摘要:

J . Chem. SOC., Furaday Trans. I, 1986,82, 189-204 Chemisorption and Catalysis by Metal Clusters : Characterisation and Chemisorption Properties of Ru,(CO),, and H,Ru,(CO),, Supported on Alumina, Silica and Titania David J. Hunt, Richard B. Moyes and Peter B. Wells* Department of Chemistry, The University, Hull HU6 7RX S. David Jackson* and Robin Whyman ICI New Science Group, P.O. Box 11, Runcorn WA7 4QE Ru,(CO),, and H,Ru,(CO),, have been impregnated on to silica, alumina and titania and characterised in the freshly impregnated state and in states achieved by subjecting the freshly impregnated material to washing and to heating to 523 K. The original clusters interact with the support surfaces and are converted to a species of empirical formula Ru,,(CO),,C,,Z,,, where Z represents adsorption sites, n zs 6, 0 d x 3 2.8, 0.2 d y 3 1.5 and 0.2 < z 3 1.6.Retention of metal-metal bonding in this species is demon- strated by ultraviolet/visible reflectance spectra while the infrared spectra contain bands indicative of the presence of carbonyl ligands bonded to ruthenium atoms in formal zero, partial negative and partial positive oxidation states. Carbon monoxide adsorption isotherms have a conventional appearance, but most are composed of a primary and a secondary region. Material adsorbed in the secondary region is removed by evacuation at room temperature whereas that adsorbed in the primary region is removed by evacuation at elevated temperatures. From the experimental evidence it is deduced that carbon monoxide and oxygen each absorb molecularly.In the primary region carbon monoxide adsorption occurs at ruthenium sites on the cluster framework, whereas in the secondary region it occurs at ligand-carbon bonded to ruthenium. For oxygen, adsorption is again at ruthenium sites on the cluster framework. ~ ~ _ _ _ ~ _ _ _ _ _ _ ~ ~ ~~~~~ In four earlier papers, the physical, chemisorption and catalytic properties of Os,(CO), *, Os,(CO),, and H,Os,(CO),, impregnated on to silica, alumina and titania were described.' This paper extends our study of chemisorption and catalysis by metal clusters to include silica-, alumina- and titania-supported Ru,(CO),, and H,Ru,(CO),,. The physical characterisation and chemisorption properties of these materials are reported here and subsequent papers will describe the characteristics of these materials as catalysts for ethene hydrogenation, CO hydrogenation, CO, hydrogenation and ethane hydrogenolysis.Much infrared characterisation of ruthenium clusters supported on common oxides has been published;5-7 however, in our view, the interpretation of such results is ambiguous, hence we have used a variety of physical techniques in an attempt to obtain a clearer understanding of the nature of these materials. A major concern of this study has been to determine whether the ideas and interpretations advanced in ref. (1)-(4), concerning the physical state, chemisorption properties, and activity of catalysts derived from osmium cluster compounds, apply also to the corresponding materials derived from the ruthenium analogues Ru,(CO),, and H,Ru,(CO),,. 189190 Suppo r ted R u then ium Clusters Experiment a1 Instrumentation Electron probe microanalysis was carried out using a Link System model 290-2KX energy-dispersive spectrometer attached to a Cambridge Instruments Geoscan.Temperature-programmed decomposition (t.p.d.) was recorded by the following procedure. A sample weighing 0.300 g was placed on a sintered glass disc in a vertical tube through which helium flowed at 23 cm3 (s.t.p.) min-l. A furnace around the tube enabled the sample temperature to be raised at a steady state of 10 K min-l from 293-523 K. Temperatures were measured by means of a thermocouple located in the catalyst bed. The gas stream leaving the decomposition chamber contained He, CO, CO,, C,H,, and C2H,; the hydrocarbons and CO, were condensed in a trap at 77 K, and the CO passed on to a chromatograph where its instantaneous concentration was detected.Later, the trap was warmed and the yields of CO, and C,H, were estimated. Small quantities of hydrogen, if evolved, would not have been detected. 1.r. spectra were obtained by use of a Perkin-Elmer 580B spectrophotometer; details of the calibration, conditions of measurements and claimed accuracy are given in ref. (1). U.v.-visible spectra were obtained by use of a Pye-Unicam SP 700 spectrophoto- meter fitted with an SP 735 diffuse reflectance attachment; the cell was constructed according to the pattern published by Zecchina et Electron microscopy was carried out using a Jeol JEM lOOC instrument capable of a resolution of 0.5nm; when used to examine conventional metal-silica catalysts this instrument normally renders visible metal particles down to ca.0.9 nm in diameter. CO chemisorption was measured volumetrically; the catalyst sample was placed in a glass vessel of known volume and connected to a stainless steel high vacuum system capable of a base pressure of 1 0-8 Torr (1 Torr FZ 133.3 Pa). Pressures in the sample vessel and in attached calibrated volumes were measured by a pressure transducer, or a calibrated LKB thermal conductivity gauge. Catalyst Preparation Ru,(CO),, and H,Ru,(CO),, were prepared by literature method^.^, lo Each was recrystallised and its purity examined by i.r. spectroscopy. No impurities were detected. Supports used were alumina [Aluminium Oxide C (Degussa)], silica (Cab-0-Sil) and titania (Tioxide).The titania, which was prepared specially for this investigation from an organic titanate, was anatase; its purity was better than 99.9% TiO, and it had a silica content of less than 100 ppm. Each support was dried by heating to 773 K for 16 h in a stream of dry nitrogen. The surface areas of the dried materials, measured by the B.E.T. method of N, physisorption, were: silica, 1 10 5 m2 g-l; titania, 43 5 m2 g-l. Supports were impregnated with solutions of Ru,(CO),, or H,Ru,(CO),, in trichloromethane, the solutions being added to suspensions of support. Evaporation to dryness was carried out in a stream of dry nitrogen at 293 K; all of the cluster compound appeared to be impregnated on to the support. Weights of cluster compound and support were such as to give 2.0% by weight of ruthenium in all catalysts.The supported materials, irrespective of their subsequent treatment, are referred to as, e.g. R~~(CO),~/alumina, H,Ru,(CO),,/titania, etc. This nomenclature is intended to convey that they were prepared by impregnation of the stated cluster carbonyl onto the support; it is not intended to denote the stoichiometry of the supported ruthenium- containing en ti ties. 5 m2 g-l; alumina, 97D. J . Hunt et al. 191 Results Characterisation of Impregnated Materials The impregnation procedure gave an uneven initial distribution of ruthenium species on the support for all six freshly impregnated materials. Low-power optical microscopy (magnification 2000 x ) revealed the presence of small coloured crystals on the white supports and electron-probe microanalysis confirmed, for all samples, that there were small areas of support at which the average ruthenium concentration was high and larger areas at which the average ruthenium concentration was low. Fig.1 presents the solid-state i.r. spectrum of Ru,(CO),, as a nujol mull and the spectra of Ru,(CO),, impregnated onto the three supports. Spectrum (a) contains sixteen bands [in excellent agreement with published spectra for crystalline Ru,(CO),,] ;ll the same sixteen bands are present in the other three spectra. Fig. 2 shows analogous spectra for H,Ru,(CO),, as a nujol mull and for H,Ru,(CO),, impregnated on to the three supports. Good overall agreement between spectra was obtained, although some weak bands in the reference spectrum were not observed in the spectra of the impregnated materials.U.v.-visible diffuse reflectance spectra were examined. For Ru,(CO),,, bands at 390 and 320nm have been assigned, respectively, to o+o* and O*’-+CT* transitions involving orbitals associated with ruthenium-ruthenium bonds.’, Thus these spectra can in principle provide evidence for or against the retention of a clustered state at each stage of catalyst preparation. Silica- and alumina-supported materials were examined (spectra of similarly treated reference supports contained no absorption bands). The technique was not applicable when titania was used as the support because this material is opaque to radiation in the range 220-400 nm. Table 1 shows that both the freshly impregnated materials and the thermally activated materials exhibited at least one, and usually both, of the absorption bands in the range 305-410 nm, and this provides evidence for the retention of cluster integrity.However, the spectra of all the supported materials contain bands above 450 nm which are not present in the spectrum of the corresponding starting material. The spectrum of H,Ru,(CO),, in solution shows one further difference from that of the supported materials, viz. the band at 394 nm in the reference spectrum is very intense [comparable with the MLCT (metal-ligand charge transfer) band for M-CO]. This intensity was absent in the spectra of the supported samples. The lability of the hydrogen in the hydrido-cluster was examined by exchange with deuterium.The supported hydrido-cluster rapidly exchanged lH for ,H ; in contrast, the unsupported cluster did not exchange hydrogen. Characterisation of the Impregnated Materials after Heating to 523 K Chemical changes occurred on heating the impregnated materials to 523 K. The stoichiometry of the change and the products formed were examined by t.p.d. and information concerning the transformed species was obtained spectroscopically. Electron probe microanalysis of the materials heated to 523 K, either in an inert gas stream or under vacuum, showed an even distribution of ruthenium-containing species over the surface of the supports. Thus a spreading of ruthenium species over the surface of the support occurred during this thermal treatment, as was also observed in the analogous osmium system.,? Temperature-programmed Decomposition Each of the six cluster-compound/support combinations were heated in a flow of helium from 293 to 523 K, and the evolved gases were analysed (table 2).The manner of the 7 FAR 1192 Supported Ruthenium CIus ters I I I 2100 2000 1900 wavenumber/cm-' Fig. 1. I I I 2100 2000 1900 wave n um be r/ mi - Fig. 2. Fig. 1. 1.r. spectra (nujol mulls) of (a) Ru,(CO),,; (b) Ru,(CO),,/alumina; (c) Ru,(CO),,/silica; (d) Ru,(CO) , / ti tania. Fig. 2. 1.r. spectra (nujol mulls) of (a) H,Ru,(CO),,; (b) H,Ru,(CO),,/alumina; (c) H,Ru,(CO),,/silica ; (d) H,Ru,(CO),,/titania. progressive loss of carbon monoxide is shown in fig. 3. All samples gave CO and CO, as major products of this heat treatment. Carbonyl ligand loss was very high and the extent of such loss was particularly high when Ru,(CO),,/silica and H,Ru,(CO),,/titania were so heated.Evolution of ethene and ethane indicates that reaction occurs between the cluster compounds and hydroxyl groups present on the support surfaces. D @se Refec tan ce Spectroscopy U.v.-visible diffuse reflectance spectra were obtained for materials heated in vacuum to 323, 373,423,473 and 523 I(. The spectra did not alter significantly as the samples were heated and bands associated with both ligand-metal and metal-metal bonding were present after heating to the highest of these temperatures (table 1).D. J . Hunt et al. 193 Table 1. Peak maxima, Amax, in u.v.-visible absorbance and reflectance spectra assignments Amax/nm metal-ligand samplea charge-transfer unassigned o*l -+ o* o + o* unassigned 240 243 3/A" 225 3/Dc 227 - 4/Ac 226 - 4/sc 225 - 3/A" 225 3 /De 226 - 4lA.f 23 1 - 4/se 227 4/Ag 230 ~ 4/sg 240 - Ru3(C0)12b - H4Ru4(CO),2b - - - - 276 318 32 1 31 1 272 337 313 276 3 52 287 323 333 313 - 306 342 - - - - - - ~ - - - - - 397 394 370 472 532 - 73 5 405 472 - - 71 5 370 454 - 649 - 397 455 - - - 408 481 - ~~ 74 1 375 495 - - - 394 - 543 - - 476 ~ 641 - 694 - 394 617 - 418 .- -~ ~~ - - ~- - - a Code: 3 = Ru3(CO),,, 4 = H,Ru4(CO),,, A = alumina, S = silica.samples. In cyclohexane solution. Heated to 473 K. Heated to 423 K. f Heated to 523 K. g Washed Freshly impregnated. Infrared Spectroscopy As samples were heated so their i.r. spectra changed from the sixteen [RU~(CO)~~] or twelve [H,Ru,(CO),,] band systems to simple two, three, and four band systems (fig. 4,5).Table 2 shows that all spectra exhibit bands at 2060 f 5 and 1991 f 9 cm-l; the relative intensities of the bands are also given in this table. No bands attributable to carbonate or carboxylate were observed. Electron Microscopy Ru,(CO),,/silica and H,Ru,(CO),,/silica, after heating to 523 K, each gave micrographs containing ill-defined features which suggested the presence of very small ruthenium- containing particles that were not properly resolved. Characterisation of Washed Impregnated Materials Freshly impregnated samples were washed at room temperature to remove Ru,(CO),, or H,Ru,(CO),, which had not interacted with the support surfaces and re-examined spectroscopically to determine whether ruthenium carbonyl entities were retained on the supports.Washed H,Ru,(CO),,/alumina were examined by u.v.-visible diffuse reflectance spectroscopy (table 1); spectra obtained are similar, except in respect of a reduced intensity, to those observed for the heated samples. 1.r. spectra of washed samples are shown in fig. 4 and 5 ; band positions and relative band intensities in the spectra of the washed materials are very similar to those obtained for the same materials after heating to 523 K. 1-2Table 2. Temperature-programmed decomposition of supported Ru,(CO),, and H,Ru,(CO),, over the range 293-523 K, and the relative intensities of the bands in the i.r. spectra of the species so produced h -_____ products of t.p.d.(%) number of CO ligands $ s lossa retained per Ru atom % sample (%I co C2H4 C2H6 present ( f 0.1) (1 /;l)/cm-l [relative band intensities (% )] 8. Ru,(CO), ,/silica 70 4 8 52 0 trb 0.0 no measureable intensity z - 1987 (42) 1950 (12) $ ki Ru3( CO) /alumina 19 40 55 2 3 2.8 2060 (40) - 1985(44) 1950(16) 5 Ru,(CO), 2 / ti tania 61 70 24 0 6 1 .o 2065 (46) H,Ru,(CO),,/alumina 20 65 25 5 5 2.3 2055 (28) 2030 (19) 1982 (30) 1946 (23) 3 - - H, Rug( CO), ,/silica 34 55 44 1 0 1.5 2062 (50) 2000 (50) 2 H,Ru,(CO),,/titaniaC 80 57 42 0 1 0.1 2062 - I987 1955 Total carbon loss expressed as the percentage of CO ligands converted to products. tr = trace = <0.5%. Very weak i.r. spectrum; intensities not measureable.D. J. Hunt et al. 195 1 0 / 1 110 n E .- B 0/1 $ 8 1/0 W v) 72 - 011 r - - - - 523 / / / / / 473 / / I Ru,(CO),~/S~ O2 / / / // 323 / / I I I I 1 0 10 20 30 40 time/min Fig.3. Carbon monoxide evolutions in temperature-programmed decompositions. Chemisorption Properties CO Chemisorption on Materials derived from H,Ru,(CO),, No adsorption was observed on the freshly impregnated materials at 293 K. After heating to 523 K under vacuum, all samples adsorbed CO [fig. 6(a)]; the isotherms showed a region in which the extent of adsorption rose steeply at low equilibrium pressures (< 0.2 Torr) followed by a region where further CO was adsorbed (0.2-1 .O Torr). After the measurement of isotherm A for H,Ru,(CO),,/alumina the adsorption vessel was evacuated at 293 K to lop8 Torr for 24 h after which the adsorption measurement was repeated (isotherm D).The displacement of curve D with respect to curve A shows that a considerable quantity of adsorbed CO [ca. 150 pmol (g catalyst)-l] was too strongly adsorbed to be removed by this pumping procedure at room temperatures. After196 Supported Ruthenium Clusters si on 10 " l o t r a n s m i s sio n 1 1 I 1 1 2100 2000 1900 2100 2000 1900 wavenum berlcm-' Fig. 4. wavenumberlcm-' Fig. 5. Fig. 4. 1.r. spectra of various Ru,(CO),,/support combinations : (a) washed Ru,(CO),,/alumina; (b) washed Ru,(CO),,/titania; (c) washed Ru,(CO),,/silica; (d) Ru,(CO),,/alumina after heating to 523 K ; (e) Ru,(CO),,/titania after heating to 523 K; (f) Ru,(CO),,/silica after heating to 523 K. Fig. 5. 1.r. spectra of various H,Ru,(CO),,/support combinations: (a) washed H,Ru,(CO),,/alumina; (b) washed H,Ru,(CO),,/silica; (c) H,Ru,(CO),,/titania after heating to 523 K ; (d) H,Ru,(CO),,/alumina after heating to 523 K ; ( e ) H,Ru,(CO),,/silica after heating to 523 K.measurement of isotherm D, the sample was pumped briefly at 293 K, heated under vacuum to 523 K, and then cooled to 293 K. Further absorption of carbon monoxide gave an isotherm identical to curve A, from which we conclude that the heat treatment caused the desorption of all the strongly adsorbed CO. A similar procedure for H,Ru,(CO),,/titania gave isotherms C and E. These procedures of adsorption, partialD. J . Hunt et al. 197 a I m 0 u I a' - 0 3 E 0 0.2 0 . L 0.6 0.8 1.0 "0 0.2 0.4 0.6 0.8 1.0 P/Torr Fig. 6. (a) Partial isotherms for CO adsorption at 293 K.(A) (open points) adsorption on H,Ru,(CO),,/alumina; (B) adsorption on H,Ru,(CO),,/silica; (C) (closed points) adsorption on H,Ru,(CO),,/titania; (D) adsorption on H,Ru,(CO),,/alumina after the measurement of isotherm (A) and thorough evacuation of the adsorbent at 293 K ; (E) adsorption on H,Ru,(CO),,/titania after the measurement of isotherm (C) and thorough evacuation of the adsorbent at 293 K (for details see text). (b) Isotherms for 0, adsorption at 293 K. (F) adsorption on H,Ru,(CO),,/silica; (G) adsorption on H,Ru,(CO),,/titania. Filled points on (A) and open points on (C) represent the translation of (D) and (E) respectively to new coordinates as described in the discussion. Isotherms (D) and (E) represent adsorption in secondary regions.or complete desorption, and regeneration of the sample in its initial state could be repeated time and again. It thus became clear that CO was adsorbed in two states distinguishable by their strength of adsorption; the region in which adsorption was strong we term the primary region, and that in which adsorption was relatively weak, the secondary region. The transition from the primary to the secondary region occurred at ca. 0.2 Torr. The extents of adsorption in each region are given for all six adsorbents in table 3. The effect on the u.v.-visible reflectance spectra of adsorbing carbon monoxide was examined for H,Ru,(CO),,/alumina. The sample was heated in vacuum to 523 K, cooled, and the spectrum recorded; 100 Torr CO was added at 293 K and the spectrum recorded; finally the sample was evacuated, heated to 523 K in vacuum and the spectrum recorded.A band at 400 nm was observed when carbon monoxide was adsorbed. (This new band totally eclipsed the band already present at 397 nm.) No other band in the spectrum was affected and the spectra before CO adsorption and after CO desorption were identical. Oxygen Adsorption on Materials derived from H,Ru,(CO),, Oxygen was adsorbed at 293 K on H,Ru,(CO),,/titania and H,Ru,(CO),,/silica samples which had previously been heated to 523 K under vacuum [fig. 6(b)]. The quantity of oxygen adsorbed shows good agreement with that of carbon monoxide (table 3). Adsorbed oxygen could be desorbed and readsorbed repeatedly in a manner identical to that described above for carbon monoxide. No secondary region was observed for oxygen adsorption.198 2 0- 10- Supported Ruthenium Clusters Table 3.Carbon monoxide and oxygen adsorption at 293 K CO chemisorptionb samplea primary" secondarj chemisorptionb - oxygen Ru,(CO),,/alumina 41.5 20.9 - rt Ru,(CO),,/silica 61 .O 9.0 ~ Ru,(CO),,/titania 68.9 14.5 __ H,Ru,(CO),,/alumina 154.0 36.0 - H,Ru,(CO),,/silica 15.5 4.5 15.5 H,Ru,(CO),,/titania 60.0 22.4 65.0 a All samples heated to 523 K in uacuo. " Equilibrium pressure = 0.2 Torr. Unitspmol (g sample)-'. Not measured. 90 1 0 0.2 Oi.4 0 . 6 0 . 