 10 107 97 - 5 129 134 -24 146 170 40 21 84 63 22 1 1 1 89 16 132 116 8 150 142 45 28 85 57 31 114 83 22 140 118 20 156 136 can be interpreted as a result of higher resistance to pressure of the structured water around the surfactant in the monomeric form than that of water in the bulk. Increasing the temperature brings aboyt a loosening of str-uctured water around the monomer and, therefore, an increase in Kl.An increase in K, can be attributed to the loosening of hydration +ell around polar head group as a result of an increase in temperature. The value of AKm increases with increasing alkyl chain length and decreases with increased temperature. Discussion The most probable explanation for the present results may be given by assuming two types of hydration, i.e. hydrophilic (or ionic) and hydrophobic, on the monomeric surfactant ion, and a single type of hydration, i.e. hydrophilic, on the surfactant ions forming micelles. The hydrophilic hydration occurs around the ionic head group which directs the hydrogen or oxygen atom of the water molecule towards the ionic group. The hydrophobic hydration is caused by strong association of water molecules or formation of hydration shell ('iceberg structure ') around alkyl chains.For the octyl and decyl derivatives the hydrophilic hydration is so extensive in the monomeric forms that there is no room for hydrophobic hydration around the hydrocarbon moiety. On the contrary, the hydrophilic hydration itself is strongly influenced by the hydrocarbon moiety, so that the compressibility further increases with increased temperature, as the alkyl chain become? longer. The contribution of one methylene group of the hydrocarbon chain to the ap,/aT coefficient is (4.2 0.7) x lo-* bar-l K-l. For the dodecyl and tetradecyl derivatives, the hydrophobic hydration is strong around the terminal methyl and methylene groups of the hydrocarbon moiety.The structure of the hydration shell around them is partially destroyed by melting at around 30-40 "C. Above and below this temperature range the apparent adiabatic com- pressibility of these higher homologues seems to vary with temperature in a manner similar to that of the lower homologues. On the other hand, the apparent adiabatic compressibility of the surfactant in _the micellar form increases linearly with increased temperature, and its rate of change, t)Pm/ aT, is 1.1 x bar-l K-' for the octyl derivative, while its average rate is 2.5 x lop6 bar-l K-l for the higher homologues. These increases in the apparent adiabatic conipressibilities of micelles can be attributed mostly to increased loosening of the micellar structure as well as to the gradual dehydration of the ionic head group with increased temperature.The lower rate of change for the octyl derivative can be attributed to the partial 6-2162 Compressibility of Surfactants penetration of water into the micelle interior. The micelle has an aggregation number as low as 2320921 for the octyl derivative in water at 25 "C. Furthermore, an increase in temperature generally causes a decrease in the micelle aggregation number for ionic surfactants.22'23 The low hydrophobicity of short alkyl chain and the small size of the micelle makes the micelle structure relatively loose and penetrable to water. Penetration of water molecules into the micelle interior has been discussed by several workers, and the existence of some bound water in the micelle interior in the vicinity of a few carbon atoms near the ionic head group was suggested by them.24-30 The temperature coefficient of the adiabatic compressibility of water, @?o/aT, in the same range of temperatures has an average value of - 1.2 x lo-' bar-' K-l.It seems reasonable to suppose that the lower value of the C@,,,/aTcoefficient for the octyl derivative can be, at least partially, caused by penetration of water molecules into the micellar interior. For the higher homologues there is almost a common micellar structure so that the rate of change in the apparent adiabatic compressibility with temperature may correspond to that of the structural loosening of the micelle, but its degree of loosening would be lower than that of liquid hydrocarbon^.'^ It is relevant to point out here again the observed difference in the types of micelle formation between the octyl derivative and higher derivatives.While it is not necessarily sharp, as seen in fig. 3 and 5, for the octyl derivative, micelle formation is clearly manifest with the higher homologues. The former behaviour means that the micelle formation does not take place like pseudo-phase formation because the size of the micelles is not sufficiently large, so that it is less sharp. For the latter, the micelle size is sufficiently large that micelle formation is well approximated by the pseudo-phase model. We thank Professor Shin Tsuge and his collaborators of Nagoya University for gas chromatographic analysis of the samples used in this study.R.Z. thanks the Ministry of Education, Science and Culture of Japan for the scholarship. References 1 K. Shinoda and T. Soda, J. Phys. Chem., 1963, 67, 2072. 2 S. Kaneshina, M. Tanaka and T. Tomida, J. Colloid Interface Sci., 1974, 48, 450. 3 B. Swaroop, Z. Phys. Chem. (Lkpzig), 1975, 256, 913. 4 G. M. Musbally, G. Perron and J. E. Desnoyers, J. Colloid Interface Sci., 1976, 54, 80. 5 0. Nomoto and H. Endo, Bull. Chem. SOC. Jpn, 1970, 43, 3722. 6 E. Vikingstad, A. Skauge and H. Hsiland, J. Colloid Interface Sci., 1979, 72, 59. 7 R. De Lisi, C. Ostiguy, G. Perron and J. E. Desnoyers, J. Colloid Interface Sci., 1979, 71, 147. 8 S. Backlund, H. Hsiland, 0. J. Kvammen and E. Ljosland, Acta Chem. Scand., Ser. A, 1982, 36, 9 R. Zielinski, S. Ikeda, H. Nomura and S.Kato, J. Colloid Interface Sci., 1987, in press. 698. 10 G. Stainsby and A. E. Alexander, Trans. Faraday Soc., 1952, 46, 587. 11 K. Shinoda and E. Hutchinson, J. Phys. Chem., 1962, 66, 577. 12 K. Shinoda, T. Nakagawa, B. Tamamushi and T. Isemura, Colloidal Surfactants: Some Physico- chemical Properties (Academic Press, New York, 1963), ch. 1. 13 K. SchaEer, in Lmdolt-Bornstein : Numerical Data and Functional Relationships in Science and Technology, New Series, ed. K. H. Hellwege (Springer-Verlag, Berlin, 1977), vol. 1, part B, p. 1. 14 M. E. Friedmann and H. A. Scheraga, J. Phys. Chem., 1965, 69, 3795. 15 D. M. Bloor, J. Gormally and E. Wyn-Jones, J. Chem. Soc., Faraday Trans. I , 1984, 80, 1915. 16 E. Vikingstad, J. Colloid Interface Sci., 1979, 68, 287. 17 V. K. Sachdeva and V. S. Nanda, J. Chem. Phys., 1981, 75,4745. 18 E. Vikingstad, A. Skauge and H. Hsiland, J. Colloid Interface Sci., 1978, 66, 240. 19 S. Cabani, G. Conti and E. Matteoli, J. Solution Chem., 1979, 8, 11. 20 H. J. L. Trap and J. J. Hermans, Proc., K. Ned. Acad. Wet., Ser. B, 1955, 58, 97. 21 H. V. Tartar, J. Colloid Sci., 1959, 14, 1 15. 22 M. N. Jones and J. Piercy, J. Chem. SOC., Faraday Trans. I , 1972, 68, 1839. 23 A. Malliaris, J. Le Moigne, J. Sturm and R. Zana, J. Phys. Chem., 1985, 89, 2709. 24 H. Sasaki, H. Okuyama and S. Saito, Bull. Chem. SOC. Jpn, 1956, 29, 752.R. Zielinski, S. Ikeda, H . Nomura and S. Kato 163 25 J. Clifford and €3. A. Pethica, Trans. Faraday SOC., 1964, 60, 1483. 26 L. Benjamin, J. Phys. Chem., 1966, 70, 3790. 27 J. M. Corkill, J. F. Goodman and T. Walker, Trans. Faraday SOC., 1967, 63, 768. 28 T. Walker, J. Colloid Interface Sci., 1971, 45, 372. 29 K. A. Zachariasse, B. Kozankiewicz and W. Kiihnle, in Surfactants in Solution, ed. K. L. Mittal and 30 K. N. Ganesh, P. Mitra and D. Balasubramanian, J. Phys. Chem., 1982, 86,4291. B. Lindman (Plenum Press, 1984), vol. 11, p. 565. Paper 71167; Received 30th January, 1987

ISSN:0300-9599    

DOI:10.1039/F19888400151

出版商:RSC    

年代:1988

数据来源: RSC

摘要:

