摘要:

J . Chem. SOC., Faraday Trans. 1, 1988, 84(7), 2465-2475 Fourier-transform Infrared Investigation of Intermediates in the Oxidation of Toluene on V,O,/TiO, Hisashi Miyata," Tadashi Mukai and Takenhiko Ono Department of Applied Chemistry, University of Osaka Prefecture, Sakai, Osaka 591, Japan Takashi Ohno" and Fumikazu Hatayama School of Allied Medical Sciences, Kobe University, Suma, Kobe 654-01, Japan The adsorption and oxidation of toluene on V-Ti oxide prepared by the gas-phase method (2.1 to 5.6 wt %) has been studied by F.t.i.r. spectroscopy as well as the microcatalytic method. It is found that toluene is bonded as a n-complex to the vanadium ion. The CH, groups of the n-complex are dehydrogenated by the V=O species to form benzyl species at room temperature. The formation of aldehyde-like intermediate before the oxidation of benzyl species to benzoate has been confirmed.The reactivity of surface V=O species of V-Ti oxide prepared by a gas-phase method is greater than that of the impregnation catalyst. Vanadium oxide catalysts in combination with various promoters are widely used for the selective oxidation of hydrocarbons. V-Ti oxide catalysts are especially well known for the practical use of their high selectivity and activity. The acid-base properties of V-Ti oxide and the adsorption of various basic and acidic molecules have been In addition, a number of workers have studied the oxidation of hydrocarbons on V-Ti oxide using i.r. spectroscopy. However, few i.r studies of surface structures formed in the oxidation of aromatic hydrocarbons have appeared in the literat~re.~, lo More recently van Hengstum et all1 have reported the oxidation mechanism of toluene and o-xylene on V-Ti oxide catalysts.However, our understanding of the surface-adsorbed species in the oxidation reaction is far from complete. In an earlier paper12 we reported that on catalysts prepared by the gas-phase method a new phase of vanadia is formed and that the character of the surface vanadia varies with the preparation method. In the present work we therefore employed F.t.i.r. spectroscopy in the study concerning the adsorption and the intermediates in the oxidation of toluene over V-Ti oxide catalysts prepared by the gas-phase method. Experimental The titania (Japan Aerosil, P25) was used to support vanadium oxide.Two preparation methods were used. The first was a gas-phase preparation method (GVTi-2.1-GVTi-5.6, 2.1-5.6 wt O/O vanadium as V20,). The second was standard wet impregnation of the support with an oxalic acid solution of ammonium metavanadate (VTi-5.0, 5.0 wt YO vanadium). Details of the methods of preparation of catalysts have been described previously.12 The physical parameters of those catalysts are listed in table 1. Toluene, benzaldehyde and benzene were purified by freeze-pumpthaw cycles and vacuum distillations prior to use. [2H,]Toluene (Merck Co., 99.7 YO) and benzoic acid were used without further purification. Two types of apparatus were used. The first was a conventional closed-circulation system equipped with an i.r. cell in the circulation loop.The second was a pulse 24652466 F. T.I.R. of Intermediates in Toluene Oxidation Table 1. Physical properties of vanadium oxide catalysts preparation V,O, surface catalyst methoda (wt %) area/m2 g-' GVTi-2.1 gas (1) 2.1 50 GVTi-3.3 gas (2) 3.3 50 GVTi-4.0 gas ( 1 ) 4.0 51 GVTi- 5.6 gas (8) 5.6 30 VTi- 5 .O wet 5.0 42 a Gas, gas-phase preparation ; wet, wet impregnation method. Parentheses show repeat time of VOCl, circulation. microreactor system directly connected with a gas chromatograph for analysis of the reaction products. The i.r. cell was a quartz tube of 2.5 cm diameter and 40 cm in length. The KRS-5 windows were sealed to the cell with wax. A disc of V-Ti oxide (20 mm diameter, ca. 100 mg) was placed in the i.r. cell and the disc temperature was slowly increased from room temperature to 673 K under evacuation and kept at that temperature for several hours.Then, heating under circulation of oxygen (ca. 4 kPa) at 673 K for 2 h, followed by degassing at the same temperature for 2 h, was repeated several times. Finally the disc was cooled to room temperature in oxygen. The amount of each reactant adsorbed was ca. 0.7 cm3 g-l. Details of the apparatus and procedures have been described previously. 13, l4 F.t.i.r. spectra were recorded at ambient temperature before and after each treatment of the catalyst disc, such as heating or admission on reactant, with the use of a Shimadzu FTIR-4000 spectrometer operating at 2 cm-l resolution and in single-beam mode. This prevents a temperature rise of the catalyst disc.After 100 accumulations had been stored the data were transferred on an IEEE-488 bus line to a master computer (PC-9801 VX2, NEC). In order to obtain quantitative information on the spectral behaviour, some of the spectra were subtracted and deconvoluted by using a data-acquisition system. The details of the data acquisition and analysis system have been described previously.l29 15, l6 For a pulse microreactor study, the dried catalysts were tested in a fixed-bed reactor. The catalyst charge of 0.1-0.3 g was preheated with flowing He for 1 h at 673 K, followed by flowing He for 1 h at reaction temperature (423-673 K). Following pretreatment, a gaseous mixture of tolyene, oxygen and helium (2.1 : 6.5 : 91.4 vol YO) was fed over the catalyst. Gaseous reaction products were measured by gas chromatography with a column of MS-13X or GCPAC-54.Results Microcat alytic Results Two catalysts (GVTi-4.0 and VTi-5.0) were tested for toluene oxidation in the temperature range 423-673 K. A fresh sample was used for each reaction temperature. Fig. 1 shows the amount of products. The reaction products vary with the reaction temperatures and with preparation methods of catalysts. GVTi-4.0 shows the highest activity to benzaldehyde at 473 K. The selectivity to benzaldehyde at 473 K was 36%. Above 573 K total oxidation takes place significantly. In contrast, VTi-5.0 shows a marked increase in the formation of benzaldehyde at 573 K. On both catalysts above 573 K a small amount of benzene is formed. The results suggest that the GVTi catalysts abstract hydrogen from toluene more readily than VTi catalysts.H.Miyata et al. 2467 400 500 600 700 TIK Fig. 1. Products in toluene oxidation. Open symbols, GVTi-4.0; closed symbols, VTi-5.0. 0, Benzaldehyde ; 0, benzene ; A, CO + CO,. 40 30 17 15 13 11 wavenumber/ 1 O2 cm-’ Fig. 2. 1.r. spectra of toluene adsorbed on GVTi-5.6. After introduction of toluene followed by 1 h evacuation at room temperature, (a) GVTi-5.6, (b) GVTi-2.1, (c) TiO,. In fig. 2-7, the spectra shown are after subtraction of catalyst itself.2468 F. T.I.R. of Intermediates in Toluene Oxidation Adsorption of Toluene and [*H,jToluene Fig. 2 shows the spectra of toluene adsorbed on GVTi catalysts together with titania after subtraction of the catalyst spectrum. 100 accumulations using single-beam mode gave spectra with good signal-to-noise ratios and made it possible to subtract the spectra.Exposing a GVTi sample to toluene led to an increase of the band of the surface hydroxyl group, the absorption near 3070 cm-' in the C-H stretching region and the bands in the bending modes. Degassing at 298 K for 1 h caused the absorption in the v(C-H) region to decrease [fig. 2(a)], the residual bands being near 3070 cm-l with the bands in the bending region. The bands in the region 1700-1200 cm-l are assigned to v(C=C) of the skeletal ring or d(C-H) of the CH, groups." In the case of GVTi-2.1 [fig. 2(6)], only weak bands remained at almost the same positions as those observed on GVTi-5.6, while on TiO, toluene was easily removed by evacuation at room temperature for 30 min [fig.2(c)]. The results suggest that the surface vanadia provides the active sites for toluene adsorption, as will be discussed below. The relatively large increase of surface OH band on GVTi samples shows that there was a reaction of the adsorbate with the GVTi samples, leading to OH group formation. Similar experiments were carried out using [2H,]toluene (fig. 3). Bands near 2350 cm-l are due to carbon dioxide molecules in air owing to the measurements being made in single-beam mode. In the region 1700-1200 cm-l only skeletal ring vibrations are expected in the spectra because of their isotopic shifts for d(C-D) bands." An increase of the period of adsorption caused the absorption due to v(C-D) of the methyl group to decrease, while the v(0-D) band together with the bands of the skeletal ring vibration were intensified [fig.3(b) and (c)]. In addition, the v(C-D) band in the benzene ring remained unchanged. On evacuation at room temperature for 1 h the bands due to CD stretching together with a broad OH band disappeared, while simultaneously a new OD band appeared at 2570 cm-' [fig. 3(d)]. This suggests that surface OD groups were formed, in agreement with results for [lH,] toluene, and that the methyl group of toluene was dehydrogenated by surface oxygen to form the hydroxyl group and the dissociated adsorbed species of toluene on the surface. A typical ring vibration around 1600 cm-l was observed in all spectra. This strongly suggests that during hydrocarbon adsorption the aromatic nucleus remains unaffected.The i.r. bands of toluene adsorbed on GVTi catalysts together with those for liquid toluene are summarized in table 2. Interaction of Oxygen with Adsorbed Toluene Fig. 4 shows the spectra of toluene adsorbed on GVTi-5.6 at higher temperatures. The temperature of the disc containing toluene was raised in stages under vacuum. Note that in the presence of gaseous oxygen (ca. 2.7 kPa) almost the same spectral behaviour as shown in fig. 4 was observed, although the spectral changes occurred at slightly lower temperatures. At 373 K new bands at 1630 and 1567 cm-' appeared and were intensified at 423 K, while the bands at 1590 and 1360 cm-l reduced in intensity. Increasing the temperature of the disc to 493 K caused appearance of the bands at 1520 and 1400 cm-l and disappearance of the bands at 1630 and 1567 cm-l.In the case of GVTi-2.1 (fig. 5) the spectral changes with increasing the temperature of the catalyst are essentially the same as those observed on GVTi-5.6. In order to identify the intermediate species formed at higher temperatures, the spectral behaviour of some compounds adsorbed on catalysts was studied. Fig. 6 shows the spectra of benzaldehyde adsorbed on GVTi-5.6. Adsorption of benzaldehyde did not give the typical aldehyde CH absorption band around 2850 cm-' or band due to v(C=O) of the free aldehyde around 1710 cm-l. This is in agreement with the observations of previous After adsorption of the aldehyde three intense bands were observed around 1630, 1600 and 1580 cm-l [fig. 6(a)]. At 373 K the bandsH.Miyata et al. 2469 40 30 17 15 13 1 1 wavenumber/ 1 O2 cm-' Fig. 3. 