8 1:O PITorr Fig. 7. Partial isotherms for CO adsorption at 293 K: (a) adsorption on Ru,(CO),,/titania after heating to 523 K; (b) adsorption on Ru,(CO),,/silica after heating to 523 K; (c) adsorption on Ru,(CO),,/alumina after heating to 523 K.CO Chernisorption on Materials derived from Ru,(CO),, Isotherms for carbon monoxide adsorption at 293 K were obtained using the freshly impregnated materials; adsorption by similarly treated blank support, where it occurred, has been subtracted. No adsorption was observed for freshly impregnated Ru,(CO),,/alumina or Ru,(CO),,/titania; however, Ru,(CO),,/silica adsorbed about 40 pmol (g catalyst)-'. Subsequent examination of this Ru,(CO),,/silica sample showed that its i.r. spectrum in the carbonyl stretching region was similar to that given by samples that had been heated to 523 K (table 2), from which we conclude that the vacuum treatment used in preparation for the adsorption measurement had caused partial decomposition of the Ru,(CO),,.D. J .Hunt et al. 199 After heating to 523 K under vacuum, all samples adsorbed carbon monoxide at 293 K; the isotherms (fig. 7) showed a steeply rising extent of adsorption with low equilibrium pressures ( c 0.2 Torr) followed by a region where additional CO was adsorbed (0.2-1 .O Torr). This behaviour mirrors exactly that for samples derived from Carbon monoxide adsorption on Ru,(CO),,/alumina was also examined by u.v.-visible reflectance spectroscopy in the manner described above for H,Ru,(CO),,/alumina. Again a band at 400 nm was observed on adsorption of carbon monoxide. No other bands were affected and the spectra before CO adsorption and after CO desorption were iden tical. H4Ru4(C0)12. Adsorption of oxygen on materials derived from Ru,(CO),, was not studied. Discussion The Freshly Impregnated State The impregnation procedure used gives a heterogeneous distribution of ruthenium- containingentities over the surface ofeach support (optical microscopy and electron-probe microanalysis) and where local ruthenium density is high the species present is Ru,(CO),, or H,Ru,(CO),, [i.r.spectroscopy (fig. 1 and 2)]. U.v.-visible spectra (table 1) of Ru,(CO),, derived materials are consistent with the presence of the original cluster compound, but bands at wavelengths > 450 nm indicate that a second species is present in addition. U.v.-visible spectra of supported H,Ru,(CO),, showed two differences from that of H,Ru,(CO),, in solution; first, the bands at >450 nm indicating the presence of a second species, and second, the loss of intensity of the band at 394 nm on going from solution to supported sample.From the intensity of this band in solution and from previous studies on H,OS,(CO),,~ this band may be assigned to a MLCT transition between the metal and bridging hydride ligands. The intensity of this MLCT band is such that the band due to metal-metal bonding transitions in this region is overshadowed. The loss of this intense band in the spectra of the supported material and the ability to exchange the H-ligands rapidly suggests that the bonding of the H-ligand is altered. Comparing the freshly impregnated state obtained with Ru,(CO),, and H,Ru,(CO),, with those of Os,(CO),, and H,OS,(CO),,~ it is clear that these ruthenium clusters behave in a similar manner to that observed for osmium clusters. The Washed State The nature of the second species, referred to in the previous section, became more apparent when Ru(CO),, or H,Ru,(CO),, crystals were removed from the supports by washing.A low concentration of evenly distributed ruthenium-containing entities was retained on the support giving two- or three-band i.r. spectra (fig. 4 and 5) fundamentally different from the spectra of the starting materials for which no evidence remained. U.v.-visible spectra of washed materials (table 1) confirmed that the evenly distributed material was responsible for the bands above 450 nm. Such spectra also contained bands below 450 nm that were not previously resolved when original cluster compounds were present. Again spectroscopic data obtained from the materials in the washed state are directly comparable with those obtained with washed supported samples derived from Os,(CO),, and from H,OS,(CO),,.~~ Heated Samples The intensities of the bands in the i.r.spectra were such as to show that the second species was a minority species in the freshly impregnated samples. Above 373 K, however, changes occurred so that by 423 K no spectral features attributable to Ru,(CO),, or200 Supported Ruthenium Clusters H,Ru,(CO),, remained, but carbonyl bands attributable to the second species were observed. Although subsequent carbonyl ligand loss occurred on heating between 423-523 K, no effect was observed on the relative band intensities. No crystals of the initial cluster compound were visible by optical microscopy in materials heated to 523 K and electron-probe microanalysis showed an even spread of ruthenium-containing material over the supports.Thus the heating procedure facilitated both the diffusion of ruthenium-containing entities over the supports and the conversion of the original clusters to the second species. U.v.-visible spectra showed that ruthenium-ruthenium bonding was retained by the second species (table 1). These entities could not be removed from any of the supports by washing, which suggests the presence of a metal-support interaction. The amount of carbonyl ligand loss from each cluster-support combination during t.p.d. measurements depended upon both the complex and support (table 2). The effect of heating these ruthenium samples appears to be directly comparable to the effects observed with the appropriate osmium Therefore, at this point we will define the second species using a general formula of Ru,(CO),,C,,Z,, (where Z represents adsorption sites).Retention of Ruthenium-Ruthenium Bonding in Ru,(CO),,C,Z,, Gray and coworkers12 have shown that u.v.-visible spectra provide evidence con- cerning metal-metal and metal-ligand bonding in (unsupported) Ru,(CO),, and other cluster compounds. By analogy, for washed and heated samples containing only Ru,(CO),,C,,Z,,, bands below 450 nm are assigned to metal-ligand charge-transfer transitions and bands in the range 300-400 nm to transitions involving orbitals associated with ruthenium-ruthenium bonds (table 1). Therefore these results suggest that Ru,(CO),, CynZz, contains a ruthenium cluster framework.Examination by electron microscopy of Ru,(CO),,/silica and H,Ru,(CO),,/silica, which had been heated to 523 K, indicated the presence of ruthenium entities having diameters near the limit of resolution; suggesting that the cluster has a nuclearity in the region of 6-10 ruthenium atoms. Temperature-programmed Decomposition The heating of unsupported cluster compounds in the absence of a pressure of carbon monoxide usually results in aggregation to give higher nucleari ty clusters, c1.g. Os, + Os, -+ O S , , ; ~ ~ Ru, + RU,;" Rh, + Rh, + Rh,.ls The electron microscopy results suggest that a similar process occurs when the clusters are supported. By use of t.p.d. of the freshly impregnated state the thermal stability of the supported clusters was examined.Carbon monoxide, carbon dioxide and C, hydrocarbons were produced during t.p.d. The pattern of carbon monoxide loss, shown in fig. 3, suggests that the loss is discontinuous and that the pattern of loss is similar for the same support; indicating that the support plays a part in the thermal degradation of the original cluster. The production of CO, and C, hydrocarbons is linked because CO,, formed by a Boudouard reaction (2CO + CO, + C), supplies active carbon atoms which can be easily hydrogenated.l6. l 7 Direct hydrogenation of the carbonyl ligands at these temperatures ( < 523 K) is unlikely to be favoured relative to hydrogenation of this active carbon. Reactions of CO in the presence of ruthenium to give CO, and C are well documented both for CO absorbedD.J . Hunt et al. 20 1 on supported ruthenium catalystsl87 l9 and in ruthenium cluster chernistry.l, Guczi et al.,O who studied the t.p.d. of Ru,(CO),,/silica also concluded that the production of CO, was by a Boudouard reaction involving carbonyl ligands. Infrared Spectra of Ru,(CO),,C,,Z,, The adsorption of CO on supported ruthenium catalysts has been extensively studied by i.r. s p e c t r o ~ c o p y ; ~ ~ - ~ ~ however, only recently has any consensus been reached in terms of interpretation. Bands observed at 2150 and 2080 cm-l have been assigned to CO linearly bonded to RuO and Rus+ Os-, respectively; the band at 2030 cm-l is assigned to Ru-CO, i.e. zero-valent ruthenium. Interactions with the supports give rise to Rud+Os- species, with a degree of &positive charge depending on the basic strength of the support and Rud-Ms+ species (where M = Al, Si, Ti), with a degree of &negative charge depending on the acid strength of the metal. The spectra obtained from the heated samples and the washed samples (fig.4 and 5 ) have a similar pattern of bands to that for CO adsorbed on ruthenium and are interpreted using Blyholder’s model of back-bonding for CO adsorbed on metals.24 The band at ca. 2060 cm-l, observed in all samples, is assigned to CO linearly bonded to a Ru atom having an induced &positive charge due to the effect of a basic oxygen atom of the support, i.e. OC - Rus+ Oh-. We assign the bands at 2000 cm-l [H,Ru,(CO),,/silica], 1987 cm-I [Ru,(CO),,/titania and H,Ru,(CO),,/titania], 1985 cm-l [Ru,(CO),,/ alumina] to Ru-CO with the ruthenium in a zero-valent state.The shifts are indicative of the basicity of the supports. The interaction may be envisaged as partial electron- transfer from the basic sites of the supports to the ruthenium cluster resulting in an overall decrease in v(C0). Alumina and titania are more basic than silica and hence the frequency of the bands should be in the order silica > titania > alumina as is observed. The shoulder observed at ca. 1950 cm-l in Ru,(CO),,/alumina, Ru,(CO),,/titania and H,Ru,(CO),,/titania can be assigned to CO bonded to ruthenium with a &negative charge which arises as a result of the inductive effect of an aluminium or titanium atom of the support, i.e.OC-RU~-A~~+ (or Tis+). No effect is observed for silica supported samples owing to the smaller electronegativity difference between Ru and Si. The spectrum of H,Ru,(CO),,/’alumina shows some differences from the other spectra owing to the presence of H-ligands. After H,Ru,(CO),,/alumina had been activated by heating to 523 K in the usual way and cooled to room temperature, it was exposed to deuterium gas, whereupon rapid hydrogen isotope exchange occurred. The extent of the exchange catalysed by H,Ru,(CO),,/alumina exceeded that shown by activated H,Ru,(CO),,/titania (by 40%) and by activated H,Ru,(CO)l,/silica (by 80%). Thus, of these activated materials, H,Ru,(CO),,/alumina contains the highest concentration of H-ligands. Hence, whereas the bands at ca.2060 and 1982 cm-l may be assigned as previously, those at 2030 and 1946 cm-l are assigned to OC-Rus+HOs- and OC-Rua-HMs+, respectively. The effect of hydrogen on v(C0) is to lower the frequency by ca. 20 cm-125 and is thought to be short range in effect.26 Chemisorption Properties Isotherms shown in fig. 6, for CO and 0, adsorption on materials prepared from H,Ru,(CO),,/support, exhibit two regions; a secondary region in which adsorbate can be removed chemically unchanged by evacuation of the adsorbent at room temperature, and a primary region in which adsorbate undergoes desorption but only at elevated temperatures. Fig. 6 shows that isotherm D, which represents the secondary region of carbon monoxide adsorption on Ru,(CO),,C,,Z,, derived from H,Ru,(CO),,/alumina, lies on isotherm A if its origin is translated to the point [PITorr = 0.2, n/pmol (g sample)-l =202 Supported Ruthenium Clusters 1501.Similarly, isotherm E, which represents the secondary region of CO adsorption on Ru,(CO),,C,,Z,, derived from H,Ru,(CO),,/titania, lies on isotherm C if its origin is translated to [P/Torr = 0.2, n/ymol (g sample)-l = 501. These observations demon- strate that the onset of the secondary region occurs at 0.2 Torr and that the primary region is complete at that pressure. Material adsorbed in the primary region was not removed by pumping at room temperature for 24 h, hence it is unlikely that these adsorbed species were in equilibrium with those in the secondary region. A threshold value of 0.2 Torr for regions of secondary adsorption has been observed for osmium cluster adsorbents and it was shown that the adsorbates in the primary and secondary regions were not in equilibrium.From the shape of the isotherms obtained for CO adsorption on Ru,(CO),,C,,Z,, derived from Ru,(CO),, (fig. 7) it is apparent that they too have primary and secondary adsorption regions. When a given Ru,(CO),,C,,Z,, is prepared by thermal activation, its adsorption capacity in respect of the primary adsorption region is similar irrespective of whether the adsorbate is CO or 0, (table 3). When CO was adsorbed at 293 K on to a sample of H,Ru,(CO),,/alumina or Ru,(CO),,/alumina previously heated to 523 K, a band in the u.v.-visible spectrum at 400 nm was observed whereas no other band in the spectrum was affected.Zecchina et a1.’ also used u.v.-visible reflectance spectra and obtained a band centred on 400 nm when the heated Ru,(CO),,/alumina system (sample preparation was significantly different from that used by us) was pressurised with CO. This band was interpreted as involving orbitals associated with Ru-Ru bonding and was used as evidence for the presence of a new surface species. However, no other band was affected and this is not easily reconciled with the proposal that a new Ru species was formed. Therefore we discount an analogous interpretation for our work and propose the following. The metal-ligand charge-transfer band at 231 nm was not affected and this demonstrates that the band at 400 nm is associated with adsorbed CO and not ligand CO. Equally the band at ca.340 nm assigned to o*l+ o* and the bands above 450 nm were all unaffected by CO adsorption. Therefore in keeping with Gray and coworkers12 who propose that metal-ligand charge-transfer transitions involving bridged carbonyls should occur in the visible region, this molecularly adsorbed CO may have a bridged structure (1,2) or capped structure (3) which is distinguishable from the linear structure of the carbonyl ligands: c=o I \ 1 RU - Ru Ru-RU I/ \\ I (3) I asdorbed / 0 Ill C Ru (4) ligandD. J . Hunt et al. 203 This interpretation is further supported by our study of 0, adsorption by u.v.-visible spectroscopy when a band at 400 nm was also observed. It is unlikely that the same species would be formed on treating the sample with CO or O,, whereas an adsorbed species may easily have similar structure and energetics. The state of adsorbed oxygen may in principle be represented by structures (5), (6) or (7).However, the u.v.-visible spectrum of Ru,(CO),,C,,Z,, containing adsorbed 0, gave no new change-transfer bands below 250 nm (nor enhancement of those already present) such as might accompany the formation of structure (5). P i RU - RU I ( 5 ) I 0- 0 I I Ru - RU O T O Ru I ( 7 ) The complete removal of adsorbed 0, by heating to 523 K in vacuo also suggests the presence of non-dissociatively adsorbed 0,. Furthermore from measurements of the catalytic activity of these samples it is known that these materials are not poisoned by air, in sharp contrast to the behaviour of conventional polycrystalline supported ruthenium catalyst^.,^ Therefore structure (5) is discounted as a model for 0, adsorption; similarly structure (7) may be discounted as such a mode of adsorption would not satisfy the experimental observation that the quantity of adsorbed CO equals that of adsorbed 0,.Hence structure (6) is proposed as a model for 0, adsorption on Ru,(CO),,C,,Z,,. This chemisorptive behaviour mirrors exactly that found for supported Os,(CO),,, Os6(CO),, and H,Os,(CO),, samples where both a primary and secondary adsorption region were identified and the states of adsorbed CO and 0, were molecular and bridge-bonded. Empirical Formulae of Run( CO),,C,,,Z,, Heating either supported Ru,(CO),, or H,Ru,(CO),, to 523 K results in a species with the general formula Ru,(CO),,C,,Z,, where Z represents adsorption sites and, in addition, some ruthenium atoms are involved in cluster-support interactions.From tables 2 and 3 (assuming all CO, formation occurs by a Boudouard reaction and that the number of adsorption sites is twice the number of CO molecules adsorbed) the empirical formulae are : Ru,(CO),,/alumina Ru,(C0)2.8,C0.,72Z,.,, Ru,(CO),,/silica RU,(CO)~.,~C,.,~~Z,.,~~ RU,(CO)~,/ titania Ru 072 '0.6 72'0. 7 12 H,Ru4(C0),2/a1umina Ru7Z(Co)2. 3 12'0. 2 72'1.6 12 H,Ru,(CO),,/silica H , Ru , ( C 0) , /tit ani a RU,(CO)O. 5 n c o . J O . 2n Ru,(CO),.n,C,.,,Zn.,n For the number of carbon atoms to be integral and for Z to 2 2 , n has a value in the region of 10, in agreement with the conclusions from electron microscopy.204 Supported Ruthenium Clusters Conclusion The physical character and chemisorption properties of these supported ruthenium carbonyl clusters mirrors that previously reported for osmium cluster compounds.l . 2? * However, the ruthenium clusters undergo loss of carbon monoxide and interaction with the support at temperatures below that found for osmium clusters. The species finally attained may be similar in form to that proposed for osmium,1 but with interstitial carbon atoms which are well known in ruthenium chemistry.28 From a comparison of the results presented here with those on apparently similar systems, e.g. Ru,(CO),,/sili~a,~ Ru,(CO)12/A120,6~ it is clear that the preparation method is of major importance and that catalytically active species finally obtained on the support is dependent on that method.We thank Dr D.Urwin of Tioxide for a gift of pure titania. Electron microscopy was performed by P. Worthington, electron-probe microanalysis by F. C. F. Wilkinson and i.r. and u.v.-visible spectroscopy by G. Collier and I. A. Pickering. This work was carried out in the context of a Joint Research Scheme funded by ICI and carried out at the University of Hull. References 1 G. Collier, D. J. Hunt, S. D. Jackson, R. B. Moyes, I. A.Pickenng, P. B. Wells, A. F. Simpson and 2 D. J. Hunt, S. D. Jackson, R. B. Moyes, P. B. Wells and R. Whyman, J . Catal., 1984, 86, 333. 3 S. D. Jackson, R. B. Moyes, P. B. Wells and R. Whyman, J . Catal., 1984, 86, 342. 4 D. J. Hunt, S. D. Jackson, R. B. Moyes, P. B. Wells and R. Whyman, Proceedings of the VIIIth International Congress on Catalysis (Verlag Chemie, Basel, 1984), vol. V, p. 27. 5 G. Webb and J. Robertson, Proc. R. Soc. London, Ser. A , 1974, 341, 383. 6 V. L. Kuznetsov, A. T. Bell and Y. I. Yermakov, J. Catal., 1980, 65, 374. 7 A. Zecchina, E. Guglielminotti, A. Bossi and M. Carnia, J. Catal., 1982, 74, 225. 8 A. Zecchina, M. G. Lofthouse and F. S. Stone, J . Chem. Soc., Faraday Trans. I , 1975, 71, 1476. 9 B. F. G. Johnson and J. Lewis, Inorg. Synth., 1972, 13, 92. R. Whyman, J . Catal., 1983, 80, 154. 10 S. A. R. Knox, J. W. Koepke, M. A. Andrews and H. D. Kaesz, J . Am. Chem. Soc., 1975, 97, 3942. 11 D. M. Adams and I . D. Taylor, J . Chem. Soc., Faraday Trans. 2. 1982, 78, 1561. 12 D. R. Tyler, R. A. Leverson and H. B. Gray, J . Am. Chem. Soc., 1978, 100, 7888. 13 C. R. Eady, B. F. G. Johnson and J. Lewis, J . C'hem. Soc., Dalton Trans., 1975, 2606; P. F. Jackson, B. F. G. Johnson, J. Lewis, M. McPartlin and W. J. H. Nelson, J . Chem. Soc., Chem. Commun., 1980, 224. 14 B. F. G. Johnson, R. D. Johnston and J. Lewis, J. Chem. Soc., Chem. Commun., 1967, 1057. 15 P. Chini, Gazz. Chim. Ital., 1979, 109, 225. 16 P. R. Wentreck, B. J. Wood and H. Wise, J. Catal., 1976, 43, 363. 17 P. Biloen, J. N. Helle and W. M. H. Sachtler, J . Catal., 1979, 58, 95. 18 J. W. A. Sachtler, J. M. Kool and V. Ponec, J . Catal., 1979. 56, 284. 19 G. G. Low and A. T. Bell, J. Catal., 1979, 57, 397. 20 L. Guczi, Z. Schay, K. Matusek, I. Bogyay and G. Steffler, Proceedings of the VIIth International Congress on Catal-psis (Elsevier, Amsterdam, 1980). p. 21 1. 21 R. A. Dalla Betta, J . Phys. Chem., 1975, 79, 2519. 22 M. F. Brown and R. D. Conzalez, J. Phys. Chem., 1976, 80, 1731. 23 A. A. Davydor and A. T. Bell, J. Catal., 1977, 49, 332. 24 G. Blyholder, J . Phys. Chem., 1964, 68, 2772. 25 H. Hyne and F. C. Tompkins, J . Chem. SOC., Faraday Truns. I , 1967,63, 1274; R. A. Dalla Betta and 26 M. Primet, J. M. Basset, M. V. Mathieu and M. Prettre, J . Catal., 1973, 29, 213. 27 D. J. Hunt, S. D. Jackson, R. B. Moyes, P. B. Wells and R. Whyman. J . Chem., Soc., Chem. Commun., 28 Transition Metal Clusters, ed. B. F. G. Johnson (John Wiley, New York, 1980). M. Shelef, J. Catal., 1977, 48, 1 1 . 1982, 85. Puper 5/822; Received 16th May, 1985