J. Chem. Soc., Faraday Trans. I , 1988, 84(1), 165-174 The Behaviour of Uranium Metal in Hydrogen Atmospheres Geoffrey C. Allen and Julian C. H. Stevens Central Electricity Generating Board, Berkeley Nuclear Laboratories, Berkeley, Gloucestershire GL13 9PB The reaction between commercial H, and uranium metal leads to the formation of UO, due to traces of water vapour or oxygen. When extremely pure H, is used uranium hydride may be formed but, even with 99.9999% H,, uranium dioxide forms preferentially. The present work identifies the presence of UH, in the X-ray photoelectron spectrum of a uranium sample which has been exposed to ca. 1O'O Lt H, at ca. 200 "C. This spectrum indicates that the hydride possesses a high degree of covalency, since the oxidation state of uranium in UH, appears to be ca.1.4. The reaction of uranium with water vapour and air has been the subject of a number of investigations but recently, in a study of the initial oxidation of uranium foil in water vapour plus oxygen mixtures using X.P.S., oxygen appeared to inhibit the reaction of uranium with water.' Moreover, the oxide produced resulted from oxidation by hydroxide ions rather than oxygen : U + 20H- -+ UO, + H, + 2e-. This observation was supported by the isotopic studies of Baker et al.' and offered the possibility that the hydrogen produced could encourage the formation of uranium hydride. This pyrophoric compound is known to form when the metal is stored under damp conditions but no clear evidence for the presence of uranium hydride was seen during experiments involving the oxidation of uranium metal in oxygen-water vapour mixtures.' The present work was undertaken to investigate the effect of hydrogen-water vapour atmospheres on a clean uranium surface and to see if the chemical processes that occur are consistent with the formation of uranium hydride.Experiment a1 The AEI ES200B electron spectrometer used in these studies has been described previ~usly.~ A sheet of natural uranium (Goodfellow Metals 99.98%) was cut into strips measuring 19 x 6 x 0.18 mm. At this stage, the uranium strips had a dull, black finish due to surface oxidation to UO, and so they were treated with a 50:50 mixture of concentrated nitric acid and water in an ultrasonic bath for about 5 min, to remove the surface oxide. After having been rinsed with distilled water each strip was electropolished in a solution containing equal parts of sulphuric acid, phosphoric acid and water using a current of 0.4 A.After ca. 2 min of electropolishing, the samples had a shiny metallic appearance and, following a rinse with distilled water and methanol, were transferred to the spectrometer where the residual oxide was removed by repeatedly heating the sample to 200 "C and etching the surface with argon atoms. Etching was achieved with an IONTECH saddle field fast atom bombardment source (FAB1 INF) which was typically -f 1 L = low6 Torr s (1 Torr = 101 325/760 Pa). 165166 operated at 6 kV under a dynamic pressure of 1 x Torr argon (BOC Research Grade). The hydrogen used in these experiments was of high purity (BOC, 99.9999 YO), and the major contaminants in this were water (< 0.5 vpm), nitrogen (< 0.4 vpm) and oxygen (< 0.1 vpm).Uranium Metal in Hydrogen Atmosphere Results and Discussion The interaction of hydrogen with a clean uranium surface was studied by monitoring photoelectron peaks due to uranium and oxygen. Since the most prominent core level features in the photoelectron spectrum are the uranium 4&,, and 4f,,, peaks, these were studied together with their associated satellites and the oxygen 1s peak. The Uranium 4fand Oxygen 1s Regions As mentioned previously, the uranium strip appeared shiny and metallic prior to its insertion into the spectrometer but subsequent photoelectron analysis produced spectra which showed the uranium 4fpeaks at binding energies of 380.3fO.l and 391.2kO.l eV with accompanying satellites at a separation of 6.9 eV to higher binding energy.The positions of these peaks were consistent with those of 380.2 f 0.1 and 391.2 f 0.1 eV obtained by Allen et aL4 for UO, in equilibrium with uranium metal, thus indicating the presence of a surface film of UO,. After the uranium strip had been heated and etched the uranium 4&,, and 4f7,, binding energies were measured and found to be 388.1 fO.1 and 377.2 f 0.1 eV respectively, and both peaks had halfwidths of 1.9 eV. These values match those previously reported for a clean uranium surface using standard A1 K, radiation4 Various experiments were performed in which the clean uranium surface was exposed to a given static pressure of pure hydrogen (BOC 99.9999%) for a given time, and the U4fregion, valence region and 0 1s region were monitored. At hydrogen exposures of less than ca.lO'L of hydrogen the U4f peaks were seen to broaden so that their halfwidths increased by ca. 10%. However, this effect was also noted from a sample of clean uranium which had been allowed to stand in the preparation chamber of the spectrometer for 10 min. In this case the surface had been oxidised by the residual gases in the vacuum chamber and the U4fpeaks were seen to broaden on the high binding energy side due to small contributions from UO, at 380.3 and 391.2 eV. The peak broadening observed after exposure to ca. lo7 L hydrogen was, therefore, probably due to oxidation by the 0.5 vpm water impurity in the hydrogen.This is consistent with the work of Bloch et ~ 1 . , ~ who concluded that exposure of uranium to hydrogen (ca. 36 L) had the same effect on the Auger electron spectrum as had exposure to the residual gas atmosphere of the u.h.v. system. This conclusion also agrees with the changes observed in the 0 1s region of the spectrum where the intensity of the 0 1s peak was seen to increase on exposure to hydrogen. However, one striking difference observed between the samples left in u.h.v. and those left in hydrogen was the intensity of the U4f photoelectron signal observed. The reduction in the signal intensity observed after exposing the samples to hydrogen was much greater than that recorded for the u.h.v. samples, which was attributed, in the latter case, to the formation of a surface film on these samples. On exposing the uranium surface to increasing amounts of hydrogen the oxide contribution in the U4fregion continued to increase until, at exposures of ca.3 x lo8 L, the oxide peaks were clearly visible as distinct shoulders to the high binding energy side of the metal 4fpeaks. Increasing the hydrogen exposure still further produced spectra in which the oxide and metal contributions in the U4f region appeared to be approximately equal in intensity. However, comparison of these spectra with thoseG. C. Allen and J. C. H. Stevens 167 1,000 - I v) c, E $ 500 390 380 370 4 d 0 ' ' ' ' ' ' ' ' ' ' ' binding energy/eV Fig. 1. Comparison of the U4fphotoelectron region for uranium foil oxidised in UHV (- - - -) and hydrogen (-).i ' 500 I 390 380 370 0 400 binding energy/eV Fig. 2. U4fphotoelectron region after exposure of clean uranium foil to ca. 10'O L hydrogen at ca. 200 "C. obtained from a clean uranium surface that had been allowed to oxidise in the absence of hydrogen revealed clear differences (fig. 1). First, examination of the U4f712 and U4f& regions showed that, in each case, the separation between the metal peak and the oxide peak was not so clearly defined for the sample treated with hydrogen. Furthermore, the hydrogen-dosed sample showed a higher signal intensity between the 4f712 and 4f5,, regions and to the high binding energy side of the 4f5,, peaks. It should also be noted that, for ease of comparison, the intensity of the spectrum produced from the sample which oxidised in the absence of hydrogen is reduced by a factor of two in fig.1, showing that the loss in signal intensity observed at low hydrogen exposures increased dramatically with increasing hydrogen dosage. At this stage it is instructive to consider the factors which might be responsible for the changes observed in the U4fregion after the sample had been exposed to hydrogen. The spectra obtained were consistent with the presence of uranium metal having an oxide film at the surface but the characteristic U4f photoelectron region appeared to be modified when hydrogen was present. Since hydrogen is a very small molecule it should168 Uranium Metal in Hydrogen Atmosphere n E c, .I. d 9 2 v Y .* x e Y .* 3 ao 375 n E c, .I x c, .- 5 c, E ._ (ii) 3 80 375 binding energy/eV binding energy/eV Fig.3. Deconvolution of U4f7,,, photoelectron region for hydrided uranium : (a) two-peak synthesis, (b) three-peak synthesis; (i) experimental spectrum, (ii) synthesized spectrum. Table 1. Binding energies for the U4f7,,, peak U4f,,, binding energy/eV ref. U (metal) UD, (single crystal) UN UH3 uo2 UBr, UC1, UF4 377.1 & 0.1 378.5 f 0.2 379.1 379.4 & 0.2 380.2 f 0.2 380.7 & 0.2 381.5 f0.2 382.8 & 0.2 4 this work 8 14 4 13 12 12 be able to diffuse readily through the oxide film to the bulk metal. Indeed, a study by Wheeler' showed that hydrogen readily diffused through UO, single crystals. Wheeler also remarked that the radius of the hydrogen molecule (0.12 nm) is slightly smaller than that of an oxygen ion (0.14 nm) and argued that hydrogen molecules could enter vacant anion sites or unoccupied interstitial sites in UO, without affecting the valency of the uranium atoms.Such filling of vacancies within the oxide lattice by hydrogen molecules might be expected to reduce the inelastic mean free path of the photoelectrons and this could account for some of the observed loss in signal intensity. Evidence for the incorporation of considerable quantities of hydrogen into the uranium sample was provided by a consideration of the outgassing which occurred when the sample was dosed with hydrogen and subsequently heated. Thus, exposing the sample to ca. 5 x lo9 L H, typically resulted in gas evolution for a period of 20 min. During this time the pressure in the chamber rose by two orders of magnitude before falling to the base pressure of the spectrometer. Nevertheless, the sole incorporation of neutral hydrogen molecules into the metal and its oxide film would not be expected to bring about the changes in the U4f photoelectron region seen in fig.1 . The oxide film probably results from the presence of the 0.5 vpm water in the hydrogen used, since uranium has been observed to react more rapidly with water than with oxygen' but the small amount of oxygen present (0.1 vpm) might also affect the formation of this film. No evidence was found for the presence of hyperstoichiometric UO,, there being no additional satellite in the photoelectron spectrum at ca. 8.2 eV fromG. C. Allen and J . C. H . Stevens 169 rystal) I I / 0 U 4f 7,2 binding energy/eV Fig.4. Correlation of U4fbinding energy with oxidation state, assuming U4+ in UF,. ., Ref. (4); V, this work, V, ref. (8); 0, ref. (12); 0, ref. (13); 0, ref. (14). the main U4f peak,' and it is unlikely that UO,-, could be formed under these conditions. It therefore seems probable that the changes observed in the U4f photoelectron region are due to uranium hydride formation. Uranium deuteride has been studied using X.P.S., and both valence-band and core- level spectra obtained from a fresh surface exposed at ca. 10-l' Torr by fracturing the sample.8 The U4f doublet binding energies were measured as 379.1 and 389.8 eV although no core-level spectra were presented. The valence-band spectrum showed a well defined U5fpeak at a binding energy of 2 eV which was overlapped by a broader band situated between ca.8 eV and the Fermi level. This broad band was assigned to bonding levels derived from H 1s and U6d orbitals. Thus the positions of the U4fpeaks in uranium hydride would be expected to occur between the 4fpeaks in uranium metal and those in UO,, in agreement with the spectrum observed here. However, large contributions from UO, make it difficult to identify these peaks with certainty and it was decided to try to limit the oxide contamination. Thermogravimetric studies of the temperature dependence of the reaction between uranium and hydrogen have shown that the problem of surface oxide contamination is at its worst for temperatures below 150 0C9,10 and so it was decided to investigate the effect of heating the uranium sample. Since the rate of reaction of metallic uranium with hydrogen is known to increase with temperature and reach a maximum at ca.250 O C l l170 Uranium Metal in Hydrogen Atmosphere Table 2. Relative intensities of the U4f components recorded during the etching process signal intensity cumulative etch time/min metal oxide hydride 0 1 2 3 4 19 49 56 30 14 66 19 15 70 17 13 75 15 10 80 11 9 80 12 8 86 7 7 1 I I I J 1 2 3 4 5 cumulative etch time/min Fig. 5. Variation of relative signal intensities for uranium metal, UO, and UH,, as the hydrogenated metal surface is etched with argon atoms: 0, metal; ., UO,; A, UH,. the temperature was kept below this value. Spectra recorded after exposing the sample to ca. 10" L of hydrogen at ca. 200 "C showed an 0 1s signal which was considerably less intense than that expected following a similar hydrogen dose at room temperature, However, while the oxide contamination appeared to have been reduced by heating the uranium during exposure to hydrogen, the measured U 4f photoelectron region still contained a significant contribution from the oxide (fig.2). Consequently we tried to separate the components from uranium metal and stoichiometric uranium oxide in the U4f photoelectron spectrum by deconvolution. Since the U 4f5/, region contains contributions from the U4f,/, shake-up satellite at 6.9 eV4 and U4f5,, signals from both the metal and oxide, it was decided only to attempt deconvolution of the U4f,,, region shown in fig. 2. The results are shown in fig.3(a). The spectrum was inadequately fittedG. C . Allen and J . C. H . Stevens 171 by two contributions but the addition of a third Gaussian component greatly improved the situation [fig. 3 (b)]. Furthermore, for a series of spectra obtained from hydrogenated uranium, both the position and halfwidth of the third peak derived from the best fit with the experimental spectra were found to be constant at a binding energy of 378.5 f 0.2 eV (f.w.h.m. 1.6f0.2eV). The improvement obtained by the inclusion of this peak in the hydrogenated spectra, and the consistency of its position indicated the presence of a third uranium environment. The binding energy for this third peak (378.5 eV) is in good agreement with that of 379.1 eV obtained by Ward et al.