1.r. spectra of [2H,]toluene adsorbed on GVTi-5.6. (a) Immediately after introduction of [2H,]toluene at room temperature; (b) followed by 30 min evacuation; ( c ) followed by 2 h evacuation; ( d ) followed by 1 h evacuation. Table 2. Wavenumbers (Flcrn-') of the bands of toluene adsorbed on V-Ti oxide liquid vibrational toluene [2H,]toluene toluene modea 3450 3070 - - - 1590 1500 1485 1450 1428 1360 1295 1190 2470 2278 2240 2212 2120 1548 1400 1378 - - - 1280 - - 3070 2910 2860 1607 1496 1460 1378 1310 1212 - - - v(0-H) or v(0-D) v(C-H) or v(C-D) ~as(CH3)or vas(CD3) Vs(CH3) or V,(CD3) - C-C stretching C-C stretching C-C stretching 6a,(CH3) 6,(CH,) - C-C-stretching 6(-H) in-plane a In ref. (17).at 1630 and 1580 cm-I reduced in intensity, while simultaneously new bands around 1540 and 1400 cm-' were intensified. The latter bands became stronger at 473 K. The spectra at 373 and 473 K [fig. 6(b), (c)] resemble those of toluene on GVTi-5.6 at 423 and 493 K, respectively. The corresponding bands with GVTi-2.1 [fig. 7(a)] appeared at slightly higher wavenumber. On titania, the spectrum of adsorbed benzaldehyde exhibits bands in higher positions than in the case of GVTi catalysts [fig. 7 (b)]. On these catalysts, weak or shoulder bands around I680 cm-' were observed. Judging from the wavenumber x i FAR I2470 F. T.I.R. of Intermediates in Toluene Oxidation 10.1 1 I 1 I I I I I I I I wavenumber/ lo2 cm-' Fig. 4. 1.r. spectra of toluene adsorbed on GVTi-5.6. (a) After adsorption of toluene [(a) in fig.21 followed by I h evacuation at 373 K; (b) followed by I h evacuation at 423 K; (c) followed by 1 h evacuation at 493 K. 40 30 17 15 13 1 1 I 1 1 I I I I I I I I I 40 30 17 15 13 1 1 wavenumberj lo2 cm-' Fig. 5. 1.r. spectra of toluene adsorbed on GVTi-2.1 (a) After adsorption of toluene [(b) in fig. 21 followed by 1 h evacuation at 373 K; (b) followed by I h evacuation at 423 K; (c) followed by 1 h evacuation at 493 K. shift of 20-30 em-' these are attributable to weakly adsorbed benzaldehyde species. van Hengstum et al. l1 reported similar assignments. On adsorption of benzaldehyde, the increase in the surface OH band is very small compared with the adsorption of toluene on GVTi samples.This indicates that during the adsorption of benzaldehyde at low temperature no dissociative adsorbed species are formed. Band-separation techniquePH. Miyata et al. 247 1 40 30 17 15 13 11 wavenumber/ 1 O2 cm-' Fig. 6. 1.r. spectra of benzaldehyde on GVTi-5.6 after adsorption of benzaldehyde; (a) followed by I h evacuation at 293 K; (b) followed by 1 h evacuation at 373 K; ( c ) followed by 1 h evacuation at 473 K. were applied to the spectra of benzaldehyde adsorbed on GVTi catalysts together with titania in the range 1800-1 500 cm-l. The resulting curves are shown in fig. 8. In the case of GVTi-2.1, the band at 1650 cm-l separated into two peaks at 1650 and 1630 cm-l. GVTi-5.6 and titania exhibit only a single band at 1630 and 1650 cm-l, respectively. Considering that GVTi-2.1 has vanadia as well as titania on its surface, it is concluded that the band at 1630 cm-l is due to v(C=O) of the species interacting with vanadia and that at 1650 is due to v(C=O) interacting with titania.The F.t.i.r. spectrum of benzoic acid adsorbed on GVTi-5.6 is also shown in fig. 7. The bands resemble those of toluene adsorbed on GVTi-5.6 at 493 K. Discussion Adsorption Sites of Toluene on GVTi at Low Temperature As mentioned above, a band due to hydrogen-bonded OH (ca. 3450 cm-l) appeared on all the catalysts, irrespective of the vanadium content. Considering that GVTi-5.6 itself showed only a weak band due to hydroxyl groups, adsorbed toluene interacts with the surface oxygen to form' surface hydroxyl groups. P r e v i o ~ s l y ~ , ~ ~ we reported that vanadium ions act as Lewis-acid centres and the 0'- ions in V=O as basic centres.Since carrier TiO, exhibits no strong basicity,'O the occurrence of dehydrogenation on GVTi is attributable to hydrogen abstraction by the 02- ions in the V=O species acting as a base. Thus surface OH species and benzyl species are formed. Similar surface reactions have been reported by Chang and 81-22472 I;. T.I.R. of Intermediates in Toluene Oxidation I I I I I 1 I I I I I 40 30 17 15 13 1 1 wavenumber/ 1 O2 cm-' Fig. 7. 1.r. spectra of benzaldehyde on GVTi-2.1 and TiO, after adsorption of benzaldehyde (a) on GVTi-2.1 followed by 1 h evacuation at 373 K; (b) on TiO, followed by 1 h evacuation at 373 K; ( c ) after adsorption of benzoic acid on GVTi-5.6 followed by 1 h evacuation at 393 K.17 16 15 17 16 15 wavenumber/ 10' cm-' Fig. 8. 1.r. spectra of benzaldehyde on GVTi catalysts and separated peaks. (a) GVTi-5.6, (b) GVTi-2.1, (c) TiO,. (1) Original, (2) separated peaks.H. Miyata et al. 2473 0 0 li II v-0-v-0 -v / / / F1/0 ?P ?? v-0-v-0-v Fig. 9. Reaction scheme. KokesZ1 for toluene on ZnO. They also suggest that the benzyl species on ZnO is adsorbed with its benzene ring parallel to the surface because of the lower wavenumber shift of the aromatic ring during the dissociation. A similar n-complex was also reported in the toluene/montmorillonite system.22 More recently van Hengstum et aZ.ll suggested a similar reaction in the toluene/V-Ti oxide system, although they did not obtain direct evidence. As mentioned above, the bands due to the vibrations of the aromatic ring (1590 cm-l) in toluene adsorbed on GVTi catalysts showed a lower wavenumber shift compared with liquid toluene (table 2).17 In addition, the ratios of the intensities of ring vibration of toluene adsorbed on GVTi catalysts together with titania are different from those in the liquid phase.This strongly indicates that a n-complex is formed on GVTi catalysts. A similar situation is also expected for the adsorption of benzene on GVTi and on titania. Accordingly, a possible mechanism of surface reactions at lower temperature is proposed in fig. 9. The first step is n-interaction with a Lewis-acid site followed abstraction of a hydrogen atom by a basic site. If there are free surface hydroxyl groups adjacent to the site containing the n-complex, a hydrogen atom in a methyl group interacts with an oxygen atom in a hydroxyl group to form the hydrogen-bonded species.Mechanism of Toluene Oxidation on GVTi As shown above, the bands in the bending region change in intensity at higher temperatures. Fig. 10 shows the changes in the band intensities on increasing the temperature of GVTi. This indicates that there are at least two types of adsorbed species originating from adsorbed toluene at higher temperatures. The species which show maximum intensities at ca. 400 K are assigned to aldehyde-like intermediates, as reported2474 F. T.I.R. of Intermediates in Toluene Oxidation 300 400 500 TIK Fig. 10. Change in the intensities of the bands with increasing temperature. (0) 1590, (@) 1360, (A) 1630, (A) 1567, (0) 1520, (m) 1400cm-'.by van Hengstum et a/." The species formed above 400 K seem to have a carboxylate structure, although some uncertainty in the positions and the intensities remains because of their severe overlapping and broadening. Thus it is confirmed that the benzoate species are formed via benzyl species, as reported by Niwa et al.9 and by van Hengstum et aZ.ll From the above considerations the oxidation mechanism shown in fig. 9 is proposed. Niwa et al.9 proposed that surface benzoate species are formed on the alumina surface. In the present study such species were also observed on the titania surface in the oxidation of benzaldehyde. However, the symmetric vibration of a benzoate species was observed at a higher position (1425 cm-l).In addition, considering that on both GVTi-5.6 and GVTi-2.1 the bands due benzoate species appeared at almost the same position [fig. 4(c), fig. 5(c)], such a situation is excluded. In the mechanism proposed by van Hengstum et a1.l' two types of benzoate complex, monodentate and bidentate, are formed. Although we did not obtain direct evidence for such species, the GVTi-5.6 catalyst exhibits four types of surface V=O species on the surface, as reported previously using F.t.i.r. study.12 Such species may also contribute to the formation of the different benzoate species. ' As mentioned above, the GVTi catalyst shows a higher activity for benzaldehyde formation (fig. 1). This can be explained as follows : GVTi-4.0 has a single V=O band, at 985 cm-l, while VTi-5.0 exhibits a V=O band (1021 cm-l) due to crystalline vanadia.12 As regards the reactivity of surface V=O species of surface vanadia, GVTi-4.0 is more active than VTi-5.0, as proposed by Jonson et aZ.23 for toluene oxidation. Thus the results in fig.1 are expected. In this paper we concentrated on the oxidation reaction at low temperature or the initial step in the oxidation, although at higher temperature appreciable total oxidation takes place. Further work is necessary to discuss this problem. References I M. Takagi, T. Kawai, M. Soma, T. Onishi and K. Tamaru, J . Cataf., 1977, 50, 441. 2 Yu. Sh. Goldberg, I. G. Iovel and M. V. Shimanskaya, React. Kinet. Cataf. Lett., 1978, 8, 327. 3 Yu. V. Belokopytov, K. M. Kholyavenko and S. V. Gerei, J .Catal., 1979, 60, 1 . 4 M. Takagi-Kawai, M. Soma, T. Onishi and K. Tamaru, Can. J . Chem., 1980, 58, 2132.H. Miyata et al. 2475 5 Y. Murakami, M. Inomata, A. Miyamoto and K. Mori, Proc. 7th Int. Congr. Catalysis (Kodansha/ 6 A. Anderson, J. Catal., 1982, 76, 144. 7 H. Miyata, Y. Nakagawa, T. Ono and Y. Kubokawa, J . Chem. SOC., Faraday Trans. I , 1983, 79, 8 H. Miyata, Y. Nakagawa, T. Ono and Y. Kubokawa, Chem. Lett., 1983, 1141. 9 M. Niwa, H. Ando and Y. Murakami, J . Catal., 1977, 49, 92. Catal. (Akademiai Kiado, Budapest, 1968), vol. I , p. 454. .I. Catal., 1986, 101, 323. I , 1987, 83, 675. Elsevier, Tokyo/Amsterdam, 1981), p. 1344. 2343. 10 W. M. H. Sachtler, G. J. H. Dorgelo, J. Fahrenfort and R. J. H. Voorhoeve, Proc. 4th Int. Congr. 1 1 A. J. van Hengstum, J. Pranger, S . M. van Hengstum-Nuhuis, J. G. van Ommen and P. J. Gellings, 12 H. Miyata, K. Fujii, T. Ono, Y. Kubokawa, T. Ohno and F. Hatayami, J . Chem. Soc., Faraday Trans. 13 H. Miyata, K. Hata and Y. Kubokawa, J. Catal., 1977, 49, 8. 14 T. Nakajima, T. Sonoda, H. Miyata and Y. Kubokawa, J . Chem. SOC., Faraday Trans. I , 1982, 78, 15 H. Miyata, K. Fujii, S. Inui and Y. Kubokawa, Appl. Spectrosc., 1986, 40, 1177. 16 H. Miyata, Y. Nakagawa, S . Miyagawa and Y. Kubokawa, J . Chem. SOC., Farada-v Trans. I , 1988, 17 N. Fuson, C. Garrigou-Lagrange and M. L. Josein, Spectrochim. Acta, 1960, 16, 106. 18 S. Pinchas and I. Laulicht, Infrared Spectra of Labelled Compounds (Academic Press, London, 19 Y. Nakagawa, T. Ono, H. Miyata and Y. Kubokawa, J . Chem. Soc., Faraday Trans. I , 1983,79,2929. 20 T. Nakajima, H. Miyata and Y. Kubokawa, J. Chem. Soc., Faraday Trans. I , 1983, 79, 2559. 21 C. C. Chang and R. J. Kokes, J . Catal., 1975, 38, 491. 22 T. J. Pinnavaia and M. M. Mortland, J . Phys. Chem., 1971, 75, 3957. 23 B. Jonson, R. Larsson and B. Rebenstorf, J . Catal., 1986, 102, 29. 555. 84, 2129. 197 1). Paper 711782; Received 5th October, 1987