ISSN:0300-9599    

DOI:10.1039/F19868200189

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

J . Chem. SOC., Faraday Trans. I , 1986, 82, 205-212 Electron Microscopy of Pt, Pd and Ni Particles in a NaX Zeolite Matrix Andreas Kleine and Peter L. Ryder Materials Physics and Structure Research Group, University of Bremen, 0-2800 Bremen 33, Federal Republic of Germany Nils Jaeger and Gunter Schulz-Ekloff Applied Catalysis Research Group, University of Bremen, 0-2800 Bremen 33, Federal Republic of Germany The morphology, size distribution and crystallography of metal precipitates up to 10 nm in diameter in an NaX zeolite (Si/Al = 1.2) loaded with Pt, Pd and Ni have been studied by electron microscopy. Using special techniques to avoid specimen damage by the electron beam, it was shown by bright and dark field microscopy, selected area diffraction, microdiffraction and lattice imaging that all three metals precipitate as equiaxial single crystals having the normal crystal structure of the bulk metal (f.c.c.) in an intact zeolite matrix.Pt and Pd showed a preferred orientation relationship, with greater scatter in the case of Pd. The thermal treatments used led to monodispersed metal phases with uniform particle size distributions, in the case of Ni to a double distribution. Increasing the degree of ion exchange led to an increase in the density rather than the size of the particles. The results are discussed with regard to possible nucleation and growth mech- anisms, and an explanation for the observed orientation relationship is proposed. Crystalline aluminosilicates of the zeolite type with an open lattice structure are used industrially as molecular sieves and as carriers for metal cata1y~ts.l~~ The present investigation was concerned with zeolite catalysts containing metallic platinum, palladium and nickel particles.The starting material was NaX zeolite, in which the contents of the unit cell may be represented by the sum formula Na,(AlO,),(SiO,),, .kH,O, with n+m = 192 and k z 264. X zeolites have Si/A1 ratios (m/n) in the range 1-1.5. The material used in the present investigation had m/n = 1.2. The structure is.cubic (space group Fd3m), and the lattice parameter for this composition is a, = 2.495 nm. The basic structural units are tetrahedra of oxygen atoms grouped around silicon or aluminium, each oxygen atom being shared by two adjacent tetrahedra, so that the basic formula units are (SO,) and (AlO,)-.In NaX zeolites the excess negative charge is compensated by Na+ ions. The linked tetrahedra form an open network structure with two types of roughly spherical cavities, the supercages (1.3 nm in diameter) and the cages (0.66 nm in diameter), connected by window^.^ The water molecules are free to move through the cavities. In the production of catalysts, the Na+ ions are partially replaced by ions of transition metals such as platinum, palladium or nickel. This ion exchange is followed by dehydration and reduction treatments, leading to the precipitation of the metal in form of finely dispersed particles. The dehydration causes no basic change in the framework structure. Research and development work in this field is directed towards understanding the factors affecting the reactivity and dispersion of the metal particles.The aim is to produce a fine, uniform size distribution which is maintained during the reaction being catalysed. 205206 Electron Microscopy of Ni, Pd, Pt in Zeolite Matrices Some of the factors which have been found to affect the size distribution and the stability of the metal particles the silicon-aluminium ratio (m/n) of the starting material, the degree of ion exchange, and the temperature and duration of the dehydration and reduction treatments. Since the metal particles are generally only a few nm in diameter, transmission electron microscopy (TEM) is the only method by which they can be observed directly.11-18 In addition to the high-resolution imaging of the particles and the surrounding matrix, TEM can provide a great deal of further information which is relevant to the investigation of nucleation and growth mechanism.The crystal structures of the matrix and the precipitates may be determined by electron diffraction which, in addition to direct lattice imaging, also indicates whether the precipitation processcauses damage to the surrounding zeolite lattice. With the aid of modern microdiffraction techniques, crystallographic information may even be obtained from single particles. Orientation relationships between the precipitates and the matrix, which may throw considerable light on possible nucleation mechanisms, may also be detected and evaluated by electron diffraction. Finally, information concerning the chemical composition may be obtained with high spatial resolution by X-ray microanalysis.Despite these advantages, relatively few high-resolution TEM investigations of zeolite catalysts have been published. This is probably due to the fact that the preparation of these materials is difficult and time-consuming, and to the sensitivity of zeolites to radiation damage in the electron beam. The present investigation has shown, however, that these difficulties are not insurmountable. Using suitable preparation and operating techniques it was found that all the usual TEM imaging and diffraction methods could be applied successfully to these materials. Even lattice imaging was possible in some cases without the use of special preparation techniques, such as the substitution of uranyl ions.l9? 2o The present paper describes the techniques used and the results obtained in an investigation of platinum, palladium and nickel precipitates in NaX zeolites, the aim of which was to elucidate the possible nucleation and growth mechanisms of the metal particles. Experimental Specimen Preparation The materials investigated and the ion exchange and reduction treatments used are summarized in table 1 .Platinum and palladium were introduced in the form of their tetrammine complex ions from aqueous solutions of the chlorides. In the case of platinum, four degrees of ion exchange, ranging from 13-52% were investigated. Two palladium-loaded materials (7 and 43% ion exchange) and one material with nickel (23.5%) and calcium (25%) were also studied.Calcium is thought to impede the coarsening of the metal particles by sintering during operation as a ~ata1yst.l~ Nickel and calcium were exchanged from solutions of the acetates. In the case of the platinum and palladium loaded specimens, the dehydration and reduction treatment was carried out at 600 "C for the former and 300 "C for the latter under vacuum or an inert gas. In these materials the precipitation takes place by temperature programmed reduction.21T 22 Nickel was reduced with hydrogen at 300 "C. The specimen number 5 had the same composition as 3, but was reduced for a longer time (4 h) and under helium instead of in a vacuum. After ion exchange and reduction all specimens were sealed in glass capillaries under vacuum to protect them from oxidation during storage.The TEM specimens were embedded in epoxy resin (Epon 812) and sectioned in an ultramicrotome. This embedding material was found to give excellent results, but has the disadvantage that it has to be left for three months at room temperature to harden completely. For a firstA . Kleine, P. L. Ryder, N . Jaeger and G. Shulz-Eklofl 207 Table 1. Materials and treatments degree of ~ ~~ ~ sample ion exchange reduction no. metal (atom % j ion-exchange agent treatment 6 7 8 Pt Pt Pt Pt Pt Pd Pd Ni Ca 13 1 heated to 600 "C at 5 "C min-l held 15 min, vacuum held 4 h, He held 16 h, Ar 5 "C min-l to 600 "C 5 "C min-l to 300 "C 5 "C min to 300 "C, held 25 h, H, 23.5 Ni(CH,COO), 25 Ca( CH,COO), investigation specimens of lesser quality were therefore prepared by one of the following methods: the resin was hardened rapidly (1 week) at 60 "C.Alternatively, a suspension of the fine crystalline powder was dropped onto support films on copper grids and allowed to dry. The edge regions of some crystals were found to be thin enough for electron microscopy. Electron Microscopy The micrographs and diffraction patterns shown in this paper were taken in a Philips EM 420 electron microscope, equipped with an LaB, cathode, at 120 kV. For dark field imaging, tilted-beam illumination was used. Several techniques were used for obtaining electron diffraction patterns and lattice images of the beam-sensitive zeolite crystals. In the first place, the exposure times required were minimized by slightly underexposing and using special developing methods.In searching for suitable diffraction patterns showing both the metal phase and the matrix, the specimen stage was moved while observing the screen in diffraction mode with the lowest possible illumination intensity, and an exposure was made as soon as a satisfactory pattern was observed. For high magnification imaging the focusing was carried out using a very small spot size, so that only a small area of the specimen (50-100 nm in diameter) was irradiated. The specimen was then shifted slightly in the horizontal plane and left for a few minutes with the beam fully defocused in order to eliminate specimen drift during exposure. The lowest possible magnification compatible with the desired resolution was used. Results Platinum Plate 1 shows a typical bright-field electron micrograph from each of the four platinum loaded materials after autoreduction in vacuum (specimens 1 4 in table 1).The platinum precipitates are visible as approximately spherical particles. The effect of increasing the degree of ion exchange is to increase the density and slightly reduce the mean diameter of the particles. It is difficult to measure the diameter of the smallest particles, but a rough estimate indicates that the mean particle diameter is reduced by a factor of ca. 2 from 3 nm to 1.5 nm when the degree of ion exchange is increased from 13 to 52% . In all cases there is only a single particle size distribution with a fairly narrow range of sizes within the zeolite matrix.23208 Electron Microscopy of Ni, Pd, Pt in Zeolite Matrices Increasing the duration of the reduction treatment from 15 min to 4 h (under helium) results in a slight coarsening of the particles.The mean diameter of the platinum particles in specimen 5 (plate 2) was ca. 4 nm. Selected area diffraction from all five materials showed weak diffraction rings which fitted the face-centred cubic structure of metallic platinum with a, = 0.392 nm. In addition, spot patterns from the zeolite lattice were observed, which disappeared after a few minutes exposure to the electron beam at normal intensities. Using the techniques described above, it was, however, possible to obtain a few diffraction patterns showing both the metal and the zeolite matrix. An example (from specimen 5) is shown in plate 3, together with a schematic diagram indicating the Miller indices of the diffraction spots.The diffracted intensity from the metal particles is not distributed evenly around the rings, as one would expect if the orientation were completely random, but is concentrated in certain directions, indicating a preferred orientation of the particles with respect to the zeolite lattice. As can be seen in plate 3, the maxima on the { 1 1 1 ), (200) and (2201 rings of the platinum diffraction pattern lie on reciprocal lattice vectors of the same type with respect to the zeolite lattice. The intensity maxima may therefore be indexed as a single crystal pattern, and the orientation relationship is simply (loo), / / (loo),, (Z = zeolite), i.e. the crystallographic axes of the platinum crystals tend to be parallel to those of the zeolite matrix.Further crystallographic information concerning the precipitates may be obtained from dark field micrographs. As an example, plate 4 shows a bright field micrograph from material 2, the corresponding selected area diffraction pattern and two dark field micrographs from the same area, one of them taken with the objective aperture on an intensity maximum of the { 11 l),, ring and the other with the aperture on the same ring, but halfway between the maxima, As is to be expected, a greater number of particles appears bright in the dark field micrograph from the intensity maximum. However, a certain fraction of the particles seems to have no special orientation relationship with the matrix. The study of a large number of dark field micrographs from this and other materials indicates that there is no visible difference, e.g.in size or shape, between particles close to the preferred orientation and those with an apparently random orientation. In the dark field images, most of the particles showing strong diffraction are evenly illuminated, indicating that they consist of single crystals. The internal structure seen in some of the particles may be due to twins or subgrain boundaries. Microdiffraction showed no evidence for the existence of polycrystalline particles. Plate 5 shows a diffraction pattern in which the spots come mainly from a single platinum particles and belong to a single zone. The splitting of the spots may be due to diffraction from neighbouring particles or to the presence of low-angle grain boundaries in one particle.Palladium The palladium-loaded specimens (materials 6 and 7, table 1) both showed metal particles ca. 5 nm in diameter, similar in appearance to the platinum particles (see plate 6). Again, the main effect of increasing the degree of ion exchange, in this case from 7 to 43%, is to increase the density rather than the size of the particles. Plate 7 shows a selected area electron diffraction pattern from the material 7 (43% ion exchange). The rings may be attributed to a face-centred cubic phase with a, = 0.389 nm, being the normal crystal structure of metallic palladium. The spot pattern arises from the zeolite matrix. The intensity distribution in the ring pattern shows that palladium, like platinum, also has a preferred orientation with respect to the zeolite matrix.The orientation relationship is the same, i.e. (loo),, / / (loo),. The scatter of the individual orientations is, however, greater than in the case of platinum.J . Chrm. Soc., Faradajt Trans. I , Vol. 82, part 1 Plate 1 Plate 1. Transmission electron micrographs (bright field) of Pt-loaded NaX zeolites with different degrees of ion exchange reduced I5 rnin at 600 "C in vacuum (see table I , materials I to 4). Degree of ion exchange: ( a ) 13, ( b ) 25, ( c ) 42 and (d) 52';:). Magnification: 300000 x . A. KLEINE et crl. (Fucing p . 208)J . Chem. Soc., FurudujJ Trims. I , Vol. 82, purt I Plates 2 and 3 Plate 2. Transmission electron micrograph (bright field) of material 5 (42':,, ion exchange.reduced 4 h at 600 "C under He). Compare with plate 1 (c). Magnification 410000 x . Plate 3. Selected-area electron diffraction pattern from the specimen shown in plate 2, with Bragg reflections from the platinum particles and the matrix. The Miller indices are given in the schematic diagram on the right. 'The spots near the centre are from the matrix, and those on the rings from the metal phase. A. KLEINE cf NI.J. Chem. SOC., Faraday Trans. 1, Vol. 82,part I Plate 4 Plate 4. Material 2 (Pt, 25 ion exchange, 15 min 600 "C), (a) bright field, (b) diffraction pattern, (c) tilted-beam dark field from maximum on { 1 1 1 }Pt ring, (d) tilted-beam dark field with aperture between maxima on { 1 1 1 }Pt ring. Magnification 300000 x . A. KLEINE et al.J.Chern. SOC., Faraday Trans. I , Vol. 82, part I Plates 5 and 6 Plate 5. Microdiffraction pattern from the metal phase in material 2. Spot size ca. 20 nm. Plate 6. Transmission electron micrographs (bright field) of Pd loaded specimens with different degrees of ion exchange: (a) material 6 (7"10), (b) material 7 (43"<,). Magnification 4100C3 x . A. KLEINE et al.J . Chem. SOC., Faraday Trans. 1, Vol. 82, part 1 Plates 7 and 8 Plate 7. Selected-area electron diffraction pattern from the material shown in plate 6(h), with Bragg reflections from the palladium particles and the matrix. Plate 8. (a) Transmission electron micrograph (bright field) of nickel precipitates in material 8 (table l), showing two particle size distributions with mean diameters of c’u.10 and 1.5 nm, respectively. Magnification 300000 x . (b) Selected area electron diffraction pattern from the same specimen, showing Bragg reflections from the metal particles and the zeolite matrixJ . Chem. Sac., Faraday Trans. 1, Vol. 82,part 1 Plate 9 Plate 9. High magnification electron micrograph (bright field) of the specimen shown in plate 8. The electron beam is parallel to the [110] zone axis, and both sets of { 1 1 1 ) planes in this zone are clearly visible. Magnification 580000 x . A. KLEINE et al.A . Kleine, P. L. Ryder, N . Jaeger and G. Shulz-Eklof 209 Nickel The appearance of the nickel particles in the material loaded with nickel and calcium (material 8, table 1) is shown in plate 8(a). In this case two particle size distributions are observed: relatively coarse particles with a mean diameter of the order of 10 nm and fine particles with diameters in the range 1-2 nm.Selected area electron diffraction again showed rings from the metal phase and spots from the matrix [plate 8(b)]. The ring pattern was compatible with the usual structure of metallic nickel (face-centred cubic, a,, = 0.352 nm). In this case, however, no preferred orientation relationship was observed. Of all the materials investigated, the nickel-calcium specimens were the most stable under irradiation by the electron beam, and lattice imaging of the zeolite matrix was possible. Plate 9 shows a high magnification image of a zeolite crystal viewed along a [ 1101 zone axis. Particularly good contrast is seen for the two sets of (1 11) planes in this zone.There is no visible destruction of the zeolite lattice in the neighbqurhood of the metal particles. Discussion and Conclusions Summary of Results The results presented above and in previous investigations 11- show that electron microscopy provides a powerful tool for investigation of the form, size distribution and crystallographic properties of metal particles in zeolites. The experimental observations may be summarized as follows. The thermal treatments used (see table 1) lead to the precipitation of fine particles, 5 nm in diameter or smaller, distributed evenly throughout the zeolite matrix. In the case of nickel, two maxima in the particle size distributions are observed, differing by almost one order of magnitude in diameter. The effect of increasing the degree of ion exchange is to increase the density rather than the diameter of the particles.In the case of platinum, a slight reduction in particle diameter is observed. Slightly coarser platinum particles were obtained by prolonging the thermal treatment to 4 h in a helium atmosphere. In contrast to particles which have been observed to have nucleated on the surfaces of the zeolite crystaW4 the internally nucleated precipitates have no particular crystal- lographic form, but are approximately spherical in shape. All precipitates have the equilibrium crystal structure of the corresponding metallic phase. There is no evidence for the occurrence of intermediate, metastable phases during the reduction process. The platinum particles, and to a lesser extent the palladium precipitates, show a preferred orientation near the orientation relationship (lOO), / / ( (m = metal), whilst the nickel crystals are more or less randomly orientated.Although the metal particles sometimes show internal structure, they nevertheless consist essentially of single crystals. Single particles consisting of polycrystalline aggregates were never observed. The metal particles nucleate and grow in an intact zeolite matrix. Particle Growth Mechanisms The fact that the particles are single crystals shows that each precipitate grows from a single nucleus. Since the resulting particles are larger than the supercages, their growth must be associated with material transport within the zeolite lattice. It was proposed in a previous paperls that the accommodation of the debris in defect zeolite lattice sites and210 Electron Microscopy of Ni, Pd, Pt in Zeolite Matrices their ultimate saturation may be the factor limiting particle growth.This could explain why the degree of ion exchange and apparently also the duration of the thermal treatment have little effect on the final particle size. The reason for the existence of a bimodal particle size distribution in the case of nickel is not known. The observed distribution of the particles in the microtome sections shows clearly that they are situated within the bulk and not on the surface. This has been confirmed by X.P.S. meas~rements.~~ The maintenance of an intact zeolite crystal structure during reduction may thus play an essential role in limiting particle growth.The present electron diffraction and lattice imaging observations show that there is no local destruction of the zeolite lattice even in the immediate neighbourhood of the metal particles. Crystallography The existence of the equilibrium metal structure in the precipitates has been demonstrated by X-ray diffraction. In the case of very small particles (ca. 1 nm), however, the diffraction lines from the metallic phase are so weak and broad that they are often undetectable. Nevertheless, by careful analysis of the measured intensities, Galle~ot~~v 25 succeeded in demonstrating the metallic structure of platinum particles with a diameter of 1 nm in Y zeolites. In the light of the results presented in this paper it may now be asserted with some confidence that the direct formation of the equilibrium crystal structure is a general property of the formation of metal particles in zeolites.The occurrence of an orientation relationship between the metal particles and the zeolite matrix cannot be explained by lattice matching between the two phases, since the structures have no recognizable similarity, and the precipitates exhibit no crystalline form. Since the larger particles embedded in the zeolite lattice cannot change their orientation appreciably during further growth, the process determining the orientation relationship must take place during the nucleation or the very first stages of growth. Any model for this process must account for: ( a ) the occurrence of the particular orientation relationship observed, (b) the fact that the orientation relationship is not obeyed exactly and ( c ) the differences between platinum, palladium and nickel.These facts are explained, at least qualitatively, by the following proposed mechanism. Since the metal atoms are free to move through the channels of the zeolite structure, it may be assumed that the first clusters form within the (super- or p) cages. These clusters are free to change their orientation within the cages and may at first adopt a random orientation. Once the particle size has reached the cage dimensions, however, further growth requires removal of material from the surrounding zeolite matrix. Since this process requires additional activation energy and is therefore slower, the addition of atoms to the cluster will continue as long as possible without destruction of the zeolite lattice, i.e.the initial growth process will proceed in such a way that the maximum possible number of metal atoms are fitted into the intact cage. Assuming that the metal atoms are arranged from the very beginning of cluster formation on the sites of a face-centred cubic lattice, the number of atoms fitting within the geometric limits of the cage will depend on the orientation of the metal lattice. As long as the cluster is free to change its orientation within the cage, the further addition of atoms will therefore lead automatically to that orientation which admits the maximum number of atoms. Both types of cages have cubic symmetry and may be represented by truncated cubo-octahedra or truncated octahedra, re~pectively.~ It may be assumed that optimum space filling is attained when the symmetry of the metal lattice matches that of the cage, which would give the observed orientation relationship.Once the maximum number of atoms have been packed into a cage, the exactness of the fit will depend on the size of the metal atoms in relation to the cage dimensions. For certain atom sizes the fit will be exact, leading to a precise orientation relationship. IfA . Kleine, P. L. Ryder, N . Jaeger and G. Shulz-Eklof 21 1 the atoms are a little smaller than this critical size, the cluster will have a certain freedom of movement, even when there is no room for a further atom. This may be the explanation for the observed scatter in the orientations and for the difference between atoms of different sizes, e.g.platinum (atomic radius 0.1387 nm) and palladium (0.1375 nm). As is predicted by the model, the slightly smaller atom shows the greater scatter. The particularly large scatter observed in the case of nickel, which has an even smaller radius (0.1246 nm) may also be a size effect, but there is another possible explanation in this case. It must be assumed that both platinum and palladium nucleate in the supercages, since the tetrammine complex ions do not fit into the smaller cages. Nickel, however, does not suffer from this restriction and may therefore nucleate also in the cages. As will be shown in a quantitative treatment (to be published), the scatter in the orientation resulting from a given size misfit is greater, the smaller the number of atoms in the cavity. Whereas about 100 nickel atoms can be accommodated in a super- cage, few more than 10 fit into a p cage.These qualitative arguments show that a study of the orientation relationship may provide important information concerning the mechanism of nucleation and growth of these particles. Outlook Computer calculations are being carried out to verify ( a ) that the observed orientation relationship is indeed that which allows the maximum number of metal atoms to fit into the cavity and (b) the observed dependence of orientation scatter on the atomic size ratios. The results of these calculations, when compared with the experimental results, njill provide a critical test of the model proposed above. The influence of the reduction conditions on the particle size also requires further investigation.It is not known, for example, whether the effect observed in the case of material 5 was due to the prolonged heat treatment or to the helium atmosphere. An influence of the inert gas atmosphere on the reduction reaction has been demonstrated by temperature-programmed desorption and differential thermal analysis22 and is thought to be due to interference of the gas with the diffusion of the reaction products in the channels of the zeolite structure. We thank Miss G. Wildeboer for providing the zeolite samples, Mrs U. Boseck for the preparation of the electron microscopy specimens, Mr D. Exner for making available unpublished measurements of size distributions, and a referee for his careful and thorough reading of our manuscript.N. J. and G. S-E. gratefully acknowledge financial support of the Deutche Forschungsgemeinschaft. References 1 A. P. Bolton, in Zeolite Chemistry and Catalysis, ed. J. A. Rabo, ACS Monograph 171 (American 2 E. Gallei, Chem. Ing. Techn., 1980, 52, 99. 3 H. Heinemann, in Catalysis: Science and Technology, ed. J. R. Anderson and M. Boudart (Springer, 4 D. W. Breck, Zeolite Molecular Sieves (J. Wiley, New York, 1974), p. 29. 5 Kh. M. Minachev and Ya. I. Irakov, in Zeolite Chemistry and Cataljsis, ed. J. A. Rabo, ACS Monograph 171 (American Chemical Society, Washington, 1976), p. 552. 6 J. B. Uytterhoeven, Acta Phys. Chem. (Szeged), 1978, 24, 53. 7 P. A. Jacobs, Carboniogenic Activity of Zeolites (Elsevier, Amsterdam, 1977).p. 183. 8 M. Briend-Faure, J. Jeanjean, M. Kermarec and D. Delafosse, J . Chem. Snc., Faraday Trans. 1, 1978, 9 M. Briend-Faure, J. Jeanjean, D. Delafosse, P. Gallezot, J. Phys. Chem.. 1980, 84, 875. Chemical Society, Washington D.C., 1976) p. 714. Berlin, 1981), vol. 1, p. 1. 74, 1538. 10 N. I. Jaeger, P. L. Ryder and G. Schulz-Ekloff, in Structure and Reactirity cfMon’lJedZeolites, ed. P. A. Jacobs et al., (Elsevier, Amsterdam, 1984) p. 299. 11 P. Gallezot, in Catalysis by Zeolites, ed. B. Imelik et al., (Elsevier, Amsterdam, 1980) p. 223.212 Electron Microscopy of Ni, Pd, Pt in Zeolite Matrices 12 D. Exner, N. Jaeger and G. Schulz-Ekloff, Chem. ing. Techn., 1980, 52, 734. 13 D. J. Elliott and J. H. Lunsford, J . Catal., 1979, 57, 11. 14 D. Exner, N. Jaeger, R. Nowak, H. Schriibbers and G. Schulz-Ekloff, J . Catal., 1982, 74, 188. 15 D. Exner, N. Jaeger, K. Moller, R. Nowak, H. Schriibbers, G. Schulz-Ekloff and P. L. Ryder, in Metal 16 F. Schmidt, in Metal Microstructures in Zeolites, ed. P. A. Jacobs et al. (Elsevier, Amsterdam, 1982), 17 D. Exner, N. I. Jaeger, R. Nowak, G. Schulz-Ekloff and P. L. Ryder, in Proceedzngs cf the 6th 18 P. Gallezot, in Proceedings of the 6th International Zeolite Conference, ed. D. Olson and A. Bisio 19 L. A. Bursill, J. M. Thomas and K. J. Rao, Nature (London), 1981, 289, 157. 20 J. M. Thomas, G. R. Millward, S. Ramdas And M. Audier, ASC Syrnp. Ser., 218, 1983, p. 181. 21 W. J. Reagan, A. W. Chester and G. T. Kerr, J . Catal., 1981, 69, 89. 22 D. Exner, N. Jaeger, K. Moller and G. Schulz-Ekloff, J . Chem. SOC., Faraday Trans. 1, 1982,78, 3537. 23 G. Schulz-Ekloff, D. Wright and M. Grunze, Zeolites, 1982, 2, 70. 24 P. Gallezot, A. 1. Bienenstock and M. Boudart, Now. J. Chim.. 1978, 2 , 263 25 P. Gallezot, Zeolites, 1982, 2, 103. Microstructures in Zeolites, ed. P. A. Jacobs et al. (Elsevier, Amsterdam, 1982), p. 205. p. 197. Zntrrnafional Zeolite Conference, ed. D. Olson and A. Bisio (Butterworths, Guildford, 1984), p. 387. (Butterworths, Guildford, 1984), p. 387. Puprr 51290; Remired 20th February, 1985