* for UD, and supports the view that uranium hydride is present. To test this view it is useful, as a guide, to rationalise the binding energy of the U4f7,, peak attributed to UH, with reference to the corresponding binding energy for other binary uranium compounds.Table 1 compares the binding energy for UH, with those obtained for UF, and UC1,,12 UBr4,l3 UO,, and UN,', with the metal taken as a point of reference. It is important to recugnise that although the binding energy shifts for the U4fpeak are dependent on the oxidation state of the metal ion, there are other factors to consider. In general, the binding energy is a measure of the potential field experienced by the ionised electron, and so differences in crystal structure and concomitant Madelung effects can also alter its value markedly.Thus, too much emphasis should not be placed on the comparison of binding energy shifts obtained from different lattices. Nevertheless if it is assumed that UF, contains the U4+ ion, the plot shown in fig. 4 is obtained. By extrapolation from this plot the oxidation state of uranium in UD, may be estimated as 1.4 and the value obtained for UY, in the present work is a little lower. The U-D distance has been determined as 2.32A by neutron diffra~tion'~ and using this distance together with Pauling's rule and radii16 Rundle estimated the valence of uranium in UD, as 1.7 for one site and 1.8 for the other, in respectable agreement with that derived above. A review article by Waber17 describes some experiments which provide evidence for the formation of uranium hydride below an oxide layer.Photomicrographs of a cross- section of uranium metal surface which had been exposed to a helium-water mixture (with a 50: 50 partial pressure ratio) at 75 "C for 1689 h revealed a triangular region projecting into the bulk metal. This was identified as /3-UH, by microbeam X-ray techniques." The hydride was thought to have formed from the interaction with hydrogen produced by the reaction in eqn (1) and it was suggested that non-uniform attack gave rise to spikes of hydride below an oxide layer. As a result, it was decided to see if the oxide contamination produced after exposing the sample to hydrogen could be removed by gradually etching the sample surface. After treating a sample with ca. lolo L hydrogen at 200 "C and recording the U4f and 0 1s photoelectron regions, the surface was bombarded with argon atoms for 1 min and these regions were scanned again. Argon etching was found to decrease the 0 1s signal by ca.30% and the oxide contribution in the U4fregion also decreased as would be expected if a surface oxide was removed. However, the U4f region showed no clear evidence for the presence of uranium hydride. The etching process was continued for a further 48 rnin during which time three additional spectra were recorded at 1 min intervals, a fourth after a further 15 rnin and a final spectrum 30 rnin later. The U4f7,, regions were deconvoluted in the same way as that described and shown in fig. 3 (b), and the area of each component peak taken as a percentage of the total area to find the relative signal intensities due to metal, oxide and hydride after each stage of the etching process.The results obtained are shown in table 2 and the initial progress plotted as a function of etch time in fig. 5 . Here it can be seen that, after the surface had been etched for 1 min, only the oxide signal was reduced and so oxide must have been present at the surface of the sample. On etching the surface for a further minute, however, both the oxide and the hydride signals were found to decrease while the metal signal continued to increase. This suggests that both172 Uranium Metal in Hydrogen Atmosphere Oxide ................. ...................................... Fig. 6. Diagrammatic representation of the uranium surface after exposure to ca.1 O l o L hydrogen at ca. 200 "C. binding energy/eV Fig 7. Valence photoelectron spectra for (a) clean uranium, (b) uranium exposed to ca. 3 x lo8 L hydrogen at room temperature and (c) uranium exposed to ca. 1O'O L hydrogen at ca. 200 "C. oxide and hydride were present in a mixed layer on top of the metal. Further etching resulted in a further increase in the metal signal at the expense of the two other components. This admittedly rather qualitative picture, supports the evidence presented by Waber17 that showed hydride growth occurring beneath a thin oxide layer. The fact that successive etching removed both oxide and hydride showed that uranium hydride was not produced as a uniform layer beneath an even layer of oxide. Rather, attack by hydrogen would appear to be localised giving rise to hydride spikes projecting into the bulk metal.The rate of removal of surface layers by fast atom bombardment has been investigated by Allen et a1.18 and the etch rate estimated as a function of the power of the atom source. Their figure of 0.42 & 0.04 nm W-I min-l was obtained during argon-atom sputtering of various oxides of iron but can be used to give an estimate of the etch rateG. C. Allen and J. C. H . Stevens 173 used for the uranium system. In this work the atom source was run at a power of 10 W and so the rate of removal of the surface was ca. 4.2 nm min-l. Since the hydride peak intensity started to decrease after 1 min (or less) of argon etchicg, the overlying oxide must have been 5 4 nm thick. An 0,- ion has a diameter6 of 2.8 A and so, if this is taken to represent the thickness of a monolayer of UO,, 4 nm will correspond to ca.14 monolayers of UO,. Thus, as a result of this profile study, the surface region of the sample after exposure to ca. 1O1O L H, at 200 "C can be visualised as shown in fig. 6. Valence Region Valence spectra were recorded for pure uranium metal. A typical measurement is shown in fig. 7(a). The sharp feature at ca. 0.5 eV is due to ionisation of U5felectrons and is in good agreement with earlier meas~rements.~ On exposure of the clean sample to ca. 3 x 1 0 8 L hydrogen another sharp feature appeared at ca. 2eV together with a broad band to higher binding energy [fig. 7 ( b ) ] . The U4fregion recorded after this level of exposure indicated the presence of roughly equal amounts of pure metal and UO, (fig.1) and so the peak at ca. 2 eV in the valence spetrum was probably due to U5felectrons in UO,. This is substantiated by the valence spectra of UO, recorded by Allen et aL4 which showed the U5f peak at 1.5 eV. On being exposed to ca. 10 L hydrogen at 200 "C, the U4fspectra show less oxide contamination and the U Sfpeak from UO, also decreased [figure 7 (c)]. Valence spectra of UD, obtained by Ward et a1.8 are similar to those for UO, except that the U Sfpeak is seen at 2 eV and not 1.5 eV. Since these two positions are fairly close it is not possible to say whether there is a contribution due to uranium hydride in the valence spectra recorded here (fig. 7 ) . Conclusions The results of this investigation clearly show that uranium hydride is formed when uranium is left in a hydrogen-rich atmosphere.Oxidation was also observed as a result of a low level of water (< 0.5 vpm) which was present as an impurity in the hydrogen. The resulting oxide was identified as stoichiometric UO, and shown to be present as a surface film of ca. 14 monolayers. Uranium oxide and uranium hydride were identified in a mixed layer between this surface film and the bulk metal. In the light of this current work it is interesting to consider, briefly, some kinetic studies performed by Bloch et al." These workers characterised the initial stages of the uranium-hydrogen reaction using a hot-stage microscope. They observed an induction period, during which the rate of reaction was immeasurably slow, which was preceded by rapid formation of product 'spots' at gain boundaries and slip planes.Bloch et al. claim that these are uranium hydride spots, and that their subsequent growth is inhibited by an increase in the stress field produced by the initial product growth. This was suggested to be the reason for the induction period, which terminated when the strained overlayer ruptured, allowing further growth. However, the results of the present study show that, in the first instance, the clean uranium surface acts as a 'getter' for oxygen- containing species in the system. Since no chemical analysis was performed by Bloch et al." the rapid formation observed by these workers during the pre-induction period may have been due to UO, spots forming at grain boundaries and slip planes.In this case, the induction period observed subsequently could be associated with the time taken for H, to diffuse through the oxide and react with the metal to form UH,. This view is supported by the work of Alire et a1.l' whose thermogravimetric studies of the uranium-deuterium reaction showed that the use of purified deuterium, at temperatures in the range 175-300 "C, eliminated the induction period altogether. The implication of this work in the area of nuclear fuel storage is that any hydride174 Uranium Metal in Hydrogen Atmosphere formation is likely to be preceded by the formation of a uranium oxide overlayer. The rate at which hydrogen difuses through this oxide film will therefore be a factor which influences the extent to which UH, will form under these conditions. This rate will, in turn, be influenced by the thickness of the oxide film, its integrity, and the hydrogen pressure above it.Increased hydrogen pressures might be expected to increase hydride formation, but the partial pressures of H,O and 0, present under normal storage conditions would be far in excess of the H, partial pressure, and thus should give rise to a substantial oxide overlayer. It is concluded, therefore, that the pyrophoric compound UH,, is unlikely to be present to any great extent under these conditions, although hydride formation could be accelerated by oxide spallation during fuel storage or transport. The work was carried out at the Berkeley Nuclear Laboratories of the Technology Planning and Research Division and the paper is published with permission of the Central Electricity Generating Board. References 1 G. C. Allen, P. M. Tucker and R. A. Lewis, J. Chem. SOC., Faraday Trans. 2, 1984, 80, 991. 2 M. McD. Baker, L. N. Less aiid S. Oman, Trans. Faraday SOC., 1966, 62, 2525. 3 G. C. Allen and N. Holmes, J. Chem. SOC., Dalton Trans., 1987, 3009. 4 G. C. Allen, I. R. Trickle and P. M. Tucker, Philos. Mag.B, 1981, 43, 689. 5 J. Bloch, V. Atzmony, M. P. Daniel, M. H. Mintz and N. Shamir, J. Nucl. Mater., 1982, 105, 196. 6 V. J. Wheeler, J. Nucl. Muter., 1971, 40, 189. 7 G. C. Allen, P. M. Tucker and J. W. Tyler, J. Phys. Chem., 1982, 86, 224. 8 J. W. Ward, L. E. Cox, J. L. Smith, G. R. Stewart and J. H. Wood, J. Phys. (Paris), 1979, C4, 15. 9 J. B. Condon, J. Phys. Chem., 1975, 79, 392. 10 R. M. Alire, B. A. Mueller, C. L. Peterson and J. R. Mosely, J. Chem. Phys., 1970, 52, 37. 11 W. M. Albrecht and M. W. Mallet, J. Electrochem. SOC., 1956, 103, 404. 12 G. C. Allen, P. M. Tucker and J. W. Tyler, Philos. Mag. B, 1983, 48, 63. 13 G. C. Allen and J. W. Tyler, J. Chem. SOC., Faraday Trans. I , 1987, 83, 1355. 14 G. C. Allen and N. Holmes, 1986, unpublished data. 15 R. E. Rundle, J. Am. Chem. Soc., 1951, 73, 4172. 16 L. Pauling, J. Am. Chem. SOC., 1947, 69, 542. 17 J. T. Waber, Los Alamos Scientific Laboratory Report, LA-2035 (1958). 18 G. C. Allen, A. G. Warner and A. R. Jones, Part. Charact., 1986, 3, 89. 19 J. Bloch, F. Simca, M. Kroup, A. Stern, D. Shmariatu, M. H. Mintz and S. Hadari, J. Less-Common Met., 1984, 103, 163. Paper 71204; Received 5th February, 1987