ISSN:0300-9599    

DOI:10.1039/F19888402465

出版商:RSC    

年代:1988

数据来源: RSC

摘要:

J . Chem. SOC., Faraday Trans. I, 1988, 84(7), 2477-2482 Cr3+ Electron Spin Resonance Linewidth in ZnO-ZnCr,O,-(Pd) Solid Mixtures Lucio Forni" and Cesare Oliva Dipartimento di Chimica Fisica ed Elettrochimica and Centro CNR, Universita di Milano, Via C. Golgi, 19 I-20133 Milano, Italy The e.s.r. signal generated by a solid catalyst, based on a ZnO, ZnCr,O, and Pd mixture, has been studied as a function of the concentration of the components. The linewidth broadening and narrowing of the signal is strongly dependent on both the Zn/Cr ratio and the Pd concentration and is connected with a change in the relative distance between the Cr3+ ions. One of the best catalysts for the synthesis of pyrazines is a mixture of ZnO and ZnCr,O, to which a few wt% of Pd may be added as a promoter.' During our exploratory research, it has been observed that the selectivity to the desired product strongly depends on both the Zn/Cr ratio and the Pd concentration in the finished catalyst.Only two phases are detected in the Debye X-ray diffraction patterns, relative to both unpalladiated and palladiated catalysts : the spinel ZnCr,O, and ZnO (zincite). Furthermore, such an analysis showed that activation of the catalyst, carried out after a preliminary calcination in air, by careful reduction in dilute hydrogen and under controlled conditions, is always accompanied by a considerable decrease in crystallinity, together with the formation of an enlarged-cell spinel-like ZnCr,O, phase. Thus it was expected that an examination of the e.s.r. signal generated by the solid and connected with the presence of Cr3+ ions could afford useful information on the characteristics of the catalyst.The present work deals with the analysis of the dependence of this e.s.r. signal on the composition of the catalyst, in terms of the Zn/Cr ratio and Pd concentration. Experiment a1 Catalyst samples have been prepared either by coprecipitation, by adding a solution of NaHCO, to a solution of Zn2+ and Cr3+, or by wet-mixing of a mixture of ZnO and CrO,,, followed by drying (373 K), calcination in air (703 K) and reduction in H, (593 K)., To this solid, when desired, a solution containing the appropriate amount of PdSO, was added, followed by further calcination and reduction. The composition of the samples examined is shown in table 1.Results and Discussion In the present case all the samples gave a broad, symmetric line, close to Lorentzian in shape and centred at g = 1.98, characteristic of magnetically interacting /3-Cr3+ ions., The analysis of these spectra is not straightforward, owing to the considerable number of species and phenomena involved. Between 133 and 3 13 K a strong temperature dependence of the linewidth was always observed with our solids (see e.g. fig. 1). The observed trend is analogous to that reported by others4y5 for Cr3+ samples reduced in hydrogen. However, a large difference is seen between the range of linewidth change reported in ref. (4) and that observed by us. The 24772478 E.S. R. Linewidth in ZnO-ZnCr,O,-(Pd) Mixtures Table 1. Composition of the samples Zn/Cr atomic sample ratio Pd (wt %) A1 A2 A3 A4 Bl B2 B3 B4 c1 c 2 c 3 c 4 c 5 C6 c 7 1.55 2.1 1 3.32 4.36 1.55 2.1 1 3.32 4.36 2.49 2.49 2.49 2.49 2.49 2.49 2.49 - - - - 0.82 0.83 0.73 0.56 0.24 0.78 1.01 1.51 1.94 3.14 - 20‘ 1 I 100 200 300 TlK Fig. 1.Typical change in e.s.r. linewidth as a function of temperature; Sample B3, table 1. former usually ranges between ca. 200 and 80 mT, the latter (see fig. 1) between 36 and 25 mT, within the same temperature range. On the other hand, the same trend (but with a linewidth < 30 mT) has been reported6 for antiferromagnetic ZnCr,O,, having a normal spinel structure, at temperatures ranging from 100 to 300 K, in qualitative agreement with the theory of Huber’ and Maleev8 for antiferromagnetic systems above their Nee1 temperature.Recently a similar trend has been notedg for AgCrS,, another compound whose properties are strongly affected by the Cr-Cr distance. For instance, if S, is substituted by Se,, a completely different type of (ferromagnetic) behaviour is seen. In our case we observed that a small variation in the geometry strongly affects the properties of the system. Cr3+ ions seem to be clustered in such a way that there is sufficient spin exchange coupling to cause exchange-narrowing of the resonance and a partial ‘washing out’ of the line-broadening which is due to the dipolar interaction.100 b E 1 a $50 L. Forni and C. Oliva I I I I I 2479 0 1 2 3 4 5 6 Zn/Cr atomic ratio Fig. 2. Change in e.s.r. linewidth at a constant temperature (295 K) as a function of the Zn/Cr atomic ratio: 0 , without Pd; ., with Pd (ca.1 wt%). When these two competitive effects are active, the width of the absorption line is given by'O Pp is the mean square magnetic field produced by a dipolar interaction between electron spins, and He is the average exchange field produced by the exchange interaction. As a first approximation, PP is proportional to the number, Z, of Cr3+ nearest-neighbours to each Cr3+ ion and He is proportional to the exchange integral, J , between Cr3+ ions. (2) Hence, eqn (1) leads to the relation 6H oc Z I J . Presumably Z does not change significantly with temperature. On the other hand, a change in the magnetic concentration of Cr3+ ions is reflected in the average value of Z.'l Kittel and Abrahams12 proposed that the linewidth for random occupancy of lattice sites by magnetic centres is proportional to the square root of the concentration if more than ca.10% of the lattice sites are filled magnetically, while, for lower concentrations the width should be proportional to the concentration. This result was extended to systems containing mixed-valence series13 and holds if we assume no variation in the lattice parameter and a molecular-field treatment which relates J and He. However, a variation in the exchange field, resulting from a reorganisation in structure, cannot be always ruled out, and the trend in the variation of linewidth with magnetic concentration can sometimes be reversed merely by changing the diluting diamagnetic ion.l4 Furthermore, when the linewidth is strongly exchange-narrowed, it is safe to predict that the first effect of dilution will not be a decrease, but instead an increase in the linewidth, since spin-exchange narrowing effects have a shorter range than the dipolar interaction, and the paramagnetic ions are separated on di1uti0n.l~ Indeed, this has been observed in our case, as shown in fig.2, where the linewidth is plotted us. the Zn/Cr atomic ratio. When a stronger spin exchange linewidth narrowing is observed at high Cr3+ concentration (samples with 1 wt% Pd) the first effect of dilution is an increase in the e.s.r. linewidth. On the other hand, when a smaller exchange-narrowing effect is observed at high Cr3+ concentration (samples without Pd), line narrowing is the first effect of dilution. 6H = Hi/He (1)2480 E.S.R. Linewidth in ZnO-ZnCr,O,-(Pd) Mixtures Fig. 3. Change of em-. signal (295 K) as a function of Zn/Cr atomic ratio. (A) Samples of the A- series with gain: (a) 5000, (b) 5000, (c) 500 and (d) 5000. (B) Samples of the B-series, with gain: (a) 5000, (b) 250, (c) 250 and (d) 500 (see table 1). The extent of the effect can also be appreciated by a direct inspection of the spectra (fig. 3). It may be seen that 1 wt % Pd ‘pushes’ the Cr3+ ions close together, yielding a stronger spin-exchange effect. The dilution of such ions, afforded by increasing the Zn/ Cr ratio, can compensate this effect, and the two graphs (fig. 2) intersect at Zn/Cr z 2.5. A further analysis of the effect of Pd was carried out by examining the trend in e.s.r. linewidth when varying the Pd concentration at a fixed Zn/Cr ratio (ca. 2.5) (fig.4 and 5). A spin exchange-narrowing effect is observed up to ca. 1 wt YO Pd, in agreement with the result shown in fig. 2. Note that no difference is observed between the unpalladiated sample and samples containing up to 1 % Pd, since the Zn/Cr ratio corresponds to the crossing point of the two diagrams of fig. 2, relative to palladiated and unpalladiated catalysts, respectively. A strong linewidth broadening follows, for concentration up to ca. 2 wt%, probably indicating that Cr3+ ions have been moved, so that the spin exchange become less effective. A further linewidth narrowing is finally observed at higher Pd concentrations, when the dipolar broadening also becomes less effective. A change in the crystal structure of the solid is also produced by a change in the Zn/ Cr ratio.It is known5’l6 that an increase in the Zn/Cr ratio leads to an increase in the cryst$ lattice parameter of the ospinel phase (ZnCr,O,) present in these solids,o from 8.36 A (Zn/Cr = 0.5/1) to 8.41 A (Zn/Cr = 3/1), followed by a decrease to 8.39 A with a further increase in the ratio to Zn/Cr = 5.7. This trend also contributes to the behaviour observed in the linewidth of the e.s.r. signal for the unpalladiated samples.L. Forni and C. Oliva 248 1 1 DPPH Fig. 4. Change of e.s.r. signal (295 K) as a function of Pd concentration; C-series, table 1, with gain (a) 320, (b) 250, (c) 250, (d) 500, (e) 5000, ( f ) 25000 and (g) 250. Pd (wt %) Fig. 5. E.s.r. linewidth us. Pd concentration.Data as for fig. 4.2482 E.S.R. Linewidth in ZnO-ZnCr,O,-(Pd) Mixtures Finally, the addition of Pd, entailing the abovementioned significant reduction in the degree of crystallinity of the ZnCr,O, spinel phase, may afford a further contribution to the change in the distances between the Cr3+ ions, and so to the reported variation of the e.s.r. spectrum. We are indebted to Bracco Industria Chimica, Milano, for a research contract by means of which this work was made possible. References 1 L. Forni, G. Stern and M. Gatti, Appl. Catal., 1987, 29, 161. 2 A. V. Stiles, Catalyst Manufacture (Dekker, New York, 1983), p. 126. 3 R. J. J. Williams and R. E. Cunningham, Ind. Eng. Chem., Prod. Res. Dev., 1974, 13, 49. 4 D. E. O’Reilly and D. S. MacIver, J .Phys. Chem., 1962, 66, 276. 5 M. Bertoldi, G. Busca, B. Fubini, E. Giamello and A. Vaccari, Proc. VIZtalian-Czechoslovak Symp. on Catalysis, Sanremo, Sept. 1987 (Italian Chemical Society, Rome, 1987), p. 156; M. Bertoldi, G. Busca, B. Fubini, E. Giamello, F. Trifiro and A. Vaccari, J . Chem. SOC., Faraday Trans. I , in press. 6 M. Baran, S. Piechota and A. Pajaczkowska, Acta Phys. Polon., Part A , 1981, 59, 47. 7 D. L. Huber, J . Phys. Chem. Solids, 1971, 32, 2145. 8 S. V. Maleev, Phys. Lett., Part A , 1974, 47, 11 1. 9 R. E. Benfield, P. P. Edwards and A. M. Stacy, J . Chem. Soc., Faraday Trans. I , 1987, 83, 3527. 10 P. W. Anderson and P. R. Weiss, Rev. Mod. Phys., 1953, 25, 869. 11 R. P. Gupta, M. S. Seehra and W. E. Vehse, Phys. Rev. B, 1972, 5, 92. 12 C. Kittel and E. Abrahams, Phys. Rev., 1953, 90, 238. 13 R. Scaringe and G. Kokoszka, J . Chem. Phys., 1974, 60, 40. 14 A. A. Samokhvalov, V. S. Babushkin, V. G. Bamburov and N. I. Lobachevskaya, Sou. Phys. Solid 15 P. B. Ayscough, ESR in Chemistry (Methuen, London, 1967), p. 113. 16 G. Del Piero, M. Di Conca, F. Trifiro and A. Vaccari, in Reactivity of Solids, ed. P. Barret and State, 1972, 13, 2530. L-C. Dufour (Elsevier, Amsterdam, 1985), p. 1029. Paper 711810; Received 9th October, 1987