ISSN:0300-9599    

DOI:10.1039/F19868200205

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

J . Chem. SOC., Furaday Trans. 1, 1986, 82, 213-232 Influence of Cocations on the Location and Coordination Geometry of Cu2" in M+A Zeolites (M+ = Na, K, T1, Rb, Cs, NH, and CH,NH,) An Electron Spin Resonance and Electron Spin-Echo Modulation Spectroscopic Study Mysore Narayana and Larry Kevan" Department of Chemistry, University of Houston, Houston, Texas 77004, U.S. A . The changes in the locations and coordination geometries of Cu2+ in various cationic forms of A zeolite has been examined by electron spin resonance and electron spin-echo modulation techniques. It is observed that Cu?+ interacts with six deuterons (three water molecules) in NaA in the x-cages, while it interacts with only two deuterons (one water molecule) in the w a g e in the other monovalent ion-exchanged A zeolites.The changes in the coordination of Cu2+ as a function of hydration conditions have been monitored and an anomalous irreversible change in the immediate environ- ment of Cu2+ in KA and TlA is observed. After dehydration at T > 353 K in KA and T1A rehydration does not restore the spectra seen in fresh samples. Instead the spectra in such rehydrated samples are identical to those in NaA. It is also found that on ammonia-exchange and partial dehydration Cu2+ exhibits markedly different behaviour in NaA when compared with exchange in NaA or in NH,A zeolites. The difference in the initial Cu2+ coordination in NaA as against the other A zeolites is explained in terms of a-cage crowding by the monovalent cations. ~ ~ ~~~ Because of the efficiency exhibited by copper-exchanged zeolites in several catalyst- promoted reactions such as oxidation,l cracking2 and is~merization,~ numerous studies have been made to characterize the active species in these zeolites.A prerequisite for such characterization of Cu2+ is a detailed knowledge of its location and immediate environment. Several electron spin resonance (e.s.r.) reports have been published about the behaviour of Cu2+ in ~eolites.~-* Except for an indication that Cu2+ ions are distributed over several sites in the zeolite, e.s.r. has not been successful in unambiguously determining details of the immediate environment in these polycrystalline samples. In recent years the pulsed magnetic resonance technique, electron spin-echo modulation (e.s.e.m.) spectroscopy has been successfully exploited9-13 to garner critical information about short-range order around paramagnetic species which cannot be obtained from continuous-wave e.s.r. measurements.In particular. interactions of cupric ion with various adsorbates in activated NaA zeolite have been e1ucidated.l4 From Mossbauer studies Dickson and Rees15 concluded that in partially hydrated Fez+-exchanged A zeolites the metal ion is in tetrahedral coordination. We have been able to show by analysis of 133Cs modulation16 and deuterium mod~lationl~ for Cuz+ in several A zeolites that Cu2+ exists in an unusual tetrahedral coordination even in fully hydrated zeolite samples. Pafomov et aI.l* studied the e.s.r. spectra of Cu2+ in A zeolites with different cocations but were unable to draw any conclusions about the site location or coordination geometry of the majority of the observed Cu2+ species.In this paper we present results of e.s.r. and e.s.e.m. studies on Cu2+ under various 213214 Coordination Geometry of Cu2+ in M+A Zeolites hydration conditions with different cocations in A zeolites. It is clearly demonstrated that the bulkier cations force severe changes in the coordination sphere of Cuz+ in these A zeolites. Experiment a1 Linde 4A zeolite was washed with 0.1 mol dmW3 sodium carbonate solutions and ion-exchanged with 0.05 mol dm-, solutions of T1+, Cs+, Rb+, NH,t and CH,NH: with either NO; or C1- as the counterion to obtain Tllo.5 Na,.,-A, Cs,Na,-A, Rb,,,Na,-A and (NH,),,-A. The numbers are with reference to the total number of exchangeable cations in the primitive unit cell of A zeolite and are obtained by atomic absorption analysis.The actual number of methylammonium cations has not been analysed, but from the literature it is known that at best only ca. 77% of the sodium ions can be replaced by CH3NH$.l9 Linde 3A was washed with 0.1 mol dm-, potassium carbonate solutions to obtain K,,-A. These zeolites will be referred to as NaA, TlA, CsA, RbA, NH,A and CH,NH,A. Ca. 0.3% of Cu2+ was exchanged into these zeolites at room temperature using lop4 mol dm-, cupric nitrate solutions at a pH of 5.2-5.5. Several samples of NaA were exchanged with ammoniacal copper nitrate solution ( mol dm-, pH 1 1) which yielded pale blue samples, indicating some complexation of Cu2+ with NH,. This colour disappeared on leaving the zeolite-in air (relative humidity 60%) for 1 day.Such ammoniacally exchanged samples will be referred to as Cu(NH,),-NaA. For the various e.s.r. and e.s.e.m. experiments samples were subjected to the following treatments. (1) For studying the coordination changes by e.s.r., zeolite samples were loaded in 3 mm 0.d. Suprasil quartz tubes and evacuated in a vacuum line to a residual pressure of 2 x lop5 Torrt for various time intervals at 293 (room temperature), 323,373,673 and 773 K, and sealed at that temperature. ( 2 ) Samples heated at any of the above-mentioned temperatures were exposed to H,O or D,O vapour (ca. 25 Torr at room temperature) for 4 7 h before sealing. For samples treated at 673 and 773 K oxidation with 400 Torr of oxygen for 3 h was necessary to regenerate all of the Cuz+, as part of it was reduced on dehydration at such temperatures. Considerable difficulty was experienced in preparing rehydrated Cu-RbA samples.When RbCl was used for ion exchange some of the samples lost crystallinity on dehydration at 773 K.,O Also, the RbA zeolite turned out to be an excellent scavenger of Mn2+ in the RbCl solutions, which interferes considerably with both e.s.r. and e.s.e.m. experiments. None of the other cationic forms (Na, K, T1, Cs) showed either loss of crystallinity or specific scavenging ability for Mn2+. (3) For e.s.e.m. measurements samples were soaked in D,O and dried in a vacuum desiccator; this was repeated three times to ensure complete deuteration. Then the samples were sealed in 3 mm 0.d.quartz tubes. Some samples were deuterated by exposing them to D,O after evacuation in a vacuum line at different temperatures. (4) KA and T1A zeolites were also prepared by using the respective acetate solutions from NaA, as some anomalous results were observed in the dehydration-rehydration cycle. These results were reproducible, irrespective of the type of exchange solution used indicating different behaviour of these two zeolites compared with the others. This point will be elaborated in a later section. E.s.r. measurements at room temperature or at 77 K were made with a modified Varian E-4 spectrometer. E.s.e.m. spectra at 4.2 K were recorded with a home-built spectrometer considerably modified from the earlier reported version.21 Some of these modifications have been already published elsewhere.22 A Tracor Northern signal averager (TN17 10 or TN 1550) was modified to start the scan of the pulse intervals in the auto mode and also to drive a 0-180" phase shifter at the input of the high-power microwave amplifier.Phase-shifting the microwave pulse on every alternate scan by 180" cancelled out the base- line drift and also reduced the dead time of the spectrometer.22b The other major t 1 Torr = 101 325/760 Pa.M . Narayana and L. Kevan 215 additions were a double balanced mixer and an Ortec 726 timing amplifier on the receiver side, both of which considerably improved the signal-to-noise ratio. Finally the digitized data were transferred from the signal averager to a Tektronix 4052 microcomputer system, where they were processed for plotting in the time domain and could also be fast- Fourier-transformed using a Tektronix program. Theory and Analysis The theory and analysis of electron spin echoes and the nuclear modulations on the echo decay envelopes is well do~umented.~-ll When a paramagnetic system is subjected to resonant microwave pulses in suitable sequence, microwave echoes are generated due to reformation of macroscopic magnetization. When the time intervals between the microwave pulses are swept these echoes decay due to various relaxation processes.The weak dipolar, isotropic hyperfine and quadrupolar interactions experienced by the un- paired spin with the surrounding magnetic nuclei often show up as periodic variations in the echo decay envelope.These modulations are characteristic of the Larmor frequencies of the interacting nuclei and thus can be quantitatively analysed to give information about the radial distribution of the interacting nuclei, their number and the contact hyperfine coupling when present. l3 Deuterium nuclei have a slower precession frequency, and the modulation from them can be more accurately analysed than that from protons. Thus, in most of the samples used in this study H,O has been exchanged with D20. In fact, when adsorbate molecule geometry is the object of the study one can selectively deuterate a molecule like CH,OH to determine accurately the orientation of the with respect to the unpaired spin. Recently generalized expressions in the zero quadrupole interaction approximation were obtained by Dikanov et al.23 for two- and three-pulse electron spin-echo modulation.These expressions can be briefly summarized as follows for three-pulse echo experiments. For an unpaired spin with S = 1/2 interacting with a nucleus of spin I = 1/2 and for a nucleus of spin I where a and P are the electron up and down states. The final expression for the stimulation echo amplitude is J q T , T ) = 1/2[vy(r, T)+ Vf(Z, T ) ] . In these expressions K = ( C O ~ B/w, COD)^ C O ~ , ~ = [(CO~ f A/2)’+ B2/4]”’ A = D(3 cos20- 1)+2na B = 3 0 sin0 C O S ~ D = ggn PPn/fir3* If the unpaired spin is interacting with N nuclei, the overall modulation is given by21 6 Coordination Geometry of Cu2+ in M+A Zeolites To consider the interactions with N equivalent nuclei with uncorrela use the methodology suggested by Mim~.~y lo s < w, T))# = n < I / I k ( Z , T)n.k=l In these expressions Y is the electron-nucleus vector making an angl ed orientations we ! 0 to the direction of the external magnetic field Ho and a is the isotropic hyperfine coupling. The basic approximations made in obtaining and using these expressions are ( I ) the local dipolar interaction is smaller than the nuclear Zeeman terms, (2) the nuclear quadrupolar interaction is smaller than the dipolar term, (3) the unpaired electron spin is sufficiently localized to fit a point-dipole description and (4) the nuclei are considered to be uncorrelated as far as the interacting electron spin is concerned, so that spherical averaging can be used. In most cases these are all excellent approximations, except when the interacting nucleus has a strong nuclear quadrupole moment.In such a case even with fairly regular symmetry at the site of the nucleus one has the quadrupolar interaction almost of the same order as the dipolar term, and the Hamiltonian will have to be exactly diagonalized to analyse the modulation rigorously. For example, in a recent two- dimensional n.m.r. study Samoson and LippmaaZ4 have shown that even in hydrated NaA zeolite there is sufficient anisotropy in the electric field gradient at the site of aluminium nuclei to give a quadrupole coupling of 1.1 MHz with an asymmetry parameter of 0.75. The average dipolar interaction in most cases is a few MHz, and thus one can no longer ignore the quadrupolar interaction in such a zeolite sample.However, in principle one can measure strong quadrupole interactions directly when present from the fast Fourier transform of the time-domain electron spin+cho modulation. This has been suggested in a study of the chlorophyll-a cation in glassy systems by Dikanov et aZ.25 In the analysis of the time-domain data the decay inherent in the modulation spectra is fitted to a polynomial and divided out before comparing with the theoretically simulated spectra as a function of: n, the number of equivalent interacting nuclei, Y, their distance to the unpaired electron spin and a, the isotropic hyperfine coupling constant. For good-quality data the errors in determination of the parameters are : n to the nearest integer for n 5 10, Y to kO.01 nm when < ca.0.45 nm and a to f 10%. If a satisfactory fit of at least three different spectra with T as the variable for three-pulse echoes and T as the variable for two-pulse echoes is not obtained for the same set of parameters n, Y and a, then a two-shell model is used in which two groups of inequivalent nuclei are considered to be interacting with the paramagnetic species. Results Cu-NaA Several reports are already available on the e.s.r. spectrum of Cu2+ in NaA zeolite.E However, only recently have the specific characteristic coordination geometries under various hydration conditions been identified.27c In freshly prepared as well as in rehydrated samples there is only one type of Cu2+ present under low loading conditions. Sometimes, if the rehydration time is shortened, one can simultaneously see Cu2+ species coordinated to zero and three water molecules.In fig. 1 the e.s.r. of spectrum of one such rehydrated sample is shown, along with that of a partially dehydrated sample. In the latter a conspicuous feature of the spectrum is a third Cu2+ species with ‘reversed’ g values8 (g,, < gl). The three types of Cu2+ thus far described will be referred to as follows, in which the roman numeral subscript denotes the number of coordinated waters: CUIII for a copper ion coordinating to three zeolite oxygens and three water molecules, CuII for a trigonal bipyramidal copper in the six-membered ring windowsM. Narayana and L. Kevan 217 Y 91' c urn) Fig. 1. E.s.r. spectra at 77 K of (a) partially dehydrated and (h) partially rehydrated Cu-NaA.The sharp line at g = 2.004 is a colour centre in the e.s.r. Dewar; (a) was recorded on a sample dried at 383 K in air for 24 h, while (b) was recorded on a sample dehydrated and oxidized at 673 K before exposing it to water vapour for 2 h at room temperature. coordinated to three zeolite oxygen atoms and two water molecules, and Cu, for the copper species in the fully dehydrated zeolite coordinating to only three six-membered-ring oxygen atoms. CU,,, is partly converted into CUII on evacuating Cu-NaA samples for a prolonged time at ambient temperature or by drying in air at 383 K for 2 h. The most interesting feature of this CUII is that it cannot be easily dehydrated further to form Cu, or rehydrated to form CUIII.Prolonged exposure of the partially dehydrated sample to water vapour does not restore CUI,, as the sole copper species in the zeolite; this can only be achieved by evacuation at T 2 383 K and then rehydration at room temperature. CU-KA, Cu-TlA Unlike in NaA, freshly prepared Cu-KA samples exhibit a completely different dominant copper species. Using e.s.e.m. analysis of the deuterated samples this has been identified by us1' as Cu2+ coordinating to only one water molecule. This will be referred to as Cu,. In the fresh samples of Cu-KA a small amount of CUIII is always present, but it disappears if the sample is evacuated at 323-353 K and then rehydrated. Upon such a treatment Cu, reforms as the sole Cu2+ species in the zeolite. Note that after 2 h of evacuation at room temperature Cu, completely disappears.It is converted into CUII, the g and A values of which are identical to those of CuII in NaA. However, if the samples are evacuated at 7' 2 383 K and then rehydrated, there is no trace of Cu,, a species similar to CU,,, being the only copper complex observable by e.s.r. spectroscopy. This is218 Coordination Geometry of Cu2+ in M+A Zeolites 1 77 K I u I * w S p , ~ 200 G , H Fig. 2. Comparison of e.s.r. spectra at 293 and 77 K of (a) fresh Cu-NaA, (6) fresh Cu-KA and ( c ) Cu-KA rehydrated at room temperature after dehydration at 383 K. Note the formation of Cu,,, instead of Cu, in rehydrated Cu-KA and that only fresh Cu-KA shows different e.s.r. spectra at 77 and 293 K. demonstrated in fig. 2, where the 293 and 77 K e.s.r.spectra of Cu-NaA are compared with those of Cu-KA before and after the activation-rehydration cycle. A similar behaviour of Cu2+ is observed in TlA, the only exception being that in the rehydrated samples Cu, is present in traces while in KA it is no longer present. Fig. 3 shows the e.s.r. spectra of Cu-TlA before and after the activation-rehydration cycle. The anomalous disappearance of Cu, on activation is further elaborated in the next section. In Cu-TlA the reversed g-value spectrum also appears on partial dehydration and has features identical to those in Cu-NaA and in Cu-KA. Cu-RbA and Cu-CsA In both these zeolites the dominant copper complex is CuI.16 In both it disappears on evacuation at room temperature and Cu,, is formed. However, dehydration at any temperature followed by exposure to water vapour completely restores Cu, as the major species.As mentioned in the Experimental section, some of the samples of RbA were irreversibly transformed into amorphous material on dehydration at 773 K. No such problem was encountered with CsA samples. Note that in neither of these was CU,,, observed either in freshly made samples or in rehydrated samples. The e.s.r. parameters of Cu2+ in freshly prepared, variously dehydrated and rehydrated samples of several of these A zeolites are given in table 1 . Note that Cu-RbA has theM . Narayana and L. Kevan 219 1 - 200 G , H Fig. 3. Comparison of e.s.r. spectra at 77 K of (a) freshly prepared Cu-T1A and (b) Cu-T1A rehydrated at room temperature after dehydration at 483 K.For increased clarity the spectra in the gl region are also recorded at higher gain. Note the decrease in the intensity of Cu, in the rehydrated sample associated with a corresponding increase in Cu,,,. highest gI1 value and also the lowest hyperfine coupling among the Cu, species seen in the other A zeolites. It was rather difficult to determine the gl value for Cu, in all these zeolites because of severe overlap of the e.s.r. features of all three species in the gl region of the e.s.r. spectra. In fact, in ref. (17) the gl signal of CUII, was mistakenly identified as that of Cu,. It is unclear whether those spectra are axially symmetric; a rhombic symmetry assignment may also be made. Cu-NH,A and CU-CH,NH,A In NH,A and CH,NH,A the major copper species is CU, as in KA and T1A.If partial dehydration is performed by evacuation between room temperature and 353 K, or by drying in air at 383 K, Cu, is restored as the dominant species on rehydration at room temperature. Partial dehydration produces a ‘reversed’ g-value spectrum (gl > g,,), but with a gl different from that of the reversed g-value spectrum denoted as CuII, which is formed in NaA, KA, T1A and RbA. In the case of NH,A an additional hyperfine splitting is seen in fig. 4 in the g 2.1 region, which is probably due to interaction with two nitrogen nuclei (14N, Z = 1). The appearance of this additional splitting was found t o be completely reversible in Cu-NH,A subjected to several dehydration-rehydration cycles with T > 353 K. On dehydration at higher temperature the e.s.r.signal intensity falls off rapidly and is not restored on rehydration. In CH,NH,A another ‘reversed’ g-value spectrum with yet a different gl is obtained on evacuation at T > 353 K or by drying in air at 383 K. In fact in CH,NH,A two different copper species with ‘reversed’ g values are observed as shown in fig. 5. However, no additional splittings are seen in either of those two species. As in NH,A, the Cu2+ signal intensities in CH,NH,A fall off rapidly when the dehydration temperatures are higher than 353 K, and the Cu2+ signals are not restored on rehydration.220 Coordination Geometry of Cu2+ in M+A Zeolites Table 1. E.s.r. parameter9 of Cu2+ at 77 K in several A zeolites under various hydration conditions zeolite NaA KA T1A RbA NH,A CH,NH,A Cu(NH,),-NaA major species hydration statusb gll g, A1 minor species probable Cu2+ configurations" 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 4 1 2 4' 1 2 4' 2.352 2.352 2.388 2.349 2.479 2.004 2.391 2.345 2.469 2.004 2.394 2.342 2.502 2.004 2.391 2.498 2.474 2.004 2.471 2.49 2.004 2.491 2.309 2.004 1.998 2.067 2.067 2.070 1.069 2.129 2.289 2.065 2.067 2.129 2.293 2.067 2.065 2.130 2.290 2.067 2.134 2.130 2.243 2.130 2.134 2.71 2.130 2.060 2.23 1 2.289 159 158 131 133 100 84 133 168 103 83 130 171 90 85 130 90 102 132 102 95 149 95 178 126 93 - 2.004 - - 2.345 2.349 - - 2.352 2.346 2.469 2.350 - - - ~ 2.022 2.3 17 2.022 2.024 1.959 2.020 - - 2.350 - 2.289 __ - 2.070 2.070 - - - 2.060 2.130 2.070 ~~ - - ~~ 2.270 2.060 2.280 2.289 2.23 2.292 - - 2.070 - 84 - - 166 154 - - 163 165 103 158 - -~ - - 102 161 102 134 105 131 - ~.15 a Errors in g values are k0.003 and in A values are 1 x lo-* crn-l; units of A are 1 x lo-, ern-'. Hydration status numbers mean the following: (1) freshly prepared samples; (2) partially dehydrated, either by evacuation at room temperature or by drying in air at 383 K; (3) fully dehydrated at 673 K and oxidized; (4) rehydrated after (3); (4') rehydrated after (2). The subscripts for the configuation of Cu2+ have following meaning: Cu,,, means coordinated to three waters, CUI, to two waters, Cu, to one water, and Cu, not coordinated to any water. Cu,,. represents the several new reversed g-value spectra which are not identical to CU,,. CUII- refers to coordination to two hydroxyls in place of two waters.Cu(NH,),NaA When Cuz+ is ammoniacally exchanged into NaA zeolite none of the copper species, CU~II, Cu,, or Cu,, seen in the freshly prepared samples of other A zeolites is observed. Instead a new Cu, species with gll z 2.31, gl = 2.06 and a large copper hyperfine splitting of 178 x lop4 cm-l is the major copper species, as shown in fig. 6. While no explicit splitting due to interaction with nitrogen nuclei is seen, all the hyperfine lines are considerably broader than those of CUIII, CUII or Cu,, which suggests some unresolved superhyperfine coupling. Also, the low value of gll is indicative of ammonia or an ammonium ion being in the first coordination sphere of C U ~ + . ~ ~ Any dehydration treatment of these ammoniacally treated samples was irreversible.In fig. 6 the e.s.r. spectra of freshly prepared Cu(NH,),-NaA and those in partially dehydrated and rehydrated samples are shown. A ' reversed ' g-value spectrum dominates on evacuation at room temperature or higher, but it is different from the spectrum assigned to CUII as well as from the reversed g-value spectra seen in partially dehydratedM . Narayana and L. Kevan 22 1 Fig. 4. E.s.r. spectra at 293 and 77 K of Cu-NH,A: (a) 29.3 K spectrum of fresh sample and (h) 77 K spectrum of fresh sample. (The e.s.r. spectra in Cu-NH,A rehydrated at T < 343 K are identical to this.) (c) 77 K spectrum of Cu-NH,A air dried at 383 K for 48 h. If this sample is exposed to air or water vapour (h) is restored completely. Fig. 5. E.s.r. spectra (a) at 293 and 77 K of fresh Cu-CH,NH,A and (h) at 77 K of Cu-CH,NH,A dried in air at 383 K for 48 h.The spectra seen in (a) are completely restored if sample (b) is exposed to water vapour for 3 h at room temperature.222 Coordination Geometry of Cu2+ in MSA Zeolites s,c c urn 1 Fig. 6. E.s.r. spectra of Cu(NH,), in NaA at 77 K: (a) fresh sample, (b) after the sample has been evacuated at 293 K for 6 h, (c) after drying the fresh sample at 383 K for 24 h, ( d ) when samples (b) or (c) are exposed to air or water vapour for 3 h at room temperature. Note the formation of Cu,,, in (a) while almost no Cu,,, is observable in (a). Cu-NH,A and Cu-CH,NH,A. On rehydration another reversed g-value spectrum is formed which looks like the Cu,, species, although its A , , value is a little larger.In fact this reversed g-value species is similar to that observed in C U - C ~ X ~ ~ ~ and C U - C ~ X ~ ' ~ in which hydroxyls have replaced water as the two apical ligands2* The other species in the rehydrated sample is CuIII, which was not seen in the fresh samples of Cu(NH,),-NaA. Fig. 7 shows three-pulse e.s.e.m. spectra of fresh and rehydrated ( T = 383 K) Cu-KA samples prepared in D,O which clearly reflect the differences observed in their e.s.r spectra. In fig. 8 the three-pulse e.s.e.m. spectrum of Cu-RbA exchanged with D,O is shown. This is almost identical to the spectra observed in fresh samples of Cu-KA17 or Cu-TlA exchanged in D,O. The analysis of the e.s.e.m. data indicates that Cu2+ interacts with only two deuterons at 0.26 nm with a small isotropic coupling of 0.2 MHz.In contrast, analysis of the e.s.e.m. spectra in the rehydrated ( T = 383 K) samples of Cu-KA or Cu-TlA in fig. 9 indicates that Cu2+ interacts with six deuterons at 0.28 nm with an isotropic coupling of 0.2 MHz, which is identical to the results in either