ISSN:0300-9599    

DOI:10.1039/F19888400165

出版商:RSC    

年代:1988

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I , 1988, 84(1), 175-185 The Thermodynamics of Solvation of Ions Part 3.-The Heat Capacity for Solvation of Gaseous Ions in Methanol at 298.15 K Michael H. Abraham" and Yizhak Marcus"? Department of Chemistry, University of Surrey, Guildford, Surrey GU2 5XH Kenneth G. Lawrence Department of Chemistry, Birkbeck College, Malet Street, London WClE 7HX Literature values of the standard partial molar heat capacities of 1 : l electrolytes in methanol have been divided into ionic contributions using the assumption that Cr(Ph,As+) = C:(BPh;). Combination with c"p values for gaseous ions then leads to single-ion values for the solvation of gaseous ions in methanol, AsOlvG. These latter values are then broken down into a neutral term N , an electrostatic term E and a configurational term C.It is shown that the above single-ion division leads to configurational single-ion quantities that agree well with other single-ion parameters. Using this division, all the inorganic cations and anions can be regarded as structure- makers that decrease the fluidity of the solvent. The tetra-alkylammonium ions are also structure-makers, but exhibit ' solvophobic solvation ', analogous to, but quantitatively much smaller than, the corresponding hydrophobic hydration in water. In part 1 of this series1 we surveyed standard partial molar heat capacities, Cp", of electrolytes in water and combined these with q values for gaseous ions to give standard molar heat capacities of hydration, Ahyd Cz. -A set of single-ion values was constructed based on the assumption that C:(Ph4P+) = C;(BPhi), and the single-ion Ahyd q values were analysed and compared to various other parameters such as standard molar entropies of hydration, viscosity B-coefficients, and partial molal volumes of ions in water.It was shown that the single-ion division according to C;(Ph,P+) = C:(BPhJ was quite compatible with divisions based on B(Ph,P+) = B(BPh,) and on P(Ph,P+) = p(BPh,), as well as with previous single-ion entropy values. It seemed of interest to attempt a similar analysis of heat-capacity results in non- aqueous solvents, and in the present paper we do so for the solvent methanol. The most reliable method of obtaining C: values is that of flow calorimetry, and Jolicoeur et a1.2 and also Criss et al.3*4 have recorded values determined in this way.Criss et have also obtained C: values by the alternative method of combining Aq(so1id -+ MeOH), found through the temperature variation of the enthalpy of solution, with the heat capacity of the solid: (1) C,"(MeOH) = Aq(so1id -+ MeOH) + Cz(so1id). Details of the heat-capacity results are in table 1, where values from the two alternative methods are separately recorded. Only few checks for additivity can be made with the data in table 1, which leads to differences (Br- - C1-) = 26 J K-l mol-1 and (Na+ - Li+) = 62 J K-l mol-l, but in order to list single-ion values, some extra thermodynamic assumption has to be made. Quite often, the assumption that single-ion values for Ph4P+ and BPh, are the same is treated t Visiting from the Department of Inorganic and Analytical Chemistry, The Hebrew University of Jerusalem, Jerusalem 9 1904, Israel.175176 Thermodynamics of Solvation of Ions Table 1. Standard partial molar heat capacities of electrolytes in methanol, in J K-' mol-1 at 298 K electrolyte flow calorimetry via eqn (1) LiCl LiBr LiBBu, NaCl NaBr NaI NaClO, NaBPh, KF KBr KI CsF Me,NBr Et,NBr n-Pr,NBr n-Bu,NBr n-pent y1,NBr n- hept y1,NBr Ph,PCl Ph,PBr Bu,PBr Ph,AsCl - 1 193 5643 -613 -35,3 -30, - 1003 - 536,2 597, - 753 - 203 03 - 8g3 - 485,2 4773 5943 8433 44g2 482, 500, 4942 - 4808 500* Table 2. Heat capacities of solvation of gaseous ions in methanol, in J K-l mol-' at 298 K ion Li' Na+ K+ cs+ Me,N+ Et,N+ Pr,N+ Bu,N+ pentyl,N+ heptyl,N+ Bu,P+ Ph,P+ Ph,As+ F- c1- Br- I- BBu; BPh, ClO, r/nm C;(MeOH) 0.059 -71 0.102 -9 0.138 5 0.170 -8 0.280 87 0.337 249 0.379 38 1 0.413 506 0.443 619 0.493 868 525 0.424 503 0.425 545 0.133 - 80 0.181 - 51 0.196 - 25 0.220 -5 0.240 32 635 0.420 545 - - 21 21 21 21 99 191 283 375 467 65 1 369 372 21 21 21 21 62 367 - - ASOl" G - 92 - 30 - 16 - 29 - 12 58 98 131 152 217 134 173 - 101 - 72 - 46 - 26 - 30 - - 178 - 62 19 63 86 - 79 - 194 -471 - 762 - 984 - 786 - 638 - 623 - 35 5 35 45 - 14 - - - 596 ~~ ~~~ ~ a Heat capacities of transfer from water to methanol.M .H . Abraham, Y. Marcus and K . G . Lawrence I77 as equivalent to the assumption of similar single-ion values for Ph,As+ and BPh,. Inspection of table 1 shows that this can hardly be the case for C," values in methanol, since Cr(Ph4PCl) = 448 J K-l mol-l, whereas the value for Ph,AsCl is 494 J K-' mol-', a difference of 46 J K-' mol-'.For C," in water' the corresponding difference is 27 J K-l mol-'. Not only is there a difference of 46 J K-' mol-' between Ph,PCl and Ph,AsCl, but there is also a discrepancy in the recorded values of Cr for NaBPh,, uis. 536 J K-' mol-' by Jolicoeur et al.' and 597 J K-l mol-' by Criss et We have no means of deciding which is the better experimental value, but we shall show later that a single-ion division based on C," = 536 J K-' mol-' for NaBPh, and 494 J K-' mol-1 for Ph,AsCI leads to ionic values more compatible with single-ion partial molal volumes and single-ion viscosity B-coefficients. Our single-ion values are given in table 2. The most recent values given by Shin and Criss4 differ from those in table 2 by +61 J K-l mol-' (with cations being more positive and anions more negative than those in table 2) owing to the different experimental values and assumptions used.Also in table 2 are single-ion heat capacities of transfer from water to methanol, A,CZ, obtained by combination of the present values with those listed for water previously.' Transfer values for the inorganic ions are rather small, and in view of the possible uncertainty in the actual single-ion values in methanol do not lead to any substantial conclusion. However, the very negative A, Ci values for the large organic ions is clear evidence for the hydrophobic structure-making effect of these ions in water; this effect results in a decrease in fluidity in water, with a consequent decrease in C," (water) and increase in A,G.'The single-ion C,"(MeOH) values may be combined with values1 of Ci(gas) to yield heat capacities for the transfer of single ions from the gas phase to methanol, AsOlv Ci : C,"(MeOH) - Ci(gas) -+ Asolv Ci. These heat capacities of solvation are listed in table 2. Following our analysis of AhYd Ci values1 it is useful to be able to subtract out the so-called 'neutral term', N , calculated as the heat capacity of solvation of a neutral solute of the same size as the ion in question. Unfortunately, the only directly determined CP(Me0H) values for inert neutral solutes are for the larger alkanes, determined by French and Criss.' These may be combined with well known'' values of Ci (gas) to yield AsolvG for a number of alkanes (see table 3).For permanent gases the usual method of obtaining AhydCi or ASolv Ci is through the temperature variation of solubilities of the gaseous solutes, or of Henry's constants, KH. The simplest fitting equation to In KH that allows for a non-zero Aso,v Ci value is eqn (3), from lnKH = A+B/T+ClnT (3) which Asolv Ci is given as -RC where R is the gas constant, 8.314 J K-' mol-'. We have applied eqn (3) to the original data of Lannung" for the solution of helium, neon and argon in methanol over a quite wide temperature range, obtaining values of 0, 0 and 24 J K-l mol-I, respectively. Similarly, application of eqn (3) to data on the solution of hydrogen,12,13 or nitrogen12 in methanol leads to Asolv Ci values of 8 and 17 J K-l mol-f, respectively.All the heat-capacity results for neutral inert solutes are given in table 3, together with the radii of the various solutes, and a plot of Asolv Cz against r is shown in fig. 1. Bearing in mind the (considerable) possible error in the ASolv values for the permanent gases, it seems as though a straight line is obtained up to a radius of ca. 0.2 nm, followed by a smooth curve afterwards.178 Thermodynamics of Solvation of Ions Table 3. The heat capacity of solvation (gas-+MeOH) of neutral solutes in J K-' mol-' at 298 K helium neon hydrogen argon nitrogen n- heptane 2,4-dime th ylpen tane 2,2-dimethylpentane 2,3 -dime t h ylpen tane 2,2,4- trimethylpentane 2,3,4-trimethylpentane 3-methyloctane 2,6-dime thy1 hep t ane 2,4-dimethylheptane 3,4-dimethylheptane 3,3-dimethylheptane n-decane heptamethylnonane n-heptadecane 0.131 0.139 0.144 0.175 0.I85 0.3 13 0.3 145 0.314 0.31 1 0.325 0.321 0.334 0.335 0.334 0.332 0.329 0.344 0.391 0.40 1 - 251.3 249.6 244.7 241.5 266.4" 273.2 307.0 306.4 31 1.4 306.3 307.2 342.4 496.2 560.6 166.0 166.0 166.0 166.0 188.9 188.9 212.6 210.5 206.1 206.0 204.2 234.6 366.0 394.7 Od Od 8d 24d 1 7d 85 84 79 76 78 84 94 96 105 100 103 108 1 60 166 a From ref. (l), or calculated using the Stearn-Eyring formula. Ref. (9). Ref. (10). See text. " Shin and Criss4 give a value of 271.1 J K-I mol-I. 200 r 0.1 0.2 0.3 0.4 r/ nm Fig. 1. Plot of AsOlvG for neutral inert solutes us. the solute radii r/nm.M . H . Abraham, Y. Marcus and K. G . Lawrence 179 Analysis of the AsolVG Results In this section we follow exactly the procedure adopted’ to deal with results in water, and consider AsolvCi to consist of an electrostatic term, E, a neutral, N , and a ‘configurational’ term that cannot be estimated and is obtained by difference in eqn (4): AsOlv Ci = N + E+ C.(4) There are various possible summations of the terms in eqn (4): we give now two summations, as before, distinguished by subscripts 1 and 2. In the first summation, the neutral term, N,, is calculated (by extrapolation where necessary) from Asolv values for the rare gases, and hence represents only a ‘size’ effect, without any contribution from ‘ solvophobic solvation ’.t The value of N , in J K-’ mol-’ is given by 317r-41 when r is in nm. The electrostatic term, El, is calculated as the total electrostatic heat capacity of solvation of the ion, and is obtained from the ‘one-layer ’ solvation model of Abraham et aZ.149 l5 through In this equation z is the ionic charge, E , the solvent bulk dielectric constant, E , the dielectric constant of the first solvation shell of thickness ( b - r ) and r is the ionic radius.The value of (b-r) is usually taken as that of the radius of a solvent molecule, and following Abraham and Liszi15 we chose E , as 2.0, 6~,/6T as - 1.60 x K-l, d2&,/ 6T2 as zero, and (b-r) as 0.20 nm for methanol. In eqn ( 5 ) the second term gives the contribution of El of the bulk solvent outside the first layer, and involves the temperature variation of the bulk dielectric constant, 6~,/6T and d2~,/6T2. For nearly all non-aqueous solvents the second differential with respect to temperature is not at all well established.We have collected values of E , for methanol as shown in table 4, and have fitted these values to various equations in polynomials in T/K (from T2 to T2) and in log T: E, = 138.840f0.715-(0.51500~0.00486) T+(0.5327+0.0082) x T 2 . (6) Of the various possible equations the best fit in terms of the minimum standard deviation is given by eqn (6), whence at 298.15 K, E , = 32.65, &,/dT = -0.1973 K-’ and &,/ 6T2 = (0.1066+0.0016) x lo-’ K-’. We have considered this problem in some detail, because the second term in eqn ( 5 ) is equally as important as the first term. We set out in table 5 details of the first summation of N , and El, and the resulting deduced value of C,. This configurational contribution should be related to structure-making or structure-breaking effects of ions, or in other words to the effect of ions on the solvent fluidity.Within the series of cations (table 5) values of C, can be compared without reference to any single-ion division. A plot of C, against r for the cations yields a shallow parabolic curve, reminiscent of the same plot in water [cf. fig. 2 with fig. 3 in ref. (l)] in which the organic ions have appreciably larger C, values than would be expected by extrapolation of the C, values for the inorganic cations. However, the larger-than- expected C, values in methanol are very much less than the corresponding values in water. The latter were attributed to structure-making or fluidity-decreasing properties of the organic ions due to ‘hydrophobic hydration’, and it is possible that a similar but smaller effect of ‘solvophobic solvation’ takes place with the organic cations in methanol.It is not easy to estimate the experimental uncertainty in the derived C, term that arises through errors in Asolv q, N , and El. However, even if the error in C, t This is a possible counterpart to hydrophobic hydration in water.180 Thermodynamics of Solvation of Ions Table 4. Dielectric constants of methanola 261.99 40.48 293.15 33.65 273.15 37.92 294.19 33.42 274.19 37.68 298.15 32.65 278.15 36.88 