ISSN:0300-9599    

DOI:10.1039/F19888402477

出版商:RSC    

年代:1988

数据来源: RSC

摘要:

J . Chem. SOC., Faraday Trans. 1, 1988, 84(7), 2483-2498 Influence of some Factors affecting the Stationary Value of the Electric Birefringence of Aqueous Solutions of Poly(styrene sulphonates) in the Presence of 0.01 mol dm-3 NaCl Sybren S. Wijmenga and Michel Mandel* Department of Physical and Macromolecular Chemistry, Gorlaeus Laboratories, Leiden University, P.O. Box 9502, 2300 RA Leiden, The Netherlands The stationary values of the electric birefringence obtained by applying rectangular pulses to aqueous Na-poly(styrene sulphonates) - 0.0 1 mol dmP3 NaCl solutions have been studied for five different samples of polyelectrolytes ranging from M, = 1.8 x lo5 to 12 x lo5 g mol-'. The corresponding Kerr constants determined at low electric field strengths were not linearly dependent on the polyelectrolyte concentration in the concentration region explored, which covered both the the dilute and semi- dilute regimes.The specific Kerr constant at infinite dilution increased with molar mass to the power 0.7. In the semi-dilute regime a scaling law was found of the form (K/K*) - (C/C*)o.8, where K* is the Kerr constant in the dilute concentration regime extrapolated to the critical concentration C*. This concentration and molar mass dependence cannot yet be fully understood in terms of proposed theories. The study of the field strength dependence of the electric birefringence indicates that smaller parts of the chain contribute to the orientation mechanism responsible for the induced macroscopic anisotropy. It is likely that progressive deformation of the more or less flexible chains by high electric fields also plays a role.Recently results on the birefrigence relaxation of Na-poly(styrene sulphonates) (NaPSS) in aqueous 0.01 mol dm-3 NaCl solutions after applying rectangular pulses of relatively low intensity have been reported. 1, The relaxation times derived from these birefringence decay curves exhibited a different concentration and molar mass dependence in the two concentration regions corresponding to the dilute and the semi-dilute regimes, respectively, in agreement with theoretical predictions for flexible polymers3 and their extension to charged chain^.^ In the dilute regime the results could be interpreted by assuming that the physical properties of NaPSS solutions, which show little concentration dependence, are determined essentially by the behaviour of only weakly interacting, individual wormlike chains with a persistence length much smaller than the contour length (at least for molar masses M larger than 1.5 x lo5 g mol-').In the semi-dilute regime, on the contrary, cooperative effects consistent with the formation of a transient network by overlapping chains have been found responsible for the enhanced concentration and molar mass dependence observed for the relaxation times of the birefringence decay. A semi-quantitative explanation for the concentration and molar mass dependence of the longest relaxation time (or the mean relaxation time, which is proportional to it) in terms of the reptation model, as introduced by Edwards5 and de Gennes,6 has been proposed.Additional information on the birefringence relaxation for these NaPSS solutions was obtained in a study in which pulsed rectangular waves of increasing frequencies up to 24832484 Electric Birefringence of Poly(styrene sulphonate) 250 kHz have been used. A preliminary account of these experiments has been published.' The most salient feature is the decrease, above a certain frequency, of the mean relaxation time with frequency and the disappearance of the distinction between dilute and semi-dilute regimes at the upper limit of the frequency range explored. Here we present results concerning the stationary values of the electric birefringence obtained by applying rectangular pulses to the same NaPSS systems as used in our previous publications.In particular, the electric field-strength dependence of the induced stationary birefringence will be examined, as well as the concentration and molar mass dependence of the ensuing Kerr constants. A satisfactory interpretation of the Kerr constant will not be possible, because no theory is at present available for the solutions containing flexible polyions for which the orienting torques arise from an induced dipole moment involving the electric field-perturbed distribution of small ions around the charged macromolecules, as will be discussed briefly in the theoretical section. A comparison of the experimental results with naive theoretical predictions for a Gaussian polyelectrolyte chain, as published elsewhere,' shows the latter to be inadequate for a quantitative understanding of the Kerr constants observed.This emphasizes the need for a more sophisticated model for the electric polarizability and a better treatment for the Kerr constant of flexible polyelectrolyte systems. Theory For solutions containing rigid molecules which can be oriented in an electric field E by an induced and/or a permanent dipole moment, the steady-state value of the electric birefringence, Ano, can be represented by the following equation :9,10 Ano = Ans@. (1) Here An, is the saturation value to which Ano tends in very high fields ( E + 00) and @ is the orientation function, which ranges from 0 to 1 for E + 0 and E + co, respectively. For axially symmetric rigid molecules in dilute solutions the saturation value of the electric birefringence can be repressed in terms of the elements of the diagonalized optical polarizability tensor of the solute molecules, provided corrections for the solvent have been applied and interaction effects between solute molecules may be neglected : ( 2 ) Here n is the refractive index of the solution, p is the number density of solute molecules (number of molecules per unit volume), a,O is the element of the polarizability tensor along the main axis and at that in the perpendicular direction. The factor (n2 + 2 ) / 3 is an 'internal field' correction, valid in a first approximation only, and AaO = az-a; represents the optical anisotropy of the solute molecules.If the main axis of the diagonalized electrical polarizability tensor of the axially symmetric molecule also coincides with the geometrical axis, then the electrical anisotropy Aa" may be written as follows : Here a: and a: are the elements of the electrical polarizability tensor ae defined analogously to the elements of the optical polarizability tensor ao.With the additional assumption that the direction of the permanent dipole moment is parallel to the main geometrical axis of the molecule, the orientation function may, for sufficiently weak fields, be expressed in terms of the electrical anisotropy and the permanent dipole moment p : This leads to the well known result for the Kerr constant K : Ans = (27r/n) [(n2 + 2)/3J2p(a,0 - a:) = ( 2 z / n ) [(n2 + 2)/3l2pAao. Aae = a:-- a:. ( 3 ) @ = [(Aa"/kT)+(p/kT)'] ( E 2 / 1 5 ) . (4) K = (An,/EZ),,, = An,[(Aae/kT) + (,~/kT)~]/15.S .S. Wijmenga and M. Mandel 2485 (Note that this is not th_e classic expression for the Kerr constant k, which is usually defined as K = K/n or K/Ibo, with Lo the wavelength of the light in vacuo.) If interactions are negligible, a concentration-proportional Kerr constant is found which is determined essentially by molecular parameters, as follows from eqn (2) and (5) : K = [(n2 + 2)/312 (2nln) ( N / 15kT) (Acc"/M) [Ad + ( p 2 / k T ) ] C. (6) Here the number density p has been expressed as a function of the concentration C in g per unit volume and the molar mass M of the solute, with N representing Avogadro's constant; p = C N / M . Unlike rigid molecules, flexible macromolecules may be deformed as well as oriented by an applied electric field.In addition, both the electrical and optical properties of the molecules must be treated as conformationally dependent quantities. It has been shown, however, that the Kerr constant, i.e. the initial slope of the birefringence curves us. E', can be expressed in terms of averages for chains unperturbed by the electric field." At higher fields this is probably no longer true, as deformations also may contribute to the measured birefringence. No theory is available at present to treat these effects. The relation for the Kerr constant of non-interacting, non-rigid macromolecules can be put into the following form :11 K / C = [(n2+2)/3]2(n/n)(N/15RTM) [(3 < ,u-a.,u-Tr(p2a"))/kT + 3 Tr(ao.ae) - ((Tra") (Tr a"))]. (7) The permanent dipole moment, as well as the optical and electrical polarizabilities, are defined for each conformation of the macromolecular chain.The brackets (. . . ) indicate averaging over all conformations in the absence of the field. [Note that the values of the averaged scalar quantities appearing between brackets in eqn (7) are independent of the coordinate system in which the tensorial quantities are expressed.] Eqn (8) is of general validity for small field strengths and applies to rigid macromolecules as well. For example, it reduces to eqn (6) in the case of a rigid polymer for which the direction of the main axis of the electrical and optical polarizability tensors coincides with the direction of the permanent dipole moment. In the particular case of a flexible chain, for which the permanent dipole moment as well as the electrical and the optical polarizabilities are respectively the vector and tensor sums of constant contributions from the individual axially symmetric subunits of the macromolecule, i.e. p,, a," and a:, eqn (7) reduces to the following expression, with terms involving one, two or three subunits: K / B = N N C p:Aa,"(3(cos Oikcos O j k ) - (COS Oij))/kT+ C AaZ Aaz(3(cos 20ij) - 1) (8a) i , j .k = l i,j=l B = [(n2 + 2)/312 (n/n) ( N / 15kT) ( C / M ) . (8 4 Here AaS and Aaz represent the optical and electrical anisotropy, respectively, of a subunit, Bij denotes the angle between the axis of subunits i and j , and N is the total number of subunits. For a Gaussian chain the only terms contributing to the sums in eqn (8 a) are those with equal indices, leading to a molar mass-independent Kerr constant as K cc N / M .On the other hand, for a rigid rod with all angles between subunits equal to zero the Kerr constant will be molar mass-dependent, with a leading term proportional For uncharged macromolecules the assumption that the contribution to K from permanent dipole moments and electric (atomic) polarizabilities are sums of constant contributions from the individual subunits of the macromolecules is a reasonable one. For charged macromolecules it is, however, commonly accepted that the main contribution is due to the polarizability arising from the perturbation by the electric field to N 3 / M K M 2 .2486 Electric Birefringence of Poly(styrene sulphonate) of the small-ion distribution around the polyions.Several theoretical approaches to this specific polarization have been proposed [for a discussion see e.g. ref. (12)], but are mostly based on a rigid cylindrical or a broken-rod model. Experimental evidence2* 13, l4 has, however, confirmed theoretical predictions4i "-17 that in the presence of an excess of low-molar-mass electrolyte the polyelectrolyte behaviour can be described in a first good approximation by that of wormlike chain. The only expression for the Kerr constant taking into account the flexibility of the charged macromolecule is that for a Gaussian, uniformly charged chain which has been derived recently :' K = [(n' + 2)/312 ( n / n ) [e2N/kT)'] (Acx&JMmOn) L2fZ C. (9) Here e stands for the elementary charge, Ask,, for the optical polarizability, M,,, for the molar mass of a monomeric unit, L for the (total) persistence length of the chain,f for the fraction of associated counterions (i.e.counterions staying in the close neighbourhood of the charged chain) and Z for the total number of elementary charges borne by the macromolecule. In deriving eqn (9) it has been assumed that only the associated counterions contribute to the electric polarizability (two-phase approxi- mation), which is again split up into contributions of individual subunits. Also, other simplifying assumptions had to be made. According to eqn (9) the Kerr constant of a charged Gaussian chain should be proportional to the molar mass as 2 x M. It will also depend on the ionic strength of the solution, mainly through the persistence length, which increases with decreasing ionic low-molar-mass electrolyte concentration.".l6 It will be interesting to see if this theoretical expression is helpful in understanding the experimentally found Kerr constants for aqueous solutions of fully charged poly(styrene sulphonates) of various molar masses as used in this investigation. Materials and Methods The measurements were performed at 21 "C with five different, nearly monodisperse commercial NaPSS samples (according to Pressure Chemical Company Mw/Mn < 1. l), all dissolved into aqueous 0.01 mol dm-3 NaCl (of analytical grade). The samples were the same as used previ~uslyl,~*' and had molar masses of 1.8 x lo', 4.0 x lo', 5.9 x lo', 6.5 x lo5 and 10.3 x lo5 g mol-l, respectively, as determined by static light scattering.The polyelectrolyte concentrations have been chosen in such a way as to cover the range of the dilute and the semidilute regime.' The electric birefringence measuring equipment has been described in detail elsewhere.2 A pulse generator and amplifier (Cober Electronics) producing voltages between 0 and 2.5 kV was used. Because of the high conductivities of the solutions, only voltages > 500 V as measured directly at the electrodes of the Kerr cell were applied. The electric field in this cell was assumed to be given by E = V/d (10) where V is the voltage and d the spacing of the electrodes in the cell (d = 1.77 or 1.66 mm). Thus any voltage drop due to electrode effects was neglected, as is commonly done. The determination of the birefringence signal An(t) has also been discussed in ref.