fresh or rehydrated samples of Cu-NaA. Thus both the e.s.r. and e.s.e.m. spectra clearly show some irreversible rearrangement of Cu2+ surroundings in KA and TlA on dehydration at T > 353 K.M. Narayana and L. Kevan 223 Fig. 7. Comparison of three-pulse e.s.e.m. spectra of fresh Cu-KA (---) and Cu-KA rehydrated at room temperature after dehydration at 383 K (-). Note the difference in the depths of deuterium modulation of the two samples; this indicates an increase in the number of interacting deuterons in the rehydrated sample. The spectra were recorded at H, = 3080 G with a first interpulse time z = 0.25 ps.I I I I I I I I I 1 I 2 3 4 5 TIP Fig. 8. (---) Experimental and (-) calculated three-pulse e.s.e.m. spectra of fresh Cu-RbA. The spectrum was recorded at H, = 3040 G with a first interpulse time, 7 = 0.26 ps. The decay function used was G(T) = exp (2.5 -0.3 T+ 0.026 T2). n = 2, Y = 0.26 nm and a = 0.15 MHz. 8 F A R 1224 Coordination Geometry of Cu2+ in M+A Zeolites I I I I I I I I I 1 I 2 3 4 5 TIPS Fig. 9. (---) Experimental and (-) calculated three-pulse e.s.e.m. spectra of Cu-KA rehydrated at room temperature after dehydration at 383 K. This spectrum was recorded at H, = 3 180 G with a first interpulse time, z = 0.25 ps.The decay function used was G(T) = exp (2.7 - 0.11 T). n = 6, Y = 0.28 nm and a = 0.2 MHz. Fig. 10 shows frequency-domain spectra obtained by fast Fourier transformation (f.f.t.) of the time-domain e.s.e.m. spectra of fresh and rehydrated (383 K) Cu-KA. Both these spectra have one strong peak near 2 MHz and another weak peak at ca. 4 MHz. The free-precession frequency of deuterons in a magnetic field of ca. 3180 G is ca. 2.1 MHz, which accounts for the main peak in the f.f.t. spectra. The origin of the weak peak at 4 MHz is unclear at this time. All the simulations done with such large hyperfine couplings yield very poor fits to the experimental time-domain spectra. Samples of Cu(ND,),-NaA/D,O were prepared with ND, and/or with D,O. We observed identical deuterium modulation for both preparations.The three-pulse e.s.e.m. spectrum and its f.f.t. of these samples are shown in fig. 11. Good fits to the time-domain spectra could not be obtained with one or two shells of equivalent interacting deuterons. The f.f.t. clearly shows that Cu2+ is interacting with at least two types of deuterons, with one type having an isotropic coupling of ca. 0.5 MHz or greater.13 This failure to obtain a good fit to the time-domain data with one or two shells of equivalent deuterons suggests that either the point-dipole approximation has failed owing to large delocalization of the unpaired spin into the ligands or that all the ligands of Cu2+ are completely inequivalent. In fig. 12 two-pulse e.s.e.m. spectra of freshly prepared Cu-NaA/H,O and fully dehydrated Cu-NaA are compared.The considerable damping of the modulation in the dehydrated sample is probably due to stronger quadrupolar interactions caused by stronger electric field gradients at the sites of nuclei. Part of the reason for such an increase in the electric field gradient could be the severe distortion of the six-membered ring caused by Cu2+ movement off the trigonal axis as a result of Jahn-Teller dist~rtion.,~ Fig. 13 compares the f.f.t. spectra obtained from the two-pulse e.s.e.m. spectra ofCu-NaA/H,O and Cu-RbA/H,O. Both the spectra have three well defined peaks above the noise level, at ca. 3.5, 7 and 13 MHz. The first two peaks correspond well toM . Narayana and L. Kevun 225 5 10 15 20 frequency/ M Hz Fig. 10.Comparison of the frequency-domain spectra obtained by fast Fourier transformation of the time-domain three-pulse e.s.e.m. spectra of (a) fresh Cu-KA and (b) Cu-KA rehydrated at room temperature after dehydration at 383 K. The spectra were recorded at H,, = 3180 G with a first interpulse time, z = 0.29 pus. 10 8 2 frequency/MHz I I I I I 0 I 2 3 4 5 TIPS Fig. 11. (a) Three-pulse e.s.e.m. spectra of Cu(NH,),-NaA prepared in D,O and (b) its f.f.t. The spectrum was recorded at H,, = 3150 G with a first interpulse time, z = 0.25 ps. Note the clean resolution of two peaks, which is indicative of two types of deuterons, one with 0.5 MHz isotropic hyperfine coupling and the other with no isotropic coupling. 8-2226 Coordination Geometry of Cu2+ in M+A Zeolites 8 1, Fig.12. Comparison of two-pulse e.s.e.m. spectra of (a) fresh Cu-NaA and (b) fully dehydrated Cu-NaA showing 27Al modulation. Note the drastic decrease in the 27Al modulation amplitudes, presumably due to increase in the nuclear quadrupole interactions. the expected positions of the free-precession 27Al frequency and its second harmonic. The third corresponds well to the expected position of protons. The proton peak is considerably stronger for the Cu-NaA sample. This difference in the proton peak intensities in Cu-NaA as against Cu-RbA is authentic, as verified by recording e.s.e.m. two-pulse spectra at a variety of magnetic-field positions and using several sets of new samples. Such a difference is also consistent with the picture obtained by analysis of deuterium modulations in the samples of Cu-NaA/D,O and Cu-RbA/D,O, the former showing an interaction of Cu2+ with six deuterons while the latter shows only two.Analysis of the aluminium modulations such as shown in fig. 12(a) in the zero quadrupolar interaction approximation indicates a Cu2+ interaction with three aluminium nuclei at a distance of 0.36-0.38 nm for CUIII as well as for Cu,. However, these distances are probably shorter owing to modulation damping by quadrupolar 31 E.s.e.m. spectra of the dehydrated samples could not be analysed owing to a lack of sufficient modulation depths. Discussion We have already established27c via analyses of e.s.e.m spectra that in fresh and rehydrated samples of Cu-NaA, Cu2+ is coordinated to three water molecules and is located at site S2* in the a-cages; this Cu2+ species is denoted as CUIII.The cation site locations are shown in fig. 14. On partial dehydration part of the intensity of this complex is lost and a new spectrum with reversed g values (g,, < gl) appears which is denoted CUII. It is wellM . Narayana and L. Kevan 1-3.19 2 1-3.38 17.05 113.60 5 10 15 20 0 frequency/MHz 227 Fig. 13. Comparison of f.f.t. spectra of two-pulse e.s.e.m. spectra in (a) fresh Cu-RbA and (b) fresh Cu-NaA. The spectrum of Cu-RbA was recorded at Ho = 3030 G and that of Cu-NaA at H , = 3165 G. Note the decrease in the intensity of the proton peak. Fig. 14. Site nomenclature in A zeolite. Site S2* projects into the a-cage above a six-membered ring window while S2' projects into the Q-cage.S2 is in the plane of the 0, oxygens of the six ring. S5 is in the plane of the eight-membered ring window of the a-cage, while S3 is an ill-defined site in front of the four-membered rings in the a-cage. known8tZ6 that for Cu2+ to exhibit such reversed g values the )3z2-r2) state should be the ground state for the unpaired spin which occurs in the following geometries: (a) compressed tetrahedral, (b) compressed tetragonally or rhombically distorted octahedral, (c) cis distorted octahedral, ( d ) compressed square pyramidal and (e) trigonal bipyramidal. E.s.e.m. analysis of the Cu2+ species in partially dehydrated Cu-NaA indicates that228 Coordination Geometry of Cu2+ in M+A Zeolites Cu2+ interacts with four deuterons (consistent with two water molecules) thus indicating the trigonal-bipyramidal configuration to be most probable.It should be emphasized that on exposure of water vapour to such a partially dchydrated sample Cu,, is not converted into CUIII. This can be achieved only by complete dehydration at higher temperatures followed by rehydration. The specific reasons for the stability of CUII are not known at this stage. On complete dehydration deuterium modulation can no longer be used as a tool to monitor the location of Cu2+, and the aluminium modulations are also considerably damped, as shown in fig. 12, presumably through increased nuclear quadrupolar interactions. An indirect conclusion from such a damping is that Cu2+ causes distortions in the six-membered ring on dehydration, partly owing to the Jahn-Teller instability, which forces Cu2+ off the trigonal axis and removes the ground-state degeneracy.Analysis of the aluminium modulation in the zero quadrupole interaction approximation for the fresh or rehydrated Cu-NaA samples shows that Cu2+ interacts with three aluminium nuclei at 0.38 nm. Such a result places Cu2+ cu. 0.2 nm above the plane of the 0, oxygens in the six-membered ring windows. However, the actual distance is probably shorter, since quadrupole interactions for 27Al cannot be completely While expressions incorporating quadrupole interaction using a first-order perturbation approach are available, we have shown3’ that such an approach incorporates approximations in the primary dipolar interaction which appear to largely defeat the purpose of the apparently more rigorous analysis.In Cu-KA and Cu-TlA fresh samples the major species is conspicuously different from the Cu,,, species found in Cu-NaA. By analysis of deuterium modulations this species was shown to be interacting with one water and can be denoted Cu,. Another important difference between Cu-NaA and Cu-KA or Cu-T1A is that the reversible formation of Cu, in the dehydration-rehydration cycle is dependent on the dehydration temperature. In Cu-NaA, CUII, is reformed on rehydration following dehydration at any temperature between 373 and 773 K. In Cu-KA and Cu-TlA, Cu, is irretrievably lost if the dehydration is carried out at T > 353 K. Instead, rehydration in both these zeolites results in the formation of CUIII as the major species. In both Cu-KA and Cu-TlA room-temperature evacuation is sufficient to completely destroy Cu,.A strong ‘reversed’ g-value spectrum identical to that in partially dehydrated Cu-NaA; hence Cu,, appears. However, unlike the case in Cu-NaA, on exposure to water vapour CUII disappears in these partially dehydrated Cu-KA and Cu-T1A samples and Cu, is restored as the major species. It is clearly seen in fig. 7 that Cu2+ interacts with more deuterons in a rehydrated ( T > 353 K) than in a fresh sample; the e.s.e.m. results for the rehydrated sample in fig. 9 indicate six interacting deuterons, which translate into three water molecules as expected for CuIr1. Note that such a drastic change takes place only for samples rehydrated after dehydration above 353 K. Since very little Cu,,, is seen in fresh samples of Cu-KA, it is clearly not the lack of water molecules that prevents CUI~I from being the dominant species in the fresh samples or in samples dehydrated at temperatures below 353 K and then rehydrated at room temperature.When bulkier cations are present in the a-cage, such as K+, NH;, Cs+, Rb+ and T1+, they crowd the a-cage considerably compared to the case of NaA. For example, in hydrated KA the eight K+ associated with the six-membered windows project ca. 0.15 nm into the a-cage above the plane of the 0, oxygens, while the corresponding displacement of Na+ from such a plane in NaA is only ca. 0.05 nm. Thus in freshly prepared or rehydrated samples of NaA, a Cu2+ at an S2* site does not experience much electrostatic repulsion from the two adjacent Na+ which are at S2 sites.However, in Cu-KA Cu has to compete with K+ for the S2* positions and will experience more electrostatic repulsion from the two adjacent K+, which are at S2* sites. This repulsion causes Cu2+ in S2* sites to be coordinated to only one water molecule in fresh CU-KA.~~ This kind of repulsion in the a-cages becomes more severe for Cu2+ in NH,A,,, CsA, RbA and CH,NH,A. Consequently in all these zeolites Cu2+M . Narayana and L. Kecan 229 is forced to occupy S2* positions, thus explaining the occurrence of Cur as the dominant or only species. On dehydration at high temperatures it has been established that some of the K+ ions move into the P-cage.,,9 35 No crystal-structure data are available for rehydrated KA, but it is possible that these K+ ions cannot move back into the a-cages.Thus if one or more of the K+ ions at S2* sites were to be irreversibly pushed into the P-cages in the dehydration-rehydration cycles the electrostatic repulsion for Cu2+ in the a-cages is reduced, possibly promoting the formation of CUIII at S2* sites. That Cu, reversibly returns as the dominant species when the dehydration-rehydration cycle is carried out at milder temperatures could indicate an energy barrier for K+ to go through the six-ring windows, since the ionic radius of K+ and the opening radius or a six-ring window are comparable (K+ ionic radius = 0.133 nm and the average six-ring opening radius is 0.1 1-0.13 nm). A corresponding analogy cannot be extended to explain the irreversible decrease of Cu, in T1A zeolite.Riley et aZ.36 studied the structure of Tl,,Na,A. Both in the hydrated and dehydrated forms they placed 7 TI+ ions projecting ca. 0.15 nm into the a-cages above the 0, plane of the six rings and one Tl+ recessed into the a-cage ca. 0.17 nm below the 0, plane. The position of Na+ was not given. Thus no T1+ ions were found to move into the P-cages on dehydration. Thus a better explanation may be that CU~II is formed in S2' on r e h y d r a t i ~ n . ~ ~ We have already shown that in Cu-CsA Cu, is the dominant species in fresh samples.16 In Cu-CsA the e.s.e.m. of 135Cs modulations have been reinterpreted from ref. (16) to indicate that Cur is most probably in the a-~age.,~ Since the modulation frequencies of deuterium and caesium are of the same order, it was not possible to obtain unambiguously the number of water molecules from an e.s.e.m.analysis of deuterium interactions in this zeolite. This difficulty was not a problem in Cu-RbA zeolite. The magnetic isotopes *"b and s7Rb have strong quadrupole moments and their Larmor frequencies are considerably different from that of deuterium. No modulations assignable to these Rb nuclei were observed in hydrated or dehydrated samples. Fig. 8 is typical of the e.s.e.m. results we have obtained for deuterated Cu-RbA, which unambiguously indictates that Cu2+ in this zeolite interacts only with two deuterons both in fresh as well as in rehydrated samples, irrespective of the temperature of dehydration. In Cu-NH,A and Cu-CH,NH,A Cu, is the dominant species, but we did not carry out any deuterium e.s.e.m.studies because of the ease with which the NH, protons exchange with the deuterons; one cannot easily distinguish the interactions of water deuterons from NH, deuterons. This was also the situation in e.s.e.m. studies of Cu(NH,),-NaA. The e.s.r. spectrum in partially dehydrated samples of Cu-NH,A clearly indicates that some NH,+ ions are quite close to Cu2+, as reflected by the nitrogen superhyperfine splitting. However, the two-pulse aluminium modulations recorded in these two samples were identical to these in fresh Cu-RbA and in fresh Cu-KA, thus indicating the same S2' location for Cu2+ in all these zeolites. It is not clear why the reversed ' g-value species observed in partially dehydrated Cu-NH,A and Cu-CH,NH,A are different from one another and different from the Cu,, observed in other A zeolites.One possibility could be that in the trigonal-bipyramidal configuration the ligand in the a-cage in these two zeolites is not a water molecule but the respective cation, NH; or CH,NHi. In Cu(NH,),-NaA we could not arrive at a satisfactory model for the Cu2+ deuterium interactions of Cu,. As shown in fig. 9 (b), the frequency-domain spectra clearly indicate that there are at least two types of deuterons in the Cu2+ coordination sphere. The e.s.r. g-values indicate that this Cu2+ is probably in square-planar or distorted-octahedral coordination with some NH, groups in the first coordination sphere, since the gll value of Cu2+ decreases when Cu-N bonds are present compared with g values of Cu2+ with only oxygens as ligands.26 On increasing the copper concentration in the exchange230 Coordination Geometry of Cu2+ in M+A Zeolites solution we were not able to see any clean resolution of nitrogen superhyperfine splitting, but the spectral features were considerably broader than in other A zeolites.On partial dehydration this copper species rapidly loses intensity and the sample changes from pale blue to white, indicating loss of NH, ligands. The irreversible nature of this loss as observed by formation of CUIII on rehydration indicates that NH, is involved in the coordination sphere of this Cu, species. An interesting feature in the rehydrated Cu(NH,),-NaA samples was the appearance of a well resolved ' reversed' g-value spectrum which is not similar to other reversedg-value species seen in any of the A zeolites but is almost identical to that seen in Cu-CaX2ia and Cu-CdX.2ib In both these X zeolites we were able to identify Cu2+ to be in trigonal- bipyramidal coordination in six-ring windows coordinating axially to two hydroxyls, one each in the a- and /&cages, respectively. Lee et from their crystallographic studies came to the conclusion that in partially dehydrated Cu(NH,),-NaA the major copper species is trigonal-bipyramidally coordinated in the six-ring windows with two hydroxyls as the axial ligands. However, they were not able to see CUII,, which e.s.r.and e.s.e.m. show to be present in these samples on rehydration. We have not attempted an analysis of deuterium e.s.e.m.spectra in this sample because we cannot distinguish deuterium between ND, and D,O ligands in Cu(NH,),-NaA. A few words are in order regarding fig. 13, where the fast Fourier transform of the time-domain data for the aluminium modulations is shown. The nuclear Zeeman frequency for 27Al in a field of ca. 3150 G is 3.88 MHz, and the main peak in the f.f.t. spectra is close to this frequency. Thus the nuclear quadrupole interaction in these zeolites is not strong enough to shift significantly the ENDOR energy levels and cause new frequencies, as has been observed in the chlorophyll cation.25 However, the quadrupole interaction is not negligible24 compared with the dipolar interaction because there is a sharp decrease in the intensity of the second harmonic of the 27A1 peak in fig.13. Conclusions E.s.r. and e.s.e,m. studies have been carried out for fresh, various vacuum treated, rehydrated and deuterated samples of Cu-NaA, Cu-KA, Cu-TlA, Cu-CsA, Cu-RbA, Cu-NH,A, Cu-CH,NH,A and Cu(NH,),-NaA zeolites. Three major copper species are identified and characterized from their deuterium and aluminium modulations. In fresh and rehydrated Cu-NaA only one type of Cu2+ exists, CUIII, coordinating to three water molecules at site S2* in the a-cage. On partial dehydration two of the water molecules in the a-cage are lost, Cu2+ moves into the plane of the 0, oxygen atoms in the six ring and interacts with another water in the P-cage. This complex is not reconverted into Cu,,, on exposure to water vapour. This CUII is coordinated to two water molecules and can be distinguished from the trigonal bipyramidal complex coordinated to two hydroxyls characterized by e.s.e.m.analysis in Cu-CaX and Cu-CciX.2i Unlike in NaA, the major copper species in all the other A zeolites studied has a very high gll value. This is assigned to a tetrahedrally coordinated Cu2+ in the a-cages at site S2* and is substantiated by e.s.e.m. analysis of Cu-Cs interactions in CU-CSA.,~ Deuterium modulation analysis for this species in Cu-KA, Cu-TlA and Cu-RbA clearly shows that Cu2+ interacts with only two deuterons, corresponding to one water molecule. On partial dehydration Cu2+ moves from S2* to S2 and interacts with another water molecule in the a-cage forming trigonal-bipyramidal coordination identical to that in Cu-NaA. However, unlike in NaA, on exposure to water vapour CUII is readily converted into Cu, in Cu-KA, Cu-TlA, Cu-RbA and Cu-CsA.This preferential formation of Cu, in these four zeolites at S2* in the a-cage is attributed to the electrostatic repulsion by the bulky cations K+, Tl+, Cs+ and Rb+ in the a-cages. In Cu-KA and Cu-TlA Cu, is irreversibly lost when dehydration is carried out atM . Narayana and L. Kevan 23 1 temperatures > 353 K. Rehydration after such a treatment results in the formation of Cu,,, in both these zeolites, as evidenced by changes in the e.s.r. and e.s.e.m. spectra. In Cu-KA this irreversible change could be due to Cu2+ moving into the P-cages on dehydration and then being rehydrated there. In Cu-CsA and Cu-RbA the loss of Cu, on dehydration is reversible on rehydration, irrespective of the rehydration temperature.In Cu-NH,A and Cu-CH,NH,A the major species is also Cu, in fresh samples. The reversed g-value spectra obtained on partial dehydration in these zeolites are different from that of Cur, observed in the alkali-metal ion zeolites. In Cu-NH,A a clear indication of Cu-N interaction is seen from hyperfine structure in the e.s.r. spectra on partial dehydration. Thus the difference in the reversed g-value spectra in Cu-NH,A and Cu-CH,NH,A as against the other A zeolites could be due to replacement of the a-cage water in the trigonal-bipyramidal configuration by a NH,+ or CH,NH: ion. Cu, is restored as the major species for dehydration below 353 K in both these zeolites. Dehydration above 353 K results both in drastic reduction of the Cu2+ signal intensities and in loss of crystal structure.In NaA zeolite in which copper is ammoniacally exchanged, a new copper complex not seen in the other A zeolites is observed. While we could not quantitatively analyse the deuterium modulation in samples hydrated with D,O, this complex is most likely square planar or a distorted octahedral complex in the a-cages. On mild dehydration and rehydration this complex is lost and Cu,,, and a new reversed g value Cu2+ appear. The latter is identical to the trigonal bipyramidal complex with two apical hydroxyl ligands seen in Cu-CaX and Cu-CdX zeolites. This work was supported by the U.S. National Science Foundation, The Robert A. Welch Foundation and the Energy Laboratory of the University of Houston.References 1 I. Mochita, S. Hoyata, A. Kato and T. Seiyama, J . Catal., 1969, 15, 314. 2 K. Tsutsumi, S. Fuji and H. Takahashi, J . Catal., 1972, 24, 146, 3 C. Dimitrov and H. F. Leach, J. Catal. 1969, 14, 336. 4 I. R. Leith and H. F. Leach, Proc. R . Soc. London, Ser. A , 1972, 330, 247. 5 E. F. Vansant and J. H. Lunsford, J. Phys. Chem., 1972, 76, 2860. 6 R. A. Schoonheydt, P. Peigneur and J. B. Uytterhoeven, J . Chem. Soc., Furaduy Trans. 1, 1978, 74, 7 D. R. Flentge, J. H. Lunsford, P. A. Jacobs and J. B. Uytterhoeven, J . Phys. Chem., 1975, 79, 356. 8 R. Herman, Znorg. Chem., 1979, 18, 995. 9 W. B. Mims, Phys. Rev. B, 1972, 5, 2609. 2550. 10 W. B. Mims, J. Peisach and J. L. Davis, J . Chem. Phys., 1977, 66, 5536. 11 L. Kevan in Time Domain Electron Spin Resonance, L. Kevan and R. N. Schwartz (Wiley-Interscience, New York, 1979) chap. 8. 12 T. Ichikawa, L. Kevan, M. K. Bowman, S. A. Dikanov and Yu. D. Tsvetkov, J . Chem. Phys., 1979, 71, 1167. 13 P. A. Narayana and L. Kevan, J . Magn. Reson., 1982,46, 84. 14 T. Ichikawa, and L. Kevan, J. Am. Chem. Soc., 1981, 103, 5355. 15 B. L. Dickson and L. V. C. Rees, J . Chem. Soc., Furaday Trans. I , 1974. 74. 2038. 16 M. Narayana and L. Kevan, J. Chem. Phys., 1981, 75, 3269. 17 M. Narayana and L. Kevan, J . Phys. C., 1983, 16, 361. 18 N. N. Pafomov, V. A. Silchenko and Yu. A. Bratashevskii, Teor. Eksp. Khim., 1978. 14, 269. 19 R. M. Barrer and W. M. Meier, Trans. Faraday Soc., 1958, 54, 1074. 20 J. J. Pluth and J. V. Smith, J. Am. Chem. Soc., 1983, 105, 2621. 21 T. Ichikawa, L. Kevan and P. A. Narayana, J. Phys. Chem., 1979. 83, 3378. 22 (a) P. A. Narayana and L. Kevan, Photochem. Photobiol., 1983, 37, 105; (b) P. A. Narayana and 23 S . A. Dikanov, A. A. Shubin and V. N. Parmon, J . Magn. Reson.. 1981,42,4748. 24 A. Samoson and E. Lippmaa, Chem. Phys. Lett.. 1983, 100, 205. 25 S. A. Dikanov, Yu. D. Tsvetkov, M. K. Bowman and A. V. Astashkin, Chcm. Phys. Lett., 1982, 90, 26 B. J. Hathaway and D. E. Billing, Coord. Chem. Rer., 1979, 5, 143. L. Kevan, Magn. Reson. Rev., 1983, 1, 234. 149.232 Coordination Geometry of Cu2+ in M+A Zeolites 27 (a) M. Narayana and L. Kevan, J. Chem. Phys., 1982, 78, 3573; ( h ) M. Narayana and L. Kevan, unpublished results; ( c ) L. Kevan and M. Narayana, Am. Chem. Soc. Sq’mp. Ser., 1983, 218, 283. 28 H. S. Lee, W. V. Cruz and K. Seff, J. Phys. Chem., 1982, 86, 3562. 29 H. A. Jahn and E. Teller, Proc. R. SOC. London, Ser. A , 1937, 161, 220. 30 A. A. Shubin and S. A. Dikanov, J. Mugn. Reson. 1983, 52, 1. 31 M. Romanelli, M. Narayana and L. Kevan, J. Chem. Phys., 1984, 80,4044. 32 M. Anderson and L. Kevan, J. Am. Chem. SOC., submitted for publication. 33 L. B. McKusker and K. Seff, J . Am. Chem. Soc., 1981, 103, 3441. 34 P. C . W. Leung, K. B. Kunz, K. Seff and I. E. Maxwell, J. Phys. Chem., 1975, 79, 2157. 35 J. J. Pluth and J. V. Smith, J . Phys. Chem., 1979, 83, 741. 36 P. E. Rdey, K. Seff and D. P. Shoemaker, J. Phys. Chem., 1972, 76, 2593. Paper 5/615; Receiwd 12th April, 1985