308.15 30.74 279.49 36.45 318.15 28.92 288.15 34.70 328.15 27.21 a From B. W. Davidson, Can. J. Chern., 1957, 35, 458 ; Landoldt-Bornstein, (Springer, Berlin, 1959), vol.1116, p. 632; P. S. Albright and L. J. Gosting, J. Am. Chern. Soc., 1946, 68, 1061. Table 5. The summation of AsOlvG by methods 1 and 2 ion Li+ -23 -272 Na+ -9 -171 K+ 3 -133 cs+ 13 -108 Me,N+ 48 -73 Et,N+ 66 -62 Pr,N+ 79 -56 Bu,N+ 90 -52 n-pentyl,N+ 99 -49 n-heptyl,N+ 115 -45 Ph,P+ 93 -51 Ph,As+ 94 -51 F- 1 -137 Cl- 16 -106 Br- 21 -99 I- 29 -90 ClO; 35 -83 BPh; 92 -51 157 -23 -56 -13 150 -9 -51 30 114 3 -48 29 66 13 -45 3 13 57 -38 -31 54 100 -35 -7 75 142 -33 -11 93 182 -32 -19 102 222 -30 -40 147 298 -29 -52 88 196 -31 -35 130 198 -31 6 35 1 -48 -54 18 16 -44 -44 32 21 -43 -24 35 29 -41 -14 18 35 -40 -25 137 191 -31 18 amounts to 30 YO, the values for the R,N+ cations still lie on a completely different curve to that of inorganic ions (see fig. 2).Note that whilst results on Gibbs energies, enthalpies and entropies of solvation of gaseous non-electrolytes in methanol have not revealed any such effect of ' solvophobic solvation',16 the Asolv Ci values of French and Criss9 (see table 3) do show an upward trend for the higher alkanes that might be attributed to such an effect. Since the heat capacity is a more sensitive probe of structural effects than are G, H or S, it is feasible for alkanes to behave regularly with respect to these thermodynamic functions, but not with respect to heat capacity. Because these solvophobic solvation effects in methanol, if they exist at all, are quite small, it is useful to examine other parameters that have been used to probe structural effects of ions in solution. Abraham and L i s ~ i ' ~ * ' ~ divided entropies of solvation of gaseous ions into neutral and electrostatic contributions, eqn (7), and showed l7 that the ASE values provided a quantitative measure of structure-making (negative AS,) and structure-breaking (7) AsOlv So = ASN -k ASEM. H.Abraham, Y. Marcus and K. G. Lawrence 181 180 r 6ot "\ \ 01 1 I I 1 0.1 0.2 0.3 0.4 r/ nm Fig. 2. Plot of C, us. ionic radii r/nm. 0, Inorganic cations; 0, inorganic anions; 0, R,N+. (positive AS,) effects. We have recalculated AS, values in methanol, see table 6, and if these AS, values are plotted against r, there is also a break between the inorganic and organic cations. Partial molal volumes of ions in methanol were set out by Abraham and Liszi,17 based on P(H+, methanol) = - 16.6 cm3 mol-l as suggested by Criss and Salomon.18 An expanded set, this time based on the nearly equivalent procedure P (Ph,P+) = p (BPh;), is given in table 6,29 3 9 l7 and again when values of p are plotted against r, values for the organic cations progressively deviate from the line for the inorganic cations.Other measures of structural or fluidity effects are the B'-coefficients of Engel and Hertz," and viscosity B-coefficients. There are no B'-coefficients in methanol for organic ions, but there are available''. 2o values for viscosity B coefficients. We have ourselves determined B-coefficients for Ph,PBr, NaBPh, and NaBr in methanol as well as for NaClO,, thus enabling a division of B-coefficients to be made based on B(Ph,P+) = B(BPh,), as given also in table 6.t On plotting B-coefficients against r there is a definite break between the inorganic and organic univalent cations.Thus all three parameters, AS,, v and B-coefficients suggest that the organic ions are anomalous in methanol with respect to the inorganic cations. The C, values indicate that the organic ions have a structure-making or fluidity-decreasing solvophobic effect, and this is confirmed by their AS, values (more negative than expected), their v" values (more positive than expected) and their B-coefficients (more positive than expected). We stress, however, that these solvophobic effects are much less than the corresponding hydrophobic hydration effects in water. In addition, note that the viscosity B-coefficients reflect a kinetic rather than a thermodynamic process.In the above comparisons between the univalent inorganic cations and the univalent organic cations, any particular division into single-ion values is irrelevant. This is not so, however, for comparisons between cation and anion values, and any comparison between C, values for cations and C, values for anions must involve the single-ion assumption used. These C, values calculated in table 5 suggest that all the inorganic univalent cations and anions are structure-makers (C, positive), in exact accord with previous conclusions reached by Abraham and Liszi" and by Engel and Hertz." For ions in water, we showed' that the single-ion division based on Cp(Ph,P+) = C,"(BPh,) led to C, values that were entirely compatible with single-ion V' values and single-ion viscosity B coefficients derived from the corresponding assumption. In the same way, plots of C , for single ions in methanol against r, ASE, p and B (see fig.2-5) lead to the t It is interesting that the division in table 6 is exactly the same as that originally proposed by Abraham and Liszi. l7182 Thermodynamics of Soluation of Ions Table 6. Single-ion values of ASE, and B- coefficients in methanol at 298 K ion ASEa V b B" H+ Li+ Na+ K+ Rb+ cs+ Me,N+ Et,N+ Pr,N+ Bu,N+ Pe,N+ Hept 4N+ Bu,P+ Ph,P+ Ph,As+ F- c1- Br- I- BBu, BPh; ClO, - - 220 - 188 - 139 - 118 - 95 - 94 -110 - 191 - - - - - - 121 - 42 - 33 - 12 - 26 - - - 128 - 20.8 - 20.2 - 18.8 - 8.8 -4.8 1.7 65.1 123.8 193.8 264.0 336.7 472.9 280.1 261.8 267.5 1.8 15.5 22.8 30.5 39.1 278.0 261.8 - 0.63 0.59 0.56 0.44 0.36 0.15 0.30 0.48 0.67 0.80 - - 0.92 - - 0.20 0.18 0.1 1 0.03 0.92 - ~ a In J K-lmo1-l based on the Ph,As+/BPh; assumption using Ahyd s" values from Y.Marcus and A. Loewenschuss, J. Chern. Soc., Faraday Trans. I, 1986,12,993 and Y. Marcus, J. Chern. SOC., Faraday Trans. I, 1986,82,233, At s" values from M. H. Abraham, T. Hill, H. C. Ling, R. A. Schulz, and R. A. C. Watt, J. Chern. SOC., Faraday Trans. I, 1984, 80, 489, and a revised equation for the neutral contribution: A S (in J K-lmo1-l on the mole fraction scale) = - 27 1.76r/nm - 20.4 14. In cm3 mol-1 from ref. (2): (3) and (17). In dm3 mol-1 from ref. (17) and (20), and this work (Ph,P+, BPh;, CIO;). conclusion that the single-ion division we have carried out for C; values for ions in methanol is quite compatible with all these other sets of single-ion values.In particular, all the univalent inorganic ions (Li+ to Cs+ and F- to C1OJ are structure-makers with positive C, values, negative AS, values and positive viscosity B-coefficients, and in the plots of C, against r, ASE, Yo and B, all the above ions lie on the same line (within, it must be said, a rather large experimental error). However, with the recent single-ion division of Shin and Criss,* this would not be the case, there being a difference of ca. +60 J K-l mol-' between this division and that given in table 2, mainly arising because Shin and Criss use a value of 597 J K-' mol-1 for C;(NaBPh,) whereas we have used the alternative value of 536 J K-' mol-', see table 1. In the second summation, N , is calculated from the curve in fig.1, so that this neutral component should then include any effects of solvophobic solvation in the outer solvation sphere. If we then calculate E, as the electrostatic effect in this outer sphere, the resulting value of C, should then refer only to the inner solvation sphere in whichM. H. Abraham, Y. Marcus and K. G. Lawrence 180 120 c1 60 0 -240 -180 -120 -60 0 ASE Fig. 3. Plot of C, us. AS,/J K-l mol-'. Symbols as in fig. 2. 180 r 60 0 -20 0 20 40 - 60 80 100 120 V Fig. 4. Plot of C, us. partial molal volumes V"/cm3 mol-l. Symbols as in fig. 2. 180 120 c1 60 0 0 0.2 0.4 0.6 0.8 B Fig. 5. Plot of C, us. viscosity B-coefficients, B/dm3 mol-'. Symbols as in fig. 2. 183184 Thermodynamics of Solvation of Ions case values for all the cations should be comparable.It is not too easy to estimate a radius for the outer-sphere commencement: as before,' we took a value of twice the solvent radius and for methanol calculated E2 as the electrostatic effect beyond a radius, b, of (r +0.4) nm. Calculations on these lines are given in table 5. If C, is due to a decrease in heat capacity of the centrally orientated solvent molecules near the ion,l then as the ionic radius increases, the number of solvent molecules thus orientated increases, and hence C, should become more negative as r becomes larger. A plot of C, against r shows that for all the cations there is a roughly linear connection between C, and r, so that this method of summation has removed any possible effect of solvophobic solvation, as indeed it should.Conclusions The initial assumption that C;(Ph,As+) = C,"(BPh;) in methanol yields single-ion values that are in reasonable agreement with other parameters that relate to structural effects of ions. After subtracting out a neutral term, N,, and an electrostatic term, El, there results a configurational heat-capacity term C, that is positive for all the (univalent) ions studied, indicating that all these ions are structure-makers in methanol. The connection of C, with ASE, P and B for the inorganic univalent cations and anions is particularly strong evidence that the ions (Li+ to Cs+ and F- to ClOh) decrease the fluidity of the solvent. However, the various plots of C, against r, AS,, P and B show also that the R,N+ ions behave differently to the inorganic cations, this difference increasing as the size of the R,N+ ions increases. The R,N+ ions are still structure makers and still decrease the fluidity of the solvent, and our suggestion is that these ions interact with methanol through ' solvophobic solvation ', an effect analogous to, but much smaller than, 'hydrophobic hydration' in water.If this effect is taken into account, as with the configurational quantity C,, then in a plot of C, against r, all the cations including the R,N+ cations lie on roughly the same line. One very marked difference between results in methanol (table 6) and our previous results for univalent ions in water is that in methanol all the ions we have studied are structure-makers that decrease the fluidity of the solvent.In water, on the other hand, large inorganic ions such as Cs+, I- and ClO, are structure-breakers that increase the fluidity of the solvent. As with our previous single-ion division with water as solvent, two separate single-ion assumptions are involved in the calculation of C, and C,. First, the single-ion C," values depend on the assumption that C;(Ph,As+) = C;(BPhJ, and secondly the calculations of the E and N terms involve the implicit assumption that cations and anions of equal radii (and equal charge) can be treated similarly. These two assumptions seem to be quite compatible, as judged from fig. 2-5. If we had used the assumption that C;(Ph,P+) = CF(BPh,) there would have been a discrepancy of ca. +20 J K-' mol-' (using NaBPh, = 536) or ca.f 50 J K-' mol-' (using NaBPh, = 597) between cation and anion values. However, in view of the experimental uncertainties in CF values, and having regard to the other particular assumptions and calculations involved, we do not feel that these differences are too significant. References 1 M. H. Abraham and Y. Marcus, J . Chem. Soc., Faraday Trans. I , 1986, 82, 3255. 2 C. Jolicoeur, P. R. Philip, G. Perron, P. A. Leduc and J. E. Desnoyers, Can. J . Chem., 1972, 50, 3 A. J. Pasztor and C. M. Criss, J. Solution Chem., 1978,7,27; R. N. French and C. M. Criss, J. Solution 4 C. Shin and C. M. Criss, J . Solution Chem., 1986, 15, 307. 5 M. Mastroianni and C. M. Criss, J. Chem. Eng. Data, 1972, 17, 222. 6 C. Shin and C. M. Criss, J . Solution Chem., 1978, 7, 205. 3167. Chem., 1982, 11, 625.M . H . Abraham, Y . Marcus and K . G. Lawrence 185 7 Y. S. Choi and C. M. Criss, Faraday Discuss. Chem. SOC., 1978, 64, 204. 8 C. Shin and C. M. Criss, unpublished results quoted in ref. (4). 9 R. W. French and C. M. Criss, J. Solution Chem., 1981, 10, 231. 10 D. R. Stull, E. F. Westrum and G. C. Sinke, The Chemical Thermodynamics of Organic Compounds 11 A. Lannung, J. Am. Chem. SOC., 1930, 52, 68. 12 Solubility Data Series, vol. 5/6, ed. C. L. Young (Pergamon, Oxford, 1981); vol. 10, ed. R. Battino 13 K. Takeuchi, K. Matsumura and K. Yaginuma, Fluid Phase Equilibria, 1983, 14, 255. 14 M. H. Abraham, E. Matteoli and J. Liszi, J. Chem. SOC., Faraday Trans. I , 1983, 79, 2781. 15 M. H. Abraham and J. Liszi, J . Chem. SOC., Faraday Trans. I , 1978, 74, 1604; 2858. 16 M. H. Abraham, J. Am. Chem. SOC., 1982, 104, 2085. 17 M. H. Abraham, J. Liszi and E. Papp, J . Chem. SOC., Faraday Trans. I , 1982, 78, 197. 18 C. M. Criss and M. Salomon, in Physical Chemistry of ionic Solutions, ed. B. E. Conway and R. G. 19 G. Engel and H. G. Hertz, Ber. Bunsenges. Phys. Chem., 1968, 72, 808. 20 C. M. Criss and M. Mastroianni, J . Phys. Chem., 1971, 72, 2532. (Wiley, New York, 1969). (Pergamon, Oxford, 1982). Barradas (Wiley, New York, 1966). Paper 71205; Received 5th February, 1987