(2). If the electric field is applied for a sufficiently long time, the birefringence reaches a steady-state value, Ano. In practice this was the case for pulse durations P approximately five to six times the mean relaxation time z of the birefringence decay. The value of An, was obtained as the average value of An(t) fluctuating around its plateau value, which corresponds to a time interval representing the last 8 4 0 % of P. The maximum field strengths used in the experiments were ca. 2.6 kV cm-' (nominal value) for the concentrations C below 1 .O-2.0 g dm-3. At higher concentrations the maximum field strengths applied were lower, down to a value of 0.3 kV cm-' for C = 20 g dm-3, in order to prevent heating effects due to the increase in conductance.S.S. Wijmenga and M . Mandel 6ool L 67 2487 / / ct" 333- 0 , - U P oOo'o O- 67 H I 1 I 1 I I I I I I 61 123 18L 2L5 307 368 L29 L91 552 613 E2/10-8 V m-2 o/o'o / O 0 200- - 0 I 1 675 Fig. 1. Variation of Ano with the square of the field strength E2 for NaPSS, M = 5.9 x lo5 g mol-' at C = 0.25 g dm-3. The curve through the experimental points is the polynomial fit of Ano us. E 2 and the solid line represents Ano = Ke2, where K is the Kerr constant as given in table 3 . The sensitivity of the measuring equipment was such that no significant measurement could be performed below E = 0.2 kV cm-l at the highest and below E = 0.3 kV cm-l at the lowest concentration. Results The steady-state value of the birefringence of all solutions investigated was negative, as has also been found for solutions of NaPSS in the absence of low-molar-mass electrolyte in aqueous ~ o l u t i o n s ~ ~ ~ ~ ~ and for DNA fragments in buffer solutions of low ionic strength.20J21 For all solutions of the lowest molar mass NaPSS (M = 1.8 x lo5 g rnol-l) Ano increased linearly with E 2 .For the other samples, except for the solutions with the higher concentrations where the range of field strengths applied had to be lower, Ano was found to be non-linear in E 2 (see fig. 1). The variation of An, with E 2 was fitted to a polynomial in E2. Because of the tensorial nature of the refractive index it may be quite generally assumed that An, can be represented as a power series in E 2 , provided the electrical polarizabilities are independent of the field strengths.The coefficient of the first term of such an expansion is the Kerr constant. The coefficients of the higher-order terms depend on the type of orientation (permanent and/or induced dipole moment) and on whether the molecules are rigid or flexible. In the case of the rigid molecules theoretical expressions for these coefficients have been This is, however, not the case for flexible molecules, for which only the Kerr constant can be related to molecular quantities, as discussed in the theoreticaI section. Use of the same criteria for the fitting procedure as in ref. (2) (where decay curves had to be fitted to a sum of exponentials) leads to the conclusion that whenever Ano was non- linear in E 2 a polynomial fit with two terms only was satisfactory (the number of separate points for each curve ranged from 20 to 70 and a level of 95 YO was chosen for the probability that the addition of an extra term was or was not due to random errors).2488 Electric Birefringence of Poly(styrene sulphonate) Table 1.First coefficient in the polynomial fit An, us. E 2 . corrected for the Kerr constant of water (M = 1.8 x lo5 g mol-l, C* = 3.9 g dm-3) C/g dmP3 - K / 1O-ls m2 V-' ( - K / C ) / lo-'* dm3 m2 V-' g-' 0.0324 0.08 1 0.36 0.52 0.94 2.40 5.75 6.27 12.67 17.8 0.054 f 0.001 0.1 36 & 0.00 1 0.476 f 0.004 0.66 f 0.01 1.06 f 0.01 1.60 k 0.01 1.86 f 0.02 1.80 f 0.02 2.29 f 0.04 2.94 f 0.05 1.66 & 0.03 1.68 _+ 0.01 1.33 0.01 1.28 f 0.01 1.13 & 0.01 0.67 k 0.01 0.323 & 0.004 0.287 & 0.003 0.181 k0.003 0.164 & 0.003 Table 2.First and second coefficients in the polynomial fit An, us. E 2 ; first coefficient corrected for the Kerr constant of water ( M = 4 x lo5 g mol-', C* = 2.0 g dm-3) C/g dmA3 - K/10-l8 m2 V-' (- K/c)/ dm3 m2 V2 g-' b2/ m4 V4 0.166 0.55 f 0.01 3.30 & 0.06 0.37 f 0.03 0.331 0.99 k0.04 3 .OO f 0.1 2 0.54 f 0.29 0.661 1.69 f 0.04 2.57 & 0.06 1 .o +o. 1 1.32 3.03f0.15 2.29 & 0.1 1 1.9f 1.8 2.64 3.65 f 0.05 1.38 k0.02 5.30 7.0 & 0.4 1.32 f 0.08 10.6 13.30f0.06 1.26 & 0.06 - Table 3. First and second coefficients in the polynomial fit An, us. E 2 ; first coefficient corrected for the Kerr constant of water ( M = 5.9 x lo5 g mol-', C* = 1.5 g dm-3) C/g dmP3 - K / 10-l8 m2 V-2 ( - K/C)/ 10-ls dm3 m2 VP2 g-l b2/10-29 m4 V-4 0.048 0.08 1 0.25 0.58 1.42 3.03 3.29 6.39 7.3 1 21.6 0.173&0.002 0.259 f 0.007 0.9 14 & 0.006 1.89 f 0.01 3.1 1 k0.03 6.8 fO.O1 6.8k0.1 12.2 k 0.3 14.7 f 0.2 25.3 f 0.1 3.60 f 0.04 3.64 k 0.09 0.19 k0.03 3.66 f 0.03 0.31 kO.01 3.26k0.02 1.13 0.07 2.19f0.02 l.Ok0.4 2.24_+ 0.03 2.07 _+ 0.03 1.91 k0.05 2.01 f 0.03 1.17 f 0.01 0.073 f 0.004 It has also been checked carefully that each fitted curve does not deviate significantly from zero for E2 -+ 0.This ensures that the coefficient of the first-order term may indeed by considered to be the Kerr constant [with the proviso that its absolute value should be considered with some care in view of the assumption expressed by eqn (1 O ) ] . The results of the fitting procedures are collected in tables 1-5.In these tables b, represents the coefficient of the second-order term of the polynomial fit if significantlyS. S. Wijrnenga and M. Mandel 2489 Table 4. First and second coefficients in the polynomial fit Ano us. E 2 ; first coefficient corrected for the Kerr constant of water ( M = 6.5 x lo5 g mol-', C* = 1.4 g dmw3) C/g dmP3 - K / 1 0-l8 m2 VP2 (-K/C)/10-18 dm3 m2 VP2 g-l b2/10-29 m4 V-4 0.0378 0.075 0.27 0.54 1.17 2.33 4.97 9.93 0.195+0.014 0.26 k0.03 0.99 k 0.02 1.88 k 0.03 4.0 k 0.2 7.7 + 0.4 12.4 f 0.9 26f 1 5.15 k0.37 0.14 f. 0.03 3.42 & 0.40 0.16 0.02 3.67 f 0.07 0.50 f 0.03 3.46 f 0.06 l.0f0.3 3.4 k 0.2 3.3 & 0.2 3.3 f0.2 2.6 _+ 0. I Table 5. First and second coefficients in the polynomial fit of An, us.E 2 ; first coefficient corrected for Kerr constant of water (M = 10.3 x lo5 g mol-', C* = 1 .O g dmP3) ~~ ~ C/g dm-3 -K/10-18 m2 V-2 ( -K/C)/10-l8 dm3 m2 VP2 g-' b2/10-29 m4 V4 0.109 0.38 0.91 1.84 3.79 6.5 11.9 23.3 0.490 f 0.007 1.52 + 0.01 3.1 1 & 0.05 5.8 +O. 1 14.0 f 0.2 21 .o kO.1 32f 1 47k I 4.50 -L 0.06 3.99 & 0.03 3.42 & 0.05 3.1 5 k0.05 3.68 0.02 2.23 0.1 2.7f0.1 2.0 * 0.1 0.19 f 0.02 0.75 k0.04 2.1 k0.4 4 k 3 differing from zero, and K the coefficient of the first-order term b, corrected for the influence of the water signal : Here Kw is the Kerr constant of water, which we measured to be 1.78 x lo-'' m2 V-', and is quite small as compared to the values of b, for the polyelectrolyte solutions. In the tables the calculated values for the critical concentrations C*, separating the dilute from the semi-dilute regime of these solutions, have also been indicated.The values of C* have been obtained with the help of the theoretical expression for polyelectrolyte-salt solutions proposed by Odijk4, in which C* decreases with molar mass but increases with ionic strength : C* cc 1-4/5(L~-1)-3/5, in which I is the contour length of the polyelectrolyte and I C - ~ the Debye screening length. In a solution without interactions the Kerr constant should be linearly dependent on the concentration of the solute. Close inspection of the K values for all molar masses investigated reveals that this is not the case even in the dilute toncentration range. In fact the apparent specific Kerr constant KL = K/C is found to decrease with increasing concentration (see tables 1-5).A second-degree polynomial can be fittzd to K in the dilute regime (see table 6 and fig. 2), although the number of relevant points is rather (too) small: The coefficient h, is small and, in as far as it differs significantly from zero, is probably due to some residual signals besides that of the solvent, for which a correction has been applied. [Even the largest value of h, (for M = 4.0 x lo5 g mol-') is less than 10% of the K = b,+Kw. (1 1) K = h , + e C + h 2 C 2 . (12)2490 Electric Birefringence of Poly(styrene sulphonate) Table 6. Coefficients of a least-squares second-order polynomial fit of the Kerr constant K us. the concentration C in the dilute regime [see eqn (12)] h,/ 10-ls dm6 m2 M/105 g mol-' h , / l O - l H m2 V-, - ~ f / l O - ' X dm3 m2 V-, g- V-"-* LO '? 3.0 > E 2 N m I I .7 2.0 1.0 C 1.8 -0.0 16 & 0.005 1.40 f. 0.0 1 0.308 f. 0.006 4.0 - 0.16 & 0.06 2.5 & 0.2 0.3 & 0.1 5.9 0.020 & 0.003 4.05 t- 0.02 1.30 kO.1 6.5 - 0.04 _+ 0.03 3.5 k0.2 0.07 _+ 0.1 10.3 - 0.03 4.3 ~ -- ~ ~~ 1:o 210 C/g dm3 L.0 3.0 N I > E 2 2.0 2 % N . 1 .c 0 1 .o I 0 1.0 2.0 3 .O Clg dm-3 Fig. 2. Concentration dependence of the Kerr constant in the dilute regime. The drawn curves represent the fitted curve according to eqn (12). Arrows on the x-axis indicate the positions of the critical concentrations, C*. ( a ) M = 5.9 x lo5 g mol-' (0) and M = 6.5 x lo5 g mol-' (A). (6) M = 1.8 x 10' g mol-l (+), M = 4 x lo5 g mol-' (0) and A4 = 10.3 x lo5 g mol-l (0).specific Kerr constant for the dilute regime g , however.] By this fitting procedure more reliable values than K:, for the specific Kerr constant are thus obtained for all molar masses investigated. (Note that the specific Kerr constant, as defined here,-does not correspond to the classic specific constant Ksp, which is the Kerr constant K per unit volume fraction of the solute: where C, is the volume fraction of the solute molecules.) Reliable values for can be used to check the molar mass dependence of the specific Kerr constant. Inspection of Ksp = R/C\ (13)Mandel S. S. Wijrnenga and M . -1 7.0. - a o 4 - -17.5- - 0 5d105 lb6 M 249 1 I 1 I 5.0 6 .O log A4 Fig. 3. Variation of the specific Kerr constant g, as defined by eqn (12), as function of the molar mass, and the corresponding logarithmic plot.table 6 and fig. 3 reveals that the absolute value of possible to put this dependence in the form of a power law: increases with molar mass. It is gccM";c<c*. (14) From the least-squares linear fit of l o g e us. logM the exponent is found to be v = 0.7kO.l. In as far as the coefficient h, is concerned, its significance can hardly be discussed, in view of the reduced number of points that are available for the fitting procedure represented by eqn (12). In the semi-dilute regime the Kerr constants also seem to follow a power law in the concentration. It turns out that within a good approximation a simple scaling law is obeyed independently of the molar mass: ( K I P ) - (C/C*)Y; c > c* (1 5) with K* the Kerr constant calculated with the help of eqn (12) at the theoretical critical concentration C* (see fig.4). A linear least-squares fit of log (K/k*) us. log (C/C*) yields y = 0.82 f 0.06. The intercept is negligibly small, 0.04 & 0.04, which confirms the validity of the scaling law. If we use the theoretical expression for C* cc M-O.* and assume, for simplicity, that K* cc C* we reach the conclusion that in the semi-dilute regime the Kerr constant depends on the molar mass with a power slightly larger than 0.5. Discussion First we shall discuss the molar mass and concentration dependence of the experimentally found Kerr constants and then inspect more closely the non-linearity of Ano us. E in as far as it has been observed. As pointed out in the theoretical section, only a limited number of theories for the Kerr constant are available for flexible macromolecules, and none of them takes into account intermolecular interactions.This leaves us, strictly speaking, only with the possible interpretation of c. If PSS were to behave as a flexible Gaussian chain with identical contributions of permanent and/or conventional induced electric moment located in each axially symmetric subunit of the macromolecule, the Kerr constant at low concentrations would be independent of molar mass [see eqn @)I. If the expression for the Kerr constant derived for a uniformly charged flexible Gaussian chain with an electric polarizability arising from the perturbation by the electric field of the small-ion distribution around the polyion were to apply, we would expect to a rough approximation that the specific Kerr constant is2492 1.0- h * --.Q 3 00 0 - Electric Birefringence of Poly(styrene sulphonate) I 1 I I 0 0.5 1.0 log(C/C*) Fig. 4. Concentration dependence of the Kerr constant with concentration in the semi-dilute regime according to eqn (15). Symbols are the same as in fig. 2. proportional to M [see eqn (9)]. For a uniformly charged rod the molar mass dependence would be a power of M larger than unity. Obviously none of these predictions applies to the solutions investigated, although the molar mass dependence is not too far from being proportional to the first power of M. This is not surprising, because there is independent experimental evidence for the fact that neither is PSS in 0.01 mol dm-3 NaCl solutions a rodlike particle (provided the contour length I is much larger than the total persistence length L, as is the case for all the molar masses investigated), nor does it behave as a Gaussian chain.In fact intramacromolecular interactions, particularly of electrostatic origin, have been shown to influence its average dimensions.2. 13, l4 However, excluded-volume effects alone cannot explain the molar mass dependence found experimentally for e. It may be useful to examine more closely the assumptions underlying the model leading to eqn (9). Besides assuming (a) the absence of interactions between the subunits of length L, and (b) the absence of contributions from permanent dipole moments and conventional atomic polarizabilities, the most important simplifying suppositions are (c) the description of the electric polarizability due to the perturbation of the small-ion distributions around the flexible uniformly charged macromolecules as the sum of individual contributions from all subunits, (d) the assumption that the ionic atmospheres around each unit do not overlap with those of its neighbours, and (e) the neglect of any radial motion of the associated counterions.The three last assumptions are responsible for the fact that, in contrast to what is expected for an uncharged Gaussian chain, the model calculation leads to a molar mass dependence of the Kerr constant through the mean-square radius of gyration.8 More precisely, assumptions (c) and (d) make the contribution to the total electric polarizability of each subunit not only dependent on its orientation but also on its relative position.Combination of (c), ( d ) and (e) leads to the proportionality of K, to the square of the length of the subunit, L2, as well as to the total charge 2 of the macromolecule. All three assumptions are rather poor even for a Gaussian chain, and probably account for the incorrect prediction concerning the molar mass dependence of the specific Kerr constant. Additionally, ion fluxes into the ionic atmosphere surrounding a charged chain, which have been neglected [assumption (e)], have been shown to be important for the electric polarizability of uniformly charged rodlike particles.22 Note that the correction for the internal field as used in eqn (9) (and in all preceding theoretical expressions quoted) not only is a first approximation but may also be quite inadequate in a system in which anisotropy appears through (partial) orientation of loosely coiled, charged macromolecules.There is no way, at present, toS. S. Wijmenga and M . Mandel 2493 take into account deviations with respect to the mean Lorentz field (from which the internal field factor as used is derived) because of theoretical difficulties, particularly for the internal field in complex systems as considered. Whereas the effect of such deviations is probably small in the dilute regime, it may become important at higher concentrations, the more so the lower the ionic strengths and the higher the field intensities. It seems likely that for a charged, flexible chain the electrical polarizability due to the perturbation by the electric field of the distribution of small ions cannot be represented as the sum of contributions of individual subunits, as is the case for the optical polarizability, because the ionic atmospheres around all charges on the chain do overlap.If this is so, the correlation between the electrical and optical polarizability tensors may be weak and, as a first approximation, may be neglected. Assuming furthermore that both tensors are diagonal after averaging over all conformations with a fixed end-to-end vector Y, with only two different elements one of which is in the direction of Y and the other in the directions perpendicular to it, we may put the general expression for the Kerr constant of non-interacting macromolecules [eqn (7)] into the following simplified form : Here B is defined by eqn (8) and contributions of permanent dipole moments have been omitted.For a wormlike chain in the Gaussian limit the optical anisotropy Aao may be considered to be independent of the molar mass and to be entirely determined by the optical anisotropy of a subunit of length L;' K = B (Aa") (Aa'). (16) with Aa;,,,,, the optical anisotropy of a monomeric unit of length a. Assuming the same relation to hold for a wormlike chain even if not Gaussian and considering that B - M-l, one has to conclude that the average electrical polarizability anisotropy would depend on the molar mass as a power slightly smaller than two (in fact (Aa') - M1.7). This is compatible with the expression for the electric polarizability derived by Fixman22 for a uniformly charged cylinder.This assessment cannot be used to derive any further consequences, however, as long as the influence of the chain flexibility in the theoretical treatment of Fixman has not been established. It certainly would be oversimplified to assimilate, even in a first approximation, the coil-like polyelectrolyte to an equivalent uniformly charged cylinder with a characteristic length of, say, the root mean-square end-to-end distance. One should note, in this context, that Stellwagen2' has found, for solutions of DNA fragments of various lengths in an aqueous buffer of ionic strength lower than that of the present investigation, that the specific Kerr constant increases quadratically with M for fragments containing up to 250 base pairs, but that for higher molar masses this dependence on M is less, albeit rather complicated.For the same kind of polyions Elias and Eden2" have also concluded that the electrical polarizabilities, derived with the help of an equation analogous to eqn (16), increase for short DNA fragments up to 124 pairs as the third power of the total length, but that for larger fragments they fall from the cubic dependence, approaching a power law with exponent 1.1 for sizes > 267 base pairs. Although there is some