ISSN:0300-9599    

DOI:10.1039/F19868200213

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

J . Chem. SOC., Faraday Trans. I, 1986,82, 233-242 The Hydration Entropies of Ions and their Effects on the Structure of Water Yizhak Marcus Department of Inorganic & Analytical Chemistry, The Hebrew University of Jerusalem, 91904 Jerusalem, Israel The conventional entropies of hydration are interpreted in terms of con- tributions from conversion to the absolute scale, compression from the gas to the solution standard states, long-range electrostatic interactions, immobilization of the solvent near the ion, and water-structural effects. A simple but effective method for the estimation of the entropy of solvent immobilization is presented. Correlations between the entropy of hydration of ions and various measures of the effects of ions on the structure of water, applied in the past mainly to monoatomic ions, are extended to polyatomic ones.Water differs from most other liquids by having a pronounced degree of structure, due to the extended three-dimensional network of hydrogen bonds. This structure is affected by solutes,' including ions that result from the dissociation of dissolved electrolytes. Some ions are known to enhance the 'structuredness' of the water, in whatever manner this might be defined, and others are known to break this structure down. To the latter, structure-breaking category belong large ions of low charge, such as Cs+ or I-, and to the structure-making category belong ions having a strong electrostatic field, such as Li+, Mg2+, or La3+. The structure-affecting properties of the ions are manifested in such properties of their aqueous solutions as the viscosity 2, (structure-breaking ions lower it), the rate of exchange of water molecules4 (structure-breaking ions lower its energy of activation), and the longitudinal relaxation rate of the water molecules, measured by n.m.r.5 (structure-breaking ions increase it).Another property of solutions of electrolytes that has been related to the effect of ions on the structure of water is the standard partial molar entropy of the ions. Since these include intrinsic entropies of the ions, a better measure of their structural effects is their entropy of hydration. The conventional standard molar entropy of hydration of an ion Xz is Ahyd con So(xz) = s&m(xz, as) - so(x*, g) (1) where z is the charge on the ion, S:on(XZ, as) is its conventional standard partial molar entropy in the aqueous standard state, and So(Xz,g) is its standard molar entropy in the gaseous standard state.The former standard state is the hypothetical ideal 1 mol dmP3 aqueous solution, with the convention that S&(H+, aq) = 0 at all temperatures, and the latter standard state is the ideal gas at 0.1 MPa pressure. This pressure and the temperature of 298.15 K are understood to prevail in all the systems to be discussed here. It was realized by Franks and Evans2 that even Ahydcon So(Xz) involves contributions extraneous to the structural effects: uiz. from the change in the volume at the disposal of the ions in the two standard states, from the coordination of water molecules to the ion, and that arising from electrostatic interactions. Frank and Evans,2 as well as the authors who followed considered practically only the monoatomic alkali metal and halide ions in their deliberations. It is instructive to demonstrate that these considerations pertain also to polyatomic ions, and that the water-structural entropy effects for these ions also correlate with the viscosity, exchange rate, and longitudinal 233234 Hydration Entropies of Ions relaxation rate, where known.The present paper considers the available $E,,(X*, aq) and So(Xz, g) data for a large set of ions, and specifies the compression, immobilization, and electrostatic contributions to Ahyd s"(Xz), obtained by the conversion of the conventional values of the standard molar entropies of hydration to the absolute scale in an appropriate manner.After subtraction of the contributions listed above from the absolute standard molar entropy of hydration, the water-structural entropy effects of the ion are obtained as the difference, as is commonly done, there being no direct method for their estimation. Data Employed Conventional standard partial molar entropies of the ions considered here are available in the NBS Tables.12 The standard molar entropies of the gaseous monoatomic ions are calculated by the Sackur-Tetrode equation, which for 298.15 K and the 0.1 MPa standard state pressure takes the numerical expression So(X&onoatomic,g) = 12.4715 In M,+ 108.85 J K-l mol-1 where M , is the relative atomic mass of the ion. The standard molar entropies of the gaseous polyatomic ions considered here are from Loewenschuss and Marcus.13 These data are shown in table 1, as are the derived conventional standard molar entropies of hydration, obtained by means of eqn (1).A few comments on the data shown in table 1 are in order here. The uncertainties of the SzOn(Xz, aq) values follow the convention of being 8-80 times the unit of the last digit reported.12 Those of the So(Xz,g) values follow the code of being 2-10 times the unit of the last digit reported. The values of Sz,,[(CH,),N+, aq] and $~,,[(C,H,),N+, as], not reported in the NBS Tables,12 are from Johnson and Martin.14 For two of the ions in table 1, AuBr; and Hg(CN)i-, the values of So(Xz, g) are not from Loewenschuss and Marcus,13 but have been calculated in the present study.The former of these two complex anions is square planar, having an Au-Br distance of 0.257 nm and a complete set of vibration frequencies given by Goggin and Mink.15 The latter of the two anions is tetrahedral, having an Hg-C distance of 0.222 nm (and the C-N distance taken as 0.11 5 nm)16 and a complete set of vibration frequencies given by J0nes.l' Both sets of frequencies were fully assigned by the authors, i.e. were accompanied by the pertinent degeneracies, including estimates of the frequencies of non-observed vibrations. The entropies of the gaseous ions were calculated according to the manner described previously. l3 Calculation of the Structural Entropy Effect The conventional standard molar entropies of hydration are converted to the corres- ponding absolute values by the addition of z times the absolute standard partial molar entropy of the aqueous hydrogen ion: This quantity was estimated by various methods, as reviewed and discussed by Conway,lS who recommended the value - 22.2 & 1.2 J K-l mol-l for P(H+, aq).The derived absolute standard molar entropies of hydration of the ions considered here are also shown in table 1. The change in the volume at the disposal of the ion, when it is transferred in the process of hydration from its gaseous standard state to its solution standard state, the so-called 'compression term' is: Acomp So = R In (1 dm3 mo1-I x 0.1 MPaIRT) = - 26.7 J K-l mol-1 (4) for all ions. (The volume in the ideal gas is RTIO.1 MPa; in solution it is 1 dm3.) Others have suggested different values, as discussed in the Discussion section below.c3 3 u s s; E8- IA 16- m L- 9z - m OP - 62- I € - I PI - 9E 8s LS - L9 - 2 - 1 9 91 6 PI E E l - S I - Z- S s1- 1- 1 1- 691 - 08 - 82- ( S L ) 8E 61 E LP - 6 I I - EZ S - 82 8Z 61 0 (81) PEI- OLI- 89 - 88 - SL - LS - LS - sz - IP- oz - E 621- 291 - 9E - PE - OE - 61 - 92- ZZ - t E - 211- 09 - PP- IP- 9s - LP - I E - SP - S61- 011- SL- 19- I- PZ- 6E - I01 - ZPI - OPI - 9E - 08 - €2- 62- 8E - PL - €01 - os.'nJ75v *osPAqv OLI - 982- WI - 991 - I Z I - I6- 16- 85 - 5L- EL- ES - 181 - 002- I L - 69 - w- ES - 09- LS - 99 - Wl- 56 - 08 - 9L - 16- 28 - 99 - 222 - WI - Z l l - 221- 9E - 65 - SL- LEI - 602- soz - ZL- L I I - 6S - 59 - PL- I l l - ZPI - on - OSPhSV 1'261 - 8'PLE - Z'ZIZ- 2-2!3- 0991 - 0 E l I - 0 E l I - E'08- 6'96- 8'LI I - 0'16- Z'IEZ- 8'PPZ - L'26- 016- 9'98 - O'EL - 5-28 - 0'6L - 0'88 - 8.69 I - O'LI I - 0.201 - 9'86 - €'El 1 - €'pol - E'88 - 9'201 - 0'002 - 0'22 1 - 9'68 - L'99 1 - 1'8S- 2'18- 6'96 - 9'65 I - 0'59 1 - 8'09 I - 8'6P - 9L.P6- IK9E - 16'2P- I'ZS- 0.68- 9'61 1 - osuoJpAqv L'8LZ 8'69P S'Z8P 8'WP S'OLP S'P82 S'P82 E'91P KE9E L'6LC 0' I PE P'182 9X92 1.962 5'882 8'LLZ O'L62 I 'Z8Z O'E92 6'L92 2'882 L'8LZ E'WL Z'SPZ E'9EZ Z'ZIZ S'ZEZ L'91 I O'E8P 6'IEE E'981 PI'ZSI 9S.691 LS'E9I OP'E s 1 6S'SPI 6P'S L 1 SE'OL 1 ZE'SLI W'L91 98'69 I I P'P9 I E9'PS I 86'LP I 66'2EI 4.98 0'56 E'OLZ 9'2EZ O'SOF 0'2 L I O'ZL I 0'9EE 6'992 6'192 E'PPZ 2.0s 8'8 I C'IOZ S'L61 2'161 0.2zz 9'66 I O'P8 I 0'08 I 9'81 I L'19I E'Z91 9'9PI O'EZ 1 6'LOI E'PPI I'PI O'E82 0'0 I2 6'96 9'PI - €'I11 P'Z8 S'9S 8'El - S'OI 9'6 S'SZI 89'ZL SO'EEI OS'IZI S'201 0'65 P'E I (be)""&236 Hydration Entropies qf Ions The electrostatic contribution to the entropy is considered to pertain beyond the first hydration shell.Any electrostatic effects that pertain to distances closer to the ion are taken up by the solvent immobilization term. The Born equation is invoked for the calculation of the electrostatic term : ABorn So(Xz) = [Ne2(C1~/2T)p/871~0 c2] z 2 / ( r X + 2r,) (5) where N is Avogadro's constant, e the charge of the proton, E , the permittivity of free space, E the relative permittivity of water and (?E/t)T)p its isobaric temperature coefficient, r x is the radius of the ion and rw is the radius of a water molecule. When the relevant numerical values are inserted, eqn ( 5 ) becomes: (6) ABorn So(Xz) = - 4067 z2[(rx/pm) + 2801 J K-' mol-l where the diameter of the water molecule in the first hydration shell is taken to be 280 pm.It is now necessary to specify the ionic radius r x to be used. For the monoatomic ions the Pauling crystal ionic radii (coordination number 6), as given by Shannon and Prewitt,lg are employed. For the polyatomic ions the thermochemical radii, as given by Jenkins and Thakur,20 are employed as far as they are available. In the other cases, the distance between the central atom and the peripheral atom that has been used13 for the calculation of the rotational contribution to the entropy of the gaseous ion is scaled by a factor obtained from the known thermochemical radii of isostructural ions.These factors are 1.60 for triatomic, 1.33 for tetra-atomic, 1.53 for tetrahedral and 1.41 for octahedral ions. The electrostatic contributions, according to eqn (6), and the radii rx used in parentheses, are shown in table 1 . An interim quantity, Ahyd So*, results when Acomp So and ABorn S o ( X r ) are subtracted from the absolute standard molar entropy of hydration of the ion : A,,, So* = Ahyd So(Xr) - Acomp So - ABorn So(Xz). (7) For monoatomic ions, both those included in table 1 and all the others for which the relevant data are available, AhydSo* is a more or less smooth function of lzl/rx, see fig. 1. This is interpreted as arising from the two remaining contributions to the entropy of hydration: the translational immobilization of the solvent in the first hydration shell of the ion and the effect of the ion on the structure of the water, and in particular from the first.The translational immobilization of the water is due to its coordination to the ion in the first hydration shell around the bare ion, containing n molecules of water on the (8) average : Xr(bare, aq) + n H,0(1) + X(H,O)Z, (aq). From the initial n+ 1 particles participating in reaction (8) only one, the hydrated ion, retains translational freedom. If it is assumed that the non-translational degrees of freedom of the water molecules are not affected, then the amount nS&ns,(H,O,l) of entropy is lost in reaction (8), and a small amount of translational entropy is gained, owing to the increase in the mass of the hydrated ion, compared with the bare one: (3/2) R In (M[X(H,O),]/M(X)).The translational entropy of water in the liquid is taken to be: (9) The standard molar entropy of water in the liquid and gaseous states is from the NBS Tables,12 and the translational entropy of the gaseous water is obtained from eqn (2). The translational immobilization entropy of the solvent is thus Atrim So(Xz) = = 12.4715 In (M[X(H,O],]/M(X))-26.0 n J K-I mol-*. (10) It remains to specify the number n of water molecules that are translationally immobilized by the ion. In view of the behaviour shown in fig. 1, and accepting as a fact S&nsl(H20, 1) = S0(H20, 1) - [So(H20, g) - StOransl(H20, g)] = 26.0 J K-l mol-l.Y. Marcus 237 I I I I I V 4 e OHf Zr I I I I 1 0 0.01 0.02 0.03 0.04 0.05 0.06 I z l/(r/pm) Fig.1. Difference between the hydration entropies of ions and the sum of the compression and electrostatic contributions thereof, Ahyd So*, plotted against the field strengths of the ions, lzl/rx, for monoatomic ions (e, cations, 0, anions). that the sodium ion is neither structure-making nor structure-breaking89 21 the following empirical relationship is proposed: n = 355 Izl/(rX/pm). (1 1) This value of n, together with eqn (10) above and (12) below, produce the value of AstrucSo(Na+) = 0 as specified. Finally, AtrimSo(XZ) is subtracted from Ahy9So*, and the difference is taken to represent the entropic results of the effect of the ion on the structure of the water: The values of the quantities appearing in eqn (12) are presented in table 1, those of n obtained according to eqn (11) being shown in parentheses after the value of Negative values of Astrue So(Xz) denote structure-making properties of the ions whereas positive values denote structure-breaking properties.The quantitative data of table 1 are translated into these qualitative terms (- for structure-making, + for structure-breaking ions) in table 2, for the sake of comparison with the results from other approaches. Since Astrue So(Xz) has considerable cumulative errors from all the terms that come into its calculation [estimated at k4(1+z2)4 J K-l mol-'1, values that are between -6 and +6 J K-l mol-l are denoted by in table 2, signifying that they are borderline cases. Atr im So(XZ)* Correlation with Other Aspects of Ionic Effects on the Structure of Water Table 2 presents, again in a qualitative manner, the results from other lines of investigation concerning the effects of ions on the structure of water.To the ions in table 2 must be added a long list of other ions, all multiply charged and not very big, that are assigned to the structure-making category by all the methods. These include both monoatolnic cations, such as Mg2+, Ca2+, Zn2+, Hg2+, La3+, Zr4+, etc. and polyatomic anions, such as Cog-, SO:-, PO:-, etc. as far as they have been studied by the methods considered in table 2. The picture arising from table 2 (and the added list of ions238 Hydration Entropies of Ions Table 2. Water structure-affecting properties of ions" ion S V E N ion S V E N Li+ Na+ K+ R b+ cs+ T1+ Ba2+ Pb2+ F- c1- Br- I- S2- NH; &+ (CH,),N+ (C,H,),N+ N3 CN- SCN- NO, NO; CH,CO; ClO, BrO; I 0 3 BF, c10, BrO; 10, MnO; TcO; ReO, Au(CN); AuCl; AuBr; CrOz- Cr20:- Co(CN)i- Fe( CN)g- Fe(CN):- Ag(CN), so:- S @ - Hg(CN):- + + - + + + + + + - + + + + + + + - + + + - + + + + - - + + + + - + - - + + + - a S, from the entropy of hydration, this work.V, from viscosity data, ref. (3), ( 5 ) , (22) and (23). E, from the energy of activation for water exchange, ref. (4). N, from the n.m.r. longitudinal relaxation rate, ref. ( 5 ) . mentioned above) is of general agreement among the methods, with some minor exceptions. Numerical correlations were determined between the quantities expressing the effects of the ions according to the various methods included in table 2.24 The results are: (13) (14) (15) B(n.m.r.) = -0.01 -0.0024A,,,,, So; n' = 28; reor, = -0.936 B(viscosity) = 0.14-0.025A,,,,,So; n' = 27; rcorr = -0.955 E(exchange) = - 0.9 1 - 0.0 1 8Astruc So ; n' = 16; rcorr = - 0.857 where B(n.m.r.) = [(zx/zo) - 11 (n VHzo) is the analogue5 of the Dole-Jones B(viscosity) coefficient3 (both in dm3 mol-l), E is the activation energy for water exchange from the immediate vicinity of the ion with the bulk water (in kJ mol-l), n' is the number of pairs of data, and r,,,, is the correlation coefficient for the linear regression.The correlations of these three measures of the effects of ions on the structure of water with the structural entropy contribution obtained by eqn (12) in this work are as good as they are among themselves. In all Ae cases there are items that fall very far from the linear regression curve, and these items have been excluded from correlations (13E(15). The more important qualitative discrepancies noted in table 2 are the apparent positive structural entropy changes (structure-breaking properties) obtained for Li+ and S2-, which are known or expected to be structure-making, and the negative structural entropy change (structure-making property) obtained for NHZ, which other methods assign to the opposite category.In the case of Li+ the fault may rest with the overlarge value of n, the number of waterY. Marcus 239 molecules immobilized, which is larger than the usual coordination number of 4. If this is taken to be the value of n applicable in eqn (lo), then the translational immobilization term becomes Atrim S"(Li+) = - 74 J K-l mol-1 and the structural entropy change becomes Astrue So(Li+) = - 30, as expected for this structure-breaking ion.In the case of S2- the fault may rest with the unrealistically small radius used for this ion in solution (only 184 pm) compared with the 18 1 pm of the isoelectronic singly charged C1- ion. If the value of the radius is increased by 20% (to the value of the univalent radius), then the Born term is 3 and the immobilization term is 15 J K-l mol-l less negative, yielding for Astrue So(S2-) the smaller positive value of 7 J K-l mol-l, which is more reasonable for such a large, though doubly charged, anion as S2-. Another possible source for the discrepancy could be an incorrect value of Sg,,(S2-,aq), owing to the difficulties encountered with this readily hydrolysed anion.The fault in the case of NH: may be only apparent, and may not be due to a shortcoming of the structural entropy method. The ammonium cation may not be structure-breaking, as the other methods lead one to think, since it fits so well into the structure of the water itself that it cannot be discerned by X-ray diffraction studies of aqueous solutions of ammonium salts. The negative value of Astruc S"(NH;) is, therefore, to be expected from the fact that the ammonium ion forms four hydrogen bonds with the water molecules surrounding it. Discussion The occurrence of negative values for Ahyd So(Xz), both for the ions appearing in table 1 and for all the others that have ever been examined, is generally interpreted as being due to a summation of contributions, most of which are negative.It is also generally accepted that electrostatic effects, alone or in combination with a 'neutral term' (accounting for the change in the standard states on hydration, among other effects), are inadequate for explaining the entire negative magnitude of the entropy of hydration. They are accepted, however, as accounting for some part of it. In order to allow for a positive contribution to the entropy of hydration for ions known to be structure-breakers, there must exist some major negative contribution to the entropy of hydration other than the electrostatic term. The present study proposes the notion of the translational immobilization of solvent (water) molecules around the ion to take this into account.Of the many authors who concerned themselves with the interpretation of the entropy of hydration of 6 - 1 1 ~ 2 5 ~ 26 most have implicitly or explicitly taken approaches similar to the present one, but have emphasized different aspects. The entropy changes that occur when an ion is first coordinated by a certain number of water molecules and the resulting hydrated ion is subsequently dissolved in water were calculated.6* i v 26 The results of these quite complicated calculations, based on unverifiable assumptions and arbitrary choices of parameters, did not give impressive agreement with the experimental entropies of hydration, even for the monoatomic alkali-metal and halide ions. The number of water molecules coordinated to the ions in these calculations was fixed arbitrarily67' or was left open as a free parameter with a wide range of values.26 The other approaches are, essentially, variants of two lines.The approach of Frank and Evans,2 adopted with slight changes by Friedman and Krishnan,* regarded the (absolute) entropy of hydration of an ion as the sum of the same four contributions as considered in the present study : compression, electrostatic, immobilization, and structural contributions (though not necessarily under the same names). Frank and Evans2 used the unit mole fraction standard state for the solute ion in the solution, which is an unreasonable It was, indeed discarded by Friedman and Krishnan,s who employed the molal concentration scale and the hypothetical ideal 1 mol kg-l standard state for the ion in the solution.Owing to the nearness of the density of water at 298.15 K to unity, in terms of kg dm-3, the compression term of these latter240 Hydration Entropies of Ions authors is the same as that used in the present work. Their electrostatic term is also the same as that used here, being applied beyond the distance of the Pauling radius plus the diameter of a water molecule. The translational immobilization of the solvent near the ion was set in the approaches used by the authors named above as a constant coordination number (four) times one half of the entropy lost by a mole of water on freezing (at 298.15 K). These are arbitrary choices that lead operationally to the value of Astruc So(Xz) obtained as the difference.Other choices, such as complete freezing, have been made,lo9 22 and by ignoring any structural contributions to the entropy have led to operationally defined hydration numbers. Krestov and Abrozimovg used a similar approach, but lumped the compression term and the translational immobilization of the solvent term together, in a term called by them AS;. This is obtained by allowing the ion to retain the same fraction of its (translational) entropy in the gas phase after its transfer into the solution as does a rare gas atom on the average (without regard to the size of the latter). The remainder of the entropy of hydration, called includes electrostatic and structural contributions (long- and short-range entropy effects), but their individual evaluation has not been specified.The results of this approach concerning the effects of the ions on the structure of water are limited to monoatomic ions, and are in qualitative agreement with the results of the present work, as shown in table 2. again limited their considerations to univalent ions and mainly to monoatomic ones [the ions (CH,),N+, (C,H,),N+, CN-, and C10; were included in some of their tables]. They developed the electrostatic interaction approach to a multilayer model. The ‘compression term’ was replaced by a ‘neutral term’, obtained from the average entropy of solution of gaseous non-polar solutes of the same size as the ion. This seems to take care of some of the solvent immobilization too, since Asoln So of the rare gases in water (from the 0.1 MPa to the 1 mol dm-, standard states) is considerably more negative than - 26.7 J I C 1 mol-l.However, neither a single-layer model nor a two-layer one accounted satisfactorily for the electrostatic contribution to the entropy of solvation, obtained as the difference between the total ionic entropy of solvation and the neutral term, in the cases of hydrogen-bonded solvents in general and water in particular. For the two-layer model, the electrostatic contribution for the second layer was obtained as the difference between the total entropy of hydration and the combined contributions from the first layer (corresponding roughly to the translational immobilization of the solvent in the present work), the bulk solvent (corresponding roughly to the electrostatic term in the present work), and the neutral term (corresponding roughly with the compression term in the present work).Conceptually, therefore, the entropy change due to this second layer could correspond to the structural effects of the ion. However, the detailed calculations of Abraham et a1.l1 produced positive values of this term for all the ions considered, both structure-breaking ones (e.g. Cs+ and I-) and a structure-making one (F-). The term was less positive for F- and more positive for K+, Rb+, Cs+, C1-, Br-, I-, and C10; than for Na+, and was in linear correlation with the Dole-Jones B-coefficient of the viscosity. This was cited” as evidence that this contribution to the entropy from a disordered second layer around the ion did have structural connotations. The arrival at all-positive values, however, depends on the rather arbitrary choices of the parameters that have gone into the calculations of the contributions from the neutral, first layer, and bulk terms.It is seen that the approach to the problem taken by other authors each has its own arbitrary choice of models and parameters. The approach proposed here is not different in this respect. Little can be said against the choices made here regarding the ‘compression term’ and the ‘electrostatic term’, so that any criticism should be pointed against the ‘translational immobilization term’, Atrim So(Xz). There are two main aspects that can be criticized: one concerns the limitation to translational entropy changes, the other concerns the empirical calculation of n, the number of water molecules taken to be translationally immobilized. Abraham etY.Marcus 24 1 Beyond admitting that both lines of criticism are valid, the following arguments can be raised to counter them. The rotational degrees of freedom lost by the coordinated water molecules are expected to be compensated for by degrees of freedom of rotation around the ion-water bond and of libration about it, and by the rotational entropy of the hydrated ion, compared with the lack of it for a monoatomic bare ion. How adequate such a compensation is must be left to very complicated calculations, to which the unsuccessful examples reported so far6y ' 9 26 are a deterrent. The estimation of the translational entropy of water in the liquid by eqn (9) assumes, in essence, that the non-translational degrees of freedom of water are not affected on its condensation from the gaseous to the liquid states.Again, a compensation of positive and negative entropy effects is invoked as a justification. The simple-minded estimation of the number n by eqn (11) could, perhaps, have been improved at the expense of requiring a more complicated expression, but it must be conceded by critics that the resulting values of n, shown in table 1 in parentheses in the column of Atrim So(Xz), are reasonable as hydration numbers. This is true not only for the ions shown in this table but also for those not included there: monoatomic ions, e.g. Mg2+(9.9), Mn2+(8.6), Zn2+(9.5), Hg2+(7.0), La3+( lO.Z), Lu3+(12.4), Fe3+(16.5), and Zr4+( 19.7); and polyatomic ones, e.g.OH-(2.7), H,0+(2.7), P043-(4.5), among others. A different choice of the numerical constant would change these numbers proportionally and cause a shift of the water structure-breaking and structure-making assignments of the ions. For instance, if the numerical constant in eqn (1 1) is set at 240, then potassium becomes the ion to which the structure of water becomes indifferent [ie. Astrue S"(K+) = 01, and sodium and lithium become structure-making [Astruc S"(Na+) = - 22 and (Xstruc S"(Li+) = - lo], but the resulting hydration numbers are then on the low side [e.g. Mg2+(6.7) and H,O+( 1.8)]. The main merit of the present approach is that in spite of its simplicity it yields estimates of the effects of the ions on the structure of water that are in agreement with those obtained by other methods.Another merit is that it can be readily extended to a large number of ions not studied by these methods, but for which S&, (Xz, aq) and So(Xz, g) data are available or can be estimated. These include many complex ions, for which the structural effects can now be predicted. References 1 Y. Marcus and A. Ben-Naim, J. Chem. Phys., 1985, in press. Abs. 7th Intl. Symp. Solute- 2 H. S. Frank and M. W. Evans, J. Chem. Phys., 1945. 13, 507. 3 R. W. Gurney, Ionic Processes in Solution (McGraw-Hill, New York, 1953). 4 0. Ya. Samoilov, The Structure qf Electrolyte Solutions and the Hydration of Ions (Izd. Akad. Nauk, 5 G. Engel and H. G. Hertz, Ber. Bunsenges. Phys. Chem., 1968, 72, 808. 6 S. Goldman and R. G. Bates, J. Am. Chem.SOC., 1972, 94, 1476. 7 J.O'M. Bockris and P. P. S. Saluja, J. Phys. Chem., 1972, 76, 2291. 8 H. L. Friedman and C. V. Krishnan, in Water: A Comprehensiue Treatise, ed. F. Franks (Plenum Press, 9 G. A. Krestov, Zh. Strukt. Khim., 1962,3, 137; G. A. Krestov and V. K. Abrozimov, Zh. Strukt. Khim., Solute-Solvent Interactions, Reading, 1985. USSR, 1957; Consultants Bureau, New York, 1965). New York, 1973), Vol. 3. 1964, 5, 510; V. K. Abrozimov, Zh. Strukt. Khim., 1973, 14, 211; 1976, 17. 838. 10 A. G. Ryabukhin, Zh. Fiz. Khim., 1981, 55, 1670. 1 1 M. H. Abraham and J. Liszi, J. Chem. SOC., Faraday Trans. I , 1978, 74, 2858; 1980,76, 1219; M. H. Abraham, J. Liszi and E. Papp, J. Chem. SOC. Faraday Trans. I, 1982,78, 197; M. H. Abraham, J. Liszi and L. Meszaros, J. Chem. Phys., 1979, 70, 2491. 12 D. D. Wagman, W. H. Evans, V. B. Parker, R. H. Schumm, I. Halow, S. M. Bailey, K. L. Churney and R. L. Nuttall, The NBS Tables of Chemical Thermodynamic Properties (Am. Chem. SOC. and Am. Inst. Phys., Washington, 1982). 13 A. Loewenschuss and Y. Marcus, Chem. Rev., 1984,84, 89. 14 D. A. Johnson and J. F. Martin, J. Chem. SOC., Dalton Trans., 1973, 1585. 15 P. L. Goggin and J. Mink, J . Chem. Soc., Dalton Trans., 1974, 1479.242 Hydration Entropies of Ions 16 W. P. Griffiths, Coord. Chem. Rev., 1975, 17, 177. 17 L. H. Jones, Spectrochim. Acta, 1961, 17, 188. 18 B. E. Conway, J. Solution Chem., 1978, 7, 721. 19 R. D. Shannon and C. T. Prewitt, Acta Crystallogr. Sect. B, 1969, 25, 925; 1970, 26, 1046. 20 H. D. B. Jenkins and K. P. Thakur, J . Chem. Educ., 1979, 56, 576. 21 V. K. Abrozimov, Radiokhimiya, 1972, 14, 916. 22 E. Asmus, 2. Naturforsch., Teil A , 1949, 4, 589. 23 J. G. Mathieson and G. Curthoys, Aust. J . Chem., 1975, 28, 975. 24 Y. Marcus, IUPAC Conf. Chem. Thermod., London, 1982, Abstr. p. 45. 25 H. Ulich, 2. Elektrochem., 1930, 36, 497. 26 R. Gonzalez Maroto, D. Posadas, M. I. Sosa and A. J. Arvia, Anales Assoc. Quim. Argentina, 1982, 27 A. Ben-Naim and Y. Marcus, J . Chem. Phys., 1984, 80,4438; 1984,81, 2016 70, 979. Paper 51644; Received 18th April, 1985