ISSN:0300-9599    

DOI:10.1039/F19888400175

出版商:RSC    

年代:1988

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I , 1988, 84(1), 187-196 Molecular Motions in (CH,),CCl by lH Spin-Lattice Relaxation Tooru Hasebe and Sachie Ohtani Department of Chemistry, Faculty of Education, Fukushima University, Mat uska wa- mac h i , Fuk ush ima 960- 12, Japan Molecular motions in (CH,),CCl have been further investigated by 'H n.m.r. relaxation times (T, and T ) measured in all solid and liquid phases. In the lowest temperature phase &hase IV), two minima are observed at Tlp and are attributed to uniaxial molecular and methyl reorientation. Two minima at Tl were also indicated but only one was clearly visible. Analysis of the Tlp and TI data in phase IV gives the activation parameters E, = 20.3f::; kJ mol-' and z, = (2.8 & ;::) x s for methyl reorientation and E, = 15.8 f 0.1 kJ mol-l and z, = (6.7 f s for uniaxial molecular re- orientation. Uniaxial molecular reorientation is found to be faster than methyl reorientation as in the case of (CH,),CCN.In phase I11 one minimum at T, is observed due to methyl reorientation with activation parameters of E, = 17.64-i:: kJ mol-' and z, = (1.7f ::;) x lo-', s. In phase I1 we could not obtain enough variation in Tl and Tlp to be able to establish the details of the molecular motions since the existence of this phase only extends over ca. 2 K, between 217.7 and 219.5 K. In phase I (plastic phase), T, and TIP are governed by translational self-diffusion with an activation energy of 36.2 & i:: kJ mol-' and z, = (1.8 & :::) x s. The mean jump time of the molecules at the melting point is 7.4 x lO-'s.The activation energy for the molecular motion in the liquid phase is 10.9 & 1.4 kJ mol-'. x The compounds of the type (CH,),CX (X=Cl, Br, I, NO, or CN) have similar molecular shapes and show similar molecular motions and thermodynamic behaviour. Without exception, these compounds exhibit at least one solid-solid phase transition. At temperatures below the transition from phase I to I1 (phases are designated I, 11,111 etc. in order of decreasing temperature), overall molecular reorientation is much more restricted. On cooling, overall molecular reorientation is usually the first motion to be restricted. This is followed by restriction of the internal motions of the t-butyl group (the uniaxial molecular reorientation about the C-X axis) and the methyl group (methyl group reorientation about the C-CH, axis).The relative rates of the latter two motions have been the subject of much recent debate, for example in (CH,),CCN1-3 and (CH3)3CC1.4-8 The debate in the case of (CH,),CCN has been settled recently with the characterisation of molecular motion in the various phases., However, the quantitative characterisation of molecular motion in (CH,),CCl, has not been settled to our knowledge. Recently it has been reported that solid (CH,),CCl under its own vapour pressure, has three solid-solid phase transition^.^ Phase I is an orientationally disordered phase (plastic crystal phase) and the others are brittle crystal phases. In order to obtain a more complete understanding, we measured 'H n.m.r.spin-lattice relaxation times (7'' and Tip) in liquid and solid (CH,),CCl over a wide temperature range and we give a detailed analysis of the molecular motions in each phase. 187188 Molecular Motions in (CH,),CCl Experimental 2-Chloro-2-methylpropane obtained from Wako Pure Chemical Industries Ltd was purified by normal distillation (b.p. 324.0-324.1 K), followed by two-stage vacuum distillation and degassed by the freeze-pumpthaw technique. The specimen was sealed under vacuum in a glass ampoule of outside diameter 10 mm. No impurities in this specimen were detected by gas-chromatographic analysis or by high-resolution 'H n.m.r. measurements of the neat liquid sample, indicating a purity of greater than 99.99 %. The melting point and three solid-solid transition points of this specimen were determined to be 248.2, 219.5, 217.7 and 182.9f0.3 K, respectively, by differential thermal analysis.The proton spin-lattice relaxation times in the laboratory frame, T,, and in the rotating frame, Tlp, were measured using a Bruker CXP 4-60 MHz pulsed n.m.r. spectrometer. Tl was measured at 7.0, 10.0 and 15.0 MHz using the saturation recovery method [(71/2-z1),-n/2-z-71/2 pulse sequence with the condition T,* < z, < TIJ. The Tip. values were measured at 15.0 MHz in a resonant rotating radiofrequency (r.f.) magnetic field (B,) of 2.0 mT for all solid phases and also of 1.28 and 0.2 mT for phase I. The spin-locking sequence employed a 71/2 pulse followed immediately by a long pulse whose r.f. phase was shifted by 7r/2 relative to the first pulse.The T, and Tlp measurements were carried out between roughly 90 K and the melting point, 248 K, (and up to 280.4 K for Tl at 10.0 MHz). The errors in T, determinations were estimated to be less than The temperature was measured with a chromel-P/constantan thermocouple to an accuracy of f0.3 K and was controlled to within 0.2 K. 10 % and those in Tlp were less than f 15 YO. Results Fig. 1 shows the temperature dependence of T, at 7, 10 and 15 MHz for the various phases of (CH,),CCI. Discontinuities in the Tl curve were observed at 219.5 (I1 ++ I) and 248.2 K (melting) due to first-order phase transitions. However, discontinuities were not observed at the IV c-, I11 (at 182.9 K) and I11 ++ I1 (at 21 7.7 K) transitions, which were also first-order phase transitions.Two minima at Tl were observed in the temperature region of phases I11 and IV as shown in fig. 1. A double minimum of T, was observed in the lowest-temperature phase (phase IV) at 7 MHz. The minimum values of TI were 3.6k0.1 ms at 129.7 K and 15 f 1 ms at 182.9 K. The ratio of these minima, 4.2, is important and will be discussed in the next section. Only one minimum at T, was observed in phase IV at the other frequencies; this was 5.0f0.2 ms at 133.3 K for 10 MHz and 7.3 f0.2 ms at 136.5 K for 15 MHz. Another minimum in Tl was observed in phase 111, which was 20.5f0.5 ms at 192.3 K for 10 MHz (and 34f 1 ms at 200.0 K for 15 MHz). In phase I1 no significant temperature dependence of TI was observed because the phase extends over only 2 K.At the transition from phase I1 to I (at 219.5 K), T, shows a jump from 38 ms (48 ms for 15 MHz) to 5.0 s (6.0 s for 15 MHz), and then decreases with increasing temperature. At the meltifig point (248.2 K), Tl jumps from 1.5 s (2.4 s for 15 MHz) to 4.2 s, and then increases gradually with increasing temperature. The rotating frame relaxation time, Tlg, in various phases at the r.f. spin-locking field intensity of B, = 2.0 mT (also B, = 1.28 mT and 0.20 mT in phase I), together with T, at 15 MHz, are shown in fig. 2. Two minima in Tlp were clearly observed in phase IV; they are 150 f 5 ,us at 101.3 K and 670 f 20 ,us at 135.1 K. Again we note the ratio of the minimum values of Tlp, 4.5, and that this is equal to the ratio of the T, minimum values within experimental error.No discontinuity was observed at the IV-I11 or 111-11 phase transition. At the transition from phase I1 to I, Tlp jumps from ca. 34 ms to 6.6 ms, andT. Hasebe and S. Ohtani I * l o , \ I ' a 88 A+a 0 1 : 0 A A$. AA 0 A A .oOO 2 10-1 = - l(r2: Tnl TII-I11 TIII-IV I I 1 1 I 1 I I 189 l o t ' ' ' ' ' , I # - ** .I I - . . a I T190 Molecular Motions in (CH,),CCl then decreases to a minimum value (3.5 rfr 0.1 ms at 233.4 K) with increasing temperature. After passing the minimum, Tlp increases with increasing temperature. A less distinct minimum was observed at 1.28 and none at 0.20 mT. Analysis and Discussion Phase IV Earlier 'H n.m.r. measurements4* on solid (CH,),CCl reported double minima at TI and Tlp in the lowest-temperature phase (phase IV).These features were assigned to two motions, methyl reorientation (C, reorientation) and uniaxial molecular reorientation (C; reorientation). The Tl and Tlp minimum in the low-temperature region was attributed to C3 reorientation and the Tl and Tlp minimum in the high-temperature region was attributed to C; reorientation. Furthermore it was stated that the methyl reorientation should be faster than the uniaxial molecular reorientation in the higher temperature phase (phase 111). These assignments were challenged by the recent neutrOn scattering study by Frost et al.7 A careful and detailed analysis of our Tl and Tlp data resolves this discrepancy as discussed below. The previously published 'H n.m.r. lineshape investigationlo indicated that the dominant motional modes which govern Tl and Tlp are the C, and C; reorientations in phase IV.We therefore calculate the relaxation ratet due to the dipolar interaction modulated by C, and C; reorientations as where rli is the interproton distance between proton 1 and proton i in a methyl group, lIj is the distance between proton 1 and proton j of other methyl groups on the molecules, z, and z, are the correlation times for the C, and C; reorientations, (2) respectively, and z, is defined by The spectral density function B(z) is, for simplicity, assumed to have the form (3) where ci) is the resonant angular frequency. In eqn (1) we assume that the C, and Ci reorientations are mutually independent and ignore any correlation between these motions. We now consider the two different cases following the approach of ref.( 3 ) : (a) C, reorientation is fast and C; reorientation is slow i.e. z, < z, in eqn (1); (h) that C; reorientation is faster than C, reorientation i.e. z, > z, in eqn (1). In case (a) the magnetic nuclear dipoledipole interaction modulated by a C, reorientation is calculated to be 0.158 mT2 from eqn (1) with z, -P GO and appropriate molecular geometry. $ Then the residual magnetic interaction modulated by the subsequent motion (C, reorientation) is calculated to be 0.065 mT2. Here we have assumed that the C, reorientation is much faster than the C; motion and that the three methyl protons may be treated as if at a distance, Zij, from other methyl protons which is 0.310 nm. This is the distance between the centres of the equilateral triangles formed by each methyl group.The ratio of these interaction strengths is calculated to be 2.4 which should correspond to the ratio of the Tl minimum values, TIIImin/TILmin. The subscripts L and H denote molecular motions which are responsible for Tl in the low- temperature region and in the high-temperature region, respectively. If we include intermolecular contributions, the ratio is estimated to be 2.9. conclusions stated in a previous paper are unchanged. used the geometry data of an analogous molecule (CD,),CCN." l/z, = l/z, + 1/TM. B(z) = z/( 1 + cu2z2) + 42/( 1 + 4u2t2) 7 In our previous paper3 we overestimated the intermethyl contribution to the relaxation rate. However the $ There are no experimental molecular geometry data for (CH,),CCl reported, to our knowledge, so weT.Hasebe and S. Ohtani 191 In case (b) the magnetic interaction modulated by the C; reorientation is expected to be 0.183 mT2 from eqn (1) (with z, + a). The interaction modulated by the slower C, reorientation is then calculated to be 0.047 mT2. The ratio of these values obtained was 3.9, so this predicts a ratio of TIHmin/TILmin (= 3.9). Including intermolecular interactions the ratio becomes 4.4 Using the same assumption and approximations we have obtained an expression for Tlp in the weak-collision limit12 by converting the spectral density function B(z) to G(z) in eqn (I), which gives (4) y"2 9 y4A2 i - 2 120r;$ 4 0 j 4 { j G(T) = 2.52/( 1 + 02z2) + z/( 1 + 402z2) + I .5z/( 1 + 4 ~ 3 ~ ) K ; = z - [8G(zm)+8G(z,)+ 19G(~,)]+- C -G(zM) where G(z) has the form and w, = yB,.Recently Ripmeester and Katcliffe8i13 pointed out that the C, and C; motions are practically indistinguishable from 'H n.m.r. lineshapes and T, measurement alone. However, as mentioned above, if one observes a double minimum at T, or Tip, one can easily distinguish between them by comparing the relaxation times at the minima. Unfortunately, if one does not obtain such a double minimum of Tl or Tlp, then it is impossible to distinguish between them clearly. Fortunately we could observe clear double minima at T, and Tlp in (CH,),CCl (with ratios 4.2 for Tl and 4.5 for Tip), and these values clearly demonstrate that the uniaxial molecular reorientation (C;) should be faster than the methyl group reorientation (C,) in phase IV. The corresponding minimum values of T, and Tlp for four t-butyl compounds [(CH,),CCN,3 (CH,),CCl, (CH,),CI14 and (CH,),CBr15] are listed in table 1 as reference data.The first three compounds show that the C; motion is faster than the C, motion in their lowest temperature phase. Activation parameters for these two types of motion were obtained by fitting the data of fig. 1 to eqn (l), using the molecular geometry mentioned above, and assuming the Arrhenius equation for the temperature dependence of the correlation time for both the ( 5 ) z = z, exp (E,/RT) C, and the C; reorientation: where T, is the pre-exponential factor and E, is the activation energy. The activation parameters obtained for the methyl group reorientation and the uniaxial molecular reorientation are listed in table 2 and the Tl values reproduced with these parameters are shown in fig.3 by the solid lines for 7 and 10 MHz, respectively. The dotted lines in fig, 3 are the T, values calculated at 7 and 10 MHz, respectively, for the case of z, > z, using activation parameters E, = 15.8 kJ mol-l and z, = 6.7 x s for uniaxial molecular reorientation. The solid lines (for the case z, > zM) in fig. 3 show much better agreement with the experimental data over the whole temperature range. The Tlp values calculated at 15 MHz and B, = 2.0 mT using eqn (4H6) and activation parameters listed in table 2 are shown in fig. 2 by a solid line. The calculated Tlp in fig. 2 is also in good agreement with the experimental Tlp data. Our analysis of T, and Tlp data therefore shows that the C, reorientation and the C; reorientation are distinguishable and that the C; reorientation is faster than the C, reorientation in phase IV.Furthermore, our result is consistent with the result of a neutron scattering s t ~ d y , ~ in that any random molecular motions are slower than 3 x s (below 165 K) in phase IV. We estimate correlation times at 165 K to be 7.5 x lO-'s for the C, reorientation and 6.7 x 10-l'~ for the C; reorientation. s for methyl group reorientation and E, = 20.3 kJ mol-' and z, = 2.8 x 7 FAR ITable 1. Minimum values in the high- and in the low-temperature regions of TI and Tlp, and their ratio T1H min TlpH min T1L min/mS 'IMHz TIL min TlpH minlmS TlpL T1pId min - - - (CH,),CCN (56 + 1)" 8.1 kO.1 18.7 7 (CH,),CCl 15.0 5- 1 .O 3.6kO.l 7.0 4.2 0.67 f 0.02 0.150 & 0.005 { 15.0 MHz, B, = 2.0 mT) 4.5 (CH,),CI (12.3k0.3)" 3.0 f O .l 7.0 4.1 15.0 1.9 0.21 kO.01 0.085+0.003 (15.0 MHz, B, = 0.7 mT) 2.6 20.0 2.0 0.53f0.08 0.184+0.020 (15.0 MHz, B, = 2.0 mT) 2.9 (CH,),CBr 16.7 & 0.2 22.1 f 0.2 - - - - a These values were obtained from the temperature dependence of T i H which was obtained using the equation, qtbs = qA+ q;, where subscripts L and H denote molecular motions which are responsible for T, in the low-temperature region and in the high-temperature region, respectively. Table 2. Activation parameters obtained in this study for the various motions in (CH,),CCl motion parameters phase IV phase I11 phase I1 phase I uniaxial molecular EJkJ mol-' 15.8 f O .1 15.8 methyl translational EJkJ mol-I - - - (6.7 +;:;) x 10-15 6.7 x 10-15 - (2.8 f ;:;) x 10-14 - - - (1.8 k x 1 0 4 4 - - reorientation ZO/S reorientation Z,/S self-diffusion ZO/S EJkJ mol-' 20.3 f t:: 17.6 f t:; - (1.7 + ;:;) x 10-13 - - - - 36.2 & i::T. Hasebe and S. Ohtani 193 1 I 1 5 7 9 1 1 103 KIT Fig. 3. Temperature dependence of TI in phase IV of (CH,),CCl. The solid and the dotted lines are calculated values using eqn (1) as explained in the text: [(-) where t, > z,, and (-----) where z, > r,]; 0, 7 and 0, 10 MHz. Phase I11 Our Tl and Tl< data do not change abruptly at the transition point, TII,-,v. This indicates that the motion responsible for Tl and Tlp is attributed to the methyl group reorientation near the transition point.If any other motions governed Tl or Tlp at the transition, a more drastic change in Tl or Tlp would be expected. The early ‘H n.m.r. lineshape investigationlo showed that the transition to the orientationally disordered (plastic) phase occurred at 220 K. On the other hand, the ‘H n.m.r. Tl data at higher resonant frequency (43 MHz) obtained by O’Reilly et aL5 showed an abrupt jump at the transition point This can be explained because there is very little contribution from the methyl group reorientation to Tl at 43 MHz near the transition point in phase IV; in this case the relaxation conditions are wz, 8 1 and then Tlc, (43 MHz) > Tlc, (7-15 MHz), where the subscript ‘C3’ indicates the contribution to Tl from the methyl group reorientation. We then expect T, observed at 43 MHz to be determined by the Ci reorientation and this contribution will be approximately the same as at 7-15 MHz [from eqn (l)].The fact that the minimum at T, shifts to higher temperature with increased resonant frequency and the behaviour Tl (at 43 MHz) and 7-15 MHz at the transition point (182.9 K) support the assumption that the motion responsible for Tl and Tlp in phase I11 is methyl group reorientation and the results of O’Reilly et al. indicate a discontinuous increase in Ci reorientation at the IV -, 111 transition. Using eqn (1) and (6) and the activation parameters for C; reorientation obtained in phase IV, the activation parameters for the methyl group reorientation are obtained and listed in table 2. The solid lines in fig.4 represent calculated values at 7 and 15 MHz using eqn (1) and the activation parameters listed in table 2. The dotted lines are calculated values at 7 and 15 MHz when Ci reorientation governs T,, i.e. z, > z, (with Ea = 15.8 kJ mol-’ and z, = 6.7 x s for the C, reorientation and Ea = 17.6 kJ mol-1 and z, = 1.7 x lo-’, s for the Ci reorientation) in phase 111. The solid lines are in better agreement with the experimental data than the dotted lines. Our result, that C, reorientation is slower than the C; reorientation in phase 111, is 7-2194 0-' lo-* Molecular Motions in (CH,),CCl 1 I I I I I 4 5 6 lo3 KIT Fig. 4. Temperature dependence of TI in phase I11 of (CH,),CCI. The solid and the dotted lines are calculated values as in fig. 3 and also as explained in the text: (a) 0, 15 and (b) 0, 7 MHz.again consistent with the results of the neutron scattering s t ~ d y . ~ The correlation times of the two motions (C, and C;) in this phase (182.9-217.7 K) are 1.8 x 10-8-2.8 x s for the C3 reorientation and < 4 x lo-'' s for the C ; reorientation, respectively, and this result is in reasonable agreement with the results [( 1.4 x s at 193 K to 1 .O x lo-'' s at 208 K)' or (ca. 4 x lo-', s at 202 K)7 for the C; reorientation] of the neutron scattering investigations. Unfortunately the lack of Tl and T data in phase TI gives no more information than that reported in our previous pape;,% i.e. the onset of hindered molecular motion, believed to be molecular tumbling, was observed. Phase I1 Phase I is a plastic crystalline phase with a low entropy of fusion (AS = 7.99 J mol-' K-').16 A previous dielectric in~estigation'~ gave an activation energy of 4.9 kJ mol-1 for motion of the molecular dipole in this phase.A comparison with the results in the lower- temperature phases and the depth of the Tlp minimum led to the conclusion that translational self-diffusion contributes to relaxation throughout most of this phase. The assignment is supported by the fact that the variations of Tl and Tlp at high temperatures correspond to an activation energy of ca. 30 kJ mol-'. According to the isotropic diffusion modella which is a good model for plastic crystals with the low entropy of fusion, the relaxation rate c; is given as 22 202 32T. Hasebe and S. Ohtani 195 lo-' -4 phase I 6 10-4 3 4 5 lo3 KIT Fig 5.Temperature dependence of TI, (at 15 MHz) for these different values of B, in phase I of (CH,),CCl. Values of B, as in fig. 2. The lines are the calculated values of Tlp [(-.-.-) 0.2, (-----) 1.28 and (-) 2.0 mT] using eqn (7) and the activation parameters in table 2. and T;plmin = 0.25 (M2)/~1 (8) where ( M , ) is the second moment modulated by translational self-diffusion and z is the mean time between diffusional jumps of molecules. The value of (M,) = 0.008 54 mT2 obtained using eqn (8) and the Tlp minimum value at B, = 2.0 mT is in good agreement with the calculated value of 0.00890 mT2 using crystal structure data (f.c.c.; a = 0.865 nm at 238 K).19 The activation parameters for the translational self-diffusion were found to be E, = 36.2 &::: kJ mol-and z, = (1.8 j-:::) x s from the Tlp data and eqn (6H8).The lines in fig. 5 shows the Tlp values calculated using eqn (7) and these activation parameters. They are well fitted to the experimental TIP data. The mean jump time at the melting point was then found to be 7.5 x lo-' s, which is similar to the mean jump times found in many plastic crystals.20 It is interesting that the translational self-diffusion in the plastic phase of analogous compounds shows almost the same activation energy : E, = 36.4 kJ mo1-1 for (CH,), CBr21 with A S = 7.48 J mol-1 K-l l6 and E, = 35.3 kJ mo1-1 for (CH,),CI.'* Liquid The activation energy for molecular motion in the liquid state obtained from the slope of the T, (fig. 1) was 10.9+ 1.4 kJ mol-'. Furthermore, the activataion energy is more than double the value of overall molecular tumbling in liquid state reported by O'Reilly196 Molecular Motions in (CH,),CCl et aL5 It is very interesting that the ratio of the activation energy of the motion to that of molecular reorientation obtained by dielectric measurements is 2.2, which is the same as that found for (CH,),CCN3 and (CH3)3CN02.22 Conclusions In this paper molecular motions responsible for lH n.m.r.relaxation in each phase were identified and their activation parameters were reported. In phase I the motion controlling Tlp and TI was assigned to molecular self-diffusion. The activation parameters obtained are comparable to those found in analogous compounds. The range of phase I1 was inadequate to allow much information to be obtained.In the lower temperature phases, methyl group ( C3 reorientation) and uniaxial molecular reorien- tation (Ci reorientation) were discussed and we established that the C; reorientation is faster than the C, reorientation in phases I11 and IV. The results are in good quantitative agreement with the results of previous neutron scattering studies.6* We are grateful to Professor J. H. Strange (University of Kent at Canterbury) for helpful discussions. References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 Z. M. El Saffar, P. Schultz and E. F. Meyer, J. Chem. Phys., 1972, 56, 1477. J. C. Frost, A. J. Leadbetter and R. M. Richardson, Faraday Discuss. Chem. SOC., 1980, 69, 32. T. Hasebe and J. H. Strange, J. Chem. SOC., Faraday Trans. 2, 1985, 81, 735. E. 0. Stejskal, E. D. Woessner, T. C. Farrar and H. S. Gutowsky, J. Chem. Phys., 1959, 31, 55. D. E. O’Reilly, E. M. Peterson, C. E. Scheie and E. Sayfarth, J. Chem. Phys., 1973, 59, 3576. P. S. Goyal, W. Nawrocik, S. Urban, J. Domoslawski and I. Natkaniec, Acta Phys. Pol. A, 1974, 46, 399. J. C. Frost, A. J. Leadbetter and R. M. Richardson, Philos. Trans. R. SOC. London, Ser. B, 1980,290, 567. J. A. Ripmeester and C. I. Ratcliffe, J. Chem. Phys., 1985, 82, 1053. S. Ohtani and T. Hasebe, Chem. Lett., 1986, 1283. J. G. Powles and H. C. Gutowsky, J. Chem. Phys., 1953, 21, 1704. J. C. Frost, A. J. Leadbetter, R. M. Richardson, R. C. Ward, J. W. Goodby, G. W. Gray and G. S . Pawley, J. Chem. SOC., Faraday Trans. 2, 1982, 78, 179. D. C. Look and I. J. Lowe, J. Chem. Phys., 1966, 44, 2995. C. I. Ratcliffe and J. A. Ripmeester, Can. J. Chem., 1986, 64, 1348. T. Hasebe, J. M. Chezeau and J. H. Strange, to be published. T. Hasebe, J. H. Strange, N. Nakamura and H. Chihara, J. Chem. SOC., Faraday Trans. 2, 1985, 81, 749. S . Urban, Adv. Mol. Relaxation Processes, 1981, 21, 221. S. Urban, J. A. Janik, J. Lenik, J. Mauer, T. Waluga and S . Wrobel, Phys. Status Solidi A, 1972, 10, 271. H. C. Torrey, Phys. Rev., 1953, 92,962. S. Urban, J. Domoslawski and Z. Tomkowicz, Muter. Sci., 1978, 4, 91. J. M. Chezeau and J. H. Strange, Phys. Rep., 1979, 53, 1. T. Hasebe and S . Ohtani, Fukushima Univ. Sci. Rep., 1986, 38, 39. T. Hasebe, N. Nakamura and H. Chihara, Bull. Chem. SOC. Jpn, 1984, 57, 179. Paper 71206; Received 5th February, 1987