quantitative disagreement between the results of both investigations, the common conclusion remains that for DNA the influence of the chain length I on the Kerr constant is complicated and may include an effect due to an increasing ratio l / L .This is consistent with our findings for PSS. With these difficulties in mind, and without any theoretical support for the effect of intermacromolecular interactions on the Kerr constant or even on the polarizability due to the perturbation of the distributional small ions by the electric field, it seems premature to try to explain the observed concentration dependence of K in the dilute concentration regime, not to speak of the scaling law, eqn (15), observed in the semi- dilute regime. It is, however, noteworthy that the distinction between the two different2494 Electric Birefringence of Poly(styrene sulphonate) I I I I C/g dm-3 0 0.5 1.0 1.5 Fig. 5. Influence of concentration and molar mass on the characteristic ratio b , J q for deviations of An, from its linear dependence on ,!? [see eqn (18)].Symbols are the same as fig. 2. concentration regimes for the Kerr constant is consistent with the similar distinction for the mean relaxation time of the electric birefringence decay, as pointed out in preceding papers.l* Whereas the latter could be understood semi-quantitatively in terms of the overall motion of the flexible chain, here even a qualitative comprehension of the concentration effects observed and of the origin of the scaling law found for the semi- dilute regime is, not surprisingly, still lacking. Further theoretical work is badly needed for a satisfactory understanding of all the effects and their possible bearing on even the qualitative aspects of the ion-distribution polarizability and electric birefringence of non-rigid polyelectrolyte chains in aqueous salt solutions.As has already been pointed out, the electric birefringence deviates from a simple E2 behaviour for the four highest molar masses investigated in the rather low concentration range C < 1-2 g dm-3. This indicates that the orientation function CD defined by eqn (1) contains contributions depending on a higher power of the electric field than E2, even at the low field strengths used. If the electric field dependence of the steady-state birefringence is written as a power series in E2 An, = - 14 E2[ 1 - ( b 2 / 1 4 ) E 2 + . . .] (18) the ratio (b2/Iq) is characteristic for the deviation from the E 2 linearity of An,. This ratio, calculated with the help of the values given in tables 2-5 for 14, the absolute value of the Kerr constant, does not seem to depend significantly on the concentration for each molar mass investigated and not even on the molar mass itself (see fig.5). This absence of molar-mass specificity could indicate that the deviations observed are not due t o the orientation of the complete macromolecules but to part of the chains only. It should be emphasized that we could only measure a small part of the orientation function, and that we thus did not obtain any information about the behaviour of the orientation function near the saturation value of the electric birefringence. Not much is known at present about the orientation function of non-rigid particles except that it should differ from that of rigid, axially symmetric molecules.lS Various studies on the birefringence and electric dichroism of DNA solutions have shown that for this polyelectrolyte no simple orientation function can be predicted2', 24-28 except for low-molar-mass fragments, where this function seems to correspond to that of a rigid particle with an axially symmetric induced dipole moment.For high ratios l / L and very low field strengths the measured orientation function for DNA resembles that theoretically predicted for rigid particles, but with a permanent dipole. However, the downwards curvature of An, us. E2 starts at values of An,,/Ans < 0.5, the value expected for rigid particles with a permanent dipole moment. It is also difficult to understand the origin of the rather high permanent dipole moment of the double-stranded helix of DNA.To explain this behaviour several different effects have been suggested, allS. S. Wijmenga and M . MandeI 2495 0 10 100 1000 E2/10-8 V2 m-2 Fig. 6. Electric field dependence of the longest relaxation time z, and the relative amplitude of its contribution a, obtained from three exponential fits to decay curves for A4 = 10.3 x lo5 g rnol-l, C = 0.38 g dm-3. excluding the existence of a real permanent dipole moment but attributing the orientational torque to an induced dipole moment, arising principally from the distribution of the small ions in the ion atmosphere around the polyions. One of the possibilites envisaged was to assume that the induced dipole moment saturates at a much lower field strength than the orientation f~nction.~'-~' However, other mechanisms have been proposed involving, e.g.distinct contributions which are ' fast ' or ' slow ' to be established. The latter would be absent at high field strength~,~~-~' without it being altogether clear why this is the case. It seems to us that in all the various models proposed not enough attention was given to the specific influence of the chain flexibility (not to speak of the influence of intermacromolecular interactions on the orientation functions, which were completely ignored). There are indications that the chain flexibility affects the electric birefringence of charged macromolecular chains in solution, in addition to the effect already discussed above. Thus it has been shown that if I/L >, 1 the decay curves of the electric birefringence cannot be fitted to a single exponential. This was observed for aqueous salt solutions of DNA2l~ 2 9 7 31 and of PSS2 At small values of E the longest relaxation time may be considered to be the time constant for the rotational diffusion of an individual coil unperturbed by the electric field, at least in the dilute regime.The shorter relaxation times are probably related to the internal modes of the more or less flexible chain. The relative contribution of the longest relaxation decreases, however, at higher field strengths, probably because deformation of the chain sets in, although the value of the relaxation time remains unaffected (see fig. 6). It may be conjectured that for E+ co the macromolecule becomes more or less stretched under the influence of the electric field.This could explain why the saturation values of the specific electric birefringence (AnJC) oberved by Stellwagen for various DNA fragments,21 including molecules for which I B L, seem to be independent of the contour length. It follows from eqn (2) that the specific birefringence would be proportional to the optical polarizability per unit molar mass, Aa"/M. For rigid axially symmetric particles Aa" cc I, hence the specific birefringence is expected to be independent of molar mass. On the contrary, for a Gaussian coil the optical anisotropy would be independent of molar mass [see e.g. eqn (1 9)] and the specific saturated birefringence is inversely proportional to M. Although2496 Electric Birefringence of Poly(styrene sulphonate) I I / / E2/10-8 V2 m-* Fig.7. Plots of An, (0) and alAn, (a) us. F for A4 = 10.3 x lo5 g mol-l, C = 0.38 g dm-3. The dashed line represents An, = K P and the solid line An, = a,KE2. excluded-volume effects would invalidate the pure proportionality between An,/C and M-l, a molar mass dependence of the specific saturated birefringence remains certain. Thus the observation that An,/C is essentially independent of M would be consistent with the assumption that chains for which l / L 9 1, which in the absence of an electric field or small values of E are coiled, are being stretched under saturating condition^.^^^^^ It has also been noted" that for PSS solutions in the absence of low-molar-mass electrolyte, where stretched conformations are more likely to occur even in the absence of an electric field, An,/C reaches a constant value for molar masses equal or superior to 2 x lo5 g mo1-l.Although in the present investigation it has been impossible to reach saturation values An,, and thus to study the influence of molar mass on the saturation values of the specific birefringence, our findings seem to be consistent with deformation by the electric field. This deformation of the charged chain may also explain the high values of Ans observed or inferred from the plot of An, us. F. In the case of DNA it has been noted that, if instead of Ano the values of a,An, are considered, where a, is the relative contribution to the total birefringence decay curve of the longest relaxation time (which is a measure for the time constant of the overall motion of the macromolecular coil unperturbed by the electric field), much lower saturating values ( a , Ano)E+x = An", are observed.21~29~31 In view of the meaning of a, we may assume that An", represents that part of the saturated electric birefringence which should be due to the complete orientation of the macromolecule if unperturbed by the electric field.This is supported by the finding that for DNA with Z/L > 1 the specific value Afi,/C becomes roughly proportional to the dependence of a , on E (as represented in fig. 6) we have calculated alAn, in the case of our sample with M = 10.3 x lo6 g mol-1 and C = 0.38 dm-3, and also have found that it already reaches a constant plateau within the range of field strengths applied (see fig.7). A value of An", = 1 x lo-' can thus be calculated, corresponding to a specific value of 2 . 6 ~ dm3 g-'. This is only a small fraction of the specific birefringence at saturation which was obtained for NaPSS in aqueous solutions without salt,lg AnJC = 2.8 x dm3 g-l corresponding roughly to the saturation value of fully oriented stretched chains. From the value of At?, and the Kerr constant M-1 , 21 as expected for the saturation value of an unperturbed Gaussian coil. UsingS . S . Wijrnenga and M . Mandel 2491 corresponding to t)e birefringence determined solely by the orientation of the unperturbed chain K = qC(aJE+,,, the value of (Aa) may be estimated with the help of the following equation based on the same assumptions as used in obtaining eqn (16): K - ~n,(&xe)/15 kT. (19) Using the value of of table 5 we thus find for the electrical polarizability anisotropy of the chain, neglecting any specific contribution of deformation, an estimate of lo-" C m2 V-l.The expression of the mean electric polarizability of a Gaussian charged chain calculated in ref. (8), based on the highly simplified model discussed above, is given by (Aa") = ( e 2 f l / 2 k T ) (r') = ( e 2 f l / 2 k T ) ( S ' ) . (20) Here ( P ) and ( S ' ) represent the mean-square end-to-end distance and the mean-square radius of gyration, respectively. Taking for 2 the degree of polymerization of the sample M = 10.3 x lo5 g mol-.', i.e. 2 = 5 x lo3, for f the value close to that predicted by condensation theory, i.e.f z 0.6, and for ( S 2 ) the theoretical values as calculated before ( 2 ) , i.e. ( S 2 ) = 6.2 x nm, we find an estimate for the electrical polarizability anisotropy of 6 x lo-" C m2 V1. Both estimates are of the same order of magnitude. This is more than probably fortuitous. As we have pointed out above, there are good reasons to doubt the validity of eqn (20), which in turn is not entirely consistent with eqn (1 9). Furthermore, the value of the mean-square radius of gyration which has been used to estimate the anisotropy according to eqn (20) was the theoretical value corresponding to a chain which is not Gaussian. The coincidence of both estimated values, incidental as it may be, seems to add some weight to the assumption that the deviations of Ano from pure linearity with I?, already at rather low field strengths, and the high values of birefringence reached with increasing E may be due to contributions of deformations in the case of relatively flexible chains such as PSS in aqueous 0.01 mol dmP3 NaCl solutions.Conclusions The steady-state electrical birefringence of PSS solutions in aqueous 0.01 mol dm-3 NaCl solutions shows a different concentration and molar mass dependence in the Kerr region for the dilute and for the semi-dilute regime. The experimentally found concentration and molar mass dependence of the Kerr constants is difficult to understand, even in the dilute concentration range, because of the lack of an adequate theory for the Kerr constant of wormlike polyelectrolyte chains. Such a theory should include the motion of both counter- and co-ions in the ion atmosphere and in the bulk around non-rigid charged macromolecules, taking into account in a more appropriate way, if possible, the influence of local field effects leading to deviations from the mean Lorentz field.'The deviations of the field dependence of the electric birefringence from simple linearity in E 2 indicate that in the orientation mechanism responsible for the appearance of macroscopic anisotropy, contributions of smaller parts of the chains must exist, in agreement with the observations on the birefringence decay of the same systems. It is also likely that progressive deformation of the more or less flexible chains plays a role. This is further demonstrated by the field-strength dependence of a, An,,, interpreted as being that part of the birefringence that is due only to the contribution of the overall orientation of the unperturbed macromolecular coil.This curve reaches a plateau, indicating saturation, at values of E where the complete birefringence of the solutions is still far from its saturation value.2498 Electric Birefringence of Poly (s ty rene sulp hona te) These investigations have been carried out under the auspices of the Netherlands Foundation for Chemical Research (SON) and with financial aid from the Netherlands Organization for the Advancement of Pure Research (ZWO). We also acknowledge the contributions of Dr F. van der Touw to the experimental parts of this study. References 1 S. S. Wijmenga, F. van der Touw and M. Mandel, in Physical Optics of Dynamical Phenomena and 2 S. S. Wijmenga, F. van der Touw and M. Mandel, Macromolecules, 1986, 19, 1760. 3 P. G. de Gennes, Scaling Concepts in Polymer Physics (Cornell University Press, Ithaca, N.Y., 4 T. Oijk, Macromolecules, 1979, 12, 688. 5 S. F. Edwards and J. W. Grant, J . Phys. A, 1973, 16, 627. 6 P. G. de Gennes, J . Chem. Phys., 1971, 55, 572. 7 S. S. Wijmenga, F. van der Touw and M. Mandel, Polym. Commun., 1985, 26, 172. 8 S. S. Wijmenga and M. Mandel, J . Mol. Liq., 1987, 36, 119. 9 C. T. O’Konski, K. Joshioka and W. H. Orttung, J. Phys. Chem., 1959, 63, 1558. Processes in Macromolecular Systems, ed. B. Sedlacek (W. de Gruyter, Berlin, 1985), p. 87. 1980). 10 E. Fredericq and C. Houssier, Electric Dichroism and Electrical Birefringence (Clarendon Press, 11 K. Nagai and T. Ishikawa, J . Chem. Phys., 1965, 43, 4508. 12 M. Mandel and T. Odijk, Annu. Rev. Phys. Chem., 1984, 35, 75. 13 R. S. Koene and M. Mandel, Macromolecules, 1983, 16, 220. 14 R. S. Koene, T. Nicolai and M. Mandel, Macromolecules, 1983, 16, 227; 232, 15 T. Odijk, J . Polym. Sci., Polym. Phys. Ed., 1977, 15, 477. 16 J. Skolnick and M. Fixman, Macromolecules, 1977, 10, 944. 17 T. Odijk and A. C. Houwaart, J . Polym. Sci., Polym. Phys. Ed., 1978, 16, 627. 18 K. Kikuchi and K. Yoshioka, J . Phys. Chem., 1973, 77, 2101. 19 M. Tricot and C . Houssier, Macromolecules, 1982, 15, 854. 20 J. G. Elias and D. Eden, Macromolecules, 1981, 14, 410. 21 N. Stellwagen, Biopolymers, 1981, 20, 399. 22 M. Fixman, Macromolecules, 1980, 13, 71 1. 23 M. J. Shah, J . Phys. Chem., 1963, 67, 2215. 24 S. Sokerov and G. Weill, Biophys. Chem., 1979, 10, 161. 25 E. Neumann and A. Katchalski, Proc. Natl Acad. Sci. USA, 1972, 69, 993. 26 S. Dickmann, W. Hillen, M. Jung, R. D. Wells and D. Porschke, Biophys. Chem., 1982, 15, 157. Oxford, 1973). 27 S. Dickmann, W. Hillen, B. Morgeneyer, R. D. Wells and D. Porschke, Biophys. Chem., 1982, 15, 263. 28 C. Houssier, J. Bontemps, X. Edmonds-Alt and E. Fredericq, Ann. N.Y. Acad. Sci., 1977, 303, 170. 29 B. Roux, J. C. Bernengo, C. Marion and M. Hanns, J . Colloid Interface Sci., 1978, 66, 421. 30 X. Edmonds-Alt, C. Houssier and E. Frederique, Biophys. Chem., 1979, 10, 27. 31 C. Marion, B. Perrot, B. Roux and J. C. Bernengo, Makromol. Chem., 1984, 185, 1665. 32 R. S. Wilkinson and G. B. Thurston, Biopolymers, 1976, 15, 1555. Paper 7/1820; Received 12th October, 1987