ISSN:0300-9599    

DOI:10.1039/F19868200233

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I , 1986, 82, 243-246 Application of the Flory-Huggins Theory to the Solubility of Solids in Glyceryl Trioleate Cary T. Chiou* U.S. Geological Survey, Water Resources Division, Box 25046, M S 407, Denver Federal Center, Denver, Colorado 80225, US. A . Milton Manes Department of Chemistry, Kent State University, Kent, Ohio 44242, U.S.A. The conventional thermodynamic deviation for ideal solid-liquid solubilities is modified by substituting the Flory-Huggins model for Raoult’s law. A comparison of published data for eleven solids in glyceryl trioleate with the predictions of the conventional and modified equations shows that the significantly higher athermal solubilities from the modified equation are in much better agreement with the experimental data.This suggests that discrepancies between the data and the predictions of the conventional model for ideal systems result from the inappropriate use of Raoult’s law for systems with significant solute-solvent size disparity rather than from specific interactions. Whereas Raoult’s law accounts for the behaviour of solutions of small molecules of comparable size, the Flory-Huggins model provides a more accurate treatment for systems containing macromolecules. In this report, we consider the solubilities of some solids in glyceryl trioleate (triolein), in which the solvent is only of moderately larger molar volume than the solute. In the conventional derivation of the ideal solubility of a solid in a liquid solvent, one assumes that Raoult’s law applies at all temperatures to mixtures of the solvent and molten solid, including supercooled molten solid below its melting point.Starting with an equilibrium mixture of solid and melt at the melting point, and cooling to equilibrium temperature, one calculates the reduction in chemical potential (or activity) of the solid, relative to the supercooled liquid, and equates it to the reduction in chemical potential in the liquid phase that results from dilution with so1vent.l The simplified equation for the solid activity (given the reasonable assumption of constant A P f ) is where as is the activity of the solid (with reference to the supercooled liquid at the same temperature), AHf is the molar heat of fusion of the solid, R is the gas constant, and Tm and Tare the melting and equilibrium temperatures (in K).If Raoult’s law is assumed to apply to the dissolved solid (which is considered in effect to be a dissolved supercooled liquid), the activity of the solute relative to pure supercooled liquid at the same temperature is equal to its mole fraction, i.e. At saturation (solid-liquid equilibrium) the chemical potentials (and activities) of solid and dissolved solute become equal, leading to the well known equation for the ideal solubility of the solid, i.e. al = X1. (2) -AH, T,- In X:d = - ~ R ( (3) 243244 Flory-Huggins Theory for Lurge Solutes For solvents of very high molecular weight, Raoult’s law leads to gross overestimates of the solute activity (and therefore gross underestimates of the solute ~olubility),~-~ and these effects are generally ascribed to the effect of solute-solvent size disparity on the entropy of mixing rather than to exothermic solute-solvent interactions.The Flory- Muggins model uses the volume fraction, rather than the mole fraction, for estimating the entropy of mixing of a component in polymer solution, and gives the activity of a solid substance in a macromolecular solvent phase as where q? is the volume-fraction solubility, q5p is the volume fraction of the macromolecular solvent phase, Y and pp are the molar volumes of solute and solvent and x is the Flory-Huggins interaction parameter, the sum of excess enthalpic kH) and excess entropic (ils) contributions to the solute-solvent interaction. The limiting athermal solubility* of a solid substance in a macromolecular solvent is obtained by setting x = 0 and substituting the right-hand side of eqn (4) for lnX:d in eqn (3) to give The limiting athermal volume-fraction solubility, #it, can therefore be solved numerically according to eqn (5).As noted, whereas eqn ( 5 ) gives the same predicted solubility as eqn ( 3 ) when V = Vp (in which case, bit = xd), the values defined by the alternative equations diverge with increasing size disparity between and Vp. Although the Flory-Huggins model is clearly superior to Raoult’s law for polymer solutions of high molecular their relative merits have not been sufficiently verified for systems with only moderate size disparity. The comparison is conveniently made on those solute-solvent pairs, such as are found for some solutes in lipids, that may be expected to have minimal specific interaction and in which the size disparity ( Vp/ V ) is large enough to produce significantly different predictions of solubilities. For this purpose we compare the data of Patton et af.7 and of Dobbs and Williams8 on the solubilities of some relatively non-polar solids in lipid triolein with predictions from eqn (3) and (5).The results are given in table 1. The size disparity between triolein (molecular weight 885.4; & = 966 cm3 mol-l) and the solutes is in the range of &/ V = 3.9-8.5. Data for solids with high melting points and high heats of fusion (AHf) have been excluded from consideration, since the calculated activity of the solid is sensitive to uncertainties in ARf and melting point.Since the selected solids and triolein have similar chemical compositions and polarities, the solutions should be reasonably close to being ‘ ideal ’. The observed (mole-fraction) solubilities of the solid substances in triolein are higher than the predictions of eqn (3) by as much as 100%. On the other hand, the observed solubilities in volume fractions are either close to or lower than the respective athennal volume-fraction solubilities given by eqn (5). The results are therefore in much better agreement with the Flory-Huggins model. Of particular significance are the results with lindane, fluoranthene and DDT, which exhibit only moderate size disparities with triolein ( Vp/ V z 4-5). The magnitude of the negative deviation from Raoult’s law is beyond the uncertainty of observed and calculated solubility data.Since the experimental data can be well accommodated by the Flory-Huggins model (withx = 0) and since no convincing evidence suggests any strong specific interaction of these relatively non-polar solutes with triolein, the observed negative deviation from Raoult’s law appears to be merely an artifact of the model. * We refrain from using the term ‘ideal solubility’ in order to avoid confusion with the conventional definition by eqn (3).Table 1. Solubility of solid organic solutes in triolein and related physical properties of the solutes: & = melting point; AHf = molar heat of fusion ; A4 = molecul'ar weight; V = molar volume solubilityC A f l f a Vb (gper 100 g compound k / " C /kcal mo1-I M /cm3 mol-' triolein) t/OC X:,, X ;(p naphthalene p-dichlorobenzene acenaph t hene biphenyl 2,6-dimethylnaphthalene 2,3-dimethylnaphthalene fluorene p henan t hrene fluoranthene lindane p,p'-DDT 80 53 96 71 110 105 116 101 111 113 109 ______ 4.61 128.19 130 4.35 147.01 114 5.02 154.21 171 4.43 154.21 155 5.78 156.23 155 4.73 156.23 155 4.51 166.23 165 4.45 178.24 170 4.53 202.26 200 5.80 290.8 186 6.07 354.49 250 18.41 k3.32 56.70 f 7.92 63.3 10.30f 1.81 19.93 & 6.47 41.5 5.47 f 1.64 7.91 k2.31 9.56k 1.01 10.22 _+ 2.37 7.82 _+ 1.75 9.28 Ifl0.22 8.00 k 3.10 9.54f0.57 15.3 10.05 23 23 37 23 23 37 23 23 23 23 23 25 37 23 25 37 ~ 0.556 k 0.045 0.771 k0.024 0.792 0.369+0.041 0.5 19 * 0.083 0.705 0.234 f 0.054 0.304 & 0.063 0.337 & 0.024 0.333 f0.052 0.253 f 0.043 0.224 * 0.004 0.3 18 0.163 fr 0.054 0.192 & 0.008 0.200 0.282 0.506 0.707 0.185 0.350 0.49 1 0.107 0.175 0.160 0.206 0.171 0.107 0.157 0.0979 0.105 0.156 0.146 & 0.023 0.286 f 0.028 0.310 0.0948 & 0.01 52 0.153 f 0.043 0.277 0.0473 f 0.0 136 0.0671 kO.0185 0.0801 k 0.0080 0.0818+0.0175 0.0662 & 0.0 140 0.0526 f 0.001 1 0.0824 0.0487 _+ 0.0183 0.0579 k 0.0032 0.0608 #ite 0.133 0.265 0.427 0.0871 0.175 0.265 0.0928 0.152 0.0742 0.0980 0.0828 0.0498 0.0742 0.0485 0.0520 0.0789 ~ a Values for acenaphthene, 2,3-dimethylnaphthalene, 2,6-dimethylnaphthalene, fluorene, fluoranthene, and p,p'-DDT are from those cited by Patton ct af.? and the rest from those cited by Chiou et al.!' Molar volumes of the supercooled liquids are approximated by calculations based on molecular weights and solid densities for all compounds (except lindane and DDT).For lindane and DDT the densities at their melting points are determined and used in the calculations. The 23 "C solubility values are from Patton el aL7, the 37 "C values from Dobbs and Williams8 and the 25 "C values from this work. Values under columns x z b and #zb are the corresponding values expressed in mole fraction and in volume fraction. The uncertainty of the 37 "C values is ca. 10% according to the source. The ideal mole-fraction solubility by Raoult's law [eqn ( 3 ) ] . The limiting athermal volume-fraction solubility by the Flory-Huggins model [eqn (511.246 Flory-Huggins Theory f o r Large Solutes The use of Raoult’s law for defining the ideal solubility of solid solutes in triolein (and which assumes unit activity coefficients for liquid solutes) has also led to gross underestimates of the observed triolein-water partition coefficients of both solid and liquid solutes1* that have molar volumes comparable with those in the present study.Thus the fractional activity coefficient that expresses the supposed non-ideality from the point of view of Raoult’s law for solutes in triolein lacks physical justification. Whereas the same effect could occur in other systems with similar solute-solvent size disparities, the effect might well escape recognition in systems in which there is significant solute-solvent incompatibility. In such cases the incompatibility and size-disparity effects may cancel each other and the experimental data may be erroneously attributed to ideal solubility The data in table 1 manifest that the entropies of mixing of the aromatic solutes with lipid triolein are in better agreement with predictions by the Flory-Huggins theory than by Raoult’s law.Although Shinoda and Hi1debrandl1-l3 have illustrated a contrary effect for some binary mixtures with holar-volume ratios as high as 9: 1, their results were obtained with mixtures of globular and compact molecules which do not conform to the Flory-Huggins postulate for chain-like molecules. As Flory14 has pointed out, such systems do not fulfill the condition of equal accessibility of the total volume to segments of either molecule. It is therefore not surprising that their behaviour may be better accounted for by Raoult’s law.In triolein the segments of the hydrocarbon chains appear to be relatively free to interact individually with other segments and with solute molecules, despite the fact that the chains are connected at one end. The data we present, which are more consistent with the Flory-Huggins model, supplement the earlier work by Shinoda and Hildebrand on essentially different systems. In summary, the results presented here demonstrate the distinct superiority of the Flory-Huggins model over the conventional model (Raoult’s law) in accounting for the solubilities of solid substances (and solute activity coefficients) in lipid systems, where the solute-solvent size disparity is only quite moderate. It is also timely to recognize the potential error in applying Raoult’s law in systems (except for special cases as mentioned) where the size disparity cannot be safely ignored. by eqn (3). References 1 J. H. Hildebrand, J. M. Prausnitz and R. L. Scott, Regukur und Related Solutions (Van Nostrand- Reinhold, New York, 1970), pp. 21-22. 2 P. J. Flory, J . Chem. Phys., 1942, 10, 5 1. 3 M. L. Huggins, Ann. N. Y. Acad. Pi., 1942, 43, 1. 4 P. J. Flory. Principles of Polymer Chemistry (Cornell University Press, Ithaca, , Neu York, 1953), 5 R. L. Scott, J. Chem. Phys., 1949, 17, 268. 6 D. Patterson, Y. B. Tewari, H. P. Schreiber and J. E. Guillet, Macromolecules, 1971, 4, 356. 7 J. S. Patton, B. Stone, C. Papa, R. Abramowitz and S. H. Yalkowsky, J . Lipid Res., 1984, 25, 189. 8 A. J. Dobbs and N. Williams. Chemosphere, 1983, 12, 97. 9 C. T. Chiou, D. W. Schmedding and M. Manes, Enuiron. Sci. Tcchnol., 1982, 16, 4. 495-520. 10 C. T. Chiou, Environ. Sci. Technol., 1985, 19, 57. 11 K. Shinoda and J. H. Hildebrand, J . Phys. Chem., 1957, 61, 789. 12 K. Shinoda and J. H. Hildebrand, J. Phys. Chem., 1958, 62, 292. 13 K. Shinoda, Principles of Solution and Solubility (Marcel Dekker, New York, 1978), pp. 11 1-1 16. 14 P. J. Flory, Discuss. Faraday Soc., 1970, 49, 13. Puper 5/623; Receiced 15th April, 1985

ISSN:0300-9599    

DOI:10.1039/F19868200243

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. 1, 1986,82, 247-253 Multicomponent Diffusion of Double Salts Sodium Hydrogen Sulphate in Aqueous Solution Betty Wiens and Derek G. Leaist* Department of Chemistry, The University of Western Ontario, London, Onturio, Canada N6A 5B7 Diffusion coefficients for the double salt sodium hydrogen sulphate in aqueous solution have been determined at 25 "C from infinite dilution to 2 mol dm-3 using limiting ionic conductances, the Harned restricted-diffusion method and Gouy optical data reported previously for aqueous sodium sulphate-sulphuric acid mixtures. Because the total flux of hydrogen ions exceeds the flux of less-mobile sodium co-ions, diffusion of aqueous NaHSO, forms a ternary mixture of NaHS0,-H,SO, in front of the diffusion boundary while a ternary mixture of NaHS0,-Na,SO, remains behind.Diffusion in aqueous solutions of NaHSO, can be described by a coupled flow of H,SO, in addition to the flow of NaHSO,. At concentrations below 0.1 mol dm-3 the coupled flow of H,SO, exceeds the flow of NaHSO,. Expressions are developed to predict ternary transport coefficients for diffusion of double salts. Although a solution of a double salt [such as AlK(SO,),-H,O] at thermodynamic equilibrium is a two-component system,l? diffusion of a double salt is a multicomponent pro~ess.~ In a solution of salt MNX for example, the equilibrium concentrations of M+, N+ and X2- are all identical and the composition of the system is defined by a single concentration variable. However, if a gradient in concentration of the salt is formed, co-ions M+ and N+, which have different mobilities, may be expected to diffuse at different rates.If M+ is the more mobile co-ion the extra flow of M+ will lead to a coupled flow of M2X in addition to the main flow of MNX. Diffusion would then resolve an initial binary solution of MNX into a ternary mixture of MNX + M2X in front of the diffusion zone and a ternary mixture of MNX+N,X behind it. These considerations apply to transport of numerous important electrolytes including acid salts of polyprotic acids and double salts of mineralogical or biological significance. No multicomponent diffusivities have been reported for these salts. This research was undertaken to measure diffusion coefficients for the double salt sodium hydrogen sulphate in aqueous solution. It is shown that diffusion coefficients for this salt can also be obtained from ternary diffusion coefficients reported previously for aqueous sodium sulphate-sulphuric acid mixtures.*~ The results for sodium hydrogen sulphate are compared with binary diffusion coefficients for aqueous sodium sulphate5-' and sulphuric acid.*.Experimental Diffusion coefficients were determined at 25 "C using a simplified version of Harned's restricted-diffusion lo In this technique salt fluxes are followed by monitoring changes in electrical conductance near the top and bottom of short diffusion columns. The conductimetric method allowed measurements to be made at low concentrations ( < 0.1 mol dm-3) where bisulphate is appreciably dissociated. Initial concentration gradients were formed by injecting NaHSO, solutions into the lower end of cylindrical diffusion channels filled with dilute NaHSO, solutions.At the 247248 Multicomponent Diflusion of NaHSO, same average cell compositions separate experiments were performed with initial gradients in the concentration of H2S0,. This was accomplished by introducing NaHS0,-H,SO, solutions into cells filled with NaHS0,-Na,SO, solutions; equimolar amounts of H,SO, and Na,SO, were added to each cell to ensure that the average cell composition corresponded to NaHSO, + H20. Eigenvalue analysisll of the conductance versus time data from the NaHS0,- and H,SO,-gradient experiments gave ternary diffusion coefficients for NaHSO,(c,~H,SO,(c, = 0)-H,O. Details of the equipment and procedure have been reported.9* l1 All materials were reagent grade.Na,SO, was dried in a vacuum oven at 130 "C. Concentrations of stock solutions of NaHSO, and H2S0, were determined by poten- tiometric titration against standardized NaOH. Solutions were made up in calibrated volumetric flasks using twice-distilled, deionized water. Results Transport equations Diffusion in aqueous sodium hydrogen sulphate solutions entails transport of Na+ and rapidly interconverting HSO;, HS and SO:- species HSO, = H+ + SO:- K2(25 OC)l2? l3 = 0.0105 mol dm+. Diffusion of the salt may be described using flow equations3? l4 for each of the four solute species j i = - C & k ~ & . (1) Here j , is the molar flux density of species i, V,i& is the gradient in electrochemical potential of species k, and lik are ionic Onsager coefficients.Owing to the constraints imposed by local chemical equilibrium 4 k=1 - only two solute fluxes are independent. Isothermal diffusion in aqueous NaHSO, solutions is therefore a ternary', process. Because the single ion potentials appearing in eqn (1) are difficult to measure it is more convenient to analyse diffusion of NaHSO, in terms of fluxes Ji of neutral electrolyte components rather than fluxes j i of charged species. Owing to the exceptional mobility15 of H+, the total flux of H+ (free H+ plus bound Hi as HSO;) may be expected to exceed the flux of Na+ co-ion. Since any extra flux of H+ accompanying the main NaHSO, flux would constitute coupled flow of H,SO,, we will describe diffusion in aqueous NaHSO, solution in terms of fluxes of neutral NaHSO,(I) and H,SO,(2) components.In this representation the flux of the H,SO, component equals one-half of the total H+ flux in excess of the Na+ flux. J2 = +&+ -&a+)P (4) The flux of Na+ and the flux of the NaHSO, component are identical. J1 = jNa+. If a denotes the degree of bisulphate dissociation, a = [SOi-]/([HSO,] + [SO:-])B. Wiens and D. G . Leaist 249 the remaining species’ fluxes are given as follows: jH+ = d1+(1+a)J, (7) JHSOT = (1 - a) (4 + J,) (8) Jsoi- = a( J1+ J,). (9) Interacting flows of neutral NaHSO, and H,SO, components are related to measurable concentration gradients by an extension of Fick’s equation16 - J1 = Dll VC, + D12 VC, - J, = D,, VC, + D,, VC, (10) (1 1) in which Di, are ternary diffusion coefficients and ci are component concentrations in moles per unit volume.Diffusion Coefficients Ternary diffusion coefficients for NaHSO,(c,)-H,SO,(c, = 0 j H , O were determined conductimetrically at four NaHSO, concentrations ranging from 0.005-0.1 mol dm-3. The results are listed in table 1. Experimental precision is quoted as & two standard deviations estimated from runs performed in triplicate with different cells. Wendt4t5 has used the Gouy optical method to measure precise ternary diffusion coefficients for Na,SO,(O.S mol dm-3)-H,S0,(0.5 mol dm-3FH,0 and for Na,SO,( 1 .O rnol dm-3)-H,S0,( 1 .O mol dmM3)-H,0 at 25 “C. Bearing in mind the equilibrium Na,SO, + H,SO, = 2NaHS0, + H20, these solutions are identical in composition to NaHSO,( 1 .O mol dm-3)-H,S0,(0.0 rnol dmP3)-H,O and NaHS0,(2.0 mol dm-3)- H,SO,(O.O mol dmP3)-H,O, respectively.The optical data for equimolar Na,SO,-H,SO, mixtures should be closely related to diffusion coefficients for NaHSO,. The relationship can be established by noting that an aqueous solution of composition NaHSO,(c,)- H,SO,(c,) is identical to a solution of composition Na,SO,(c~)-H,SO,(c~) provided c; = c,/2 and ci = (c,/2) + c,. Concentrations c: are thus linear combinations of ci 2 where A , , = 0.5, A,, = 0, A,, = 0.5, and A,, = 1. It is easily shownll that ternary diffusion coefficients Di, for NaHSO,(c,)-H,SO,(c,) mixtures are obtained from diffusion coefficients Dik for Na,SO,(c~)-H,SO,(c~) mixtures by the linear transformation D = A-’D’A which provides D,, = D;,+D’,, (13) D,, = 2D;, (14) D2, = (-Oil - Di2 + Oil + Dk2)/2 (15) D,, = - D;, -I- D;,.(16) Ternary diffusion coefficients for NaHSO,(cl~H,SO,(c, = 0) mixtures that were derived from Wendt’s data for Na,SO,-H,SO, mixtures by use of eqn (13)-(16) are listed in table 1. Discussion It is evident from table 1 that values of D,, for dilute NaHSO, solutions are large and positive. Since D,, measures the coupled flow of H,SO, generated per unit gradient in concentration of NaHSO,, diffusion of aqueous NaHSO, is indeed a ternary process.250 Multicomponent Difusion of NaHSO, Table 1. Ternary diffusion coefficients for NaHSO,(c,)-H,SO,(c, = O)+H,O at 25 "C cJmol dm-3 4 1 a m2 s-' D2 1 /lop9 m2 s-l 0.000" 0.005b O.OIOb 0.050b 0.1 OOb 1 .00OC 2.0ooc 1.00 0.571 0.81 0.56+0.01 0.73 0.55 f 0.07 0.55 0.69 f 0.03 0.48 0.75 f 0.09 - 0.93 0.56 - - 1.476 -1.41 f0.12 - 1.21 f0.05 - 1.15k0.02 - 1.05 f 0.02 -0.35 - 0.06 1.710 1.43 &O.11 1.37 f0.15 0.91 k0.05 0.72 0.01 0.08 0.14 D22 / m2 s-l 4.899 4.27 & 0.04 4.01 k0.12 3.60 & 0.04 3.43 * 0.08 1.45 0.88 ________. ._ a Limiting value calculated from eqn (27)-(30). measurements on Na2S0,-H2S0, mixture^.^^ Conductimetric data. Derived from optical At 0.005 mol dm-3 NaHSO,, for example, D,, is twice as large as Dll, the diffusivity of the NaHSO, component. Although this solution contains no added H,SO,, the gradient in concentration of the NaHSO, component produces a parallel flow of H,SO, (- D,, Vc,) that is twice as large as the initial flow of the NaHSO, component (- D,, Vc,).This behaviour stands in sharp contrast to the diffusional properties of chemically inert in which a gradient in solute 1 is unable to generate coupled flow of solute 2 in a solution that is free of added solute 2, i.e. D,, -+ 0 as c,/c, + 0. At high NaHSO, concentrations where dissociation of bisulphate is less important, Na+ and HSO; are the major transporting species for the NaHSO, component. In this region the gradient in concentration of NaHSO, is incapable of producing significant additional flow of H+ ; values of D,, are therefore small and diffusion of NaHSO, is more nearly a binary process. When sulphuric acid diffuses in water, a relatively large electric field is induced in order to slow down highly mobile H+ and avoid significant charge separation. In aqueous NaHSO, solutions the electric field produced by diffusion of H,SO, has the additional effect of driving countertransport of Na+.For this reason values of D,, are negative. Limiting Diffusion Coefficients A detailed interpretation of diffusion of aqueous NaHSO, solutions is difficult because it involves two interacting solute flows consisting of Na+, HSO;, H+ and SO:- in proportions that vary with composition. Ternary diffusivities for the system are thus complicated weighted averages over the diffusivities of various species. To develop a model for diffusion of NaHSO, and other double salts we will make the dilute solution approximationls? l9 &k = 6,kEiDi/RT (17) where 6ik is the Kronecker delta, R is the gas constant, Tis the temperature, and ti and Di are the concentration and diffusion coefficients for species i.Diffusion coefficients for the solute components can then be estimated using the identity,, where Vik and Cim are stoichiometry coefficients which denote the number of moles of constituent ion i per mole of component k and species m, respectively, and ,uik = api/ac, is the derivative of the chemical potential of solute component i with respect to theB. Wiens and D. G. Leaist 25 1 concentration of solute component k. Superscript T indicates the transpose of a matrix. We will number the components, constituent ions, and species as follows component constituent ions species 1 NaHSO, 1 Na+ 2 H,SO, 2 H+ 3 HSO; which provides 1 0 0 P = O 1 0 i 0 0 1 Degrees of dissociation and derivatives Pik used from the exmessions 1 Naf 2 Hi 3 HSO, 4 so;- -;).1. to predict values of D i , were evaluated Pl = P? + RT ln [(I - 00 C l h + c2)A 931 f l 2 = Pi + RT In [(aci + (1 +a) c2) (1 -a) (C1-k c,)92.?3]. (22) (23) Corrections were made for departures from ideal solution thermodynamics using activity coefficients for the species estimated from the relation20 (24) I = (1+2a)(c,+c,). (25) l n j i = - 1.17~?Z~’~/(l +I1/,) where I denotes the ionic strength In fig. 1, ternary diffusion coefficients predicted for dilute NaHSO, solutions are compared with experimental results. Agreement is close, generally within +0.2 x m2 s-l. For D,,, however, predicted values are too low by ca. 65 x m2 s-l at concentrations above 0.05 mol dm-3. The calculations employed limiting diffusion coefficients for each solute species: D,(Na+) = 1.334 x D,(H+) = 9.3 15 x m2 s-l.These values were obtained from published limiting molar ionic cond~ctances~~~ 21 using the relation At high concentrations the effects of non-zero off-diagonal lik values are primarily responsible for discrepancies between calculated and observed Dik values. At infinite dilution the analysis of diffusion in NaHSO, solutions is simplified by complete dissociation of bisulphate. Furthermore, departures from the dilute solution model vanish. In this limit eqn (17)-(26) provide the following exact expressions for the ternary diffusion coefficients of aqueous NaHSO, D3(HS0,) = 1.363 x and D4(SOi-) = 1.065 x Di = RTil;/(zifl2. (26) Dl D4 D, i- D, i- 40, D, + D, + 40, DYl = 6 D1(D4 - D 2 ) DY2 = 2 = 0.511 x m2 s-l = - 1.476 x 1 0-9 m2 s-I252 Mult icomponen t Diflusion of N a H SO, I 1 I - 2 I-== 0.0 0.1 0.2 0.3 c,Y2/rno1112 drn-Y2 Fig.1. Comparison of observed and predicted ternary diffusion coefficients for NaHSO,(c,)- H,SO,(c, = 0)-H,O at 25 "C: 0, 0, conductimetric data; B, limiting values calculated from eqn (27)-(30); solid line, values predicted by eqn (17)-(26). = 1.710 x m2 s-l D,(D, - 0,) DZ1 = 3 D, + D, +4D, 2 0 , D2 - D, D, + 5D2 D, Do = = 4.899 x 10 -9 m2 s-l. D, + D, -+ 4 0 , 22 As shown in fig. 1 the measured ternary diffusivities extrapolate correctly to these limits. Although dilute aqueous NaHSO, is extensively dissociated to sulphate and rapidly diffusing hydrogen ions, the diffusivity of the NaHSO, component is surprisingly small, ca.one-half as large as the binary diffusivity for aqueous Na,SO, solution~.~-~ This apparent discrepancy can be understood by recalling that diffusion of NaHSO, generates flow of sulphuric acid. Because the electric field induced by diffusion of the acid drives countertransport of Na+, the ternary diffusivity of the NaHSO, is much smaller than experience with binary diffusion would suggest. By contrast, the ternary diffusivity of H,SO, in dilute NaHSO, solutions is almost twice as large as the binary diffusivitys*g of H,SO,. In binary solutions, however, hydrogen ions are constrained by electroneutrality to travel at the same rate as less mobile bisulphate ions, whereas hydrogen ions, and thus sulphuric acid, can diffuse more rapidly in NaHSO, solutions at the expense of counterflow of Na+.The limiting cross-coefficient Oil measures the flux of H,SO, generated by a gradient in fully dissociated NaHSO,. As mentioned earlier, the flux of H+ in excess of the flux of Na+ along a gradient in concentration of NaHSO, is responsible for the coupled flow of H2S0,. From eqn (29) we see that the limiting value for D2, is directly proportional to D, - D,, i.e. the diffusivity of H+ in excess of the diffusivity of Na+. This result stresses the importance of the difference in co-ion diffusivities in determining the multicomponent nature of diffusion of a double salt.B. Wiens and D. G. Leaist 253 Conclusions Isothermal diffusion in a solution of a double salt is a ternary process. A conductimetric procedure suitable for measuring diffusion coefficients of these salts has been successfully tested on aqueous sodium hydrogen sulphate solution.Previously published data for equimolar Na,SO,-H,SO, are a source of additional information on diffusion of NaHSO,. The experimental results together with limiting diffusion coefficients derived from ionic conductances provide ternary diffusion coefficients for the salt at 25 "C from infinite dilution up to 2 mol dmP3. In this system the large mobility of H+ relative to the mobility of Na+ leads to coupled flow of H,SO, whenever NaHSO, diffuses. Below 0.1 mol dm-3 the coupled flow of H,SO, produced by diffusion of NaHSO, exceeds the parent flow of NaHSO,. The authors gratefully acknowledge financial support by the Natural Sciences and Engineering Research Council of Canada.References 1 K. S. Pitzer and J. C. Peiper, J. Phys. Chem., 1980, 84, 2396. 2 G. Scatchard and R. C . Breckenridge, J . Phys. Chem., 1954, 58, 596. 3 W. H. Stockmayer, J . Chem. Phys., 1960, 33, 1291. 4 R. P. Wendt, J . Phys. Chem., 1965, 69, 1227. 5 R. P. Wendt, J . Phys. Chem., 1962,66, 1279. 6 H. S. Harned and C. A. Blake, J . Am. Chem. Soc., 1951, 73, 5880. 7 J. A. Rard and D. G. Miller. J. Solution Chem., 1979, 8, 755. 8 M. R. Savino and V. Vitagliano, Ric. Sci., 1962, 2, 341. 9 D. G. Leaist, Can. J . Chem., 1984, 62, 1692. 10 D. G. Leaist, J . Chem. Soc., Faraday Trans. I , 1984, 80, 3041. 11 R. A. Noulty and D. G. Leaist, Can. J. Chem., 1985, 63, 476. 12 A. K. Covington, J. V. Dobson and W. F. K. Wynne-Jones, Trans. Fararlay Soc., 1965, 61, 2057. 13 K. S. Pitzer, R. N. Roy and L. F. Silvester, J . Am. Chem. Soc., 1977, 99, 4930. 14 D. G. Leaist, J . Chem. Soc., Faraday Trans. I , 1982,78, 3069. 15 R. A. Robinson and R. H. Stokes, Electrolyte Solutions (Academic Press, New York, 2nd edn, 1959), 16 D. D. Fitts, Non-equilibrium Thermodynamics (McGraw-Hill, New York. 1962), chap. 7 and 8. 17 H. Kim, G. Reinfelds and L. J. Gosting, J . Phys. Chem., 1972, 76, 3419. 18 J . M. Creeth and R. H. Stokes, J . Phys. Chem., 1960, 64, 946. 19 C. W. Garland, S. Tong and W. H. Stockmayer, J . Phys. Chem., 1965, 69, 1718. 20 E. A. Guggenheim and J. C. Turgeon, Trans. Faraday Soc., 1955, 51, 747. 21 M. Kerker, J . Am. Chem. Soc., 1957, 79, 3664. Appendix 6.2. Paper 5/71 1 ; Receiced 30th April, 1985