ISSN:0300-9599    

DOI:10.1039/F19888400187

出版商:RSC    

年代:1988

数据来源: RSC

摘要:

J. Chem. Soc., Faraday Trans. I, 1988, 84(1), 197-205 Effect of Surface Hydroxyl Density on Photocatalytic Oxygen Generation in Aqueous TiO, Suspensions Yoshinao Oosawa? and Michael Gratzel" Institut de Chimie Physique, Ecole Polytechnique Fe'de'rale, CH-1015 Lausanne, Switzerland TiO, powders with various surface hydroxyl densities have been prepared by thermal treatment (600-1000 "C), and their activity for photocatalytic oxygen evolution in aqueous suspensions has been investigated in the presence of various electron acceptors, i.e. AgNO,, K,PtCl, and KBrO,. The activity increased drastically with the thermal treatment. The oxygen evolution rate was found to be inversely proportional to the surface concentration of OH-, indicating an inhibitory effect of surface hydroxyl groups with respect to photocatalytic oxygen evolution.On highly hydroxylated samples the formation of surface bound peroxides appears to be favoured over oxygen generation. Oxygen evolution from TiO, electrodes by band-gap irradiation is well known as a water decomposition process utilizing light energy. Furthermore photocatalytic decom- position of water also occurs over TiO, powders loaded with precious metals., In these cases it has been assumed that oxygen evolution is brought about through the oxidation of surface hydroxyl groups by positive holes3 photogenerated in the valence bond. Hydrogen is always detected during photocatalytic water decomposition over TiO, powders, but often oxygen does not appear in the gas phase.4 The lack of oxygen release in the gas phase can be rationalized in terms of the generation of surface-bound peroxides by the hole reaction.** Alternatively, oxygen formed by the hole reaction might be photoadsorbed by TiO, and hence escape detection in the gas phase.'.' It is known that hydroxyl groups at the TiO, surface participate in the photoadsorption of oxygen by Ti0,.7 Thus the role of surface hydroxyl groups in photoinduced surface reactions over TiO, is important.The density of hydroxyl groups on the TiO, surface can be varied by thermal treatment.* In the present study we use this effect in order to explore how the surface hydroxyl density affects the efficiency of photocatalytic 0, generation from TiO, (rutile) dispersions in the presence of electron acceptors such as PtCl:-, BrO; and Ag+.We have previously published a short communication dealing with the Ag+ system. Experiment a1 Material In most experiments, TiO, rutile (a gift from Dr Panek, Bayer AG, Federal Republic of Germany) with a surface area of 80 m2 g-l was employed as the starting material for thermal treatment. A few investigations were carried out with commercial TiO, (anatase) samples supplied by Fluka, Fuji-Titanium (TP-2), Merck (LAB) and Degussa (P-25). The latter is a mixture of 60 % anatase and 40 % rutile. The duration of the thermal treatment carried out at a fixed temperature was 7 h. A stream of Ar or 0, was passed over the samples during heating. Visiting scientist from the National Chemical Laboratory for Industry at Tsukuba, Japan. 197198 Photocatalytic Oxygen Evolution Table 1.Surface properties of TiO, (Bayer, rutile) after thermal treatment specific surface surface area OH density" T/"C atmosphere /m2 g-' /FS 8-l 79.5 333 600 Ar 21.3 106 700 Ar 22.3 67 800 Ar 23.2 43 900 Ar 13.3 17 1000 Ar 5.1 6 1000 0 2 9.0 13 as supplied - " See text. Apparatus Investigations were carried out with 20 cm3 samples contained in Pyrex glass vials (37 cm3 in volume) equipped with a septum. Prior to irradiation, the TiO, dispersions were sonicated and freed from air by purging with argon. The dispersions were vigorously stirred during irradiation. A Hanau Suntest lamp, (global irradiance 70 W ern-,) whose emission spectrum closely mimics solar radiation, was used as a light source. The temperature of the dispersion rose to ca.45 "C during irradiation. Analysis Oxygen evolved was determined by gas chromatography using a molecular-sieve 5 A column and argon as a carrier gas. Ag+ concentration was determined by potentiometric titration with HC1 on a Metrohm Titroprocessor 636 and Dosimat E635. The specific surface area was measured by the B.E.T. method using a Micro-metrics surface-area analyser (model 2205) with argon as an adsorbate. X-Ray powder diffractions were taken on a Siemens D-500 X-ray diffractometer. Surface hydroxyl densities were measured by NaOH ion titration as described by Boehm." Results are summarized in table 1. Since rutile is the stable form of TiO,, there is no phase transformation induced by the thermal treatment. While the hydroxyl density decreases monotonically with increasing temperature, the specific area decreases in a stepwise fashion.Between 600 and 800 "C the surface area is about a factor of 4 lower than that of the starting material. It decreases to 8 m2 g-l at 1000 "C. mol dm-3).10 Usually 100-500 mg of TiO, was added to 10 cm3 of the NaOH solution in a tightly closed Pyrex glass bottle and the mixture was stirred for 15 h. The mixture was centrifuged and the supernatant liquid was titrated with HCl(aq). Only the acidic OH groups can be titrated by using NaOH in this concentration range. Therefore, the value obtained was divided by 2.4 to obtain the density of basic hydroxyl groups,ll and the two values were added. The surface OH density of the untreated sample obtained in this manner was 333 peq g-', which is close to reported values for similar TiO, samples.'l Surface OH density was measured by surface titration with NaOH(aq) (2 x Results Light-induced 0, Evolution with Ag+ as Electron Acceptor Fig. 1 shows the effect of TiO, concentration on the amount of 0, generated during the first 5 h of photolysis in the presence of 5 x lo-, mol dm-3 Ag+ electron acceptor.TheY. Oosawa and M. Gratzel 199 Fig. 1. Effect of TiO, concentration on the volume (all volumes refer to 1 atm and 25 "C) of oxygen generated during the first 5 h of illumination of a TiO, dispersion (20 cm3) containing 5 x IO-, mol dm-, Ag NO,, T = 50 "C: @, untreated TiO,; 0, TiO, heated at 1000 "C under Ar. 1.2 yr) E \ B n 0 O L I I I I 1 0.005 0.015 0.05 0.15 concentration/mol dm-3 Fig. 2. Effect of AgNO, concentration on the volume of 0, generated during the first 5 h of illumination of a TiO, dispersion (20 cm3) containing 100 mg TiO,, T = 50 "C: @, untreated TiO,; 0, TiO, heated at 1000 "C under Ar.rutile sample treated at 1000 "C shows a four-fold increase in the 0, volume upon increasing the amount of TiO, from 3 to 30 mg per 20 ~ m - ~ , while the untreated sample exhibits very little increase in the same concentration range. Presumably, the high- temperature treatment produces sintering of the particles, resulting in a lower dispersion. As a consequence, a larger amount of TiO, powder is required to absorb the incident200 Photocatalytic Oxygen Evolution L r n 0 * 0 10 20 30 40 50 Fig. 3. Time course of the 0, evolution during illumination of a TiO, dispersion (100 mg in 20 cm3) containing 5 x lo-, mol dm-3 Ag NO,, T = 50 "C; initial pH and calcination temperatures are indicated below in parentheses.The TiO, samples were calcinated in Ar. 0, TiO, (untreated, pH 2.5); e, TiO, (700 "C, pH 2.1); A, TiO, (900 "C, pH 1.8); A, TiO, (1000 "C, pH 1.6). time/h light. All the experiments described hereafter were carried out with 100 mg of TiO, in 20 an3 solution, i.e. an amount corresponding to the plateau region in fig. 1. Fig. 2 illustrates the effect of Ag+ concentration on the initial rate of photoinduced 0, generation. The Ag' concentration was varied between 5 x lo-, and 0.15 mol dm-,, and the volume of 0, produced during the first 5 h light exposure of the TiO, dispersion was determined. For untreated rutile the 0, evolution rate changed very little in the investigated concentration range.On the other hand, the rutile sample subjected to the heat treatment at 1000 "C under Ar showed a pronounced increase in V(0,) between 5 x lo-, and 1.5 x lo-, mol dmP3 Ag+. All further experiments were carried out at 5 x lo-, mol dm-, AgNO,, i.e. a concentration which is situated well in the plateau region of fig. 2. Fig. 3 illustrates the time course of photoinduced oxygen evolution in Ag+ containing suspensions of TiO, heated at different temperatures. The flattening of the curves is due to the decrease in the pH of the solution during the photoreaction. Fig. 4 shows the effect of calcination on the amount of 0, evolved during the first 5 h of illumination. The volume of 0, drastically increases for rutile powders treated at 700 "C and above. TiO, calcinated at 900 "C exhibits the highest activity.The volume of 0, obtained is here 10 times higher than that observed with the untreated powder. The fact that there is no difference in the activity of TiO, treated at 1000 "C in Ar and 0, shows that doping (partial removal of lattice oxygen, rendering the TiO, particles conducting) cannot explain the observed phenomena, because TiO, cannot be n-doped in a stream of oxygen of 1 atm.t (One referee pointed out that TiO, can be doped to have an oxygen excess at 1000 "C.) The molar ratio of silver ions reduced and oxygen evolved during the photoreaction was also found to be affected by the thermal treatment of the TiO, powder. Table 2 lists some representative results.The samples were analysed after 21 h of irradiation. Both the amount of Ag+ in solution and that of Ag deposited onto the TiO, were analysed. 1 atm = 101 325 Pa.Y. Oosawa and M . Gratzel 20 1 Y as 600 700 800 900 1000 1100 supplied treatment temperaturel'c Fig. 4. Effect of calcination temperature on the volume of 0, generated from a suspension of TiO, (Bayer, rutile) in water, pH 4.5, during the first 5 h of illumination. Calcination at 1000 "C was performed under Ar and 0,. Table 2. Analysis" of Ag+, Ag and O,* during photocatalytic oxygen generation in aqueous TiO, dispersions sample unreacted 0, Ag (metal) Ag+ Ag+Ag+ nAg/nO, I (TiO,, untreated)d 6.0 53 936 989 8.8 I1 (TiO,, untreated)d 9.2 61 923 984 6.6 I11 (TiO,, 1000 "C, Ar) 101 417 546 963 4.1 " Quantities are given in pmol.The analysis of photodeposited Ag metal comprised oxidation by Na,S,O, and subsequent determination of Ag+ in solution by potentiometric titration with HC1. In all the experiments the initial amount of AgNO, was 1000 pmol. The adsorption of Ag+ on the surface of TiO, was found to be negligibly small. For example, of the 1000 pmol Ag+ added initially to 20 cm3 of dispersion containing 100 mg of TiO,, only 3.9 pmol were found to be adsorbed after 15 h of stirring at room temperature. From table 2, the TiO, powder treated at 100 "C, i.e. the sample with low OH- density gives a molar ratio of nAg/nO, of 4.1 which is close to that expected from the stoichiometry of the photoreaction hv, TiO, Ag+ + 2H,O --+ 4Ag + 4H++ 0,.By contrast, for unheated TiO, powders which have a high hydroxyl density the molar ratio significantly exceeds 4. For the untreated TiO, powders one finds nAg/nO, = 8.8 (sample I) and 6.6 (sample 11). It can be excluded that the nAg/nO, 3 4 values observed with highly hydroxylated powders are due to the presence of carbonaceous impurities acting as hole scavengers. Elemental analysis showed the carbon content of the TiO,202 Photocatalytic Oxygen Evolution Table 3. Comparison of the photocatalytic activity of different TiO, powders as a function of treatment temperature TiO, crystal volume of 0, sample treatment structure evolved/cm3 a Bayer Fluka Fuji titanium Tp-2 Merck Lab. Degussa p-25 as supplied 1 OOO/ Ar as supplied 1000/Ar as supplied 1000/0, 1 oo/o, 1000/Ar as supplied 1000/0, 1000/Ar as supplied 600/Ar 800/Ar lOOO/Ar 1000/0, rutile rutile rutile anatase rutile rutile anatase rutile rutile anatase rutile rutile/anatase 0.4b rutile rutile rutile - 0.16 I .29 1.29 0.23 1.79 1.56 0.60 4.23 3.36 0.25 1.43 0.43 0.92 1.15 0.49 0.5 1 a After 5 h irradiation.Rutile to anatase ratio. employed to be negligible (0.01 wt%). Other impurities, such as C1 or S , were only present in trace amounts and were not affected by the high temperature treatment. Elemental analysis gave 34 ppm of C1 and 4 pprn of S in the TiO, samples as received. After treatment at 1000 "C, 26 ppm of C1 and 13 ppm of S were still present. Therefore it is concluded that water is the source of electrons which reduce Ag+ to Ag. Reaction (1) is associated with an increase in the proton concentration and it was examined to which extent this affects the oxygen evolution rate.The natural pH of a 5 x lo-, mol dm-, AgNO, solution is ca. 4.5. Irradiation for 1 h of 20 cm3 solution containing 100 mg TiO,, previously subjected to thermal treatment at 1000 "C, led to the formation of 580 mm3 (s.t.p.) of O,, the pH decreasing simultaneously to 2.2, in agreement with the stoichiometry of eqn (I). When the initial pH was adjusted with H,SO, to 2 or 1 the volume of 0, photogenerated during 1 h illumination decreased to 270 or 20mm3, respectively. Similar effects were observed with the untreated TiO, powder, although the effect of pH on the 0, evolution rate appeared to be less pronounced in this case. A similar pH dependence of the photocatalytic oxygen evolution from AgNO,/TiO, dispersions was observed by Kagiya et a1.l' It is ascribed to the increase in the gain in free energy of the reaction as the pH decreases.Enhanced activity for photocatalytic oxygen generation after calcination at high temperature was also observed with a variety of other TiO, samples. A comparison is made in table 3. In all cases illuminations were performed with dispersions containing 100 mg TiO, and 0.05 mol dmP3 AgNO, in 20 cm3 water. The volume of oxygen evolved after thermal treatment at 1000 "C was ca. 6-8 times larger than that obtained with the untreated samples. An exception is TiO,-P25, a mixture of anatase and rutile, the activity of which is optimal when heated at 800 "C and declines at higher pretreatment temperatures.Y. Oosawa and M .Gratzel 203 0.6 1 16 0 as 600 700 800 900 1000 1100 supplied treatment temperature/'C Fig. 5. Volume of 0, evolved during the first 5 h of illumination of a dispersion (20 cm') containing: 0, 5 x mol dm-3 K,PtCl, and 30 mg TiO, (Bayer, rutile); 0, 5 x lo-, mol dm-, KBrO, and 100 mg TiO, (Bayer, rutile), T = 50 "C. Photoinduced Oxygen Evolution from TiO, Dispersions containing PtCli- or BrO, as Electron Acceptor An enhancement of oxygen evolution activity after heat treatment of the TiO, rutile was found also with PtC1;- or BrO, as electron acceptors. Results are shown in fig. 5, where the volume of 0, generated during the first 5 h of photolysis of the TiO, dispersion is plotted as a function of pretreatment temperature.For temperature up to 900 and 1000 "C the photocatalytic activity of the TiO, increases in a similar fashion for the two acceptors. Between 900 and 1000 "C the BrO, acceptor system shows a decline in the oxygen output, while that obtained with PtClt- remains constant. BrO, is reduced by conduction band electrons to Br- ions: BrO, + 6e- + 6H+ -+ Br- + 3H,O E" = 1.44-0.059 pH The Br- ions are likely to impair oxygen evolution by competing with water for the scavenging of valence-band h01es.l~ It is possible that this inhibition is more effective on TiO, surface having a very low density of hydroxyl groups. Discussion The present investigation has illustrated that the calcination of TiO, has a remarkable influence on its activity for the photogeneration of oxygen in the presence of electron acceptors. The reaction was strongly enhanced by increasing the temperature of the heat treatment.Since the rutile crystal structure of the TiO, was not modified by the calcination, the effects probably originate from changes in the surface properties of the oxide. Table 1 shows that the parameter that is most significantly affected by the heat treatment is the density of hydroxyl groups. This is 470 peq g-l for untreated TiO, (Bayer rutile) powder, while it is 60 times smaller for the sample calcinated at 1000 "C under Ar.204 Photocatalytic Oxygen Evolution 6 10.6 P) E 2 2 > 0 I I n 0 1 2 3 4 5 (OH density)-'/ 1 O4 g per eq OH Fig. 6. Correlation between the oxygen generation rate (data from fig. 3-5) and the reciprocal density of OH groups at the TiO, surface (taken from table I): 0, AgNO,; 0, KBrO,; A, K,PtCl,.Further analysis showed that there is a linear relation between the reciprocal density of surface OH groups and the rate of oxygen generation. This correlation was obtained for all three acceptors investigated in the present study, (fig. 6). Thus it appears that at low surface OH coverage the photogeneration of oxygen is efficient and quantitative with respect to the stoichiometry of the photoreaction. Conversely, when the OH density is high the efficiency of oxygen generation is decreased and its quantity is less than stoichiometric. The oxygen deficiency is attributed to the formation of surface bound peroxides during the oxidation of water by valence-band holes : H,O + 2h+(TiO,) + O%(TiO,) + 2H+.The formation of peroxides was confirmed recently by photoelectrochemical in- vestigation~.~ Cyclic-voltammetric analysis of the products of the photo-oxidation of water on polycrystalline anatase films revealed that the titanium peroxo complexes formed differ from that produced by chemisorption of hydrogen peroxide at the surface of Ti0,.5a Similar peroxides are also produced during water photolysis with TiO, dispersions, l5 and they have been analysed colorimetrically using redox indicators. l6 The data in table 2 show that in the photodeposition of silver on calcined TiO,, the molar ratio of Ag to 0, is close to the stoichiometric value of 4. This implied that peroxides are not formed on TiO, surfaces with a very low hydroxyl density.Thus the OH groups play an important role in the formation of these peroxo species, presumably by providing surface states for the trapping of valence-band holes. l7 From fig. 6 it can be inferred that the surface OH groups exert an inhibitive effect on the rate of oxygen generation. The OH groups scavenge valence-band holes, leading to trapped hole states and subsequently to peroxo-titanium complexes. Both the trapped hole state and the surface-bound peroxide are recombination centres for charge carriers18 that decrease the efficiency of the photoreaction. This work was supported by grants from the Swiss National Energy Foundation (NEFF) and the Gas Research Institute, Chicago (subcontract SERI, Golden,Y. Oosawa and M. Gratzel 205 Colorado). We are grateful to Dr Panek, Bayer AG, for providing us with the rutile sample. References 1 Energy Resources through Photochemistry and Catalysis, ed. M.Gratzel (Academic Press, New York, 1983). 2 (a) S. Sat0 and J. M. White, Chem. Phys. Lett., 1980,72,83; (6) T. Kawai and T. Sakata, Chem. Phys. Lett., 1980,72,87; (c) E. Borganello, J. Kiwi, E. Pelizetti and M. Gratzel, J. Am. Chem. SOC., 1981,103, 6324; (d) E. Yesodharan and M. Gratzel, Helv. Chim. Acta, 1983,66,2145; (e) J. Kiwi and M. Gratzel, J . Phys. Chem., 1984, 88, 1302. 3 (a) H. van Damme and W. K. Hall, J. Am. Chem. SOC., 1979, 101,4373; (6) B. Parkinson, F. Decker, J. F. Juliao, M. Abramovich and H. C. Chagas, Electrochim. Acta, 1980, 25, 521 ; (c) P. Salvador and L. Gutierrez, Chem. Phys. Lett., 1982, 86, 131. 4 K. Kalyanasundaram, M. Gratzel and E. Pelizzetti, Coord. Chem. Rev., 1986, 69, 57. 5 (a) I . J. Ferrer, H. Muraki and P. Salvador, J. Phys. Chem., 1986, 80, 2805; (b) M. Ulmann, N. R. 6 D. Duonghong and M. Gratzel, J. Chem. SOC., Chem. Commun., 1984, 1547. 7 (a) A. M. Boonstra and C. A. H. A. Mutsaers, J. Phys. Chem., 1975, 79, 1694; (b) G. Munera, V. 8 H. P. Boehm and M. Herrmann, Z. Anorg. Allg. Chem., 1967, 352, 156. 9 Y. Oosawa and M. Gratzel, J. Chem. SOC., Chem. Commun., 1984, 1629. Tacconi and J. Augustinsky, J. Phys. Chem., 1986, 90, 523. Rives-Arnau and A. Sancedo, J. Chem. Soc., Faraday Trans. I , 1979, 75, 736. 10 H. P. Boehm, Discuss. Faraday SOC., 1971, 52, 264. I 1 J. A. Rob van Veen, F. T. G. Veltmaat and G. Jonkers, J. Chem. Soc., Chem. Commun., 1985, 1656. 12 S. Nishimoto, B. Ohtani, H. Kajiwara and T. Kagiya, J. Chem. Soc., Faraday Trans. 1, 1983, 79, 13 Y. Oosawa, Chem. Lett., 1982, 424. 14 J-M. Herrmann and P. Pichat, J. Chem. SOC., Faraday Trans. I , 1980, 76, 1138. 15 B. Gu, J. Kiwi and M. Gratzel, Nouv. J. Chim., 1985, 9, 539. 16 J. Kiwi and M. Gratzel, J. Mol. Catal., 1987, 39, 63. 17 G. Rothenberger, J. Moser, M. Gratzel, N. Serpone and D. K. Sharrna, J. Am. Chem. SOC., 1985, 107, 18 G. T. Brown and J. R. Darwent, J. Phys. Chem., 1984, 88, 4955. 2685. 8054. Paper 7/2 18 ; Received 6th February, 1987

ISSN:0300-9599    

DOI:10.1039/F19888400197

出版商:RSC    

年代:1988

数据来源: RSC