ISSN:0300-9599    

DOI:10.1039/F19888402483

出版商:RSC    

年代:1988

数据来源: RSC

摘要:

J . Chem. SOC., Faraday Trans. I , 1988, 84(7), 2499-2509 Solvation of Thiols An Infrared and Nuclear Magnetic Resonance Study of Ethanethiol Martyn C. R. Symons and Geoffrey P. Archer Department of Chemistry, The University, Leicester LEI 7RH The i.r. spectra of solutions of ethanethiol in CCl, in the S-H fundamental (v) and first overtone (2v) regions have been measured as a function of thiol concentration. A dimerisation constant of 0.038 dm3 mol-' was estimated at 22 "C, the S-H stretching frequencies for free and mono-hydrogen-bonded thioI being at 2585 and 2570 cm-', respectively, in the fundamental, and 5056 and 4993 cm-' in the overtone regions. Beer's law was obeyed for the monomer in dilute solutions, giving absorbances of 9.4 and 0.53 dm2 mol-1 in the v and 2v regions.Spectra for the pure thiol in the fundamental region were analysed in terms of these two bands and a third at ca. 2559 cm-', assigned primarily to the central molecule of linear trimers. The results show that dimers dominate at room temperature, in contrast to results for methanol, which is highly polymerised at room temperature. Also for methanol in CC1, solutions, dimers never constitute a major component, the main equilibria being between monomeric and cyclic units. Thus the phenomenon of cooperativity in hydrogen bonding is far less for thiols than for alcohols. This reflects the reduced hydrogen- bond ' basicity ' of sulphur and the reduced 'acidity' of the S-H protons. There is an inversion in the relative oscillator strengths for (SH)f,ee and (SH),,u,k oscillators on going from the fundamental (v) to the first overtone (2v).For the former, there is a marked increase as the hydrogen-bond strength in RSH-----B units increases, whereas in the 2v(SH) region, the 'free' .band is much stronger. Both regions have been used to obtain a more complete analysis of the compositions of solutions. In basic aprotic solvents only one i.r. feature was detected, which is assigned to EtSH-----B units. We have constructed a correlation between the i.r. and the proton shift n.m.r. data, which is satisfactorily linear over the range of co-solvents from MeCN to Et2NH. There is an abrupt discontinuity for the non-basic solvents CCl, and C6HI2, for which the i.r. shifts are less than those predicted from the corresponding n.m.r.shifts. This is reflected in plots involving the solvent donor numbers. We estimate that the pure liquid at 22 "C contains ca. 49% (SH)Pree units. This result requires that there is an average of one hydrogen bond per thiol molecule, which accords well with the results of a recent Monte Carlo calculation for thiols. Whilst simple thiols are not widely used as solvents, the way in which S-H groups are solvated is a matter of considerable importance in biochemistry. However, these groups are invariably part of relatively complex biomolecules, and in order to centre attention on the thiol group we consider it necessary to study a simple thiol, our choice being ethanethiol (EtSH). This unfortunately rules out the study of aqueous systems because of insolubility. However, the present results form the basis for a study of biologically related thiols.There is an extensive literature on the i.r. and proton magnetic resonance spectra of thiols,l-s most of which has been concerned with dilute solutions in 'inert' solvents or with the pure materials. Particularly relevant studies are those of Spurr and Byers,' 24992500 Solvation of Thiols Forsen,2 Bulanin et al.,3 Marcus and Miller4 and Bicca de Alencastro and S a n d ~ r f y . ~ These studies have established the presence of the monomer S-H band in the 2580-2590 cm-I region and, upon cooling, dimer and oligomer bands in the regions of 2540 and 2525 cm-l, re~pectively.~ The molar absorptions of the dimer and oligomer bands are far greater than those of the monomer bands./ \\\ 2 RSH' R-S S-R Spurr and Byersl invoked equilibrium (1) to explain their results for several thiols in CCl,. The resulting dimer association constants are included in table 1 with our own data and those of other workers. However, Bulanin et aL3 rejected the cyclic dimer and decided that the unassociated S-H bond of the open dimer in equilibrium (2) must contribute to the absorption in the region of the monomer band. This approach, of course, gives rise to quite a different equilibrium constant. Later Marcus and Miller,4 using the S-H proton resonance, attempted to allow for the formation of higher oligomers, using the analytical method of Saunders and Hyneg applied to equilibrium (3). Kn nRSH e (RSH), (3) Details of this analysis are given below. By introducing certain modifications and using an iterative convergence technique they concluded, after all, that n = 2.From their results it is possible also to calculate a shift for the dimer and, if the open structure of eqn (2) is accepted, a shift for the bound S-H proton can be calculated to be 6.16 ppm. Bicca de Alencastro and Sandorfy5 have used i.r. spectroscopy to study self- association of thiols in the mixed (1 : 1) solvent CFC13-C,F4Br2, which they symbolise as FR. Their results are particularly pertinent to the present study, and are considered in detail below. Also pertinent are the computer-simulation studies of Jorgensen" and the matrix isolation studies of Barnes et a1.l' Experimental I.R. Spectra 1.r. spectra were recorded on a Perkin-Elmer 68 1 double-beam spectrometer.Demountable cells with Teflon spacers and calcium fluoride plates were used for the sample solutions. Identical cells containing solvent only were used in the reference beam. Path lengths of 0.5 or 0.025 mm were used, depending on the concentration of the solution being recorded. N.M.R. Spectra Proton resonance spectra were recorded using a Jeol PS 100 spectrometer operating at 100 MHz. Frequencies were measured relative to that of the protons of the internal TMS reference. The spectra were calibrated using TMS sidebands induced at a known frequency from TMS. Chemical shifts are the average of three or more measurements for each concentration. A variable-temperature probe cooled by nitrogen gas generatedM. C. R. Symons and G. P. Archer 250 1 from liquid nitrogen was used for all variable-temperature work.The temperature was thermostatically maintained by electric heating coils to within k0.5 K. First-overtone Near-infrared Spectra First-overtone spectra were recorded on a Perkin-Elmer 340 spectrometer. Silica cells of 1 cm path length were used. The sample cell contained pure EtSH or EtSH in solution. The corresponding reference cell was left empty or contained the solvent only. For variable-temperature work the cells were cooled by applying Drikold or liquid nitrogen to the copper cell mountings. The temperature was thermostatically maintained to within & 0.1 K by a heating coil which surrounds the mountings. Purification of Materials EtSH, used without further purification, was obtained from Aldrich at 98 '/o +purity.Solvents were of the purest grade available and were purified further, usually by fractional distillation over CaH, and storing over CaH, or molecular sieves. Water was purified by de-ionisation followed by passage through a millipore ' Milli-P' system. Results and Discussion For bulk EtSH, we distinguish between three types of thiol molecules, two terminal sites symbolised as (SH)f,ee and (lone-pair),,,, [(LP),,,,] and fully bound units (SH),,,,, as in (1:). R I R I R I R I R I We expect that all dihydrogen-bonded units will be spectroscopically indistinguishable, with the S-H bond in (LP)frec units intermediate in strength between the free and bound units. The (SH),,,, oscillator is expected to be indistinguishable from that of the monomer under our conditions.Thus we expect to detect, under ideal conditions, three i.r. bands, one for (SH)f,ee groups, one for (LP)free groups and one for (SH)bulk groups, at progressively lower frequencies. The oscillator strengths for these bands are expected to increase in this order in the fundamental v(S-H) region, but to decrease in this order in the overtone, 2v(S-H) region, by analogy with the results for 0-H oscillators. On the n.m.r. timescale resonances from the three classes of protons will merge to give a single, weighted-average line. In our work on methanol systems we found a linear correlation between the OH proton resonance shifts and the i.r. shifts, which enabled us to predict values for unknown resonances, such as that for the methanol (LP)free terminal groups.12* l3 One of our aims for this study was to search for such a correlation for t hiols. The' S-H Stretching Band v(SH) Pure Solvents Dilute solutions of EtSH in a range of pure aprotic solvents have a single v(SH) feature which shifts progressively to lower frequencies as the basicity of the solvent increases, 82 FAR I2502 So hat ion of Th io Is Fig. 1. Relationship between the S-H stretching frequency and solvent donor number for EtSH in a range of pure solvents. DMF = dimethylformamide, DMA = dimethylacetamide, HMPA = hexamethylphosphoramide and DMSO = dimethyl sulphoxide. The heavy line is the best fit for all solvents except pure EtSH, CCl, and C,H,, (cyclohexane). The thin line is the best fit for all solvents. Note the unexpected insensitivity of v(SH) to solvents from MeCN to C,H,, (clashed line).with a concomitant increase in width and oscillator strength. When these shifts are displayed as a function of the solvent donor number (DN)14 a linear correlation is obtained up to the weak base, MeCN (fig. 1). However, this line does not include results for the non-polar solvents CC1, or C6HI2, for which the shifts in the S-H band are smaller than predicted. Also the point for pure EtSH is well off the correlating line. The latter result is expected, since the overtone i.r. data show that there is a high concentration of (SH)f,ee groups present in the pure liquid, but because of their relatively low oscillator strength this has almost no effect upon the fundamental v(SH) band. If a proper weighting were used the deviation would be greatly reduced (see below).The former result, that the spectrum hardly changes on going from MeCN to C6H12, is discussed after considering the n.m.r. results. Mixed Solvents Only the binary system CC1,-EtSH has been studied in detail. Dilute solutions show a narrow feature at 2585 cm-l which is assigned to the v(SH) of monomeric EtSH molecules. This accords well with assignments made by others. This band obeys Beer’s law, giving an absorbance of 9.4 dm2 mol-l. Thus this is a relatively weak spectral feature, and other bands in this region are also apparent (fig. 2). These bands seem to be independent of changes in solvent and temperature, and are probably combination bands not involving the S-H unit; their presence is allowed for in the computer analysis1’ (fig.2), but no attempt at assignment has been made. Over most of the mole fraction range, the band a t 2570 cm-l dominates, but as pureM. C: R. Symons and G. P. Archer 2503 I I I J 270 260 2 50 2 40 v(SH)/lO cm-' 1.0 a, 0.6 e % : L) 0.2 1.0 r 2 70 2 60 2 50 240 v(SH)/ 10 cm-' Fig. 2.1.r. spectra showing the v(SH) band for EtSH in CC1,. (a) 2.25 mol dmP3, (b) 6.75 mol dm-3. Note the presence of other features which are not associated with the S-H unit (a, /3 and y). EtSH is approached, it was necessary to include an extra band with vmax at ca. 2559 cm-l in order to simulate the spectra. For the pure solvent at 22 "C this band makes a sig- nificant contribution, although the band maximum remains close to 2570 cm-' (fig.2). The (SH)free band at 2585 cm-l is undetectable for the pure solvent, which might suggest that only cyclic oligomers or relatively high polymers are present. This, however, cannot explain why the band at 2570 cm-l dominates. In fact, because of its relatively low absorbance, a considerable concentration of (SH)free groups (estimated at 49 %, see below) can be present, with no significant effect upon the bandshape of the fundamental feature. We therefore studied the first overtone, 2v(S-H), which favours the more weakly bonded units. The S--H Overtone Band Dilute solutions of EtSH in CC1, have a band at 5056 cm-', which is assigned to the 2v(SH) of monomeric EtSH. This band obeys Beer's law, the results giving an absorbance of 0.53 dm2 mol-'. Thus the intensity is reduced by only a factor of ca.17 on going from v to 2v. At higher concentrations a band at 4993 cm-l grew in, which is assigned to hydrogen-bonded units. There was no clear differentiation between features for (LP)free units and (SH)bulk units, even for pure EtSH (fig. 3). Other bands in the region were unaffected by changes in temperature, solvent or concentration, and are assumed to be combination bands not involving the S-H oscillator. On cooling, the (SH)free band lost intensity in favour of the band assigned to bound units. This band shifted to low frequencies as its intensity increased, but there was still no clear growth of a new low-frequency band. It is probable that the single band detected at 4993 cm-l comprises both the expected bands. This band is almost twice as broad as the corresponding fundamental band, as expected if the widths of these bands stem largely from a distribution of hydrogen bond strengths.The important result is that the presence of (SH)f,ee groups in pure EtSH is clearly established. 82-22504 Soivation of Thiols I I I 540 5 2 0 t t 500 480 ' y o Y 2 v ( S H ) / 10 cm-' Fig. 3. First-overtone i.r. spectrum for EtSH in CCl,, showing the (SH)f,ee band. Other features (a, j? and 7) are not associated with the S-H unit. Comparison with other I.R. Studies In their matrix-isolation studies of EtSH in argon, Barnes et all1 were able to resolve the monomer band into two components (2597 and 2600 cm-l), which were assigned to trans andgauche isomers. They also detected low-frequency bands at 2576 and 2552 cm-l, which were assigned to open-chain dimers and cyclic tetramers, respectively.In the light of our discussion given above, it seems to us possible that these are due to dimers [2576 + (2597/2600) cm-'1 and trimers [2576 + (2597/2600) + 2552 cm-l]. Bicca de Alencastro and Sandorfy studied primarily a range of RSH compounds in their solvent mixture FR. This solvent is favoured because of its low solidification point and the fact that it forms good glasses. However, results obtained with the medium are often unusual owing to the presence of bromine in their solvent. They reported the formation of two distinct bands (2540 and 2525 cm-l) upon cooling a 0.47 mol dm-3 solution of propane-1-thiol in FR. They assigned these bands to the dimer, and a more associated species, respectively.Clearly these assignments are at lower frequencies than our own. This may be explained by extrapolating their dimer frequency [fig. 4(a)] from ca. - 185 to 25 "C. The frequency of the dimer band extrapolates to ca. 2573 cm-l at 25 "C. When this is taken into account their results are in reasonable agreement with our own. However, their conclusion that 'we find free molecules even at the lowest temperatures ' is perhaps questionable, especially for the pure liquids. Proton Resonance Shifts Single-proton resonances were obtained for the S-H proton in all systems studied. For the pure compound the shift was linear with temperature [fig. 4(b)] over a range of ca. 120 K, giving a rate of shift equal to 0.0029 ppm K-'. Shifts for a range of pure solvents are plotted against the fundamental v(SH) maxima in fig.5 , and against the solvent DN in fig. 6. The dashed correlation line in fig. 5 is used to estimate the percentage ofM. C. R. Symons and G. P. Archer 2505 I I I I & I 2530 2 540 2 550 2560 2570 & 2580 wavenumber of dimer S-H band/cm-' 40 25 0 -40 V oc -00 -120 -160 0 -50 V F 3 --. -1oc - 1Lf f -2.0 -1.94 -1.75 - 1.5 6 (SH)/ppm Fig. 4. (a) Wavenumber of the S-H dimer band maxima (propane-1-thiol) as a function of temperature. The frequency extrapolated to 25 "C is ca. 2573 cm-'. (b) Shift for the S-H proton resonance as a function of temperature. The shift extrapolated to the m.pt of EtSH (- 144 "C) is - 1.94 ppm.2506 Solvation of Thiols v (SH)/cm -’ Fig. 5. The proton resonance shift relative to the i.r.band maxima for a range of solvents (key as in fig. 1). The dashed line is drawn as the best fit for all solvents excluding CCl, and C6HI2, and the full line is for all solvents. 60 - 40 - L p” 8 5 0 U 20 - -1.0 (a) -20 -3.0 - 4.0 6 (W (PPm) Fig. 6. The proton resonance shift as a function of solvent DN. Note that the abnormality seen in fig. 1 and 5 is not present. For CCl, the line passes close to the mean of the two literature values, (a) and (b). Note also that the result for pure EtSH now lies on the correlating line.M. C. R. Symons and G. P. Archer 2507 (SH)f,ee present in pure EtSH at 22 "C by utilising expression (4); the calculation assumes that there are no monomers present in the pure liquid. 1.46 = 1 .1 6 ~ + 1 . 7 5 ~ + 1.89(1-2~) x = 0.49 i.e. there is ca. 49% (SH)free at 22 "C. The contrast between the n.m.r. and i.r. results is noteworthy. For the proton resonance results all the data are linearly dependent on the DN. This strongly suggests that it is the i.r. results that give rise to the anomalous insensitivity noted for solvents MeCN, CCl, and C,H,, in fig. 1 and 5. It is well established that i.r. shifts are very insensitive for very weak hydrogen bonds. These results suggest that the hydrogen bonds formed in MeCN are so weak that they induce effectively no i.r. shift. However, proton resonance shifts remain sensitive in this weakly bonding region. This is an important conclusion which, if generally true, suggests that n.m.r. is a better tool for studying very weakly hydrogen-bonded systems than i.r.spectroscopy. It is also noteworthy that the shift for pure EtSH at 22 "C fits well on the n.m.r. shift us. DN plot (fig. 6), in contrast with the i.r. shift us. DN plot (fig. 1). This result accords with the 2v(SH) result, which shows that there is a relatively high concentration of (SH)f,ee groups. These must make a strong upfield contribution to the proton resonance, but because of the low absorbance of the (SH)free band there is no significant contribution to the i.r. shift. Correcting for this moves the EtSH point in fig. 1 to high frequencies, i.e. closer to the correlating line. In view of the shift abnormality in the region (fig. 1) we do not feel that the results can be used to derive an independent estimation of the percentage of (SH)f,ee groups in the system.The EtSH-CCl, System The proton shift is remarkably linear throughout the whole mole fraction range (fig. 7). This is an unexpected result, considering the range of species involved (EtSH monomers, dimers and trimers probably). It is in marked contrast with our results for the MeOH-CCl, system, where the OH proton resonance remains close to that of bulk methanol in the 0-0.6 mole fraction (CCl,) range. Indeed, for the MeOH-CH2Cl, system at low temperatures the resonance hardly changes from the bulk MeOH value in the 0-0.9 mole fraction range, but then shifts rapidly to the MeOH monomer value. This extraordinary behaviour can only be understood if cyclic oligomers dominate in fairly dilute solutions of MeOH in CCl, or CH2Cl,.By cyclising, the and (LP)free terminal groups, both of which would contribute strong upfield shifts, are removed so that reinforcement applies to all bands. Conversely, we argue that for the EtSH-CCl, system such cyclic oligomers are not formed. Thus (SH)free and (LP)free groups are already abundant in pure EtSH, and their concentration increases upon dilution. By using fig. 7 to obtain the proton shift for the (LP)free group the average proton resonance of the open dimer can be calculated to be 1.45 ppm. This was then used in the limiting- slope treatment2 and a dimerisation constant for equilibrium (2) was calculated (table 1). It must be stressed that this dimerisation constant is only valid for solutions at concentrations at which higher polymers can be ignored.We believe that our estimation of the proton resonance for the dimer is more reasonable than the shift used by Marcus and Miller (3.8 ppm). Their shift suggests that the hydrogen bond formed in the dimer is stronger than that formed to HMPA, a result which appears to us to be highly improbable.2508 Solvation of Thiols I I I I I 0 0.4 0.8 1 1.6' mole fraction CC14 Fig. 7. Linear shift in the SH proton resonance for EtSH as a function of dilution by CCI,. Table 1. Dimerisation equilibrium constants for the equilibrium 2 RSH (RSH), K,/dm3 mo1-1 ref. 0.02 1 1 0.012 2 0.0056 4 0.038 this work Efect of Temperature upon EtSH On cooling, both the i.r. and n.m.r. results indicate a marked increase in hydrogen bonding. There is a clear loss of (SH)free groups, the 2v(SH),,,, band being almost non- existent close to the melting point.Hence at the melting point, polymers or cyclic structures must dominate. The average proton shift at the melting point (- 1.94 ppm) gives an enhanced DN of ca. 24 using the correlation of fig. 6. In the overtone region, in addition to a decrease in the (SH)free band, there is a corresponding shift in the band for bound species to low frequencies on cooling. This must in part correspond to the loss of (LP)free (terminal) molecules, and in part to a general increase in bond strength on cooling. Comparison with Monte Carlo Calculations Jorgensen has reported Monte Carlo calculations for H,S and a range of RSH, RSR and RSSR derivatives, including EtSH. For our purposes, the most significant results relate to the extent of hydrogen bonding in the pure compound at ambient temperatures.M . C. R. Symons and G. P. Archer 2509 The results suggest that there is a near-linear set of hydrogen bonds and a substantial occurrence of linear dimers and trimers, with an average of one hydrogen bond per molecule. Cyclic oligomers or higher polymers made no significant contribution. Our results lend experimental support for those findings. References I R. A. Spurr and H. F. Byers, J . Phys. Chem., 1958, 62, 425. 2 S. Forsen, Acta Chem. Scand., 1959, 13, 1472. 3 M. 0. Bulanin, G. S. Denisov and R. A. Pushkina, Opt. Spectrosk., 1959, 6, 754. 4 S. H. Marcus and S. I. Miller, J . Am. Chem. SOC., 1966, 88, 3719. 5 R. Bicca de Alencastro and C. Sandorfy, Can. J. Chem., 1972, 50, 3594. 6 M. M. Rousselot, C.R. Shances Acad. Sci., Ser. C, 1966, 262, 26. 7 M. M. Rousselot and M. Martin, C.R. Seances Acad. Sci., Ser. C, 1966, 262, 1445. 8 M. M. Rousselot, C.R. SPances Acad. Sci., Ser. C, 1966, 263, 649. 9 M. Saunders and J. B. Hyne, J . Phys. Chem., 1959, 31, 269. 10 W. L. Jorgensen, J. Phys. Chem., 1986, 90, 6379. 11 A. J. Barnes, H. E. Hallam and J. D. R. Howells, J . Chem. Soc., Faraday Trans. 2, 1972, 68, 737. 12 M. C. R. Symons, V. K. Thomas, N. J. Fletcher and N. G. Pay, J. Chem. SOC., Faraday Trans. I , 1981, 13 M. C. R. Symons and V. K. Thomas, J . Chem. Soc., Faraday Trans. I , 1981, 77, 1899. 14 V. Guttman, Coordination Chemistry in Non-aqueous Solutions (Springer-Verlag, Berlin, 1968), 77, 1890. chap. 2. Paper 7/1837; Received 12th October, 1987