ISSN:0300-9599    

DOI:10.1039/F19868200247

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

J . Chem. SOC., Faraday Trans. I , 1986,82, 255-261 A Comparison of the Conformation of Micellar and Non-micellar Sodium Octanoate and Octanoic Acid using 13C-lH and lH-lH Vicinal Spin-Spin Coupling Constants Frank Heatley Department of Chemistry, University of Manchester, Manchester M13 9PL lH-'H and lH-13C vicinal spin-spin coupling has been studied at 21 "C for micellar and non-micellar aqueous solutions of sodium octanoate, aqueous solutions of sodium butanoate and pentanoate, and CDCI, and CD,OD solutions of octanoic acid. For the octanoyl subjects, there is no detectable variation in the proportion of trans conformer with solvent or degree of association except for the CH,-CH, bond adjacent to the carboxy group which shows an increase in fractional trans content of ca.0.1 on micellisation. In sodium butanoate, the trans fraction is cu. 0.1 lower than in sodium pentanoate owing to the absence of four-bond steric conflicts. _____ Although a knowledge of the conformation of chains in mesophases is of considerable importance in understanding mesophase structure, the little experimental evidence available is conflicting. A Raman spectroscopy study of sodium octanoatel was interpreted in terms of an increase in trans conformer (fig. 1) on micellisation, though a similar study of various lipids2 concluded that the alkyl chain in the mesophases differs little from liquid hydrocarbons. From 13C chemical shifts, it was suggested that micellisation induces an increase in trans content throughout the chains in sodium octanoate and nonylammonium bromide.A similar inference was drawn from a computer model4 of the packing of a nonyl chain in micelles. However, studies of micellar sodium octanoate5 and sodium mono-octyl hydrogen phosphate6 using 13C: relaxation by paramagnetic ions indicate that there is a significant increase in trans content only for the two bonds next to the polar group, while the remaining bonds are similar to liquid hydrocarbons. H H G+ G- Fig. 1. Conformations of a -CH,--CH2-- unit. T In these studies the relationship between the measured quantities and the chain conformation is not simple and direct. A more satisfactory approach is to make use of vicinal spin-spin coupling constants whose variation with dihedral angle is well d~cumented.~ For surfactant alkyl chains, most CH, signals overlap in a single broad peak, so lH-lH coupling constants are generally not accessible.The exception is spin-spin 9 255 F A R 1256 Conformational Study by N.M. R. Spectroscopy coupling across the -CH,-CH,X bond, where X is the polar head-group, since the CH2X signal is usually well resolved owing to deshielding by X. In 13C n.m.r., however, most 13C signals are clearly resolved because of the larger chemical-shift range, thus giving the opportunity to pursue lH-13C spin-spin coupling as a conformational probe. With recent improvements in instrumental sensitivity and techniques for resolution enhancement it is feasible to obtain well resolved coupled natural abundance 13C spectra even at quite low concentrations such as those below the sodium octanoate c.m.c.$ of 0.4 mol dm-3.To observe such coupling clearly, it is necessary to eliminate dipolar coupling and to minimise natural broadening. These requirements are most easily obtained by studying micellar systems of small surfactants where the rapid molecular motion and isotropic character averages dipolar coupling to zero and gives relatively small natural linewidths of one or two Hz. Although of equal interest, liquid-crystalline phases suffer from non-zero dipolar coupling owing to their anisotropic structure. This coupling could be removed in principle by magic-angle r ~ t a t i o n , ~ though it is doubtful whether it could be reduced sufficiently to reveal conformationally significant spin-spin splittings of a few Hz. This paper reports a study of lH-lH and lH-13C spin-spin coupling constants in sodium octanoate in D20 above and below the c.m.c. This system was chosen because its relatively high c.m.c.facilitates the study of the free-molecule conformation, and because its low molecular weight increases the intensity per carbon site, reduces the chance of peak overlap and minimises the natural linewidth. For comparison we have also studied octanoic acid in CDC1, and CD30D, and sodium butanoate and pentanoate in D20. Experimental 'H and 13C spectra were recorded on Varian Associates SC-300 and XL-300 spectrometers. respectively, both operating at 300 MHz for protons and at a temperature of 21 "C. In order to make use of the 13Cg1H) nuclear Overhauser enhancement (n.0.e.) to improve the signal-to-noise ratio, the 13C spectra were acquired with the lH decoupler gated off (to preserve coupling) following a preparation period with the decoupler on (to establish the n.0.e.).Resolution enhancement was performed using the Lorentzian-Gaussian transformation technique.1° Spectra were run for solutions of sodium octanoate in D,O of concentration 8.13 mol dm-3 (non-micellar) and 0.85, 1.45 and 2.03 mol dmP3 (micellar), and for 0.5 mol dm-3 solutions of octanoic acid in CDC1, and CD,OD, and sodium butanoate and pentanoate in D,O. Results In the following, carbons are numbered according to their position in the chain, C-1 being the carboxy carbon. C-n designates the methyl carbon, C-(n - 1) the CH, adjacent to the methyl, and so on. Protons are numbered according to the carbon to which they are attached .lH-lH Coupling Values of lH-lH vicinal coupling constants are listed in table 1. The H-2 protons give a well resolved triplet at ca. 2.36 in all cases. The coupling constants are averaged over the trans (T) and gauche (G) conformations of a -CH2-CH2- fragment (fig. 1). Thus the H-2 and H-3 protons actually comprise an AA'BB' systemll with two distinct A-B coupling constants. If J H H and JkH are not too different, the spectrum may appear 'deceptively simple',12 i.e. each nucleus gives apparently a 1 : 2: 1 triplet as if JHH = J&. The separation of the triplet outer lines gives (JHH + JhH) exactly, and this is reported here. Similarly, the H-3-H-4 coupling patternsI;. Heatley 257 Table 1. lH-lH vicinal coupling constants for alkanoic acids and sodium alkanoatesa (JHH + J L d JHH ~ _ _ _ _ _ solute solvent conc./mol dm-, H-2-H-3 H-3-H-4 H-(n - l)-H-n - 7.4 k 0.1 0.5 14.6f0.1 14.9 0.1 14.8 f0.2 7.4 & 0.1 0.5 14.9 & 0.1 14.4k0.3 octanoate D2O 0.13 - 0.85 15.4kO.l - - 1.45 15.6 &O.1 14.4k0.3 - 2.03 15.6 kO.1 - - octanoic acid CD,OD 0.5 14.9 0.1 15.0 +0.3 - butanoate D2O pentanoate D2O octanoic acid CDC1, 0.5 14.9k0.1 - - ~ ~~~~ ~ _ _ _ _ _ _ _ _ _ _ _ ~~~~~ ~ a All values in Hz at 21 "C. in valerate and octanoates are also deceptively simple, the H-3 protons giving a quintet at ca. 1.6 6 (severely broadened in the octanoatesj indicating that the H-3-H-4 couplings are close to those between H-2 and H-3. In table 1, the values listed for (JHH+JhH) for H-3 and H-4 were obtained from the separation of the second and fourth lines of the quintet.The H-3 protons in butanoate and H-4 protons in pentanoate give sextets, since the methyl coupling is practically equal to the mean of the JHH and J;IH couplings to the preceding methylene. The H-4 to H-7 protons inclusive in octanoates give a broad singlet at ca. 1.3 6, while the CH, groups give a triplet (broadened in the octanoatesj at ca. 0.96. From table 1, it is seen that (JHH +JhHj for H-2-H-3 coupling varies somewhat with the system. The values for pentanoate, non-micellar octanoate and octanoic acid solutions are identical, but butanoate is slightly lower and micellar octanoate, interestingly, is rather higher. 13C-lH Coupling Even in the octanoates, each carbon is clearly resolved, owing to the wide dispersion of 13C chemical shifts.A chemical shift assignment for sodium octanoate has been reported.13 Because of the close match of the spectra, the same assignment was used for octanoic acid. The assignment of the lower salts was straightforward. Good coupled 13C spectra were obtained for all systems which with resolution- enhancement showed considerable geminal and vicinal C-H coupling. Some 13C nuclei gave first-order coupling patterns, or apparently so, but unfortunately many were strongly second-order, owing to strong coupling in the associated proton system. Fig. 2 and 3 illustrate these effects for 0.85 mol dm-3 micellar sodium octanoate. Fig. 2 shows the spectrum of C-1, a first-order triplet of triplets due to coupling to H-2 and H-3.Selective decoupling of H-2 proved the vicinal coupling to be the smaller. Fig. 3 shows the four lines of the methyl carbon, C-8, due to one-bond coupling to the attached protons. Although there are some further splittings due to geminal and vicinal coupling, note that the splittings are not symmetrical nor do they correspond to those expected for first-order effects. Similar, though less marked, effects were observed for most of the CH, carbons in the octanoyl systems. However, an important exception was C-2, where the longer-range couplings resulted in a first-order pentuplet pattern, indicating equivalence of the C-2-H-3 and C-2-H-4 couplings. In the lower salts, the 13C spectra were first-order except for C-3 and C-4 of pentanoate. The 13C-lH vicinal coupling constants readily obtainable are summarised in table 2.The C-2-H-4 butanoate coupling is distinct in being an average over three identical 9-2258 Conformational Study by N.M.R. Spectroscopy Ti I 1 1 1 I I 1 I I I I 1 1 I I I 1 1 1 I 1 1 1 1 1 I lln I 1 I I I I I 1 1 1 I 1 1 ~ 1 r 20 10 0 -10 -20 Hz Fig. 2. 'H-coupled 13C spectrum of the carbonyl signal of 0.85 mol dmP3 sodium octanoate at 21 "C. The f.i.d. was weighted with an exponential of time constant - 3.18 s. k- L -llllllrlrlrnllrlrlllrnlrrrrlrllrpllrlrlr-- 200 190 180 170 80 70 60 50 40 - 5 0 -60 -70 -80 -170-180 -190-200 Hz Fig. 3. lH-coupled 13C spectrum of the methyl carbon of 0.85 mol dm-3 sodium octanoate at 21 "C. Each line of the quartet due to the directly bonded protons is expanded separately. Resolution- enhancement has been applied to the f.i.d.using an exponential of time constant 0.07 s and a Gaussian of time constant 0.2 s.F. Heatley 259 Table 2. l3C--lH vicinal coupling constants for alkanoic acids and sodium alkanoatesa solute solvent conc./mol dm-3 C-1-H-3 C-2-H-4 C-n-H-(n - 2) butanoate D,O 0.5 4.4 5.5 4.3 pentanoate D,O 0.5 4.2 3.9 4.0 octanoate D,O 0.13 4.2 3.8 0.85 3.7 3.8 1.45 3.5 3.7 - 2.03 3.3 3.8 - - - octanoic acid CD30D 0.5 4.3 3.9 - ~~~~ ~ ~ ~~ ______ a All values in Hz at 21 "C with an uncertainty of & 0.1 Hz. conformations of a methyl group, whereas the others are averaged over the T and G conformations as for H-H coupling. The variation of Jc.l with the system parallels that of JH 2-H 3, as expected, and in addition shows a small decrease with increasing concentration in the micellar solutions.However the variation of J , 2-H is insignificant. In the case of the remaining 13C resonances which give second-order spectra, it was impractical to obtain accurate values of the coupling constants by simulation, owing to the large number of spins involved and the lack of sufficient resolved transitions. However, it was observed that for the octanoyl systems, the second-order effects were of the same form for all solutions. Hence it is possible at least to compare the conformations qualitatively by comparing the magnitude of corresponding fine structure splittings, even though these splittings do not necessarily represent actual coupling constants. It was found that for all practical purposes, the splittings were identical in both octanoic acid and sodium octanoate solutions. Discussion From the results above, it appears that apart from the C-2-C-3 bond, there is no evidence for significant changes in conformation between micellised and non-micellised sodium octanoate, between sodium octanoate in D20 and octanoic acid in other solvents, or between alkyl chains of different lengths. However both 'H-lH and 13C-lH vicinal coupling constants across the C-2-C-3 bond show some variations.Making the reasonable assumption that the coupling constants for trans and gauche dispositions are constant, the observed variations in the averaged values must arise from a change in the proportions of T and G conformations. If it is assumed that all gauche lH-lH coupling constants are the same, then the averaged value of (JHH + J;HH) is given by where PT is the fraction of the T conformation, and JtHH and JhH are coupling constants between trans and gauche proton pairs, respectively.Similarly the averaged vicinal 13C-lH coupling constant is Since it is well established that Jg << Jt for both lH-lH and 13C-lH coupling^,^ then (JHH+J;HH) increases and JCH decreases on increasing PT. Comparing the data for unassociated solutions, PT for bond C-2-C-3 is somewhat lower in sodium butanoate than in the higher alkyl chains. This result is understandable in terms of the so-called ' pentane effect' described by Flory.14 In a fragment, X-CH,-CH,-CH,-CO; (X = CH, in pentanoate and -CH, in octanoate), the conformations G+G- and G-G+ JCH = $ ( J t c H + J & r H ) + ~ P T ( J ~ H - J t c , ) .260 Conformational Study by N.M.R.Spectroscopy engender a severe steric conflict between the CO; and X groups which effectively suppresses these conformations. Although G* states are allowed in conformations TGk, G+T and G*Gk, the trans content is enhanced compared to butanoate where this interaction does not occur. Comparing data for micellar and non-micellar sodium octanoate, micellisation leads to a significant increase in PT for bond C-2-C-3. The small decrease in JC.l-H with increasing concentration in the micellar solutions is probably due to changes in the balance between free and micellised states. Exchange between these states is rapid on the n.m.r. timescale and the observed spectrum is a weighted average of the two populations.According to either the pseudophase separation or mass action models of micelle formation,lS above the c.m.c. the concentration of free monomers remains roughly constant at a value equal to the c.m.c., the remaining surfactant forming micelles. Thus the ratio of micellised to free molecules changes from c'u. 1 : I in the 0.85 mol dm-3 solution to ca. 5: 1 in the 2.03 mol dmP3 solution. Using these ratios, the extrapolated value of Jc 1pH.3 in the micelles is 3.0 Hz. To obtain quantitative estimates of PT, it is necessary to know Jt and JR. Unfortunately, these are not known accurately, but reasonable estimates for J h H and JgH can be obtained from a relationship given by PachleP between JHH, the dihedral angle and substituent electronegativity in substituted alkanes.For a C-CH,-CH,-C unit, this gives J h H = 13.3 and JhH = 3.8 Hz, and hence these values of PT: 0.47 (butanoate); 0.54 (pentanoate, non-micellar octanoate and octanoic acid) ; 0.64 (0.85 mol dm-3 octanoate) ; 0.68 (1.45 and 2.03 mol dmP3 octanoate). As a check on the reliability of this method, we may calculate the CH,-CH, coupling in butanoate or pentanoate which is averaged over symmetrical methyl rotation: JHH = #JAH +2JW,). For a C-CH,-CH, unit, Pachler's relationship yields J& = 13.6 and JhH = 4.0 whence JHH = 7.2 Hz, slightly less than the experimental value of 7.4 Hz. The absolute values of PT given above are therefore possibly over-estimated, though differences in PT are still reliable.For the C-3-C-4 bond in pentanoate, using the same assumptions, we obtain The angular and substituent dependence of 13C-lH coupling is quantitatively much less well understood than 'H-lH, though the overall behaviour appears to be similar. Using the values of PT derived above, the observed values of J , in butanoate, pentanoate and octanoate are reproduced to within 0.1 Hz for J& = 10.5 and JEH = 2.2Hz, values which are consistent with those quoted by Bystrov.' Although it is not possible to determine PT for other bonds in the octanoate chain, the fact that the fine splittings in the micellar and non-micellar 13C spectra are the same within experimental error of 0.1 Hz indicates that PT differs by < 0.05 between the two environments . The results presented here are in qualitative agreement with those reported for sodium octanoates and sodium mono-octyl hydrogen phosphate6 using 13C relaxation by paramagnetic ions.Unfortunately quantitative data for the octanoates were limited to the imprecise statement that 'chain confirmations differ from those of a hydrocarbon by their confinement in the non-polar core of the micelle, which gives rise to an increase in the trans population from 60% up to more than 80% at the chain end'. It is not clear which chain end is meant, but presumably it is the polar end. For the phosphate,6 more precise information was provided in the form of AEGT for each bond where AEGT is related to Rr = 0.5. Pr = [ 1 + 2 exp (- AEGT/RT)]-l. From the data given for AEGT, we obtain PT values at 21 "C of 0.58 and 0.48 for the first and second CH,-CH, bonds, respectively (relative to the polar end), and 0.43 forF.Heatley 26 1 the remaining four bonds, this last value being characteristic of liquid hydrocarbons. It was however noted that the relaxation data were not sensitive to the conformation of bonds near the methyl group, and the results quoted were ‘rather arbitrary’. The present results are therefore the most relevant to the conformation of this part of the chain. Numerical agreement is reasonable considering differences in molecular structure and various assumptions used in the analysis of experimental data. In contrast to the conformational pattern deduced from coupling constants and 13C relaxation, the I3C chemical shift changes3 on micellisation were interpreted in terms of an increase in PT for every bond, the increase being least at the extremities.It is likely that the observed shifts are principally due not to changes in conformation, but to changes in the medium, to which chemical shifts are also susceptible. The computer model of nonyl chains in micelles4 results also in a more uniform change in PT than found experimentally, from PT z 0.7 in a free chain to x 0.8 in the micelle. This discrepancy therefore raises reservations about the validity of the model’s description of the details of chain structure, though the overall micelle structure may be adequately represented. Finally we note that the results presented here depict, for the most part, a highly disorganized micelle structure. As has been pointed out previo~sly,*-~ the substantial probability of gauche conformations enables the methyl group to sample all regions of the micelle. It is interesting that such conformational changes as do occur on micellisation are confined to the first bond. It is possible that these results from a degree of crystal-like correlation between molecules necessary to obtain a reasonably coherent micelle surface. References 1 J. B. Rosenholm, K. Larsson and N. Dinh-Nguyen, Colloid Pnlym. Sci., 1977, 255, 1098. 2 K . Larsson, Chem. Phys. Lipids, 1972, 9, 181. 3 B-0. Persson, T. Drakenberg and B. Lindman, J . Phys. Chem., 1976, 80, 2124. 4 D. W. R. Gruen, Prog. Colloid Polym. Sci., 1985, 70, 6. 5 T. Zemb and C. Chachaty, Chem. Phys. Lett., 1982, 88, 68. 6 Y. Chevalier and C. Chachaty. J . Phys. Chem., 1985, 89, 875. 7 V. F. Bystrov, Prog. Nucl. M a p . Reson. Spcrtrosc., 1975, 10, 41. 8 B. Lindman and H. Wennerstrom, Top. Curr. Chem., 1980, 87, 1. 9 E. R. Andrew, Int. Rev, Phys. Chem., 1981, 1, 195. 10 A. G. Ferrige and J. C. Lindon, J . M a p . Reson., 1978, 31, 337. 11 N. Sheppard and J. J. Turner, Proc. R. Soc. London, Ser. A , 1959, 252, 506. 12 R. J. Abraham and H. J. Bernstein, Can. J. Chem., 1961, 39, 216. 13 T. Ahlnas, 0. Soderman, C . Hjelm and B. Lindman, J. Phys. Chem., 1983, 87, 822. 14 P. J. Flory, Statistical Mechanics of Chain Molecules (Interscience, New York, 1969), p. 138. 15 H. Wennerstrom and B. Lindman, Phys. Rep., 1979, 52, 1. 16 K. G. R. Pachler, J . Chem. SOC., Perkin Trans. 2, 1972, 1936. Paper 51960; Receiced 6th June, 1985

ISSN:0300-9599    

DOI:10.1039/F19868200255

出版商:RSC    

年代:1986

数据来源: RSC