ISSN:0300-9599    

DOI:10.1039/F19888402499

出版商:RSC    

年代:1988

数据来源: RSC

摘要:

J. Chern. Soc., Faraday Trans. I , 1988, 84(7), 2511-2517 Reactions of 1,1,3,3-Tetramethylcyclobutane on Evaporated Metal Films John K. A. Clarke and Bernard F. Hegartyt Department of Chemistry, University College Dublin, BeFeld, Dublin 4 , Ireland John J. Rooney Department of Chemistry, Queen 's University, David Keir Building, Stranmillis Road, Belfast BT9 5AG, Northern Ireland Reactions of 1,1,3,3-tetramethylcyclobutane (TMCBf-hydrogen mixtures on evaporated metal films have shown that both ring scission to 2,2,4- trimethylpentane (TMP) and ring enlargement to 1,1,3-trirnethylcyclo- pentane (TMCP) are dominant on sintered platinum films, while ring scission to TMP predominates on sintered palladium and on unsintered molybdenum films. Unsintered tantalum and molybdenum films gave a large production of iC, hydrocarbon, particularly above ca.500 K: ring enlargement was a minor reaction found particularly with tantalum. The homogeneous reaction of TMCB giving a large iC, production sets in at 600 K. The possibility, suggested by the product distribution, that the reaction of TMCB on Mo and Ta is a metal-assisted free-radical reaction is examined. While it may not be completely excluded that the catalysed ring enlargement on Pt and Pd is a free-radical reaction it is argued that only the previously proposed Rooney-Samman bond-shift mechanism accom- modates without added qualifications published facts on bond shifts, including those at quaternary carbon atoms, and also ring enlargements. Ring enlargement of hydrocarbons on platinum metals has attracted less study than have other hydrocarbon skeletal reactions.An earlier debate centred on whether ring enlargement of methylcyclopentane on platinum takes place by ring opening to a c6 chain followed by 1,6-ring closure yielding benzene/cyclohexane or whether the reaction is simply a bond shift in which bond switching occurs in a single catalysed step.' A programme of 13C-labelling experiments concluded that the former route makes a contribution.' The mechanism of the bond-shift reaction for acyclic alkanes is now substantially agreed. 3-5 ,4 number of experimental observations may be cited as arguments against a mechanistic equivalence between the two types of reaction. Thus, ring enlargement has been found to be more robust in the face of incorporation of a B sub-group metal as an alloying component in Pt or Rh than is a simple alkane bond-shift reaction, at least when evaporated metal film catalysts are used.'" The distinction is most marked with reactants having a quaternary carbon centre and a dimethylcyclopentadiene inter- mediate having non-classical bonding to a surface metal atom was suggested for ring enlargement of 1,l-dimethylcyclopentane on Pt and Pt-Au.' The reaction of 1,1,3,3-tetramethylcyclobutane (TMCB) on platinum and other metallic surfaces offers the opportunity of extending the available information.This choice was encouraged by an aim, not completely realised, of probing the degree of dehydrogenation of the ring-enlargement intermediate on Pd and Pt using deuterium as a tracer as previously re~iewed.~ Reactions are reported on these and a number of other f Present address : Department of Chemical and Nuclear Engineering, University of New Mexico, Albuquerque, NM 87131, U.S.A.251 12512 1,1,3,3- Tetramethylcyclobutane on Catalytic Films metallic elements located throughout the transition-metal block prepared as evaporated films. Discussion of the ring-opening reaction to 2,2,4-trimethylpentane (TMP) is also of interest mechanistically because neither the selective nor the non-selective route of ring opening (SCM, NSCM, in the terminology of Gault5) is available to TMCB for reasons which have been detailed.3 It has previously been shown that in the equally strained reactant 1,l -dimethylcyclobutane the endocyclic C-C bonds adjacent to the quaternary substituents do not undergo scission on platinum films at 573 This has long been the experience with larger rings.g Experimental A static reaction system made from Pyrex glass was used both for an initial study of the homogeneous (uncatalysed) thermal decomposition of TMCB and for the catalytic experiments.Single metal or alloy films were prepared by evaporation from appropriate metal filaments as previously de~cribed.~ Metals were of best research grade and deuterium/hydrogen was palladium-diffused. The TMCB was synthesised by Wolff-Kischner reduction" of 1,1,3,3-tetra- methylcyclobutane-2,4-dione. The major impurity following fractionation was 3-4 YO of 2,2-dimethylpentane, and this material was used directly. Ta, Ti, Zr and Mo films were examined for catalytic activity directly after deposition at 273 K ('unsintered').Pt, Pd, Rh and Tr films laid with the vessel wall at 773 K were then heated in hydrogen or deuterium at the same temperature for 2 h before admitting the reaction mixture ('sintered'). The standard reaction mixture contained 1 : 10 hydrocarbon to hydrogen (or deuterium) partial pressures, with 0.6 Torrf. TMCB in the 600 cm3 reaction vessel. Analysis of hydrocarbon mixtures was effected during reaction by g.1.c. sampling (column of OV1 on Chromosorb P at 303 K) and by in situ mass spectrometry (by Kratos MSlOc2). Deuterium contours of reaction products was determined by later g.1.c.-m.s. analysis of vessel contents on a separate instrument, as previously detailed.3 A number of confirmatory analyses of products of the TMCB-hydrogen reaction were also carried out by the same means.Product distributions to be presented cannot distinguish between isobutane and isobutene ; for the homogeneous reaction, however, literature reports make it clear that the iC, product here is isobutene. All results to be presented are for initial reactions because conversions were kept below 5 % (except for Ta at higher temperature). Results The Homogeneous Reaction Results of blank-run experiments on the gas-phase decomposition of TMCB in excess hydrogen (table 1) show the major product to be isobutene with a smaller proportion of lighter hydrocarbons at the lower temperatures and minor production of 1,1,3- trimethylcyclopentane (TMCP) at higher temperatures. This homogeneous decom- position commenced at 600 K or a little higher.The activation energy derived for the production of isobutene of 240 kJ mol-' agrees satisfactorily with published values of 273 kJ mol-' l1 and 270 kJ mo1-l.12 Workers have favoured a biradical mechanism for the uncatalysed reaction as rationalised by Woodward and Hoffmann. l3 Expected products from the biradical intermediate are isobutene, 2,4,4-trimethylpent- 1-ene and TMCP of which the former dominates and the second has been observed.12 t Torr = 101 325/760 Pa.J . K . A . Clarke, B. F. Hegarty and J . J . Rooney 2513 Table 1. ‘ Blank ’ (uncatalysed) reaction and initial product distributions of TMCB/hydrogen on unsintered films of early transition metals (TMCB/H, ratio of 1 / 10) catalyst metal T/K Cl-c, iC, TMP TMCP k / % min-lU no catalyst 623 658 68 5 715 Ti 693 Zr Mo Tab 585 647 697 465 498 542 487 516 35 18 5 1 - 6 65 82 93 97 99 100 100 92 tr 76 91 81 I00 0.003 0.008 0.039 0.374 0.23 0.01 1 0.014 0.52 0.88 0.07 0.40 0.90 0.62 Per 10 mg metal in the runs with metal catalysts.With tantalum traces of partially demethylated TMCB species were observed. Catalysis by Early Transition Metals Unsintered Ti and Zr films are not considered to catalyse the reactions of TMCB at all because rates of reaction and product distributions are indistinguishable from those of the homogeneous reaction (table 1). Although self-poisoning was a marked feature (as it was to a lesser extent in all the catalytic runs), the reaction of TMCB on Mo was clearly metal catalysed, the products being those of ring-opening and hydrogenolysis. Hydrogenolysis was the dominant feature of the tantalum-catalysed reaction. Catalysis by Platinum Metals Representative results are given in table 2.In these experiments which were designed primarily to suggest suitable reaction temperatures and times for deuterium runs (vide infra), C,-C, products were not resolved from iC, in the g.1.c. except in the case of early transition metals and Pt-Cu alloy films where a different column condition was used. The dominant process on rhodium films was extensive hydrogenolysis with a trace of ring enlargement. The latter feature was somewhat irreproducible from run to run. Extensive hydrogenolysis was found on iridium at all reaction temperatures; a run at a lower hydrogen/TMCB ratio (2.7 : l), prompted by a study of methylcyclopentane reactions,l* did not reveal any TMCP or TMP product.On palladium, ring scission to TMP was by far the major process, with further break- up to C,-C, a lesser reaction and ring enlargement to TMCP a still more minor route. With platinum selectivity for ring opening to TMP and for ring enlargement to TMCP are comparable and together dominate the reaction. Two Pt-Cu alloy films, prepared as previously described3 and having widely different Pt/Cu ratios, gave ring enlargement as a major fraction of the reaction, even at 4 atom% Pt. Simple ring scission to TMP was relatively reduced (to ca. 25 ?” ring enlargement) on both alloy films. However, there is further break-down to lighter hydrocarbons, particularly on the high-copper film, but the temperature required for reaction is now > 550 K.2514 1,1,3,3- Tetramethylcyclobutane on Catalytic Films Table 2.Initial product distributions of TMCB/hydrogen on sintered platinum-metal films (TMCB/H, ratio of 1 / 10) cat a1 y s t metal T/K Cl-c, a TMP Rh 542 625 Ir 462 495 523 567 Pd 475 534 Pt 447 472 502 88 Pt-12 CU 457 533 4 Pt-96 CU 555 584 96 100 100 100 100 100 24 12 35 8 11 -, CU. 80' -, 15 62, - 47, - k/% min-l TMCP (10 mg metal)-l ~ 1 ' 2 - - - - - 3 9 19 45 45 ca. 20 70 38 42 0.15 0.9 1 0.03 0.04 0.08 0.27 0.04 0.05 0.08 0.19 0.2 1 0.003 0.03, 0.003 0.002 a Partially demethylated TMCB (see table 1). iC, which were resolved in the g.1.c. conditions used for Pt-Cu experiments. measurable due to analysis limitations.Respective production of C,-C, products and of Small, not Deuterium Study The final part of this work was the g.l.c./m.s. analysis of TMCB/deuterium reaction products on sintered films of those noble metals (Pt and Pd) which were active for non- destructive processes. A wide range of catalyst pretreatments failed to produce a metal surface on which deuterium exchange in the products TMCP and TMP had not proceeded too rapidly to permit observation of the lowest possible deuterium content (which would correspond to deuterogenation of the surface intermediate). Such pretreatments included extended sintering, varying the reaction time and temperature and predosing with carbon monoxide. The lowest deuteroisomers observed for Pd were [2H4]TMP and [2H4]- TMCP: the shape of the deuterium-containing product contours and some lack of reproducibility of the cut-off point around 2H4 give rise to doubt as to whether the lowest deuterium content in the two products had been observed.Discussion Earlier review^'^^ l6 have developed the speculation that a metal atom having d-orbitals of appropriate symmetry may offer a catalytic pathway to reaction of cyclobutane. The bulky geminal dimethyl groups of TMCB clearly prevent such interaction with a metal centre, as can be demonstrated by a molecular model. Such mechanisms, without the need to pass judgement on their validity, may thus be excluded from further consideration in the discussion of TMCB reactions. The similarity of reaction products on Ta and Mo to those in the homogeneous reaction of TMCB which takes place at higher temperatures might suggest that the catalysis occurs by a free-radical reaction mediated by the metal. Reaction schemes may be constructed, related to those proposed in the literature for the homogeneous decomposition''.l2 and also considered for reactions of metal c~mplexes,~' whichJ. K. A . Clarke, B. F. Hegarty and J. J. Rooney 2515 Pt 1 R-S Pt p ’ 2 q 1 I \ Pt a Scheme 1. Free-radical reactions mediated by metal. The Rooney-Samman route (R-S), referred to in the discussion, is also shown. P Pt Scheme 2. rationalise the production of isobutene, 2,4,4-Trimethylpent- 1 -ene and of TMCP (scheme 1). Such a mechanism cannot at present be totally excluded, and in its favour is the rather specific production of isobutene (isobutane) in the catalysis.The preference must lie, however, with mechanistic proposals based on surface-bound radicals which have general validity for ring opening and ring enlargement of cyclic alkanes and bond shift of acyclic alkanes. We note that a free-radical-type mechanism which rationalises the hydrogenolysis and isomerisation of neopentane is difficult to devise. Hydrogenolysis and ring enlargement will now be considered in turn in relation to catalytic mechanisms already argued in the literature. The persistent formation of isobutene (isobutane) in the TMCB reaction originates, we believe, in the symmetrical scission of a metallocyclopentane intermediate (scheme 2) which is now well documented in organometallic reactions.6.Thus, the a y adsorbed2516 1,1,3,3- Tetramethylcyclobutane on Catalytic Films species formed initially leads through a carbene-alkene intermediate either (i) to TMP by total hydrogenation or (ii) by limited addition of hydrogen atoms to give the metallocyclopentane shown which cracks symmetrically, yielding two isobutene fragments. Looking ahead to the following discussion, we note that bond-shift is not possible by the metathesis-type mechanism5’ l9 for this structure, that is by direct 1,5- cyclisation of the carbene-alkene intermediate mentioned. This would be impossible sterically (scheme 2). The ring enlargement of TMCB to TMCP is entirely analogous to that of the adamantene dimer which has geminal substitution at C- 1 and C-3 of a cyclobutane ring.This reaction takes place rapidly at 483 K on platinum catalysts and similarly on palladium.20 Some observations and assessments will now be made on available information on bond shift and ring enlargement followed by comments on the present reactant. The supposed uniqueness of platinum as a catalyst metal for bond shift of quaternary carbon centres has been shown in recent years to be untrue. At the same time the difference in reactivity in the bond-shift reaction of non-quaternary and quaternary carbon centres has been shown not to represent a fundamental difference in kind. Thus palladium has been shown to be initially active for neopentane rearrangement but shows rapid self-poisoning due to carburisation. 21 A rhodium-tin evaporated alloy film was active in homologation of neopentane to benzene at 600 K, implying bond-shift rearrangement of the reactant in an initial step.22a Muller and G a ~ l t ~ ~ reported 1,1,3- trimethylcyclopentane ring enlargement at the quaternary centre to give xylenes as 88 O h of aromatics production at 573 K on platinum and as 10-20% of aromatics produced on palladium, rhodium and iron at a similar temperature.An early report of neopentane isomerisation on was questioned. 25 Subsequently, sintered iridium films were reported to produce xylenes from 1,1,3-trimethylcyclopentane, but only above ca. 630 K, and a meaningful xylene/toluene ratio could not be observed because of cracking.22b In a later, comprehensive study, the neopentyl-group bond shift of 2,2,4,4-tetramethylpentane was found to be measurable below 470 K on each of the platinum metal^.^ Metals other than Pt have, then, non-negligible activity in isomerisations at quaternary centres and further there is no good reason to make a distinction in kind between the bond shift of alkanes and ring-enlargement processes. The usually more rapid ring enlargement achieves relief of ring strain and consequently has a greater driving force.Bond shift at a quaternary centre appears from general experience to be retarded by carburisation and to be more prone to self-poisoning. The similar persistence of TMCB ring enlargement on a Pt-Cu alloy of high percentage Cu to the 1,l-dimethylcyclopentane ring enlargement on Pt-Au at high percentage Au shows that the dimethylcyclopentadiene- type intermediate postulated for the latter6 is unnecessary.No analogous structure is possible for TMCB. The Rooney-Samman mechanism1 (scheme 1, ‘ R-S ’) rationalises both reactions. Finally, a comment is offered on the product distributions for the Pt-Cu alloy films. The trend toward deeper cracking within the hydrogenolysis route when platinum is extremely diluted by copper (4 Pt-96 Cu film of table 2) is in harmony with previous studies of Pt-Cu alloy catalysts. Ponec and co-workers have reported that similar dilution of iridium, palladium, nickel or platinum with copper leads to catalysts having increased activity for hydrogenolysis: this is in contrast to the effect of dilution with gold, silver or tin.26*27 In particular, a greater proportion of methane in the hydrogenolysis products with Pt-Cu is found.28 We believe that the copper acts by furnishing a matrix in which a very high dispersion of platinum is promoted and maintained.Analogous behaviour of phosphorus in platinum catalysts derived from ‘Chatt ’ clusters and which showed very selective demethylation activity has been indicated previo~sly.~~ Mechanistically, the sites giving rise to a y adsorption have in the present study at 4 Pt-96 Cu been sufficiently modified to lead to other forms of multipleJ. K. A . Clarke, B. I;. Hegarty and J . J. Rooney 2517 bonding than those depicted in scheme 2. The platinum has, in effect, assumed nickel- type behaviour. One of us (J. J. R.) thanks the S.E.R.C. for purchase of the g.1.c.-m.s. equipment. Thanks are extended to Dr J.G. Hamilton for synthesis of the TMCB. References 1 J. K. A. Clarke and J. J. Rooney, Adv. Catal., 1976, 25, 125. 2 V. Amir-Ebrahimi and F. G. Gault, J . Chem. Soc., Faraday Trans. I , 1981, 77, 1813. 3 0. E. Finlayson, J. K. A. Clarke and J. J. Rooney, J . Chem. Soc., Faraday Trans. I , 1984, 80, 191. 4 Z. Karpinski, Nouv. J . Chim., 1980, 4, 561. 5 For an earlier review see F. G. Gault, Adv. Catal., 1981, 30, 1. 6 A. F. Kane and J. K. A. Clarke, J . Chem. Soc., Faraday Trans. 1, 1980, 76, 1640. 7 A. Peter and J. K. A. Clarke, J. Chem. Soc., Faraday Trans. 1, 1976, 72, 1201. 8 J. J. Rooney, J . Catal., 1963, 2, 53. 9 B. A. Kazanskii, Usp. Khim., 1948, 17, 641. 10 H. L. Herzog and E. R. Buchman, J . Org. Chem., 1951, 16, 99. 11 A. T. Cocks and H. M. Frey, J . Chem. Soc.A , 1969, 1671. 12 T. A. Babcock, J. Am. Chem. SOC., 1969, 91, 7622. 13 R. B. Woodward and R. Hoffman, The Conservation of Orbital Symmetry (Academic Press, New 14 Z. Karpinski and J. K. A. Clarke, J . Chem. Soc., Faraday Trans. 1, 1975, 71, 2310. 15 G. N. Schrauzer, Adv. Catal., 1968, 18, 373. 16 F. D. Mango, Adv. Catal., 1969, 20, 291. 17 T. C . Flood and J. A. Statler, Organometallics, 1984, 3, 1795. 18 R. H. Grubbs and A. Miyashita, J . Am. Chem. Soc., 1978, 100, 1300 and subsequent papers from the 19 J. M. Muller and F. G. Gault, J . Catal., 1972, 24, 361. 20 W. Burns, M. A. McKervey, J. J. Rooney, N. G. Samman, J. Collins, P. von R. Schleyer and E. Osawa, J . Chem. Soc., Chem. Commun., 1977, 95. 21 Z. Karpinski and T. KoScielski, J . Catal., 1979, 56, 430. 22 J. F. Taylor, Ph.D. Thesis (National University of Ireland, Dublin, 1976), (a) p. 74, (b) p. 72. 23 J. M. Muller and F. G. Gault, Proc. 4th Znt. Congr. Catal., Moscow, 1968, preprint no. 15, p. 1850. 24 M. Boudart and L. D. Ptak, J . Catal., 1970, 16, 90. 25 J . R. Anderson, Adv. Catal., 1973, 23, 1. 26 H. C. de Jongste and V. Ponec, J . Catal., 1980, 63, 389. 27 H. C. de Jongste, V. Ponec and F. G. Gault, J. Catal., 1980, 63, 395. 28 e.g. H. C . de Jongste, F. J. Kuijers and V. Ponec, Proc. 6th Znt. Congr. Catal., ed. G. C. Bond, P. B. 29 J. J. Rooney, Faraday Discuss. Chem. Soc., 1981, 72, 87. York, 1970). same group. Wells and F. C. Tompkins (The Chemical Society, London, 1977), vol. 2, p. 915. Paper 7/ 1940; Received 2nd November, 1987

ISSN:0300-9599    

DOI:10.1039/F19888402511

出版商:RSC    

年代:1988

数据来源: RSC